Cephalopod Cognition

Cephalopods are generally regarded as the most intelligent group among the inverte-
brates. Despite their popularity, relatively little is known about the range and function of
their cognitive abilities. This book fills that gap, accentuating the varied and fascinating
aspects of cognition across the group.
Starting with the brain, learning and memory, Part I looks at early learning, mem-
ory acquisition and cognitive development in modern cephalopods. An analysis of the
Chambered Nautilus, a living fossil, is included, providing insight into the evolution
of behavioural complexity. Part II surveys environmental responses, especially within
the active and learning-dependent coleoids. The ever-intriguing camouflage abilities
of octopuses and cuttlefish are highlighted, alongside bioluminescence, navigation and
other aspects of visual and cognitive competence.
Covering the range of cognitive function, this text underscores the importance of
the cephalopods within the field of comparative cognition generally. It will be highly
valuable for researchers, graduates and senior undergraduate students.

Anne-Sophie Darmaillacq is a researcher in the Group Memory and Behavioural Plasticity

Unit at Caen University in France. She studies embryonic cognition and early learning
systems in the cuttlefish Sepia officinalis and lectures in ethology at graduate and
undergraduate levels.

Ludovic Dickel is Professor of Behavioural Biology at Caen University, managing the

Group Memory and Behavioural Plasticity Unit. His research interests are focused on
brain development and cognition in cuttlefish as well as collaborative work on embryonic
neuroethology in oviparous species.

Jennifer Mather is Professor in the Department of Psychology at the University of

Lethbridge in Alberta, Canada. She is active in researching many aspects of cephalopod
cognition and behaviour, especially in octopuses, including ground-breaking work in
personality, play and consciousness.
Cephalopod Cognition
Edited by

ANNE-SOPHIE DARMAILLACQ
University of Caen Basse-Normandie, France

LUDOVIC DICKEL
University of Caen Basse-Normandie, France

JENNIFER MA THER
University of Lethbridge, Alberta, Canada
University Printing House, Cambridge CB2 8BS, United Kingdom

Cambridge University Press is part of the University of Cambridge.

It furthers the University’s mission by disseminating knowledge in the pursuit of
education, learning and research at the highest international levels of excellence.

www.cambridge.org
Information on this title: www.cambridge.org/9781107015562

C Cambridge University Press 2014
This publication is in copyright. Subject to statutory exception
and to the provisions of relevant collective licensing agreements,
no reproduction of any part may take place without the written
permission of Cambridge University Press.
First published 2014
Printed in the United Kingdom by TJ International Ltd. Padstow Cornwall
A catalogue record for this publication is available from the British Library
Library of Congress Cataloguing in Publication data
Cephalopod cognition / edited by Anne-Sophie Darmaillacq, University of Caen
Basse-Normandie, France, Ludovic Dickel, University of Caen Basse-Normandie,
France, Jennifer Mather, University of Lethbridge, Alberta, Canada.
pages cm
Includes bibliographical references and index.
ISBN 978-1-107-01556-2 (hardback)
1. Cephalopoda – Behavior. 2. Invertebrates – Behavior. 3. Cognition in animals.
I. Darmaillacq, Anne-Sophie, 1977– II. Dickel, Ludovic. III. Mather, Jennifer A.
QL785.C34 2014
594 .5 – dc23 2014009749
ISBN 978-1-107-01556-2 Hardback
Cambridge University Press has no responsibility for the persistence or accuracy of
URLs for external or third-party internet websites referred to in this publication,
and does not guarantee that any content on such websites is, or will remain,
accurate or appropriate.
To the next generation

ASD and LD: To Nitzan and Pia

JM: To Charlotte and Nicky

 Contents

List of contributors page x

Preface xiii
Acknowledgements xvi
Tribute to Martin J. Wells xvii
Part I Cognition, brain and evolution

1 Cuttlefish preschool or how to learn in the peri-hatching period 3

Anne-Sophie Darmaillacq, Christelle Jozet-Alves, Cécile Bellanger and Ludovic Dickel

1.1 Introduction 3
1.2 Development of the sensory systems 4
1.3 Early learning about prey 6
1.4 Lateralization 12
1.5 Phenotypic plasticity and defensive behaviour 15
1.6 The brain and its neurotransmitters 17
1.7 Conclusion 23

2 Evolution of behavioral and neural complexity: learning and memory in

Chambered Nautilus 31
Jennifer Basil and Robyn Crook

2.1 Introduction and comparative approaches 31

2.2 Sensory ecology 32
2.3 Neural systems 35
2.4 Associative conditioning 37
2.5 Spatial navigation 42
2.6 Evolution 46
2.7 Conclusions 48

3 Learning from play in octopus 57

Michael J. Kuba, Tamar Gutnick and Gordon M. Burghardt

3.1 Introduction 57
3.2 Exploration 58
3.3 Play 61
3.4 Conclusions 67
viii Contents

4 The neurophysiological basis of learning and memory in an advanced

invertebrate: the octopus 72
Binyamin Hochner and Tal Shomrat

4.1 Introduction 72
4.2 The anatomy of the vertical lobe system 74
4.3 Neurophysiology of the vertical lobe 76
4.4 Synaptic plasticity in the vertical lobe 80
4.5 Mechanism of LTP induction in the octopus vertical lobe 81
4.6 Neuromodulation in the vertical lobe 84
4.7 Are the vertical lobe and its LTP involved in behavioral learning and
memory? 86
4.8 Conclusion: a system model for the octopus learning and memory 88

5 The octopus with two brains: how are distributed and central representations
integrated in the octopus central nervous system? 94
Frank W. Grasso

5.1 The relationship between cognition and neuroscience in cephalopod

cognition 94
5.2 The problem of soft bodies controlled by complex brains 96
5.3 Representations and connectivity 96
5.4 The organization of the octopus central nervous system 99
5.5 The computational roles of the arm module 104
5.6 A network of homogeneous ganglia 105
5.7 Relationships between a network of local brachial modules and the
cerebral ganglia 112
5.8 The octopus with two brains 117

Part II Cognition and the environment

6 Foraging and cognitive competence in octopuses 125

Jennifer A. Mather, Tatiana S. Leite, Roland C. Anderson and James B. Wood

6.1 Introduction 125

6.2 Octopuses as generalist foragers 126
6.3 Foraging as cognitively demanding 130
6.4 Predation avoidance and learning 137
6.5 Conclusion 143

7 Navigation in cephalopods 150

Christelle Jozet-Alves, Anne-Sophie Darmaillacq and Jean G. Boal

7.1 Introduction 150

7.2 Sensory systems 150
7.3 Dispersal 153
7.4 Migrations 156
 Contents ix

7.5 Local movements 158

7.6 Summary and future directions 166

8 Camouflage in benthic cephalopods: what does it teach us? 177

Noam Josef and Nadav Shashar

8.1 Brief historical review 177

8.2 Definitions and nomenclature 177
8.3 Dynamic versus static camouflage 179
8.4 Neural physiology 182
8.5 Body patterns 182
8.6 Camouflage complexities 183
8.7 Camouflage limitations 184
8.8 Communication and courtship 185
8.9 Quantifying camouflage 185
8.10 How do cephalopods decide what camouflage strategy to use? 188
8.11 Camouflaged to resemble what? 189
8.12 Concluding remarks 190

9 Cuttlefish camouflage: vision and cognition 197

Sarah Zylinski and Daniel Osorio

9.1 Cephalopod camouflage and comparative cognition 197

9.2 Animal visual cognition 198
9.3 Expression of body patterns 201
9.4 Cuttlefish vision 203
9.5 Making decisions about body patterns 214
9.6 Concluding remarks 217

10 Visual cognition in deep-sea cephalopods: what we don’t know and

why we don’t know it 223
Sarah Zylinski and Sönke Johnsen

10.1 The other cephalopods 223

10.2 Why do we know so little about the other 95%? 224
10.3 Do mesopelagic cephalopods need to be visually cognitive? Is there
evidence to suggest that they are? 227
10.4 Body pattern repertoire in the open ocean and deep sea 228
10.5 Evidence for complex visual behaviours and body patterning in
mesopelagic cephalopods 231
10.6 Bioluminescence: using and detecting 232
10.7 Concluding remarks 236

Index of species 242

Index 244
Contributors

Roland C. Anderson
Former Biologist, Seattle Aquarium, Seattle, WA, USA

Jennifer A. Basil
Associate Professor, LIBE Laboratory, Department of Biology, Brooklyn College,
CUNY, Brooklyn, New York, NY, USA

Cécile Bellanger
Assistant Professor, Groupe Mémoire et Plasticité comportementale, EA4259,
Université de Caen Basse-Normandie, Caen, France

Jean G. Boal
Professor, Department of Biology, Millersville University, Millersville, PA, USA

Gordon M. Burghardt
Professor, Departments of Psychology and Ecology and Evolutionary Biology,
University of Tennessee, Knoxville, TN, USA

Robyn Crook
Research Fellow, Department of Integrative Biology and Pharmacology, University of
Texas Health Science Center, Houston, TX, USA

Anne-Sophie Darmaillacq
Assistant Professor, Groupe Mémoire et Plasticité comportementale, EA4259,
Université de Caen Basse-Normandie, Caen, France

Ludovic Dickel
Professor, Groupe Mémoire et Plasticité comportementale, EA4259, Université de
Caen Basse-Normandie, Caen, France

Frank W. Grasso
Associate Professor, Biomimetic and Cognitive Robotics Laboratory, Department of
Psychology, Brooklyn College, CUNY, Brooklyn, New York, NY, USA
 List of contributors xi

Tamar Gutnick
Doctoral Fellow, Department of Neurobiology, Institute of Life Science, Hebrew
University, Jerusalem, Israel

Binyamin Hochner
Professor, Department of Neurobiology, Silberman Institute of Life Sciences and the
Interdisciplinary Center for Neural Computation, Hebrew University, Jerusalem, Israel

Sönke Johnsen
Professor, Biology Department, Duke University, Durham, NC, USA

Noam Josef
Doctoral Fellow, Department of Life Sciences, Ben-Gurion University of the Negev,
Eilat Campus, Eilat, Israel

Christelle Jozet-Alves
Assistant Professor, Groupe Mémoire et Plasticité comportementale, EA4259,
Université de Caen Basse-Normandie, Caen, France

Michael J. Kuba
Research Fellow, Department of Neurobiology, Institute of Life Science, Hebrew
University, Jerusalem, Israel

Tatiana S. Leite
Professor, Departamento de Oceanografia e Limnologia, Federale Universite do Rio
Grande do Norte, Via Costeria s/n Mae Luiza, Natal, Brazil

Jennifer A. Mather
Professor, Department of Psychology, University of Lethbridge, Lethbridge, AB,
Canada

Ronald O’Dor
Professor of Biology, Dalhousie University, Halifax, NS, Canada

Daniel Osorio
Professor, School of Life Sciences, University of Sussex, Sussex, UK

Nadav Shashar
Professor, Department of Life Sciences, Ben-Gurion University of the Negev, Eilat
Campus, Eilat, Israel
xii List of contributors

Tal Shomrat
Postdoctoral Fellow, Department of Neurobiology, Silberman Institute of Life
Sciences, Hebrew University, Jerusalem, Israel

James B. Wood
Mounts Botanical Garden, West Palm Beach, FL, USA

Sarah Zylinski
Lecturer, School of Biology, University of Leeds, Leeds, UK
Preface

In July of 2012, at the Francis Crick Memorial Conference on Consciousness in Human

and Non-Human Animals, a group of researchers signed The Cambridge Declaration
on Consciousness. What was important about this declaration was that it firmly stated
that consciousness can emerge in animals that evolved along different evolutionary
tracks, including cephalopod mollusks. This recognized what researchers in cephalopod
behaviour, starting with the ground-breaking work of J. Z. Young and Martin Wells over
50 years ago, have been maintaining, that these animals are ‘learning specialists’ and
capable of ‘complex cognition’. The chapters of this book will help define and discuss
this cephalopod cognition, giving insight not just into how their brain and behaviour
work, but also a comparative view that can be contrasted to research in ‘traditional’
vertebrates.
Cognition has been difficult to define and has been described more or less broadly
over the years (Shettleworth, 2010). In general terms, Neisser (1967) defined it as ‘all the
processes by which the sensory input is transformed, reduced, elaborated, stored, recov-
ered and used’ (and see chapter 6, this volume, by Mather et al.). Deukas (1998) defined
it similarly as ‘the neuronal processes concerned with the acquisition and manipulation
of information by animals’. Richardson’s (2010) definition that cognitive systems ‘dis-
ambiguate confusing, incomplete and rapidly changing data, while generating equally
unique, yet adaptive responses’ reminds us that this information processing is not static
and works across time. Shettleworth (2010) starts with a definition that is similar to
these, but points out that many believe it to be too broad. She notes that cognition might
need to be declarative (knowing that) rather than procedural (knowing how). Similarly,
cognition might need to act on mental representations. First-order operation (operating
directly on perceptual input) would not be cognitive, whereas second-order ones (act-
ing on first order ones) would be. She reminds her readers that without experimental
analysis, it is impossible to tell what kinds of processes are reflected in a behaviour, a
point nicely analyzed by Webb (2012) for insects. She also notes that different kinds of
animals might accomplish functionally similar behaviour in different ways, an important
point for the comparison of cephalopod cognition with that or animals from other phyla
(see the chapter 7, this volume, by Jozet-Alves et al.). All of these ideas are incorporated
in different ways in the chapters of this book.
Although comparative cognition was probably not even defined when Martin J. Wells
was doing research in the 1950s and 1960s, his work definitely fits into that category
and was also the basis for much of the work we are doing on cephalopod cognition now.
xiv Preface

His initial orientation was physiology, but Wells moved the study of octopus learning
from visual to tactile, and located a second area in the brain concerned with learning. He
conducted many of the lesion studies that laid the groundwork for our understanding of
the brain’s control of behaviour (see chapter 4, this volume, by Hochner & Shomrat). But
his investigation of the function of the optic gland was a pioneering step in the under-
standing of invertebrate neuroendocrinology. He was always interested in cardiovascular
physiology and his work was the foundation for later study in this area, particularly by
Ronald O’Dor. Later in his career he became intrigued with the lifestyle and adaptation
of nautiloids, and while he may have underestimated their ability (see chapter 2, this
volume, by Basil & Crook), he laid the foundation for further research in this area too.
And we still find his book, written in 1978, a useful source of information (Wells, 1978).

Part I: cognition, brain and evolution

Cephalopods are often described as amazing brainy creatures; this is probably why they
are the first invertebrates considered as ‘sensitive’ beings and, hence, the use of them is
now regulated by a European directive. This is fully explored in the first part of this book.
Given the absence of parental care to the eggs and juveniles, Darmaillacq et al. (chapter 1)
highlight how the cuttlefish is a suitable model to address developmental questions. They
review the remarkable memory and other cognitive skills from embryonic stages. From
a comparative point of view, Basil and Crook (chapter 2) illustrate learning in the
ancient cephalopod Nautilus pompilius. This chapter addresses how Nautilus can be
used for a better understanding of evolution of cognition and neural complexity in its
more derived relatives, the coleoid cephalopods (octopuses, cuttlefish and squid). In
contrast, Kuba et al. (chapter 3) underline the interest in the ‘sophisticated’ octopus
as a model to study object play, a high-order behavioural emergence in the animal
kingdom. Hochner and Shomrat (chapter 4) underline the spectacular functional and
structural similarities of memory systems between octopus and mammals, despite very
different brain organization. Lastly, chapter 5 by Grasso gives an authoritative account of
current understanding of the octopus nervous system and a clearly personal view of the
functional architecture. The hierarchical system proposed in this chapter could be applied
in robotics and cybernetics, and the ideas are challenging and may be controversial.

Part II: cognition and the environment

Of course cognitive abilities evolved through and are used in the natural environment.
But it is not always easy to study them or even to theorize how they are used in the marine
environment. Given that cephalopods are mostly not social animals, their cognitive ability
is not, as it is in the ‘higher’ vertebrates, used to solve social problems and understand the
actions of conspecifics. Instead, Mather et al. (chapter 6) argue that foraging in the face
of predator pressure is the foundation for the development of intelligence and perhaps
consciousness, particularly of octopuses. All cephalopods are mobile, and Jozet-Alves
 Preface xv

et al. (chapter 7) point out the sensory processing and cognitive competence involved in
both long-distance and local movement, including laboratory investigation of the latter
activities. And one of the most fascinating aspects of the cephalopods’ lives is their
skin display system. Josef and Shashar (chapter 8) discuss the basic aspects of benthic
cephalopods’ camouflage and Zylinski and Osorio (chapter 9) review the recent in-depth
research on how cuttlefish use visual information about the background to produce the
precise matching that they perform so well. Finally, Zylinski and Johnsen (chapter 10)
challenge us to look more widely at the many species across the class. They point out
that the deep-sea cephalopods, though little known, have abilities in visual cognition and
body pattern production quite different from the well-known near-shore species. This
range of adaptations leaves us much more to investigate about cephalopod cognition as
it is adapted to a wide variety of habitats.

References

Deukas, R. (ed.) (1998). Cognitive Ecology. Chicago, IL: University of Chicago Press.
Neisser, U. (1967). Cognitive Psychology. New York, NY: Appleton-Century-Crofts.
Richardson, K. (2010). The Evolution of Intelligent Systems. London: Palgrave Macmillan.
Shettleworth, S. J. (ed.) (2010). Cognition, Evolution and Behaviour. New York, NY: Oxford
University Press.
Webb, B. (2012). Cognition in insects. Philosophical Transactions of the Royal Society of London
B, 367: 2715–2722.
Wells, M. J. (1978). Octopus – Physiology and Behaviour. Chichester, Sussex: John Wiley and
Sons, Inc.
Acknowledgements

This book emerges from the symposia about cognition in cephalopods co-organized
by A.-S. Darmaillacq, L. Dickel (University of Caen, France), N. Shashar (University
of Ben Gurion, Israel) and Y. Ikeda (University of Ryukyus, Japan) at the 31st Inter-
national Ethological Conference held in Rennes (France) in 2009, entitled ‘Studying
cephalopods’ brain and behaviours: when molluscs allow a better understanding of the
mechanisms and evolution of cognition: Part I and Part II’. We are very grateful to the
scientific committee of the International Ethological Conference for having accepted
these symposia. The editors would like to thank all of the contributors for their hard
work and enthusiasm while producing this volume. They also thank all the reviewers
whose comments have made valuable contributions to each of the chapters. We would
like to thank Martin Griffith of Cambridge University Press for his support, enthusi-
asm and encouragements, and Megan Waddington for her patience. It is encouraging to
see a range of contributors from both young researchers to more established experts in
cephalopod cognition.
Financial support for the research of A.-S. Darmaillacq and L. Dickel on cognition
in cuttlefish was supported by the French ministry of research, the regional council of
Basse-Normandie and the University of Caen Basse-Normandie. We thank the Société
Française pour l’Etude du Comportement Animal for promoting behavioural research.
Tribute to Martin J. Wells

Martin Wells, who died aged 80 in 2009, was not only a distinguished biologist with a
passion for invertebrates, but also a colourful and stimulating personality who enthused
generations of students with the sheer excitement and beauty of studying animals,
especially cephalopods – squids and octopuses. He began his research into cephalopod
learning when he and his wife, Joyce, abandoned their PhD programs at Cambridge
to move to the Stazione Zoologica in Naples, Italy, as the Director in 1953. Acting
on a suggestion from J. Z. Young, who had discovered a way of training octopuses
to make visual discriminations, they began studying tactile learning in octopus. They
soon showed that these animals could discriminate between objects on the basis of touch,
using the suckers on their arms, and that octopus suckers contain chemoreceptors so they
can learn to ‘taste’ what they touch. This ‘tasting by touching’ is extremely sensitive,
enabling octopuses to distinguish between snails and stones as their arms explore their
surroundings at night or in murky waters.
Collaborations with J. Z. Young continued into the 1970s, building up a map of octopus
brain function based on surgically ‘removing bits’, seeing what functions changed and
then documenting exactly what was removed microscopically. This eventually resulted
in octopus brains becoming the best-known non-vertebrate biological data processor
ever studied. Ultimately, cephalopods’ brains were so different that they were referred
to as ‘the only alien intelligence that humans have encountered’. A bit later Martin tried
to work backward towards the origins of this intelligence by studying ancient Nautilus
brains, but they turned out to be not very smart, so he began referring to them as ‘racing
snails’.
Martin also discovered that in the cephalopod brain there is an analogue of the verte-
brate pituitary gland: the optic gland, closely associated with sexual maturation. Martin
and Joyce’s 1959 paper in the Journal of Experimental Biology on this topic became
a classic in the literature of invertebrate endocrinology. Martin’s interest in the 1970s
turned to various aspects of cephalopod cardiovascular and respiratory physiology. Col-
laborating with colleagues from around the world, he published a series of challenging
papers attempting to relate physiology to the life of the whole animal in its environment.
His more recent studies on nautiluses yielded fascinating data about the physiology of
an animal that regularly moves from the surface to depths of more than 700 m, and also
taught us much about the reasons for the eventual failure, from an evolutionary perspec-
tive, of the shelled cephalopods (the ammonites and belemnites) that once dominated
the ancient seas.
xviii Tribute to Martin J. Wells

On the basis of his work in Naples, Martin was elected to a prize fellowship at Trinity
College, Cambridge, in 1956, and in 1959 he was appointed a university demonstrator
in the Cambridge zoology department. He soon became one of five founder fellows of
Churchill College, and a tutor and director of studies in biology. In 1966 Martin was
awarded a Cambridge ScD and the silver medal of the Zoological Society of London
in 1968. He was made a university reader in 1976. With Joyce, he travelled extensively
in search of cephalopods, and colleagues with whom to study them. Among many
destinations that he visited (often as a visiting professor) were Duke University in North
Carolina, Hawaii, Ghana, Dalhousie in Canada, Papua New Guinea, Australia, Texas
and Uganda.
Martin approached marine biology as a ‘way of life’. He used his position in Cam-
bridge to create a global cadre of marine biologists of the cephalopod persuasion. Post-
doctoral fellows and graduate students from around the world were welcomed to his
laboratory, his college, his house, his marine stations and his boats. They were expected
to make wine and, occasionally, garden in the Bury, his home. Perhaps only half of the
world’s cephalopod biologists enjoyed Martin and Joyce’s hospitality over the years, but
this community will never forget their influence. Several generations now hark back
to it. Martin’s approach was not just a scientific but also an intellectual exercise that
expanded the minds of those lucky enough to be involved. He was a writer of popular
science books, essays and rigorous papers, a novelist, painter and a yachtsman, talents
he came by honestly as the grandson of H. G. Wells. Living with the entire Wells family
shaped minds.
Martin’s philosophy was always, ‘Why tackle easy questions when the hard ones
are so much more interesting?’ I think the hardest question he ever asked was, ‘Why
do cephalopods need to be so smart when they die so young?’ The phrase ‘live fast,
die young’ has become a popular descriptor of the cephalopod lifestyle, but perhaps
the phrase should be expanded to ‘live fast and smart, to leave your offspring fewer
enemies’. Martin died on the first day of Darwin’s bicentenary, so a Darwinian answer
seems appropriate.

Ronald O’Dor
Part I

Cognition, brain and evolution

1 Cuttlefish preschool or how to learn
in the peri-hatching period
Anne-Sophie Darmaillacq, Christelle Jozet-Alves, Cécile Bellanger
and Ludovic Dickel

1.1 Introduction

One of the greatest challenges faced by many precocial young after hatching is that of
finding food without being eaten. This challenge is particularly important in species
where offspring receive no parental care. In these species, newly hatched young have
at least two possible options in response to the problem of prey and predators: ‘come
prepared’ (e.g. unlearned behaviour), which is safe but rigid, or ‘learn as you go’ (e.g.
trial-and-error learning), which is risky if they do not learn quickly enough, but more
flexible for adapting to a changing environment.
For a better understanding of how a newborn animal deals with environmental chal-
lenges, we need to study what is perceived and learned before hatching. Our understand-
ing of prenatal cognitive activity is comparatively recent, compared for example with
philosophical trends of the seventeenth century. A cognitive system can be defined as
one which is able to perceive and extract information from its environment, and then
use that information for the purpose of making appropriate decisions and developing
suitable behaviour (Shettleworth, 2010; Vauclair, 1996). There is no longer any doubt
about the embryo as a cognitive system, since it is now well established that fetal sen-
sory experience has a crucial role in behavioural and cognitive development (Krasnegor,
Blass, Hofer & Smotherman, 1987).
Although there are 700 cephalopod species, the common cuttlefish Sepia officinalis has
been the focal subject of intensive developmental studies in neuroethology. Cuttlefish,
like many other cephalopod species, start life under strong selective pressure from their
environment; neither eggs nor juveniles benefit from parental care (Boletzky, 1987), and
so young individuals must find their own means of feeding themselves and avoiding
predators. Female S. officinalis lay hundreds of eggs in shallow water on various rigid
supports: algae stems, tubeworms, ropes, nets, etc. (Boletzky, 1983). Eggs laid in clusters,
popularly known as ‘les raisins de mer’ (sea grapes), are abandoned after the death of the
mother. The egg capsule consists of a chorion and spirally coiled black envelopes, which
protect the embryos against microbial attack and predation (Boletzky, 2003). Throughout
embryonic development, the capsule is enlarged as osmotic pressure increases due to
water entering the perivitelline space, becoming more translucent shortly before the

Cephalopod Cognition, eds A.-S. Darmaillacq, L. Dickel and J. Mather. Published by Cambridge University
Press. 
C Cambridge 2014.
4 Anne-Sophie Darmaillacq, Christelle Jozet-Alves, Cécile Bellanger and Ludovic Dickel

embryos hatch. Once the animals have hatched, egg envelopes decay after some weeks
of degradation (Billings, Sullivan & Vine, 2000). Unlike many other cephalopods, there
is no larval or paralarval stage in cuttlefish development; the hatchlings are very similar
to adults in their general form, which facilitates comparative development studies. They
possess the same benthic life style as adults; they can also swim in the water column,
achieve elaborate forms of crypsis by colour pattern and posture changes, and use
the same prey-catching strategies as adults, by striking out with their tentacles or by
jumping on prey. They seem to possess a behavioural repertoire comparable with that
of adults, although some adjustments in both predatory and defensive behaviour occur
during development (for review, see Dickel et al., 2006). The brain of early juveniles has
the same structure as that of adults, and all the brain lobes are present from hatching.
Neurogenesis is intense during development (e.g. in octopus, the adult brain possesses
a thousand times more neurones than that of the hatchling) and some brain structures
develop faster than others during post-embryonic maturation (for a review, see Nixon
& Young, 2003). From an experimental point of view, these characteristics make the
cuttlefish a unique animal model in developmental neuroethology; firstly, because the
absence of parental care allows a precise control of the experiential history of embryo and
juvenile; and, secondly, because the absence of a paralarval stage in young cuttlefish
allows longitudinal and comparative studies of brain and behaviour from hatching to
adulthood.
This chapter is a review of the most recent studies addressing the influence of early
experience in S. officinalis, from embryonic stages, to post-hatching behaviour and
cognition. It will first describe the development of the sensory systems and the central
nervous system in the embryo. It will then focus on the factors affecting the development
of prey preference and the efficiency of primary defences in newly hatched cuttlefish
and juveniles. It will also address the ontogeny of lateralization in cuttlefish, which may
afford juveniles greater behavioural efficiency by enabling them to look out for escape
routes while hunting. Finally, we will try to describe the scenario of the beginning of the
life of a cuttlefish.

1.2 Development of the sensory systems

The study of cognition is concerned with how animals process information, beginning
with how information is acquired by the senses (Shettleworth, 2010). As all of an animal’s
experience is founded on its ability to sense changes in its environment, measurement
of the embryo’s sensory capabilities allows us to define the limits of its potential for
modifying its behaviour through experience (Smotherman & Robinson, 1992). The onset
of sensory system function follows an invariant sequence in birds and mammals – i.e.
tactile, vestibular, chemical, auditory and visual (Gottlieb, 1971; Lickliter, 1993). This
process remains to be investigated in invertebrates. The sense organs of adult cuttlefish
have been well described (Hanlon & Messenger, 1996), but we do not know at what
point they begin to function during embryonic development. The stages of embryonic
development in the cuttlefish have been precisely described by Lemaire (1970), from
 Cuttlefish preschool or how to learn in the peri-hatching period 5

Figure 1.1 Sepia officinalis embryonic stages according to Lemaire (1970) from left to right:
(A) stage 23, (B) stage 25 and (C) stage 30 (copyright L. Bonnaud and A.-S. Darmaillacq).
See plate section for colour version.

segmentation to organogenesis. Organogenesis ranges from stage 18 to stage 30 (i.e.

hatching). From stage 23 (about 4 weeks before hatching, incubation temperature 18°C,
embryo size about 1 mm; Figure 1.1, left), rhythmic mantle contractions are visible
through the outer covering of the egg (Romagny, Darmaillacq, Guibé, Bellanger &
Dickel, 2012); stage 25 (about 3 weeks before hatching, incubation temperature 18°C,
embryo size about 2 mm; Figure 1.1, middle) is easily identifiable by the reddish
colouration of the retina and stage 30 (a few days before hatching, incubation temperature
18°C, embryo size from 7 to 9 mm; Figure 1.1, right), corresponds to the hatching stage
where embryos resemble juveniles (Lemaire, 1970).
Variations in the frequency of mantle contractions in adult cuttlefish have been demon-
strated to be a good indicator of stimulus detection (odour: Boal & Golden, 1999; visual
stimuli: King & Adamo, 2006). Romagny, Darmaillacq, Guibé, Bellanger & Dickel
(2012) used this behavioural characteristic to investigate whether tactile, chemical and
visual sensory systems are functional in cuttlefish embryos. From stage 23, these authors
observed an increase of the rate of mantle contractions immediately after a gentle prick
on the ventral mantle of the embryo with a blunt needle through the membrane of the
egg. At the same developmental stage, a predator odour introduced to the immediate
environment of the egg induces the same behavioural response. Response to visual
stimuli (sudden light) did not occur earlier than stage 25. This study shows for the first
time that in cuttlefish sensory systems do not all become functional at the same stage of
embryonic development. As in vertebrates (Lickliter, 1993), visual system maturation
in the cuttlefish embryo occurs later than that of the tactile and chemical systems. This
similarity between cuttlefish and vertebrates could merely be a coincidence but it could
also suggest a unique evolutionary origin or, more likely, a similar convergent evolu-
tion. The latter hypothesis is reinforced by the other known evolutionary convergences
between the sensory systems of cephalopods and vertebrates (e.g. eyes, statocysts, lateral
line; Budelmann, 1994; Nixon & Young, 2003; Packard, 1972). Similar environmental
pressures could have led to the same developmentally conserved sequence in vertebrates
and in invertebrates. Further investigation on behavioural embryology in invertebrates
could provide important insights into whether this phenomenon is conserved throughout
different taxa.
6 Anne-Sophie Darmaillacq, Christelle Jozet-Alves, Cécile Bellanger and Ludovic Dickel

One expression of structural and functional developments in the sensory systems

of embryos is the ability to learn. Three types of learning have been documented,
mainly in vertebrates, as originating from prenatal sensory stimulation by chemosensory
and auditory stimuli: these are habituation, associative learning and exposure learning.
Associative and exposure learning can only be assessed after hatching even if the
process of acquisition occurs before hatching (see section 1.3.5) whereas habituation
can be directly tested with embryos. Habituation, understood to be a simple form of
learning, is defined as a decreasing response to an often-repeated stimulus. This has
been demonstrated in mammal fetuses (Granier-Deferre & Busnel, 1983; van Heteren,
Boekkooi, Jongsma & Nijhuis, 2001; Schaal, 1988; Smotherman & Robinson, 1992). In
S. officinalis embryos, habituation to a light stimulus presented every 30 minutes occurs
at stage 30 but not at stage 25 (Romagny, Darmaillacq, Guibé, Bellanger & Dickel,
2012). This learning is disrupted with a dishabituatory stimulus (a prick). Habituation
is a central process which allows the subject to filter irrelevant information and focus on
biologically meaningful stimuli. Before the egg capsule dilates and becomes transparent,
the embryo develops in a visually buffered dark environment, in which light is attenuated
by the capsule before reaching the developing embryo. This may explain why the visual
habituation processes are not mature at stage 25. In the experiments described above, the
outer dark layer of the eggs was removed. It, therefore, seems likely that habituation to
tactile or chemical stimuli can occur before stage 30. There is a rich and varied array of
stimuli in the prenatal environment and so it seems likely that the embryos would make
‘sense’ of these opportunities for learning about their future world. Firstly, a decrease
of the response to repeated non-relevant stimuli may serve the embryo to avoid loss of
energy. Secondly, its capabilities to memorize visual information from the environment
perceived in the egg may potentially optimize behavioural adaptation of hatchlings
to environmental pressure met at the hatching sites (available shelters, predators and
prey).

1.3 Early learning about prey

1.3.1 Prey preference and early experience

Cuttlefish are active predators, feeding on different types of prey including fish and
crustaceans (Guerra, 2006). They search for prey and find it mainly by visual means
(Messenger, 1968), although one cannot exclude the possibility that other sensory means
may be involved, and prey movement is of real significance. The question is: how do
newly hatched cuttlefish recognize safe prey in the absence of parental assistance and/or
observational learning? Wells (1962) presented newly hatched cuttlefish having no prior
post-embryonic experience of food with a series of models, either rotating or moving
up and down, to newly hatched cuttlefish without prior post-embryonic experience with
food. He showed that they preferred elongate shapes moving along their longer axis
(mysid-like prey) to rounded shapes (crab-like prey) that they almost never attacked.
 Cuttlefish preschool or how to learn in the peri-hatching period 7

Unfortunately Wells (1962) did not actually record the environmental conditions in
which the eggs developed, conditions that might have influenced prey choice. When
naı̈ve 1-day-old cuttlefish are given a choice between shrimp, crabs and fish, again
they prefer shrimp to the other two types of prey (Darmaillacq, Chichery, Poirier &
Dickel, 2004). This finding not only confirms that cuttlefish have an innate preference
for shrimp-like prey. It also suggests that prey recognition is based on identification
processes of a higher order than the simple selection of elongate prey moving along
their long axis, since hatchling cuttlefish prefer shrimp to fish which nevertheless share
the same basic characteristics. This preference for shrimp continues throughout the first
month of life but then seems to extend to other prey types even when cuttlefish are only
fed shrimp. It is known that early experience exerts a particularly potent influence on
subsequent food selection. For example, turtles retain a preference for the first food they
experience in their life (Burghardt, 1967; Burghardt & Hess, 1966). This phenomenon,
known as the primacy effect, has also been shown in other precocial species such as
the lynx spider (reviewed in Stasiak, 2002 and see also Punzo, 2002). In cuttlefish,
Darmaillacq, Chichery, Poirier & Dickel (2004) showed that juveniles that ate crab as
their first meal 3 days after hatching subsequently preferred crabs to shrimp. The ability
to learn the positive consequences of food ingestion may be adaptive to a young animal
needing to select safe prey alone. However, it may be a risky option for the young to
be genetically programmed to seek out a single type of prey, especially in the case of a
changing environment.

1.3.2 Associative learning

Cuttlefish actively prey upon shrimp, capturing them by shooting out their two tentacles
for a strike. This behaviour, visually driven, has been extensively used in the studies of
learning and memory capabilities in cuttlefish since the 1940s. Shrimp are presented
behind glass (or inside a glass tube; Figure 1.2); the cuttlefish then attacks the prey
but does not obtain it (Agin, Chichery & Chichery, 2001; Agin, Chichery, Maubert
& Chichery, 2003; Bellanger, Dauphin, Chichery & Chichery, 2003; Dickel, Boal &
Budelmann, 2000; Dickel, Chichery & Chichery, 1997, 1998, 2001; Messenger, 1971;
Sanders & Young, 1940; Wells, 1962).
Under these conditions, S. officinalis learns not to attack the shrimp and, hence, to
inhibit the predatory motor pattern: the number of capture attempts (tentacle strikes)
decreases with stimulus presentations. This learning has been recognized as a form
of associative learning (Agin, Dickel, Chichery & Chichery, 1998; Agin et al., 2006;
Messenger, 1973). Following a massed procedure (a single continuous 20-min training
phase), with various retention times (between 2 min and 2 days), Messenger (1971)
reported differential performance of memory recall in adult cuttlefish. He considered
this retention curve to be a product of two memory stores: a labile short-term memory
(STM) lasting for a period of some minutes, and a long-term memory (LTM) lasting at
least 2 days, and probably more. The retention curves obtained in 30- and 90-day-old
cuttlefish bear a close resemblance to those recorded in adults, suggesting that LTM is
8 Anne-Sophie Darmaillacq, Christelle Jozet-Alves, Cécile Bellanger and Ludovic Dickel

Figure 1.2 One-week-old cuttlefish striking at shrimp enclosed in a glass apparatus. The prey
movement is kept constant by a continuous water flow (copyright L. Dickel).

fully operational from one month of age (Agin et al., 2006). STM appears earlier, as early
as 8 days of age (Dickel, Chichery & Chichery, 1998), but 1-week-old cuttlefish have
poor 60-minute retention performance. This improves progressively between 15 and
60 days of age (Dickel, Chichery & Chichery, 1998). These results suggest that there is
a time lag between the establishment of short- and long-term memory systems during
post-embryonic development and that LTM matures later than STM. Moreover, there is
an improvement in the acquisition of learning ability during the first 2 months of life,
as well as an increase of 24-hour retention performance between 30 and 90 days of
age (Dickel, Boal & Budelmann, 2000; Dickel, Chichery & Chichery, 2001). As first
suggested by Wirz (1954), this poor LTM performance in 8-day-old cuttlefish has been
related to the immaturity of the brain’s vertical complex (see below, section 6). This is
particularly true of the vertical lobe, which, although the smallest lobe of the vertical
complex at hatching, nearly doubles in size during the first month of post-embryonic
development (Dickel, Chichery & Chichery, 1997, 2001).
However, this seems in contradiction with the findings of Darmaillacq, Chichery,
Poirier & Dickel (2004) where cuttlefish fed on crab at 3 days of age, subsequently
showed a preference for crab rather than their naturally preferred prey 4 days later. One
can imagine that it is much more adaptive for a growing individual to learn quickly
that available food is a positive alternative when their naturally preferred food is not
accessible. Another possibility, not mutually exclusive with the first, is that such learning
depends on different mechanisms and involves different brain structures than those
involved in the retention of an associative task. This point will be discussed later in this
chapter.
 Cuttlefish preschool or how to learn in the peri-hatching period 9

1.3.3 Food imprinting

Cuttlefish hatch with internal nutritive reserves. Even if active predation can begin
before the inner yolk is entirely used up (Boletzky, 1987; Dickel, Chichery & Chichery,
1997), these reserves allow young cuttlefish to go without food for a few days. This
period is probably critical in the life of young cuttlefish, since hatchlings can collect
information about their environment (potential food, predation risk, shelters) before
they start foraging. In cuttlefish from the English Channel, the prey-pursuit behaviour
does not begin before the end of the first week after hatching (Dickel, Chichery &
Chichery, 1997). From these data, it has been shown that the preference for crabs could
be induced simply after visual familiarization with them during a sensitive period within
the first hours of life after hatching (Darmaillacq, Chichery & Dickel, 2006; Darmaillacq,
Chichery, Shashar & Dickel, 2006).
Interestingly, the efficiency of this familiarization depends on the length of the expo-
sure as well as the density of prey exposed, i.e. the flow of information perceived during
the sensitive period. Finally, it appears that the primacy of the early familiarization out-
weighs the untrained preference for shrimp (Darmaillacq, Chichery, Shashar & Dickel,
2006). This learning of the visual characteristics of a potential prey meets all the criteria
of imprinting (Sluckin, 2007): no reinforcement, sensitive period, persistence (Darmail-
lacq, Chichery, Shashar & Dickel, 2006) and generalization (Guibé, Poirel, Houdé &
Dickel, 2012; see next section). Food imprinting could account for the prey preference
observed in 1-week-old cuttlefish, an effective compromise between the flexible but
precarious strategy of trial and error, and the rigid, genetically driven response of non-
learning. Such early learning capabilities would allow juveniles to take advantage of a
changing environment and to deal with a world where shrimp (i.e. their “innate” food
preference) was unavailable. To our knowledge, there is no available literature about
fluctuation in prey availability on the English Channel coasts. Unlike filial or sexual
imprinting, food imprinting might seem ineffective and even disadvantageous in the
long term (Healy, 2006), in particular for long-lived individuals in a changeable envi-
ronment. However, it could be helpful for juveniles to find available prey attractive in
the weeks following hatching before they begin to forage.
It also appears that, if we compare these results with those obtained from the prawn-
in-the-tube protocol, the two different types of learning may depend on the presence of
different neural substrates.

1.3.4 Prey generalization

The preference for different types of prey observed in cuttlefish suggests that they are
capable of interspecies discrimination (crab vs shrimp). Both crab and shrimp belong to
the Class Crustacea, but they differ greatly in characteristic movement and morphology.
A recent study has shown that naı̈ve hatchlings preferred black crabs to white ones
(Guibé, Poirel, Houdé & Dickel, 2012); this result is consistent with those obtained
in the adult octopus, which spontaneously prefers to attack dark over white artificial
objects (Fiorito & Scotto, 1992; Messenger & Sanders, 1972). However, if newly hatched
10 Anne-Sophie Darmaillacq, Christelle Jozet-Alves, Cécile Bellanger and Ludovic Dickel

cuttlefish have been exposed to white crabs, their spontaneous preference changes to
white crabs. This demonstrates that cuttlefish have the ability to discriminate between two
crabs from the same species but with different cephalothorax luminance. Moreover, if
they are familiarized with white crabs at hatching, they subsequently prefer black crabs to
shrimps. This means that they can generalize the learning of the characteristics of a prey
to which they have been familiarized to a novel prey that shares some morphological
features (Guibé, Poirel, Houdé & Dickel, 2012), and suggests a capacity for prey-
generalization and possibly categorization in hatchling cuttlefish. This cognitive process,
which consists in grouping distinguishable objects or events on the basis of a common
feature or set of features and consequently to respond in a similar way to new stimuli, is
a cognitive economy. Perceptual categorization has been extensively studied in pigeon,
monkeys and honeybees using operant conditioning (Wasserman & Zentall, 2006), but
still poorly investigated in marine molluscs.
It should be mentioned here that late juvenile cuttlefish (dorsal mantle length, DML,
from 100 to 140 mm) from the wild systematically catch crabs (cephalothorax length,
CL, from 40 to 60 mm) by the jumping strategy, although they capture shrimp by
striking with their tentacles (Chichery, 1992). Results show that fewer than 60% of
crab captured by inexperienced 5-day-old cuttlefish are caught by the jumping strategy
(Dickel, 1997). These observations show that, even though crab identification is possible
at hatching, there seems to be a correlation between the decision-making processes for
catching prey and the experience and/or later brain maturation of juvenile cuttlefish.
The frequency with which the jumping strategy is adopted in crab predation increases
progressively during post-embryonic development to reach more than 80% after a month
of life (Dickel, 1997). A possible hypothesis is that the later maturation of the vertical
lobe system is a factor in these processes (see below).

1.3.5 Embryonic learning of visual features of potential prey

In S. officinalis, the egg envelopes are stained black with the ink of the female. As the
embryo grows and the osmotic pressure of the perivitelline fluid increases (Boletzky,
1983), the elastic envelopes are dilated and some parts peel off so that the envelopes
become more transparent. By the end of embryonic development the tactile, chemical
and visual systems are all functional (see above, section 1.2), so that it is very likely that
the embryo approaching the end of its development can perceive stimuli coming from its
environment and, in particular, potential prey. Darmaillacq, Chichery, Shashar & Dickel
(2006) showed that the sensitive window for food imprinting was open at hatching;
given the characteristics of the egg capsules and the embryo’s cognitive competencies,
this sensitive period is likely to begin before hatching. Based on this assumption, Dar-
maillacq, Lesimple & Dickel (2008) exposed cuttlefish embryos to crabs for at least a
week before hatching (Figure 1.3). They were able to show that these cuttlefish subse-
quently preferred crab to shrimp, unlike a control group of cuttlefish that had not seen
crab during the sensitive period before hatching. Hence, embryos are able to learn the
general shape of a potential prey in the late stages of their development. This preference
was observed whether the outer layer of the egg envelope had been removed or not.
 Cuttlefish preschool or how to learn in the peri-hatching period 11

Figure 1.3 Embryonic exposure to crabs. Note the absence of pigmentation of the egg capsule
usually stained in black in Sepia officinalis (copyright A.-S. Darmaillacq). For details about the
apparatus see Darmaillacq, Lesimple and Dickel (2008).

Cephalopod ink is generally supposed to function in defence as a visual stimulus,

either as a smoke screen or as a distraction for predators (Hanlon & Messenger, 1996). It
has been hypothesized that it could function in the chemical realm as well (Derby, 2007)
and be repellent to predators (e.g. moray eel: Hanlon & Messenger, 1996; green turtle:
Caldwell, 2005). The ink contained in the egg envelope may also have the same function,
providing, in addition to the visual screen, a repulsive smell to ward off predators. Other
evidence that supports this idea is that female Sepia pharaonis, which lay transparent
eggs, hide them in crevices, corals or under any potential shelter, whereas S. officinalis lay
eggs in open fields without any additional protection. In Sepia esculenta, the eggs have
a sticky clear exterior that accumulates sand and other material for cryptic colouration
(Hanlon & Messenger, 1996). The results obtained by Darmaillacq, Lesimple & Dickel
(2008) suggest that the envelopes, although black in appearance, do not prevent embryos
from seeing outside during the last days of embryonic development, while continuing
to protect them from predation. Guibé and collaborators recently confirmed that visual
information can be used during the very last week, and specified that a 4-hour exposure
to crabs either in the morning or in the afternoon was enough (L. Dickel & M. Guibé,
unpublished data). From these results, one can argue that the sensitive period probably
starts shortly before hatching.
Shrimp differ from crabs in morphology, way of moving and luminance. But to what
extent does the embryo perceive the details of prey presented outside the egg? A partial
answer to this question has recently been found. When offered a choice between white
and black crabs, cuttlefish visually familiarized in ovo to white crabs preferred them,
12 Anne-Sophie Darmaillacq, Christelle Jozet-Alves, Cécile Bellanger and Ludovic Dickel

unlike the control cuttlefish that preferred black crabs; they also prefer black crabs to
shrimp, unlike the control cuttlefish that prefer shrimp (Guibé, Poirel, Houdé & Dickel,
2012). This confirms that juvenile cuttlefish have the ability to categorize and generalize
prey, but, interestingly, it also shows that embryos are capable of perception of one
particular feature of prey, its luminance.
If vision is the predominant sense in cuttlefish, chemical perception is also important
in the recognition of prey items, predators and conspecifics (Boal & Golden, 1999;
Boal & Marsh, 1998; Hanlon & Messenger, 1996). In embryos, the chemical system is
functional before the visual one (Romagny, Darmaillacq, Guibé, Bellanger & Dickel,
2012). To evaluate the effect of a chemical exposure on visual preference after hatching,
embryos were exposed to odours from shrimp (Crangon crangon; preferred prey), crabs
(Carcinus maenas; unpreferred prey), molluscs (Mytilus edulis; non-prey) or a seawater
control (Guibé, Boal & Dickel, 2010). They were then tested for a visual preference
between crabs and shrimp. Cuttlefish that had previously smelt shrimp had a visual
preference for crabs, whereas cuttlefish that had previously smelt crabs preferred shrimp
and cuttlefish that had previously smelt bivalves had no preference. To explain these
puzzling results, the authors pointed out a cross-modal effect between the chemical
and visual systems. Moreover, they hypothesized that an overstimulation of the chem-
ical system during the embryonic development could have disturbed the onset of the
visual system and, hence, the visual perception in juveniles (Honeycutt & Lickliter,
2003).

1.4 Lateralization

Cerebral lateralization is a universal and evolutionarily ancient trait in vertebrates (Val-

lortigara & Rogers, 2005). It is often revealed behaviourally by motor (e.g. handedness)
and perceptual (e.g. preferential eye use) asymmetries. It now appears that in inver-
tebrates, too, cerebral lateralization is probably widespread (Frasnelli, Vallortigara &
Rogers, 2012). This suggests that lateralization of the nervous system may be a common
feature of complex brains, even with completely different evolutionary histories. Very
recent research has also begun to suggest evidence of lateralization in the cuttlefish brain.
Several studies have investigated whether behavioural asymmetries are already present
at hatching, whether the development of lateralization is dependent on both genes
and environment, as in vertebrates, and the neural correlates of visual lateralization
(see below, section 1.6).

1.4.1 Ontogenesis of lateralization

Visual lateralization
Adult cuttlefish show an untrained side-turning preference (preference for turning right
or left) in a T-maze (Alves, Chichery, Boal & Dickel, 2007). Side-turning preference
may be the result of an eye use preference (i.e. visual lateralization) in cuttlefish, as has
been shown in octopus (Byrne, Kuba & Griebel, 2002; Byrne, Kuba & Meisel, 2004).
 Cuttlefish preschool or how to learn in the peri-hatching period 13

Exp. 1 Exp. 2

Sand and shadow

Le-turning Exp. 2
preference
Exp. 1
Right-turning
preference
Post-embryonic
development

Hatching 1 month

Figure 1.4 Postembryonic development of side-turning preference in Sepia officinalis. Exp. 1:

Cuttlefish are tested in an empty apparatus; Exp. 2: Cuttlefish are tested with a shelter (i.e. sand
and shadow) on both sides of the apparatus (Jozet-Alves et al., 2012).

The cuttlefish visual system consists of two eyes sitting laterally on the head (Nixon &
Young, 2003) and asymmetry of eye use seems to be ubiquitous among vertebrates with
laterally placed eyes (Vallortigara & Bisazza, 2002).
Jozet-Alves et al. (2012b) investigated the age at which the right- or left-turning
preference appeared by testing cuttlefish in a T-shaped apparatus between 3 and 60 days
post-hatching. To find out whether side-turning preferences were the result of an eye use
preference, cuttlefish were tested either inside a plain T-shaped apparatus or in one with
an attractive shelter on each arm of the T-maze (Figure 1.4). In the first case, juveniles
did not show any side-turning preference. However, when shelters were available in both
arms of the T-shaped apparatus, the authors observed a progressive development from 3
to 60 days post-hatching of a bias to turn leftwards, even though when going out of the
start box, cuttlefish simultaneously saw both shelters, one on each side. In the natural
environment, an eye-use preference may be very adaptive for cuttlefish to decide rapidly
which shelter to choose.
The emergence of visual lateralization in juvenile cuttlefish coincides with other onto-
genetic events such as the progressive diversification of their diet (Darmaillacq, Chich-
ery, Poirier & Dickel, 2004), increasing efficiency in matching the background (Han-
lon & Messenger, 1988) and dispersal (Dickel et al., 2006). One potential advantage
14 Anne-Sophie Darmaillacq, Christelle Jozet-Alves, Cécile Bellanger and Ludovic Dickel

of visual lateralization is the ability to process two types of information simultane-

ously. It is possible that the lateralization affords the cuttlefish greater behavioural
efficiency at the time of dispersal, by enabling them to look out for escape routes while
hunting.

Lateralization of food imprinting

In invertebrates, there is growing evidence for perceptual and motor lateralization, but
few examples of lateralization of learning are available. Researchers have demonstrated
asymmetrical performance favouring the right antenna when bees have to learn to
respond to particular odours to obtain a food reward (in honeybees: Letzkus et al., 2006;
in bumble bees: Anfora et al., 2011). More recently, Letzkus, Boeddeker, Wood, Zhang
& Srinivasan (2008) also found that honeybees learn a colour cue better with their right
eye to obtain a food reward.
C. Jozet-Alves and A.-S. Darmaillacq (unpublished data) investigated lateralization
of food imprinting (see above) in cuttlefish. They familiarized cuttlefish with crab with
the right eye, the left eye or both eyes. This study revealed that the efficiency of the
familiarization depends on the eye exposed to crab: only cuttlefish exposed to crab with
their right eye were significantly more likely than control cuttlefish to choose crab. This
study indicates: (i) that some types of behavioural lateralization are already established
in hatchlings; and (ii) that the right-eye pathway is sufficient for food imprinting. The
right eye may play a predominant role in learning the characteristics of a potential
prey. It is then possible that the right visual field is more generally specialized for
foraging and feeding behaviour, as in many species of vertebrates (such as chicks and
fish; Rogers, 2007) and in honeybees (Letzkus, Boeddeker, Wood, Zhang & Srinivasan,
2008).

1.4.2 Influence of early experience on visual lateralization

In vertebrates, some well-known examples of lateralization indicate that the ontogeny of
lateralization depends on the interaction of genes and experience. For example, in fish,
the degree of lateralization varies between populations of the same species that have
been exposed to different ecological pressures (e.g. predation pressure: Brown, Western
& Braithwaite, 2007). In invertebrates, no study has yet explored the influence of early
experience on the development of lateralization. The cuttlefish appears to offer an ideal
opportunity for addressing questions pertaining to the ontogenesis of lateralization in
invertebrates as their incubation and rearing conditions are easy to control.
Recently, Jozet-Alves and Hébert (2013) investigated a modulation of lateralization
by chemical signals of predation in cuttlefish. Cuttlefish embryos differed according
to the type of olfactory stimulation [predator odour (sea bass), non-predator odour
(sea urchins), no odour (plain sea water)] to which they were exposed before hatch-
ing. Whereas hatchlings that had not been exposed to predator odour during embry-
onic development (control and non-predator groups) showed no side-turning bias in a
T-shaped apparatus, as found previously by Jozet-Alves et al. (2012b; see above), hatch-
lings that had been exposed to predator odour (predator group) preferentially turned
 Cuttlefish preschool or how to learn in the peri-hatching period 15

to the left in the T-maze. The exposure to predator odour could possibly be an acute
predictor of a high predation risk in the environment. Differences in strength of later-
alization might provide fitness benefits to cuttlefish that hatch in high-predation areas,
enabling them to hunt and watch predators simultaneously. Indeed, it has been shown that
strongly lateralized individuals possess cognitive advantages over their less-lateralized
counterparts both in vertebrates (e.g. Magat & Brown, 2009, Rogers, Zucca & Val-
lortigara, 2004) and invertebrates (drosophila: Pascual, Huang, Nevue & Préat, 2004).
Cerebral lateralization may enhance brain efficiency in cognitive tasks that demand the
simultaneous but different use of both hemispheres.

1.4.3 Discussion on lateralization

In vertebrates, the right visual field is usually associated with the response to an identified
target such as a prey item, while the left visual field is specialized for the control of rapid
responses (e.g. response to a predator). This common pattern of visual lateralization
has led some researchers to argue in favour of a homology of cerebral lateralization, at
least in vertebrates (Rogers, 2007). The different studies on lateralization in cuttlefish
described above seem to indicate a dominance of the left visual field (i.e. left-turning
preference in a T-shaped apparatus to reach a shelter) in a defensive context, and a
dominance of the right visual field in learning the characteristics of a potential prey.
Moreover, as in vertebrates, visual lateralization in cuttlefish is markedly influenced by
early experience; genetic influences on lateralization are not paramount even in inverte-
brates. The similarities observed between vertebrates and cuttlefish could merely be a
coincidence or be indicative of a common origin of visual lateralization in vertebrates
and invertebrates. This last hypothesis implies that lateralization evolved once in the
common ancestor. Regardless of whether the various types of lateralization are homol-
ogous or not, allocating some functions to one half of the brain, and other functions
to the other half appears a common solution to enhance brain’s capacity. This solution
appears to be more widely adopted in the animal kingdom than an increase in brain
size. Evidence of lateralization now available from invertebrates, and more specifically
from cephalopods, will offer an instructive comparison for a better understanding of
lateralization in vertebrates.

1.5 Phenotypic plasticity and defensive behaviour

Predation is a constant threat faced by prey animals, especially young individuals, and
Brown and Chivers (2006) have suggested that there may be strong selection pressure
for the early detection and avoidance of potential predators. Another option is to avoid
being detected, a strategy that includes primary defensive behaviour. In the cuttlefish,
primary defence is generally called crypsis (see chapter 9, this volume, by Zylinski
& Osorio) and comprises camouflage and sand digging (Hanlon & Messenger, 1996).
Cuttlefish hatchlings spontaneously show the Disruptive pattern (see chapter 9, this
volume, by Zylinski & Osorio) whatever the substrate they settle on (Uniform grey or
16 Anne-Sophie Darmaillacq, Christelle Jozet-Alves, Cécile Bellanger and Ludovic Dickel

Figure 1.5 Sand-digging behaviour in 15-day-old cuttlefish. From left to right: cuttlefish not
buried, partially buried (50%) and completely buried (the arrows indicate the eyes) (copyright
C. Di Poi).

a chequerboard). Then, throughout the first 2 months of post-embryonic development,

the cuttlefish’s ability to blend into a Uniform background improves, resulting in closer
general background resemblance and a much reduced display of the Disruptive pattern;
this is especially true of cuttlefish reared in enriched conditions (i.e. tanks provided with
sand, shelters, rocks, conspecific, etc.: Poirier, Chichery & Dickel, 2005). This suggests
that sensory and cognitive enrichment affects the maturation of defensive strategies. The
same result was obtained for sand digging behaviour, where cuttlefish reared on sand had
shorter latencies to dig and were much better covered with sand than those reared without
sand (Poirier, Chichery & Dickel, 2004; Figure 1.5). Although this behaviour is at least
partially experience-dependent, what really stands out is that camouflage behaviours in
hatchlings are immature and hence inefficient.
We have already pointed out that embryos were able to perceive external information
concerning potential prey from inside the egg. However, a question remains: if embryos
can see potential prey, can they also see the risk of predation to which they will be
exposed after hatching? A very recent study showed cuttlefish that, as embryos, had
been exposed to sea bass, Dicentrarchus labrax, and crabs, Carcinus maenas (which
can be considered as potential predators), displayed a more effective camouflage on
a Uniform substrate or on sand (A.-S. Darmaillacq & L. Dickel, unpublished data).
This suggests that sensory-motor maturation could have begun during the last stages of
embryonic development. The type of stimulus to which cuttlefish respond (movement,
shape, size, etc.) is still not clear, and more experiments are needed. Nevertheless, in
cuttlefish, camouflage is used to prevent being detected by not only predators but also
by prey. As a consequence, it could be very advantageous for young cuttlefish to be
well camouflaged, so that they are effective in catching prey present in the vicinity of
the eggs. Such behavioural plasticity makes sense for a small and highly vulnerable
organism. When predation risk is high, it is likely that there is more movement around
the eggs and, hence, better camouflage at hatching will be highly adaptive, whereas this
is not necessary in an area of low predation risk. This suggests a cost of camouflage.
Body colouration may also serve functions other than defence (communication, hunting
strategy, etc.) that do not require the expression of an efficient camouflage.
 Cuttlefish preschool or how to learn in the peri-hatching period 17

1.6 The brain and its neurotransmitters

1.6.1 Brain patterning

The cephalopod brain originates from four pairs of molluscan ganglia, known as the
pedal, palliovisceral, cerebral and optic ganglia. Baratte and Bonnaud (2009) described
ganglionic proneural cells forming dendrites at stage 16 of embryogenesis, which demon-
strate early neural differentiation. These ganglia converge at stage 21 to become a cen-
tral circumoesophageal brain, lying between the eyes. Thus, the central nervous system
comprises supra- and sub-oesophageal masses connected by peri-oesophageal lobes.
On each side of the peri-oesophageal lobes are the large optic lobes, connected to the
supra-oesophageal mass by short optic tracts. The cephalopod brain has been described
in great detail (for a review, see Nixon & Young, 2003). A review of brain development
requires a preliminary summary of its anatomy and the main functions of the lobes in
adult cuttlefish. The central mass lobes all have the same pattern: a central neuropil
(area of nerve fibres) and a surrounding cortex (cell bodies). The optic lobes comprise
an outer cortex and a central medulla. They are involved in higher-order visual pro-
cessing, in the control of motor programmes (Chichery & Chanelet, 1976) as well as in
memory (in connection with the vertical system, see below). In the hilum of the optic
lobes, the peduncle lobes are small structures lying horizontally and dorsally to the optic
tracts. The peduncle lobes are higher motor centres that may play a role in the fine
control of locomotion, body colour and pupil dilatation (Chichery & Chanelet, 1978).
The vertical system corresponds to the dorsal part of the supra-oesophageal mass and
contains various structures such as the vertical, sub-vertical and superior frontal lobes,
involved in learning and memory (Sanders & Young, 1940), and the inferior frontal lobe
involved in sensory-motor integration (Young, 1979). The ventral part (basal lobes) of
the supra-oesophageal mass comprises some higher centres implicated in the control of
motor programmes. The sub-oesophageal mass (brachial, pedal and palliovisceral lobes)
and the peri-oesophageal mass (magnocellular lobes) contain the premotor and motor
neurones responsible for swimming, escape behaviour, ink ejection and arm and tentacle
movements (Nixon & Young, 2003).

1.6.2 Brain maturation

The central nervous system in cephalopods grows throughout life, but some lobes
develop in size or in function earlier than others (Nixon & Mangold, 1998; Figure 1.6).
This brain maturation has been investigated in juvenile cuttlefish from hatching until the
age of 3 months through three sets of studies: (i) measurement of the relative growth of
the different lobes (volume of a lobe/total brain volume) (Dickel, Chichery & Chichery,
1997); (ii) quantification of cell proliferation in the brain of juvenile cuttlefish (5-bromo-
2ʹ-deoxyuridine incorporation) (Poirier, 2004); (iii) Determination of the presence of
heavy neurofilaments (pNF-H, 220 kD) in the optic lobes of late embryos (Darmaillacq,
2005) and in the brain of early juveniles (Dickel, 1997). These highly phosphorylated
neurofilaments in the cytoskeleton are considered as strong markers of a stabilized neural
18 Anne-Sophie Darmaillacq, Christelle Jozet-Alves, Cécile Bellanger and Ludovic Dickel

SF

sV

IF

PB
AB

sV
SF

IF

PB AB

Figure 1.6 Supra-oesophageal mass of the brain: cuttlefish hatchling (top) and 3-month-old
cuttlefish (bottom) (sagittal sections). Note the dramatic increase of the vertical lobe (V) size.
Abbreviations: AB, anterior basal lobe; IF, inferior frontal lobe; PB, posterior basal lobe; SF,
superior frontal lobe; sV, subvertical lobe; V, vertical lobe. Upper scale bar = 200 µm; lower
scale bar = 400 µm (copyright L. Dickel).
 Cuttlefish preschool or how to learn in the peri-hatching period 19

network in both the vertebrate and the invertebrate nervous system (Darmaillacq, 2005;
Dickel, 1997; Grant, Tseng, Gould, Gainer & Pant, 1995; Riederer, 1992).

Optic lobes
Each optic lobe receives input from the retina and their cortex corresponds to the deeper
layers of the vertebrate retina (second and third-order neurones; Nixon & Young, 2003).
The behavioural response of embryos to a visual stimulus (see section 1.2) and their
ability to learn visual features of potential prey (see section 1.3.5) indicate that the visual
system, including the optic lobes, is functional before hatching. The presence of pNF-
H in the plexiform layer of the cortex at embryonic stage 29 suggests that first-order
visual inputs (plexiform layer) are already mature (Darmaillacq, 2005). At hatching, the
optic lobes are similar in relative size and tissue organization to those of adults. Their
connexions with the arms are already mature at hatching. This early maturation could
be linked to the early visuo-motor abilities of hatchlings. As described in section 1.3.4,
newly hatched cuttlefish are able to perform prey-generalization and categorization
(Guibé, Poirel, Houdé & Dickel, 2012). Investigations of neural circuits involved in prey
identification are still very preliminary, but the optic lobes seem likely to be the neural
substrates of these early processes of visual integration. Apart from their involvement in
the integration of visual information, the optic lobes have been supposed to be the site
of the visual memory store in octopus (Sanders 1975; Young, 1965). No study has been
undertaken so far in cuttlefish, but here, again, they seem likely candidates for neural
substrates of food imprinting based on visual familiarization (see section 1.3.3). These
structures are more developed in newly hatched cuttlefish than the vertical lobe system
(see above). In addition to their early neural and functional maturation, the optic lobes
also undergo further post-embryonic maturation. Even though the rate of cell divisions
in these lobes decreases from hatching to age 2 months, this decrease is less marked
than in other brain structures (Poirier, 2004). The increase in size of the optic lobes is
mainly due to the development of neurites; indeed, plexiform layers of the deep retina
thicken during the first days of post-hatching development. pNF-H immunostaining in
the cortex and neuropil is of low intensity at birth and the neurites show a gradual
post-embryonic stabilization during the first 6 days (Dickel, 1997). This post-embryonic
maturation means that the development of the optic lobe can still be affected by the
environment of the juvenile. Accumulating evidence in vertebrates shows that sensory
experience modulates brain development during a critical period, especially before the
maturation of the brain wiring (Bengoetxea et al., 2012).

Motor centres
Several lobes of the supra-, sub- and peri-oesophageal mass control the various actions
of the cuttlefish. Those lobes are gathered into lower, intermediate and higher motor
centres. The lower and intermediate motor centres (sub-oesophageal, chromatophore
and magnocellular lobes) contain the motor neurones, whereas the higher motor centres
(peduncle lobes and basal lobes of the supra-oesophageal mass) are responsible for
the combined movements of different effectors. At hatching, juveniles display prey
capture and cryptic behaviours. Brain studies have shown that the lobes involved in such
20 Anne-Sophie Darmaillacq, Christelle Jozet-Alves, Cécile Bellanger and Ludovic Dickel

behaviours (e.g. basal, chromatophore and sub-oesophageal lobes) are well developed, as
are their connections. Moreover, the sub-oesophageal lobes show some mature neurones
at embryonic stage 29. The higher motor centres are mature soon after hatching since
a high staining of pNF-H has been observed in their neuropil (basal part of the supra-
oesophageal mass and peduncle lobes). Cell proliferation decreases gradually (peduncle,
chromatophore and basal lobes) until the age of 2 months. This can be linked to the
improvement of predation and camouflage strategies observed during the first 2 months
(see Section 1.5).

The vertical lobe system (memory system)

The vertical system contains associative structures receiving signals from different
sensory systems and is known to be involved in learning and memory (Agin et al.,
2006; Graindorge et al., 2006). If the optic lobes have been suggested as a location
for information storage, Young (1965) hypothesized that this stored information is not
available for use in the absence of the vertical lobe system. This seems to be particularly
true in the case of associative learning. Short- and long-term synaptic plasticity has
recently been shown in the vertical lobe of adult cuttlefish (Shomrat et al., 2011).
Apart from the inferior frontal lobe, which shows some mature neurons from embryonic
stage 29, the vertical system is quite immature at hatching. The associative nervous
structures (vertical and superior frontal lobes) are the most poorly developed regions of
the brain in hatchlings. Their volume increases during the first months of life, and the
post-embryonic growth of the vertical lobe is dramatic when compared to that of other
supra-oesophageal lobes (7% of the total brain volume in hatchlings against 12% in
12-month-old animals; Figure 1.6). As first suggested by Wirz (1954), poor long-term
memory performance of an associative learning task in 8-day-old cuttlefish has been
related to the immaturity of the vertical complex. Moreover, the growth of the superior
frontal and vertical lobes appears to be significantly linked with the improvement of long-
term retention performances (Dickel, Chichery & Chichery, 2001, see section 1.3.2).
This increase in volume may be due to neuritic growth and cell proliferation, especially
during the first months of life. In vertebrates, an increase in neurogenesis underlies brain
growth in early postnatal development (during the first week of life in rats); afterwards
neurogenesis decreases to reach a relatively stable state in adulthood (Bandeira, Lent
& Herculano-Houzel, 2009; Jabès, Lavenex, Amaral & Lavenex, 2010). Even if the
neuropil of the frontal superior lobe and its connexions to the vertical lobe show some
pNF-H stained fibres at hatching, the fibres in this lobe keep on stabilizing, at least in
the week after hatching (intensification of the pNF-H staining). In the sub-vertical lobe,
cell proliferation increases during the 2 weeks after hatching, the rate then decreasing
until 2 months of age, as in the other lobes of the vertical system. The tracts connecting
the vertical lobe to the sub-vertical lobe – sites of long-term potentiation (Shomrat
et al., 2011) – are lacking at hatching (Dickel, Chichery & Chichery, 1997). These
nervous tracts grow between day 1 and day 9 and appear really stable at the age of 1 month
(Dickel, Chichery & Chichery, 1997). From a functional standpoint, their appearance
is concomitant with the maturation of short-term retention of some visual information
during the first days of life (see section 1.3.2). In addition to this involvement in memory
 Cuttlefish preschool or how to learn in the peri-hatching period 21

processes, one can also hypothesize that the late maturation of the vertical lobe system
could be related to decision-making processes for catching prey (see section 3.4).
Brain maturation is strongly influenced by environment. The maturation of the
vertical–subvertical lobe tracts depends on the water temperature (Dickel, Chichery
& Chichery, 1997), and when juveniles are reared in enriched conditions (in large tanks,
with conspecifics, sand, shelters, seaweed), the vertical system develops faster, espe-
cially during the first month of life (Dickel, Chichery & Chichery, 2001). Poirier (2004)
showed that enrichment did not modify the intensity of cell proliferation in the vertical
system, so it might optimize neurite growth. The development of the optic lobes is
also environment-sensitive. When cuttlefish are reared in the dark, there is a significant
decrease in cell proliferation in the optic lobe neuropils at 15 days compared to cuttlefish
reared in a natural light–dark cycle, whatever the other rearing conditions (Poirier, 2004).
On the contrary, at 30 days of age enriched rearing conditions induce an increase in cell
proliferation in the deep retina and in the neuropil of optic lobes. These results highlight
the importance of sensory stimulation on brain maturation.

1.6.3 Brain correlates of visual lateralization

Little is known about how the ontogenesis of behavioural asymmetry is related to the
development of cerebral asymmetry. Jozet-Alves, Romagny, Bellanger & Dickel (2012a)
have investigated whether anatomical or neurochemical asymmetries are present in the
cuttlefish brain at 30 days post-hatching, and whether these asymmetries correlate with
side-turning preferences. The authors found individual variation in the size of the optic
lobe and the degree of vertical lobe asymmetry. Some cuttlefish possessed relatively
symmetrical optic and vertical lobes, and others possessed a larger right optic lobe and
right half of the vertical lobe. A striking correlation emerged from the behavioural results:
the larger the right optic lobe and the right part of the vertical lobe, the stronger the bias to
turn leftwards. The absence of correlation between other brain asymmetry measurements
and behavioural lateralization indicates that these correlations are exclusive to these
structures and not part of a more general brain asymmetry. It remains to be determined
how the optic/vertical lobes interact with other neural systems to produce lateralization
of behaviour.

1.6.4 Neurotransmitters in the developing brain

Neurotransmitters have diverse roles in the developing nervous system from embryonic
developmental functions in morphogenesis to synapse formation (Buznikov, Shmukler &
Lauder, 1996). The few neurotransmitters studied in the developing brain of S. officinalis
will be reviewed here.

Acetylcholine
The central cholinergic system is involved in cognitive processes, as extensively reported
in vertebrates (Everitt & Robbins, 1997) and, to a lesser extent, in invertebrates (e.g.
insects: Fresquet, Fournier & Gauthier, 1998; molluscs: Mpitsos, Murray, Creech &
22 Anne-Sophie Darmaillacq, Christelle Jozet-Alves, Cécile Bellanger and Ludovic Dickel

Barker, 1988). Acetylcholine has been studied in the cuttlefish central nervous system
(Bellanger, Dauphin, Belzunce & Chichery, 1998; Messenger, 1996) including during
memory formation (Bellanger, Dauphin, Chichery & Chichery, 2003). Unlike the mus-
carinic type, nicotinic-like receptors are widely present in the brain of adult cuttlefish.
We have not yet investigated the development of these nicotinic-like receptors during
post-embryonic life but preliminary results show that, except in the optic and superior
frontal lobes, density and distribution of the nicotinic-like receptors in the different brain
structures are relatively unchanged from 3 months to 15 months of age (Bellanger, Halm,
Dauphin & Chichery, 2005).

Serotonin
In addition to its physiological role as a neuromodulator (e.g. in mood regulation, mem-
ory, food motivation, etc.), serotonin also modulates cellular migration and cytoarchitec-
ture throughout development (Daubert & Condron, 2010). The distribution of serotonin
immunoreactive cells in the brain of newly hatched cuttlefish is quite similar to that
described in the adult’s brain (unpublished data; Boyer, Maubert, Charnay & Chichery,
2007). Monoamines were quantified in the whole head of juveniles (by high perfor-
mance liquid chromatography). Circulating monoamines may have been included in our
measurements but are likely to be in negligible amounts (little remaining haemolymph).
The median serotonin content was 0.43 ± 0.07 ng/mg tissue at hatching and seems to
have remained quite stable during the first week of life, but it was subject to considerable
inter-individual variations (0.35 ± 0.34 ng/mg tissue at day 7; unpublished data). The
spatio-temporal distribution of serotonin was investigated in the visual system during
embryogenesis by immunohistochemistry (Romagny, 2010). Serotonin is not present in
the embryos before stage 26. Few immunoreactive fibres were observed in the retina
and optic lobes (cortex and neuropil) from stages 27. Then the number of serotonin
immunoreactive fibres increased gradually until hatching. A strong light stimulus sig-
nificantly increased serotonin staining in the visual system in newly hatched cuttlefish.
This argues for the implication of serotonin in the functioning of this sensory system
in embryos. Nevertheless, we cannot yet determine whether the increase in serotonin
content results from activation of the visual pathways or to a general reaction of the
young to a stressor.

Catecholamines
In the brain, catecholamines are known to be involved in major functions in vertebrates
including motor control, cognition, emotion, positive reinforcement and visceral regula-
tion (Iversen & Iversen, 2007). They contribute to the development of neuronal pathways
and synapse formation (Dobson, Dmetrichuk & Spencer, 2006). Catecholamines occur
in the embryonic ganglia of cuttlefish at stage 16 in some proneural cells exhibiting
neurites (Baratte & Bonnaud, 2009). Baratte and Bonnaud (2009) suggested an early
differentiation of catecholamine-containing neurones in the future brain and in the pri-
mary retina. Catecholamines may also play a modulatory role, such as axonal guidance,
during gangliogenesis. Dopamine and particularly noradrenaline are more concentrated
 Cuttlefish preschool or how to learn in the peri-hatching period 23

in the head of newly hatched cuttlefish than serotonin (0.78 ± 0.42 ng dopamine/mg
tissue and 19.51 ± 496.27 ng noradrenaline/mg tissue).

Oxytocin
Brain oxytocin plays an important role in behavioural regulation in mammals and may
influence the development of the central nervous system during the postnatal period
(Carter, 2003). The spatio-temporal distribution of a peptide has been investigated in
the central nervous system during the postnatal development of S. officinalis (Bardou
et al., 2010). The central oxytocin-like system is well developed at birth and continues
to mature throughout the postnatal weeks until it reaches clear maturity in 2-month-
old animals. After hatching very few changes occur in the relative abundance of fibres
containing oxytocin-like elements; the maturation of this neuropeptidic system is mainly
seen in the spatial distribution of cell bodies. Most of the central lobes show a mature
oxytocin-like pattern in 15-day-old animals. Surprisingly, the central vertical lobe and
the medulla of the optic lobes have the earliest maturation. Indeed, no change in the
relative abundance of oxytocin-like cell bodies was observed between hatching and age
3 months. The sub-oesophageal lobes continue maturation until the age of 2 months.
Our knowledge of the processes of brain development in cuttlefish is incomplete.
Nevertheless, we can see that the structures involved in motor programme and sensory
integration mature earlier than the associative lobes. This heterogeneity can certainly be
linked to the behavioural abilities of hatchlings and the post-embryonic optimization of
learning capacities.

1.7 Conclusion

The starting point of this chapter was to find out whether, as in vertebrates, prenatal
experience could shape postnatal behaviour in a sophisticated invertebrate and if so,
whether cuttlefish embryos could be considered to have complex cognitive systems.
In fact, the studies reviewed here show that cuttlefish, which are known to have good
learning and memory capabilities as adults, seem already to be very good learners as
embryos. In the absence of parental care, young cuttlefish have to find edible prey and
protect themselves from predators. The studies showed that embryos can collect visual
information concerning prey and predators and use it after hatching. This validates the
existence of continuities in learning from prenatal to postnatal life even in invertebrates.
For this, embryos need efficient sense organs, the onset of whose functioning follows
the same sequence as in vertebrates, and which are functional even before they have
completed their maturation. They also need a sophisticated nervous system that will
integrate and process all the information collected by the sensory system, which is
indeed the case. Cuttlefish are at the same time potential prey and predators, so that they
need to have both efficient anti-predator and predatory behaviour; this requires cognitive
abilities such as prey/predator categorization and generalization. In the cuttlefish which,
like the other coleoids, ‘live fast and die young’, such cognitive abilities seem to be
established very early, even at the embryonic stage. Female cuttlefish lay relatively large
24 Anne-Sophie Darmaillacq, Christelle Jozet-Alves, Cécile Bellanger and Ludovic Dickel

eggs with abundant yolk reserves. As a consequence embryos have resources for neural
development in the egg compared with other molluscs, which seem to have a more
random approach (more offspring and some will survive by luck – see chapter 2, this
volume, by Basil & Crook). Embryos could then benefit from a large brain and adapt to
a wide range of environments.
At this point it might be useful to describe the early stages of the life of a cuttlefish,
as it is now understood. The cuttlefish embryo develops inside an opaque egg, sheltered
from predators. As the embryo grows, the envelope of the egg stretches and becomes
thinner and more transparent. This allows the embryo to become familiar with its future
biotic and abiotic environment through its chemical and visual senses as each perceptual
system becomes functional. At hatching, the cuttlefish is able to dig into the sand and
to camouflage itself, although it will probably prefer to find a shelter (Guibé & Dickel,
2011), until its more classical defensive behaviour becomes more efficient. Meanwhile,
since from inside the egg it has had a ‘preview’ of prey type, it is prepared to feed in
the immediate environment on what is available immediately after hatching, switching
to its innately preferred (elongate-shaped) prey if this becomes available.
Cuttlefish are usually described as animals that receive no parental care. However, it
is still not known whether a kind of indirect parental care exists, in that the mother could
in laying her eggs choose a place where suitable prey will be available, or she may, as
birds are known to do, deposit chemical information (steroid hormones, prey odours,
etc.) in the yolk (Henriksen, Rettenbacher & Groothuis, 2011).
In conclusion, cuttlefish are both the most important cephalopod fishery resource
in Europe, and also an important component in the marine ecosystem. At the present
time, cuttlefish in Europe face a major problem. Each year, billions of eggs are laid on
fishermen’s traps, which are an ideal available surface for females, but they are destroyed
during harvest when the traps are cleaned. To counterbalance this, future programmes
for the reintroduction of eggs/hatchlings will soon be necessary. In depth ethological
knowledge, as described in this chapter, will be necessary to ensure that conditions on
the spawning ground do not interfere with embryos and hatchling viability. Survival
of young animals in the wild depends largely on their behavioural plasticity and their
embryonic experience. For this reason, as a major contribution to the programme’s
success, attention will have to be given to the quality of the released individuals, in
particular ensuring the development of strategies to include the encouragement of the
best possible performance and robustness from the eggs or hatchlings concerned.

Acknowledgements

We are very grateful to Frédéric Guyon who takes care of the cuttlefish at the Marine
Station (CREC, Luc sur Mer, France) and to Jean-Paul Lehodey who helps us design
and build the experimental apparatus. We also thank all the voluntary students (from
undergraduate to graduate levels) who participated in some of the experiments. Lastly,
we are particularly grateful to Prs Raymond and Marie-Paule Chichery who worked for
decades to develop cephalopod research in Normandy.
 Cuttlefish preschool or how to learn in the peri-hatching period 25

References

Agin, V., Chichery, R. and Chichery, M.-P. (2001). Effects of learning on cytochrome oxidase
activity in cuttlefish brain. Neuroreport, 12: 1–4.
Agin, V., Chichery, R., Chichery, M-P., Dickel, L., Darmaillacq, A.-S. and Bellanger, C.
(2006). Behavioural plasticity and neural correlates in adult cuttlefish. Vie et Milieu, 56: 81–
87.
Agin, V., Chichery, R., Maubert, E. and Chichery, M.-P. (2003). Time-dependent effects of cyclo-
heximide on long-term memory in the cuttlefish. Pharmacology Biochemistry and Behavior,
75: 141–146.
Agin, V., Dickel, L., Chichery, R. and Chichery, M.-P. (1998). Evidence for a specific short-term
memory in the cuttlefish, Sepia. Behavioural Processes, 43: 329–334.
Alves, C., Chichery, R., Boal, J. G. and Dickel L. (2007). Orientation in the cuttlefish Sepia
officinalis: response versus place learning. Animal Cognition, 10: 29–36.
Anfora, G., Rigosi, E., Frasnelli, E., Ruga, V., Trona, F. and Vallortigara, G. (2011). Lateralization
in the invertebrate brain: left-right asymmetry of olfaction in bumble bee, Bombus terrestris.
Public Library of Science One, 6: e18903.
Bandeira, F., Lent, R. and Herculano-Houzel, S. (2009). Changing numbers of neuronal and non-
neuronal cells underlie postnatal brain growth in the rat. Proceedings of the National Academy
of Sciences of the United States of America, 106: 14108–14113.
Baratte, S. and Bonnaud, L. (2009). Evidence of early nervous differentiation and early cate-
cholaminergic sensory system during Sepia officinalis embryogenesis. Journal of Comparative
Neurology, 517(4): 539–549.
Bardou, I., Maubert, E., Leprince, J., Chichery, R., Dallérac, G., Vaudry, H. and Agin, V. (2010).
Ontogeny of oxytocin-like immunoreactivity in the cuttlefish, Sepia officinalis, central nervous
system. Developmental Neuroscience, 32(1): 19–32.
Bellanger, C., Dauphin, F., Belzunce, L. and Chichery, R. (1998). Parallel regional quantification
of choline acetyltransferase and cholinesterase activity in the central nervous system of an
invertebrate (Sepia officinalis). Brain Research. Brain Research Protocols, 3: 68–75.
Bellanger, C., Dauphin, F., Chichery, M.-P. and Chichery, R. (2003). Changes in cholinergic
enzyme activities in the cuttlefish brain during memory formation. Physiology and Behavior,
79: 749–756.
Bellanger, C., Halm, M.-P., Dauphin, F. and Chichery, R. (2005). In vitro evidence and age-related
changes for nicotinic but not muscarinic acetylcholine receptors in the central nervous system
of Sepia officinalis. Neuroscience Letters, 387(3): 162–167.
Bengoetxea, H., Ortuzar, N., Bulnes, S., Rico-Barrio, I., Lafuente, J. V. and Argandoña, E. G.
(2012). Enriched and deprived sensory experience induces structural changes and rewires
connectivity during the postnatal development of the brain. Neural Plasticity, 2012: 1–10.
Billings, V. C., Sullivan, M. and Vine, H. (2000). Sighting of Thysanoteuthis rhombus egg mass
in Indonesian waters and observations of embryonic development. Journal of the Marine
Biological Association of the United Kingdom, 80: 1139–1140.
Boal, J. G. and Golden, D. K. (1999). Distance chemoreception in the common cuttlefish, Sepia
officinalis (Mollusca cephalopoda). Journal of Experimental Marine Biology and Ecology, 235:
307–317.
Boal, J. G. and Marsh, S. E. (1998). Social recognition using chemical cues in cuttlefish (Sepia
officinalis Linnaeus, 1758). Journal of Experimental Marine Biology and Ecology, 230: 183–
192.
26 Anne-Sophie Darmaillacq, Christelle Jozet-Alves, Cécile Bellanger and Ludovic Dickel

Boletzky, S. v. (1983). Sepia officinalis. In Cephalopod Life Cycles, Species Accounts, Boyle
P. R. (ed.), vol. 1. New York: Academic Press.
Boletzky, S. v. (1987). Juvenile behaviour. In Cephalopod Life Cycles, Comparative Reviews,
Boyle P. R. (ed.), vol. 2. New York: Academic Press.
Boletzky, S. v. (2003). Biology of early stages in cephalopod molluscs. Advances in Marine
Biology, 44: 143–203.
Boyer, C., Maubert, E., Charnay, Y. and Chichery, R. (2007). Distribution of neurokinin A-like
and serotonin immunoreactivities within the vertical lobe complex in Sepia officinalis. Brain
Research, 1133(1): 53–66.
Brown, C., Western, J. A. C. and Braithwaite, V. A. (2007). The influence of early experience on,
and inheritance of, cerebral lateralization. Animal Behaviour, 74: 231–238.
Brown, G. E. and Chivers, D. P. (2006). Learning about danger: chemical alarm cues and the
assessment of predation risk by fishes. In Fish Cognition and Behavior, Brown, C., Laland,
K. and Krause, J. (eds.). Oxford (UK): Wiley-Blackwell.
Budelmann, B. U. (1994). Cephalopod sense organs, nerves and the brain: adaptations for
high performance and life style. Marine and Freshwater Behaviour and Physiology, 25: 13–
33.
Burghardt, G. M. (1967). The primacy effect of the first feeding experience in the snapping turtle.
Psychonomic Science, 7: 383–384.
Burghardt, G. M. and Hess, E. H. (1966). Food imprinting in the snapping turtle, Chelydra
serpentina. Science, 151: 108–109.
Buznikov, G. A., Shmukler, Y. B. and Lauder, J. M. (1996). From oocyte to neuron: do neu-
rotransmitters function in the same way throughout development? Cellular and Molecular
Neurobiology, 16: 533–559.
Byrne, R. A., Kuba, M. and Griebel, U. (2002). Lateral asymmetry of eye use in Octopus vulgaris.
Animal Behaviour, 64: 461–468.
Byrne, R. A., Kuba, M. J. and Meisel, D. V. (2004). Lateralized eye use in Octopus vulgaris shows
antisymmetrical distribution. Animal Behaviour, 68: 1107–1114.
Caldwell, R. L. (2005). An observation of inking behavior protecting adult Octopus bocki from
predation by green turtle (Chelonia mydas) hatchlings. Pacific Science, 59: 69–72.
Carter, C. S. (2003). Developmental consequences of oxytocin. Physiology and Behavior, 79(3):
383–397.
Chichery, M.-P. (1992). Approche neuroéthologique du comportement prédateur de la seiche
Sepia officinalis. Doctorat d’Etat de l’Université de Caen Basse-Normandie.
Chichery, R. and Chanelet, J. (1976). Motor and behavioural responses obtained by stimula-
tion with chronic electrodes of the optic lobe of Sepia officinalis. Brain Research, 105: 525–
532.
Chichery, R. and Chanelet, J. (1978). Motor responses obtained by stimulation of the peduncle
lobe of Sepia officinalis in chronic experiments. Brain Research, 150: 188–193.
Darmaillacq, A.-S. (2005). Plasticité des préférences alimentaires chez la seiche Sepia officinalis:
approches ontogénétique et neuro-ethologique. Unpublished thesis (PhD), Université de Paris-
Nord.
Darmaillacq, A.-S., Chichery, R. and Dickel, L. (2006). Food imprinting, new evidence from the
cuttlefish Sepia officinalis. Biology Letters, 2: 345–347.
Darmaillacq, A.-S., Chichery, R., Poirier, R. and Dickel, L. (2004). Effect of early feeding expe-
rience on subsequent prey preference by cuttlefish, Sepia officinalis. Developmental Psychobi-
ology, 45: 239–244.
 Cuttlefish preschool or how to learn in the peri-hatching period 27

Darmaillacq, A.-S., Chichery, R., Shashar, N. and Dickel, L. (2006). Early familiarization over-
rides innate prey preference in newly-hatched Sepia officinalis cuttlefish. Animal Behaviour,
71(3): 511–514.
Darmaillacq, A.-S., Lesimple, C. and Dickel, L. (2008). Embryonic visual learning in the cuttlefish,
Sepia officinalis. Animal Behaviour, 76: 131–134.
Daubert, E. A. and Condron, B. G. (2010). Serotonin: a regulator of neuronal morphology and
circuitry. Trends in Neuroscience, 33(9): 424–434.
Derby, C. D. (2007). Escape by inking and secreting: marine molluscs avoid predators through a
rich array of chemicals and mechanisms. The Biological Bulletin, 213: 274–289.
Dickel, L. (1997). Comportement prédateur et mémoire chez la seiche (Sepia officinalis),
approches développementale et neuro-éthologique. Unpublished thesis (PhD), Université de
Caen Basse-Normandie.
Dickel, L, Boal, J. G. and Budelmann, B. U. (2000). The effect of early experience on learning
and memory in cuttlefish. Developmental Psychobiology, 36(2): 101–110.
Dickel, L., Chichery, M.-P. and Chichery, R. (1997). Postembryonic maturation of the vertical
lobe complex and early development of predatory behavior in the cuttlefish (Sepia officinalis).
Neurobiology of Learning and Memory, 67(2): 150–160.
Dickel, L., Chichery, M.-P. and Chichery, R. (1998). Time differences in the emergence of short-
and long-term memory during post-embryonic development in the cuttlefish, Sepia. Behavioural
Processes, 44: 81–86.
Dickel, L., Chichery, M.-P. and Chichery, R. (2001). Increase of learning abilities and matura-
tion of the vertical lobe complex during postembryonic development in the cuttlefish, Sepia.
Developmental Psychobiology, 39(2): 92–98.
Dickel, L., Darmaillacq, A.-S., Poirier, R., Agin, V., Bellanger, C. and Chichery, R. (2006).
Behavioural and neural maturation in the cuttlefish Sepia officinalis. Vie et Milieu, 56: 89–
95.
Dobson, K. S., Dmetrichuk, J. M. and Spencer, G. E. (2006). Different receptors mediate the
electrophysiological and growth cone responses of an identified neuron to applied dopamine.
Neuroscience, 141(4): 1801–1810.
Everitt, B. J. and Robbins, T. W. (1997). Central cholinergic systems and cognition. Annual Review
of Psychology, 48: 649–684.
Fiorito, G. and Scotto, P. (1992). Observational learning in Octopus vulgaris. Science, 256: 545–
247.
Frasnelli, E., Vallortigara, G. and Rogers, L. J. (2012). Left–right asymmetries of behaviour
and nervous system in invertebrates. Neuroscience and Biobehavioral Reviews, 36: 1273–
1291.
Fresquet, N., Fournier, D. and Gauthier, M. (1998). A new attempt to assess the effect of learn-
ing processes on the cholinergic system: studies on fruitflies and honeybees. Comparative
Biochemistry and Physiology B, 119(2): 349–353.
Gottlieb, G. (ed.) (1971). Ontogenesis of Sensory Function in Birds and Mammals. New York:
Academic Press.
Graindorge, N., Alves, C., Darmaillacq, A.-S., Chichery, R., Dickel, L. and Bellanger, C. (2006).
Effects of dorsal and ventral vertical lobe electrolytic lesions on the spatial learning and
locomotor activity in Sepia. Behavioral Neuroscience, 120: 1151–1158.
Granier-Deferre, C. and Busnel, M. C. (1983). Stimulations acoustiques au cours du
développement du système auditif et adaptation au bruit chez la souris. International Journal
of Audiology, 22: 280–310.
28 Anne-Sophie Darmaillacq, Christelle Jozet-Alves, Cécile Bellanger and Ludovic Dickel

Grant, P., Tseng, D., Gould, R. M., Gainer, H. and Pant, H. C. (1995). Expression of neurofilament
proteins during development of the nervous system in the squid Loligo pealei. Journal of
Comparative Neurology, 356(2): 311–326.
Guerra, A. (2006). Ecology of Sepia officinalis. Vie et Milieu, 56: 97–107.
Guibé, M., Boal, J. G. and Dickel, L. (2010). Early exposure to odors changes later visual prey
preferences in cuttlefish. Developmental Psychobiology, 52: 833–837.
Guibé, M. and Dickel, L. (2011). Embryonic visual experience influences posthatching shelter
preference in cuttlefish. Vie Milieu, 61(4): 243–246.
Guibé, M., Poirel, N., Houdé, O. and Dickel, L. (2012). Food imprinting and visual generalization
in embryos and newly hatched cuttlefish (Sepia officinalis). Animal Behaviour, 84: 213–217.
Hanlon, R. T. and Messenger, J. B. (1988). Adaptive coloration in young cuttlefish (Sepia offic-
inalis L.): the morphology and development of body patterns and their relation to behaviour.
Philosophical Transactions of the Royal Society of London. Series B, 320: 437–487.
Hanlon, R. T. and Messenger, J. B. (eds.) (1996). Cephalopod Behaviour. Cambridge (UK):
Cambridge University Press.
Healy, S. D. (2006). Imprinting: seeing food and eating it. Current Biology, 13: R501.
Henriksen, R., Rettenbacher, S. and Groothuis, T. G. G. (2011). Prenatal stress in birds: pathways,
effects, function and perspectives. Neuroscience and Biobehavioral Reviews, 35: 1484–1501.
van Heteren, C. F., Boekkooi, P. F., Jongsma, H. W. and Nijhuis, J. G. (2001). Fetal habituation to
vibroacoustic stimulation in relation to fetal states and fetal heart rate parameters. Early Human
Development, 61: 135–145.
Honeycutt, H. and Lickliter, R. (2003). The influence of prenatal tactile and vestibular stimulation
on auditory and visual responsiveness in bobwhite quail: a matter of timing. Developmental
Psychobiology, 43(2): 71–81.
Iversen, S. D. and Iversen, L. L. (2007). Dopamine: 50 years in perspective. Trends in Neuroscience,
30(5): 188–193.
Jabès, A., Lavenex, P. B., Amaral, D. G. and Lavenex, P. (2010). Quantitative analysis of postnatal
neurogenesis and neuron number in the macaque monkey dentate gyrus. European Journal of
Neuroscience, 31: 273–285.
Jozet-Alves, C. and Hébert, M. (2012). Embryonic exposure to predator odour modulates visual
lateralization in cuttlefish. Proceedings of the Royal Society. B: Biological Sciences, 280(1752):
20122575.
Jozet-Alves, C., Romagny, S., Bellanger, C. and Dickel, L. (2012). Cerebral correlates of visual
lateralization in Sepia. Behavioural Brain Research, 234: 20–25.
Jozet-Alves, C., Viblanc, V., Romagny, S., Dacher, M., Healy, S. D. and Dickel, L. (2012). Visual
lateralization is task- and age-dependent in cuttlefish (Sepia officinalis). Animal Behaviour, 83:
1313–1318.
King, A. J. and Adamo, S. A. (2006). The ventilatory, cardiac and behavioural responses of resting
cuttlefish (Sepia officinalis L.) to sudden visual stimuli. Journal of Experimental Biology, 209:
1101–1111.
Krasnegor, N. A., Blass, E. M., Hofer, M. A. and Smotherman, W. (eds.) (1987). Perinatal
Development: A Psychobiological Perspective. New York: Academic Press.
Lemaire, J. (1970). Table de développement embryonnaire de Sepia officinalis L. (Mollusque
céphalopode). Bulletin de la Société Zoologique de France, 95: 773–782.
Letzkus, P., Boeddeker, N., Wood, J. T., Zhang, S. W. and Srinivasan, M. V. (2008). Lateralization
of visual learning in the honeybee. Biology Letters, 4: 16–19.
Letzkus, P., Ribi, W. A., Wood, J. T., Zhu, H., Zhang, S. W. and Srinivasan, M. V. (2006).
Lateralization of olfaction in the honeybee Apis mellifera. Current Biology, 16: 1471–1476.
 Cuttlefish preschool or how to learn in the peri-hatching period 29

Lickliter, R. (1993). Timing and the development of perinatal perceptual organization. In Devel-
opmental Time and Timing, Turkewitz G. and Devenny D. A. (eds.). Hillsdale, NJ: Erlbaum.
Magat, M. and Brown, C. (2009). Laterality enhances cognition in Australian parrots. Proceedings
of the Royal Society. B: Biological Sciences, 276: 4155–4162.
Messenger, J. B. (1968). The visual attack of the cuttlefish, Sepia officinalis. Animal Behaviour,
16: 342–357.
Messenger, J. B. (1971). Two stage recovery of a response in Sepia. Nature, 232: 202–
203.
Messenger, J. B. (1973). Learning performance and brain structure: a study in development. Brain
Research, 58: 519–523.
Messenger, J. B. (1996). Neurotransmitters of cephalopods. Invertebrate Neuroscience, 2(2): 95–
114.
Messenger, J. B. and Sanders, G. D. (1972). Visual preference and two-cue discrimination learning
in Octopus. Animal Behaviour, 20: 580–585.
Mpitsos, G. J., Murray, T. F., Creech, H. C. and Barker, D. L. (1988). Muscarinic antagonist
enhances one-trial food-aversion learning in the mollusc Pleurobranchaea. Brain Research
Bulletin, 21(2): 169–179.
Nixon, M. and Mangold, K. (1998). The early life of Sepia officinalis, and the contrast with that
of Octopus vulgaris (Cephalopoda). Journal of Zoology, 245(4): 407–421.
Nixon, M. and Young, J. Z. (eds.) (2003). The Brains and Lives of Cephalopods. New York: Oxford
University Press.
Packard, A. (1972). Cephalopods and fish: the limits of convergence. Biological Reviews, 47:
241–307.
Pascual, A., Huang, K.-L., Nevue, J. and Préat, T. (2004). Brain asymmetry and long-term memory.
Nature, 427: 605–606.
Poirier, R. (2004). Expérience précoce et ontogenèse des comportements défensifs chez la seiche
(Sepia officinalis): approches comportementales et neurobiologiques. Unpublished thesis (PhD),
Université de Caen Basse-Normandie.
Poirier, R., Chichery, R. and Dickel, L. (2004). Effect of rearing conditions on sand digging
efficiency in juvenile cuttlefish. Behavioural Processes, 67(2): 273–279.
Poirier, R., Chichery, R. and Dickel, L. (2005). Early experience and postembryonic maturation
of body patterns in cuttlefish (Sepia officinalis). Journal of Comparative Psychology, 119(2):
230–237.
Punzo, F. (2002). Food imprinting and subsequent prey preference in the lynx spider, Oxyopes
salticus. Behavioural Processes, 58: 177–181.
Riederer, B. M. (1992). Differential phosphorylation of some proteins of the neuronal cytoskeleton
during brain development. The Histochemical Journal, 24: 783–790.
Rogers, L. J. (2007). Lateralization in its many forms, and its evolution and development. In The
Evolution of Hemispheric Specialization in Primates, Special Topics in Primatology, Hopkins,
W. D. (ed.), vol. 5. Amsterdam: Elsevier.
Rogers, L. J., Zucca, P. and Vallortigara, G. (2004). Advantages of having a lateralized brain.
Proceedings of the Royal Society. B: Biological Sciences, 271(Suppl): S420–422.
Romagny, S. (2010). Ontogeny of the sensory systems and early learning in the cuttlefish, Sepia
officinalis: behavioural and neurobiological approaches. Unpublished thesis (MSc), Université
Paris 13.
Romagny, S., Darmaillacq, A.-S., Guibé, M., Bellanger, C. and Dickel, L. (2012). Feel, smell and
see: emergence of perception and learning in an immature invertebrate, the cuttlefish embryo.
Journal of Experimental Biology, 215: 4125–4130.
30 Anne-Sophie Darmaillacq, Christelle Jozet-Alves, Cécile Bellanger and Ludovic Dickel

Sanders, F. K. and Young, J. Z. (1940). Learning and other functions of the higher nervous centres
of Sepia. Journal of Neurophysiology, 3: 501–526.
Sanders, G. D. (1975). The cephalopods. In Invertebrate Learning, Corning, W. C., Dyal, J. A.
and Willows, A. O. D. (eds.), vol 3. New York: Plenum Press.
Schaal, B. (1988). Olfaction in infants and children: developmental and functional perspectives.
Chemical Senses, 13: 145–190.
Shettleworth, S. J. (ed.) (2010). Cognition, Evolution and Behaviour. New York: Oxford University
Press.
Shomrat, T., Graindorge, N., Bellanger, C., Fiorito, G., Loewenstein, Y. and Hochner, B. (2011).
Alternative sites of synaptic plasticity in two homologous “Fan-out Fan-in” learning and memory
networks. Current Biology, 21: 1773–1782.
Sluckin, W. (ed.) (2007). Imprinting and Early Learning, 2nd edn. Piscataway, NJ: Aldine Trans-
action.
Smotherman, W. P. and Robinson, S. R. (1992). Prenatal experience with milk: fetal behavior and
endogenous opioid systems. Neuroscience and Biobehavioral Reviews, 16: 351–364.
Stasiak, M. (2002). The development of food preferences in cats: the new direction. Nutritional
Neuroscience, 5: 221–228.
Vallortigara, G. and Bisazza, A. (2002). How ancient is brain lateralization? In Comparative Ver-
tebrate Lateralization, Andrew R. J. and Rogers L. J. (eds.). Cambridge: Cambridge University
Press.
Vallortigara, G. and Rogers, L. J. (2005). Survival with an asymmetrical brain: advantages and
disadvantages of cerebral lateralization. Behavioral and Brain Sciences, 28: 575–633.
Vauclair, J. (1996). La cognition animale. Que sais-je? Paris: Presses Universitaires de France.
Wasserman, E. A. and Zentall, T. R. (eds.) (2006). Comparative Cognition: Experimental Explo-
rations of Animal Intelligence. New York: Oxford University Press.
Wells, M. J. (1962). Early learning in Sepia. Symposia of the Zoological Society of London, 8:
149–159.
Wirz, K. (1954). Etudes quantitatives sur le système nerveux des Céphalopodes. Comptes rendus
de l’Académie des sciences, Paris, 238: 1353–1355.
Young, J. Z. (1965). Influence of previous preferences on the memory of Octopus vulgaris after
removal of the vertical lobe. Journal of Experimental Biology, 43: 595–603.
Young, J. Z. (1979). The nervous system of Loligo. V: the vertical complex. Philosophical Trans-
actions of the Royal Society of London. Series B, 285: 311–354.
2 Evolution of behavioral and neural
complexity: learning and memory in
Chambered Nautilus
Jennifer Basil and Robyn Crook

We study learning in the ancient cephalopod Nautilus pompilius as a way of understand-

ing how cognition and neural complexity may have evolved in its more derived relatives,
the coleoid cephalopods (octopuses, cuttlefish and squid), whose brains are heavily
invested in learning and memory. Nautilus retains a relatively primitive brain compared
with the brains of coleoids, which are among the most complex of all the invertebrates.
Coleoid brains contain several discrete lobes dedicated to learning and memory storage –
the vertical and frontal lobe complexes. The nautilus central nervous system lacks known
homologs or analogs of these regions, and instead the brain more likely approximates
those of the ancestors of the two extant subclasses. Nautilus is, therefore, uniquely placed
in this lineage for studies of the evolution and origin of cognition, including learning and
memory. Our laboratory identifies and characterizes learning, memory formation, com-
position and retrieval in Nautilus through controlled classical-conditioning studies and
behavioral assays with free-moving animals. In studies of classical conditioning where a
light predicted the reward of food, nautiluses learned rapidly, expressing temporally sep-
arated short-term and long-term memory. Short-term memory in Nautilus was similar to
that of coleoids, but long-term memory was of shorter duration. Spatial-memory studies
in mazes reveal far more complex learning and memory than the “primitive brain” of
nautiluses might predict. Providing increasingly complex cue arrays to be learned for
spatial navigation to a goal, we tested nautiluses in cognitive tasks similar to those stud-
ied in coleoids. Our studies suggest that despite lacking the dedicated neural structures
of its more derived relatives, nautiluses perform surprisingly well at cognitive tasks, and
show flexibility in what they can learn and remember, even simultaneously. Perhaps the
antecedent of the vertical or frontal lobe may be present in the Nautilus brain or there is
an analogue in existing structures. Either of these possibilities may be identified through
future electrophysiological and neuroanatomical experiments.

2.1 Introduction and comparative approaches

Cephalopod mollusks are a unique taxon for the comparative study of cognition because:
(1) they exhibit complex behavior; (2) there is substantial variety in the natural histories
Cephalopod Cognition, eds A.-S. Darmaillacq, L. Dickel and J. Mather. Published by Cambridge University
Press. 
C Cambridge 2014.
32 Jennifer Basil and Robyn Crook

of the cephalopods that may contribute to behavioral differences; (3) their neuroanatomy
shares many convergent features with vertebrate organization; (4) the brain of extant
Chambered Nautilus retains primitive features of the cephalopod lineage, while the
coleoid brain is highly derived and specialized; yet (5) we have discovered complex
behaviors in nautiluses that overlap with those of their relatives, offering an exceptional
opportunity for phylogenetic analysis of cognition not available in many other closely
related lineages. To this end we characterize underlying factors shaping associative
learning of ecologically relevant stimuli in nautiluses. Our aim is to reveal adaptive con-
straints contributing to cognition, large brains and fast and flexible processing by careful
comparative study of a class that expresses both the ancestral and derived conditions in
extant species.
Cephalopods are a diverse and successful group of mollusks. All living species belong
to either the coleoidea (internal, reduced or absent shell: squids, octopuses, cuttlefishes,
600–800 species) or the nautiloidea (external, chambered shell: limited to two genera
and perhaps five total species (Bonnaud, Ozouf-Costaz & Boucher-Rodoni, 2004; Saun-
ders & Landman, 1987; Ward & Saunders, 1997)). A major contributing factor to the
divergence of these two subclasses was a period of intense competition with vertebrates
about 380 million years ago (MYA) (Aronson, 1991; Chamberlain, 1993; Grasso &
Basil, 2009; Packard, 1972). The complex behavioral abilities of coleoids are thought
to have developed from a “cognitive radiation” during this period, which minimized
competition with fishes, as coleoids cannot outperform them (Chamberlain, 1993). Nau-
tiloids also avoid fishes, but in a fundamentally different way. Coleoids opt for a “live
fast/die young,” r-selected, reproductive lifestyle characterized by rapid growth, fast
metabolic rates, active hunting and voracious feeding. Nautiloids rely instead upon a
“live slow/die old,” k-selected lifestyle, where they dwell in cold, deep waters, avoiding
predation and competition with most teleosts by scavenging in darkness rather than
active hunting. Because they spend most of their time at depth, little is known of the
cognitive behavior of the Nautilus and the theory that a “cognitive radiation” in coleoids
served to minimize niche overlaps with day-hunting teleosts (Aronson, 1991; Chamber-
lain, 1990, 1993; Packard, 1972) developed largely in the absence of much information
on the behavior of nautiluses. In the past few years, researchers have identified sur-
prising cognitive abilities in Nautilus, casting these theories in a new light. Perhaps the
cephalopod brain entering this period of intense competition millions of years ago was
already large and capable of complex behaviors (Grasso & Basil, 2009) having previ-
ously coped with repeated periods of rapid diversification and competition following
mass die offs.

2.2 Sensory ecology

An understanding of the internal world view of Nautilus (Figure 2.1) is essential to

develop hypotheses about their cognitive abilities and to identify relevant ecological
pressures shaping their behavior (Kamil, 1994; Tinbergen, 1963). Nautilus lives among
 Learning and memory in Chambered Nautilus 33

Shell

Hood

Tentacle
sheath

Eye

Rhinophore
(under shell)
Ocular Digital tentacle
tentacle

Figure 2.1 Chambered Nautilus, with pertinent anatomical features labeled. External shell is
chambered, with the animal residing in the front chamber only. Hood can close, protecting the
animal inside the first chamber. Up to 90 distinct tentacles can be retracted into sheaths when at
rest. Olfactory rhinophores lie below each eye, exposed to the exterior by a narrow pore.

the coral reefs of the Indo-Pacific, remaining in dimly lit waters at depths up to 300 m
for most of its life. They undergo daily vertical migrations, traveling to shallower,
warmer waters (75 m) to forage amidst coral reefs during night-time hours and make
brief excursions throughout the day (Carlson, McKibben & De Gruy, 1984; Dunstan,
Ward & Marschall, 2011; Ward, Carlson, Weekly & Brumbaugh, 1984). At depth nau-
tiluses are likely to rely heavily upon chemical and tactile sensory modalities, especially
since they are primarily nocturnal. The structure of the large but primitive pinhole eye
of Nautilus also suggests vision may not be an essential sensory system for foraging
(Messenger, 1991) as it is for other modern cephalopods (Hanlon & Messenger, 1996;
Messenger, 1991; Muntz, 1986, 2010a; Muntz & Raj, 1984). Coleoid cephalopods have
well-developed eyes with acute vision (Budelmann, 1995) and large optic lobes for
processing visual input (Nixon & Young, 2003). While the eyes of Nautilus lack lenses,
they are extremely large to capture as much light as possible in a dim environment
and tuned to wavelengths (470 nm) commonly produced by bioluminescent organisms.
Nautiluses are also positively phototactic (Muntz, 1986, 2010a, 2010b). Recent labora-
tory experiments (Crook & Basil, 2008a, 2013; Crook, Hanlon & Basil, 2009) reveal
that nautiluses can use visual stimuli to learn simple associations and also to locate
goals in complex spatial arrays (horizontal shallow-water mazes and vertical artificial
reefs). Vision may play a greater role in their lives than previously thought, and it may
34 Jennifer Basil and Robyn Crook

be that food items such as decaying organisms provide both visual and olfactory cues
to a scavenging nautilus in the wild. Nautiluses associating these kinds of stimuli to
predict the presence or location of a food item based upon experience is likely important
for their survival. Recalling good hiding locations in the coral reefs where they live
would also be adaptive for an animal with few defenses other than an external shell
(nautiluses lack ink, are relatively slow moving and do not have the capacity for active
crypsis).
Nautiluses seem to be specialized for a “smelling and groping” lifestyle (Hanlon
& Messenger, 1996; Saunders, 1985). A pair of olfactory organs (rhinophores), one
located below each eye, is open to the exterior by a narrow pore (Figure 2.1; Barber
& Wright, 1969; Ruth, Schmidtberg, Westermann & Schipp, 2002). Coleoids also bear
rhinophores (Gilly & Lucero, 1992; Woodhams & Messenger, 1974) and have extensive
tactile and chemical sensitivity in their suckers (Graziadei, 1964). The rhinophores of
nautiluses are similar to the olfactory organs in Octopus and other cephalopods but
are significantly larger, as are the olfactory lobes in the brain of Nautilus to which
they project (Young, 1965a). The epithelium of their rhinophores possesses cells that
are similar to chemoreceptors located in the sucker of Octopus, the olfactory structure
of squids and the lip of Sepia (Emery, 1975; Gilly & Lucero, 1992; Graziadei, 1964;
Lucero, Horrigan & Gilly, 1992; Ruth, Schmidtberg, Westermann & Schipp, 2002). The
paired rhinophores are essential for orientation to food odor, especially at a distance
(Basil et al., 2005; Basil, Hanlon, Sheikh & Atema, 2000), and nautiluses use odor to
discriminate between males and females at short range (Basil, Lazemby, Nakanuku &
Hanlon, 2002; Westermann & Beuerlein, 2005). Odor on a variety of spatial scales is,
therefore, an important source of information to nautiluses in their complex coral-reef
environment. Olfactory memory for predators and prey, along with possible mates, may
be important to their survival in the wild.
Cephalopod arms and tentacles actively collect sensory information from their envi-
ronment and nautiluses are no exception. Coleoid cephalopods have either 8 or 10
sucker-bearing appendages, while nautiluses possess a large array of up to 90 tentacles,
lacking suckers and retracted into sheaths when the animals are at rest. These tentacles
are covered in both mechanosensory (Kier, 1987) and chemosensory structures (Fukuda,
1980). Ruth, Schmidtberg, Westermann and Schipp (2002) described ciliated cells on the
epithelium of the tentacles, suggesting they serve both a chemosensory and mechanosen-
sory function at a variety of spatial scales. When the rhinophore is stimulated by odor,
nautiluses extend their tentacles up to a shell length in a stereotyped posture and swim
in the odor plume toward the source of the odor (Basil, Hanlon, Sheikh & Atema, 2000;
Bidder, 1962), from up to 10 m away. Tentacle extension is in general a reliable and
quantifiable measure of arousal in nautiluses (Basil et al., 2005; Basil, Hanlon, Sheikh &
Atema, 2000; Bidder, 1962; Crook & Basil, 2008a; Ruth, Schmidtberg, Westermann &
Schipp, 2002). We have demonstrated in a naturalistic odor-tracking paradigm, and by
reversible rhinophore blockage, that Nautilus: (1) detects highly dilute odors at a dis-
tance of up to 10 m; (2) uses turbulent odor-plume information to locate the distant
odor source; and (3) relies upon its paired rhinophores to track the odor accurately
(Basil et al., 2005; Basil, Hanlon, Sheikh & Atema, 2000). Stimulation of the digital
 Learning and memory in Chambered Nautilus 35

tentacles initiates near-field food-searching behavior (digging, etc; Basil et al., 2005).
Clearly, complex odor and tactile stimuli are of great importance to these animals in
their deep-sea, coral-reef habitat where little light penetrates.
Another possible source of environmental information in deep water is vibration
and nautiluses detect and respond to underwater vibrations (Soucier & Basil, 2008).
Nautiluses slowed their breathing in response to vibrations, especially when the source
of the vibration was within 20 cm. When the amplitude and frequency of vibrations
increased, nautiluses decreased their respirations further (Soucier & Basil, 2008). Poten-
tial receptors for environmental vibratory stimuli (e.g. snapping shrimp, etc.) include
the mechanosensory receptors found on their tentacles and also at the base of their
rhinophores (Ruth, Schmidtberg, Westermann & Schipp, 2002). Based on their responses
to environmentally relevant stimuli, nautiluses can collect detailed information on many
scales about a suite of environmental factors. Learning to associate these stimuli and
remember their location would be advantageous in an animal with few defenses living
in a complex, deep-sea environment.

2.3 Neural systems

Coleoid cephalopods have developed heavily modified brains that include dedicated
learning and memory centers (Figure 2.2; vertical lobe and superior frontal lobe; Agin,
Chichery, Maubert & Chichery, 2003; Boycott & Young, 1955; Dickel, Chichery &
Chichery, 2001; Graindorge et al., 2006; Hochner, Shomrat & Fiorito, 2006; Young,
1960, 1988; Young & Boycott, 1971). In fact, coleoid brains have been proposed to
converge with the vertebrate brain in organization (see Hochner, Shomrat & Fiorito,
2006 for review; Boyle, 1986; Kandel, 1976; Nixon & Young, 2003; Wells, 1978;
Young, 1965b, 1991), with shorter connectives and synaptic morphology consistent
with fast and flexible neural processing (Hochner, Shomrat & Fiorito, 2006). When the
vertical lobe is ablated or lesioned in octopuses and cuttlefishes, learning and memory of
certain stimuli are abolished (Boycott, 1961; Boycott & Young, 1955; Graindorge et al.,
2006; Maldonado, 1965; Wells, 1978; Young, 1991). Observational learning, a complex
cognitive behavior exhibited by coleoids (Fiorito & Scotto, 1992), also requires the
vertical lobe (Fiorito & Chichery, 1995). Ecology impacts the organization of the coleoid
brain. There is parametric variation in memory abilities and neural systems among
coleoids depending upon their primary habitat. Many octopods have a well developed
chemo-tactile memory system (subfrontal lobe) suited for their benthic habitat. This
area of the brain is absent or poorly developed in pelagic octopods, which do not
display complex chemo-tactile learning and memory, illustrating an ecological impact
on brain and behavior (Young, 1964). On a cellular level, long-term memory formation in
cuttlefish is dependent upon de novo protein synthesis at distinct time points after learning
(Agin, Chichery, Maubert & Chichery, 2003) as in other animals (Davis & Squire,
1984; Stork & Welzl, 1999). Short-term memory in vertebrates and in invertebrates is
mediated by transient changes in synaptic morphology (for a review of invertebrates
and vertebrates, see Stork & Welzl, 1999). The neuroanatomy of dedicated learning
36 Jennifer Basil and Robyn Crook

Figure 2.2 Comparison of cephalopod brains. (a) Nautiloids, (b) Octopods, (c) Cuttlefishes,
(d) Squids. The vertical lobe complex (VL) plays a substantial role in learning and memory in
coleoids, but is not found in nautiluses. The laminated area of the olfactory lobe (LAM) in
nautiloids lies in a similar position to the vertical lobe complex in coleoids. SFL = subfrontal
lobe complex. (Modified from Maddock & Young, 1987; figure reproduced from Grasso & Basil,
2009, with permission from S. Karger AG, Medical and Scientific Publishers, Basel.)
 Learning and memory in Chambered Nautilus 37

and memory structures of coleoids supports the hypothesis of anatomical convergence

with vertebrates, especially since the cellular spiking properties and neuroanatomical
arrangement conserve ancient molluskan features (for a review, see Hochner, Shomrat &
Fiorito, 2006).
Nautiluses have a simple ring-shaped brain with little internal differentiation
(Figure 2.2; Maddock & Young, 1987; Young, 1965a) and with only 13 lobes in contrast
to the 38 found in squid and 40 in Octopus (Hochner, Shomrat & Fiorito, 2006; Young,
1965b; Young & Boycott, 1971), though the nautilid brain is considerably larger than the
brains of other mollusks (Aplysia: Carew & Sahley, 1986; Kandel & Schwartz, 1982;
Hermissenda: Alkon et al., 1990; Blackwell, 2006; Snails: Benjamin, 2000; Gelperin
& Culligan, 1984). Recent embryological evidence (Sasaki, Shigeno & Tanabe, 2010;
Shigeno et al., 2008; Shigeno, Takenori & Boletzky, 2010) suggests that nautilus retains
features of the neural arrangement of the common cephalopod ancestor, which may
help to illuminate how brain and cognitive complexity arose in the cephalopod lineage
as a whole.
Of crucial importance is the apparent absence in nautilus of any defined structures
analogous to the vertical or superior frontal lobe of coleoids (e.g. Boycott & Young,
1955; Young, 1965a). It is not surprising that the brain of Nautilus has a greatly expanded
olfactory lobe, with their reliance on olfactory stimuli. One possible analog to coleoid
learning-and-memory centers is the laminated area of the olfactory lobe (Figure 2.2,
LAM) and surrounding cerebral nerve cord, located in a similar region in nautilus as
the vertical lobe in coleoids (Young, 1965a). This laminated area receives input from
the plexiform region (Figure 2.2), which integrates information from the visual and
tactile systems of the animal (Young, 1965a). Based on its location and projections, the
plexiform region is a likely comparator to the superior frontal lobes of coleoids. Perhaps
the basic neural arrangement of nautiloids is conserved, although greatly expanded in
coleoids, or nautiluses use another area of the brain entirely to support their learning
(Grasso & Basil, 2009). There are currently no studies identifying regions in the nautilus
brain essential for learning and memory as there are for coleoids.

2.4 Associative conditioning

Animals with simple neural organization, including mollusks, have proven invaluable
to our understanding of learning and memory (for a review, see: Burrell & Sahley,
2001; Carew & Sahley, 1986; Milner, Squire & Kandel, 1998). A number of feeding
and avoidance reflexes in the marine mollusk Aplysia can be classically conditioned
and the neural, synaptic and cellular properties mediating learning have been well
characterized (Baxter & Byrne, 2006; Brembs, Baxter & Byrne, 2004; Carew & Sahley,
1986; Kandel, 1976; Kandel & Schwartz, 1982). The marine mollusk Hermissenda
can learn to associate a light with turbulence, which is adaptive for an organism that
must brace itself against incoming wave action during storms (Blackwell, 2006; Crow,
2004; Crow & Alkon, 1978). Importantly, simple invertebrate learning systems share
fundamental physiological and molecular features in common with variety of animal
38 Jennifer Basil and Robyn Crook

groups (e.g. Burrell & Sahley, 2001; Kandel & Tauc, 1965; Hermissenda: Blackwell,
2006; Aplysia: Walters, 1991; Walters, Alizadeh & Castro, 1991; Lymnaea: Benjamin,
2000; Snails: Gelperin & Culligan, 1984; Insects: Davis, 2005, including vertebrates,
e.g. Fanselow & Poulos, 2005). Even in animals with simple neural systems, learning
can occur rapidly and memory can persist for long periods of time. This is true of
Nautilus (Crook & Basil, 2008a, 2008b, 2013; Crook, Hanlon & Basil, 2009; Grasso
& Basil, 2009) and also coleoids. It is probably adaptive for animals to detect salient
features in their environment and predict them, but there are also costs from a high
investment in learning and memory (and supporting structures). Since nautilus retains
pleisiomorphic neural features of the cephalopod lineage – a lineage known for its
advanced learning capabilities – a thorough understanding of the capabilities of nautilus
allows us to predict and test what adaptive role learning and memory has played in
shaping cephalopod cognition.
Coleoids exhibit a range of cognitive behaviors (reviewed in Alves, Chichery, Boal &
Dickel, 2007; Hanlon & Messenger, 1996; Mather, 1995; Nixon & Young, 2003; Wells,
1978) including imprinting (Darmaillacq, Chichery & Dickel, 2006; Darmaillacq,
Chichery, Shashar & Dickel, 2006), associative learning (Agin, Chichery, Dickel &
Chichery, 2006; Agin, Dickel, Chichery & Chichery, 1998; Agin, Poirier, Chichery,
Dickel & Chichery, 2006; Darmaillacq, Dickel, Chichery, Agin & Chichery, 2004), dis-
criminative learning (Boal, 1996; Cole & Adamo, 2005; Hvorecny et al., 2007; Suther-
land, 1963), observational learning (Fiorito & Scotto, 1992; Suboski, Muir & Hall, 1993)
and short- and long-term memory (Agin, Poirier, Chichery, Dickel & Chichery, 2006;
Sanders, 1970). Behavioral changes as a result of learning are long lasting in relation
to coleoids’ typically brief lifespan (Agin, Dickel, Chichery & Chichery, 1998; Alves,
Chichery, Boal & Dickel, 2007; Boal, 1991; Boal, Dunham, Williams & Hanlon, 2000;
Darmaillacq, Dickel, Chichery, Agin & Chichery, 2004; Hanlon & Messenger, 1996;
Messenger, 1973). Coleoid cephalopods can be operantly conditioned to avoid attacking
a prawn contained within a hard Plexiglas tube (Agin, Chichery, Dickel & Chichery,
2006; Cartron, Darmaillacq & Dickel, 2013) and with this paradigm researchers have dis-
covered that cuttlefish possess temporally separated short and long-term memory stores
(Agin, Dickel, Chichery & Chichery, 1998; Agin, Poirier, Chichery, Dickel & Chich-
ery, 2006; Messenger, 1973). For comparison, we have used associative-conditioning
techniques to probe the memory profile of Chambered Nautilus.
In classical associative conditioning, an initially neutral conditioned stimulus (CS,
e.g. bell) becomes associated with an unconditioned stimulus (US, e.g. food) and sub-
sequently elicits a response on its own (the conditioned response, CR, e.g. salivation)
previously only initiated by the US (Pavlov, 1927). Models of classical conditioning
(e.g. Pavlov, 1927; Rescorla & Wagner, 1972) posit that: (1) learning can be thought of
in terms of a change in association strength between the CS and the US; and that (2)
variation in stimulus salience and CS/US timing or pairing can contribute to changes
in the strength of the association. A hallmark of classical conditioning is that the CS
must precede the US in a reliable way. The period between them should not be too
great or the associative strength will lessen. This is true across a variety of animal phyla
(e.g. Bees: Bitterman, Menzel, Fietz & Schäfer, 1983; Menzel, Leboulle & Eisenhardt,
2006; Mollusks: Blackwell, 2006; Vertebrates: Rescorla & Wagner, 1972). According
 Learning and memory in Chambered Nautilus 39

Mirror

Tank surround

Experimental
arena

Camcorder

Blind

Depth adjuster

Tripod
Pipette guide tube
Harness

Figure 2.3 Classical conditioning aquarium. A small light is flashed onto the back of the tank
(conditioned stimulus) to create a dispersed 470 nm illumination of the tank for 1 s and to avoid
direct contact with the eye of the animal. The harness holding the shell (inset) aids in controlled
delivery of the odor reward (unconditioned stimulus) via the guide tube, 1 s later. Tentacles are
free to respond to either of the stimuli and behavior is captured on video by a camera 0.5 m from
the harness. (Reproduced from Crook and Basil (2008a), with permission from Company of
Biologists, Ltd.)

to models like those of Rescorla and Wagner (1972), after the association is learned
associative strength will decline if the CS is occasionally not followed by the US in a
predictable way, or at all. When there is no US present as expected, a discrepancy arises.
This results in both unlearning of the association and a learning of the absence of the
US. The CS subsequently loses associative strength on each trial where this is the case
until it has none at all and the animal no longer responds (extinction).
For our studies, we used a pairing of light (CS) to signal a food-odor reward (US)
in an experimental aquarium where the nautilus was held in place in a harness by its
shell, allowing it tentacles to freely extend (CR) (Figure 2.3, conditioning apparatus;
Crook & Basil, 2008a). In forward-paired conditioning, or CS+ trials, 2 ml of fish-head
stimulus (US) was taken into a pipette, which was inserted into a guide tube attached
to the harness for release onto the rhinophores. A blue light (470 nm, CS) was flashed
for 500 ms against a predetermined spot on the back of the experimental tank. With a
40 Jennifer Basil and Robyn Crook

Hood length Tentacle length

Ocular tentacles Tentacle sheath Digital tentacles

Figure 2.4 Tentacle extension response (TER), calculated from tentacle length/hood proportion.
TER is used as a conditioned response in associative conditioning as a measure of learning the
association between a light pulse (conditioned stimulus) and an odor reward (unconditioned
stimulus). Tentacles are retracted when at rest and can extend up to a full body length from their
sheaths when the animals are aroused. (Reproduced from Crook and Basil (2008a), with
permission from Company of Biologists, Ltd.)

delay of 1 s, the odor reward was released from the pipette (Figure 2.3, inset). We ranked
stereotyped tentacle extension over the following 10 min, as the CR during acquisition
and then at various delays after training, to probe the nature of short-term (ST) and
long-term (LT) memory (tentacle extension response [TER], Figure 2.4; modified from
bee proboscis extension response [PER]; Bitterman, Menzel, Fietz & Schäfer, 1983;
Chandra, Hosler & Smith, 2000; Menzel, Leboulle & Eisenhardt, 2006; Smith, 1997;
Smith & Cobey, 1994).
Memory testing began after the last trial of training to map the time course of
short-term and long-term memory. We tested the following memory-retention intervals:
3 min, 30 min, 1 h, 6 h, 12 h and 24 h post training (counterbalanced to control for any
experiential effects). For each retention interval, separate animals were tested with a
single unrewarded presentation of the light (CS) and their tentacle extension responses
measured for 3 min. Excitatory conditioning is thought to be pairing dependent, with
the CS predicting the US in a reliable way. To determine that excitatory conditioning
had occurred, we also assessed learning and memory with an explicitly unpaired group,
where the US was presented at random intervals relative to the CS (CS–). After training,
retention was tested in this group at the same intervals as above.
 Learning and memory in Chambered Nautilus 41

Memory curves over time

10
Ventilation rate (cycles 5s–1)

CS –
8 CS +
Baseline
STM
6 LTM

0
0 .5 1 3 6 9 12 15 18 21 24
Time (h)
Figure 2.5 Biphasic memory profile during associative conditioning. Y axis: Proportion of trials
with a tentacle extension response; X axis: Time after initial training. In excitatory forward-
paired (CS+) , trials (black), there is a peak of memory at 1 h after training that drops and then
rises again at about 6 h, declining until 24 h. No such learning curve is detected in explicitly
unpaired (CS–) , trials (grey). Bars indicate SEM.

Our studies revealed that nautiluses exhibit a biphasic learning curve with identifiable
short-term and long-term memory peaks (Figure 2.5, Crook & Basil, 2008a) as found
in cuttlefishes (Agin, Poirier, Chichery, Dickel & Chichery, 2006; Messenger, 1973).
Duration of short-term memory in nautiluses is comparable to other cephalopods (up to
1 h; Agin, Poirier, Chichery, Dickel & Chichery, 2006) but long-term memory is much
shorter, at least under these conditions (12 h in Nautilus versus up to a week in Octopus,
(Young, 1961; Figure 2.5). In contrast, in explicitly unpaired controls, where the delivery
of odor was not reliably signaled by the flash of light, there was no increase in tentacle
extension during training or retention testing. More tentacle extension (CR) in response
to the light (CS) during training in the excitatory-conditioning group demonstrates that
excitatory learning occurs only when there is a predictable temporal relationship between
the light (CS) and the odor (US).
Nautiluses exhibit almost no memory for the task 6 h after initial training, hypothet-
ically because that is when short-term memory is being consolidated into long-term
memory. We do not know the exact time-course or mechanism of consolidation between
1 and 6 h. We currently determine if the memory stores are separate, as may be the
case in cuttlefish, or if there is some form of functional intermediate-term memory
(McGaugh, 2000) as in the snail Lymnaea (Lukowiak, Adatia, Krygier & Syed, 2000;
Sangha, Scheibenstock, McComb & Lukowiak, 2003). Long-term memory formation in
cuttlefishes is dependent upon translational processes at critical time points after learn-
ing (Agin, Chichery, Maubert & Chichery, 2003; Pedreira & Maldonado, 2003), but
the timing of long-term memory formation (and perhaps intermediate-term memory) in
42 Jennifer Basil and Robyn Crook

nautiluses is unknown and we continue to probe their memory processing in a variety of

tasks. Identifying the location and processes guiding learning and memory in nautiluses
(e.g. cuttlefishes, Agin, Chichery & Chichery, 2001) will allow us to understand how
behavioral complexity emerges from the variety of neuroanatomical solutions found in
this lineage.

2.5 Spatial navigation

For questions of adaptation and cognition, an analysis of the spatial abilities of cephalo-
pods is illustrative because: (1) a direct link can be made to spatial problems they solve
in the wild; (2) given the variability in natural environments, some form of flexibility
is required of any animal navigating; and (3) similar tasks have been tested in both the
coleoids and nautiloids, so direct comparisons can be made within this lineage. Coleoid
cephalopods are adept at solving spatial problems (e.g. Wells, 1964) – not surprising,
given the complexity of their natural environment and pressures they experience in the
wild (predator avoidance, food localization, returning to known locations and dens, and
finding conspecifics (for a review, see Alves, Boal & Dickel, 2008; Hanlon & Messenger,
1996; see also chapter 7, this volume, by Jozet-Alves et al.). Learning the complex layout
of an environment requires: (1) exploration of the space (Gallistel, 1990; O’Keefe &
Nadel, 1978) – with animals often exhibiting (2) habituation to the space over time
once the layout has been learned (e.g. voles: Shillito, 1963; hamsters: Poucet, Chapuis,
Durup & Thinus-Blanc, 1986; blind cave fish: Teyke, 1985, 1988, 1989; crayfish: Basil
& Sandeman, 2000). If a familiar environment is altered in some way, (3) the level
of exploratory behavior increases (dishabituation), presumably because the previously
gathered environmental information must be updated (fish: Welker & Welker, 1958;
hamsters: Poucet, Chapuis, Durup & Thinus-Blanc, 1986; blind cave fish: Teyke, 1985,
1988, 1989; crayfish: Basil & Sandeman, 2000). Octopuses in a complex environment
habituate to the space over time, supporting the notion that they were actively exploring
and learning (Boal, Dunham, Williams & Hanlon, 2000; Karson, Boal & Hanlon, 2003).
After having been removed for a period from the space, they were still able to locate a
preferred den, illustrating their memory for the environment. Nautiluses, too, explore
and become habituated to a vertical, three-dimensional artificial reef over time (Crook,
Hanlon & Basil, 2009), spending more time in contact with the reef as they became
familiar with it. When the spatial configuration of the reef was changed, nautiluses altered
their swimming behavior and proximity to the reef, indicating that during exploration
they had learned something about the original configuration of the reef and had detected
it had changed.
The ability to competently navigate on different spatial scales requires an animal
detect and perhaps learn the relevant topography of its environment. They can simply
remember a series of turns to return to a goal (Gallistel, 1990), which are called response
strategies, including route memory, or, when overall distance and net direction has been
determined and remembered from the movements made on the outgoing journey, known
as dead reckoning or path integration. Alternatively, they can use single visual cues as
“beacons” (Gallistel, 1990, beacon homing), coding the location of a goal in space with a
 Learning and memory in Chambered Nautilus 43

landmark nearby (e.g. Octopus, Mather, 1991). A more complex alternative is arranging
more than one beacon in a “chain” to be followed to a goal (coleoids: Alves, Boal &
Dickel, 2008; Alves, Chichery, Boal & Dickel, 2007; arthropods: Collett, Dillmann,
Giger & Wehner, 1992; Gallistel, 1990). Animals may also use numerous landmarks
distal to the goal to cue its location (piloting or place learning, Kamil & Jones, 1997),
taking multiple bearings from these various landmarks to increase their accuracy (Kamil
& Cheng, 2001). Karson, Boal and Hanlon (2003) and Alves, Chichery, Boal and Dickel
(2007) demonstrated that cuttlefishes can use route information to find a goal location
(a dark spot with a sandy bottom). Depending upon the kinds of cues available and/or
reliable (either outside or inside a maze or spatial array), cuttlefishes can also switch
flexibly between tactics to find a familiar spatial goal, using cues within the maze or
outside the maze to varying degrees. Given their complex natural environment and
the importance of quickly locating home, food and safety, it is perhaps not surprising
octopuses and cuttlefish would have well-developed, complex spatial abilities, probably
based upon constellations of visual cues in their environment.
The spatial abilities of Nautilus are particularly interesting because of their unique
daily vertical migration in the wild, as they swim up and down coral-reef slopes during
night-time hours to forage (Carlson, McKibben & De Gruy, 1984; Dunstan, Alanis &
Marshall, 2010; Ward, Carlson, Weekly & Brumbaugh, 1984). It may be that they can
collect and use spatial information on a large scale during these vertical movements,
relying upon smaller-scale spatial cues (visual, tactile, olfactory) as they hug the coral-
reef face during foraging. In the laboratory, nautiluses are adept at solving spatial
problems and are cognitively flexible in the solutions they use: relying upon a variety of
landmark cues in various subsets to locate a learned escape hole in a shallow-water maze
(Crook, 2008; Crook & Basil, 2013; Crook, Hanlon & Basil, 2009). Nautiluses can use
visual cues as beacons or as a complex array of cues to locate the goal and demonstrate
flexibility in cue use depending upon availability and reliability (Crook & Basil, 2013;
Crook, Hanlon & Basil, 2009). They can also switch dynamically between beacon
homing and route memory to find a goal in response to changes in the environment
(Crook & Basil, 2013). Ecologically, nautiluses’ accommodation of the spatial changes
they encounter on their nightly vertical migrations along coral reefs would benefit from
such a flexible system. The simpler neural architecture of nautiloids, relative to coleoids
at least, seems to belie their true capabilities.
In our spatial experiments, when an escape hole in a horizontal, 1-m diameter, shallow-
water maze (Figure 2.6) is cued only with a simple visual beacon (a white striped ring
of bubble wrap) surrounding the hole, nautiluses learned the goal quickly and memory
persisted for up to 3 weeks, rivaling the performance of octopuses (Crook, Hanlon &
Basil, 2009; Figure 2.7). When the beacon was shifted from the goal, nautiluses first
directed their search over the beacon, only finding the goal after extensive search of
the maze. At least in this case, route-memory tactics did not compensate and allow
the animals to find the goal quickly, with the learning and presence of the beacon
taking priority over the alternate available tactic. In a later control series and with more
experience, however, route memory was likely invoked when the beacon was removed
entirely (Crook, Hanlon & Basil, 2009; Crook & Basil, 2013). In related experiments
with a constellation of objects positioned in the maze in various subsets, nautiluses could
44 Jennifer Basil and Robyn Crook

Camera

Blind

Escape hole

Water level

Maze platform
Start

Tank

Figure 2.6 Shallow-water spatial maze. Total depth of tank is 1.5 m and width is 1 m. Maze depth
is the shell diameter of the animal subject and width matches that of the tank. Goal location (exit
to deeper water) located 80 cm from animal’s start position and is 18 cm in diameter. Landmark,
or beacon, is a 5 cm-wide concentric ring of white-striped bubble wrap surrounding the escape
hole. (Reproduced from Crook, Hanlon & Basil, 2009, with permission from the American
Psychological Association.)

use each subset flexibly, within the maze (both proximate to the goal and distant from the
goal) and outside the maze, to navigate to the escape point. Prior learning of the beacon
during training did not affect the ability of nautiluses to subsequently use local intra-
maze cues alone in tests as long as they did not shift from the configuration in training
(e.g. rats: Thein, Westbrook & Harris, 2008). Nautiluses also expressed a hierarchy for
cues that shifted with cue reliability (Crook, 2008; Crook & Basil, 2013). The salience
of a particular cue and the choice to include it in decision making was dependent on
 Learning and memory in Chambered Nautilus 45

600 Escape times over successive trials

540 a

480

420
Escape time (s)

360 b

300
c
240

180

120

60

0
1 2 3 4 5 18h 24h 36h 48h 72h 96h 1w 3w

Training Test period

Figure 2.7 Long-term retention of spatial information. Time to find exit hole marked by a beacon
in the spatial maze on successive 10-minute trials improves rapidly across 5 training trials and is
retained for at least 3 weeks Points labeled with a different letter are significantly different
(P < 0.05, two-tailed). (Reproduced from Crook, Hanlon & Basil, 2009, with permission from
the American Psychological Association.)

the array of other cues that were present during learning and their reliability in testing.
Similarly, Alves, Boal and Dickel (2008) showed that cuttlefish (Sepia officinalis) were
capable of making spontaneous choices between response- and place-based strategies
based on the type of cues available and the perception of their reliability. Nautiluses also
detect and learn the layout of an artificial reef constructed from black and white cinder
blocks introduced into their home tank. Over time, nautiluses increase their exploration
of the artificial coral reef – coming into closer contact with the structure over a period
of hours, touching the reef extensively with their tentacles and, finally, attaching to the
reef with a few tentacles and remaining still. The nautiluses always detected when the
vertical three-dimensional layout of the artificial coral reef was altered (Crook, Hanlon
& Basil, 2009), swimming at a distance from the reef until, over a period of hours, they
were familiarized with it and made contact. Animals did not behave this way when the
visual pattern, the black and white components of the reef, simply swapped colors, with
the three-dimensional layout remaining the same. In the dark waters where nautiluses
live, this kind of tactile familiarization of local features could contribute to locating good
hiding or foraging locations along the reef.
In our initial studies examining beacon navigation, animals trained with a beacon (and
potentially movement/route memory information as they navigated toward the beacon)
did not seem to use their memory of the path to the beacon when the beacon was removed
during testing (Crook, Hanlon & Basil, 2009). It seemed that route information was not
46 Jennifer Basil and Robyn Crook

available to the animals in the absence of visual cues signaling the goal. However, with
subsequent training in more complex cue arrays, including a beacon, animals did find
the escape point when the landmarks and beacon were removed during testing (Crook,
2008; Crook & Basil, 2013). We inferred that perhaps route/movement information is
learned at the same time as visual cues and beacons, but at a slower rate requiring
more experience. Beacon learning can be considered a form of associative learning,
with the beacon signaling the reward of deep water. The path taken, too, results in the
reward. Models of associative learning propose that the total expectation of a reward
when learning is determined by the associative strengths of all the stimuli present
during training (elemental models: Brandon & Wagner, 1998; Rescorla & Wagner,
1972; configurational models: Pearce, 1987, 1994) and not just the temporal relationship
between the stimuli and reward. When an animal is learning more than one stimulus,
for instance a beacon and the path to the beacon, the associative strengths of each of
the components combine to equal a proportion of what is learned entirely (compound
stimulus, Brandon & Wagner, 1998; Pearce, 1987, 1994; Rescorla & Wagner, 1972;
Thein, Westbrook & Harris, 2008). If an animal is first trained to a compound stimulus
and then tested with the component stimuli on their own, they should perform less
well than with the stimuli in combination during testing (overshadowing, Rescorla &
Wagner, 1972). In essence, the two sources of information being learned share a finite
amount of memory space and each on its own only supports a portion of the information
available during testing. An alternative would be that the two sources of information
do not compete for memory space, but are stored and accessed separately, which has
been demonstrated in vertebrates (e.g. rats: Cheng, Shettleworth, Huttenlocher & Rieser,
2007; Gibson & Shettleworth, 2005; Shettleworth & Sutton, 2005). If this is so, animals
trained with a compound of beacon and route/movement information would perform less
well in tests without a beacon than those trained only with dead reckoning and no beacon
as a solution. Our experiments on the timing and nature of associations guide studies of
the neural architecture and cellular properties underlying learning in nautiluses.
In coleoid cephalopods, the use of visual cues in various learning tasks is well doc-
umented (reviewed in Mather, 2008) and unsurprising given the complex structure of
their eye and reliance on visual input in hunting, communication and defense. However,
nautiluses have simple eyes that are probably not capable of forming distinct images
(Muntz & Raj, 1984; Muntz, 2010b) and an optic lobe that is small compared with
its relatives (Young, 1965a, 1988). We were, therefore, surprised that they used visual
cues to orient and navigate in our maze studies and that they could dynamically switch
among navigation tactics to find a goal. Thus, in spatial ability it appears that both the
nautiloids and coleoids share a foundational set of abilities that may be conserved from
their common ancestor, or may reflect that both lineages had to solve similar problems.

2.6 Evolution

The divergent modern lifestyles of nautiloids and coleoids are thought to have arisen
during a period of intense competition with bony fishes before and during the
 Learning and memory in Chambered Nautilus 47

Figure 2.8 Timeline for cephalopod divergences and significant evolutionary event.
(a) Divergences; (b) timeline with significant events in cephalopod evolution above the line and
geological and speciation events below the line. MYA = millions of years ago. Note the
mass-extinction events and periods of rapid diversification thereafter. (Figure reproduced from
Grasso & Basil, 2009, with permission from S. Karger AG, Medical and Scientific Publishers,
Basel.)

Cretaceous, canalizing these two lineages into entirely different niches: “live fast, die
young” (coleoids) versus “live slow, die old” (nautilids) (Aronson, 1991; Chamberlain,
1990, 1993; Grasso & Basil, 2009; Landman & Cochran, 2010; Packard, 1972) The
cephalopod fossil record (Clarke & Trueman, 1988; Dzik, 1981; Strugnell, Jackson,
Drummond & Cooper, 2006; Strugnell, Norman, Jackson, Drummond & Cooper 2005;
Teichert, Clarke & Trueman, 1988) supports the idea that the ancestors of coleoids sep-
arated about 380 MYA, radiating into three orders and hundreds of species (Figure 2.8;
Bonnaud, Lu & Boucher-Rodoni, 2006; Carlini, Young & Vecchione, 2001; Teichert,
Clarke & Trueman, 1988; Teichert & Matsumoto, 2010; Young & Vecchione, 1996;
48 Jennifer Basil and Robyn Crook

Young, Vecchione & Donovan, 1998). The coleoid expansion in direct competition with
marine vertebrates is thought to be responsible for the evolution of their vertebrate-like
eye, their speed and their complex behavior (Hochner, Shomrat & Fiorito, 2006; Nixon &
Young, 2003; Packard, 1972). Coleoids are believed to have developed their fast-and-
flexible cognitive abilities as a means to hide and escape from their fast, visual preda-
tors. Nautiloids likely remained in deep-water niches that avoided direct competition
with vertebrates and adopted a slow-moving, scavenging lifestyle (Jurassic/Cretaceous;
Chamberlain, 1993; Packard, 1972). Today, they are represented by only two genera, Nau-
tilus and Allonautilus (Bonnaud, Ozouf-Costaz & Boucher-Rodoni, 2004; Saunders &
Landman, 1987; Ward & Saunders, 1997). It has long been assumed that their slower
lifestyle would not support much in the way of learning and memory. However, they are
long lived and while they occupy a different niche from coleoids, their niche is quite
complex, and they too must find food and avoid predation without the predatory (speed,
poison) and defensive tools (active crypsis and ink release) of the coleoids. Learning and
returning to reliable foraging and hiding places along the coral reef is likely adaptive
to long-lived nautiluses. We argue based on the learning and memory capabilities of
Chambered Nautilus and their ancestral state, that perhaps the brain that entered this
period of competition was already large and capable of complex behaviors (Grasso &
Basil, 2009) having coped with, and been refined, during periods of rapid diversification
and competition following numerous mass extinctions (Figure 2.8).

2.7 Conclusions

The study of cognitive abilities in Nautilus is valuable because of their close phylogenetic
relationship to the most neuroanatomically complex invertebrates, their unique ecolog-
ical niche and their own brain organization, which despite lacking the known learning
centers of coleoids, is more complex than those of other mollusks (for which much is
known of their mechanisms of learning and memory). Their longevity, natural complex
environment, unique foraging range and susceptibility to predation (for a review, see
Crook & Basil, 2008b; Hanlon & Messenger, 1996) are likely to promote learning and
memory in nautiluses. While the cephalopods are a perfect group for a comparative study
of adaptive constraints promoting cognitive behavior and complex brains, a complete
picture is not possible without a thorough understanding of the complex behavior of
nautiluses – providing insight into historical and current evolutionary pressures shaping
behavioral and brain complexity in all cephalopods and in other lineages with a heavy
investment in brains.

Acknowledgements

We thank the members of the Laboratory for Invertebrate Behavior and Ecology (LIBE)
for their invaluable intellectual input and for animal care. The Aquatic Research and
 Learning and memory in Chambered Nautilus 49

Environmental Assessment Center (AREAC) provided top-knotch facilities for our ani-
mals. Robert Dickie was instrumental in designing our recirculating systems. Thank
you to the Biomimetic and Cognitive Robotics laboratory (BCR) for help with sea-water
production and animal care. Doctors Binyamin Hochner and Roger Hanlon supported
recent advances in the understanding of Nautilus brains. We appreciate our extensive
discussions on this subject with Dr. Michael Kuba. The opportunity and time to write
this chapter were generously supplied by a sabbatical (J. B., Brooklyn College) in the
laboratory of Dr. Binyamin Hochner at the Hebrew University in Jerusalem.

References

Agin, V., Chichery, R. and Chichery, M.-P. (2001). Effects of learning on cytochrome oxidase
activity in cuttlefish brain. Neuroreport, 12: 1–4.
Agin, V., Chichery, R., Dickel, L. and Chichery, M.-P. (2006). The “prawn-in-the-tube” procedure
in the cuttlefish: Habituation or passive avoidance learning? Learning and Memory, 13: 97–
101.
Agin, V., Chichery, R., Maubert, E. and Chichery, M.-P. (2003). Time-dependent effects of cyclo-
heximide on long-term memory in the cuttlefish. Pharmacology Biochemistry and Behavior,
75: 141–146.
Agin, V., Dickel, L., Chichery, R. and Chichery, M.-P. (1998). Evidence for a specific short-term
memory in the cuttlefish, Sepia. Behavioural Processes, 43: 329–334.
Agin, V., Poirier, R., Chichery, R., Dickel, L. and Chichery, M.-P. (2006). Developmental study
of multiple memory stages in the cuttlefish, Sepia officinalis. Neurobiology of Learning and
Memory, 86: 264–269.
Alkon, D. L., Ikeno, H., Dworkin, J., McPhie, D. L., Olds, J. L., Lederhendler, I., Matzel, L.,
Schreurs, B. G., Kuzirian, A. and Collin, C. (1990). Contraction of neuronal branching volume:
An anatomic correlate of Pavlovian conditioning. Proceedings of the National Academy of
Sciences, 87: 1611–1614.
Alves, C., Boal, J. G. and Dickel, L. (2008). Short-distance navigation in cephalopods: A review
and synthesis. Cognitive Processing, 9: 239–247.
Alves, C., Chichery, R., Boal, J. G. and Dickel, L. (2007). Orientation in the cuttlefish Sepia
officinalis: Response versus place learning. Animal Cognition, 10: 29–36.
Aronson, R. B. (1991). Ecology, paleobiology and evolutionary constraint in the octopus. Bulletin
of Marine Science, 49: 1–2.
Barber, V. C. and Wright, D. E. (1969). The fine structure of the sense organs of the cephalopod
mollusc Nautilus. Cell and Tissue Research, 102: 293–312.
Basil, J. A., Bahctinova, I., Kuroiwa, K., Lee, N., Mims, D., Preis, M. and Soucier, C. (2005). The
function of the rhinophore and the tentacles of Nautilus pompilius L. (Cephalopoda, Nautiloidea)
in orientation to odor. Marine and Freshwater Behaviour and Physiology, 38: 209–221.
Basil, J. A., Hanlon, R. T., Sheikh, S. I. and Atema, J. (2000). Three-dimensional odor tracking
by Nautilus pompilius. Journal of Experimental Biology, 203: 1409–1414.
Basil, J. A., Lazenby, G. B., Nakanuku, L. and Hanlon, R. T. (2002). Female nautilus are attracted
to male conspecific odor. Bulletin of Marine Science, 70: 217–225.
Basil, J. A. and Sandeman, D. (2000). Crayfish (Cherax destructor) use tactile cues to detect and
learn topographical changes in their environment. Ethology, 106: 247–259.
50 Jennifer Basil and Robyn Crook

Baxter, D. A. and Byrne, J. H. (2006). Feeding behavior of Aplysia: A model system for comparing
cellular mechanisms of classical and operant conditioning. Learning and Memory, 13: 669–80.
Benjamin, P. R. (2000). A systems approach to the cellular analysis of associative learning in the
pond snail Lymnaea. Learning and Memory, 7: 124–131.
Bidder, A. M. (1962). Use of the tentacles, swimming and buoyancy control in the pearly nautilus.
Nature, 196: 451–454.
Bitterman, M. E., Menzel, R., Fietz, A. and Schäfer, S. (1983). Classical conditioning of proboscis
extension in honeybees (Apis mellifera). Journal of Comparative Psychology, 97: 107–119.
Blackwell, K. T. (2006). Subcellular, cellular, and circuit mechanisms underlying classical condi-
tioning in Hermissenda crassicornis. Anatomical Record. Part B, New Anatomist, 289: 25–37.
Boal, J. G. (1991). Complex learning in Octopus bimaculoides. American Malacological Bulletin,
9: 75–80.
Boal, J. G. (1996). A review of simultaneous visual discrimination as a method of training
octopuses. Biological Reviews, 71: 157–190.
Boal, J. G., Dunham, A. W., Williams, K. T. and Hanlon, R. T. (2000). Experimental evidence
for spatial learning in octopuses (Octopus bimaculoides). Journal of Comparative Psychology,
114: 246–252.
Bonnaud, L., Lu, C. C. and Boucher-Rodoni, R. (2006). Morphological character evolution and
molecular trees in sepiids (Mollusca: Cephalopoda): Is the cuttlebone a robust phylogenetic
marker? Biological Journal of the Linnean Society, 89: 139–150.
Bonnaud, L., Ozouf-Costaz, C. and Boucher-Rodoni, R. (2004). A molecular and karyological
approach to the taxonomy of Nautilus. Comptes Rendus de Biologie, 327: 133–138.
Boycott, B. B. (1961). The functional organization of the brain of the cuttlefish Sepia officinalis.
Proceedings of the Royal Society. B: Biological Sciences, 153: 503–534.
Boycott, B. B. and Young, J. Z. (1955). A memory system in Octopus vulgaris Lamarck. Proceed-
ings of the Royal Society. B: Biological Sciences, 143: 449–480.
Boyle, P. R. (1986). Neural control of cephalopod behavior. The Mollusca, 9: 1–97.
Brandon, S. E. and Wagner, A. R. (1998). Occasion setting: Influences of conditioned emotional
responses and configured ural cues. In Occasion setting: Associative learning and cognition
in animals. Schmajuk, N. A. and Holland, P. C. (Eds), pp. 343–382. Washington, DC, USA:
American Psychological Association.
Brembs, B., Baxter, D. A. and Byrne, J. H. (2004). Extending in vitro conditioning in Aplysia to
analyze operant and classical processes in the same preparation. Learning and Memory, 11:
412–20.
Budelmann, B. U. (1995). Cephalopod sense organs, nerves and the brain: Adaptations for high
performance and life style. Marine and Freshwater Behaviour and Physiology, 25: 13–33.
Burrell, B. D. and Sahley, C. L. (2001). Learning in simple systems. Current Opinion in Neuro-
biology, 11: 757–764.
Carew, T. J. and Sahley, C. L. (1986). Invertebrate learning and memory: From behavior to
molecules. Annual Review of Neuroscience, 9: 435–487.
Carlini, D. B., Young, R. E. and Vecchione, M. (2001). A molecular phylogeny of the Octopoda
(Mollusca: Cephalopoda) evaluated in light of morphological evidence. Molecular Phylogenet-
ics and Evolution, 21: 388–397.
Carlson, B. A., McKibben, J. N. and De Gruy, M. V. (1984). Telemetric investigation of vertical
migration of Nautilus belauensis in Palau. Pacific Science, 38: 183–188.
Cartron, L., Darmaillacq, A. S. and Dickel, L. (2013). The “prawn-in-the-tube” procedure: What
do cuttlefish learn and memorize? Behavioural Brain Research, 240: 29–32.
 Learning and memory in Chambered Nautilus 51

Chamberlain, J. A. (1990). Jet propulsion of Nautilus: A surviving example of early paleozoic

cephalopod locomotor design. Canadian Journal of Zoology, 68: 806–814.
Chamberlain, J. A. (1993). Locomotion in ancient seas: Constraint and opportunity in cephalopod
adaptive design. Geobios, 26: 49–61.
Chandra, S. B. C., Hosler, J. S. and Smith, B. H. (2000). Heritable variation for latent inhibition
and its correlation with reversal learning in honeybees (Apis mellifera). Journal of Comparative
Psychology, 114: 86–97.
Cheng, K., Shettleworth, S. J., Huttenlocher, J. and Rieser, J. J. (2007). Bayesian integration of
spatial information. Psychological Bulletin, 133: 625–637.
Clarke, M. R. and Trueman, E. R. (1988). Evolution of recent cephalopods. A brief review. The
Mollusca, 12: 11–79.
Cole, P. D. and Adamo, S. A. (2005). Cuttlefish (Sepia officinalis: Cephalopoda) hunting behavior
and associative learning. Animal Cognition, 8: 27–30.
Collett, T. S., Dillmann, E., Giger, A. and Wehner, R. (1992). Visual landmarks and route following
in desert ants. Journal of Comparative Physiology A, 170: 435–442.
Crook, R. J. (2008). Behavioral correlates of learning and memory in Chambered Nautilus,
Nautilus pompilius. New York, NY, USA: City University of New York.
Crook, R. J. and Basil, J. A. (2008a). A biphasic memory curve in the chambered nautilus, Nautilus
pompilius L. (Cephalopoda: Nautiloidea). Journal of Experimental Biology, 211: 1992–1998.
Crook, R. J. and Basil, J. A. (2008b). A role for nautilus in studies of the evolution of brain and
behavior. Communicative and Integrative Biology, 1: 18–19.
Crook, R. J. and Basil, J. A. (2013). Flexible spatial orientation and navigational strategies in
chambered Nautilus. Ethology, 119: 77–85.
Crook, R. J., Hanlon, R. T. and Basil, J. A. (2009). Memory of visual and topographical fea-
tures suggests spatial learning in nautilus (Nautilus pompilius L.). Journal of Comparative
Psychology, 123: 264–274.
Crow, T. (2004). Pavlovian conditioning of Hermissenda: Current cellular, molecular, and circuit
perspectives. Learning and Memory, 11: 229–238.
Crow, T. J. and Alkon, D. L. (1978). Retention of an associative behavioral change in Hermissenda.
Science, 201: 1239–1241.
Darmaillacq, A.-S., Chichery, R. and Dickel, L. (2006). Food imprinting, new evidence from the
cuttlefish Sepia officinalis. Biology Letters, 2: 345–347.
Darmaillacq, A.-S., Chichery, R., Shashar, N. and Dickel, L. (2006). Early familiarization over-
rides innate prey preference in newly hatched Sepia officinalis cuttlefish. Animal Behaviour,
71: 511–514.
Darmaillacq, A. S., Dickel, L., Chichery, M.-P., Agin, V. and Chichery, R. (2004). Rapid taste
aversion learning in adult cuttlefish, Sepia officinalis. Animal Behaviour, 68: 1291–1298.
Davis, H. P. and Squire, L. R. (1984). Protein synthesis and memory: A review. Psychological
Bulletin, 96: 518–559.
Davis, R. L. (2005). Olfactory memory formation in Drosophila: From molecular to systems
neuroscience. Annual Review of Neuroscience, 28: 275–302.
Dickel, L., Chichery, M.-P. and Chichery, R. (2001). Increase of learning abilities and matura-
tion of the vertical lobe complex during postembryonic development in the cuttlefish, Sepia.
Developmental Psychobiology, 39: 92–98.
Dunstan, A. J. , Alanis, O. and Marshall, J. (2010). Nautilus pompilius fishing and population
decline in the Philippines: A comparison with an unexploited Australian Nautilus population.
Fisheries Research, 106: 239–247.
52 Jennifer Basil and Robyn Crook

Dunstan, A. J., Ward, P. D. and Marshall, N. J. (2011). Vertical distribution and migration patterns
of Nautilus pompilius. Public Library of Science One, 6: e16311.
Dzik, J. (1981). Origin of the Cephalopoda. Acta Palaeontologica Polonica, 26: 161–191.
Emery, D. G. (1975). The histology and fine structure of the olfactory organ of the squid Lolli-
guncula brevis blainville. Tissue and Cell, 7: 357–367.
Fanselow, M. S. and Poulos, A. M. (2005). The neuroscience of mammalian associative learning.
Annual Review of Psychology, 56: 207–234.
Fiorito, G. and Chichery, R. (1995). Lesions of the vertical lobe impair visual discrimination
learning by observation in Octopus vulgaris. Neuroscience Letters, 192: 117–120.
Fiorito, G. and Scotto, P. (1992). Observational learning in Octopus vulgaris. Science, 256: 545–
547.
Fukuda, Y. (1980). Observations by SEM, Nautilus macromphalus in captivity. Tokyo, Japan:
Tokai University Press.
Gallistel, C. R. (1990). The organization of learning. Cambridge, MA, USA: MIT Press.
Gelperin, A. and Culligan, N. (1984). In vitro expression of in vivo learning by an isolated
molluscan CNS. Brain Research, 304: 207–213.
Gibson, B. M. and Shettleworth, S. J. (2005). Place versus response learning revisited: Tests of
blocking on the radial maze. Behavioral Neuroscience, 119: 567–586.
Gilly, W. F. and Lucero, M. T. (1992). Behavioral responses to chemical stimulation of the olfactory
organ in the squid Loligo opalescens. Journal of Experimental Biology, 162: 209–229.
Graindorge, N., Alves, C., Darmaillacq, A.-S., Chichery, R., Dickel, L. and Bellanger, C. (2006).
Effects of dorsal and ventral vertical lobe electrolytic lesions on spatial learning and locomotor
activity in Sepia officinalis. Behavioral Neuroscience, 120: 1151–1158.
Grasso, F. W. and Basil, J. A. (2009). The evolution of flexible behavioral repertoires in cephalopod
molluscs. Brain, Behavior and Evolution, 74: 231–245.
Graziadei, P. (1964). Receptors in the sucker of the cuttlefish. Nature, 203: 384–386.
Hanlon, R. T. and Messenger, J. B. (1996). Cephalopod behaviour. Cambridge, MA, USA:
Cambridge University Press.
Hochner, B., Shomrat, T. and Fiorito, G. (2006). The octopus: A model for a comparative anal-
ysis of the evolution of learning and memory mechanisms. The Biological Bulletin, 210:
308–317.
Hvorecny, L. M., Grudowski, J. L., Blakeslee, C. J., Simmons, T. L., Roy, P. R., Brooks, J. A.,
Hanner, R. M., Beigel, M. E., Karson, M. A. and Nichols, R. H. (2007). Octopuses (Octopus
bimaculoides) and cuttlefishes (Sepia pharaonis, S. officinalis) can conditionally discriminate.
Animal Cognition, 10: 449–459.
Kamil, A. C. (1994). A synthetic approach to the study of animal intelligence. Behavioural
Mechanisms in Evolutionary Ecology, 11–45.
Kamil, A. C. and Cheng, K. (2001). Way-finding and landmarks: The multiple-bearings hypothesis.
Journal of Experimental Biology, 204: 103–113.
Kamil, A. C. and Jones, J. E. (1997). The seed-storing corvid Clark’s nutcracker learns geometric
relationships among landmarks. Nature, 390: 276–279.
Kandel, E. R., 1976. Cellular basis of behavior: An introduction to behavioral neurobiology. New
York, NY, USA: W. H. Freeman.
Kandel, E. R. and Schwartz, J. H. (1982). Molecular biology of learning: Modulation of transmitter
release. Science, 218: 433–443.
Kandel, E. R. and Tauc, L. (1965). Heterosynaptic facilitation in neurones of the abdominal
ganglion of Aplysia depilans. Journal of Physiology, 181: 1–27.
 Learning and memory in Chambered Nautilus 53

Karson, M. A., Boal, J. G. and Hanlon, R. T. (2003). Experimental evidence for spatial learning
in cuttlefish (Sepia officinalis). Journal of Comparative Psychology, 117: 149–155.
Kier, W. M. (1987). The functional morphology of the tentacle musculature of Nautilus pompilius.
In Nautilus: The biology and paleobiology of a living fossil. Saunders, W. B. and Landman, N.
(Eds), pp. 257–269. New York, NY, USA: Springer.
Landman, N. H. and Cochran, J. K. (2010). Growth and longevity of Nautilus. In Nautilus:
The biology and paleobiology of a living fossil. Saunders, W. B. and Landman, N. (Eds),
pp. 401–420. New York, NY, USA: Springer.
Lucero, M. T., Horrigan, F. T. and Gilly, W. M. F. (1992). Electrical responses to chemical
stimulation of squid olfactory receptor cells. Journal of Experimental Biology, 162: 231–249.
Lukowiak, K., Adatia, N., Krygier, D. and Syed, N. (2000). Operant conditioning in Lymnaea:
Evidence for intermediate-and long-term memory. Learning and Memory, 7: 140–150.
Maddock, L. and Young, J. Z. (1987). Quantitative differences among the brains of cephalopods.
Journal of Zoology, 212: 739–767.
Maldonado, H. (1965). The positive and negative learning process in Octopus vulgaris Lamarck.
Influence of the vertical and median superior frontal lobes. Journal of Comparative Physiology
A, 51: 185–203.
Mather, J. A. (1991). Navigation by spatial memory and use of visual landmarks in octopuses.
Journal of Comparative Physiology A, 168: 491–497.
Mather, J. A. (1995). Cognition in cephalopods. Advances in the Study of Behavior, 24:
317–353.
Mather, J. A. (2008). To boldly go where no mollusc has gone before: Personality, play, thinking,
and consciousness in cephalopods. American Malacological Bulletin, 24: 51–58.
McGaugh, J. L. (2000). Memory – a century of consolidation. Science, 287: 248–251.
Menzel, R., Leboulle, G. and Eisenhardt, D. (2006). Small brains, bright minds. Cell, 124: 237–
239.
Messenger, J. B. (1973). Learning in the cuttlefish, Sepia. Animal Behaviour, 21: 801–826.
Messenger, J. B. (1991). Photoreception and vision in molluscs. Vision and Visual Dysfunction,
2: 364–397.
Milner, B., Squire, L. R. and Kandel, E. R. (1998). Cognitive neuroscience and the study of
memory. Neuron, 20: 445–468.
Muntz, W. R. A. (1986). Short communications: The spectral sensitivity of Nautilus pompilius.
Journal of Experimental Biology, 126: 513–517.
Muntz, W. R. A. (2010a). A possible function of the iris groove of Nautilus. In Nautilus: The
biology and paleobiology of a living fossil, 2nd edn. Saunders, W. B. and Landman, N. (Eds),
pp. 245–247. New York, NY, USA: Springer.
Muntz, W. R. A. (2010b). Visual behavior and visual sensitivity of Nautilus pompilius. In Nautilus:
The biology and paleobiology of a living fossil, 2nd edn. Saunders, W. B. and Landman, N.
(Eds), pp. 231–244. New York, NY, USA: Springer.
Muntz, W. R. A. and Raj, U. (1984). On the visual system of Nautilus pompilius. Journal of
Experimental Biology, 109: 253–263.
Nixon, M. and Young, J. Z. (2003). The brains and lives of cephalopods. Oxford, UK: Oxford
University Press.
O’Keefe, J. and Nadel, L. (1978). The hippocampus as a cognitive map. New York, NY, USA:
Oxford University Press.
Packard, A. (1972). Cephalopods and fish: The limits of convergence. Biological Reviews, 47:
241–307.
54 Jennifer Basil and Robyn Crook

Pavlov, I. (1927). Conditioned reflexes: An investigation of the physiological activity of the cerebral
cortex. Oxford, UK: Oxford University Press.
Pearce, J. M. (1987). A model for stimulus generalization in Pavlovian conditioning. Psychological
Review, 94: 61–73.
Pearce, J. M. (1994). Similarity and discrimination: A selective review and a connectionist model.
Psychological Review, 101: 587–607.
Pedreira, M. E. and Maldonado, H. (2003). Protein synthesis subserves reconsolidation or extinc-
tion depending on reminder duration. Neuron, 38: 863–869.
Poucet, B., Chapuis, N., Durup, M. and Thinus-Blanc, C. (1986). A study of exploratory behavior
as an index of spatial knowledge in hamsters. Learning and Behavior, 14: 93–100.
Rescorla, R. A. and Wagner, A. R. (1972). A theory of Pavlovian conditioning: Variations in
the effectiveness of reinforcement and nonreinforcement. Classical Conditioning II: Current
Research and Theory, 2: 64–99.
Ruth, P., Schmidtberg, H., Westermann, B. and Schipp, R. (2002). The sensory epithelium of the
tentacles and the rhinophore of Nautilus pompilius L. (Cephalopoda, Nautiloidea). Journal of
Morphology, 251: 239–255.
Sanders, G. D. (1970). Long-term memory of a tactile discrimination in Octopus vulgaris and the
effect of vertical lobe removal. Brain Research, 20: 59–73.
Sangha, S., Scheibenstock, A., McComb, C. and Lukowiak, K. (2003). Intermediate and long-term
memories of associative learning are differentially affected by transcription versus translation
blockers in Lymnaea. Journal of Experimental Biology, 206: 1605–1613.
Sasaki, T., Shigeno, S. and Tanabe, K. (2010). Anatomy of living Nautilus: Re-evaluation of
primitiveness and comparison with Coleoidea. In Cephalopods – present and past. Tanabe, K.,
Shigeta, Y., Sasaki, T. and Hirano, H. (Eds), pp. 35–66. Tokyo, Japan: Tokai University Press.
Saunders, W. B. (1985). Studies of living Nautilus in Palau. National Geographic Society Reseach
Reports, 18: 669–682.
Saunders, W. B. and Landman, N. H. (1987). Nautilus: The biology and paleobiology of a living
fossil, reprint with additions. New York, NY, USA: Springer.
Shettleworth, S. J. and Sutton, J. E. (2005). Multiple systems for spatial learning: Dead reckoning
and beacon homing in rats. Journal of Experimental Psychology: Animal Behavior Processes,
31: 125–141.
Shigeno, S., Sasaki, T., Moritaki, T., Kasugai, T., Vecchione, M. and Agata, K. (2008). Evolution
of the cephalopod head complex by assembly of multiple molluscan body parts: Evidence from
Nautilus embryonic development. Journal of Morphology, 269: 1–17.
Shigeno, S., Takenori, S. and Boletzky, S. v. (2010). The origins of cephalopod body plans:
A geometrical and developmental basis for the evolution of vertebrate-like organ systems.
Classical Conditioning II: Current Research and Theory, 1: 23–34.
Shillito, E. E. (1963). Exploratory behaviour in the short-tailed vole Microtus agrestis. Behaviour,
21: 145–154.
Smith, B. H. (1997). An analysis of blocking in odorant mixtures: An increase but not a decrease
in intensity of reinforcement produces unblocking. Behavioral Neuroscience, 11: 57–69.
Smith, B. H. and Cobey, S. (1994). The olfactory memory of the honeybee Apis mellifera.
II. Blocking between odorants in binary mixtures. Journal of Experimental Biology, 195:
91–108.
Soucier, C. P. and Basil, J. A. (2008). Chambered Nautilus (Nautilus pompilius pompilius) responds
to underwater vibrations. American Malacological Bulletin, 24: 3–11.
Stork, O. and Welzl, H. (1999). Memory formation and the regulation of gene expression. Cellular
and Molecular Life Sciences, 55: 575–592.
 Learning and memory in Chambered Nautilus 55

Strugnell, J., Jackson, J., Drummond, A. J. and Cooper, A. (2006). Divergence time estimates for
major cephalopod groups: Evidence from multiple genes. Cladistics, 22: 89–96.
Strugnell, J., Norman, M., Jackson, J., Drummond, A. J. and Cooper, A. (2005). Molecular phy-
logeny of coleoid cephalopods (Mollusca: Cephalopoda) using a multigene approach; the effect
of data partitioning on resolving phylogenies in a Bayesian framework. Molecular Phylogenetics
and Evolution, 37: 426–441.
Suboski, M. D., Muir, D. and Hall, D. (1993). Social learning in invertebrates. Science, 259:
1628–1629.
Sutherland, N. S. (1963). The shape-discrimination of stationary shapes by octopuses. American
Journal of Psychology, 76: 177–190.
Teichert, C., Clarke, M. R. and Trueman, E. R. (1988). Main features of cephalopod evolution.
In The Mollusca, Clarke, M. R. and Trueman, E. R. (Eds), pp. 11–79. San Diego, CA, USA:
Academic Press.
Teichert, C. and Matsumoto, T. (2010). The ancestry of the genus Nautilus. In Nautilus: The
biology and paleobiology of a living fossil, 2nd edn. Saunders, W. B. and Landman, N. (Eds),
pp. 25–32. New York, NY, USA: Springer.
Teyke, T. (1985). Collision with and avoidance of obstacles by blind cave fish Anoptichthys jordani
(Characidae). Journal of Comparative Physiology A, 157: 837–843.
Teyke, T. (1988). Flow field, swimming velocity and boundary layer: Parameters which affect the
stimulus for the lateral line organ in blind fish. Journal of Comparative Physiology A, 163:
53–61.
Teyke, T. (1989). Learning and remembering the environment in the blind cave fish Anoptichthys
jordani. Journal of Comparative Physiology A, 164, 655–662.
Thein, T., Westbrook, R. F. and Harris, J. A. (2008). How the associative strengths of stimuli
combine in compound: Summation and overshadowing. Journal of Experimental Psychology:
Animal Behavior Processes, 34: 155–166.
Tinbergen, N. (1963). On aims and methods of ethology. Zeitschrift für Tierpsychologie, 20:
410–433.
Walters, E. T. (1991). A functional, cellular, and evolutionary model of nociceptive plasticity in
Aplysia. The Biological Bulletin, 180: 241–251.
Walters, E. T., Alizadeh, H. and Castro, G. A. (1991). Similar neuronal alterations induced by
axonal injury and learning in Aplysia. Science, 253: 797–799.
Ward, P. D., Carlson, B., Weekly, M. and Brumbaugh, B. (1984). Remote telemetry of daily vertical
and horizontal movement of Nautilus in Palau. Nature, 309: 248–250.
Ward, P. D. and Saunders, W. B. (1997). Allonautilus: A new genus of living nautiloid cephalopod
and its bearing on phylogeny of the Nautilida. Journal of Paleontology, 71: 1054–1064.
Welker, W. I. and Welker, J. (1958). Reaction of fish (Eucinostomus gula) to environmental
changes. Ecology, 39: 283–288.
Wells, M. J. (1964). Detour experiments with octopuses. Journal of Experimental Biology, 41:
621–642.
Wells, M. J. (1978). Octopus: Physiology and behaviour of an advanced invertebrate. London,
UK: Chapman and Hall.
Westermann, B. and Beuerlein, K. (2005). Y-maze experiments on the chemotactic behaviour of
the tetrabranchiate cephalopod. Marine Biology, 147: 145–151.
Woodhams, P. L. and Messenger, J. B. (1974). A note on the ultrastructure of the Octopus olfactory
organ. Cell and Tissue Research, 152: 253–258.
Young, J. Z. (1960). The failures of discrimination learning following the removal of the vertical
lobes in Octopus. Proceedings of the Royal Society. B: Biological Sciences, 153: 18–46.
56 Jennifer Basil and Robyn Crook

Young, J. Z. (1961). Learning and discrimination in the octopus. Biological Reviews, 36: 32–95.
Young, J. Z. (1964). A model of the brain. Oxford, UK: Clarendon Press.
Young, J. Z. (1965a). The central nervous system of Nautilus. Philosophical Transactions of the
Royal Society of London. Series B, 249: 1–25.
Young, J. Z. (1965b). The Croonian lecture, 1965: The organization of a memory system. Pro-
ceedings of the Royal Society. B: Biological Sciences, 163: 285–320.
Young, J. Z. (1988). Evolution of the cephalopod brain. The Mollusca, 12: 215–228.
Young, J. Z. (1991). Computation in the learning system of cephalopods. The Biological Bulletin,
180: 200–208.
Young, J. Z. and Boycott, B. B. (1971). The anatomy of the nervous system of Octopus vulgaris.
Oxford, UK: Clarendon Press.
Young, R. E. and Vecchione, M. (1996). Analysis of morphology to determine primary sister-taxon
relationships within coleoid cephalopods. American Malacological Bulletin, 12: 91–112.
Young, R. E., Vecchione, M. and Donovan, D. T. (1998). The evolution of coleoid cephalopods and
their present biodiversity and ecology. South African Journal of Marine Science, 20: 393–420.
3 Learning from play in octopus
Michael J. Kuba, Tamar Gutnick and Gordon M. Burghardt

3.1 Introduction

The brain of cephalopods rivals that of the vertebrates in relative size, being as large as
or larger than the brains of many fish, although smaller than those of birds and mammals
(Hochner, 2008). The elaborate sensory and neural system in many extant cephalopods
enables them to exhibit complex types of adaptive behaviour (Wells, 1978; Hanlon &
Messenger, 1996; Nixon & Young, 2003; Hochner, 2008, 2010). Of the many kinds of
such complex behaviour, we focus here on exploration, play and cognition.
To adequately study exploration, or playful interaction, in cephalopods, we need to
know a good deal about their behaviour, especially ‘normal’ or ethotypic behaviour, and
have sufficient time and opportunities to observe changes to their behaviour across time
and settings. This requires keeping animals in good condition in captivity or observing
them in the wild in natural environments. This being said, most cephalopods live in the
open waters of the deep sea and the majority are unavailable for systematic observation.
Furthermore, most cephalopod species cannot be maintained regularly in laboratory
conditions. For these reasons, most studies of the behaviour of individual cephalopods
come from work on only three genera, Octopus, Sepia and Loligo, animals that live in
coastal waters. These, and a few of their close relatives, are the only cephalopods that have
been kept at all regularly in aquaria; Octopus, Sepia and, to some extent, Sepiotheuthis
alone do sufficiently well under these conditions for their observed behaviour to be
considered typical and natural (Wells, 1962; Sanders, 1975; Wells, 1978; Moynihan &
Rodaniche, 1982; Hanlon & Messenger, 1996; Nixon & Young, 2003; Mather, Griebel
& Byrne, 2010). Due to this lack of data on basic behaviours of these animals, it is often
difficult to argue which observed playful behaviours might be considered exaggerations
or modifications of normal behaviours. To date we are aware of only four papers, on two
different species of octopus, dealing with exploration and play (Mather & Anderson,
1999; Kuba, Meisel, Byrne, Griebel & Mather, 2003; Kuba, Byrne, Meisel & Mather,
2006a, 2006b). In this chapter, therefore, we will focus on octopuses, probably the most
studied group of all cephalopods.
Octopuses have fascinated observers for thousands of years and there is a wealth
of anecdotal reports of their intelligence, cunning and curiosity. They are generalist

Cephalopod Cognition, eds A.-S. Darmaillacq, L. Dickel and J. Mather. Published by Cambridge University
Press. 
C Cambridge 2014.
58 Michael J. Kuba, Tamar Gutnick and Gordon M. Burghardt

predators and detect their prey both visually and by touch (Hanlon & Messenger, 1996).
The loss of the external molluscan shell has been linked to their high behavioural
plasticity for predator avoidance, an increase in brain size and the evolution of more
effective sense organs, resulting in higher cognitive abilities (Packard, 1972; Hanlon &
Messenger, 1996). The aim of the studies on octopus exploration and play is to describe
the behavioural processes involved and to untangle the interplay between exploration,
habituation and exploratory play in these animals. These studies show, for the first time,
that an invertebrate does engage in playful interactions with objects (Burghardt, 2005).

3.2 Exploration

Exploration, in its most basic manifestation, can be defined as the acquisition of knowl-
edge about animate or innate objects, the environment and its changes through sensory
information gathering (Berlyne, 1966; Renner, 1990; Power, 2000). Exploration and
sampling the surrounding environment is one of the most basic tasks each and every
organism faces. It usually occurs when an animal is first exposed to an object or an
environment. Yet, in spite of the above, defining and recognizing exploration is not
easy. For example, exploration is not the only response to novel settings; it is affected
by motivational and other internal states, and has often been conceptually dependent
on recording easily measured changes in general behavioural activity (e.g. increased
locomotion, alternation of search patterns or general contact with novel objects Renner,
1990). Renner (1990) criticized the fact that exploration was often treated as the ani-
mals’ equivalent to Brownian random motion in molecules. Although early-learning
theorists historically could not explain exploration (Burghardt, 2012), because it was not
always associated with an obvious, primary biological need, evidence shows that there
is a functional significance of exploration (Archer & Birke, 1983; Renner, 1990). The
extraction of information from the surrounding environment (Hutt, 1966) can result in a
learning process (Baldwin & Baldwin, 1997), which would allow for the establishment
of a familiar area (territory or home range, see Russell, 1983) and the ability to monitor
changes in it. For animals in the wild this means that the maintenance of familiarity with
the environment necessitates regular inspections – i.e. exploration (Russell, 1983).
Exploration has been divided into two different types: extrinsic and intrinsic. Extrinsic
exploration is exploration for obtaining information about a conventional reinforcer
(Russell, 1983). Examples of this kind of motivated exploration include foraging for food
when hungry or searching for conspecifics at times of reproduction (Toates, 1983). Due
to the seemingly obvious motivational cause for this type of exploration, it has received
little scientific attention. Laboratory research on birds and mammals has focused on
intrinsic exploration (Archer & Birke, 1983; Huber, Rechberger & Taborsky, 2001;
Gazzaniga & Heatherton, 2003). Intrinsic exploration is defined as a behaviour directed
towards a stimulus of little biological importance (Russell, 1983). Early studies on
intrinsic exploration in rats (see Toates, 1983) and later work on farm animals (Wood-
Gush, Stolba & Miller, 1983; Day, Kyriazakis & Lawrence, 1994) focused on the effect
of hunger or punishment on exploration. Most rats did not reduce or cease exploration
 Learning from play in octopus 59

when food deprived, and would even delay eating and endure physical stress for an
opportunity to explore a Dashiell multi-arm maze filled with novel objects (Berlyne,
1960).
Captive aquatic animals typically face a lack of environmental variation and a rather
stimulus-deprived setting; thus, novel, non-threatening stimuli may evoke exploration in
intact animals of many phyla (see Archer & Birke, 1983). Therefore, stimulus seeking to
avoid boredom is an important factor modifying intrinsic as well as extrinsic exploration
(Toates, 1983). Early on, researchers (e.g. Craig, 1918; Harlow, 1953) looked into the
motivational states behind exploratory behaviour and whether external stimuli or internal
drives are more important for the manifestation of exploratory behaviours. Harlow (1950)
showed that monkeys would try to solve a three-piece puzzle apparatus even if there was
no extrinsic reward present.
Recent papers, influenced by learning theory (see Toates, 1983 for a review on theories
affecting exploration), focus on the general linkage between learning and exploration
(Toates, 1983; Renner, 1990; Renner & Seltzer, 1994; Heinrich, 1995). Learning during
exploration has the potential to benefit the animal by reducing risk factors, such as
susceptibility to predation and increasing the ability to incur positive outcomes like
finding food. However, in order to study learning and memory through exploration
the focus has to be on the form and function of the behaviour, rather than the more
obvious performance methods used in classical learning studies. Octopuses, with their
excellent learning abilities (Wells, 1978; Nixon & Young, 2003), foraging behaviour and
semi-permanent home ranges (Mather, 1991a, 1991b) provide an appropriate subject for
characterizing the behaviour in an invertebrate.
Kuba, Byrne, Meisel and Mather (2006b) divided the responses to an object dropped
into the tank of a healthy octopus, into two levels of exploratory behaviour. At level 1
the animal used one or several arms to explore the object; at level 2 the animal brought
the object to the mouth area under its interbrachial web (Figure 3.1a,b: levels 1 and 2
with objects). With non-food objects there were no significant differences in the number
of level 1 or level 2 interactions; with food objects there were significantly more level
2 interactions. Food was essentially either eaten or ignored, resulting in more level 2
and lengthier interactions, and non-food items were explored with three to four times
more contacts but with shorter interactions. When the same animals were tested 2 hours
after feeding, the number of interactions with food items was significantly lower, but the
number of interactions with non-food items remained the same (Figure 3.2a,b).
The results showed that the interaction with non-food items was not misplaced pre-
dation by an animal that cannot tell the difference between food and non-food items.
Animals specifically chose to approach and manipulate the non-food item, and this
intrinsic exploration did not change significantly under different feeding regimes. These
results were consistent with those found in studies on intrinsic exploration in rats (see
Toates, 1983 for a review), farm animals (Wood-Gush, Stolba & Miller, 1983; Day,
Kyriazakis & Lawrence, 1994) and monkeys (Harlow, 1950).
How do other cephalopods fit into this picture? Despite the lack of laboratory studies,
it seems certain that squids and cuttlefish as well as nautilus should show exploratory
behaviour. There is evidence that spatial exploration in Sepia is based on visual clues
 (a) (b)

Figure 3.1 (a) Level 1 exploratory behaviour; the animal uses one or several arms to explore the
non-food item. (b) Level 2 exploratory behaviour; the animal holds the objects in its interbrachial
web. (From Kuba, Byrne, Meisel & Mather, 2006b.)

(a)

(b)

Figure 3.2 Difference in number and duration of contacts for food (a) and non-food (b) items.
Black bars represent values 24 hours after feeding (hungry) and dark grey bars represent values
2 hours after feeding (satiated). (From Kuba, Byrne, Meisel & Mather, 2006b.)
 Learning from play in octopus 61

(see, e.g. Alves, Chichery, Boal & Dickel, 2007). Nautilus, however, with a predom-
inantly olfactory-based sensory world, would be an interesting future challenge for
testing exploration in an evolutionary-old organism (Grasso & Basil, 2009). Despite
their short lifespan, many cephalopods develop complex sensory and motor abilities and
behaviours. This could be exploited to understanding the adaptiveness of behavioural and
cognitive traits and the role of experience. Theoretically, this would enable researchers to
conduct longitudinal studies throughout the lifespan of an individual to get an in-depth
look at the change/progression of exploratory behaviour in that individual. An especially
promising area could be the investigation of motivation for exploration, from looking
for food and suitable home ranges in juveniles to looking for mates in adults.

3.3 Play

‘Who would dare study play?’ These were the first words in the introduction to a book on
play and its evolution edited by Bruner, Jolly and Sylva (1976). In an effort to formalize
the study of play, Fagan (1981) described three different types of play: locomotor-
rotational play (typically solitary), object play and social play. However, these are not
absolutely separate and all can occur together in spite of having, in some species, different
developmental trajectories. Locomotor-rotational play consists of movements matching
the traditional criteria for play activity (Byers & Walker, 1995) and includes playful
running and twisting in ungulates to somersaulting in monkeys (Müller-Schwarze, 1984;
Sommer & Mendoza-Granados, 1995). Object play, which may be either solitary or
social, involves manipulating an object. This type of play has been linked to exploratory
behaviour and is sometimes also referred to as diversive exploratory play (Berlyne, 1960;
Hutt, 1966; Drickamer, Vessey & Meikle, 1996; Power, 2000). Social play involves
chasing, wresting, sexual and other interactions, mostly performed by two or more
conspecifics, but can occur across species (Bekoff & Byers, 1998; Pellis & Iwaniuk,
2004). Although octopuses, with their solitary lifestyle and opportunistic cannibalism
(Hanlon & Messenger, 1996), cannot be expected to engage in social play, object and
locomotor play may be more likely given their predatory lifestyle and capacity for an
almost infinite array of movements employing their eight arms.
The last 20 years have seen a surge in identifying play in animals other than birds
and mammals. This raises the crucial issue of quantitatively studying play as it is
exhibited differently by a wide variety of animals. Burghardt (1999, 2005) formulated
five criteria to standardize the research on play behaviour. These criteria were the first
to offer scientists working on different groups of animals the opportunity to find a
common guideline to compare their findings. The first criterion is that play behaviour is
incompletely functional in the context in which it is expressed. The second criterion states
that play behaviour is spontaneous and pleasurable (‘done for its own sake’). To meet
the third criterion, play has to differ from ‘regular’ behaviour in being exaggerated or
modified. The fourth criterion says that play is repeatedly produced, but not stereotypical.
According to the fifth criterion, play is observed in healthy subjects and in a stress-free
condition.
62 Michael J. Kuba, Tamar Gutnick and Gordon M. Burghardt

One of the biggest problems when dealing with an animal that has such a radically
different, and little understood, behaviour repertoire is to discern what might or might not
be play. In the first paper to broach this subject, Mather and Anderson (1999) observed
play-like behaviour in Enteroctopus dofleini. In a series of 10 daily trials over 5 days,
they presented their octopuses with object stimuli and recorded their behaviour on a
rising scale of object interaction. The most striking result was that two of the octopus
used water jetting, through the exhalant funnel, to manipulate the object in a play-like
manner. While jetting by octopus is used as a means of removing an unpleasant or
unwanted stimulus from the body, such as food remains or annoying fish, in experiments
it was used, in a modified way, to move an object around the aquarium. The extent of
use, the ‘out of context’ circumstance and the modification of a purposeful behaviour
met the criteria for play.
Kuba and colleagues (Kuba, Meisel, Byrne, Griebel & Mather, 2003; Kuba, Byrne,
Meisel & Mather, 2006a) expanded this way of recognizing play by formulating a ‘tree’
of behaviours leading up to and including play (Figure 3.3). By presenting Octopus
vulgaris with a variety of food and non-food objects under different feeding regimes,
they defined interaction levels ranging from the more predatory/exploratory contacts to
play behaviour. The intensity of each behaviour was manifested by a rise in levels (0–4)
on a ‘play scale’. Holding the object to the mouth was designated as level 0 interaction,
this behaviour, if not for feeding, allowed for inspection by the very sensitive mouth area
suckers. Level 1 exploratory interaction, when animals held the object with the distal
suckers of one or several arms led to one of three further exploration methods, but also
sometimes to a level 0 Web Over. The three different interactions defined in level 2 are
the first out-of-context interactions. When these interactions were repeated or prolonged
they were considered play-like interactions, level 3, and if repeated or prolonged further
they were defined as level 4, play. So how did the octopuses play? The first mode of
exploration-play was towing the objects on the water surface. Animals held the object
with one or more arms and then started to move. Initially this behaviour was short and
unidirectional (level 2a), but when it was towed in more than one direction (level 3a)
it was play-like. Towing the object in more than one direction for over 30 seconds was
play (level 4a). The second type of interaction was a simple, single action, where the
animals pulled the objects closer or pushed them away, either horizontally or vertically
(level 2b). When this action was repeated up to five times it was play-like (level 3b),
and if repeated more than five times it was play (level 4b). The third mode was passing
the object from one arm to another. If the object was passed once or twice between
arms it was termed level 2c. When the object was passed three to six times between
arms, it was play like (level 3c); seven or more of such passings were considered play
(level 4c).
These two studies (Kuba, Meisel, Byrne, Griebel & Mather, 2003; Kuba, Byrne,
Meisel & Mather, 2006a) showed that 11 out of 21 tested animals exhibited play
behaviour. Similar to the findings of Mather and Anderson (1999), these behaviours
followed a certain sequence across experimental days. They began engaging in longer
and fewer exploratory contacts, followed by a period of boredom/habituation. When
encountering food items, interactions never exceeded level 2; after exploring the object,
 Learning from play in octopus 63

Figure 3.3 Five levels and three different modes of exploration and play in Octopus vulgaris.
(From Kuba, Byrne, Meisel & Mather, 2006a.)
64 Michael J. Kuba, Tamar Gutnick and Gordon M. Burghardt

Table 3.1 The five play criteria as applied to octopuses

Criterion Fit

The behaviour is incompletely functional The use of water jets for object manipulation in the
in the context in which it is expressed study by Mather and Anderson (1999) and level
3 and 4 interactions in the studies by Kuba and
colleagues (Kuba, Meisel, Byrne, Griebel &
Mather, 2003; Kuba, Byrne, Meisel & Mather,
2006b) are exhibits of natural behaviours that are
incompletely functional in the context in which
they were used
The behaviour is voluntary, spontaneous In all studies (Mather & Anderson, 1999; Kuba,
or rewarding Meisel, Byrne, Griebel & Mather, 2003; Kuba,
Byrne, Meisel & Mather, 2006b) octopuses
sought out the objects to interact with them
The behaviour differs from regular forms Tossing, towing and push/pull behaviours (Kuba,
of behaviour (e.g. exaggerated) Meisel, Byrne, Griebel & Mather, 2003; Kuba,
Byrne, Meisel & Mather, 2006b) as well as
water jets (Mather & Anderson, 1999) are
modifications of regular behaviour
Repeatedly observed but not stereotypic All studies (Mather & Anderson, 1999; Kuba,
Meisel, Byrne, Griebel & Mather, 2003; Kuba,
Byrne, Meisel & Mather, 2006b) show that
contacts with the objects are repeated in a
non-stereotypic way
The behaviour is performed in a Only relaxed and healthy octopuses are curious and
stress-free condition active and engaging in voluntary interactions
with innate or food objects

it was either eaten or abandoned. In the case of non-food objects, interest in the
object returned and in some instances led to more diversified interactions and to play
(Table 3.1; Figure 3.4).
Any interactions exeeding level 2 that the animals made with objects were consid-
ered non-functional in the specific context. They differed from behaviours that animals
show towards food objects. Food objects were always explored – either with one or
several arms, or brought the web between the arms for further chemotactile exploration
and subsequent rejection or consumption. The behaviour towards non-food items was
spontaneous and repeated, and since behaviours in levels 0–2 could be considered ‘nor-
mal’ interactions, it was thus an exaggeration/modification of that behaviour. These
behaviours, therefore, meet the five criteria set by Burghardt (1999, 2005). Additionally,
the progression of the sequence of behaviours that all papers observed resembles the
sequence preceding object-play in children (Mather & Anderson, 1999; Kuba, Meisel,
Byrne, Griebel & Mather, 2003; Kuba, Byrne, Meisel & Mather, 2006b). Hughes (1983)
suggested that a child starts exploration of an object asking the question ‘What is this
object?’ and later transforms this to the question ‘What can I do with this object?’, which
 Learning from play in octopus 65
Duration of contacts (sec)

2500 40

Number of contacts
2000 35
30 Level 4
1500 25
Level 3
20
1000 15 Level 2

500 10 Level 1
5
0 0
1 2 3 4 5 6 7 8 1 2 3 4 5 6 7 8
Experimental session Experimental session
(a) (b)

Figure 3.4 A representative graph showing the duration of contacts with the smooth Lego
R
block
for the octopus ‘Dorian’ (a). Representative graph showing the number of different levels of
interaction by ‘Dorian’ (b) for each experimental session. (From Kuba, Byrne, Meisel & Mather,
2006a.)

Figure 3.5 The exploration–play cycle. (From Hughes, 1983.)

leads to play (Figure 3.5). Similarly, the pattern of behaviours leads from learning to
manipulate an object to playing with it.
Play and play-like behaviour in O. vulgaris differs from that reported for E. dofleini
(Mather & Anderson, 1999), who used their funnels, rather than their arms, to manipulate
objects. This difference in mode of play between the two species is not surprising – even
66 Michael J. Kuba, Tamar Gutnick and Gordon M. Burghardt

closely related mammals such as rodents, carnivores and primates also show different
types and amounts of play (Pellis, 1993; Burghardt, 2005). Apart from ecological dif-
ferences, O. vulgaris is also a diurnal (Meisel et al., 2006), curious and agile species,
spending most of the time during observational periods moving around in its tank. Ente-
roctopus dofleini, on the other hand, is a nocturnal cold-water species (Anderson &
Wood, 2001). It is probably less agile and interacted less with the object when the object
was moved towards the octopus by the water current. These differences aside, both
species fulfilled the criteria for play or play-like behaviour.
Do octopuses have the necessary prerequisites to show such a complex behaviour?
Aristotle was the first to report the curiosity displayed by the Octopus, which sparked an
age-old speculation about their intelligence. Their ability to learn and perform complex
tasks points towards a highly evolved cognitive capacity (Mather 1995; Hanlon & Mes-
senger, 1996; Hochner, 2008, Mather 2008; Hochner, 2010; Gutnick, Byrne, Hochner &
Kuba, 2011; Figure 3.4). Researchers have also demonstrated the existence of personal-
ities (Mather & Anderson, 1993; Sinn, Perrin, Anderson & Mather, 2001) and, perhaps,
even archaic forms of consciousness in octopuses (Mather, 2008; Edelmann & Seth,
2009), further indicators of complex behaviour. All these factors fit into the model on
factors underlying primary process play formulated by Burghardt (2005; Table 3.1).
What might be the function of play in a cephalopod mollusc? If play has an adap-
tive (e.g. learning, training) function it should be more frequently observed in younger,
smaller animals, as proposed by various theories on the importance of play in training and
preparing animals for future behaviours (Byers & Walker, 1995; Bekoff & Byers, 1998;
Spinka, Newberry & Bekoff, 2001). While there was no difference in play by smaller
(potentially) younger versus larger (potentially) older animals (Kuba, Byrne, Meisel &
Mather, 2006a), the standard concepts of juvenile and adult life found in vertebrates do
not easily translate to cephalopods and, thus, need further scientific research. Presum-
ably, play behaviour might not have the training, or practice, role for life in an octopus
as it has in many vertebrates (Bekoff & Byers, 1998; Burghardt, 2004). In contrast to the
view of play as a preparation for the future, Hall (1904) formulated the long neglected
recapitulation theory of play. Hall (1904) stated that play is more a legacy from the past
than a preparation for the future. Burghardt (1984, 1998, 2004, 2005, 2010; Graham &
Burghardt, 2010) picked up some of the implications of Hall’s theory in his Surplus
Resource Theory (SRT) and his division of play into primary, secondary and tertiary pro-
cesses. He claimed that primary process play is most likely to occur when behaviourally
complex animals have some surplus resources including metabolic energy, time, minimal
stress and a rich repertoire of behavioural movements to deploy (Burghardt, 1998, 1999,
2004, 2005, 2010; Graham & Burghardt, 2010). Burghardt (2004) formulated his theory
on the evolutionary importance of primary process play (play without a direct selective
benefit from the behaviour itself), claiming that it might not have adaptive functions,
but can be useful to the individual if it is, to the extent it is heritable, transformed via
selection to enhance survival and reproductive fitness. Additionally, it provides variation
from which novel and complex behaviour can be derived more rapidly than through
selection of the more fixed functional aspects of an animal’s behavioural repertoire. In
other words, play can indeed become an evolutionary pump. In this case, play might
 Learning from play in octopus 67

have additional functions not easily tested. Darwish, Korányi, Nyakas & Almeida (2001)
showed that providing novel objects to aged rats reduced anxiety levels and that in both
young and adult rats, play behaviour dampens the stress response as measured by the
corticosterone level (Darwish, Korányi, Nyakas & Ferencz, 2001). In their case, play had
the function of creating bolder animals with increased behavioural flexibility (Darwish,
Korányi, Nyakas & Ferencz, 2001). As sensory deprivation tank experiments (Lilly &
Shurley, 1961) showed that humans had visual sensations of sensory inputs (hallucina-
tions) without stimulation, play might also be a way to keep a brain ‘busy’ in times of
stimulus deprivation. Play could, therefore, be a by-product of a complex nervous system
heavily depending on learning (Hanlon & Messenger, 1996; Hochner, 2008) and have
the function of maintaining a status quo in times of lesser ecological pressure and lack
of stimuli. Thus, play has been an important indicator for evaluating animal welfare in
vertebrates (for a review, see Held & Spinka, 2011). This raises the issue of enrichment
as an essential part of cephalopod welfare and, in that context, whether exploratory and
play behaviour has to be considered as an indicator for welfare. All these questions pro-
vide important and interesting avenues for research into further underlying mechanisms
in cephalopods.
What about cephalopod play in groups other than octopus? Certainly cuttlefish and
squid might have the cognitive capacity to engage in playful interactions. However, due
to the different nature of their body plan, life history and ecology, they might engage in
other activities such as locomotor play or, to the extent they are social, rudimentary forms
of social play. An especially promising candidate to study this would be cuttlefish or
the Caribbean reef squid. Both groups show highly complex social interactions ranging
from sneaker males (Hanlon, Naud, Shaw & Havenhand, 2005) to more or less complex
communication using body shape and colour (Byrne, Griebel, Wood & Mather, 2003;
Mather, Griebel & Byrne, 2010).

3.4 Conclusions

A true comparative approach to play was long neglected, as play was seen to be a
trait unique to mammals. For example, the eminent brain researcher Paul MacLean
(1985, 1990) stated that play behaviour was one of the ‘signature behaviours’ separat-
ing mammals from other vertebrates. However, his theories did not take into account
reports on avian play already documented by Fagan (1981). Burghardt, Ward and Ross-
coe (1996) reported play behaviour in a Nile soft-shelled turtle, and thereby stretched
the ‘phylogenetic boundaries’ of play even further. To date animals as diverse as
lizards, crocodilians, frogs, fishes (Burghardt, 2005), insects (Dapporto, Turilliazzi &
Palagi, 2006) and, most recently, spiders (Pruitt, Burghardt & Reichert, 2012) have
been reported to show at least some form of play behaviour. Due to the phyloge-
netic distance of the invertebrate octopus to other studied playful animals, a new win-
dow has opened for comparative studies on the evolution of not only play behaviour
but also how play may impact the evolution and instantiation of other behavioural
systems.
68 Michael J. Kuba, Tamar Gutnick and Gordon M. Burghardt

References

Alves, C., Chichery, R., Boal, J. G. and Dickel, L. (2007). Orientation in the cuttlefish Sepia
officinalis: Response versus place learning. Animal Cognition, 10: 29–36.
Anderson, R. C. and Wood, J. B. (2001). Enrichment for giant Pacific octopuses: Happy as a clam?
Journal of Applied Animal Welfare Science 4: 157–168.
Archer, J. and Birke, L. (eds) (1983). Exploration in Animals and Humans, Wokingham, Berks,
UK: Van Nostrad Reinhold.
Baldwin, J. D. and Baldwin, J. I. (1997). Behavioral Principles in Everyday Life, Upper Saddle
River, NJ: Prentice Hall.
Bekoff, M. and Byers, J. A. (1998). Animal Play: Evolutionary, Comparative, and Ecological
Perspectives, Cambridge: Cambridge University Press.
Berlyne, D. E. (1960). Conflict, Arousal, and Curiosity, New York: McGraw-Hill.
Berlyne, D. E. (1966). Curiosity and exploration. Science 153: 25–33.
Bruner, J. S., Jolly, A. and Sylva, K. (1976). Play – Its Role in Development and Evolution, New
York: Basic Books.
Burghardt, G. M. (1984). On the origins of play. In Play in Animals and Humans, Smith, P. K.
(ed.), London: Basil Blackwell.
Burghardt, G. M. (1998). The evolutionary origins of play revisited: Lessons from turtles. In
Animal Play: Evolutionary, Comparative, and Ecological Perspectives, Bekoff, M., Byers J. A.
(eds.), Cambridge: Cambridge University Press.
Burghardt, G. M. (1999). Conceptions of play and the evolution of animal minds. Evolution and
Cognition, 5: 115–123.
Burghardt, G. M. (2004). Play: How evolution can explain the most mysterious behavior of all. In
Evolution from Molecules to Ecosystems, Moya, A., Font, E. (eds.), Oxford: Oxford University
Press.
Burghardt, G. M. (2005). The Genesis of Animal Play: Testing the Limits, Cambridge, MA: MIT
Press.
Burghardt, G. M. (2010). Play. In Encyclopedia of Animal Behavior, Breed, M., Moore J. (eds.)
Oxford: Academic Press.
Burghardt, G. M. (2012). Play, exploration and learning. In Encylopedia of Sciences of Learning,
Seel, N. M. (ed.), Heidelberg: Springer.
Burghardt, G. M., Ward, B. and Rosscoe, R. (1996). Problem of reptile play: Environmental
enrichment and play behavior in a captive Nile soft-shelled turtle, Trionyx triunguis. Zoo
Biology, 15: 223–238.
Byers, J. A. and Walker, C. (1995). Refining the motor training hypothesis for the evolution of
play. The American Naturalist, 146: 25–40.
Byrne, R. A., Griebel, U., Wood, J. B. and Mather, J. A. (2003). Squid say it with skin: A
graphic model for skin displays in Caribbean Reef Squid (Sepioteuthis sepioidea). Berliner
Palaeoontologische Abhandlungen, 3: 29–35.
Craig, W. (1918). Appetites and aversions as constituents of instincts. The Biological Bulletin, 34:
91–107.
Dapporto, L., Turilliazzi, S. and Palagi, E. (2006). Dominance interactions in young adult paper
wasps (Polistes dominulus) foundresses: A playlike behavior? Journal of Comparative Psychol-
ogy, 120: 394–400.
Darwish, M., Korányi, L., Nyakas, C. and Almeida, O. F. X. (2001). Exposure to a novel
stimulus reduces anxiety level in adult and aging rats. Physiology and Behavior, 72: 403–
407.
 Learning from play in octopus 69

Darwish, M., Korányi, L., Nyakas, C. and Ferencz, A. (2001). Induced social interaction reduces
corticosterone stress response to anxiety in adult and aging Rats. Klinikai és Kı́sérletes Labo-
ratoriumi Medicina, 28: 108–111.
Day, J. E. L., Kyriazakis, I. and Lawrence, A. B. (1994). The effect of food deprivation on the
expression of foraging and exploratory behavior in the growing pig. Applied Animal Behavior
Science, 42: 193–206.
Drickamer, L.C., Vessey, S.H. and Meikle, D. (1996). Animal Behavior, 4th edn. Dubuque, IA:
William C. Brown.
Edelman, D. B. and Seth, A. K. (2009). Animal consciousness: A synthetic approach. Trends in
Neurosciences, 32: 476–484.
Fagan, R. (1981). Animal Play Behavior, New York: Oxford University Press.
Gazzaniga, M. S. and Heatherton, T. F. (2003). Psychological Science: Mind, Brain and Behavior,
New York: W. W. Norton and Company.
Graham, K. L. and Burghardt, G. M. (2010). Current perspectives on the biological study of play:
Signs of progress. Quarterly Review of Biology, 85: 393–418.
Grasso, F. and Basil, J. (2009). The evolution of flexible behavioral repertoires in cephalopod
molluscs. Brain Behavior and Evolution, 74: 231–245.
Gutnick, T., Byrne, R. A., Hochner, B. and Kuba, M. (2011). Octopus vulgaris uses visual
information to determine the location of its arm. Current Biology, 21: 460–462.
Hall, G. S. (1904). Adolescence, its Psychology and its Relation to Physiology, Anthropology, Sex,
Crime, Religion, and Education, New York: D. Appleton and Company.
Hanlon, R. T. and Messenger, J. B. (1996). Cephalopod Behaviour, New York: Cambridge Uni-
versity Press.
Hanlon, R. T., Naud, M.-J., Shaw, P. W. and Havenhand, J. N. (2005). Transient sexual mimicry
leads to fertilisation. Nature, 430: 212.
Harlow, H. F. (1950). Learning and satiation of response in intrinsically motivated complex puzzle
performance by monkeys. Journal of Comparative and Physiological Psychology, 43: 289–
294.
Harlow, H. F. (1953). Mice, monkeys, men and motives. Psychological Review, 60: 23–33.
Heinrich, B. (1995). Neophilia and exploration in juvenile common raven Corvus corax. Animal
Behaviour, 50: 695–704.
Held, S. D. E. and Spinka, M. (2011). Animal play and animal welfare. Animal Behaviour, 81:
891–899.
Hochner, B. (2008). Octopuses. Current Biology, 18: R897.
Hochner, B. (2010). Functional and comparative assessments of the octopus learning and memory
system. Frontiers in Biosciences, 2: 764–771.
Huber, L., Rechberger, S. and Taborsky, M. (2001). Social learning affects object exploration and
manipulation in keas, Nestor notabilis. Animal Behaviour, 62: 945–954.
Hughes, M. (1983). Exploration and play in young children. In Exploration in Animals and
Humans, Archer J., Birke L. (eds.), Wokingham, Berks, UK: Van Nostrad Reinhold.
Hutt, C. (1966). Exploration and play in children. Symposia of the Zoological Society of London,
18: 61–81.
Kuba, M. J., Byrne, R. A., Meisel, D. V. and Mather, J. A. (2006a). When do octopuses play?
Effects of repeated testing, object type, age, and food deprivation on object play in Octopus
vulgaris. Journal of Comparative Psychology, 120: 184–190.
Kuba, M. J., Byrne, R. A., Meisel, D. V. and Mather J. A. (2006b). Exploration and habituation
in intact free moving Octopus vulgaris. International Journal of Comparative Psychology, 19:
426–438.
70 Michael J. Kuba, Tamar Gutnick and Gordon M. Burghardt

Kuba, M. J., Meisel, D. V., Byrne, R. A., Griebel, U. and Mather, J. A. (2003). Looking at play in
Octopus vulgaris. Berliner Paläontologische Abhandlungen, 3: 163–169.
Lilly, J. C. and Shurley, J. T. (1961). Experiments in solitude in maximum achievable physical
isolation with water suspension of intact healthy persons. (Symposium, USAF Aerospace
Medical Center, San Antonio, Texas, 1960.) In Psychophysiological Aspects of Space Flight,
New York: Columbia University Press.
MacLean, P. D. (1985). Brain evolution relating to family, play, and the separation call. Archives
of General Psychiatry, 42: 504–517.
MacLean, P. D. (1990). The Triune Brain in Evolution: Role in Paleocerebral Functions, New
York: Plenum Press.
Mather, J. A. (1991a). Foraging, feeding and prey remains in middens of juvenile Octopus vulgaris.
Journal of Zoology, 224: 27–39.
Mather, J. A. (1991b). Navigation by spatial memory and use of visual landmarks in octopuses.
Journal of Comparative Physiology A, 168: 491–497.
Mather, J. A. (1995). Cognition in cephalopods. Advances in the Study of Behavior, 24: 317–
353.
Mather, J. A. (2008). Cephalopod consciousness: Behavioral evidence. Consciousness and Cog-
nition, 17: 37–48.
Mather, J. A. and Anderson, R. C. (1993). Personalities of octopuses (Octopus rubescens). Journal
of Comparative Psychology, 107: 336–340.
Mather, J. A. and Anderson, R. C. (1999). Exploration, play and habituation. Journal of Compar-
ative Psychology, 113: 333–338.
Mather, J. A., Griebel, U. and Byrne, R. A. (2010). Squid dances: An ethogram of postures
and actions of Sepioteuthis sepioidea squid with a muscular hydrostatic system. Marine and
Freshwater Behaviour and Physiology, 43: 45–61.
Meisel D. V., Byrne R. A., Kuba M. J., Mather J. A. Ploberger W. and Reschenhofer E. (2006).
Comparing the activity patterns of two Mediterranean cephalopod species. Journal of Compar-
ative Psychology, 120: 191–197.
Moynihan, M. H. and Rodaniche, A. F. (1982). The behavior and natural history of the Caribbean
reef squid Sepioteuthis sepioidea with a consideration of social, signal and defensive patterns
for difficult and dangerous environments. Advances in Ethology, 25: 1–151.
Müller-Schwarze, D. (1984). Analysis of play behaviour: What do we measure and when? In Play
in Animals and Humans, Smith, P. K. (ed.), Oxford: Basil Blackwell.
Nixon, M. and Young, J. Z. (2003). The Brains and Lives of Cephalopods. Oxford: Oxford
University Press.
Packard, A. (1972). Cephalopods and fish: The limits of convergence. Biological Reviews, 47:
241–307.
Pellis, S. M. (1993). Sex and the evolution of play fighting: A review on the behavior of muroid
rodents. Play Theory and Research, 1: 55–75.
Pellis, S. M. and Iwaniuk, A. N. (2004). Evolving a playful brain: A level of control approach.
International Journal of Comparative Psychology, 17: 92–118.
Power, T. G. (2000). Play and Exploration in Children and Animals. Mahwah, NJ: Lawrence
Erlbaum Associates.
Pruitt, J. N., Burghardt, G. M. and Riechert, S. E. (2012). Non-conceptive sexual behavior in
spiders: A form of play associated with body condition, personality type, and male intrasexual
selection. Ethology, 18: 33–40.
Renner, M. J. (1990). Neglected aspects of exploratory and investigative behavior. Psychobiology,
18: 16–22.
 Learning from play in octopus 71

Renner, M. J. and Seltzer, C. P. (1994). Sequential structure in behavioral components of object

investigations by Long–Evans rats. Journal of Comparative Psychology, 108: 81–91.
Russell, P. A. (1983). Psychological studies of exploration in animals. In Exploration in Animals
and Humans, Archer J., Birke L. (eds.), Wokingham, Berks, UK: Van Nostrad Reinhold.
Sanders, G. D. (1975). The cephalopods. In Invertebrate Learning, Corning, W.C. (ed.), New
York: Plenum Press.
Sinn, D. L., Perrin, N. A., Anderson, R. C. and Mather, J. A. (2001). Early temperamental traits
in an octopus (Octopus bimaculoides). Journal of Comparative Psychology, 115: 351–364.
Sommer, V. and Mendoza-Grandos, D. (1995). Play as indicator of habitat quality: A field study
of langur monkeys (Presbytis entellus). Ethology, 99: 177–192.
Spinka, M., Newberry, R. C. and Beckoff, M. (2001). Mammalian play: Training for the unex-
pected. Quarterly Review of Biology,76: 141–168.
Toates, F. M. (1983). Exploration as a motivational and learning system. In Exploration in Animals
and Humans, Archer J., Birke L. (eds.), Wokingham, Berks, UK: Van Nostrad Reinhold.
Wells, M. J. (1962). Brain and Behavior in Cephalopods, London: Heinemann.
Wells, M. J. (1978). Octopus. Physiology and Behaviour of an Advanced Invertebrate, London:
Chapman and Hall.
Wood-Gush, D., Stolba, A. and Miller, C. (1983). Exploration in farm animals and animal hus-
bandry. In Exploration in Animals and Humans, Archer J., Birke L. (eds.), Wokingham, Berks,
UK: Van Nostrad Reinhold.
4 The neurophysiological basis of
learning and memory in an advanced
invertebrate: the octopus
Binyamin Hochner and Tal Shomrat

4.1 Introduction

It is commonly believed that invertebrates should be used for studying general ques-
tions in neurobiology only when their specific features, such as “a small number of
large identifiable neurons” or their amenity to genetic manipulation, provide a special
experimental advantage. We do not agree. We believe it is important to study octopuses
and other modern cephalopods for totally different reasons. These invertebrates show
behavioral repertoires comparable to those of higher vertebrates (Packard, 1972; Wells,
1978; Hochner, Shomrat & Fiorito, 2006; Hochner, 2008; Grasso & Basil, 2009), yet
their brains maintain the much simpler invertebrate organization (Budelmann, 1995).
This unique combination of simpler nervous system and complex behavior is especially
advantageous for tackling the central question of how a nervous system controls complex
behaviors and cognitive functions.
Although the subject of this book is cephalopod cognition, this chapter deals with
the neurophysiological bases of learning and memory, which, in our opinion, is not
necessarily a cognitive capacity. It is true that cognitive functions are usually thought
(e.g. Wikipedia definition) to include learning and memory in addition to attention,
association, language, problem-solving, decision-making, mental imagery and more.
There is no doubt, however, that cognitive function cannot happen without learning
and memory. Some cognitive functions may emerge from mutual high level processing
of long-term stored memories and the current sensory information stored temporarily
as short-term working memory. In this mechanistic scenario memory itself, especially
in its more basic forms like associative and non-associative learning, would not be
considered a cognitive function. It is worth noting that certain animals without obvious
high cognitive abilities show excellent learning and memory, and humans with cognitive
disorders may still possess an excellent memory.
As part of their sophisticated behavioral repertoire, the octopus and other modern
cephalopods exhibit vertebrate-like learning and memory behaviors (Sanders, 1975;
Wells, 1978; Fiorito & Scotto, 1992; Hanlon & Messenger, 1996; Hochner, Shomrat &
Fiorito, 2006; Grasso & Basil, 2009). Analyzing the cellular processes and neuronal
circuitry of their learning and memory systems and comparing the results with those

Cephalopod Cognition, eds A.-S. Darmaillacq, L. Dickel and J. Mather. Published by Cambridge University
Press. 
C Cambridge 2014.
 Neurophysiological basis of learning and memory in the octopus 73

vertical l. subvertical l.

superior frontal l.
amacrine MSF cell
inferior frontal l

dorsal basal l. superior buccal l.

VL
MSFL tract
agus
esoph large cell MSFL

median basal l. basal l.

l anterior
sop hagea
suprae
geal SV
sopha
suboe
IFL

brachial nerve.

BL

anterior brachial l. BL

median pedal l.
(b)
statocyst

palliovisceral l
10mm

(a)

Figure 4.1 The morphological organization of the octopus central brain. (a) Sagittal section of the
sub- and supraoesophageal lobes of Octopus vulgaris showing the dorsal location of the median
superior frontal (MSFL)–vertical lobe (VL) system and the organization of the lobes which are
discussed in the text. (Modified from Nixon and Young, 2003.) (b) Unstained median sagittal
section through the dorsal part of the supraoesophageal brain mass with superimposed schematic
drawing of the three types of neurons in this system and their connections. This area in the
supraoesophageal lobes (see (a)) forms the main part of the VL-slice. (BL, basal lobes; IFL,
inferior frontal lobe; MSFL, median superior frontal lobe; SV, subvertical lobe; VL, vertical lobe).

of other invertebrate and vertebrate systems may advance our understanding of their
evolution and function. This comparative evolutionary approach can determine whether
mechanisms subserving cognitive functions evolved convergent across widely diverse
phyla or, alternatively, whether evolutionarily primitive mechanisms mediating simple
forms of behaviors are conserved in more advanced animals to mediate complex behav-
iors.
The octopus central brain consists of 40 lobes that maintain some simple organi-
zational features of the central nervous system of simple mollusks (Figure 4.1). The
neuron cell bodies lie in the outer region of each lobe, while their processes form the
central neuropil (Figure 4.1a; Young, 1971).
Previous experiments have indicated that the vertical lobe (VL) is involved in higher
brain functions. Stimulating it or the superior frontal lobes evoked no obvious effects,
whereas stimulating other parts of the brain caused movements of some part of the body
(Boycott, 1961; Zullo, Sumbre, Agnisola, Flash & Hochner, 2009). Removing the VL
also did not appear to affect the animal’s general behavior (Boycott & Young, 1955;
Maldonado, 1965; Sanders, 1975). Thus, the VL and the superior frontal lobe do not
seem to be engaged in simple motor functions. Deficiencies were revealed only when
74 Benny Hochner and Tal Shomrat

animals were required to learn new tasks. After removal or lesion of the VL, octopuses
continued to attack crabs in spite of receiving electrical shocks, unless the intertrial
interval was less than 5 minutes (Boycott & Young, 1955). Thus, the VL appears to be
specifically involved in long-term and more complex forms of memory.
Another complex form of learning demonstrated by octopuses is observational learn-
ing. A naive octopus learns to attack a previously positively rewarded target after only
four observations of a trained octopus attacking the same target. This is much faster
than it takes to train the demonstrator octopus (Fiorito & Scotto, 1992). Lesion study
has demonstrated that the VL is important for this advanced form of learning (Fiorito &
Chichery, 1995).
Finally, the morphological structure of the VL appears very simple, but its unique
matrix-like organization led Boycott and Young (1955) to postulate that the median
superior frontal lobe (MSFL)–VL and inferior frontal lobe (IFL) (Figure 4.1b) networks
form associative networks for learning and memory. Wells (1978) and Young (1991)
further suggested that the VL matrix is analogous to the mammalian hippocampus
memory system and the insect mushroom body (Young, 1991, 1995; Hochner, 2010). The
VL system is, thus, an exciting site for exploring the neuronal circuitry and physiological
mechanisms involved in learning and memory.

4.2 The anatomy of the vertical lobe system

The octopus VL contains only two types of neuron, amacrine cells and large neurons,
both of which are morphologically typical invertebrate monopolar neurons (Figure 4.2;
Gray & Young, 1964; Gray, 1970; Young, 1971). Twenty-five million amacrine cells,
which are the smallest neurons in the octopus brain (diameter 6–10 µm) converge onto
only 65 000 large neurons (diameter 17 µm). The amacrine processes are confined
within the VL (Figures 4.1b, 4.2), while the axons of the large neurons form the only
output of the VL, leaving it in organized axons bundles or roots, which are easy to
identify and record from (Figure 4.1b).
The VL receives inputs from the MSFL. This lobe is thought to integrate visual and
taste information. It contains only one morphological type of neuron, whose 1.8 million
axons project to the VL in a distinct tract running between the VL neuropil and its
outer cell body layer (Figure 4.1a,b). This arrangement allows each MSFL axon to
make en passant synapses with many amacrine neurons along the VL (Figure 4.2).
Young (1971) also postulated a direct connection to the large neurons but this was not
supported by Gray’s (1970) electron microscopic study, and is also not supported by
recent physiological results (see below and Shomrat et al., 2011). This special matrix-
like organization of the MSFL–VL complex is shared by the inferior frontal lobe (Figure
4.1), which plays a role in chemotactile memory (Wells, 1978; Young, 1991).
Testing Boycott and Young’s (1955) hypothesis that the VL and inferior frontal net-
works are associative networks for learning and memory clearly requires physiological
characterization of these brain circuits and their plastic properties. The new experimental
 Neurophysiological basis of learning and memory in the octopus 75

Figure 4.2 Digram showing what is thought to be the basic circuitry of the vertical lobe. The
nature of the “serial synapse” of the amacrine cells is depicted by the arrows indicating input and
output at the same amacrine dendrite. (amn, amacrine cells; amt, amacrine trunk; dc, dendritic
collaterals of large cell; dcv, dense-core vesicle; lc, large cell; msf, median superior frontal axon;
mt, microtubule; nf, neurofilaments; pa, possible “pain” axon input to the large cell; sv, synaptic
vesicles.) (From Gray, 1970.)
76 Benny Hochner and Tal Shomrat

preparations that have been developed now allow neurophysiological exploration of this
anatomically unique system.

4.3 Neurophysiology of the vertical lobe

Most of the octopus brain shows a typical invertebrate form (see Bullock & Horridge,
1965), but the VL shows a greater resemblance to vertebrae brain organization, with its
large numbers of neurons organized in layers and their processes aligned more or less in
parallel. This organization, unlike the more random organization of typical invertebrate
ganglia, generates a measurable external electric field near the active neurons. This
field potential shows local field potentials (spontaneous, ongoing background activity),
evoked potentials (EP) and event-related potentials (ERP), and a compound field poten-
tial similar to an electroencephalography (EEG) (e.g. Bullock & Basar, 1988). Bullock
(1984) suggested that the existence of such a compound field potential indicates the high
level of complexity of the octopus brain.
How does the anatomical organization allow us to record field potentials? The amount
of current flowing in the extracellular space due to neuron activity, such as an action
potential, is usually tiny and, as the extracellular impedance is very low (about 3–
4 orders of magnitude lower than the neurons’ input resistance), only summation of
local field potentials generated by many neurons becomes large enough to be reliably
monitored. Such summation accrues only if the neurons are synchronously active and
their currents flow in the same direction, otherwise they cancel each other out. This
condition is achieved only when the neurons have elongated morphology and they
are organized in the same orientation. Such an arrangement occurs in the MSFL–VL
system, where the MSFL axons input to the VL is organized in bundles. The cell bodies
of the millions of amacrine neurons lie in the outer zone of the lobe, while their axons
project in parallel into the lobe, perpendicular to the MSFL axons bundles (Figures
4.1, 4.2). Such an architecture would be expected to generate a significant local field
potential (LFP), possibly similar to that generated close to the Schaeffer collaterals in the
hippocampus.
The existence of such a local field potential was confirmed by physiological exper-
iments. Stimulating the MSFL tract with a short current pulse evokes a typical field
potential waveform (Figure 4.3). This is a tri-phasic (positive–negative–positive) tract
potential (TP) generated by the volley of action potential propagating along the stim-
ulated axons in the MSFL tract. The delay after the stimulus artifact (stim.) depends
on the distance between stimulus and recording electrodes. A mainly negative field
potential immediately follows the second positive wave of the TP. This potential is a
postsynaptic field potential (fPSP) because it disappears in zero-calcium physiologi-
cal solution (Figure 4.3a). In contrast to the action potential, which is generated by
sodium and potassium ionic currents, the release of neurotransmitter depends on cal-
cium ions and, therefore, removal of calcium blocks the part of the LFP, which is gener-
ated by postsynaptic response while leaving the TP unchanged. This fPSP is generated
by opening of postsynaptic glutamate receptors because the fPSP is also blocked by
 Neurophysiological basis of learning and memory in the octopus 77

Figure 4.3 (a) Superimposed local field potentials (LFP) in control and after blocking
postsynaptic field potential (fPSP) in Ca2+ -free EGTA ASW (artificial seawater) fPSP,
postsynaptic field potential; Stim, stimulus artifact; TP tract potential). (b) Similar to (a),
showing blockade of fPSP by AMPA-like glutamatergic receptor antagonists CNQX. Inset:
Schematic depiction of median superior frontal (MSFL)–vertical lobe (VL) slice preparation
with a possible electrodes placement for stimulation and recording of LFP.
78 Benny Hochner and Tal Shomrat

AMPA (α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid)-like antagonists

such as CNQX (6-cyano-7-nitroquinoxaline-2,3-dione) (Figure 4.3b), DNQX (6,7-
dinitroquinoxaline-2,3-dione) or kynurenate (Hochner, Brown, Langella, Shomrat &
Fiorito, 2003).
The types of fast neurotransmitters in the invertebrate nervous system are more
diversified than in vertebrates, and finding glutamate neurotransmission in the VL has
hinted to the universality of this transmitter in learning neural networks. However, as
will be shown below, this is not entirely true, because acetylcholine was found to be
another fast transmitter in the VL. In vertebrates acetylcholine is a fast neurotransmitter
mainly at the neuromuscular system. “Fast” neurotransmitters serve for fast transmission
of information (at the range of a few milliseconds) between neurons and involves the
opening of ionic channels, while “slow” neurotransmitters serve mainly in modulation
of cell activity and their action involves activation of a special class of receptors that are
coupled, via a biochemical cascade, to the activation of second messengers (e.g. cyclic
adenosine monophosphate [c-AMP]).
The short latency between the peak negativity of the TP and the fPSP onset (3 ms,
Figure 4.3a,b) suggests a monosynaptic delay. The physiological results are thus in good
agreement with Gray’s scheme (Figure 4.2), in which the MSFL axon terminals synapse
directly on the amacrine dendrites. The recent physiological results (Figure 4.3b), further
suggest that the first synaptic layer of the VL is likely glutamatergic.
As expected from the anatomical organization, the local field potential in the VL
(presynaptic TP followed by fPSP; Figure 4.3) is similar to that evoked at the stratum
radiatum of the hippocampus by stimulating the Schaffer fiber collaterals in CA1. In both
cases the collaterals/tract terminals make synapse en passant with the dendrites of the
pyramidal/amacrine cells, respectively. However, because the amacrine cells are inex-
citable as they don’t generate regenerative action potentials (Hochner, Brown, Langella,
Shomrat & Fiorito, 2003; Hochner, Shomrat & Fiorito, 2006), the fPSP has an amplitude-
independent waveform with no population spikes, in contrast to the postsynaptic field
potential of the hippocampal pyramidal neurons (Miyakawa & Kato, 1986).
The amacrine cells innervate the large efferent neurons via special “serial” synapses
(Figure 4.2). This synapse is called serial because the dendrites of the amacrine cells are
both the input site for the MSFL axonal terminals and the output site to the large cell
(arrows in Figure 4.2). To investigate the nature of this input to the large cells, infrared
differential interference contrast (DIC) microscopy was used to record intracellularly
from their cell bodies, as well as recording their spiking activity extracellularly from
their axon bundles. Figure 4.4a shows excitatory postsynaptic potential (EPSP) evoked
by stimulating the MSFL tract and spontaneous EPSPs in a large neuron. Like most
invertebrate neurons, the cell bodies of the large neurons are inexcitable, as demonstrated
by the small non-overshooting spikelets (arrowhead). These decrease in size passively
as they pass along the single neurite to the cell body from a distant spike initiation zone
at the junction between the dendritic tree and the axon.
The synaptic input to the large neurons is cholinergic, since both evoked and spon-
taneous EPSPs were blocked by hexamethonium (Figure 4.4a), a muscarinic receptor
antagonist that also blocks the synaptic potential at the neuromuscular junctions of the
 Neurophysiological basis of learning and memory in the octopus 79

(a) (b) (c)

Control

−1′
0′ 1mM-Hex

10mM-Hex
1′

1.5′
0.05mV 0.5 mV
Washout
50ms 0.5 ms
2′
10 mV
50 ms
2.5′
stim

(d) (e)
Bundle activity (normalized)

1.4
LF stimulation
1.2 P
re VL
fPSP (normalized)

c SFL
1
0.8 SFL tract
fPSP whole-cell rec.
0.6
bundles
0.4
0.2 Hex bundle activity
n=9
0
0 5 10 15 20 25 30 35
Time (min)

Figure 4.4 The synaptic inputs to the large cells are most likely cholinergic. (See (e) for the
different recording modes.) (a) Hexamethonium blocked the spontaneous and SFL tract-evoked
EPSPs recorded intracellularly from large cells (arrowheads point to example of spikelet).
(b) Hexamethonium blocked the burst of action potentials recorded extracellularly from large
cells axonal bundles. Note that twin stimuli (b,c) were used to obtain a clearly measurable bundle
response. (c) Hexamethonium had no effect on the tract potential (TP) and postsynaptic field
potential (fPSPs). Records were obtained simultaneously with bundle activity shown in (b).
(d) Summary of nine experiments as exemplified in (b) and (c). Black curve, normalized
integrated bundle activity; gray, fPSP amplitude. Here and in subsequent figures responses were
normalized to the average of 3–10 test responses at the beginning of the experiments. (Modified
from Shomrat et al., 2011.)

octopus arm (Matzner, Gutfreund & Hochner, 2000). Hexamethonium also blocked both
spontaneous and evoked spiking activity recorded from the large neuron axon bundles
(Figure 4.4b,d). As would be expected for glutamatergic synapses, the fPSP of the first
synaptic layer was unaffected by cholinergic antagonists (Figure 4.4c,d). Thus, it must
be the amacrine-to-large neuron synapse that is cholinergic.
Both cholinergic and glutamatergic antagonists blocked activity output of the large
neuron axon bundles. This finding suggests that there is no strong direct connection from
80 Benny Hochner and Tal Shomrat

MSFL input axons to the large neurons and that the main connections within the VL
are the MSFL inputs onto the amacrine cells, which then innervate the large efferents
neurons (Shomrat et al., 2011).
The VL system thus appears to be organized as a simple feed-forward fan-out fan-
in network. This type of network architecture is frequently found amongst biological
and artificial networks (e.g. a machine-learning algorithm; Vapnik, 1998) that learn to
classify when endowed with the synaptic plasticity that creates learning ability. The
first fan-out synaptic layer may create a neural representation of the incoming sensory
information in a form suitable for further processing at the fan-in layer (Nowotny, Huerta,
Abarbanel & Rabinovich, 2005; Shomrat et al., 2011). If the VL system possesses this
architecture and functions as a learning and memory network, as shown above, then we
would expect to find synaptic plasticity at one or more of its synaptic sites.

4.4 Synaptic plasticity in the vertical lobe

Examining the neurophysiological properties of synaptic transmission in the VL network

revealed clear synaptic plasticity.
The local field potential recorded at the MSFL tract demonstrates a very robust
activity-dependent plasticity. Tetanization led to a long-term potentiation (LTP) with a
4-fold increase in fPSP amplitude without affecting the amplitude of the presynaptic
TP (Figure 4.5, inset top left). The potentiation of the glutamatergic synaptic input to the
amacrine neurons is long-term as it was maintained for at least 5 hours (Hochner, Brown,
Langella, Shomrat & Fiorito, 2003). Tetanizing a second time showed that this LTP is
a saturated phenomenon, because no additional long-term enhancement is obtained
(Figure 4.5).
Does activity-dependent plasticity occur only at the fan-out glutamatergic connections
(Figures 4.1, 4.5)? Tetanization, which induced LTP of the fPSP, also caused a long-term
increase in the extracellular spiking activity of the large neuron axon bundles (Figure
4.6a,b). This result was not surprising; facilitation of the synaptic input to the amacrine
cells (the fPSP) should increase their cholinergic input to the large neurons and thus to
an enhancement of the large neurons output (i.e. increase bundle activity).
This result does not tell whether there is also synaptic plasticity at the input to the
large neurons. However, analyzing the relationship between the amplitude of the fPSP
and bundle activity showed a linear relationship that did not change after LTP induction
(Figure 4.6c, gray labels; Shomrat et al., 2011). Because the same fPSP amplitude gave
rise to the same level of bundle activity, irrespective of LTP, then LTP must occur only
at the first, fan-out, synaptic layer.
This linear input–output relationship (Figure 4.6c) has important computational con-
sequences. In a linearly operating network such as this, similar computation capacity is
obtained regardless of whether the plasticity is localized at the fan-out or the fan-in layer
(see Shomrat et al., 2011).
The cuttlefish (Sepia officinalis) and the octopus show opposite plasticity sites in their
otherwise similar network synaptic connectivity arrangements. The cuttlefish shows
 Neurophysiological basis of learning and memory in the octopus 81

Figure 4.5 Long-term potentiation (LTP) at the median superior frontal lobe (MSFL) to amacrine
cells synaptic connection. Summary of eight experiments in octopus: the development,
maintenance and saturability of LTP in octopus. LTP was induced by four high frequency (HF)
trains (20 pulses, 50 Hz, 10 s interval), postsynaptic field potentials (fPSP) were normalized
to the averages of 10 test fPSP at the beginning of the experiments. The insets show a
superimposition of LFP traces before (black) and after HF stimulation (gray). See
explanation in the text. (Modified from Shomrat et al., 2011.)

synaptic plasticity at the fan-in and not at the fan-out layer, whereas the octopus has
plasticity in the fan-out and not the fan-in layer. This led to suggest that evolutionary
or self-organizational processes, which shape nervous systems organization, select for
the network’s computational properties rather than for its specific neuronal properties
(Shomrat et al., 2011).

4.5 Mechanism of LTP induction in the octopus vertical lobe

Is activity-dependent synaptic plasticity mediated by associative or non-associative

mechanisms? That is, does the synaptic plasticity in the VL fulfill Hebb’s rule for
associative plasticity, which postulates that the synaptic connection between the pre- and
postsynaptic cells strengthens only when both are simultaneously and sufficiently active.
The N-Methyl-d-aspartate (NMDA) channel, whose discovery was one of the most excit-
ing breakthroughs in modern neuroscience, shows such a coincidence-detecting gating
property. Therefore, the possible involvement of such a Hebbian mechanism in the octo-
pus LTP has been explored. Direct tests for the involvement of an NMDA-like receptor
82 Benny Hochner and Tal Shomrat

Figure 4.6 Input–output relationships in the vertical lobe (VL). (a) Extracellular recordings
from the large cells axon bundles show long-term potentiation (LTP) of the VL output. The
development and maintenance of LTP as measured by the median superior frontal lobe (MSFL)
tract-evoked integrated bundles activity (gray line is a moving average of five test pulses). Upper
insets, superimposition of LTP before (black) and after (gray) high frequency (HF). Lower insets,
activity of the large cells axon bundles before (black) and after (gray) HF. (b) Summary of the
experiments of the type shown in (a). (c) The VL input–output relationship before (black) and
after (gray) LTP induction. The relationship is expressed as the correlation between the
integrated activity in large cells axon bundles and the postsynaptic field potential (fPSP)
amplitude generated by various stimulus intensities. The fPSP amplitudes and the bundles
activity were normalized to those obtained in controls by a test pulse at the intensity used for
LTP induction. (Modified from Shomrat et al., 2011.)

in the postsynaptic current (fPSP) or in LTP induction gave negative results; neither APV
((2R)-amino-5-phosphonovaleric acid), which blocks NMDA-like current in cephalo-
pod chromatophore muscles (Lima, Nardi & Brown, 2003), nor MK-801 blocked these
phenomena (Hochner, Brown, Langella, Shomrat & Fiorito, 2003; T. Shomrat & B.
Hochner, unpublished data).
The crucial experiment to test whether the octopus VL evolved an NMDA-independent
Hebbian plasticity is to completely block the postsynaptic response and then check
whether tetanization leads to LTP induction after washing out the antagonists. The results
from such experiments (summarized in Figure 4.7) show that the LTP in the octopus VL
 Neurophysiological basis of learning and memory in the octopus 83

Figure 4.7 Dependence of long-term potentiation (LTP) induction on postsynaptic response.

Summary of 22 experiments where the response during high frequency (HF) tetanization was
completely blocked by a mixture of 20 mM kynurenate plus 200 µM CNQX and in some
experiments also 200 µM APV. The histogram categories depict three ranges of LTP (expressed
as the percent of total LTP) induced in the presence of the blocking mixture. The distribution
shows a higher proportion of either largely blocked (0–25%) or hardly blocked (75–100%) LTP.
(From Hochner, Brown, Langella, Shomrat & Fiorito, 2003.)

uses both associative and non-associative induction mechanisms. In somewhat less than
half of the experiments, LTP induction was largely blocked by glutamatergic antagonists
(left bar); in the other experiments there was hardly any blocking effect and LTP appears
to have occurred in the absence of a postsynaptic response. Thus, if the two processes
revealed were evenly distributed among the different synaptic connections in the VL, we
would expect a normal distribution of the results. The bimodal distribution in Figure 4.7
can only be explained by the two types of plasticity being segregated in anatomically
different regions, as in the hippocampus (CA3 vs. CA1; Kandel, Schwartz & Jessel,
2000). Such morphological differentiation has not yet been demonstrated in the octo-
pus VL.
Characterizing associative/Hebbian type activity-dependent synaptic plasticity
requires determining whether LTP results from an increase in the amount of trans-
mitter released, or an increase in postsynaptic response, or both, an issue that is yet
not completely resolved in the various hippocampal synapses. Detailed analysis of the
changes in the properties of synaptic transmission accompanying LTP induction in the
VL suggested that the expression of LTP is mainly, if not exclusively, presynaptic. That
is, it occurs in the MSFL neuron terminals at their synapses with the amacrine neurons
(Hochner, Brown, Langella, Shomrat & Fiorito, 2003).
This raises a mechanistic question regarding presynaptic expression of LTP as a result
of strong enough postsynaptic response – the essence of associative/Hebbian plasticity.
The prevailing hypothesis to explain such association is by a retrograde message that is
84 Benny Hochner and Tal Shomrat

transmitted from the postsynaptic to the presynaptic site, the most popular candidate has
been nitric oxide (NO). Because of its gaseous nature NO can cross membranes readily
and rapidly and activates biochemical processes in the presynaptic terminals. Although
the involvement of NO in the induction of presynaptic LTP in mammals is supported by
several experimental results (Garthwaite, 2008), it is still not a completely resolved issue.
It is interesting that although NO was implicated first in mammalian synaptic plasticity,
including LTP, it has become apparent that NO is an important modulator of invertebrates
learning and memory (Susswein & Chiel, 2012), and NO was even implicated in the
octopus learning and memory (Robertson, Bonaventura & Kohm, 1994; Robertson,
Bonaventura, Kohm & Hiscat, 1996). It is conceivable, therefore, that the synapse that
shows that the Hebbian type of LTP induction in the VL (Figure 4.7) may involve a
retrograde messenger, such as nitric oxide, which would transfer the association signal
from the postsynaptic cell to the presynaptic terminals to induce the LTP.

4.6 Neuromodulation in the vertical lobe

The highly dynamic properties of neural networks mediating learning and memory
are not only achieved by the various types of activity-dependent plasticity but also
by neuromodulators. Neuromodulators may feed back negative or positive heterosy-
naptic reward or punishment signals that facilitate or inhibit, respectively, the activity-
dependent (homosynaptic) processes (Bailey, Giustetto, Huang, Hawkins & Kandel,
2000). The involvement of neuromodulators in associative learning is well documented
in insects (Keene & Waddell, 2007), molluscs (Kemenes, O’Shea & Benjamin, 2011)
and vertebrates (Schultz, 2007), and thus has most likely a universal role in complex
neural function.
The search for neuromodulators in the octopus VL started by testing whether serotonin
plays a role in VL function. Serotonin is a well known neuromodulator in mollusks,
involved in both short- and long-term facilitation of the sensory-motor synapse in the
defensive reflex of Aplysia californica (Kandel, 2001). Immunohistochemistry showed
serotonin reactivity in nerve terminals in the VL neuropil but not of cell bodies, indicating
that serotonin indeed may convey signals to the VL from other brain areas (Shomrat,
Feinstein, Klein & Hochner, 2010).
The short-term modulatory effects of serotonin on VL function are shown in Figure
4.8a,b. A dose of 100–200 µM 5-hydroxytryptamine (5-HT) caused an average of
3.5-fold facilitation of the fPSP. This facilitation involves presynaptic modulation
of transmitter release, as it was accompanied by a reversible reduction in twin pulse
facilitation (i.e. reduction in the ratio of the second fPSP amplitude to that of the first
in twin pulse stimulation; Figure 4.8a,c). Repeated exposure to 5-HT did not lead to
long-term modulation in the octopus VL, unlike its long-term effects on the Aplysia
sensory-motor synapse. Also in contrast to Aplysia, c-AMP does not appear to be a
major second messenger in 5-HT induced fPSP facilitation in the octopus VL (Shomrat,
Feinstein, Klein & Hochner, 2010).
 Neurophysiological basis of learning and memory in the octopus 85

Figure 4.8 Serotonin (5-HT) induced short-term synaptic facilitation. (a) Records from one
experiment. Responses to stimulation by twin pulses. fPSP, postsynaptic field potential; Stim,
stimulation artifact; TP, tract potential.) (b) Summary of seen experiments demonstrating the
facilitatory effect of 100–200 µM 5-HT and its reversal on washout. Test stimuli were applied
every 10 or 20 seconds. (c) 5-HT reversibly reduced the level of paired–pulse facilitation
measured as the ratio of the second to the first fPSP amplitudes. Unlike the example shown in
(a), the histogram in (c) includes only experiments in which the first fPSP was apparent. (From
Shomrat, Feinstein, Klein & Hochner, 2010.)

It is intriguing that serotonin has only a short-term robust facilitatory effect on a synap-
tic connection that undergoes long-term activity-dependent facilitation. Therefore, the
serotonergic system in the octopus VL may have been adapted to provide a modulatory
signal to the VL rather than inducing long-term plasticity changes as in Aplysia. The
reinforcement effect of 5-HT on LTP induction (Shomrat, Feinstein, Klein & Hochner,
2010) suggests such a mechanism.
The reinforcement effect of 5-HT was proposed on the basis of the experiment shown
in Figure 4.9. Here, instead of using 4 tetanization trains of 20 pulses (50 Hz), trains
of only 3 pulses were applied. This reduced tetanization intensity induced only a partial
and very modest LTP (19% of the final LTP; Figure 4.9a triangles and 4.9b control). In
contrast, similar triplet stimulation in the presence of 5-HT, which caused a robust short-
term fPSP facilitation, induced a much higher LTP (60.7%) (Figure 4.9a squares and
4.9b 5-HT). Since LTP induction depends on the postsynaptic response or the presynaptic
release processes (Figure 4.7; Hochner, Brown, Langella, Shomrat & Fiorito, 2003), the
simple explanation is that 5-HT augments LTP indirectly by enhancing synaptic activity.
86 Benny Hochner and Tal Shomrat

Figure 4.9 Serotonin (5-HT) reinforced activity-dependent long-term potentiation (LTP)

induction. (a) Partial LTP was induced by the triple-pulse stimulation protocol. 5-HT (gray trace,
open squares) caused large short-term facilitation (measuring the first postsynaptic field potential
[fPSP] in the triple-pulse at 30 min). After 30 min washout (60 min), high frequency (HF)
stimulation revealed the residual LTP. HF stimulation gave rise to greater facilitation of the
fPSPs in the control experiments (black trace, open triangles), indicating less recruitment of LTP
than in the presence of 5-HT. (b) Summary histogram showing the percentage of the final LTP
measured at the end of the experiments exemplified in (a). LTP was induced by the triplet
stimulation protocol with and without 5-HT. (From Shomrat et al., 2010).

That is, the serotonergic system of the VL appears to reinforce the induction of
activity-dependent plasticity and, thus, serotonin may convey reward signals to the
learning and memory network in the VL, as do dopamine in mammals (Schultz, 2007)
and octopamine in insects (Cassenaer & Laurent, 2012).

4.7 Are the vertical lobe and its LTP involved in behavioral
learning and memory?

The octopus VL may be the preparation of choice to investigate how neurons store mem-
ories and how these memories are retrieved or forgotten. The simple VL circuitry may
allow us to clarify how neurophysiological processes are utilized to build a behaviorally
related neural system. Pioneering steps in this direction were made by Young, Boycott,
Wells and their colleagues during the second half of the previous century. Their behav-
ioral, morphological and lesion experiments showed that the octopus VL is important
for learning and memory, but that it is not the sole brain lobe involved in these functions.
The discovery of activity-dependent long-term plasticity and its neuromodulation in the
VL provided physiological support for a VL function in learning and memory. However,
a direct experimental test was required to confirm the involvement of these physiological
processes in learning behavior.
To test the involvement of the LTP in the VL in learning and memory, Shomrat,
Zarrella, Fiorito and Hochner (2008) followed the experiments of Moser, Krobert, Moser
and Morris (1998). These showed that artificial saturation of LTP induced in the rat
 Neurophysiological basis of learning and memory in the octopus 87

hippocampus by tetanization impaired spatial learning when tetanization was applied

prior to learning. This technically challenging experiment in the rat is much simpler in
the octopus, as the VL lies most dorsally in the brain and is relatively easily accessible
to a large electrode for global tetanization. Such tetanization induced on average 56%
of the available LTP.
Now Shomrat, Zarrella, Fiorito and Hochner (2008) could test whether this reduction
in the available LTP induced by the tetanization affected a learning task. This was tested
after 75 minutes of recovery from anesthesia used during the tetanization. The learning
task was a passive avoidance task in which a mild electric shock to the arms “taught” the
octopus to stop attacking a red ball. The results of LTP saturation by the tetanization were
contrasted with the effects of transecting the MSFL tract to the VL, which disconnects
the VL system from its sensory input. In contrast, inducing strong LTP of the synaptic
connection between the MSFL axons terminals and the amacrine cells may be viewed as
“short-circuiting” the VL, because after LTP the same MSFL input will lead to a stronger
VL output via the large cells axonal bundles (see VL circuitry Figure 4.4e and Figure 4.6).
Neither tetanization nor transection eliminated the ability of the octopuses to learn
the task; however, they both affected the rate of learning (Figure 4.10). Transection
slowed the learning rate relative to sham-operated animals (Figure 4.10a). Unexpectedly,
saturating the LTP had the opposite effect, enhancing the learning process relative to
non-tetanized animals (sham controls). Thus, LTP of specific synaptic connections in
the VL do not appear to be involved in short-term learning. The VL output most likely
controls the process mediating short-term learning that occurs outside the VL, with
transection reducing and LTP enhancing the rate of learning by respectively modulating
the VL output.
Testing for long-term memory 24 hours after training produced more easily under-
standable results. Both treatments severely impaired the long-term memory (Figure
4.11). Sham-operated and control animals did not demonstrate perfect memory of the
task, with about 70% of the animals attacking the ball on their first trial. However,
they demonstrated a robust saving or recollection, as they stopped attacking the ball in
the following test trials (Figure 4.11a,b; open symbols). The transected animals (Figure
4.11a) remembered almost nothing of the avoidance task they had learned the day before,
while the tetanized animals showed severe impairment (Figure 4.11b). This difference
was probably due to the tetanization not completely saturating the LTP (see Shomrat,
Zarrella, Fiorito & Hochner, 2008). However, memory acquired before tetanization or
transection, such as to attack a crab or a white ball (associated with positive reward
in pre-training), was not impaired (Shomrat, Zarrella, Fiorito & Hochner, 2008). This
experiment demonstrated clearly that the VL and its LTP are not important for long-term
memory storage, confirming previous experiments showing that previous memories were
not affected by lesions of the VL; similarly to results obtained in mammals and even
humans (Corkin, 2002; Stellar, 1957) with a severed hippocampus.
This finding adds support to the suggestion that short- and long-term memory traces
are stored outside the VL, possibly in the circuitry mediating the attack behavior, and that
the VL plays an important role in controlling the processes involved in the consolidation
of short-term memory occurring elsewhere.
88 Benny Hochner and Tal Shomrat

Figure 4.10 Transection slows and tetanization enhances short-term learning of the avoidance
task. (a) The median superior frontal lobe (MSFL)-transected animals show significantly slower
learning curves than the sham-controls, with a significant difference from the fourth testing trial
(lower panel). Nevertheless, the transected animals stopped touching the red ball significantly
faster than no-shock controls (lower panel). (Lower panels show cumulative Fisher tests between
the different experiments. P in the nth trial was calculated by Fisher’s exact test of a
2 × 2 contingency table (two groups versus two outcomes), in which outcome 1 is the sum of
touches and outcome 2 is the sum of no-touches made by the group from the second to the ‘n’th
trials.) (b) Tetanized animals learn faster than the sham-operated animals; by the eighth trial the
level of cumulative Fisher’s test fell to <0.01 (= 0.0094) (lower panel). (From Shomrat, Zarrella,
Fiorito & Hochner, 2008.)

4.8 Conclusion: a system model for the octopus learning and memory

We conclude by proposing a tentative model for the octopus learning and memory system
that incorporates the physiological and behavioral findings presented here (Figure 4.12).
The rationale behind this model is as follows:

1. Sensory inputs feed in parallel to the VL system and to the circuits controlling
behavior. Transection and tetanization did not affect the behavior; the octopuses still
showed their stereotypical attack behavior.
2. Long-term memory is stored outside the VL. Both tetanization and transection did
not erase old memories.
 Neurophysiological basis of learning and memory in the octopus 89

Figure 4.11 Tetanization and transection impair long term recall tested 24 hours after training.
The animals were given five test trials without electric shock. Testing revealed no significant
difference in long-term memory but impairment in recall in consecutive tests both in transected
(a) and tetanized (b) animals. By the fifth test trial, the experimental animals showed some
retention. Cumulative Fisher’s exact test between the treated and the sham groups and the treated
groups and the non-contingent controls are shown in the lower panels (see explanation in
Fig. 4.10). (From Shomrat, Zarrella, Fiorito & Hochner, 2008.)

3. The output of the VL modulates the rate of short-term learning that takes
place outside the VL-system, possibly by inhibiting the attack mediating circuit.
Tetanization accelerated learning (inhibited the tendency to attack) and transection
slowed learning (increased the tendency to attack). However, both treatments did not
prevent the attack behavior.
4. The LTP in the VL system is crucial for the consolidation of long-term memory
outside the VL system. Both treatments prevented the consolidation of short-term
into long-term memory.
5. Serotonin conveys the reward/punishment signal to the VL which reinforces LTP
of specific synapses active during the signal. Serotonin reinforced LTP induction.

We use the passive avoidance task employed by Shomrat, Zarrella, Fiorito and Hochner
(2008) to illustrate how the model functions. In this task the octopus learns to refrain from
90 Benny Hochner and Tal Shomrat

Figure 4.12 A tentative model for the octopus learning and memory system. See explanation in
text. See plate section for color version.

attacking a red ball (actually dark, because the octopus is color-blind) by receiving an
electric shock to its arms when it attacks. The visual information on shape and brightness
is fed into yet to be delineated attack behavior circuitry that activates the natural attack
behavior. This information is fed in parallel into the MSFL. Each quality (brightness,
shape) is then transferred by different sets of MSFL neurons to the VL, most likely
creating a sparse representation of each sensory quality in the matrix-like connections
of the MSFL neurons with the amacrine interneurons. Those amacrine cells receiving
inputs from both qualities are more likely to undergo LTP due to their higher level of
activity. The “dark-round” association is highly reinforced if they are conjugated with the
pain signal possibly conveyed to the VL by the serotonergic system. The strengthening of
this association during training creates in turn a long-term enhancement of the amacrine
cells input to the set of large neurons driven by this mutual sensory representation.
The VL output generally inhibits the tendency to attack and can be regarded as an
inhibitory supervising signal. The enhancement of the output generated by the red ball
representation alone in the VL is now enough to specifically inhibit attacking the red
ball.
It is not yet clear whether similar mechanisms mediate positive-reward learning, e.g.
by decreasing the inhibitory drive of the VL.

Acknowledgments

The HUJI Octopus Group is supported by the Smith Family Laboratory in the Hebrew
University, the United States–Israel Binational Science Foundation (BSF), the Israel
Science Foundation (ISF) and the European Commission EP7 projects OCTOPUS and
 Neurophysiological basis of learning and memory in the octopus 91

STIFF-FLOP. We would also like to thank Jenny Kien for editorial assistance and
suggestions.

References

Bailey, C. H., Giustetto, M., Huang, Y. Y., Hawkins, R. D. and Kandel, E. R. (2000). Is heterosy-
naptic modulation essential for stabilizing Hebbian plasticity and memory? Nature Reviews
Neuroscience, 1(1): 11–20.
Boycott, B. B. (1961). The functional organization of the brain of the cuttlefish Sepia officinalis.
Proceedings of the Royal Society. B: Biological Sciences, 153: 503–534.
Boycott, B. B. and Young, J. Z. (1955). A memory system in Octopus vulgaris Lamarck. Proceed-
ings of the Royal Society. B: Biological Sciences, 143(913): 449–480.
Budelmann, B. U. (1995). The cephalopods nervous system: what evolution has made of the
molluscan design. In The nervous system of invertebrates: an evolutionary and comparative
approach, Breidbach, O. and Kutsch, W., (eds). Basel: Birkhäuser Verlag, pp. 115–138.
Bullock, T. H. (1984). Ongoing compound field potentials from octopus brain are labile
and vertebrate-like. Electroencephalography and Clinical Neurophysiology, 57(5): 473–
483.
Bullock, T. H. and Basar, E. (1988). Comparison of ongoing compound field potentials in the
brains of invertebrates and vertebrates. Brain Research, 472(1): 57–75.
Bullock, T. H. and Horridge, G. A. (1965). Structure and function in the nervous systems of
invertebrates, San Francisco: W. H. Freeman.
Cassenaer, S. and Laurent, G. (2012). Conditional modulation of spike-timing-dependent plasticity
for olfactory learning. Nature, 482(7383): 47–52.
Corkin, S. (2002). What’s new with the amnesic patient H. M.? Nature Reviews Neuroscience,
3(2): 153–160.
Fiorito, G. and Chichery, R. (1995). Lesions of the vertical lobe impair visual discrimination
learning by observation in Octopus vulgaris. Neuroscience Letters, 192(2): 117–120.
Fiorito, G. and Scotto, P. (1992). Observational learning in Octopus vulgaris. Science, 256(5056):
545–547.
Garthwaite, J. (2008). Concepts of neural nitric oxide-mediated transmission. European Journal
of Neuroscience, 27(11): 2783–2802.
Grasso, F. W. and Basil, J. A. (2009). The evolution of flexible behavioral repertoires in cephalopod
molluscs. Brain Behavior and Evolution, 74(3): 231–245.
Gray, E. G. (1970). The fine structure of the vertical lobe of octopus brain. Philosophical Trans-
actions of the Royal Society of London. Series B, 258: 379–394.
Gray, E. G. and Young, J. Z. (1964). Electron microscopy of synaptic structure of octopus brain.
Journal of Cell Biology, 21: 87–103.
Hanlon, R. T. and Messenger, J. B. (1996). Cephalopod behaviour. Cambridge: Cambridge Uni-
versity Press.
Hochner, B. (2008). Octopuses. Current Biology, 18(19): R897–R898.
Hochner, B. (2010). Functional and comparative assessments of the octopus learning and memory
system. Frontiers in Bioscience, 2: 764–771.
Hochner, B., Brown, E. R., Langella, M., Shomrat, T. and Fiorito, G. (2003). A learning and mem-
ory area in the octopus brain manifests a vertebrate-like long-term Journal of Neurophysiology,
90(5): 3547–3554.
92 Benny Hochner and Tal Shomrat

Hochner, B., Shomrat, T. and Fiorito, G. (2006). The octopus: a model for a comparative analysis of
the evolution of learning and memory mechanisms. The Biological Bulletin, 210(3): 308–317.
Kandel, E. R. (2001). The molecular biology of memory storage: a dialogue between genes and
synapses. Science, 294(5544): 1030–1038.
Kandel, E. R., Schwartz, J. H. and Jessel, T. M. (2000). Principles of neural science, New York:
McGraw-Hill.
Keene, A. C. and Waddell, S. (2007). Drosophila olfactory memory: single genes to complex
neural circuits. Nature Reviews Neuroscience, 8(5): 341–354.
Kemenes, I., O’Shea, M. and Benjamin, P. R. (2011). Different circuit and monoamine mechanisms
consolidate long-term memory in aversive and reward classical conditioning. European Journal
of Neuroscience, 33(1): 143–152.
Lima, P. A., Nardi, G. and Brown, E. R. (2003). AMPA/kainate and NMDA-like glutamate recep-
tors at the chromatophore neuromuscular junction of the squid: role in synaptic transmission
and skin patterning. European Journal of Neuroscience, 17(3): 507–516.
Maldonado, H. (1965). The positive and negative learning process in Octopus vulgaris Lamarck.
Influence of the vertical and median superior frontal lobes. Zeitschrift fur vergleichende Physi-
ologie, 51: 185–203.
Matzner, H., Gutfreund, Y. and Hochner, B. (2000). Neuromuscular system of the flexible
arm of the octopus: physiological characterization. Journal of Neurophysiology, 83(3): 1315–
1328.
Miyakawa, H. and Kato, H. (1986). Active properties of dendritic membrane examined by current
source density analysis in hippocampal CA1 pyramidal neurons. Brain Research, 399(2): 303–
309.
Moser, E. I., Krobert, K. A., Moser, M. B. and Morris, R. G. M. (1998). Impaired spatial learning
after saturation of long-term potentiation. Science, 281(5385): 2038–2042.
Nixon, M. and Young, J. Z. (2003). The brain and lives of cephalopods, Oxford: Oxford University
Press.
Nowotny, T., Huerta, R., Abarbanel, H. D. and Rabinovich, M. (2005). Self-organization in the
olfactory system: one shot odor recognition in insects. Biological Cybernetics, 93(6): 436–446.
Packard, A. (1972). Cephalopods and fish: the limits of convergence. Biological Reviews, 47:
241–307.
Robertson, J. D., Bonaventura, J. and Kohm, A. P. (1994). Nitric oxide is required for tactile learning
in Octopus vulgaris. Proceedings of the Royal Society. B: Biological Sciences, 256(1347): 269–
273.
Robertson, J. D., Bonaventura, J., Kohm, A. P. and Hiscat, M. (1996). Nitric oxide is necessary for
visual learning in Octopus vulgaris. Proceedings of the Royal Society. B: Biological Sciences,
263(1377): 1739–1743.
Sanders, G. D. (1975). The cephalopods. In Invertebrate learning. Cephalopods and Echinoderms,
Vol. 3, Corning, W. C., Dyal, J. A. and Willows, A. O. D., (eds). New York: Plenum Press,
pp. 139–145.
Schultz, W. (2007). Behavioral dopamine signals. Trends in Neurosciences, 30(5): 203–210.
Shomrat, T., Feinstein, N., Klein, M. and Hochner, B. (2010). Serotonin is a facilitatory neuro-
modulator of synaptic transmission and “reinforces” long-term potentiation induction in the
vertical lobe of Octopus vulgaris. Neuroscience, 169(1): 52–64.
Shomrat, T., Graindorge, N., Bellanger, C., Fiorito, G., Loewenstein, Y. and Hochner, B. (2011).
Alternative sites of synaptic plasticity in two homologous “fan-out fan-in” learning and memory
networks. Current Biology, 21(21): 1773–1782.
 Neurophysiological basis of learning and memory in the octopus 93

Shomrat, T., Zarrella, I., Fiorito, G. and Hochner, B. (2008). The octopus vertical lobe modulates
short-term learning rate and uses LTP to acquire long-term memory. Current Biology, 18(5):
337–342.
Stellar, E. (1957). Physiological psychology. Annual Review of Psychology, 8: 415–436.
Susswein, A. J. and Chiel, H. J. (2012). Nitric oxide as a regulator of behavior: new ideas from
Aplysia feeding. Progress in Neurobiology, 97(3): 304–317.
Vapnik, V. N. (1998). Statistical learning theory, New York: John Wiley and Sons, Inc.
Wells, M. J. (1978). Octopus, London: Chapman and Hall.
Young, J. Z. (1971). The anatomy of the nervous system of Octopus vulgaris, Oxford: Clarendon
Press.
Young, J. Z. (1991). Computation in the learning-system of cephalopods. The Biological Bulletin,
180(2): 200–208.
Young, J. Z. (1995). Multiple matrices in the memory system of octopus. In Cephalopod Neurobi-
ology, Abbott, J. N., Williamson, R. and Maddock, L., (eds). Oxford: Oxford University Press,
pp. 431–443.
Zullo, L., Sumbre, G., Agnisola, C., Flash, T. and Hochner, B. (2009). Nonsomatotopic organiza-
tion of the higher motor centers in octopus. Current Biology, 19: 1632–1636.
5 The octopus with two brains:
how are distributed and central
representations integrated in the
octopus central nervous system?
Frank W. Grasso

5.1 The relationship between cognition and neuroscience

in cephalopod cognition

Ο δε. πολυ, πουσ αννο, ητον µε, ν ενστι και. γα.ρ προ.σ τη.ν χει/ρα ␤αδι, ζει του/ αννθρ␻, που
καθιεµε, νην καθιεµε, νην.
The octopus is a stupid creature, for it will approach a man’s hand if it be lowered into the water.
Aristotle, The History of Animals, Book IX, translated by D’Arcy Wentworth Thompson
(2013).1

Today, two-and-a-quarter millennia after Aristotle denigrated octopuses’ intelligence

for behaviors that might be called “curiosity,” this volume reflects the attitudes and
accumulated evidence of our time in asserting that cephalopods possess cognition. And
perhaps, with all due respect to the great philosopher whose writings were the foundation
for the scala naturae behind medieval scholasticism, his definition of cognition would
be too anthropomorphic to admit an intelligence that did not agree with the typical social
conceptions of intelligence in humans. The definition of cognition that I will use in this
chapter is broader: any adaptive process which connects sensing in the world to action in
the world. The word adaptive in this definition requires some elaboration. I use adaptive
in the sense that it is used in the field of cybernetics: that of a process that changes
with experience. It is not the intergenerational sense of adaptation from evolution but
something that happens within the individual during its lifetime.
In the field of cybernetics a non-adaptive or fixed control system reliably produces the
same output every time the same input is provided. The behavior is “machine-like.” In
biological neural systems there are four types of processes that lead to adaptive behavior.
First are developmental processes in which new components, like limbs, muscles and
neurons via neurogenesis and neural systems are added as the animal ages. The other
three change the state of the central nervous system (CNS) without the addition and
integration (or maturation) of new parts. Hormonal systems modify the performance

1 I am indebted to Dimitris Tsakiris for drawing my attention to this passage and translation.

Cephalopod Cognition, eds A.-S. Darmaillacq, L. Dickel and J. Mather. Published by Cambridge University
Press. 
C Cambridge 2014.
 The octopus with two brains: the octopus central nervous system 95

of targeted functional groups of neurons in activational and organizational modes. In

octopus the action of hormonal systems on brain operation and the development of
neural systems are areas in need of a large amount of basic research. The other two have
better developed methods and a larger knowledge-base. Synaptic plasticity changes the
strength of connections between neurons. This is what is classically2 termed learning and
its product is classical “memory.” And, finally, there are internal recurrent connections
(feedback loops) between neurons which cause inputs received some time in the past to
persist in influencing behavior after the stimulus is no longer present. The cybernetics
sense of adaptive systems has focused on the latter two, synaptic plasticity and recurrent
connections. Broadly speaking, synaptic adaptation is a mechanism for long-term mem-
ory while recurrent connections are a mechanism for short-term memory. This chapter
emphasizes the role of recurrent connections in the octopus.
This definition of cognition may seem too broad at first but it has several advantages
when attempting to apply the concept to invertebrate behavior. First, the use of the word
“process” makes clear that cognition is about information processing, for which we have
many useful concepts and tools. In dealing with non-human species we are not concerned
with the refined discriminations from human social and cultural systems found in the
study of human psychology but with the biological foundations of intelligent behavior.
So secondly, the definition relates to behavior directly and not to inferred internal
states. Further, the effective adjustment of behavior is an effective selection force in the
evolution of cognitive systems and such internal states are only a means to that end.
In this chapter the term adaptive indicates that cognition is about the role of individual
experience in shaping behavior.
Perhaps the most important aspect of this definition is that it usefully points us towards
mechanism. This follows from the paragraph above, which identified development,
hormones, plasticity and recurrent network architecture as defined biological processes
that could change the brain in ways which might result in adjusting individual behavior
based on experience. In an individual line of research, instances of these mechanisms
may be profitably traced in the brains of real cephalopods to their influence on the
unfolding of the behavior of the individual animal over its lifespan. Any explanation
which cannot account for (or exclude) the contributions of these will be partial.
The study of mechanism is the bridge between cognitive processes as we conceive
them and the neurobiological reality. Many theoretical mechanisms may reasonably
reproduce a description of behavior. For this volume’s stated purpose of understanding
cognition in cephalopods we should be interested in discovering the mechanism not
just a mechanism that fits a description of behavior. To do this cephalopod cognition
studies are best served when the theories of cognitive processing can be excluded or
retained based on neurobiological facts. Discovery of the unique cognitive innovations
of cephalopods depend critically on such well-constrained approaches.

2 Adaptive processes are those that might be construed loosely as “memory” processes. They all result
in changes in behavior as a result of experience. The psychological literature uses the term narrowly in
that memory has come to be used almost exclusively for processes associated with synaptic plasticity.
Cybernetics uses the definition that changes to the system that produce different responses to the same input
at different times are adaptive. My inclusion of hormonal and developmental processes is broader still.
96 Frank W. Grasso

Research in recent years has uncovered valuable insights into the synaptic plasticity of
octopus that dovetails with research in other simpler-brained mollusks, like Aplysia and
gastropods, and with learning mechanisms in primitive cephalopods, like Nautilus. And
we possess an abundance of detailed neuroanatomical data that continues to grow with
new techniques. This chapter will use some of that architectural knowledge to argue for
a unique, decentralized information processing mechanism in octopus.

5.2 The problem of soft bodies controlled by complex brains

The neural representations that control and coordinate the behavior of cephalopods are
great scientific value. Their soft bodies, in contrast to those of vertebrates and arthropods,
which effect movement through the control of rigid (endo- or exo-) skeletons, suggest
that cephalopods have new things to teach us about ways that brains produce action
on the world that is useful to the animal. This chapter will not explain those principles
because much research is required before they become known. It will, however, frame the
problem of understanding neural representations in cephalopods and suggest one idea on
how a special balance of centralized and distributed processing might occur in octopus.
Octopuses offer an important group for direct functional comparisons between soft-
bodied and rigid-bodied creatures. They use and coordinate their eight arms in tasks
involving dexterous manipulation that are functionally similar to tasks performed by
rigid-bodied species. They can learn to open jars and uncork bottles (Fiorito, Von
Planta & Scotto, 1990; Hanlon & Messenger, 1996), they naturally open bivalve shells
(Fiorito & Gherardi, 1999; Anderson & Mather, 2007), and touch and fetch objects within
reach of their arms (Sumbre, Gutfreund, Fiorito, Flash & Hochner, 2001; Sumbre, Fiorito,
Flash & Hochner, 2006). The biomechanics that support these and other behaviors are
suitably different from those of rigid-bodied species (Hochner, Shomrat & Fiorito,
2006; Grasso & Basil, 2009), but are the central representations that the octopuses use
convergent on a common solution or, also, uniquely cephalopod in character? The bodies
of cephalopods are markedly different from those of arthropods and vertebrates, but
what of their CNS? The common ancestors of cephalopods, vertebrates and arthropods
possessed muscles, specialized sensory receptor, specialized motor neurons and central
interneurons between the sensory receptors and motor neurons. The common ancestors
of vertebrates and mollusks used neurons for the control and coordination of behavior.
Cephalopods are mollusks that separated from the molluscan line about 505 million
years ago. So, when compared with arthropods and vertebrates, cephalopods offer an
opportunity to evaluate the interplay of convergent selection pressures and phylogenetic
inertia on the evolution of brains as information processing systems.

5.3 Representations and connectivity

David Marr specified three levels of analysis that needed to be understood in order to
understand a cognitive process (Marr, 1982). They are theory, or what is being computed
by the system; representation and algorithm, or a specification of the building blocks
 The octopus with two brains: the octopus central nervous system 97

of the theory and rules as to how they can interact; and implementation, or the actual
physical apparatus in which the process could be realized in the world. He pointed out
that confusing these levels was a path to faulty understanding and that without a clear
idea of the theory neither the representation nor the implementation could be realized.
So it is essential, in this chapter on representation in the octopus CNS that we have our
architectural components well defined.
In describing neural architecture, neurobiologists identify three classes of neuron:
primary sensory neurons which transform environmental stimulus energy into neural
information (e.g. spikes), motor neurons which transmit control signals to muscle fibers
via neuro-muscular junctions to produce behavior and interneurons which lie in between
and process information. Figure 5.1 provides a primer on these concepts and a graphical
way of representing neural system architecture.
Figure 5.1a shows a complete wiring diagram for a simple system containing primary
receptor neurons and no interneurons or motor neurons. On the left side is a sketch of
an abstracted animal body showing a physical arrangement of muscles and sensors. To
the right of this sketch is the corresponding neural connectivity diagram (Anastasio,
2010). Assume that activation of muscles on the left and the right sides of the animal
are competent to steer the animal left or right or propel it forward depending on the
level of left or right muscle activation. The primary sensory receptors (gray soma)
each form a single neuromuscular junction on their respective muscles. The direction
of information flow is indicated by the arrow beside the soma. It is interesting to note
that this simple mechanism is competent to produce a taxis behavior (e.g. phototaxis,
chemotaxis, phonotaxis, etc.) as noted by Braitenberg (1984) by virtue of the crossed
connections which would produce more “propulsion” on the side opposite that with
greater stimulation. When the input from the two sides is equal then forward movement
would result instead of steering.
Figure 5.1b shows an imaginary animal of slightly greater neural complexity that adds
committed motor neurons (gray soma), internal synapses between the primary sensory
neurons and the motor neurons. Note the convention in the wiring diagram on the right
of orienting progressive layers of neurons at right angles. The synapses are noted as
circles at the intersection of output lines (i.e. axons) and input lines (i.e. dendrites).
The absence of a synapse marker indicates a potential connection that is not made,
a zero-valued connection. Note that this architecture also results in a mechanism that
can competently produce a taxis behavior without any central representation. The only
difference in behavior between Figure 5.1b and that in Figure 5.1a would be a delay in
response to the input owing to the delay in signal transmission through the synapses.
Otherwise the taxis behavior would be comparable.
Figure 5.1c is a simple configuration which, perhaps debatably,3 includes a central
representation. It is not a brain but this single layer of two interconnected interneurons
(gray soma) is the closest thing that this imaginary animal has to a brain occupying, as

3 As this was just an introduction for the specific treatment of octopus arm connectivity that follows, I’ve
not mentioned the important issue of the strength of the connections, or synaptic weight. This is a key
component of neural network performance which has received a large amount of theoretical consideration.
For a thorough general introduction see Anastasio (2010).
(a)

Sensory Sensory
receptor receptor
left right
Sensory receptor Sensory receptor
right left

Muscle Muscle Neuromuscular

left right junction
Left Right
muscle muscle

(b)
Synapse/
Sensory receptor
connection
left
Sensory Sensory
receptor receptor
left right
Sensory receptor
right

Motor Motor
Muscle Muscle neuron left neuron right
left right

Left Right
muscle muscle

(c)
Sensory Sensory Synapse/
Sensory receptor
receptor receptor connection
left
left right

Interneurons

Sensory
receptor
right
Motor
Muscle Muscle neuron right
left right

Motor
neuron left

Figure 5.1 Schematic representation of behavioral control schemes for an abstracted bilateral
creature. In each panel the left side shows the arrangement of neurons and muscles in the body,
the right the central nervous system connectivity pattern following the conventions used in this
chapter. (a) A hypothetical bilateral creature with sensory receptor cells connected directly to
muscles via a neuromuscular junction. The connections crossing at the midline cause the side of
the body with greater stimulation to stimulate greater activation on the contralateral side. This is
sufficient to produce a taxis behavior. (b) The addition of motor neurons requires the addition of
synapses, connections, between the neurons in the network. There will be no difference in taxis
behavior. (c) The addition of interneurons – neurons that are neither sensory neurons nor motor
neurons. The interneurons are interconnected. This is evident in the right connectivity schematic,
which shows feedback connections from the interneurons to each other and themselves. These
feedback connections cause past inputs to persist in the network. The
network will produce taxis behavior but that behavior will depend more on the state of the
entire network and its history (memory) of inputs.
 The octopus with two brains: the octopus central nervous system 99

it does, the brain’s position between the primary sensory and the motor neurons. The
addition of the interneurons changes the character of the network performance and the
behavior of the imaginary animal because of the recurrent connections. Note the four
additional synapses (compared with Figure 5.1b) on the connections that loop back from
the outputs of each interneuron. These supply each interneuron with the new, additional,
inputs of a copy of its own output and with the output of the other interneuron. This
is a recurrent form of “memory,” in the mathematical sense of hysteresis, which is
not in agreement with classic psychological definitions of memory. This computational
architecture corresponds to the neurobiological structure of a neuropil where neurons in
the CNS interconnect. In the neuropil the previous (historical) past activity patterns are
carried through loops to “reverberate” between them, maintaining patterns of activity
after the stimulus has ceased to activate receptor(s). Thus, the output of the interneurons
to the motor neurons and, therefore, the behavior of the hypothetical animal will depend,
in part, on the past pattern of inputs and in part on the current inputs. Brain complexity
grows (in the developmental and evolutionary sense) from the kernel of a single neuropil,
through the multiplication of neuropils, their specialization and organization. This is one
of the key building blocks of brain organization.
It is important to note that that in the third imaginary animal the change in behavior
with experience does not depend on synaptic plasticity. So, in a larger network with
proper arrangement of connectivity and connection strengths a “memory” (again in
the mathematical, reverberating sense) can be produced in the interneurons, which will
bias the performance of the network and, therefore, the behavior of the animal with
experience. The permanent changes in synaptic strength that are considered in classical
learning, when applied to these recurrent networks, could tune, adjust or even obliterate
these reverberating memories, but the parallel existence of both forms of experience-
derived performance modification indicates that an understanding of behavior must take
both into account.
It is commonly accepted that interneurons are the elementary biological units for
representing the world in the CNS of animals for the control and for the coordination of
the behavior of animals. Removed as they are from primary sensory neurons, interneu-
rons can potentially meet the important requirement of a representation in that their
activity persists after a stimulus has ceased to be present. With a recurrent connectivity
pattern between them or synaptic plasticity in the connections, interneurons can acquire
representations with experience of the world that can be stored nervous systems. The
nature of the representation depends on which mechanism is used to form the represen-
tation. Another key question is whether the representations formed in a nervous system
are localized or distributed. Interneurons can be arranged to do either based on their
connectivity and so, for clarity, I will review the organization of the CNS of the octopus
as it pertains to the question of neural representations to control soft-bodied animals.

5.4 The organization of the octopus central nervous system

By invertebrate standards the CNS of cephalopods is enormous. The octopus CNS con-
tains on the order of 5 × 108 neurons (Young, 1971). This number excludes the number
100 Frank W. Grasso

of peripheral neurons in glandular, sensory, chromatophore, digestive and respiratory

systems, which is larger still in number. This size is orders of magnitude larger than that
of the next-largest mollusks. This quantum leap in size is undoubtedly responsible in
part for the huge, individually adaptable behavioral repertoire of cephalopods.

5.4.1 The cerebral ganglia

The large size of the brain might reflect the complex requirements of controlling a
soft body. Rather than having muscles that control a finite and fixed set of joints,
cephalopods must operate muscles that overlap to reshape body surfaces at arbitrary
location – to effectively form joints where they are needed. This increased complexity
would be expected to require larger numbers of motor neurons and more motor control
nuclei and lobes in the brain. This special demand for motor control circuitry, compared
to rigid-bodied animals appears to contribute to the size of the brain as it does in
vertebrates (Finlay, Darlington & Nicastro, 2001; Herculano-Houzel, Collins, Wong
& Kaas, 2007). However, it does not account for the whole of the brain which is of
considerable complexity with 35 lobes (Young, 1971; Grasso & Basil, 2009), only some
of which are directly responsible for motor control. It appears that learning, in the
classical psychological sense, is a specialty of the octopus brain.
The exceptional learning abilities of the octopus (in the classical psychological
sense) are well documented (Wells & Wells, 1957b; Mackintosh & Mackintosh, 1963;
Wells, 1964; Mackintosh, 1965; Fiorito & Scotto, 1992; Fiorito, Agnisola, D’Addio,
Valanzano & Calamandrei, 1998). Among the 35 lobes of the octopus brain the vertical
lobe, for example, is considered a committed central memory system. It has been the
subject of focused studies of learning (Wells & Wells, 1957b) and recently in neurophysi-
ological studies (Shomrat, Zarrella, Fiorito & Hochner, 2008; Shomrat et al., 2011; Zullo
& Hochner, 2011). It has been demonstrated to be a site of long-term potentiation (LTP)
and synaptic facilitation in both octopuses and cuttlefishes (Shomrat et al., 2011). The
brains of octopuses are sophisticated cognitive systems, which have led some researchers
to suggest the species as a model system for the study of consciousness (Edelman &
Seth, 2009).
The neurons in the higher order neuropils of the brain, those that have no obvious
impact on behavior, appear to serve the acquisition of new behaviors – classical learning
and memory systems (Wells & Wells, 1957b). These findings and analysis suggest strong
analogies with the centralization of neural representation such as the presence of cogni-
tive maps in the brains of vertebrates and function in the “brains” of the cephalopods. Yet
efforts to discover a somatotopic representation or other maps in the brains of octopuses
have found none (Zullo, Sumbre, Agnisola, Flash & Hochner, 2009; Zullo & Hochner,
2011). Neurons in those areas seem to be strongly multimodal compared to their verte-
brate and arthropod counterparts. A different overall brain representation of the world
and of motor control compared to vertebrates is suggested (Zullo & Hochner, 2011).
An indication of the source of this difference is found in the fact that three-fifths
(3.5 × 108 ) of the neurons in the octopus CNS are located outside the central brain,
or cerebral ganglia as it is called. These neurons are located in the brachial ganglia that
 The octopus with two brains: the octopus central nervous system 101

extend down the length of the arm and in the plexus of interconnections between the
arms. This large collection of neurons undoubtedly contributes locally to motor control
in the arms but the organization of the ganglia suggests that these extracerebral portions
of the CNS are serving other functions as well.

5.4.2 The brachial plexus and sucker ganglia

The CNS in the arms is organized to support the functions of the 300 suckers of the
arm and the muscles of the arms. There is a central nerve cord running the length of
each arm and interconnecting the arms. It contains nerve fibers originating outside the
arm in the brain and interarm plexus, as wells as cells and neuropil throughout its length;
however, there is a significant expansion of the cord, or ganglion, associated with each
sucker, delineated by an increase in the number of neurons and the size of the neuropil.
Technically, when a set of ganglia is interconnected, it is called a plexus. So I will refer
to the interconnected nerve cords of the eight arms as the brachial plexus, and use the
shorthand term of brachial ganglion when referring to portions of it such as a whole arm
or portion or a selection of ganglia within an arm. Four smaller nerve cords that run the
length of the arm and are more eccentrically located are important for motor control of
the arm, but their role will not be discussed in this chapter which focuses on sensory
representations. Outside each brachial ganglion there is a sucker ganglion associated
with each sucker that lies eccentric from the central nerve cord just over each sucker and
outside the main arm muscle mass (Graziadei, 1971).
The connectivity of the arm CNS structure is summarized in Figure 5.2. Sensory
input to the arm CNS comes from the rims of the suckers and proprioceptors embedded
in the muscular tissues of the arm and sucker. The rim of each sucker houses on the
order of 104 chemo- and mechanoreceptors (Graziadei & Gagne, 1976). The axons from
these receptors pass from the sucker rim to the sucker corresponding sucker ganglia. The
proprioceptors of the sucker also pass in nerves to the sucker ganglia. The sucker ganglia
contains motor neurons so local reflex arcs can result from information processing there.
Each sucker ganglion is associated with a single sucker. It makes reciprocal connections
with its corresponding brachial ganglion. At present the mode of operation of the sucker
ganglia is not well investigated.
Sucker ganglia do not communicate directly with one another. Connections between
brachial ganglia exist to channel the information from one sucker to its neighbors.
Another role of the brachial ganglia is to control and coordinate the muscles of the arm
and thereby effect arm movements by means of intrinsic motor neurons. Motor neurons
in the brachial ganglia and sucker ganglia innervate muscle fibers in the arm and sucker
respectively (Graziadei, 1971; Rowell, 1966). Proprioceptors embedded in the main arm
(outside the suckers) send processes to the four smaller, eccentric nerve cords and the
central nerve cord (Graziadei, 1965). These four eccentric nerve cords contain motor
neurons of their own but their role is not well studied. Somehow the processing in the
brachial ganglia and the four eccentric ganglia produce the reshaping of the arm, which
provides it with its functional flexibility.
102 Frank W. Grasso

Descending cerebral connectives

Inter-arm connectives

Brachial Brachial Brachial

ganglion ganglion ganglion

Arm Arm
proprio- muscles
receptors
Sucker Sucker Sucker
ganglion ganglion ganglion
Proximal

Distal
Rim Sucker
receptors muscles

Segment N − 1 Segment N Segment N + 1

Figure 5.2 The organization of the ganglia in the octopus arm. Around each sucker is a section of
the arm whose neural and muscular structure repeats along the length of the arm. The central
segment, N, shows the complete set. Sensory inputs (gray) in the form of chemo-receptors on the
sucker rim and propioceptors embedded in the tissues of the arm provide input to the sucker and
brachial ganglia respectively. The brachial ganglion also receives inputs from the sucker
ganglion in its segment, descending inputs from the cerebral ganglia, inputs from the other arms
and several proximally and distally positioned brachial ganglia. The brachial ganglia process
these sources of input and produce motor commands to the muscles of the arm. A shorter
sensory motor loop exists from the sucker ganglia (which also processes input from the brachial
ganglia in its segment) to the sucker muscles. The proximal and distal segments (delineated by
dashed boxes) are included to illustrate the continuum nature of this organization proximally and
distally down the arm.

The summary shown in Figure 5.2 makes clear the modular nature of the arm con-
nectivity. I refer to these modules as local brachial modules (LBM) and they contain
the neural components of each sucker–ganglion/brachial–ganglion pair (i.e. the primary
receptors [chemo-, mechano- and proprioceptive], the motor neurons and the interneu-
rons). In Figure 5.2 and the following text I use the term segment to refer to a section of
the arm (i.e. muscle, connective tissue, non-CNS neural-components) associated with a
given LBM. This basic organization repeats down the length of the arm, one module for
each sucker. It is likely that the circuitry for controlling the arm movements forms a con-
tinuum down the arm that matches the continuum of tissue. The continuous connectivity
of the brachial ganglia in the arm naturally reflects this. Evidence for arm-sucker coor-
dination supports suggests that a modular view of the functional organization of the arm
might be appropriate for even considering the circuitry responsible for the movement of
the arm.
 The octopus with two brains: the octopus central nervous system 103

Table 5.1 Rowell’s (1966) summary of the numbers of fibers originating

from and entering into the brachial ganglia in one module

Neuron type No. of fibers

Motor neurons 2000

Sensory cells 3000
Afferent axons in the axial cord 18 000
Efferent, chromatophore fibers 2300
Efferent, non-chromatophore fibers 2000

There are on the order 105 central neurons in each segment module. This seems a very
large number of neurons when one reflects on the fact that this figure excludes the larger
numbers of peripheral neurons that contribute to the chromatophore, circulatory and
secretory systems and the primary sensory receptors. There are more neurons present
in one segment of an octopus arm than exist in the whole of successful invertebrate
animals such as the medicinal leech or Aplysia. In terms of the numbers of neurons, a
module is on a comparable scale to the size of an arthropod brain. What requirement
could produce such a commitment of neural resources?
Table 5.1 gives an indication. It lists Rowell’s (Rowell, 1966) summary of the numbers
of fibers originating from and entering into the brachial ganglia in one module. This
makes clear that 75% of the signal lines in a given module carry neither primarily
sensory, nor primarily motor, nor chromatophore signals in nature; rather the majority
of fibers carry information to and from other interneurons. Not only are the numbers of
neurons in a module comparable to those of a whole organism, each segment is organized
like the brain of a living organism (in at least the sense of Figure 5.1), with a diversity of
sensory modalities, motor neurons effecting different motor systems and large central
neuropils, which are processing centers for large amounts of information.
Yet, while a segment may resemble a brain functionally it is not a brain because by
definition brains contain anatomically delineated areas of specialized function, such as
the 35 lobes of the octopus brain. This in fact is an approximate definition of a ganglion:
a collection of interconnected interneurons. In this sense the module compares to a
segment in the vertebrate spinal cord. A recent quantitative analysis of the spinal cord
of the Red Eared turtle found that one spinal cord segment (D3) contained on the order
of 104 neurons, an order of magnitude fewer than in the octopus arm module, of which
about 75% were interneurons and 25% were motor neurons (Walløe, Nissen, Berg,
Hounsgaard & Pakkenberg, 2011). And, like the vertebrate spinal cord, the chain of
brachial ganglia form a linear series. The important structural difference is that there
are about 10 times more brachial ganglia in the arm of an octopus than there are spinal
segments in a vertebrate like a human (31 versus 300), and that the chain of brachial
ganglia in one octopus arm is connected through the brachial plexus to form a single
continuous system with the other seven arms. It is also important to note that this huge
investment in neurons and connectivity forms a network of ganglia of a more or less
homogeneous type.
104 Frank W. Grasso

5.5 The computational roles of the arm module

One way to think about the arms of octopuses is as a device for delivering suckers to a
work surface. In object manipulation and benthic locomotion the suckers play a key role.
Arms are used for other behaviors such as swimming, anchoring, sensory exploration
(Mather, 1998) and display (Packard, 1961), but I will omit a complete discussion of the
uses of the arms here. The wrapping the arm around an object without sucker attachment
has rarely, perhaps never, been seen. Generally, when an octopus uses its arms for
manipulation of objects or for locomotion the suckers are attached to a surface and
coordinated movements of the arm effect a movement of the object relative to the
animal’s body (small objects) or of the animal’s body relative to the surface (objects
more massive than the animal).
The coordination of the sucker attachment and orientation with arm movement, which
is essential for proper locomotion and object manipulation, must be a part of what the
brachial ganglia do. There is good evidence for intersucker coordination (Rowell, 1963,
1966; Altmann, 1971; F. W. Grasso, S. Nair, J. Barocas & S. Hadjisolomou, unpublished
data) and for arm–sucker coordination (Grasso, 2008), both of which must be mediated
by the brachial ganglia for the anatomical reasons stated above. Suckers themselves
are not passive agents. They have the ability to rotate and extend relative to the arm
in addition to attaching to surfaces, thanks to their extrinsic muscles (Kier, 1982).
Doubtless, this is part of the computational load of the sucker ganglia and the brachial
ganglia in combination.
In addition to the local information processing that anatomy and behavior demonstrate,
descending influences from the cerebral ganglia contribute to the coordination of arm
movement. Visual information has been shown to guide reaching and manipulation tasks
(Gutfreund, Flash, Fiorito & Hochner, 1998; Grasso, 2008; Gutnick, Byrne, Hochner &
Kuba, 2011). Integrating these and other descending influences to guide the activity of
the arms must form part of the computational load borne by the local brachial module.
But this is a condition of influence and not a necessary condition for behavior.
Various studies show that blind and even decerebrated octopuses can perform tactile
discrimination and manipulation tasks (Wells, 1964; Wells & Wells, 1957a, 1957b) and
that much of the postural and feeding behavior of the animal appears “normal” in the
decerebrate condition (Wells, 1978). Figure 5.3 shows an intact Enteroctopus dofleini
solving a tactile Y-maze task to obtain food. In this case the maze is opaque and the
animal cannot see inside. It has to rely on the information provided by the sucker
receptors and the coordination of the local brachial modules to solve the maze (F. W.
Grasso, unpublished observations). These results indicate that the network of local
brachial modules can operate independently from visual and perhaps all descending
influence. It is known that local accept and reject reflexes persist in the suckers of a
disconnected octopus arm (Altmann, 1971).
Perhaps, in light of the number of computational roles served by the LBM, it is
unsurprising that the investment in neural resources in a single LBM is comparable to
arthropod brain or vertebrate spinal cord. However, there is perhaps, an additional and
cognitive role that a network of ganglia such as the LBM can perform in the octopus:
 The octopus with two brains: the octopus central nervous system 105

Coming to the entrance Touching the first piece Removing second piece

Removing third piece Removing fourth piece Removing the last piece

Figure 5.3 Flexible grasping and object manipulation by an octopus arm. The octopus in this
series cannot see inside the Y-maze and must rely on touch and taste sensors in its arm to locate
the five food items (pieces of shrimp abdomen) inside. The progression of slides shows the
reshaping of the arm to produce bends at different orientations and at different positions along
the arm as needed to achieve the grasping and retrieval task.

The network of LBMs may be capable of forming representations of the external world.
This proposal is made based on two facts. First, that the sensory information about the
external world that is available to an LBM comes from the sucker rims. And second that
the sucker rims necessarily form a topologically-ordered spatial array. Surface contact is
required for the chemical or mechanical information to be transmitted so a given LBM is
excluded from sampling the same location as another and the sequence of suckers from
proximal to distal ensures an orderly (though not necessarily Cartesian) arrangement of
sample points.

5.6 A network of homogeneous ganglia

5.6.1 An analysis of ganglia network functions

The proposal of a representation outside the cerebral ganglia presented above poses
a daunting technical challenge: where amongst the approximately 3.5 × 108 neurons
associated with the arms should we look for this chemical or tactile representation
of space? The organizing principle of the LBM helpfully narrows the focus to about
50 000 neurons but this does not help much when we remember that there are about
2400 LBMs in a typical octopus. The literature does not presently contain the types
106 Frank W. Grasso

of simultaneous multisite recordings from octopus brachial ganglia that would inform
us on this (though Altmann, 1971; Rowell, 1963, 1966, and Young, 1991 give useful
indications). Instead a mathematical analysis grounded in the specifics of the system can
provide a demonstration of the concept and supply predictions for what we might look
for in future empirical studies.
Figure 5.4 is a connectivity diagram using the same conventions employed in
Figure 5.1. It shows three LBMs interconnected in an ordered series. It could be mathe-
matically extended to the 300 LBMs found in an octopus arm, but three will suffice for
this illustration. The labels N–1, N and N+1 throughout the figure refer to the identity of
the LBM. The network nodes labeled sucker sensory input N, the propriosensory input
N, the sucker ganglion N and the brachial ganglion N, all constitute the same LBM (N) as
appears in Figure 5.2.4 The same applies for N–1 and N+1, which are, respectively, one
LBM proximal and one LBM distal to N. The aim of this depiction, which rearranges
the clear organization of an LBM shown in Figure 5.2, is to make clear the nature of this
network as a short-term memory system, which will unfold in a series of steps in the
following paragraphs.
First, note that each node in the network represents a group of neurons and not actual
single neurons as implied in the same sort of diagram in Figure 5.1. For the purpose of
this chapter the same network principles apply for a node whether it represents a neuron
or a collection of neurons. Thus for LBM N the sucker sensory inputs (Ss ) represent all
the sensory neurons on the rim of sucker N, as similarly all the propriosensory inputs
(Sa ) represent all of the proprioceptors in the LBM N. These sensory units supply inputs
to the sucker ganglia and brachial ganglia, respectively. The groups of neurons in these
ganglia are represented as single units as well. Thus, this figure and analysis are of a
network of pools of neurons or ganglia.
The interpool connections between the sensors and the ganglia are placed on gray
background fields to suggest the two separate neuropils in which these synapses exist
in the octopus. Mathematically, both types of sensors are at the same level of process-
ing (primary receptors) so their neuropils are represented as a single connectivity (or
weight) matrix Wsg , which represents the connections between the sensors and the gan-
glia.
The connections in the Figure 5.4 for Wsg are presented in Table 5.2 in matrix form
where the rows represent the source of the signal (from) and the columns represent the
target of the signal (to) thus from sensors to ganglia is Wsg .
In Table 5.2, a value of 0 indicates no connection. The letters a and b indicate
connections of two different strengths with values other than zero. The values of these
connections will be discussed below. Note that each sucker sensor (Ss ) provides input
to just one ganglion (in its own LBM). The off-diagonal matrix entries labeled h for the
arm inputs reflect the fact that proprioceptors located between given brachial ganglia

4 The motor output units could have been arranged according to this convention. Since the motor units are not
essential to the representation question they were not enumerated in order to simplify the diagram. Matrix
mathematics aficionados will see that the equations represent them as vector outputs to preserve the LBM
identity.
 The octopus with two brains: the octopus central nervous system 107

Sucker sensory Wsg

neurons (Ss)
N –1
N –1

N
N
N+1
N+1
Arm propriosensory
Wgg neurons (Sa)

Brachial ganglia
interneuron pool
N–1 N N+1 (GB)
N–1 N N+1

Sucker motor neuron pool

(Ms)

Sucker ganglia N
interneuron pool
(GS) N

Wgm Arm motor neuron pool

(Ma)

Figure 5.4 A schematic network representation of the segmental organization of the octopus arm.
This figure represents the components of three segments of the arm as N–1, N and N+1 for the
segment of interest and two adjacent suckers. The diagram could be extended to the 300 suckers
of the arm but is reduced to this for convenience of illustration. Two sources of sensory input (s)
are represented the chemo- and mechano-sensory inputs from the sucker rim are represented
pooled as ss and the proprioceptive sensory inputs of the arm as Sa . The motor outputs are
schematized as arm Ma and sucker motor outputs Ms . The key feature in this diagram is the
feedback matrix, Wgg , through which past network states and activations my continue to affect
the output of the system (movements of the arms and suckers) at times after the application of
the stimulus. Circles at the intersection of neuron inputs and outputs indicate a connection. The
sign and magnitude are not indicated. The connectivity between, in the weights matrices (W’s:
Wsg , Wgg and Wgm ) follows that described by Graziadei in Young (1971). In this diagram neurons
are represented as polarized with an input (dendritic) line with connections, an integrating unit
(soma) and output (axon) marked with an arrow. The sum of the input lines is assumed to be
computed at the integrating unit and transmitted through the output.

provide input to both. The influence is known to spread farther than adjacent LBMs, but
for the purpose of this analysis a single LBM lateral connectivity will not detract from
illustrating the principles.
Each sucker ganglion receives input only from its sucker rim. The brachial ganglia,
Gb,n receive input from the immediate vicinity but may also receive direct sensory input
from adjacent areas. The values of the connection strengths, a and b reflect the idea
108 Frank W. Grasso

Table 5.2 The connections in the Figure 5.4 for Wsg in matrix form

GS (n–1) GS (n) GS (n+1) GB (n–1) GB (n) GB (n+1)

SS (n–1) a 0 0 0 0 0
SS (n) 0 a 0 0 0 0
SS (n+1) 0 0 a 0 0 0
Sa (n–1) 0 0 0 b h 0
Sa (n) 0 0 0 h b h
Sa (n+1) 0 0 0 0 h b

that all sensory inputs from one modality would be equal but that those from different
modalities need not be.
Returning to the Figure 5.4, there is a large set of interpool connections corresponding
to the brachial ganglia neuropil, Wgg below Wsg , is also set against a gray background.
This ganglion-to-ganglion connectivity matrix shows the connections between the gan-
glia. Finally, the sensory motor loop is completed with connections in a third neuropil
to the motor neurons of the sucker and arm.5
The key feature of this analysis is the recurrent or looped feedback connections
between the ganglia. The three sucker ganglia are placed slightly higher in the figure
and in darker gray than the brachial ganglia to aid in tracing their connectivity. Note
that on the input lines of the sucker ganglia there are just three connections, one on
each sucker ganglion. This depiction reflects the neuroanatomic fact that each sucker
only receives input from its corresponding brachial ganglion. The pattern repeats in the
inputs to the brachial ganglia units. Again corresponding to the anatomy, each brachial
ganglion receives input from only one sucker ganglion. The upper right corner of this
matrix is the most interesting. It shows the connections between the brachial ganglia.
The brachial ganglion in the middle, N, receives inputs from itself and the other two
units. The proximal brachial ganglion receives inputs from itself and the central brachial
ganglion and the same is true of the distal brachial ganglion (i.e. from itself and the
central one). This imposes the orderly linear array of brachial ganglia connectivity. If
Gb,n–1 and Gb,n+1 had been connected, the topology of the network would have been a
ring and not a linear array as occurs in the octopus arm.
Table 5.3 presents these interganglionic connections as the matrix Wgg .
In Table 5.3, note that the matrix row and column labels are the same. This means that
a non-zero connection along the left diagonal (c, c, c, f, f, f) in this matrix represents a
connection between a unit and itself. So in the upper-left corner of the matrix we see
the connection between sucker ganglia n–1 (Gs,n–1 ) and itself has a value of “c.” This
feedback means that an input to this ganglion, say from the sucker rim, will persist
after the stimulus has disappeared from the receptor and the sensor is inactive. This is

5 Speaking strictly anatomically this “third neuropil” might appropriately have its connections included in
the Wgg neuropil because some of the connections are located in the brachial ganglia. However, others are
found in the eccentric neuropils and this presentation format helps emphasize the functional properties of
the brachial ganglia that are central to this paper.
 The octopus with two brains: the octopus central nervous system 109

Table 5.3 The interganglionic connections in Figure 5.4 for W gg in a matrix form

GS (n–1) GS (n) GS (n+1) GB (n–1) GB (n) GB (n+1)

GS (n–1) c 0 0 k 0 0
GS (n) 0 c 0 0 k 0
GS (n+1) 0 0 c 0 0 k
GB (n–1) l 0 0 f m 0
GB (n) 0 l 0 m f m
GB (n+1) 0 0 l 0 m f

one of the requirements of a representation in the cognitive sense. How long it persists
depends on the value of c. If the value is 1.0, then a perfect copy of the activation
will be continuously supplied to the input of Gs,n–1 . The input will be “remembered”
indefinitely. On the other hand if c has a value less than 1.0, then the memory of the
input in that ganglion will decay with time and be “forgotten.” This is the basis by which
recurrent networks produce short-term memory, and it is consistent with the existence of
the neuropil in the sucker ganglia that reverberate input signals for some time depending
on the strength of the connections within the neuropil. We see from Table 5.3 that the
self-connections between the brachial ganglia have a value of f. If f is greater than c then
the brachial ganglia will “remember” input patterns of equal intensity longer than the
sucker ganglia will and vice versa.
The connections between the ganglia, those entries valued g in Table 5.3, are another
form of memory. If ganglion Gbn is activated by the stimulation of the receptors in its
LBM it will after some delay pass on that activation to its neighboring ganglia, N–1 and
N+1. This just follows from the neuroanatomy which indicates that adjacent brachial
ganglia are interconnected. However, because all the brachial ganglia are reciprocally
connected, that delayed copy of the activation of Gbn will, following another delay, find
its way back to Gbn . It should be clear that this connectivity pattern means that this
reverberation can continue forever through this loop.
How long will this “memory” last? Again, that depends, as it did for a, b, c and f, on
the value of g. These concepts can be summarized in the matrix equations that describe
this model.6

g(t) = s(t − Δt)Wsg + g(t − Δt)Wgg (5.1)

m (t) = g(t − Δt)Wgm (5.2)

Equation 5.1 states that the state of the ganglia at any time step depends on the recent
inputs from the primary receptors [s(t − Δt)Wsg ] and the previous state of the system

6 These equations are included as formal specifications for those familiar with matrix mathematics and can
be skipped without loss of understanding the important points of this section. These equations follow
traditional matrix mathematical forms with capital letters representing matrices and lower case letters
representing vectors. Matrix multiplication operators are implied here. Naturally, these equations may be
applied to larger matrices representing more spatially realistic models of the CNS components of the octopus
arm.
110 Frank W. Grasso

[g(t − Δt)Wgg ], makes this by definition an adaptive system. Equation 5.2 closes the
sensory-motor arch expressing the motor output of the system (the behavior of the arm
and sucker muscles) as the result of the states of the ganglia. The memory built into this
system through recurrent loops ensures that the behavior will not be determined by a
given stimulus input pattern. Rather it is a product of the current input and the history
of the system’s state. The behavior of the system is adaptive.
However, because there is now a network of interconnected ganglia we can ask a
second question: how far will a given stimulus activation travel through the network?
Remember that in this toy example we just used three LBMs. If the matrix were sized to
represent the 300 of a single arm or the 2400 of all eight arms the input to a single sucker
might spread quite far. The answer, again, is determined by the value of g. A larger g
will spread 50% of the activation level farther down the chain of brachial ganglia than a
smaller one.
There is more to this spatial aspect of the network. The stimulation of two different
suckers, adjacent or distant, will lead to a superposition of their activation patterns
in the entire network. For a simple example consider two inputs of equal intensity.
The stimulation of two nearby suckers will produce a mutually reinforcing activation
pattern in both of them that will be spatially localized and longer persisting. Two more
distant suckers will produce a mutually reinforcing activation between them that will
persist longer than a single activation. Two activations presented to the network at short
delays relative to the “forgetting” rate will produce asymmetrical mutually reinforcing
signals. And this principle can be extended to numbers of units that are greater than
two. This makes evident that in a network of ganglia with recurrent connections, such as
the octopus LBMs, patterns of activation can be stored and remembered hierarchically
across the network of ganglia. And these patterns have an ordered spatial arrangement
that reflects the attitude of the animal’s body and the state of the external world as sensed
by contact with surfaces. This exercise demonstrates that such a process is possible
within the octopus CNS given what we know of its connectivity.
Figure 5.5 shows a somewhat contrived graphical illustration of this concept. I made
it using a photograph of one of our laboratory’s favorite octopuses, taken as he was
attached to the flat surface of a tank glass. If the words of the text in the lower left
had been painted in shrimp extract (a chemical cocktail octopuses detect) on the glass
beneath the octopus, the pattern in the lower right panel would be transmitted by the
chemo-receptors on the sucker rims into the octopus’s CNS. Bear in mind that this is just
a single frame and that the network of ganglia in the arms would remember this input
for some time. Should the animal move to arrange its suckers with care the entire text
might be surveyed, the gaps filled in and been stored in the octopus memory for some
time afterwards.7

7 I do not mean to imply that an octopus could “read” this text. A problem with this example, and what makes it
somewhat contrived, is that the positions of the sucker needs to be known in order to decode the information
in this text. It is not clear, and the available evidence makes it seem unlikely, that the Cartesian positions of
the pixels in this image would be available from proprioceptive inputs for the octopus to reconstruct between
moves. Yet, according to the analysis presented here the information of activated suckers would certainly
pass and persist in the network of brachial ganglia.
 The octopus with two brains: the octopus central nervous system 111

Figure 5.5 Illustration of the chemo-tactile sensory surface of the arms of octopus. Panel (a)
shows an image of an Octopus rubescens “Benji” attached by his suckers to the glass surface of
his tank. Panel (b) shows the same image edited to just the sucker rims which contain the
chemical and tactile receptors. Panel (c) shows some sample text. Panel (d) shows the portions of
that text that would be transmitted through the collected sensory surface of the suckers visible in
this image. Note that much of the text is preserved and would be transmitted to Benji’s central
nervous system for processing.

A typical metaphor for recurrent networks of this type is of a pool of water. A

disturbance, such as a stone thrown into the pool, causes ripples throughout the entire
network that can be used at a later time to reconstruct the features of the original
stimulus. Multiple stones will cause intersecting patterns of ripples that can be similarly
interpreted. In this case the metaphor is of eight canals of water linked at a single pool.
And the brain of the octopus sits at the intersection of these canals, in a position to
interpret the ripple patterns arriving from the canals.

5.6.2 Tuning the ganglia network

Section 5.4.2 made clear that the brachial ganglia are multitasking entities. It posed the
question of how much cognitive load could the network support and answered that there
were a lot of neurons available to do a lot of computations. This leads to two possibilities
that, if true, have different implications for the evolution of the nervous system.
112 Frank W. Grasso

One possibility is that the information content extracted by and stored by the ganglia
network (as illustrated in section 5.6.1) is a by-product of tuning up the network to
do well enough in meeting the multiple constraints posed by the different functions
that were described in section 5.5. The network does not need to be optimized for the
information content or memory function on its own. In the analysis above (Equation
5.2) the observable behavior, (i.e. motor output) occurs on its own regardless of whether
or not the reverberating activation patterns in the brachial ganglia are used as a memory
mechanism. If the determination of the connectivity strength in the brachial ganglia
neuropil (the entries in the Wgg matrix of the analysis) were aimed the achieving suitable
performance of reaching and grasping and other behaviors the properties described in
section 5.5 would emerge to some extent as long as the connection strengths were not
zero.
If, on the other hand, there were sufficient neural resources and there was a sufficient
evolutionary advantage, then the connectivity of a ganglionic network could have been
optimized to some extent for its information content. Octopuses are known to make both
chemical and tactile discriminations using their sucker rim inputs (see chapter 7, this
volume). The analysis above is at a general and abstracted level. It does not, for example,
deal explicitly with the nature of the chemicals that are sensed, nor the mechanical qual-
ities encoded in the receptor neurons. If there are suitable neural resources amongst the
105 neurons of each LBM, then separate channels for different chemical qualities, spatial
localization on the sucker rim, and intensity of chemical and mechanical stimulation
could be encoded across the brachial ganglia network. The same principles could be
applied to single modality channels or their combinations.

5.7 Relationships between a network of local brachial modules

and the cerebral ganglia

5.7.1 The absence of somatotopic maps

Section 5.4.1 reviewed the literature which expressed the surprising failure to discover
somatotopic maps in a brain as complex and large (in numbers of cells) as that of the
octopus. Section 5.6.1 has demonstrated that the connectivity of the octopus brachial
ganglia could be an implementation of a memory system capable of encoding spatial
information about surfaces with which the octopus is in contact. It appears that the
absence of somatotopic representations in the cerebral ganglia might find explanation
in the LBM network.
Somatotopic maps and spatial maps in general are ubiquitous features of the brain
architectures of vertebrates. But it should not be surprising that there should be a different
representational strategy in octopuses with soft bodies and different life styles. Perhaps it
would be more surprising to find a somatotopic representation of the body of an octopus
which lacks rigid physical links of an endoskeleton. The retinotopic representation of
the octopus optic lobes makes sense for an invariant physical structure like the retina
and, indeed, we see this in a manner similar to that observed in arthropods.
 The octopus with two brains: the octopus central nervous system 113

Zullo and Hochner (2011) have suggested that the lack of a central somatotopic
representation might reflect the concept of embodiment in neuromechanical systems.
The analysis above advances a mechanism that is such an embodiment approach. It begs
the question, however, of how the cerebral ganglia might use this representation. We
know that the cerebral ganglia are profoundly interconnected with the brachial ganglia
so the access to the information contained in the brachial ganglia is not an issue. And,
one of the hallmarks of the supraesophageal lobes of the cerebral ganglia is their capacity
for learning. Persistence of the pattern of activation and temporal progression of that
pattern across the brachial ganglia would provide a trace that could be used to apply
trial-and-error learning on the descending inputs to the arm. Why make a copy when
you can work with the real thing?
The reciprocal information flow between the arms and cerebral ganglia could be quite
informationally impoverished (in the Shannon sense, a few bits) but rich in meaning (in
the Bayesian statistical sense). An ordered array of outputs from the individual LBMs to
the brain could provide considerable information about the state of the external world in a
relatively small number of bits. There is ample evidence that the rich local circuitry of the
arm and LBMs support motor programs of their own for local and distributed control.
The learned responses to the input patterns might be equally sparse in information,
relying on the fixed behavioral repertoires stored in the circuitry of the brachial ganglia.
The reverberant circuits, like the one described in section 5.6.1, produce relatively
short-lived traces, minutes or perhaps an hour at the longest, certainly long enough
to support the learning of behavioral sequences by the cerebral ganglia but not long
enough to make local changes in behavior on a longer time scale. There is no evidence
for LTP in the arm and no evidence yet against it. LTP at appropriate synapses would
be one way to lengthen the duration of the memory and tune the character of motor
responses and information content on a more or less permanent basis. On the other hand,
perhaps the predictability of fixed but dynamically complex performance characteristics
in the brachial network combined with a plastic brain for long-term changes is a better
computational solution than tasking the cerebral ganglia to learn what the brachial
ganglia has learned.

5.7.2 The “two brains” of the octopus

I chose the illustrations in Figure 5.1 for two reasons: the first was to illustrate the
conventions in representing recurrent neural networks. The second reason was to under-
score the idea that a network that connects sensors in the world to effectors in the world
is capable of producing behavior. The architecture of the LBM makes clear that the
individual LBM is capable of producing autonomous behavior by closing the sensory
motor loop in the control of sucker movements. The inputs from other LBMs do not
over-ride that autonomy so much as influence the ongoing behavior to support inter-
sucker coordination and coordination with the movements of the arm muscles which
form a continuum. Thus, the ganglionic network I’ve proposed here operates rather
like a distributed control network. It may not match the neuroanatomical definition of a
brain (a collection of specialized, interconnected ganglia) but the diversity of behaviors
114 Frank W. Grasso

demonstrated by “decerebrate” octopuses indicates that we are not far off from the truth
if we were to conceptualize it as one.
There is a parallel example in the nervous system of the starfish. Starfish engage in
coherent whole-organism behaviors like chemotactic foraging (Dale, 1999) and dexter-
ous manipulation in the opening of the shells of bivalve mollusks like clams and oysters
(Dyal, Owen & Willows, 1973). Yet these animals are without a central structure that
could even loosely be considered a brain. Instead the five arms (or more depending on
the species) contain a nerve cord organized, among other things, around the control of
the hundreds of tube feet that occupy the starfish’s oral surface. These nerve cords are
connected by a nerve ring that encircles the center of the animal and that is devoid of neu-
rons itself. The choice of direction of walking in a starfish is determined by a democratic
“vote” of the neural control units of all the tube feet which is decided by a winner-take-
all mechanism.8 Jellyfish form an even more extreme example of a distributed nervous
system, producing coherent behaviors of reproduction, feeding and swimming without
even ganglia (Prescott, 2007). They have a diffuse reticulum or diffuse nerve net spread
around their bodies to connect patterns of sensory input to muscle commands.
If the starfish, without a central brain, can forage for distant food items, find mates
and mate and manipulate objects using dexterous (if stereotyped) methods involving
the coordination of arms, why not an octopus with many orders of magnitude more
neural hardware in its brachial network alone? In this view the cerebral ganglia simply
influence but do not specify the details of ongoing sucker and arm behavior. It is also
possible and likely that the cerebral ganglia, as Zullo and Hochner (2011) assert, serve a
“gating” function on the ongoing autonomous behavior produced by the brachial ganglia
network.
The above demonstrates that computational architecture of a single LBM processes
a diversity of sensory modalities and motor outputs, and has a capacity for memory.
Many of these powerful LBM information-processing units interconnected with such a
large number of central interneurons will be a font of emergent functional properties
in the sense described since the early days of connectionism (McCelland & Rumelhart,
1986). One key aspect of that conception of cognitive information processing (the
“microstructure of cognition”) is that while each unit possesses (in this case large
amounts of) local information the properties of the entire network emerge from the
interactions of its parts. The network of LBMs in the octopus arm is, from its connectivity
and sheer size, a complex distributed information processing structure. Its observable
behavior results not from the sum of its parts but from the interactions of its parts (in
space and with intrinsic “memory” capacity in time), which is an exponentially larger
set of functions than the number of parts.
There is limited evidence from physiological recordings, but what there is suggests that
the information from individual sensory modalities is both preserved and propagated

8 It is reported that if this nerve ring is cut in two places to separate two legs from the other three the action of
the desynchronized (and strong) tube feet will eventually tear the two arms apart from the other three. I’ve
found this in the literature but not been able to find the original research report. Whether or not this story is
apocryphal its point about the distribution of competent autonomous behavior processes is useful here.
 The octopus with two brains: the octopus central nervous system 115

between neighboring LBMs. This means that the types of memory and distributed
information systems described in the analysis above are likely maintained as parallel
systems (e.g. different chemical qualities, mechanical properties, proprioception) up
and down a given arm and between arms. It is possible that these modalities also mix
in channels of their own (“interaction channels”) and that there are also efference copy
channels for motor signals that are propagated through the distributed network. In short,
the connectivity of the brachial plexus supports its operating as a distributed multimodal
representation of sensory and motor states across the arms with memory of past events,
which also has the ability to produce action on its own. Even more briefly it is a cognitive
behavioral control system.

5.7.3 Communication between cerebral and brachial ganglia

It should be clear now that the conception of the brachial ganglia as an autonomous or
semiautonomous behavior-generating explains several odd aspects of octopus dexterous
manipulation compared to rigid-bodied animals’. But it does beg the question: how do
the brachial ganglia communicate the state of the world in the absence of somatotopic
or other forms of maps?
Having raised the question, I will offer the following speculation as a possible answer.
Figure 5.6 shows two alternative communication schemes. Figure 5.6a shows how a
somatotopic map might be formed by the arm. With direct connectivity between the
LBMs and corresponding units in the brain, a literal copy of the orderly sequence from
proximal to distal would appear in the brain as a virtual progression from proximal to
distal through the brain tissue as, for example, a representation of the body surface does
across the sensory cortex (SI) of mammals. For simplicity I show just five LBM units
in a single arm, but there is no reason the principles not be extended to all the LBMs of
all eight arms. In that case one would expect to find an “Octo-munculus” in the brain
of an octopus as Wilder Penfield famously found the “Homunculus” in the brains of his
human patients.
Figure 5.6b assigns the cerebral ganglia (lobe deliberately not specified) the role of a
decoder of temporal rather than spatial patterns. In the upper left corner of the figure is
the chain of LBMs in the arm. Remember that these are interconnected recurrently and,
therefore, possessed of “memory” properties of their own to store past patterns. The
information in the most proximal node, N0 would contain information about its state at
the present time, the state of N1 with some delay in the past, N2 with some greater delay
and so on in the distal direction down the arm. The output from N0 contains therefore the
information of the state of the arm at each instant. (If the information carried in this line
included both primary sensory and motor state information it might be more useful and
certainly more compact than reconstituting the sensory state in the arm in the brain.) A
continuous time series of these values contains information about the progression of arm
states. The several units in the cerebral ganglia network (Na , Nb , Nc . . . ) all receive the
identical state-series input. The activation of one could signal the occurrence of a state in
the arm that was of behavioral importance. In this architecture the tuning of a particular
unit to encode a particular state is accomplished by differential connection strengths
116 Frank W. Grasso

Distal Proximal
(a)
Arm

LBM N4 N3 N2 N1 N0
units

(b)

N3 N2 N1 N0

Distal Proximal
Cerebral

Cerebral
ganglia

N’4 N’3 N’2 N’1 N’0

ganglia
units Cerebral

Cerebral
ganglia
ganglia Na Nb Nc Nd Ne
Plastic, adaptable units
connection

Virtual Virtual
distal proximal

Figure 5.6 An illustration of two network connectivity arrangements. Panel (a) shows one which
supports the formation of somatotopic maps. The progression of local brachial modules (LBM)
units (N1 , N2 . . . Nn ) from the distal to the proximal ends of a given arm maps in an orderly
progression that parallels the distal to proximal sequence of units in the cerebral ganglia (N’1 ,
N’2 . . . N’n ) because they are connected that way. Panel (b) shows a connectivity that supports a
temporal rather than a spatial encoding approach. Each of the internally recurrent LBM units in
one arm progressively pass their activity in a distal to proximal direction. The arm’s input to the
cerebral ganglia is passed from the most proximal LBM as the accumulated activity of all the
LBMs. The delay in propagation between the LBMs produces a temporal code of the arm’s state.
The cerebral ganglia units encode the patterns (Na , Nb , Nc . . . ) not locations in arm space. The
recurrent arrangement of the cerebral network here endows this network with “memory” that
permits the units to follow the time course of the input stream. Given that all cerebral ganglia
units receive the same input, the discrimination of patterns arm state is acquired through
Hebbian learning of the different time encoded sequences at plastic input connections.

on the feedback connections within the cerebral ganglia neuropil. In most animals the
acquisition of connectivity, such as that in the feedback matrix here, is as a combination
of inherited connections, those acquired during development or those learned through
experience. This network architecture supports all three modes but I’ve drawn them with
adaptable connections as a tip-of-the hat to the fact that LTP and Hebbian forms of
learning are known to exist in the cerebral ganglia. Neural network methods for learning
the connection strengths to store time series like this exist (Williams & Zipser, 1989;
Mak, Lu & Ku, 1995; Anastasio, 2010), but it should be remembered that it is the
connections in this matrix that make the discrimination possible.
With an architecture like that in Figure 5.6b, which classifies the arm states, the
representation in the cerebral ganglia is encoded as a time-series (stream) of events.
The cerebral ganglia could combine this compact summary with other information at its
disposal and decide which biasing signals to send to the arms to modulate (or gate) the
ongoing behavior.
 The octopus with two brains: the octopus central nervous system 117

Concluding an alternative encoding scheme from the absence of evidence of this

“Octo-munculus” is faulty logic – like concluding the null hypothesis when an experi-
ment fails to reject it. The discovery tomorrow of an “Octo-munculus” will invalidate
the event-encoding theory that excludes the existence of an “Octo-munculus.” Conse-
quently, I tender the conception in Figure 5.6 not as a conclusion but as an example of
a mechanism that fits the available evidence. There are many variations on this mecha-
nism and indeed alternative theories (e.g. Hopfield nets [Hopfield, 1982]) that could offer
explanations for encoding and representation of arm states in the octopus. Regardless
of the algorithm that is used, the theory (in David Marr’s sense) that the communication
between the brachial ganglia and the cerebral ganglia is a representation of sensory-
motor events, offers several advantages in explaining and new opportunities to study
cognition in octopuses and, perhaps, cephalopods generally.
Cognitive psychology and cognitive science has made heavy use of reaction time to
find quantitative evidence of the levels of cognitive load associated with specific tasks,
and to infer various capacities and the nature and number of processing steps involved
in the processing of those tasks. Underpinning this approach is the idea that longer
processing time reflects the investment of brain resources in the task. The processes
described in this chapter rest heavily on the temporal aspects of neural processing and
could be used to make specific predictions about performance of tasks by cephalopods
that if falsified or supported would reflect the underlying organization of brain processes.

5.8 The octopus with two brains

The two brains of the octopus I alluded to in the title of this chapter are the cerebral
ganglia and the ensemble of brachial ganglia (or brachial plexus), and by this point in
the chapter we have discussed them from a variety of perspectives. The cerebral ganglia
of the octopus are the part of the CNS that most researchers refer to as the “brain”
and for good reason – it has all the key features of a brain. The brachial plexus, on the
other hand, has some of the features of a brain, namely an aggregation of ganglia and a
capacity for the autonomous performance of behavior. I have endeavored to demonstrate
that the term “cognitive” could be used to characterize the brachial plexus in that the
information processing performed there meets the definition of being adaptive. And yet,
constituted as a homogenous chain of ganglia as it is, it does not meet the requirement
of specialized ganglia that distinguish a brain. So it is properly referred to as a plexus.
So, we can say that, as we use the word, the octopus does not have two brains.
Yet, there is an issue here beyond semantics, beyond the definition of the word
“brain.” If the conception of the octopus brain (cerebral ganglia) as an agent that biases
or gates in the ongoing behavioral processes mediated by the brachial ganglia is true,
then the ensemble of brachial ganglia that exist in octopuses is something remarkable: a
cognitive information-processing organ for which the term plexus does not seem quite
an appropriate name and perhaps not even a proper conception.
We could hedge our bets and, following J. Z. Young’s example, (and as I did in the
introduction of this chapter) refer to the brachial and cerebral ganglia as “the central
118 Frank W. Grasso

nervous system.” It makes good sense to consider the brachial plexus and brain as one
system for producing the dexterous manipulation behaviors of the octopus. But to think
that we understand the processing by claiming the CNS works as a whole is a tautology
that obscures the important distinction between the processing modes of the brain and
brachial plexus and the important issue of how the two parts of the CNS communicate
to produce coherent behavior. Even if the brachial plexus is not an equal partner with the
brain, underestimating its degree of autonomy obscures its important cognitive role in
the behavior of the octopus. Its size and complexity, a collection of about 2400 ganglia
(each composed of about 100 000 interneurons) interacting to produce coherent behavior
in eight non-rigid arms and about 2000 moveable suckers makes it a truly remarkable
computational instrument rivaling or exceeding the abilities of most whole organisms
we know. Perhaps it is the more intriguing and mysterious part than the part of the CNS
to which the generic name of plexus does not do appropriate descriptive justice and
perhaps one that in its novelty has more to teach us about nervous system function,
control systems and the evolution of nervous systems than its better-publicized partner.
I suspect that a vertebrate-centric bias lies behind the neglect of the remarkable
brachial plexus. The word “brain” draws on our intuitions about our own brains and
how they work. We draw our metaphors often from vertebrate brains because they have
been so much more extensively studied. The influence of funding agencies’ frequent
anthropocentric stance that research should provide discoveries that are useful for human
health perpetuates this vertebrate bias. Scientifically, the contrast between vertebrates and
cephalopods invites discovery through similarities and differences that are all remarkable
because the octopus’s body is so novel compared to our own. The counterpoint to the
novelty of cephalopod mechanisms is convergent evolution; the idea that animals of
different lineages converge from different starting places on the same basic design
principles because of the universal utility of those principles (e.g. the wings of bats
and birds have shapes that utilize aerodynamic lift). These confirmations of the general
utility of a given mechanism give us a sense of their greater explanatory power as
theories. It seems likely to me that the brachial ganglia network is an example of
convergent evolution that parallels the starfish ring ganglia. The utility of the nerve ring
for coordinating many ganglia is a more likely explanation of octopus brain evolution
than octopus converged on the solution of somatotopic maps that evolved in vertebrate
brains. The question is an interesting one and is at the heart of the comparative method
in biology and neuroscience. We should remember the novel solutions to problems that
cephalopods can teach us about are true discoveries rather than confirmation of the
general utility of the vertebrate approach, and may be of greater value to future human
generations than those discoveries that hold a mirror up to ourselves.
Computationally, the LBM network promises to be one of these opportunities for dis-
covering new mechanisms in three ways: first, in an understanding of how to control and
coordinate eight (or more, or fewer) soft appendages capable of dexterous manipulation;
second, in discovering the emergent properties of computational systems composed of
large numbers of very complex homogeneous units; and, third, by studying the nature
of the communication between the cerebral ganglia and LBMs we may learn about the
control of massive, complex systems with various degrees of distributed autonomy.
 The octopus with two brains: the octopus central nervous system 119

One novel idea that I see the brachial plexus as illustrating is its mode of control.
One useful engineering definition of a control system is one that is capable of bringing
a controlled system, such as the octopus arms and suckers, from its current state to
another arbitrary state with a known degree of precision and a finite amount of time.
A simple and common instance in octopus behavior might be releasing the 27th sucker
on arm R3, moving it 2 cm to the right and reattaching it. Treating the whole CNS as a
control system, then perhaps the brain and brachial plexus together meet the engineering
definition of a control system and together they are able to implement this for any of
the 2400 suckers. It seems more likely that the brain does not know anywhere near
this degree of detail and that the truly distributed processes of the brachial plexus serve
to give the octopus something just close enough to this degree of control using the
distributed computational method illustrated in the analysis above.
The communication between the brain and brachial plexus is another novel mechanism
that the octopus can teach us. The octopus, like the human species, is profoundly visual
in its behaviors. The most prominent externally visible structure in the octopus brain is
its paired, enormous, cytoarchitectonically complex visual lobes. The representations in
the brain of the octopus, which support processes like visual attention, are intrinsically
spatial in the information they encode. They are also, therefore, fundamentally different
from the distributed representations in the brachial plexus, which I have argued for
here. Young and others have argued similarly for the functional division of the octopus
brain into visual and chemotactile division in the organization of the brain, though
theirs was a far more centralized, descending control scheme (Young, 1964; Wells,
1978). Those distributed representations, including “memory,” are an adaptation to the
soft nature of the tissues in the arms and the consequent intedeterminant nature of the
sensing and motor control soft tissues. The transformation of the distributed encoding
and representation in the arms to something the brain can use and vice versa is a
fundamentally novel transformation between a localized representation to a distributed
one. The analysis above is one way that this may be achieved, there may be several
awaiting discovery in the parallel sensory modalities and levels of motor control in the
octopus.
The octopus has a plethora of massively parallel neural control systems: the chro-
matophore and mantle control systems are two other prominent examples. The non-rigid
form of the octopus’ body and the evolutionary demands for a complex behavioral reper-
toire have probably shaped several of these massively parallel and distributed control
systems in the coleoid cephalopods. There are undoubtedly variations on the central
theme of distributed control amongst them. Cephalopods may be specialists in dis-
tributed control systems. However, the behavioral and sensory-motor demands made by
the coordinated control of the arms and suckers in the octopus have made the brachial
plexus a uniquely cognitive organ, and one with much to teach us about architectures
for distributed control and computation.
Aristotle can be forgiven for his claims of octopus stupidity. He did not have access to
the abundant behavioral evidence for advanced cognition in cephalopods reviewed in this
volume. The octopus may or may not have two kinds of brains in one body. The answer to
that question will be resolved by understanding the content and modes of communication
120 Frank W. Grasso

between the arms and the brain. Pursuit of that mechanistic understanding will take us a
long way towards learning how cognition works in these creatures; and may either define
it specifically as a cephalopod innovation that is very different from vertebrate cognition
or provide a perspective on a monolithic form of cognition that transcends the details of
neural architecture.

Acknowledgements

I thank Dr. Michael Kuba for his assistance with the design and construction of the Octo-
pus Y-maze and Professor Jennifer Basil for helpful discussions of some of the concepts
in this manuscript. I am grateful to my home institution, Brooklyn College, CUNY for
the wise level of sabbatical year support that facilitated the writing of this chapter.

References

Altmann, J. S. (1971). Control of accept and reject reflexes in the octopus. Nature, 229: 203–207.
Anastasio, T. J. (2010). Tutorial on neural systems modeling, Sunderland, MA, Sinauer Associates
Inc.
Anderson, R. C. and Mather, J. A. (2007). The packaging problem: bivalve prey selection and
prey entry techniques of the octopus Enteroctopus dofleini. Journal of Comparative Psychology,
121: 300–305.
Aristotle, The history of animals, Book IX, translated by D’Arcy Wentworth Thompson (2013).
ebooks@Adelaide, The University of Adelaide. Available at http://ebooks.adelaide.edu.au/a/
aristotle/history/
Braitenberg, V. (1984). Vehicles: experiments in synthetic psychology, Cambridge, MA, Bradford
Books, MIT Press.
Dale, J. (1999). Coordination of chemosensory orientation in the starfish Asterias forbesi. Marine
and Freshwater Behaviour and Physiology, 32: 57–71.
Dyal, J. A., Owen, A. and Willows, D. (1973). Invertebrate learning: cephalopods and echinoderms.
In Corning, W. C., Dyal, J. A. and Willows, A. O. D. (eds.) Invertebrate learning: cephalopods
and echinoderms, New York, NY, Plenum Press.
Edelman, D. B. and Seth, A. K. (2009). Animal consciousness: a synthetic approach. Trends in
Neuroscience, 32: 476–484.
Finlay, B. L., Darlington, R. B. and Nicastro, N. (2001). Developmental structure in brain evolution.
Behavioral and Brain Sciences, 24: 263–308.
Fiorito, G., Agnisola, C., D’Addio, M., Valanzano, A. and Calamandrei, G. (1998). Scopolamine
impairs memory recall in Octopus vulgaris. Neuroscience, 253: 87–90.
Fiorito, G. and Gherardi, F. (1999). Prey-handling behaviour of Octopus vulgaris (Mollusca,
Cephalopoda) on bivalve preys. Behavioural Processes, 46: 75–88.
Fiorito, G. and Scotto, P. (1992). Observational learning in Octopus vulgaris. Science, 256: 545–
547.
Fiorito, G., Von Planta, C. and Scotto, P. (1990). Problem solving ability of Octopus vulgaris
Lamarck (Mollusca, Cephalopoda). Behavioral and Neural Biology, 53: 217–230.
Grasso, F. W. (2008). Octopus sucker–arm coordination in grasping and manipulation. American
Malacological Bulletin, 24: 13–23.
 The octopus with two brains: the octopus central nervous system 121

Grasso, F. W. and Basil, J. (2009). The evolution of flexible behavioral repertoires in cephalopod
mollusks. Brain, Behavior and Evolution, 74: 231–245.
Graziadei, P. (1965). Muscle receptors in cephalopods. Proceedings of the Royal Society. B:
Biological Sciences, 161: 392–402.
Graziadei, P. P. C. (1971). The nervous system of the arms. In Young, J. Z. (ed.) The anatomy of
the nervous system of Octopus vulgaris, Oxford, UK, Clarendon Press.
Graziadei, P. P. C. and Gagne, H. T. (1976). Sensory innervation of the rim of the octopus sucker.
Journal of Morphology, 150: 639–680.
Gutfreund, Y., Flash, T., Fiorito, G. and Hochner, B. (1998). Patterns of arm muscle activation
involved in octopus reaching movements. Journal of Neuroscience, 18: 5976–5987.
Gutnick, T., Byrne, R. A., Hochner, B. and Kuba, M. (2011). Octopus vulgaris uses visual
information to determine the location of its arm. Current Biology, 21: 460–462.
Hanlon, R. T. and Messenger, J. B. (1996). Cephalopod behaviour, Cambridge, UK, Cambridge
University Press.
Herculano-Houzel, S., Collins, C. E., Wong, P. and Kaas, J. H. (2007). Cellular scaling rules for
primate brains. Proceedings of the National Academy of Sciences, 104: 3562–3567.
Hochner, B., Shomrat, T. and Fiorito, G. (2006). The octopus: a model for a comparative analysis
of the evolution of learning and memory mechanisms. The Biological Bulletin, 210: 308–317.
Hopfield, J. J. (1982). Neural networks and physical systems with emergent collective computa-
tional abilities. Proceedings of the National Academy of Sciences, 79: 2554–2558.
Kier, W. M. (1982). The functional morphology of the musculature of squid (loliginidae) arms
and tentacles. Morphology, 172: 179–192.
Mackintosh, N. J. (1965). Discrimination learning in the octopus. Animal Behaviour suppl, 1:
129–134.
Mackintosh, N. J. and Mackintosh, J. (1963). Reversal learning in Octopus vulgaris Lamarck with
and without irrelevant cues. Quarterly Journal of Experimental Psychology, 15: 236–242.
Mak, M. W., Lu, Y. L. and Ku, K. W. (1995). Improved real time recurrent learning algorithms: a
review and some new approaches. Neurocomputing, 24: 13–36.
Marr, D. (1982). Vision, New York, NY, W. H. Freeman and Co.
Mather, J. A. (1998). How do octopuses use their arms? Journal of Comparative Psychology, 112:
306–316.
McCelland, J. L. and Rumelhart, D. E. (1986). Parallel distributed processing: explorations in the
microstructure of cognition. Psychological and biological models, Cambridge, MA, MIT Press.
Packard, A. (1961). Sucker display of octopus. Nature, 190: 736–737.
Prescott, T. J. (2007). Forced moves or good tricks in design space? Landmarks in the evolution
of neural mechanisms for action selection. Journal of Adaptive Behavior, 15: 9–31.
Rowell, C. F. H. (1963). Excitatory and inhibitory pathways in the arm of Octopus. Journal of
Experimental Biology, 40: 257–270.
Rowell, C. F. H. (1966). Activity of interneurons in the arm of Octopus in response to tactile
stimulation. Journal of Experimental Biology, 44: 589–605.
Shomrat, T., Graindorge, N., Bellanger, C., Fiorito, G., Loewenstein, Y. and Hochner, B. (2011).
Alternative sites of synaptic plasticity in two homologous “Fan-out Fan-in” learning and memory
networks. Current Biology, 21: 1773–1782.
Shomrat, T., Zarrella, I., Fiorito, G. and Hochner, B. (2008). The octopus vertical lobe modulates
short-term learning rate and uses LTP to acquire long-term memory. Current Biology, 18:
337–342.
Sumbre, G., Fiorito, G., Flash, T. and Hochner, B. (2006). Octopuses use a human-like strategy
to control precise point-to-point arm movements. Current Biology, 16: 767–772.
122 Frank W. Grasso

Sumbre, G., Gutfreund, Y., Fiorito, G., Flash, T. and Hochner, B. (2001). Control of octopus arm
extension by a peripheral motor program. Science, 293: 1845–1848.
Walløe, S., Nissen, U. V., Berg, R. W., Hounsgaard, J. and Pakkenberg, B. (2011). Stereological
estimate of the total number of neurons in spinal segment D9 of the Red Eared turtle. Journal
of Neuroscience, 31: 2431–2435.
Wells, M. J. (1964). Tactile discrimination of surface curvature and shape by the octopus. Journal
of Experimental Biology, 41: 433–445.
Wells, M. J. (1978). Octopus: physiology and behaviour of an advanced invertebrate, Chichester,
Sussex, UK, John Wiley and Sons.
Wells, M. J. and Wells, J. (1957a). The function of the brain of Octopus in tactile discrimination.
Journal of Experimental Biology, 34: 131–142.
Wells, M. J. and Wells, J. (1957b). Repeated presentation experiments and the function of the
vertical lobe. Journal of Experimental Biology, 34: 469–477.
Williams, R. J. and Zipser, D. (1989). A learning algorithm for continually running fully recurrent
neural networks. Journal of Neural Computation, 1: 270–280.
Young, J. Z. (1964). A model of the brain, Oxford, UK, Clarendon Press.
Young, J. Z. (1971). The anatomy of the nervous system of Octopus vulgaris, Oxford, UK, Claren-
don Press.
Young, J. Z. (1991). Computation in the learning system of cephalopods. The Biological Bulletin,
180: 200–208.
Zullo, L. and Hochner, B. (2011). A new perspective on the organization of an invertebrate brain.
Communicative and Integrative Biology, 4: 26–29.
Zullo, L., Sumbre, G., Agnisola, C., Flash, T. and Hochner, B. (2009). Nonsomatotopic organiza-
tion of the higher motor centers in octopus. Current Biology, 19: 1632–1636.
Part II
Cognition and the environment
6 Foraging and cognitive competence
in octopuses
Jennifer A. Mather, Tatiana S. Leite, Roland C. Anderson and James B. Wood

6.1 Introduction

Octopuses and other cephalopods are widely acknowledged to have cognitive ability,
where cognition is defined as “All the processes by which the sensory input is trans-
formed, reduced, elaborated, stored, recovered and used” (Neisser, 1976, see Mather,
1995). Over decades, researchers at the Stazione Zoologica in Naples have evaluated
both octopuses’ visual concept formation and the brain control of their learning (Wells,
1978; Borrelli & Fiorito, 2008). Octopuses play, have personalities and solve problems,
capacities which have led to a suggestion that they have a simple form of conscious-
ness (Mather, 2008). Recent research has begun to explore the neurobiology of these
processes (Hochner, Shomrat & Fiorito, 2006; Hochner, 2010) and the neural networks
underlying them (Williamson & Chrachri, 2004). Yet all these investigations of octopus
ability are laboratory-based, and there is a lack of information about how octopuses
might process such information in the field, in their daily lives. What is needed to fill
this gap is an assessment of the daily activities such as foraging, but in terms of their
underlying cognitive processes. The link may not be easy to make as field studies are
generally uncontrolled, but it is worth exploring as a foundation for laboratory studies.
High intelligence and complex cognitive processing are theorized to have developed
in mammals that live in close social groups for the understanding of cues about con-
specifics (Humphrey, 1976; Jolly, 1966). One important aspect of the lives of octopuses
is that, in contrast to these mammals, they are generally solitary (Boal, 2006, though
see Huffard, Caldwell & Boneka, 2010). Why would a non-social animal need acute
sensory processing and quick decision making? The most obvious area in which such
solitary animals would use their abilities is in foraging strategies and food selection,
which occurs while avoiding predation (Lima & Dill, 1990; Brown & Kotler, 2007)
especially given the complex near-shore environment in which many octopus species
live.
Classic assessment of animal foraging has been based on the simplistic evaluation
of the relative amount of energy expended in finding food versus that produced by
its consumption (Stephens & Krebs, 1986), which does not take other variables into
account. Such simple analysis is not always adequate. For example, in Octopus rubescens

Cephalopod Cognition, eds A.-S. Darmaillacq, L. Dickel and J. Mather. Published by Cambridge University
Press. 
C Cambridge 2014.
126 Jennifer A. Mather, Tatiana S. Leite, Roland C. Anderson and James B. Wood

(Onthank & Cowles, 2011), energetics does not drive prey choices between crab and
clam prey; the animals preferred the crabs but got more nutrition from and had less
handling time of the clam. The authors theorized that a difference in digestibility of prey
lipids might have driven the choice, although the octopod preference for crabs is a general
one (Hanlon & Messenger, 1996). In addition to taking little account of predator risk,
the simple energetic model has left little room for other lifestyle influences. However, a
recent updating of the area of foraging (Stephens, Brown & Ydenberg, 2007) has begun
to point out the role of many aspects of life, including the threat of predation (Bednekoff,
2007) and cognition (Adams-Hunt & Jacobs, 2007), in foraging. The present chapter
discusses the potential usefulness of the cognitive skills, some of which have been tested
in the laboratory, for the job of finding prey while evading capture.
This cognitive linkage has not been made in octopuses because they have been assumed
to be generalist foragers (see O’Brien, Browman & Evans, 1990; Cooper, 2005) and use
cruise searching (although pure cruise searching is probably rare). This means that they
would range widely across the substrate, unselectively capturing whatever prey they
encountered. However, a more detailed assessment of octopus foraging reveals: (1) that
individuals are not generalists (see Anderson, Wood & Mather, 2008); (2) they are
not cruise searchers but employ the saltatory search described by O’Brien, Browman
and Evans (1990) (see Mather, 1991a; Forsythe & Hanlon, 1997; Leite, Haimovici
& Mather, 2009), although the secondary search is tactile. Their foraging behavior
differs considerably across time and space; and (3) they forage under considerable
risk of predation (Bednekoff, 2007, see Mather & O’Dor, 1991; Hanlon, Forsythe &
Joneschild, 1999). All these conditions of foraging must be met by the exertion of
considerable cognitive ability, evidenced particularly in prey selection, spatial memory
and antipredator tactics.

6.2 Octopuses as generalist foragers

The first caveat about octopuses as unselective generalist foragers is that they are usually
not. Researchers in this area are fortunate, as octopuses consume the soft parts of prey
and discard the skeletal and shell remains near where they have been eating (Mather,
1991a). Although assessment of remains over-represents the proportion of molluscs
and under-represents that of crustaceans in the diet (Smith, 2003), it is probably more
accurate when sampling is done daily (see Mather, 1991a, for retention of different
types of prey remains). But investigators have sometimes been misled by the array
of remains of different prey species left in their middens. There were many species:
Anderson, Sinn and Mather (2008) found remains of 75 species from Octopus vulgaris,
Leite, Haimovici and Mather (2009) documented 55 for Octopus insularis, Ambrose
(1984) found 59 for Octopus bimaculatus and Vincent, Scheel and Hough (1998) found
32 for Enteroctopus dofleini. Mather (1991a), sampling in one small bay in the isolated
islands of Bermuda over a few weeks, found 28. But this array of species remains,
taken in some cases from dens of many individuals over a wide area or over several
years, is misleading. The aggregate list of prey species hides individual selectivity that
 Foraging and cognitive competence in octopuses 127

is obvious with a further assessment (Anderson, Sinn & Mather, 2008; Mather, Leite
& Batista, 2012; Scheel & Anderson, 2012). Such selectivity may be the result of the
considerable differences in octopus personality (Mather & Anderson, 1993), leading
them to different microhabitats and, thus, influencing their prey choices. Learning can
change small individual differences into much larger prey choice ones in silver perch
(Warburton, 2003). Kestrels are also generalists, yet the array of prey remains near their
nests show that even near-neighbors are selecting different prey species on which to
specialize (Costantini, Casagrande, Di Lieto, Fanfani & Dell’Omo, 2005), though the
cause is unknown. A shy octopus, avoidant in the face of predation risk (Brown & Kotler,
2007) might forage in short trips to areas well known to contain easy prey such as Lima
(Ctenoides) bivalves hidden under rocks in Bermuda (Mather 1991a). An active one
might explore new habitat areas and find unusual prey, such as clams that have to be dug
out of sand at Vancouver Island (Hartwick, Thorarinsson & Tulloch, 1978). Octopuses
also opportunistically capture unusual prey such as birds and hatchling turtles on oceanic
islands (Leite, Haimovici & Mather, 2009).
The majority of prey remains found in octopus middens comes from just a few,
often crustacean, species. Octopus teheulechus selected hermit crabs in the labora-
tory (Iribarne, Fernandez & Zucchini, 1991) and Octopus cyanea consumed a small
array of four crustacean species in Hawaii in the field (Van Heukelem, 1966; Mather,
2011). Even the octopus species mentioned in the last paragraph were selective of
prey species. Two species of crab constituted 51% of the prey remains of O. insularis
(Leite, Haimovici & Mather, 2009), the snail Tegula aureocincta comprised 37% of the
diet of O. bimaculatus (Ambrose, 1984), and four crustacean species made up 71%
of the diet of E. dofleini (Vincent, Scheel & Hough, 1998) in Alaska. But variation
amongst individuals is high (Mather, Leite & Batista, 2012; Scheel & Anderson, 2012).
To know whether this differential consumption represents decision making and pref-
erence or simply availability, researchers can use two different approaches. They can
contrast preferences in the laboratory with consumption in the field, or consumption in
the field against availability of potential prey species (a difficult and time consuming
evaluation).
Two studies compared laboratory preferences and prey remains in the field. Ambrose’s
(1984) thorough evaluation of Octopus bimaculoides showed an intersection of prefer-
ence and availability. In the laboratory, the octopuses much preferred crustaceans as
prey, and had a low preference particularly for the snail Tegula eiseni, though the species
T. aureocincta was moderately preferred. In the field crustaceans were rare, gastropods
common, with T. eiseni the most abundant. The species that constituted 37% of the
diet was not this most common one but was T. aureotincta, which combined fairly wide
availability and reasonable preference. As Ambrose (1984) commented, cruise-searching
predators should be opportunistic feeders that attack all encountered prey, and this is
clearly not the case for O. bimaculoides. With much less data, Mather (1991a) reported
preference for Pachygrapsus crabs and Lima bivalves in the laboratory over Fissurella
limpets and chitons. In the field, selection was clearly for Lima and the crustaceans Pachy-
grapsus and Mithrax. Chitons, while very common, were seldom consumed. Informal
observation showed that chitons were abundant in the intertidal area where octopuses
128 Jennifer A. Mather, Tatiana S. Leite, Roland C. Anderson and James B. Wood

foraged, despite this lack of consumption. While chitons have a strong hold on the
rocky substrate, octopuses are known for their formidable arm strength (Dilly, Nixon
& Packard, 1964) and also can drill a hole in one of the chiton valves (J. A. Mather,
personal observation) so they can inject a poison that weakens the muscle. They rou-
tinely take Astrea sp. gastropods in Trindade Island, Brazil, which have a strong shell
and also an operculum (T. S. Leite, personal observation). In this case, the laboratory
preference reflected the selection in the field, but was likely not the result of encounter
rates.
A thorough evaluation of consumption versus prey availability in the field has been
carried out for E. dofleini (Vincent, Scheel & Hough, 1998; Scheel, Lauster & Vincent,
2007). Again, at the population level octopuses were generalist, with a wide range of
prey chosen. However, preferences did not always match availability. Most shell remains
(75%) were of decapod crustaceans, and the percentages and abundance of remains in
den litter were not significantly different for the top five prey species consumed. The
crustacean Hapalogaster mertensii, while fairly common, was significantly avoided by
the octopuses. More striking, chitons represented a majority of the live suitable prey in
the area but were only found as 16 of the 295 prey remains. In further investigation,
Scheel, Lauster and Vincent (2007) found that energy content (a factor assumed to
dictate prey choice, see Stephens & Krebs, 1986) of the common prey species was not
different. As well, variations in density of these crustacean prey across several sampling
years did not affect the proportion of their remains in octopus middens. The octopus was
a generalist on the surface, but fairly selective (Scheel & Anderson, 2012), perhaps from
learning which prey are available or hunting selectivity in specific microhabitats to find
them, when preferences and availability were assessed in depth. Cognitive processing
of sensory information about places or identity of encountered prey may underlie this
selection.
Two pieces of information may reconcile the wide range of prey sampled with the
underlying selectivity of the octopuses themselves. First, Leite, Haimovici and Mather
(2009) reported a chance visual encounter between a foraging octopus and a gobiid fish
leading to its capture, and J. A. Mather (personal observation) saw a similar capture
and several visually guided attempts in Bermuda. Such opportunistic visual triggering
of attack indicated that chemotactile search (Mather, 1991a) could be modified with a
secondary strategy of using visual information either to locate moving prey or to pursue
prey that had evaded a first capture. Documenting the use of the Passing Cloud skin
display, in which a pale “Blotch” is passed along the skin surface from posterior to
anterior, Mather and Mather (2004) found that O. cyanea used such a secondary strategy
after an unsuccessful attempted capture of crab prey. This display, which is seen in
similar circumstances in cuttlefish (Adamo, Ehgoetz, Sangster & Whitehorne, 2006),
presumably startled the crab into movement, at which point it would be more easily seen
by the octopus and could be captured by a spread-arms Web Over action. An octopus
would have to process its capture failure and the likely location of the crab before
programming the skin display system to produce the Web Over, while also alerting to
future movement of the prey. The control of skin patterns is not well understood, but
this was not an automatic “flush” reaction of chromatophore expansion (which is under
 Foraging and cognitive competence in octopuses 129

Exploration of Memory of Win-shift

environment area searched strategy

Gut
Predator
empty
Memory of absence
habitat types

Visual landmark
recognition
Decision and direction
to go

Memory of Visual orienting to

capture likely locations
success

Primary search

Memory of prey
preparation
Intensive
chemotactile
search
Prey Secondary search
selection

Prey contact and

evaluation

Memory of home Distance to Decision to return

location home

Figure 6.1 A flow chart of the decisions made by octopuses during foraging.

nervous control, see Messenger, 2001), but a forward-directed apparent movement, due
to differential chromatophore expansion across the skin surface. Both of these types of
evidence show that the octopus can react to changing information and vary strategies and
form actions that increase the likelihood of prey capture. In fact, general assessment of
detailed hunting behavior sequences (Mather, 1991a; Leite, Haimovici & Mather, 2009)
showed variable opportunistic and not fixed sequences.
Second, in a more general approach to prey selection, Anderson, Sinn and Mather
(2008) documented the fact that the population of O. vulgaris on the island of Bonaire
130 Jennifer A. Mather, Tatiana S. Leite, Roland C. Anderson and James B. Wood

in the Caribbean was a den-specific generalist in prey selection, but that individuals
were often specialists; a Cardona niche breadth index of 0.08 confirmed this. As the
octopus dens at which prey remains were found lay within a small area of mixed
sand/mud, rock and rubble habitat, individuals were not constrained by hunting effort
within homogeneous local microhabitats to a restricted choice of prey. It is far more likely
that individuals used experience to select particular microhabitats for their foraging or
locations where particular types of prey were found, or whose techniques for prey
handling they had mastered (see Anderson & Mather, 2007), and in areas of high
density they might be influenced by competition. For instance, one of these octopuses
selected juvenile Strombus conch snails, who were handled by drilling through the shell
to poison and weaken the adductor muscle (Wodinsky, 1973), a time-consuming process
(Anderson & Mather, 2007; Steer & Semmens, 2003). The “decision” to select this
prey, normally found in seagrass beds distant from the octopus dens, and use of this
penetration technique would have meant greater energy expenditure but would have
gained the octopus a large volume of food.
Another way to look at prey selectivity is to ask whether foraging strategies guide
selectivity of prey size, or if octopuses take all sizes of prey encountered. Larger prey,
available to larger octopuses, should result in a bigger energy intake, and thus fulfill
the selectivity in terms of energetics postulated by Stephens and Krebs (1986). Scheel,
Lauster and Vincent (2007) found that the amount of consumable flesh increased expo-
nentially with the size of individuals of common crab prey species of E. dofleini, and
size of prey remains in middens was shifted upwards, so mean prey size was larger
than the mean size found in live crustacean samples from the same area. Thus, logically
for energetics but not logically for an unselective predator, octopuses were not taking
a representative sample of the prey population but the larger and more energetically
rewarding items, though we do not know if they refused small ones or sought areas
where larger ones were available. Scheel, Lauster and Vincent (2007) suggested that
the octopuses were rate-maximizing foragers. However, this type of selectivity is not
always found for octopuses. In O. insularis the prey size did not significantly increase
with octopus size (Leite, Haimovici & Mather, 2009) and most prey individuals were
quite small, suggesting that the animals were time minimizing their duration of foraging
under the risk of predation (also see Mather & O’Dor, 1991, for O. vulgaris). This lack
of size selectivity was also found in the laboratory for Octopus dierythraeus (Steer &
Semmens, 2003). In both cases the expenditure of energy involved in prey penetration
of larger prey items (Steer & Semmens, 2003; Anderson & Mather, 2007) might also
have been taken into consideration by the octopus.

6.3 Foraging as cognitively demanding

Classical foraging theory models (Stephens & Krebs, 1986) of decision making were
so simple that they were widely criticized, even initially (e.g. see Pierce and Ollason,
1987). Foraging theorists have begun to take note of the complexity of animals’ lives,
realizing that environments are neither static nor homogeneous (Stephens, Brown &
 Foraging and cognitive competence in octopuses 131

Figure 6.2 An octopus (Octopus insularis) during chemotactile search by many arms in the second
phase of foraging. (Photograph courtesy of T. S. Leite.) See plate section for color version.

Ydenberg, 2007; Dunlap & Stephens, 2012) and require a foraging animal such as an
octopus to spend time and effort acquiring information about them. One way in which
such information can be gained proactively is by exploration, and octopuses are highly
exploratory animals (Mather & Anderson, 1999; Boal, Dunham, Williams & Hanlon,
2000; Kuba, Byrne, Meisel, Greibel & Mather, 2006). Adams-Hunt and Jacobs (2007)
comment that cognitive aspects of foraging include perception and the formation of a
search image, conditioning affecting future prey choice, spatial cognition for where to
hunt, and acquisition of prey-handling techniques. All of these are relevant to octopus
foraging.
Another simplistic model needs to be modified to understand the role of planning
and learning in octopus foraging. Theorists originally divided foraging strategies into
ambush, when the prey comes to a predator, and cruise searching, when the forager
goes to find prey (O’Brien, Browman & Evans, 1990). Either of the choices, which
actually belong on the ends of a search-pause ratio continuum, can have a low cognitive
demand and may be suitable in a homogeneous habitat and for animals with limited
cognitive ability. But O’Brien, Browman and Evans (1990) description of saltatory
search based on the pattern of foraging fish and birds (see also Anderson, Stephens &
Dunbar, 1997; Cooper, 2005), provides a stop-and-go pattern that fits more closely with
octopus foraging movements (Mather 1991a; Leite, Haimovici & Mather, 2009). This
pattern of stopping to search for prey, then using results of the search (usually visual) to
132 Jennifer A. Mather, Tatiana S. Leite, Roland C. Anderson and James B. Wood

evaluate whether to continue in the same location or move on (Hill, Burrows & Hughes,
2003) demands information assessment about habitat and decision making on when to
leave a patch (O’Brien, Browman & Evans, 1990).

6.3.1 Foraging and search images

When some areas are more suitable for search than others, foraging animals must locate
these areas, assess their resources, find and consume prey and then move on. Yet they must
also store information about where they have moved and where shelter is, how they have
depleted the patch and where similar patches might be found. Hill, Burrows and Hughes
(2003) modeled the search of plaice fish and found that a saltatory stop-and-go pattern
was most effective, especially with patchy prey distribution. This localized intensive
search activity is assisted by the formation of a search image, a pattern of sensory input
indicating acceptable prey and focusing attention on important characteristics about it
(see Adams-Hunt & Jacobs, 2007; Shettleworth, 2010). Such search images are often
visual, and the sophistication of visual search has been studied in depth, often in birds.
Search selectivity is controlled by factors such as prey size, microhabitat type, prey
crypsis and environmental complexity (O’Brien, Browman & Evans, 1990) and cannot
be automatic.
Despite many years of concentration of studies of learning on visual figures in the
laboratory (Wells, 1978), saltatory search of octopuses is only visual in the primary
stage, but search images of likely habitat must be formed nevertheless. Mather (1991a),
continuously observing foraging of O. vulgaris in Bermuda, noted that search strategy
encompassed two phases. The first phase was visual scan (often before leaving the
sheltering home) and location of patch areas in which search could be concentrated –
under rocks, in crevices, within algae clusters and in pockets of sand. Location of
likely habitat must have been learned and T. S. Leite (personal observation) notes that
young O. insularis are much less selective of microhabitats for intensive search than
adults. In the second phase of search, octopuses employed the focal search strategy in
specific locations that was first recorded by Yarnall (1969) for O. cyanea. Octopuses
extended the flexible arms into these areas, picking up chemical and tactile cues and
grasping prey with the many suckers on the arms. The details of foraging behavior
varied with habitat to suit its physical constraints (also see Mather, 1991a; Forsythe &
Hanlon, 1997; Leite, Haimovici & Mather, 2009). For instance, octopuses approaching
a rock extended the arms around and below it, stretching the web between them in
the bag-like Web Over posture, enclosing the rock and feeling with arm tips for prey
within this balloon-like structure. This is an excellent example of processing learned
information for use, Neisser’s (1976) final cognitive step. At some giving-up time, the
octopus must have chosen to search in another patch or look for a different type of
habitat (see Brown & Kotler, 2007), but the rules for and constraints on this switch are
unknown.
In the laboratory, Sutherland’s (1963) quest for visual shape discrimination “rules”
showed how octopus visual search images could be formed. Animals learned to
 Foraging and cognitive competence in octopuses 133

discriminate between two visual shapes based on their horizontal and vertical extents.
But they could also encode differences in edge/area ratio, smoothness of margins (square
vs. circle), size and other characteristics. Muntz (1970) constructed figures that octo-
puses could discriminate that did not include any of Sutherland’s (1963) cues, and he
concluded that the octopuses were “learning how to learn” the important cues for this
particular experiment; in other words, forming a search image (see Shettleworth, 2010).
If an octopus was given an orientation discrimination that was too subtle to detect at first,
further training with closer and closer approximations could allow it to do so (Wells,
1978), thus sharpening this search image. How might octopuses use such information?
An octopus in Bonaire that dug into sand to acquire Pinna bivalves likely narrowed its
search down to sandy patches in order to find this prey species. Hartwick, Breen and
Tulloch (1978) commented that E. dofleini consumed clams that must have come from
scattered sandy patches distant from their sheltering rock homes and likely remembered
their location through a series of visits.
As octopuses search for prey in two stages, it is impossible to say at which stage of
search and which type of images are formed and used. Still, the rejection by octopuses of
nutritious potential prey such as chitons (Mather 1991a; Vincent, Scheel & Hough, 1998)
suggests selectivity at both phases. Foraging O. vulgaris followed landscape contours
(Mather, 1991a) and hunted along these likely areas in which prey might be concealed,
ignoring flat open areas with few crevices or little attached algae. Chitons were easily
visible in shaded subtidal rocky microhabitats to which octopuses might simply not
have oriented, though they sometimes foraged in these areas. But this prey might also
have been rejected at touch due to surface chemical cues. The rejection of the crab H.
mertensii by E. dofleini in Alaska (Vincent, Scheel & Hough, 1998) could easily have
been at the second stage, in chemotactile search. Its common name is the hairy crab,
and a variegated hairy surface makes grasping by the multiple suckers more difficult,
a factor used when strips of artificial turf prevent large octopuses from escaping from
their tanks in captivity (Wood & Anderson, 2004).
A stronger argument for selection of prey by search image is the extent to which
individual octopuses select a much narrower range of prey than the population to which
they belong. Mather (1991a) saw that individual O. vulgaris in Bermuda concentrated
on a few species such as the fragile-shelled Lima bivalve, which hides under rocks and
in crevices; it is not clear whether this was a prey preference or a microhabitat selection.
These individual differences were even stronger in O. vulgaris in Bonaire (Anderson,
Sinn & Mather, 2008), where Cardona’s niche breadth index indicated that, despite the
varied habitat and wide total prey choice, individual octopuses were specialists on a
few prey species. This selectivity is also true for E. dofleini across its range (Scheel &
Anderson, 2012). Dukas and Kamil (2000) describe the processing stage of attention
underlying such selectivity in blue jay birds, and Dunlap and Stephens (2012) found
that birds exposed to a frequent change of environment sampled more often and learned
more quickly. See Adams-Hunt and Jacobs (2007) in general and chapter 7, this volume,
by Jozet-Alves et al., for cephalopods, but much remains to be assessed about octopuses’
ability in this area.
134 Jennifer A. Mather, Tatiana S. Leite, Roland C. Anderson and James B. Wood

6.3.2 Spatial memory and prey selection

To find and return to productive sites of prey concentration, octopuses and other searching
predators use spatial memory (Mather, 1994). Much laboratory testing of animal learning
which used food as a reward has focused on “win-stay” strategies, testing an animal’s
return to rewarded locations or choices (though see the radial maze, Roberts, 1991).
Octopuses learned these choices in the laboratory quickly (Wells, 1978). Sometimes,
however, they seemed to learn incompletely and did not concentrate on one or a few
rewarded stimuli. For instance, Papini and Bitterman (1991) found that O. cyanea’s ratio
of choices of the continually rewarded stimulus of a pair asymptoted at 7/10, as the
animals maintained a low rate of checking the unrewarded location.
This should be a reminder that the shape of learning may vary, and West-Eberhard
(2003) reminds us that forgetting is also part of the learning process, especially in a
varied environment (Kerr & Feldman, 2003), making way for new information to be
stored. What looks like a failure to learn in the laboratory situation may also be the
result of a win-switch rather than a win-stay foraging strategy in the field, assisted by
learning where one has been and resulting in avoiding returns to a depleted foraging
location (Shettleworth, 2010). Radial mazes (Roberts, 1991) reward this approach, as
they supply food reward at the end of each arm and the animal must learn and remember
which locations are depleted of prey so as not return to them. Continuous observation
of areas chosen for intense search by O. vulgaris in Bermuda (Mather 1991a, see Figure
6.1) showed that, in a similar pattern, patches recently visited were avoided over the next
few days, fulfilling all of Neisser’s (1976) cognitive steps. If a crab was taken from under
a rock, a snail picked off algae or a clam dug from a pocket of sand, another would not
likely reappear, at least for a few days. Such a win-switch strategy, learned before their
capture, would not be so adaptive when octopuses were brought into the laboratory and
continually rewarded for choice of the same location.
One aspect of cognitive processing that is poorly understood is exploration to acquire
information. Octopuses are highly exploratory animals in the laboratory (Mather &
Anderson, 1999; Boal, Dunham, Williams & Hanlon, 2000; Kuba, Byrne, Meisel,
Griebel & Mather, 2006). Such behavior (though exploration is poorly understood, see
Renner, 1990) is a technique for long-term acquisition of the information about one’s
environment, which is necessary to make choices such as where to forage intensively or
where to find emergency shelter. This type of information gathering may be particularly
important for animals in complex environments (see Godfrey-Smith, 2002), such as the
rain forest and the tropical nearshore, but is difficult to evaluate by learning studies in
the laboratory.
While we know much about octopuses’ visual object perception and memory in the
laboratory, how do we know they use memory to control spacing of foraging in the field?
Mather’s (1991b) continuous tracking of foraging paths gave some clues. The juvenile
octopuses (a few hundred grams in weight) that she observed made half of their foraging
trips over a long time and distance, nearly an hour and 9 m, so that memory of location in
space was involved (Mather, 1991b). The angle of return to the sheltering home compared
to the outgoing path was 30 degrees, the percentage of approximate trail overlap was
 Foraging and cognitive competence in octopuses 135

32% (Mather, 1991b, see Figure 6.1). These returns from far away were jet-propelled
and the octopuses lifted off the sea bottom. This is important because gastropod snails
also return to a central “home,” but they use simple chemical trail following, which can
be seen by their trail overlap and minimal angle of deviation (Chelazzi, Innocenti &
Della Santina, 1983). The possibility that octopuses form a spatial map of their home
ranges (see Shettleworth, 2010) was enhanced by Mather’s (1991b) observation of 11
instances of displacement of octopuses from their foraging path, mostly by fish attack,
from which all the animals returned directly home, usually orienting by local landmarks
that were apparent to the human observers. Such calculation of a novel route home after
displacement suggests a cognitive map (Shettleworth, 2010).
Laboratory information about this use of spatial memory is slowly being gathered.
Initially, Wells (1964) believed that octopuses could not use a detour to catch a crab that
they could see through a window, but they may have been misled by the use of glass for
viewing. They were able to move along a corridor if they kept one side continuously in
the view of the same eye, perhaps because information was not yet available to the other
side of the brain (see Wells, 1978). Mather (1991b) reported fragmentary observations
of O. rubescens learning to orient to a beacon. Boal, Dunham, Williams and Hanlon
(2000) developed a paradigm where octopuses learn the location of deep “wells” in a
tank, escaping to one of them as a refuge when the water level in the tank was lowered.
A series of papers on this capacity in Sepia officinalis cuttlefish showed that they have
two memory strategies (Alves, Chichery, Boal & Dickel, 2006), remembering their turns
or memorizing landmarks, varying with maturity and sex. Brain lesion studies revealed
that the ventral area of the vertical lobe was involved in this memory (Graindorge
et al., 2006). Such studies will offer insight into an adaptive use of a type of learning
that has already been investigated in arthropods and vertebrates (Shettleworth, 2010). A
comparison of cognitive processes and strategies used by animals from three different
phyla on the same memory task would be an exciting one, allowing us to isolate whether
these are analogous or homologous abilities. Adams-Hunt and Jacobs (2007) already
have suggested a cross-phylum bee–gerbil contrast for this ability.

6.3.3 Learned prey handling

When a prey animal is captured, there may be several steps in the consumption process
that may influence prey choice. For instance, O. vulgaris in Bermuda were timed in
handling of the fragile Lima bivalves, at 5 minutes between capture and discarding of
the shell at the end of consumption, a factor that might have influenced their choice of this
species (Mather, 1991a). Larger prey, whose selection was seen by Scheel, Lauster and
Vincent (2007), present characteristics that prevent such easy consumption. A predator–
prey arms race has evolved over the millennia, with bivalves becoming stronger and
stronger resistors as their predators become more able to use a variety of strategies to
penetrate them (Vermeij, 1995).
Octopuses can adapt their tactics, probably due to experience, when preying upon
difficult-to-access prey like bivalves. When clams resist opening, octopuses have very
strong arms to pull the valves apart, a parrot-like beak to chip at the shell, a salivary
136 Jennifer A. Mather, Tatiana S. Leite, Roland C. Anderson and James B. Wood

papilla to drill holes and a posterior salivary gland toxin that immobilizes prey at injection
(Hanlon & Messenger, 1996). Steer and Semmens (2003) found that, faced with clam
prey, an O. dierythraeus first attempted to pull the valves apart. If this procedure failed, it
used the more time-consuming technique of drilling a hole in the shell and injecting the
paralytic neurotoxin. Anderson and Mather (2007) extended this work with E. dofleini
and found octopuses had three penetration techniques: they could pull, chip on the
valve edge or drill through the shell. Locations of chip and drill holes were not random
but near adductor muscles or over the heart, as the vulnerable areas of the clam only
occupy some of the area within the shell (also see Hartwick, Breen & Tulloch, 1978;
Ambrose & Nelson, 1983; Nixon & Maconnachie, 1988; Cortez, Castro & Guerra,
1998). This drilling location is somewhat dependent on the octopus species and age,
and is probably learned after early failures (Anderson, Sinn & Mather, 2008). When
the preferred procedure of pulling apart was thwarted by wires holding Venerupis clam
valves together, E. dofleini switched from pulling to drilling and chipping (Anderson &
Mather, 2007). Here there is no fixed procedure, but action, evaluation of results and
choice of another action.
When Strombus snails spires, the preferred drilling location near the adductor muscle
insertion, were covered by a metal coating, O. vulgaris drilled as close as possible, right
at the metal’s edge (Wodinsky, 1973). Indicative of the octopod flexibility in solving
problems, they pulled more malleable coatings off the shell, but when senescence meant
that brooding females had no digestive gland secretion, they simply pulled the snails out
of the shell (Wodinsky, 1978). Specific penetration techniques were used, their success
stored, and they were then retrieved as guides in preparation of prey for consumption
(see the cognitive steps of Neisser, 1976). This process influenced choice of prey species
in the laboratory. Octopuses’ choice amongst three bivalve species was different when
they were opened and offered “on the half shell” compared to those of intact animals
(Anderson & Mather, 2007). Mytilus valves were easily broken apart and this species was
often chosen from intact prey, whereas strong Protothaca had to be drilled or chipped and
many more were eaten when they had already been opened. The octopuses apparently
had taken effort and/or processing time into consideration.
Octopuses may offer a simpler model for the decision-making involved in foraging.
Unlike many vertebrates whose foraging is influenced by conspecifics (see Waite & Field,
2007), they are solitary for much of their lifespan and not reproductively active until
the end, with a semelparous reproductive strategy. Thus social influences are postponed
(see Huffard, Caldwell & Boneka, 2010; T. S. Leite & J. A. Mather, unpublished data).
Foraging subadult octopuses in Bermuda did not defend territories and although they
noticed conspecifics, they did not fight them and passively avoided contact (Mather &
O’Dor, 1991). Their home ranges were fairly small; under 120 m square for an animal
of 200 g weight. Octopuses remained in these small areas for a short period, under
2 weeks for juvenile O. vulgaris (Mather & O’Dor, 1991) and a month for the much
larger E. dofleini (Hartwick, Ambrose & Robinson, 1984). Such an occupancy pattern
would demand speedy acquisition of information about the environment and equally
speedy forgetting (West-Eberhard, 2003), a win-switch foraging strategy, and a relatively
quick assessment of prey density and predator risk to decide when to move on. In this
 Foraging and cognitive competence in octopuses 137

Figure 6.3 An octopus (Octopus insularis) accompanied by scavenging fish as it forages on the
rocky ground. See plate section for color version.

situation, the number of variables influencing foraging and food intake might be small.
A limited test of departure time from these home ranges against food consumption
and weight gain for O. vulgaris did not produce a significant correlation (Mather &
O’Dor, 1991). Instead it suggested another variable for consideration of influences on
foraging, time minimizing under the threat of predation (see also Ambrose, 1988, for
O. bimaculatus).

6.4 Predation avoidance and learning

Modern foraging theory has begun to acknowledge that animal foraging is often car-
ried on under the threat of predation (Lima & Dill, 1990; Bednekoff, 2007; Brown &
Kotler, 2007). Octopuses have abandoned the protective molluscan shell and, probably
in competition with the bony fishes (Packard, 1972), they have evolved an array of
defenses including stunning camouflage (Messenger, 2001), deimatic behavior (Hanlon
& Messenger, 1996), inking (Wood et al., 2010) and the cognitive ability to manipulate
them. Many animals avoid predation by hiding in shelter (Cooper, 2005) and trading off
the risk of emerging with the necessity to forage. Some octopuses are time minimizers
who shelter for much of the day, and O. vulgaris in Bermuda only hunted for 12% of the
daytime (Mather, 1988) and sheltered in homes at night. Provisioned animals or those
138 Jennifer A. Mather, Tatiana S. Leite, Roland C. Anderson and James B. Wood

in areas of abundant prey often minimize their time at risk out foraging (see Shaffery,
Ball & Amlaner, 1985, for gulls).
If shelters are so important to the avoidance of predation, have octopuses used cog-
nitive ability to select them? Field studies (Hartwick, Breen & Tulloch, 1978; Mather,
1994; Katsanevakis & Verriopoulos, 2004) demonstrated that octopuses usually shelter
in niches, under rocks and in holes in rubble. As they tend to be non-social, octopuses are
often primarily constrained in their distribution by the presence of such shelter. Octopus
joubini in northern Florida was confined mainly to seagrass beds where discarded mol-
luscan shells were available for shelter, and aggregated to a high density when artificial
provisioning with gastropod shells was carried out (Mather, 1982a). The low availability
of rocky crevice shelter can limit the density of octopuses, as Hartwick, Breen and Tul-
loch (1978) found for E. dofleini on Vancouver Island. Shelter can even allow them to
inhabit areas where they would not otherwise be found. Discarded beer bottles allowed
O. rubescens near Seattle to live in sandy areas that they would otherwise have been
eliminated from (Anderson, Hughes, Mather & Steele, 1999).
Little research has been carried out to discover if octopuses have specific sensory
preferences for particular shelter dimensions. O. joubini preferred shelters of about the
same volume as themselves and also ones that had a small aperture (Mather, 1982b),
as with their lack of fixed skeletons, octopuses could squeeze through holes that would
block out crustacean or fish rivals. Ideal homes do not appear in the natural environment,
however, and octopuses appeared instead to select likely shelters and adapt them closer
to some desired dimensions (Mather, 1994). They pushed out small rocks with their
arms, blew sand with jets of water from the funnel and pulled algae from the top of
crevices. More interesting, if an aperture was “too big,” an octopus brought small rocks
or molluscan shells to block the aperture; the correlation of aperture size with number of
rocks was statistically significant, r = 0.69. This is a clear example of tool use by Beck’s
(1980) definition and suggests a proactive strategy of assessment of area and gathering
of items to change it – though the cognitive background of tool use is not known. Such
a strategy was also observed in Indo-east Pacific octopuses on the sand-mud bottom,
who cleaned out a coconut shell and carried it with them for later use as a shelter (Finn,
Tregenza & Norman, 2009), a clear proactive strategy of retaining the shell for future
use.
Probably depending on predation success, octopuses have many short hunting trips
spaced irregularly during the day (Mather, 1988; Scheel & Bisson, 2012) or at night
(Meisel, Byrne, Kuba & Mather, 2011), during which they adopt camouflage skin
patterns (Packard & Sanders, 1971; Mather 1991a; Forsythe & Hanlon, 2007; Leite
et al., 2009). Animals at risk of predation during foraging may adjust the timing and
spacing of their hunts (Bednekoff, 2007). For instance, squirrels balance off the risk of
eating out of the home with the energy expenditure of returning to it (Lima, Valone &
Caraco, 1985) and they compromise by eating small food items outside and large ones in
shelter. O. vulgaris showed a different compromise. Perhaps because of the small home
ranges, 70% of prey were transported home to consume. Prey size did not correlate with
transport home, possibly because most prey items were small (Mather, 1991a). Instead,
octopuses chose to transport items to a home if it was nearby and eat them outside,
 Foraging and cognitive competence in octopuses 139

though in suboptimal shelter, if home was far away. This tradeoff argues again for the
storage, evaluation and use of a spatial memory of the area in which the octopuses
hunted.
How important is learning about the presence of predators and adjusting behavior to
them? Both the Caribbean reef squid (Mather, 2010) and cuttlefish (Langridge, Broom
& Osorio, 2007) carefully adjust their reactions, from camouflage and visual displays
to jet-propelled escape, depending on predator species, distance, speed and fish species.
But they do not have the shelter opportunities of octopuses. Many animals that come
out of shelter to hunt can vary either the amount of this activity or its timing. Nocturnal
rodents suppressed activity on moonlit nights when they were more visible to predators
(Daly, Behrends, Wilson & Jacobs, 1992) and whiptailed lizards similarly changed
their hunting time and pattern in the presence of a predatory lizard (Eifler, Eifler &
Harris, 2008). O. vulgaris is well known for the flexibility of timing of its activity
(Wells, O’Dor, Mangold & Wells, 1983). This activity was influenced particularly in the
laboratory by the timing of food provisioning, a kind of temporal learning common in
many animals (Shettleworth, 2010). Nevertheless, the daytime cycle of activity of this
species in Bermuda was not matched to time of day, height of tides or current in the
channel in which the octopuses lived (Mather, 1988), although intertidal animals must
be affected by the tidal cycles.
The possibility that octopuses might learn to change their activity to avoid potential
predators was explored in a laboratory study by Meisel, Kuba, Byrne, and Mather
(2013). Octopuses shared a tank with one of two fish species, separated by a diagonal
clear barrier. They shifted activity timing to more nocturnal in the presence of the
diurnal Balistes trigger fish, a facultative but not major octopod predator. In the presence
of a nocturnal moray eel, a well-known predator of octopuses (see Randall, 1967),
they did not become more diurnally active. But since the moray hunted by snaking
through rocks and rubble, a longer duration of sheltering would not avoid this predator.
In fact, the octopuses avoided this potential predator by spending more time out of
shelter, suggesting a sophisticated avoidance based on the hunting style of the specific
predator.
Animals at risk of predation often have a complex set of avoidance strategies (see
Kelley & Magurran, 2011, for fishes) and this is true of octopuses (see Figure 6.2). When
they emerge to hunt, their primary antipredator strategy is to avoid being detected.
Cephalopods including octopuses have variable camouflage (see Cott, 1940) used in
avoidance of detection by vertebrates. Packard (1972) called fish “the designers of
octopus skin” and octopuses’ appearance is matched to vertebrates’ sensory perception.
For instance, they are color blind yet produce careful color matches to the substrate on
which they move. Many cephalopod species can vary the colors and patterns on the
skin surface within 30 ms or less, change their apparent texture by formation of raised
papillae on the skin (Allen, Mäthger, Barbosa & Hanlon, 2009), and even alter their
posture, including positions of the flexible octopus arms (Packard & Sanders, 1971;
Hanlon & Messenger, 1988; Mather, 2004). When out of shelter, they can even change
the pattern of movement across the substrate (Huffard, 2006), being bunched up and
swaying on two arms and looking like a ball of algae, rather than walking normally
140 Jennifer A. Mather, Tatiana S. Leite, Roland C. Anderson and James B. Wood

Change timing Minimizing time Assessing prey

with predation risk exposed
extraction time

Decision: Take
prey home
Avoidance: or eat out
Staying in shelter

Skin
matched
to habitat
Distance to
Disruptive home
Primary Defense:
appearance
Camouflage (color,
texture, posture)
Mimics
other
Predator
species
notices

Deimatic
display makes
animal appear larger/ Secondary Defense:
threatening Appearance change

Eyebar Predator
conceals eye approaches

Tertiary Defense:
Departure

Inking conceals,
gives a distractive Sequence of
“fake,” disrupts appearances disrupts
chemoreception Search Images
Jetting away
to shelter

Figure 6.4 A flowchart of the decisions made by an octopus avoiding predation during foraging.

with the eight arms. Without knowing the transformation of sensory input to motor
output to the chromatophore muscles, cognitive control of the process is difficult to
prove.
There has been some debate as to whether octopuses use a kind of search image to adapt
their skin patterns to match the substrate over which they move when foraging, and their
repertoire of appearances is so large that this is difficult to determine. Hanlon, Forsythe
and Joneschild (1999) did not find that the large O. cyanea out foraging used background
 Foraging and cognitive competence in octopuses 141

matching camouflage. However, Leite and Mather (2008) determined a pattern repertoire
for the smaller O. insularis and evaluated whether this species could vary the set of
patterns, dependent on the substrate over which they hunted (Leite, Haimovici & Mather,
2009). Body pattern varied with depth (predicting amount of ambient light), presence
of sand or rubble and foraging behavior. Interestingly, larger octopuses (>500 g) used
less of the camouflaging Mottle pattern and more of the Disruptive Blotch, almost as
if large size (O. cyanea is one of the largest octopuses, Hanlon, Forsythe & Joneschild,
1999) meant that camouflage was less vital to predator avoidance. While walking,
octopuses used background matching, and swimming octopuses moving from patch to
patch adopted countershading camouflage. This automatic reaction to light intensity
differences prevents easy detection from above and below and is commonly used by
their cuttlefish relatives (Ferguson, Messenger & Budelmann, 1994). Even over a short-
distance swim, O. vulgaris in Bermuda changed from bottom-matching camouflage to
mud-brown and then adopted brown and white stripes (Mather & Mather, 1994). Because
of the nervous control of chromatophore muscles, pattern changes can be accomplished
in 30 ms and could pinpoint an area as small as 1 mm2 , although regularities of “patch”
and “groove” (Packard, 1995) produce overriding patterns.
Changes in skin appearance have led observers to speculate that octopuses in the Indo-
east Pacific were mimicking the appearance of common or distasteful prey items, such as
lionfish and poisonous snakes or common fish (see Hanlon, Watson & Barbosa, 2010 for
a detailed study). One of the problems with assessment of this behavior is that pattern
mimicry, such as that of palatable insect mimics for aposematic (warningly colored)
models, is generally static and no cognition by mimics is involved (Shettleworth, 2010).
Research has instead investigated the role of learning in the behavior of the potential
predator (Adams-Hunt & Jacobs, 2007), which would be fish in this case. Mimicking
the appearance of distasteful fish would involve octopus assessment of models’ patterns,
processing of the frequency of different species’ appearance, and programming of a
specific pattern output during an encounter with a potential visual predator. This is
quite a complex set of behaviors, though compatible with Neisser’s (1976) cognitive
steps of transforming, reducing and elaborating the input, storing the information and
recovering and using it. Octopuses might instead carry out the unpredictable sequence of
changes in a repertoire of patterns noted above, which Hanlon, Forsythe and Joneschild
(1999) documented in O. cyanea when threatened by a diver’s chase. Hanlon, Watson and
Barbosa (2010) delineated a series of appearance changes in Macrotritopus defilippi. One
of these posture/pattern combinations, which is postulated to be mimicry of flounders
(who are not poisonous or aversive) was a dorsoventrally flattened posture also useful
for head-first jet propulsion during search over the sandy bottom. More research must
be done to sort out the possible causes of these appearance changes. It is possible that
octopuses are simply “running through” preset sequences of patterns and not using
learning to match the appearance of other species.
The unit of production of skin patterns is smaller on the head and around the eyes
(Leite & Mather, 2008), perhaps because these areas are more obvious when the octopus
has emerged from hiding. Changes in these areas include production of “eyebars,” which
extend the horizontal pupil along the skin anterior and posterior to the eye to conceal
it (Mather, Anderson & Leite, 2009); an eye is an important search image for a visual
142 Jennifer A. Mather, Tatiana S. Leite, Roland C. Anderson and James B. Wood

predator, indicating the presence of an animal (Coss & Goldthwaite, 1995). This pattern
was produced during learned annoyance to keepers in the laboratory (Anderson, Mather,
Monette & Zimsen, 2010) and is probably a proactive response to a low level threat of
predation, making oneself look less like an animal. Similarly, paired obvious white dots
on the dorsal surface of the arms may distract the fish observer from the outline of the
octopus behind them (Packard & Sanders, 1971). Again, the control of their production
is unknown.
The control of camouflage production is better known in the related cuttlefish. This
species has a dazzling pattern repertoire (Hanlon & Messenger, 1988) and its background
matching of different patterns is particularly responsive to evaluation of contrast and
edges (Zylinski, Osorio & Shohet, 2009; see also chapter 9, this volume, by Zylinski &
Osorio). The texture variation of cuttlefish skin by papillae is also the result of matching
to the visual appearance (and not due to tactile information) of the substrate below
(Allen, Mäthger, Barbosa & Hanlon, 2009). Such matching may be the result of a
transformation of the visual image into a “2½ dimension” visual template of the
background appearance – Marr’s (1983) first step in human cognitive evaluation of
visual pattern. Cuttlefish can, however, be foiled in their background matching by color
variation, as the cephalopods are color blind (Messenger, 2001). In addition, cuttlefish
both change displays and interrupt hunting of shrimp prey when a model bird is “flown”
overhead (Adamo, Ehgoetz, Sangster & Whitehorne, 2006).
Detection by a predator despite camouflage does not mean capture, and octopuses
and other cephalopods can avoid capture by breaking a potential predator’s search image
during pursuit, as in the Hanlon, Forsythe and Joneschild (1999) study of unpredictable
pattern changes, stop-and-go movement and concealment within a cloud of dark ink
for O. cyanea pursued by a diver. This is a general secondary antipredatory response
of cephalopods, as Anderson and Mather (1996) noted camouflage matched to different
backgrounds, jet escape, inking and digging into the sand substrate in small Euprynna
scolopes sepiolid squid. Cephalopod ink can also hold together in a “blob” and these
animals can darken, squirt out an ink blob, pale and then jet away, leaving the predator
attacking the blob that still matches its search image (Hanlon & Messenger, 1996).
The selection and use of these common secondary tactics, which also have physiological
costs, have been little studied across the cephalopods, though see Hanlon and Messenger
(1996) for discussion.
Cephalopods that are threatened can change their appearance to look more menacing.
Moynihan (1975) speculated that the dymantic (deimatic) visual skin display of squid,
cuttlefish and octopuses showed a production parallel across the cephalopods. The
purpose is likely the same but the displays are different. Both cuttlefish (Langridge,
Broom & Osorio, 2007; Langridge, 2009) in the laboratory and squid (Mather, 2010) in
the sea produce eye spots on the dorsal surface. The cognitive control of this pattern was
demonstrated by their directional production towards the threatening fish and selection
towards visual predators and at lesser threat. Octopuses spread the web between the arms,
darken the area around the eyes and the edges of the web, orienting this spread dorsal
surface towards the potential threat (Packard & Sanders, 1971). However, no evaluation
of the circumstances of the production of this display has been carried out in octopuses.
Differential use of varied antipredator responses is widely documented across the animal
 Foraging and cognitive competence in octopuses 143

kingdom (Owings, Hennessy, Leger & Gladney, 1986; Owings & Morton, 1998), and
such selection of responses must come after sensory information storage, risk evaluation
and then action selection.
Little is known about the selection, learning or brain control of appearance in these
antipredator responses. The optic lobe of the brain contains the circuits controlling
pattern production (Williamson & Chrachri, 2004) and the suboesophageal lobe has an
output-controlling chromatophoric lobe. The octopuses may use open-loop production of
these camouflage patterns and not monitor their own output. Still, cephalopods certainly
utilize the visual cues about background, which Sutherland (1963) proved long ago in the
laboratory, that they can recognize, to form the behavioral sequence. As Hanlon, Forsythe
and Joneschild (1999) pointed out, this sequence of responses to predator threat must
include pattern and/or movement production, assessment of predator response, selection
of a new response and a movement option, a clear cognitive accomplishment in the
sequence proposed by Neisser (1976).

6.5 Conclusion

The results of this closer look at foraging of octopuses is both a realization of the
complexity of their behavior and a better understanding of the pressures that likely
caused the cognitive ability that we see demonstrated in laboratory situations. Prey
selectivity, saltatory search and spatial memory, as well as predator avoidance, tap the
considerable information processing capacity of the octopus (see Mather, 2008) and
demonstrate its cognitive ability.
Many of these field studies could form the basis for extensive laboratory research.
Foraging of octopuses under predatory threat, as Adamo, Ehgoetz, Sangster and White-
horne (2006) have done for cuttlefish, would be revealing. It would be possible to study
octopuses’ Search Images for prey, perhaps using a Y-maze to evaluate chemical cues.
More research needs to be carried out on prey choice, taking into account the variables of
search, penetration ability and handling time, as well as energetic costs and food value.
As octopuses live in a changing environment, laboratory studies could evaluate their
responses to frequent changes in prey cues (see Dunlap & Stephens, 2012). Develop-
mental studies could evaluate acquisition of foraging competence. The link between per-
sonality, learning and foraging competence could be established (see Warburton, 2003).
A wonderful opportunity exists for a cross-phyla comparison for spatial memory (Alves,
Chichery, Boal & Dickel, 2006). And, of course, laboratory studies should begin to evalu-
ate the nervous system control of these behaviors (see chapter 5, this volume, by Grasso).

Acknowledgements

The first and second authors would like to thank the Brazilian science funding agency,
CNPq, for a grant that allowed J. M. to travel to Brazil so the two could work on the
manuscript together. The authors would like to dedicate this chapter to Roland Anderson,
in memory of his many contributions to cephalopod biology.
144 Jennifer A. Mather, Tatiana S. Leite, Roland C. Anderson and James B. Wood

References

Adamo, S. A., Ehgoetz, K., Sangster, C. and Whitehorne, I. (2006). Signaling to the enemy?
Body pattern expression and its response to external cues during hunting in the cuttlefish Sepia
officinalis (Cephalopoda). The Biological Bulletin, 210: 192–200.
Adams-Hunt, M. M. and Jacobs, L. F. (2007). In Foraging: behavior and ecology. Chicago, IL:
University of Chicago Press, pp. 105–138.
Allen, J. J., Mäthger, L. M., Barbosa, A. and Hanlon, R. T. (2009). Cuttlefish use visual cues to
control three-dimensional skin papillae for camouflage. Journal of Comparative Physiology,
195: 547–555.
Alves, C., Chichery, R., Boal, J. G. and Dickel, L. (2006). Orientation in the cuttlefish Sepia
officinalis, response versus place learning. Animal Cognition, 10: 29–36.
Ambrose, R. F. (1984). Food preferences, prey availability, and the diet of Octopus bimaculatus
Verrill. Journal of Experimental Marine Biology and Ecology, 77: 29–44.
Ambrose, R. F. (1988). Population dynamics of Octopus bimaculatus: influence of life history
patterns, synchronous reproduction and recruitment. Malacologia, 29: 23–39.
Ambrose, R. F. and Nelson, B. V. (1983). Predation by Octopus vulgaris in the Mediterranean.
PSZN Marine Ecology, 4: 251–261.
Anderson, J. P., Stephens, D. W. and Dunbar, S. R. (1997). Saltatory search: a theoretical analysis.
Behavioral Ecology, 8: 307–317.
Anderson, R. C., Hughes, P. D., Mather, J. A. and Steele, C. W. (1999). Determination of the diet
of Octopus rubescens Berry, 1953 (Cephalopoda: Octopodidae) through examination of its beer
bottle dens in Puget Sound. Malacologia, 41: 455–460.
Anderson, R. C. and Mather, J. A. (1996). Escape responses of Euprymna scolopes Berry, 1911
(Cephalopoda: Sepiolidae). Journal of Molluscan Studies, 62: 543–545.
Anderson, R. C. and Mather, J. A. (2007). The packaging problem: bivalve prey selection and
prey entry techniques of the octopus Enteroctopus dofleini. Journal of Comparative Psychology,
121: 300–305.
Anderson, R. C., Mather, J. A., Monette, M. Q. and Zimsen, S. R. M. (2010). Octopuses (Enter-
octopus dofleini) recognize individual humans. Journal of Applied Animal Welfare Science, 13:
261–272.
Anderson, R. C, Sinn, D. L. and Mather, J. A. (2008). Drilling localization on bivalve prey by
Octopus rubescens (Cephalopoda: Octopodidae). The Veliger, 50: 326–328.
Anderson, R. C., Wood, J. B. and Mather, J. A. (2008). Octopus vulgaris in the Caribbean is a
specializing generalist. Marine Ecology Progress Series, 371: 199–202.
Beck, B. B. (1980). Animal tool behavior: the use and manufacture of tools by animals. New York,
NY: Garland Publishing.
Bednekoff, P. A. (2007). Foraging in the face of danger. In Foraging: behavior and ecology.
Chicago, IL: University of Chicago Press, pp. 305–330.
Boal, J. G. (2006). Social recognition: a top-down view of cephalopod behavior. Vie et Milieu,
56: 69–79.
Boal, J. G., Dunham, A. W., Williams, K. T. and Hanlon, R. T. (2000). Experimental evidence
for spatial learning in octopuses (Octopus bimaculoides). Journal of Comparative Psychology,
114: 246–252.
Borrelli, L. and Fiorito, G. (2008). Behavioral analysis of learning and memory in
cephalopods. In Learning and memory: a comprehensive reference. London, UK: Elsevier,
pp. 605–627.
 Foraging and cognitive competence in octopuses 145

Brown, J. S. and Kotler, B. P. (2007). The ecology of fear. In Foraging: behavior and ecology.
Chicago, IL: University of Chicago Press, pp. 437–480.
Chelazzi, G., Innocenti, R. and Della Santina, P. (1983). Zonal migration and trail following of an
intertidal gastropod analyzed by LED tracking in the field. Marine Behaviour and Physiology,
10: 121–136.
Cooper, W. E., Jr. (2005). The foraging mode controversy and foraging space. Journal of Zoology,
267: 179–190.
Cortez, T., Castro, B. G. and Guerra, A. (1998). Drilling behaviour of Octopus mimus Gould.
Journal of Experimental Marine Biology and Ecology, 224: 193–203.
Coss, R. G. and Goldthwaite, R. O. (1995). The persistence of old designs for perception. In
Perspectives in ethology, II. New York, NY: Plenum Press, pp. 83–148.
Costantini, D., Casagrande, S., Di Lieto, G., Fanfani, A. and Dell’Omo, G. (2005). Consistent
differences in feeding habits between neighbouring breeding kestrels. Behaviour, 142: 1409–
1421.
Cott, H. B. (1940). Adaptive coloration in animals. London, UK: Methuen Publishing.
Daly, M., Behrends, P. R., Wilson, M. I. and Jacobs, L. F. (1992). Behavioral modulation of
predation risk: moonlight avoidance and crepuscular compensation in a nocturnal desert rodent,
Dipodomys merriami. Animal Behaviour, 44: 1–9.
Dilly, P. N., Nixon, M. and Packard, A. (1964). Forces exerted by Octopus vulgaris. Pubblicazioni
della Stazione Zoologica di Napoli, 34: 86–97.
Dukas, R. and Kamil, A. (2000). The cost of limited attention in blue jays. Behavioral Ecology,
11: 502–506.
Dunlap, A. S. and Stephens, D. W. (2012). Tracking a changing environment: optimal sampling,
adaptive memory and overnight effects. Behavioral Processes, 89: 86–94.
Eifler D. A, Eifler, M. A. and Harris, B. R. (2008). Foraging under the risk of preda-
tion in desert grassland whiptail lizards (Aspidoscelis uniparens). Journal of Ethology, 26:
219–223.
Ferguson, G. P., Messenger, J. B. and Budelmann, B. U. (1994). Gravity and light influence the
countershading reflexes of the cuttlefish Sepia officinalis. Journal of Experimental Biology,
191: 247–256.
Finn, J. T., Tregenza, T. and Norman, M. D. (2009). Defensive tool use in a coconut-carrying
octopus. Current Biology, 19: 1069–1070.
Forsythe, J. W. and Hanlon, R. T. (1997). Foraging and associated behavior by Octopus cyanea
Gray, 1849 on a coral atoll, French Polynesia. Journal of Experimental Marine Biology and
Ecology, 209: 15–31.
Godfrey-Smith, R. (2002). Environmental complexity and the evolution of cognition. In The
evolution of intelligence. Mahwah, NJ: Lawrence Erlbaum Associates, pp. 223–250.
Graindorge, N., Alves, C., Darmaillacq, A. S., Chichery, R., Dickel, L. and Bellanger, C. (2006).
Effects of dorsal and ventral electrolytic lesions of the vertical lobe on spatial learning and
locomotor activity in Sepia. Behavioral Neurosciences, 120: 1151–1158.
Hanlon R. T., Forsythe J. W. and Joneschild, D. E. (1999). Crypsis, conspicuousness, mimicry and
polyphenism as antipredator defences of foraging octopuses on Indo-Pacific coral reefs, with a
method of quantifying crypsis from video tapes. Biological Journal of the Linnean Society, 66:
1–22.
Hanlon, R. T. and Messenger, J. B. (1988). Adaptive coloration in young cuttlefish (Sepia offic-
inalis L.): the morphology and development of body patterns and their relation to behaviour.
Philosophical Transactions of the Royal Society of London. Series B, 320: 437–487.
146 Jennifer A. Mather, Tatiana S. Leite, Roland C. Anderson and James B. Wood

Hanlon, R. T. and Messenger, J. B. (1996). Cephalopod behaviour. New York, NY: Cambridge
University Press.
Hanlon, R. T., Watson, A. C. and Barbosa, A. (2010). A “mimic octopus” in the Atlantic: flatfish
mimicry and camouflage by Macrotritopus defilippi. The Biological Bulletin, 218: 15–24.
Hartwick, E. B., Ambrose, R. F. and Robinson, S. M. C. (1984). Den utilization and the movements
of tagged Octopus dofleini. Marine Behaviour and Physiology, 11: 95–110.
Hartwick, E. B., Breen, P. A. and Tulloch, L. (1978). A removal experiment with Octopus dofleini
(Wülker). Journal of the Fisheries Research Board of Canada, 35: 1492–1495.
Hartwick, E. B., Thorarinsson, G. and Tulloch, L. (1978). Methods of attack by Octopus dofleini
(Wülker) on captured bivalve and gastropod prey. Marine Behaviour and Physiology, 5: 193–
200.
Hill, S. L., Burrows, M. T. and Hughes, R. N. (2003). The efficiency of adaptive search tactics
for different prey distribution patterns: a simulation model based on the behavior of juvenile
plaice. Journal of Fish Biology, 63 (Supplement A): 117–130.
Hochner, B. (2010). Functional and comparative assessments of the octopus learning and memory
system. Frontiers in Bioscience, 2: 764–771.
Hochner, B., Shomrat, T. and Fiorito, G. (2006). The octopus: a model for comparative analysis
of the evolution of learning and memory. The Biological Bulletin, 210: 308–317.
Huffard, C. L. (2006). Locomotion by Abdopus aculeatus (Cephalopoda: Octopodidae): walking
the line between primary and secondary defenses. Journal of Experimental Biology, 209: 3697–
3707.
Huffard, C. L., Caldwell, R. L. and Boneka, F. (2010). Male–male and male–female aggression may
influence mating associations in wild octopuses (Abdopus aculeatus). Journal of Comparative
Psychology, 124: 38–46.
Humphrey, N. K. (1976). The social function of intellect. In Growing points in ethology. Cam-
bridge, UK: Cambridge University Press, pp. 303–317.
Iribarne, O. O., Fernandez, M. E. and Zucchini, H. (1991). Prey selection by the small Patagonian
octopus Octopus tehuelchus d’Orbigny. Journal of Experimental Marine Biology and Ecology,
148: 271–281.
Jolly, A. (1966). Lemur social behavior and primate intelligence. Science, 139: 764–766.
Katsanevakis, S. and Verriopoulos, G. (2004). Den ecology of Octopus vulgaris Cuvier, 1797, on
soft sediment: availability and types of shelters. Scientia Marina, 68: 147–157.
Kelley, J. L. and Magurran, A. E. (2011). Learned defences and counterdefences in predator-prey
interactions. In Fish cognition and behavior. Chichester, UK: Wiley-Blackwell, pp. 36–58.
Kerr, B. and Feldman, M. W. (2003). Carving the cognitive niche: optimal learning strategies in
homogeneous and heterogeneous environments. Journal of Theoretical Biology, 220: 169–188.
Kuba, M., Byrne, R. A., Meisel, D. V., Griebel, U. and Mather, J. A. (2006). When do octopuses
play? Effects of repeated testing, object type, age, and food deprivation on object play in Octopus
vulgaris. Journal of Comparative Psychology, 120: 184–190.
Langridge, K. V. (2009). Cuttlefish use startle displays, but not against large predators. Animal
Behaviour, 77: 847–856.
Langridge, K. V., Broom, M. and Osorio, D. (2007). Selective signaling by cuttlefish to predators.
Current Biology, 17: 1044–1045.
Leite, T. S., Haimovici, M. and Mather, J. A. (2009). Octopus insularis (Octopodidae): evidences
of a specialized predator and a time-minimizing forager. Marine Biology, 156: 2355–2367.
Leite, T. S. and Mather, J. A. (2008). A new approach to octopuses’ body pattern analysis:
a framework for taxonomy and behavioral studies. American Malacological Bulletin, 24:
31–41.
 Foraging and cognitive competence in octopuses 147

Lima, S. L. and Dill, L. M. (1990). Behavioural decisions made under the risk of predation: a
review and prospectus. Canadian Journal of Zoology, 68: 619–640.
Lima, S. L., Valone, T. J. and Caraco, T. (1985). Foraging efficiency-predation risk trade-off in the
grey squirrel. Animal Behaviour, 33: 155–165.
Marr, D. (1983). Vision: a computational investigation into the human representation and pro-
cessing visual information. San Francisco, CA: W. H. Freeman and Co.
Mather, J. A. (1982a). Factors affecting the spatial distribution of natural populations of Octopus
joubini Robson. Animal Behaviour, 30: 1166–1170.
Mather, J. A. (1982b). Choice and competition: their effects on occupancy of shell homes by
Octopus joubini. Marine Behaviour and Physiology, 8: 285–293.
Mather, J. A. (1988). Daytime activity of juvenile Octopus vulgaris in Bermuda. Malacologia, 29:
69–76.
Mather, J. A. (1991a). Foraging, feeding and prey remains in middens of juvenile Octopus vulgaris
(Mollusca: Cephalopoda). Journal of Zoology, London, 224: 27–39.
Mather, J. A. (1991b). Navigation by spatial memory and use of visual landmarks in octopuses.
Journal of Comparative Physiology, 168: 491–497.
Mather, J. A. (1994). “Home” choice and modification by juvenile Octopus vulgaris (Mollusca:
Cephalopoda): specialized intelligence and tool use? Journal of Zoology, London, 233: 359–
368.
Mather, J. A. (1995). Cognition in cephalopods. Advances in the Study of Behavior, 24: 316–353.
Mather, J. A. (2004). Cephalopod displays: from concealment to communication. In Evolution of
communication systems. Cambridge, MA: MIT Press, pp. 193–213.
Mather, J. A. (2008). Cephalopod consciousness: behavioral evidence. Consciousness and Cog-
nition, 17: 37–48.
Mather, J. A. (2010). Vigilance and antipredator responses of Caribbean reef squid. Marine and
Freshwater Behaviour and Physiology, 43: 357–370.
Mather, J. A. (2011). Why is Octopus cyanea Gray in Hawaii specializing in crab prey? Vie et
Milieu, 61: 181–184.
Mather, J. A. and Anderson, R. C. (1993). Personalities of octopuses (Octopus rubescens). Journal
of Comparative Psychology, 107: 336–340.
Mather, J. A. and Anderson, R. C. (1999). Exploration, play and habituation in octopuses (Octopus
dofleini). Journal of Comparative Psychology. 113: 333–338.
Mather, J. A., Anderson, R. C. and Leite, T. S. (2009). “The octopus eyebar: anatomy of a skin
display.” Presented at the meeting of the Animal Behavior Society, Pirenopolis, Brazil, June,
2009.
Mather, J. A., Leite, T. S. and Batista, A. T. (2012). Individual prey choices of octopuses: are they
generalist or specialist? Current Zoology, 58: 596–602.
Mather, J. A. and Mather, D. L. (1994). Skin colours and patterns of juvenile Octopus vulgaris
(Mollusca, Cephalopoda) in Bermuda. Vie et Milieu, 44: 267–272.
Mather, J. A. and Mather, D. L. (2004). Apparent movement in a visual display: the passing cloud
in Octopus cyanea. Journal of Zoology, London, 263: 89–94.
Mather, J. A. and O’Dor, R. K. (1991). Foraging strategies and predation risk shape the natural
history of juvenile Octopus vulgaris. Bulletin of Marine Science, 49: 256–269.
Meisel, D. V., Byrne, R. A., Kuba, M. and Mather, J. A. (2011). Behavioural sleep in Octopus
vulgaris. Vie et Mileu, 61: 185–190.
Meisel, D. V., Kuba, M., Byrne, R. A. and Mather, J. A. (2013). The effect of predatory presence
on the temporal organization of activity in Octopus vulgaris. Journal of Experimental Marine
Biology and Ecology, 447: 75–79.
148 Jennifer A. Mather, Tatiana S. Leite, Roland C. Anderson and James B. Wood

Messenger, J. B. (2001). Cephalopod chromatophores: neurobiology and natural history.

Biological Reviews, 76: 473–528.
Moynihan, M. (1975). Conservatism of displays and comparable stereotyped patterns among
cephalopods. Function and evolution in behavior: essays in honor of Professor Niko Tinbergen,
F.R.S. Oxford, UK: Oxford University Press, pp. 276–291.
Muntz, W. R. A. (1970). An experiment in shape discrimination and signal detection in Octopus.
Quarterly Journal of Experimental Psychology, 22: 82–90.
Neisser, U. (1976). Cognition and reality. San Francisco, CA: W. H. Freeman and Co.
Nixon, M. and Maconnachie, E. (1988). Drilling by Octopus vulgaris (Mollusca: Cephalopoda)
in the Mediterranean. Journal of Zoology, London, 216: 687–716.
O’Brien, W. J., Browman, H. I. and Evans, B. I. (1990). Search strategies of foraging animals.
American Scientist, 78: 152–160.
Onthank, K. L. and Cowles, D. L. (2011). Prey selection in Octopus rubescens: Possible roles of
energy budgeting and prey nutritional composition. Marine Biology, 158: 2795–2804.
Owings, D. H., Hennessy, D. F., Leger, D. W. and Gladney, A. B. (1986). Different functions of
“alarm” calling for different time scales: a preliminary report on ground squirrels. Behaviour,
99: 101–116.
Owings, D. H. and Morton, E. (1998). Animal vocal communication: a new approach. Cambridge,
UK: Cambridge University Press.
Packard, A. (1972). Cephalopods and fish: the limits of convergence. Biological Reviews, 47:
241–307.
Packard, A. (1995). Organization of cephalopod chromatophore systems: a neuromuscu-
lar image generator. In Cephalopod neurobiology. Oxford, UK: Oxford University Press.
pp. 331–367.
Packard, A. and Sanders, G. D. (1971). Body patterns of Octopus vulgaris and maturation of the
response to disturbance. Animal Behaviour, 19: 780–790.
Papini, M. R. and Bitterman, M. E. (1991). Appetitive conditioning in Octopus cyanea. Journal
of Comparative Psychology, 105: 107–113.
Pierce, G. J. and Ollason, J. G. (1987). Eight reasons why optimal foraging theory is a complete
waste of time. Oikos, 49: 111–118.
Randall, J. E. (1967). Food habits of reef fishes of the West Indies. Studies in Tropical Oceanog-
raphy, 5: 665–847.
Renner, M. J. (1990). Neglected aspects of exploratory and investigatory behavior. Psychobiology,
18: 16–22.
Roberts, W. A. (1991). Testing optimal foraging theory on the radial maze. Animal Learning and
Behavior, 19: 305–316.
Scheel, D. and Anderson, R. C. (2012). Variability in the diet specialization of Enteroctopus dofleini
in the eastern Pacific examined from midden contents. American Malacological Bulletin, 30:
1–13.
Scheel, D and Bisson, L. (2012). Movement patterns of giant Pacific octopuses, Enteroctopus
dofleini (Wülker, 1910). Journal of Experimental Marine Biology and Ecology, 416–417: 21–
31.
Scheel, D., Lauster, A. and Vincent, T. L. S. (2007). Habitat ecology of Enteroctopus dofleini from
middens and live prey surveys in Prince William Sound, Alaska. In Cephalopods present and
past: new insights and fresh perspectives. New York, NY: Springer, pp. 434–458.
Shaffery, J. P., Ball, N. J. and Amlaner, C. J. (1985). Manipulating daytime sleep in herring gulls
(Larus argentatus). Animal Behaviour, 33: 566–572.
 Foraging and cognitive competence in octopuses 149

Shettleworth, S. J. (2010). Cognition, evolution, and behavior, 2nd edn. New York, NY: Oxford
University Press.
Smith, C. D. (2003). Diet of Octopus vulgaris in False Bay in South Africa. Marine Biology, 143:
1127–1133.
Steer, M. A. and Semmens, J. M. (2003). Pulling and drilling, does size or species matter?
An experimental study of prey handling in Octopus dierythraeus (Norman, 1992). Journal of
Experimental Marine Biology and Ecology, 290: 165–178.
Stephens, D. W., Brown, J. S. and Ydenberg, R. C. (Eds.) (2007). Foraging: behavior and ecology.
Chicago, IL: University of Chicago Press.
Stephens, D. W. and Krebs, J. R. (1986). Foraging Theory. Princeton, NJ: Princeton University
Press.
Sutherland, N. S. (1963). Shape discrimination and receptive fields. Nature, 197: 118–122.
Van Heukelem, W. F. (1966). Some aspects of the ecology and ethology of Octopus cyanea Gray.
Unpublished MSc. Thesis, University of Hawaii.
Vermeij, G. J. (1995). A natural history of shells. Princeton, NJ: Princeton University Press.
Vincent, T. L. S., Scheel, D. and Hough, K. R. (1998). Some aspects of diet and foraging behavior
of Octopus dofleini (Wülker, 1910) in its northernmost range. Marine Ecology, 19: 13–29.
Waite, T. A. and Field, K. L. (2007). Foraging with others: games social foragers play. In Foraging:
behavior and ecology. Chicago, IL: University of Chicago Press, pp. 331–362.
Warburton, K. (2003). Learning of foraging skills by fish. Fish and Fisheries, 4: 203–215.
Wells, M. J. (1964). Detour experiments with octopus. Journal of Experimental Biology, 41:
621–642.
Wells, M. J. (1978). Octopus: physiology and behaviour of an advanced invertebrate. London,
UK: Chapman and Hall.
Wells, M. J., O’Dor, R. K., Mangold, K. and Wells, M. J. (1983). Diurnal changes in activity and
metabolic rate in Octopus vulgaris. Marine Behaviour and Physiology, 9: 275–287.
West-Eberhard, M. J. (2003). Developmental plasticity and evolution. Oxford, UK: Oxford Uni-
versity Press.
Williamson, R. and Chrachri, A. (2004). Cephalopod neural networks. Neurosignal, 13: 87–98.
Wodinsky, J. (1973). Mechanism of hole boring in Octopus vulgaris. Journal of General Psychol-
ogy, 88: 179–183.
Wodinsky, J. (1978). Feeding behavior of broody Octopus vulgaris. Animal Behaviour, 26: 803–
813.
Wood, J. B. and Anderson, R. C. (2004). Interspecific evaluation of octopus escape behavior.
Journal of Applied Animal Welfare Science, 7: 96–106.
Wood, J. B., Maynard, A. E., Lawlor, A. G., Sawyer, E. K., Simmons, D. M., Pennoyer, K. E. and
Derby, C. E. (2010). Caribbean reef squid, Sepioteuthis sepioidea, use ink as a defense against
predatory French grunts, Haemulon flavolineatum. Journal of Experimental Marine Biology
and Ecology, 388: 20–27.
Yarnall, J. L. (1969). Aspects of the behaviour of Octopus cyanea Gray. Animal Behaviour, 17:
747–75.
Zylinski, S., Osorio, D. and Shohet, A. J. (2009). Edge detection and texture classification by
cuttlefish. Journal of Vision, 9: 1–10.
7 Navigation in cephalopods
Christelle Jozet-Alves, Anne-Sophie Darmaillacq and Jean G. Boal

7.1 Introduction

The coleoid cephalopods exhibit three classes of natural movement patterns: dispersal,
local movements and migrations. How are these movements accomplished? The coleoid
cephalopods (octopuses, cuttlefishes and squids) have complex nervous systems and
multiple types of sensory information at their disposal, including vision (contrast, polar-
ization), chemoreception (contact, distance) and mechanoreception (touch, vibration).
Based on information gleaned from field and laboratory studies, what can we infer about
cephalopod orientation and navigation?
Multiple words are in use to describe the movement patterns of organisms in space.
Orientation is a term used for directional movements, such as phototaxis or rheotaxis,
orientation with respect to light or water currents, respectively (Dunn, 1990). Complex
cognition is not necessary for orientation behavior. The term navigation is used for
the process of determining and maintaining a course or trajectory to a goal location
(Franz & Mallot, 2000). Navigation requires at least some cognition (attention, memory,
learning): learning particular landmarks, travel routes, associations between particular
environmental features, or some combination of these. If an animal uses the configuration
of several landmarks to localize the goal; this behavior is called piloting (Gallistel, 1993).
In this chapter, we first describe the sensory systems of coleoid cephalopods (octo-
puses, cuttlefishes and squids) that enable them to collect spatial information and then
explore current field and laboratory evidence for short- and long-distance navigation.
Throughout the chapter, we will discuss possible mechanisms to support such navigation.
For a discussion of orientation and navigation in Nautilus, see chapter 2, this volume,
by Basil & Crook; for excellent overviews of spatial orientation in animals, see Wehner
et al. (1996), Healy (1998), Golledge (1999), Brown and Cook (2006) and Dolins and
Mitchell (2010).

7.2 Sensory systems

Navigation requires complex sense organs to collect information, a sophisticated central

nervous system to process the information and make decisions and efficient effectors

Cephalopod Cognition, eds A.-S. Darmaillacq, L. Dickel and J. Mather. Published by Cambridge University
Press. 
C Cambridge 2014.
 Navigation in cephalopods 151

Figure 7.1 Cuttlefish Sepia officinalis. The picture has been taken in the English Channel off
Granville (Manche, France). The ocular globe is particularly voluminous. The eye is
characterized by the w-shaped papillae. (Copyright, F. Sichel.) See plate section for color version.

to carry out the behavioral response. Cephalopods possess all of these (for reviews,
Budelmann, 1995; Hanlon & Messenger, 1996). Their prodigious capabilities for com-
plex cognition are covered elsewhere (for a review, Mather, 1995), including within this
volume; here we assess the types of sensory information available to support cephalopod
navigation.

7.2.1 Vision
Eyes are one of the most conspicuous features of cephalopods (Figure 7.1). Cephalopod
eyes are structurally very similar to those of vertebrates (for reviews, Messenger, 1991;
Budelmann, 1994; Gleadall & Shashar, 2004). The visual information from the eyes is
considerable and the brain has large visual areas. The size of the optic lobes represents
85% of the whole central nervous system in Octopus vulgaris, about 140% in Sepia sp.
and more than 330% in Loligo sp. (reviewed in Nixon & Young, 2003). It is interesting to
note that the relative size of these lobes increases as the way of life changes from the most
benthic individuals (octopuses) to the most pelagic (squids). Cephalopods’ excellent
visual discrimination abilities are evident from the extensive body of research addressing
visual discrimination learning (e.g. in Sepia: Karson, 2003; in Octopus: Sutherland
et al., 1963; Mackintosh & Mackintosh, 1964; for reviews, Mather, 1995; Boal, 1996).
152 Christelle Jozet-Alves, Anne-Sophie Darmaillacq and Jean G. Boal

Experimental evidence indicates that cephalopod vision is highly sensitive to bright-

ness contrast. For example, cephalopods show an optomotor or an optokinetic response
when placed within a rotating cylinder painted with black and white vertical stripes
(in Sepia: Messenger, 1970; Groeger et al., 2005). To date, there is no evidence that
cephalopods see colors (Marshall & Messenger, 1996; Mäthger et al., 2006). Color
vision requires at least two visual pigments (Tansley, 1965) and, except in Watasenia
scintillans, a deep-sea squid, cephalopod eyes contain only a single visual pigment
(Hanlon & Messenger, 1996; Bellingham et al., 1998). Cephalopods do perceive light
polarization (Hanlon & Messenger, 1996; Horváth & Varjú, 2004), however, which
can be considered as analogous to color vision (Cronin et al., 2003). This polarization
sensitivity (PS) arises from the orthogonal positioning of the microvilli of neighboring
photoreceptors in the retina. Linear PS can be used to break camouflage by enhancing
contrast; for example, when attacking transparent or silvery prey (in Sepia officinalis:
Shashar et al., 2000). It is also possibly involved in intraspecific communication (Shashar
et al., 1996; Boal et al., 2004). Most notably, polarization is used in orientation behavior
by many animals (bees: Kraft et al., 2011; fish: Waterman, 2006; see also Shettleworth,
2010 and Marshall & Cronin, 2011); recent laboratory experiments have demonstrated
that cephalopods also can use the e-vector when orientating (S. officinalis; Cartron
et al., 2012, see below).

7.2.2 Chemoreception
Detection of chemical cues can be either through contact chemoreception (gustation)
or distance chemoreception (olfaction). Contact chemoreception is achieved through
chemoreceptors located on the suckers and lips (Budelmann et al., 1997). There are about
10 000 receptors on each sucker in Octopus and only about 100 in Sepia, illustrating
the relative importance of this sense to octopuses (Graziadei, 1964; Graziadei & Gagne,
1976). Chemotactile discrimination learning has been demonstrated in octopuses (Wells,
1978). Distance chemoreception is achieved through olfactory organs located close to
the eyes (Budelmann, 1996). Octopuses and cuttlefish are capable of sensing a variety
of different chemicals, even at very low concentrations and can orient with respect to
these odors (in Eledone: Boyle, 1986; in Octopus: Lee, 1992; Walderon et al., 2011; in
Sepia: Boal & Golden, 1999; Boal et al., 2010; in Loligo: Cummins et al., 2011; for a
review, Hanlon & Messenger, 1996).
As with the optic lobes, the relative size of the brain structures that process chemosen-
sory information depends on the animals’ way of life. Octopuses and cuttlefish, which
are benthic or necto-benthic, are more likely to use their arms for crawling or manipu-
lating prey than squids, which swim in the open water. This difference is reflected in the
organization of the brain where the sub-esophageal mass is much less condensed and
the brachio-pedal connectives much longer in squids than in cuttlefish and octopuses.

7.2.3 Mechanoreception
Cephalopods have a complex mechanoreceptive system that comprises the statocysts,
the lateral line analogue and touch and pressure receptors (Budelmann, 1996).
 Navigation in cephalopods 153

The statocysts are the most complex of all invertebrate equilibrium receptor systems
(Young, 1960; Dilly et al., 1975). Their level of sophistication rivals that of the vestibular
system of vertebrates. There are receptors for the detection of gravity, angular acceler-
ation and linear acceleration; by integrating this information, cephalopods can assess
their location and their active and passive movements in the water (Budelmann, 1980).
This sophisticated equilibration system may allow cephalopods to move efficiently in a
three-dimensional environment.
Like other aquatic species, cephalopods are highly sensitive to local water move-
ments (“sound”). The lateral line analogue (in Sepia and Lolliguncula: Budelmann &
Bleckmann, 1988; Bleckmann et al., 1991) provides cephalopods with the equivalent
of “hearing” (Hanlon & Budelmann, 1987). Physiological studies have shown that the
hair cells respond to local water movements at 0.5–400 Hz (Bleckmann et al., 1991);
behavioral responses have been documented to frequencies from 20 to 600 Hz (in Sepia:
Komak et al., 2005). Recordings from the statocysts suggest that these receptors also
play a role in detecting particle movements (Loligo: Mooney et al., 2010). It is possible
that this sensory system detects sounds from predators and prey, or the low-frequency
environmental sound signatures that may aid navigation.
Cephalopods possess also other mechanoreceptors. Octopuses are particularly sensi-
tive to touch and can learn to discriminate different degrees of roughness (Wells, 1978).
Benthic cephalopods could use texture to orient (Hvorecny et al., 2007). Cuttlefish pos-
sess mechanoreceptors in the fins; according to Kier et al. (1985), they probably use this
peripheral mechanical information during swimming to regulate and control their fin
beat.

7.2.4 Geomagnetism perception

The mechanisms underlying the perception of the geomagnetic field are not fully under-
stood; however, this ability is important to the navigation of many birds, fishes and
reptiles (Winklhofer, 2009). The mollusk Tritonia diomedea appears to use its percep-
tion of the geomagnetic field to orient towards the shore (Willows, 1999). No evidence
yet exists for whether cephalopod mollusks can detect the geomagnetic field.

7.2.5 Summary
Cephalopods have well-developed sensory systems that could support orientation and
navigation in the field. Vision is clearly useful for orienting in shallow, lit waters.
Chemoreception could allow cephalopods to locate prey, particular water currents,
spawning locations or conspecific aggregations. Mechanoreception could provide ori-
entation cues to support short-distance navigation (e.g. integration of movements in the
three-dimensional environment) as well as onshore and offshore seasonal migrations.

7.3 Dispersal

Dispersal refers to moving away from a particular location, often the natal location, that is
typically undirected and without a later return. These latter features distinguish dispersal
154 Christelle Jozet-Alves, Anne-Sophie Darmaillacq and Jean G. Boal

Figure 7.2 Giant Pacific Octopus, Enteroctopus dofleini, paralarvae. They do not have the same
way of life as adults, which are generally benthic. (Photograph courtesy of J. Kocian.) See plate
section for color version.

from migration. Dispersal is often accomplished by passive means, such as drifting with
currents (e.g. planktonic larvae, Figure 7.2). Cephalopod hatchlings, whether benthic
or planktonic, likely use similar mechanisms to disperse, but further investigation is
required (Boletzky, 2003).

7.3.1 Field observations

The lives of cephalopods between hatching and adult recruitment are poorly described,
with particular difficulties in both sampling and species identification (Boyle & Rod-
house, 2005). Field samples yield distinct cohorts (Arkhipkin, 1993) with different
physiographic distributions (Octopus: Faraj & Bez, 2007; Todarodes: Kawabata et al.,
2006), adding to the confusion.
Cephalopod adults are semelparous and a single species can spawn at different times
of the year within the same area (Loligo: Ito, 2007). The duration of the hatching period
of a single egg mass can range from 2 days (Octopus laqueus, Kaneko et al., 2006)
to 78 days (Enteroctopus dofleini, Gabe, 1975). Hatchlings may be either planktonic
 Navigation in cephalopods 155

or benthic, depending on the species (Boletzky, 2003) and the duration of a planktonic
stage can range from 3 weeks (Octopus joubini, Forsythe & Toll, 1991) to 6 months (E.
dofleini, Villanueva & Norman, 2008).
The horizontal distances covered during dispersal may range from several hundreds of
meters in small bottom dwelling species without a planktonic phase (e.g. sepiolid squids)
to hundreds of kilometers in pelagic species (e.g. teuthid squids and octopods; Boyle &
Boletzky, 1996; for a review, see Villanueva & Norman, 2008). Several approaches have
been taken to assess dispersal distance. Genetic structuring of the common cuttlefish
(S. officinalis) in the Atlantic–Mediterranean region was evaluated using cytochrome
oxidase mitochondrial DNA (Pérez-Losada et al., 2007) and isolation by distance was
shown to be the main factor driving genetic structuring of Sepia populations. This result
supports the hypothesis of restricted dispersal ability probably due to its nectobenthic
way of life along with the absence of any planktonic stage. An analysis of natural
elemental signatures in cephalopod statoliths suggested that juvenile squid (Gonatus
fabricii) inhabit surface waters while larger animals move to colder, deeper waters
(Arkhipkin et al., 2004; Zumholz et al., 2007). Few studies to date have specifically
examined the potential of mass-marking techniques for the early life-history stages of
cephalopods (Pecl et al., 2010; Payne et al., 2011). Such techniques offer considerable
potential for understanding juvenile dispersal.

7.3.2 Possible mechanisms

“Paralarvae,” planktonic stages that do not undergo metamorphosis before settlement
(Young & Harman, 1988), possess considerable potential for dispersal. Hatchlings may
disperse actively both in the water column and along the seabed (Boletzky, 2003).
Strong positive phototaxis is a common response for hatchlings and some later paralarval
(immature; see Young & Harman, 1988) stages, but among benthic species, this response
reduces, disappears or reverses after settlement (Villanueva & Norman, 2008). It is not
known what factors control settlement of the planktonic paralavae of benthic species
(Semmens et al., 2007).
All hatchlings may undergo some passive dispersal due to water movements. Soon
after hatching, squid paralarvae are active predators on other zooplankton; although
they are active swimmers using jet propulsion during this life stage, their distribution
is essentially dependent on oceanic circulation (Moreno et al., 2009). In western Iberia,
temperature and upwellings were shown to be the most important variables in modulating
seasonality and distribution of the paralarvae of Loligo vulgaris, O. vulgaris, sepiolids
and ommastrephids (Moreno et al., 2009). The influence of the physical environment
is particularly pronounced for the paralarvae of O. vulgaris, which follows distinct
patterns according to the oceanography of the western Iberia and the Gulf of Cádiz
systems (Moreno et al., 2009).

7.3.3 Individual differences

Differences between individual cephalopods could influence individual movement tra-
jectories. Animal personalities (temperaments, behavioral syndromes) are common
156 Christelle Jozet-Alves, Anne-Sophie Darmaillacq and Jean G. Boal

across taxa and have important evolutionary and ecological implications that have only
recently come under close scrutiny (e.g. Reale et al., 2007; Stamps, 2007). Individual
differences have been found in exploration, aggressiveness, reactivity and boldness in
the dumpling squid (Euprymna: Sinn & Moltschaniwskyj, 2005; Sinn et al., 2010) and
enduring individual differences in reactions have been documented in octopuses (Mather
& Anderson, 1993). Among vertebrates, variations in boldness, sociability or aggres-
siveness influence dispersal tendency (Cote et al., 2010). This possibility remains to be
explored in cephalopods.

7.3.4 Discussion and conclusions

Dispersal is an important component of spatial behavior, yet relatively little is known
about the life histories of juvenile cephalopods. Mechanisms for dispersal are unlikely
to involve complex cognition, however.

7.4 Migrations

Migration typically refers to long-distance movements, often in connection to seasonal

or ontogenetic changes. Often, but not always, migrations are cyclical and involve a
return to the original location by either the individuals or the population. Most of what
we know about migrations in cephalopods comes from fisheries data, tagging studies
(both traditional tag/recapture and electronic tracking), systematic sampling and statolith
and other tissue composition analyses (reviewed in Semmens et al., 2007).

7.4.1 Field observations

Cephalopod migrations range from diel vertical migrations of a few hundred meters (e.g.
Dosidicus gigas, Gilly et al., 2006; multiple species, Watanabe et al., 2006) to annual
horizontal movements of hundreds of kilometers (e.g. Illex argentinus, Arkhipkin, 1993;
see Boyle & Rodhouse, 2005) to the combination of vertical and horizontal components
in onshore-offshore migrations (e.g. Villanueva, 1992; Olyott et al., 2007; reviewed
in Hanlon & Messenger, 1996). Cephalopod migrations support the combined needs of
foraging, avoiding predators, mating and spawning and, in the case of vertical migrations,
managing thermal stress (D. gigas, Zeidberg & Robison, 2007; reviewed in Olson &
Young, 2007). When migrating, some species traverse the same path in both directions,
e.g. traveling north as they grow and traveling south along the same route when they
return to traditional spawning grounds as mature adults (e.g. Kiyofuji & Saitoh, 2004),
but other species traverse different routes (e.g. Choi et al., 2008).
Field observations suggest that the key factors in many cephalopod migrations
are ocean productivity and hydrographic/physiographic parameters: temperature, light,
salinity, oxygen and water currents (Sims et al., 2001; Wang et al., 2003; Gilly et al.,
2006; Stark, 2008; Chen C. S. et al., 2007, Chen X. J., 2007; Ito, 2007; Onitsuka
et al., 2010; Rosa & Seibel, 2010; Sakaguchi, 2010; for a review, Pierce et al., 2008).
 Navigation in cephalopods 157

Consistent with these observations is the apparent shift in migrations and distributions
with climate change (Sims et al., 2001; Kidokoro et al., 2010).
Most cephalopods are semelparous, spawning just once before they die, and live
only for about a year (Hanlon & Messenger, 1996), providing little opportunity to
learn from previous migratory experience. Curiously, although some populations show
clear evidence of genetic mixing (e.g. Loligo opalescens, Reichow & Smith, 2001;
O. vulgaris: Oosthuizen et al., 2004), other species show genetically distinct spawning
populations (e.g. S. officinalis: Pérez-Losada et al., 2002; Loligo forbesi: Shaw et al.,
1999; Martialia hyadesi: Brierley et al., 1993; Loligo pealeii: Buresch et al., 2006)
suggestive of population-level site fidelity in spawning locations.

7.4.2 Possible mechanisms

The mechanisms underlying long-distance migration, particularly for a first or only
transit, are not fully understood for any animal (Wehner et al., 1996). For marine migrants,
potential cues for orientation include celestial cues, the geomagnetic field, vibrations
and water currents.
Celestial cues can provide directional information for migrating animals. Navigation
based on the movements and configurations of the sun and stars has been demonstrated
in insects (e.g. Akesson & Wehner, 2002) and birds (e.g. Able & Able, 1996). Navigation
based on polarization patterns in the sky has been documented in insects (Marshall &
Cronin, 2011). An ability to perceive polarized light could also support underwa-
ter navigation, especially under restricted conditions (e.g. turbidity; Lerner et al.,
2011). Some squids exhibit spontaneous preferential swimming direction relative to the
e-vector orientation of polarized light (Eupryma morsei, Sepioteuthis lessoniana: Jan-
der et al., 1963). It remains unclear whether this polarization-based body orientation is
biologically meaningful in the context of navigation and migration. It could play a role
comparable to phototaxis (see Saidel et al., 1983 for further discussion).
The geomagnetic field can provide both directional and map information for migrating
animals (e.g. Lohmann et al., 2004). It is not known whether any cephalopods detect the
geomagnetic field.
Vibration (e.g. reef sounds, surf ) provides directional information that is used for
orientation by some marine species (Pomacentrid larvae, Tolimieri et al., 2004; decapod
crustaceans, Montgomery et al., 2006). Behavioral studies in S. officinalis showed that
young animals (<4 months post-hatching) responded to low frequencies ranging from
20 to 600 Hz by changing their body patterns, moving and trying to bury (Komak et al.,
2005). Vibration cues could be used for orientation in cephalopods, but this possibility
has still to be investigated.
Water currents can provide both directional and map information from the speed
and direction of the flow, the temperature, salinity and clarity of the water, and the
chemicals carried in the current, whether of biologic or geologic origin. Detection
of the spatial gradient of chemical cues in water can also be used to orient (Moore
et al., 1991). Laboratory experiments have demonstrated that both octopuses and cuttle-
fishes detect biologically relevant odors (different waters, conspecifics and prey items;
158 Christelle Jozet-Alves, Anne-Sophie Darmaillacq and Jean G. Boal

e.g. S. officinalis: Boal & Golden, 1999) and can use these odors to orient (S. officinalis:
Boal et al., 2010; Octopus bimaculoides: Walderon et al., 2011); presumably, squids are
similarly capable.
It is not currently understood how cephalopods locate traditional, common spawning
grounds (e.g. Sepia apama in the Spencer Gulf, South Australia: Hall & Hanlon, 2002).
Such sites may be consistent from a large scale viewpoint, but the exact locations can
vary from year to year (e.g. Doryteuthis opalescens: Young et al., 2011). Late arrivers
could be guided by odors from conspecifics or conspecific eggs (S. officinalis: Boal
et al., 2010; O. bimaculoides: Walderon et al., 2011) or, at close range, the sight of the
eggs, themselves (King et al., 2003). Early learning (imprinting) could account for this
location fidelity. Recent research has demonstrated that biologically significant learning
occurs at or even before hatching (e.g. S. officinalis: Dickel et al., 2000; Darmaillacq
et al., 2006, 2008; Guibé et al., 2010) and that adults are capable of oriented movements
in response to the odors of conspecifics or conspecific eggs (S. officinalis: Boal et al.,
2010; O. bimaculoides: Walderon et al., 2011). It is possible that young cephalopods
could learn the characteristics of the hatching location and then later return to that
location to spawn, similar to salmon (Scholz et al., 1976); this hypothesis remains to
be explored. Alternatively, developmentally timed responses to physiographic features
could allow a return to particular locations via specific water currents.

7.4.3 Discussion and conclusions

Long-distance migrations can be accomplished through sensory and developmentally
linked orientation behavior (Shea & Vecchione, 2010) as well as through true naviga-
tion. Simple diel vertical movements within the pelagic ocean probably do not involve
cognition, for example. Long-distance horizontal migrations could involve cognition if
return travel followed outbound travel routes, implicating learning and early learning of
the hatching site could facilitate a later return to that location for spawning, suggestive of
imprinting. Alternatively, endogenous biologic timing mechanisms coupled with inter-
nal biases, attention or preferences for particular physical parameters could facilitate
such movement patterns. Such mechanisms, while both complex and fascinating, do not
require complex cognition. There is ample room for further investigation of these topics.

7.5 Local movements

Local movements include travels within a home range. Such travels open the possibility
of learning local landmarks and the locations of significant resources. Such learning can
result in more efficient and potentially purposeful movements in space.

7.5.1 Field observations

At least some cephalopods navigate within a local area. Many octopus species are central
place foragers, traveling away from one or more (O. dofleini: Mather et al., 1985) home
 Navigation in cephalopods 159

Figure 7.3 Octopus cyanea in its den. Note the stones surrounding the den entrance. (Copyright,
A.-S. Darmaillacq.) See plate section for color version.

den and then returning to the den repeatedly for shelter (O. bimaculatus: Ambrose,
1982; Octopus briareus: Aronson, 1986, 1989, 1991; Octopus cyanea: Van Heukelem,
1966; Yarnall, 1969; Forsythe & Hanlon, 1997; O. dofleini: Hartwick et al., 1978, 1984;
Mather et al., 1985; O. vulgaris: Altman, 1967; Kayes, 1974; Mather, 1988, 1991a,
1991b; Mather & O’Dor, 1991; for reviews, Boyle, 1983, 1987; Figure 7.3).
The local movements of octopuses are likely supported by sophisticated spatial orien-
tation abilities. At least some octopuses travel considerable distances before returning
to their dens (up to 40 m, O. cyanea, Forsythe & Hanlon, 1997) and they sometimes
remain away for extended periods of time before returning (up to 7 h, O. dofleini, Mather
et al., 1985). They often jet-swim directly home without contacting the bottom substrate
or retracing their outbound routes (O. vulgaris, Mather, 1991a; O. cyanea, Forsythe &
Hanlon, 1997). Individuals avoid recently visited areas on subsequent hunting trips, sug-
gesting that they remember where they foraged previously (O. vulgaris: Mather, 1991a;
O. cyanea: Forsythe & Hanlon, 1997). Octopuses may use a particular den continuously
for anywhere from 1 day to as long as 5 months (O. dofleini, Hartwick et al., 1984), after
which they move to a new location, presumably because of prey depletion. Moving to
a new location with new features and spatial relationships is likely to promote flexible
and long-term spatial memory.
The natural history of cuttlefish also provides evidence for spatial learning (S. offic-
inalis: Boletzky, 1983; Hanlon & Messenger, 1988). In particular, cuttlefish appear to
160 Christelle Jozet-Alves, Anne-Sophie Darmaillacq and Jean G. Boal

minimize their hunting time and then quickly return to safe places. Recent tagging
experiments indicate that S. apama forages away from and then returns to specific rocky
outcroppings (Aitken et al., 2005). Monitored individuals showed high site fidelity and
small home ranges (between 5300 m2 and 27 300 m2 ), suggesting that the cuttlefish
confined their foraging to a restricted area. Tagged cuttlefish produced fewer data from
movements after several days. They hypothesize that the cuttlefish may have settled into
a den (rock crevice, no signal) emitting only brief signals during foraging excursions.
S. apama rest more than 95% of the time (Aitken et al., 2005), supporting the idea that
long rest periods in safety are punctuated by brief foraging bouts in the open, nearby.
Regular travels between safe resting places and good hunting grounds would promote
spatial learning.
In the field, cuttlefish frequently swim in and around vertical barriers. Spatial learning
could aid in negotiating these obstacles (Sanders & Young, 1940). In the old harbor of
Cesarea, Israel, a single cuttlefish (S. officinalis) was tracked for over 50 m, at water
depth of 5–3 m (N. Shashar, personal communication), as it swam in a straight line
until it reached a small (2 m tall) underwater wall. There, without touching any surfaces
and with great speed, it swam approximately 3 m up an inclined ledge and turned
into a small crevice, not much larger than the animal, where it went into hiding. The
animal did not seem to hesitate, change speed or collide with obstacles, indicating that
it was familiar with the spatial layout of the area. Its behavior suggests the use of
three-dimensional navigation to reach a location that was both distant and higher than its
original position. Alternatively, it could have used two orthogonal yet flat representations
of the environment, one leading to the wall and the other leading up the surface of the
wall into the crevice.
Squids are predominantly neritic (e.g. L. forbesi) or pelagic (e.g. D. gigas) and direct
underwater observations are difficult to obtain. Most of the daily activities of squids are
known from tracking experiments and results to date vary widely. Individual L. vulgaris
reynaudii were monitored as they circled egg beds in South Africa (Sauer et al., 1997).
Each dawn, the squid moved onshore (several kilometers) to mate in small and well-
defined areas near egg beds on the substrate; at dusk, most individuals moved offshore,
probably for feeding or resting. In L. forbesi, most individuals (except the largest ones)
tended to remain in the vicinity of the reef (L. forbesi: O’Dor et al., 1994; Sepioteuthis
australis: O’Dor et al., 2002) and most of the smaller individuals moved very little over
periods of up to 19 days. Tracking experiments are more scarce for pelagic species.
Individual D. gigas spent most daylight hours at depths greater than 250 m and foraged
in near-surface waters at night (Yatsu et al., 1999; Gilly et al., 2006). Similar daily
vertical migrations have been documented in other squid species (e.g. Ommastrephes
bartramii: Murata & Nakamura, 1998; Hiroyuki, 2001). Such diel migrations do not
require spatial learning.

7.5.2 Laboratory evidence of spatial memory

Octopuses are good candidates to study spatial learning. In the wild, they are central-
place foragers that relocate periodically, depending on den size and the presence of
 Navigation in cephalopods 161

preferred prey (Mather, 1994). Spatial memory in octopuses was initially demonstrated
by Walker et al. (1970), who trained octopuses (Octopus maya) to find a goal location
with water in a dry, T-shaped apparatus. The stem of the “T” served as a start box and the
two identical arms of the “T” led to the goal compartments situated at either side of the
stem. In this experiment, five octopuses learned to enter the goal compartment opposite
to their initial side-turning preference to regain access to seawater. After 27 days of
training, all octopuses were performing without errors, providing convincing evidence
of spatial learning in octopuses.
Exploratory learning is considered a natural manifestation of spatial learning (Gallis-
tel, 1993). According to O’Keefe and Nadel (1978), exploration is necessary for animals
to build maps of new environments and to update existing maps. To test the hypoth-
esis that octopuses show exploratory learning, Boal and collaborators performed two
separate experiments (O. bimaculoides, Boal et al., 2000). To determine if octopuses
spontaneously move about, they placed individual octopuses in an unfamiliar environ-
ment and recorded the octopuses’ behavior. The octopuses gradually decreased their
activity over the course of 3 days, consistent with the hypothesis that the octopuses had
been exploring. To evaluate whether they actually learned (retained information) about
their new environment during their movements, the octopuses were placed in a different
aquatic environment with two inset (deeper water) burrow sites, only one of which was
open, for a period of 23 h and then removed (Boal et al., 2000). After a 24 h delay, the
octopuses were placed back into this same environment, but with much shallower water
in the arena. Most octopuses approached the previously open burrow first, demonstrating
that they had learned the location of the open burrow, evidence for exploratory learning.
More recently, Hvorecny et al. (2007) demonstrated that octopuses can learn two differ-
ent spatial problems simultaneously (conditional discrimination). They tested octopuses
(O. bimaculoides) in two distinct maze apparatuses that differed in the visual and tactile
cues provided. As before, each maze was provided with two inset burrows, one open and
one closed. The octopuses were placed for 16 h per day in one of the two maze configura-
tions, which were presented alternately throughout training (10 days total, 5 days in each
maze). Six out of ten octopuses learned to go directly to the open burrow in both mazes.
Unlike many octopus species, cuttlefishes have not been described as having dens
(but see Aitken et al., 2005 in S. apama) and are thought to rely primarily on camouflage
for defense. This behavior pattern could explain why researchers have started to explore
the abilities of cuttlefish to orient and navigate only recently. Preliminary observations
showed that cuttlefish (S. officinalis) placed in a large artificial pond spontaneously
moved around and appeared to learn its features (Karson et al., 2003). This result
suggested that cuttlefish, too, explore a new environment. Karson et al. (2003) then
designed mazes to assess cuttlefish (S. officinalis) spatial learning abilities. Although
food reinforcement is used in mazes classically (in rats, birds, fish or even insects), escape
was chosen to make the task more comparable to a natural spatial learning problem. The
cuttlefish were trained to exit a circular testing arena with two exit holes cut on opposite
sides. The exit holes were surrounded by visual cues (striped or spotted panels of fabric),
with only one exit hole opened. The cuttlefish demonstrated the ability to solve the task;
moreover, cuttlefish improved over serial reversals (once the cuttlefish trained, the open
162 Christelle Jozet-Alves, Anne-Sophie Darmaillacq and Jean G. Boal

hole was closed and the opposite one was opened), showing that the cuttlefish were
able to quickly update their spatial memory with the most recent information. More
recently, Hvorecny et al. (2007) tested the cuttlefish S. officinalis and Sepia pharaonis in
modified versions of this same procedure. This time, the presence of a centrally located
cue indicated which of the two exit holes was open (striped or spotted; right or left).
Cuttlefish learned to select the correct exit hole, even when the learning trials with the
two different central cues were intermixed. These experiments clearly demonstrate that
complex spatial memory is well within the abilities of cuttlefishes.

7.5.3 Possible mechanisms

There are four systems in which spatial information is represented (Dolins & Mitchell,
2010). In order of cognitive complexity, these systems include: (i) use of a single cue
or beacon (self in response to a cue); (ii) use of sequential cues or route following (self
in response to individual cues within a sequence); (iii) path integration (self as a cue);
and (iv) piloting (relational and simultaneous use of multiple cues). The first three of
these are self-referential mechanisms. With (i) single cues or beacons, individuals orient
their own position in space according to one or more cues. This form of orientation
can take the form of trail following (e.g. gastropod slime trails, Davies & Blackwell,
2007), taxes (i.e. orienting with respect to a sensory gradient, Campan, 1997) or bea-
coning (i.e. movements directed towards a particular landmark, Cheng & Spetch, 1998).
(ii) The use of sequential cues or route following can be accomplished through motor
patterns that link sequences of movements or responses into stereotyped series (e.g.
arthropods, Bell, 1985). (iii) Path integration, also called “dead reckoning,” is a naviga-
tional process by which signals generated during locomotion allow the subject to update
its position as if it had summed the vectors of its outbound journey. This computation
allows the subject to return directly home without tracing the path of its outbound jour-
ney. This strategy is found in many species, from arthropods to humans (ants: Müller &
Wehner, 1988; Wehner, 2003; crabs: Layne et al., 2003; mammals: Etienne et al., 1996).
(iv) Piloting is the most sophisticated navigational mechanism and is not self-referential.
Environmental features are combined together with a directional sense to accomplish
orientation (“cognitive mapping”; Save et al., 1998). Laboratory experiments with octo-
puses and cuttlefish have explored the first two of these systems. Unfortunately, no study
has yet directly tested either path integration or piloting hypotheses in cephalopods (see
Alves et al., 2008, for further discussion about piloting in cephalopods).

Single cues
Many animals return home by secreting a chemical trail on a solid substrate on their
outbound route and then retrace that trail when returning. This strategy, called trail
following, depends on contact chemoreception (e.g. limpets, Cook, 1971; termites, Kaib
et al., 1982). In the field, Mather (1991a) and Forsythe and Hanlon (1997) observed that
octopuses (O. vulgaris and O. cyanea) traveled by jetting through the water when leaving
or returning home; consequently, they were not in contact with the substrate. In addition,
they returned directly to their den afterwards rather than tracing their outbound route.
 Navigation in cephalopods 163

As a consequence, it is unlikely that octopuses use contact chemoreception for short-

distance navigation. In support of this hypothesis, raking the gravel substrate between
trials in a laboratory maze experiment had no effect on the octopuses’ orientation
behavior (Boal et al., 2000). However, these observations do not exclude the possibility
that octopuses are able to use this strategy in particular environmental conditions (low
visibility, for example). Neither cuttlefishes nor squids maintain regular, tactile contact
with the substrate when they travel, making it unlikely that these species use trail
following, either.
For local travel, cephalopods seem more likely to orient using visual landmarks,
particularly those with high contrast and distinctive polarization. One possible orienting
strategy consists of using landmark guidance or beaconing (Gallistel, 1993; also called
stimulus-approach behaviour, Schmajuk & Thieme, 1992). In this case, the animal will
learn to go to a particular feature that marks the location of a goal, regardless of the
motor behavior involved. In a series of experiments, Mather (1991a) has investigated
octopuses’ ability to use visual landmarks to orient. These cues can be used in a variety
of ways to provide information about the location of a goal. The simplest way of using
them is to rely on a single visual cue as a beacon marking a goal: for example, goldfish
can learn to use simple objects as beacons to locate food which has been buried under a
gravel substrate in a testing tank (Warburton, 1990). Mather (1991a) trained octopuses
(Octopus rubescens) to go to a piece of plastic tubing (the beacon) to find a food
reward located within a featureless circular tank. The octopuses learned to approach the
plastic tubing even when it was moved around the tank. This experiment clearly showed
octopuses’ ability to rely on a single landmark to label the location of a goal.
Some learning experiments are designed to provide very precise information about the
features that cephalopods attend to and how they learn to solve spatial problems. These
studies used an experimental technique called the transformational approach (Tinbergen,
1972). Animals are trained to solve a spatial task and then their environment is modified
in some way (e.g. by displacing landmarks) to determine which strategy the animals
were using to solve the task (e.g. using landmarks or a motor sequence). Using this
approach, cuttlefish (S. officinalis) were tested in a cross-maze apparatus (Alves et al.,
2007). The task required animals to locate a goal compartment at the end of one arm of
the maze. The reward for solving the task was 15 min in a darkened goal compartment
with a sandy bottom into which the cuttlefish could bury. The cuttlefish could either
learn a motor sequence (e.g. turn left) or orient using distal visual cues (water-pipes,
sets of shelving in the laboratory room). At the end of training, the cuttlefish were
placed again in the maze but in the opposite starting arm. Most of the cuttlefish made
the formerly rewarded motor sequence (turn right or left), swimming in the opposite
direction of the rewarded goal compartment. Interestingly, one out of the ten cuttlefish
swam in the correct direction, apparently using the distal visual cues rather than a
motor sequence. In a second version of the experiment, the cross-maze was surrounded
by black curtains and two proximate visual cues were provided just above the water
(striped and spotted PVC panels). When tested in the opposite start arm of this maze
configuration, half of the cuttlefish used the proximate visual cues while the other half
used a motor sequence. These two experiments demonstrated that cuttlefish can use at
164 Christelle Jozet-Alves, Anne-Sophie Darmaillacq and Jean G. Boal

landmarks

- + OR + -
Training
Step 1:

Random
e-vector position of the
of the down- reward (=shelter)
welling light between trials

landmarks
Single cue tests

- + Cuttlefish succeed to
orient with
+ -
Step 2:

landmarks e-vector e-vector

OR of the down-
(9/10 trials with (9/10 trials with
correct choices) correct choices) welling light

+ +
Step 3:
Probe

e-vector landmarks
indication indication

Figure 7.4 Schematic representation of three steps of the learning procedure. See Cartron et al.
(2012) for the detailed procedure.

least two different strategies to solve spatial tasks: visual cues and motor sequences.
The availability and salience of the visual cues influenced which strategy was used.
Similar results have been obtained previously with vertebrates (Restle, 1957; Carman &
Mactutus, 2001).
Cartron et al. (2012) trained cuttlefish (S. officinalis) in a Y-maze with two kinds of
visual cues, the e-vector of a polarized light (either perpendicular or parallel) and two
PVC panels (one striped and one spotted). Cues were placed just above the water surface
(Figure 7.4). During training, both visual cues were available. At the end of training, the
experimenters provided just one of the two types of cues (i.e. either the PVC panels or
the filter linearly polarizing the light). All cuttlefish tested were able to orient with either
one of the visual cues, presented separately. This study also showed for the first time
that cuttlefish can orient either parallel or perpendicular to the e-vector of a polarized
light. The results clearly indicate that cuttlefish, like marine mammals (Gibson & Shet-
tleworth, 2005) and insects (Heinze & Homberg, 2007), acquire redundant information
simultaneously.

Sequential cues
Animals can also use a series or chain of landmarks to compose a route when multiple
landmarks are available. For example, ants learn and recognize landmarks distributed
 Navigation in cephalopods 165

along a route and they correct their course relative to these landmarks (Collett et al.,
1992). In Mather’s experiments (1991a; above), when a box and a dish were added in the
tank, the octopus (O. rubescens) moved first to the larger landmark (the box) and, next,
oriented to the plastic tubing associated with food. These results could be interpreted as
the use of a chain of landmarks to follow a route leading to the food location.
Accurate assessment of the layout of multiple objects in the immediate environment
allows animals to plan routes and to solve spatial problems, such as those posed by
obstructions. Detouring around objects can be cognitively challenging, requiring the
animal to synthesize spatial memories in potentially new ways. Octopuses, at least,
appear to be capable of this type of spatial reasoning. Schiller (1949), Boycott (1954)
and Wells (1964) tested the ability of octopuses (O. vulgaris) to detour around opaque
partitions to reach a prey item (a live crab). The crab was either (a) shown in the corner of
an opaque partition and then displaced behind that partition (Boycott 1954) or (b) directly
visible but located behind a transparent partition (Schiller, 1949; Boycott, 1954; Wells,
1964). How animals responded to their first trial in these apparatuses was indicative
of their spatial representation of the location of the goal. Some, although not all, of
the octopuses successfully reached the crab at the first trial (1 in 5 when the crab was
displaced behind the partition, Boycott, 1954; 8 in 29 when the crab was visible behind
a transparent wall, Wells, 1964). In Schiller (1949) and Wells (1964) experiments, the
transparent partition that separated the octopuses from the prey could have been such
an unnatural stimulus that the animals failed to perceive it as an obstacle (for further
discussion of possible explanations for the behavior of the octopuses that failed this
task, see Alves et al., 2008). The authors then attempted to train the octopuses that had
failed by providing repeated trials with the same detour problem. The octopuses showed
an improvement in performance over time, as measured by a decrease in the time spent
pushing and attacking the transparent partition. Performances were erratic, however,
and the octopuses never went straight to the edge of the partition (Boycott, 1954). Upon
closer analysis, Schiller (1949) and Wells (1964) found that the octopuses needed to
remain in tactile contact with the wall separating them from the goal (Schiller, 1949)
and if the tactile contact was lost, they needed to maintain a continuous visual fixation
on the wall itself (Wells, 1964).
More recently, Moriyama and Gunji (1997) tested octopuses (O. vulgaris) in a tank
with four partitioning boards, either transparent or opaque. Animals had to learn to
detour these obstacles to reach the end of the tank to obtain a food reward. With
repeated trials, octopuses reduced the time they spent detouring around the obstacles.
By studying swimming actions, the researchers asserted that octopuses “evaluated” their
environment and estimated their actions, before proceeding. Octopuses in such situations
may be planning their route. Taken together, experiments suggest that octopuses, at least,
have the ability to maintain orientation towards a goal while detouring around obstacles.

7.5.4 Individual differences

Spatial learning can vary between individuals, based on differences in age, experience
and sex. The range size hypothesis predicts sex dimorphism in spatial learning abilities
in species in which range expansion is different between males and females (e.g. sex
166 Christelle Jozet-Alves, Anne-Sophie Darmaillacq and Jean G. Boal

differences in promiscuous meadow voles) and not in other species (e.g. no sex difference
in monogamous voles; Gaulin & FitzGerald, 1986, 1989; reviewed in Jones et al.,
2003). Jozet-Alves et al. (2008) assessed spatial learning performances of male and
female cuttlefish (S. officinalis), either before or after sexual maturation, in a T-maze
(procedure described above; Alves et al., 2007). They also determined the spatial strategy
preferentially used by each cuttlefish to evaluate whether males and females use different
spatial orientation strategies. Although cuttlefish males and females did not differ in the
time they took to learn the spatial task in a T-maze, sexually mature males were more
likely to attend to the visual cues provided above the apparatus to solve the T-maze task
while sexually mature females preferentially relied on a motor sequence (right versus left
turn). This difference in strategy did not lead to a sex difference in overall performance.
The authors also showed that sexually mature males traveled a longer distance when
placed in an open-field compared with female and immature cuttlefish, suggesting a
possible greater propensity for exploring new environments by males. This study was
the first to demonstrate cognitive differences between sexes in an invertebrate. Moreover,
the data conform to the predictions of the range size hypothesis.

7.5.5 Discussion and conclusions

Both field observations and laboratory experiments provide clear evidence that at least
some cephalopods navigate within local areas for the purposes of foraging and sheltering.
These travels are accomplished by learning the locations of external cues (both single and
sequential; see above) and learning and remembering patterns of movements (turning
right/left, route following; see above). The more cognitively complex forms of orienta-
tion, path integration and piloting (see above), merit further investigation, particularly
in light of anecdotal observations indicating complex and purposeful traveling behavior.

7.6 Summary and future directions

Cephalopods show a wide range of orientation behavior in the field, ranging from pre-
cise short-distance navigation, particularly in octopuses, to long-distance orientation
with respect to physiographic characteristics, particularly in squids. Newer methods,
ranging from electronic tags (e.g. Gilly et al., 2006) to molecular analyses (see Sem-
mens et al., 2007) are likely to provide more precise information about particular species’
movement patterns in the near future. Data arising from these methods will allow the
generation of testable hypotheses for mechanisms underlying cephalopod orientation
behavior. In parallel, brain studies (e.g. electrophysiology, pharmacology and immuno-
histochemistry) have the potential to uncover how cephalopods represent space and the
sensory bases for these representations. Different spatial behavior mechanisms may be
supported by different neural representational mechanisms.
A better understanding of cephalopod movements will be of both theoretical and
practical value. For evolutionary biologists, results of research with cephalopods in
comparison with other lineages will provide clearer understanding of the evolution of
 Navigation in cephalopods 167

spatial orientation. For industrial and artisanal fisheries, such research can provide practi-
cal information useful for improving stock assessment and fisheries management (Boyle
& Rodhouse, 2005; Faraj & Bez, 2007). The greatest benefit will come from an inter-
disciplinary approach that integrates population-level analyses obtained from fisheries
investigations with individual-level analyses obtained from laboratory and field studies.
Such an approach will require greater teamwork between disparate subdisciplines than
has been typical in the past.

Acknowledgements

We would like to express our gratitude to the long history of curious and dedicated
scientists, both in the field and in laboratories worldwide, who have carefully and
painstakingly built our understanding of the natural world.

References

Able, K. P. and Able M. A. (1996). The flexible migratory orientation system of the savannah
sparrow (Passerculus sandwichensis). Journal of Experimental Biology, 199: 3–8.
Aitken, J. P., O’Dor, R. K. and Jackson, G. D. (2005). The secret life of the giant Australian
cuttlefish Sepia apama (Cephalopoda): behaviour and energetics in nature revealed through
radio acoustic positioning and telemetry (RAPT). Journal of Experimental Marine Biology and
Ecology, 320: 77–91.
Akesson, S. and Wehner, R. (2002). Visual navigation in desert ants Cataglyphis fortis: are
snapshots coupled to a celestial system of reference? Journal of Experimental Biology, 205:
1971–1978.
Altman, J. S. (1967). The behaviour of Octopus vulgaris Lam. in its natural habitat: a pilot study.
Underwater Association Reports, 1966–1967: 77–83.
Alves, C., Chichery, R., Boal, J. G. and Dickel, L. (2007). Orientation in the cuttlefish Sepia
officinalis: response versus place learning. Animal Cognition, 10: 29–36.
Alves, C., Boal, J. G. and Dickel, L. (2008). Short-distance navigation in cephalopods: a review
and synthesis. Cognitive Processing, 9: 239–247.
Ambrose, R. F. (1982). Shelter utilization by the molluscan cephalopod Octopus bimaculatus.
Marine Ecology Progress Series, 7: 67–73.
Arkhipkin, A. I. (1993). Age, growth, stock structure and migratory rate of pre-spawning short-
finned squid Illex argentinus based on statolith ageing investigations. Fisheries Research, 16:
313–338.
Arkhipkin, A. I., Campana, S. E., Fitzgerald, J. and Thorrold, S. R. (2004). Spatial and temporal
variation in elemental signatures of statoliths from the Patagonian longfin squid (Loligo gahi).
Canadian Journal of Fisheries and Aquatic Sciences, 61: 1212–1224.
Aronson, R. B. (1986). Life history and den ecology of Octopus briareus Robson in a marine lake.
Journal of Experimental Marine Biology and Ecology, 95: 37–56.
Aronson, R. B. (1989). The ecology of Octopus briareus Robson in a Bahamian saltwater lake.
American Malacological Bulletin, 7: 47–56.
Aronson, R. B. (1991). Ecology, paleobiology and evolutionary constraint in the octopus. Bulletin
of Marine Science, 49: 245–255.
168 Christelle Jozet-Alves, Anne-Sophie Darmaillacq and Jean G. Boal

Bell, W. J. (1985). Sources of information controlling motor patterns in arthropod local search
orientation. Journal of Insect Physiology, 31: 837–847.
Bellingham, J., Morris, A. G. and Hunt, D. M. (1998). The rhodopsin gene of the cuttlefish Sepia
officinalis: sequence and spectral tuning. Journal of Experimental Biology, 201: 2299–2306.
Bleckmann, H., Budelmann, B. U. and Bullock, T. H. (1991). Peripheral and central nervous
responses evoked by small water movements in a cephalopod. Journal of Comparative Physi-
ology A, 168: 247–257.
Boal, J. G. (1996). A review of simultaneous visual discrimination as a method of training
octopuses. Biological Reviews, 71: 157–190.
Boal, J. G. and Golden, D. K. (1999). Distance chemoreception in the common cuttlefish, Sepia
officinalis (Mollusca, Cephalopoda). Journal of Experimental Marine Biology and Ecology,
235: 307–317.
Boal, J. G., Dunham, A. W., Williams, K. T. and Hanlon, R. T. (2000). Experimental evidence
for spatial learning in octopuses (Octopus bimaculoides). Journal of Comparative Psychology,
114: 246–252.
Boal, J. G., Shashar, N., Grable, M. M., et al. (2004). Behavioral evidence for intraspecific signaling
with achromatic and polarized light by cuttlefish (Mollusca: Cephalopoda). Behaviour, 141:
837–861.
Boal, J. G., Prosser, K. N., Holm, J. B., et al. (2010). Sexually mature cuttlefish are attracted to
the eggs of conspecifics. Journal of Chemical Ecology, 36: 834–836.
Boletzky, S. v. (1983). Sepia officinalis. In Cephalopod life cycles. Species accounts, vol. 1.
London, Academic Press, pp. 31–52.
Boletzky, S. v. (2003). Biology of early life stages in cephalopod mollusks. Advances in Marine
Biology, 44: 143–200.
Boycott, B. B. (1954). Learning in Octopus vulgaris and other cephalopods. Pubblicazioni Della
Stazione Zoologica di Napoli, 25: 67–93.
Boyle, P. R. (1983). Cephalopod life cycles. Species accounts, vol. 1. London, Academic Press.
Boyle, P. R. (1986). Responses to water-borne chemicals by the octopus Eledone cirrhosa
(Lamarck, 1798). Journal of Experimental Marine Biology and Ecology, 104: 23–30.
Boyle, P. R. (1987). Cephalopod life cycles. Comparative reviews, vol. 2. London, Academic
Press.
Boyle, P. R. and Boletzky, S. v. (1996). Cephalopod populations: definition and dynamics. Philo-
sophical Transactions of the Royal Society of London. Series B, 351: 985–1002.
Boyle, P. R. and Rodhouse, P. (2005). Cephalopods: ecology and fisheries. Oxford, Blackwell.
Brierley, A. S., Rodhouse, P. G., Thorpe, J. P. and Clarke, M. R. (1993). Genetic evidence of
population heterogeneity and cryptic speciation in the ommastrephid squid Martialia hyadesi
from the Patagonian Shelf and Antarctic Polar Frontal Zone. Marine Biology, 116: 593–602.
Brown, M. F. and Cook, R. G. (2006). Animal spatial cognition: comparative, neural, and com-
putational approaches [online]. Available from www.pigeon.psy.tufts.edu/asc/
Budelmann, B. U. (1980). Equilibrium and orientation in cephalopods. Oceanus, 23: 34–43.
Budelmann, B. U. (1994). Cephalopod sense organs, nerves and the brain: adaptations for high
performance and life style. Marine and Freshwater Behaviour and Physiology, 25: 13–33.
Budelmann, B. U. (1995). The cephalopod nervous system: what evolution has made of the
molluscan design. In The nervous systems of invertebrates: an evolutionary and comparative
approach. Basel, Birkhäuser Verlag, pp. 115–138.
Budelmann, B. U. (1996). Active marine predators: the sensory world of cephalopods. Marine
and Freshwater Behaviour and Physiology, 27: 59–75.
 Navigation in cephalopods 169

Budelmann, B. U. and Bleckmann, H. (1988). A lateral line analogue in cephalopods: water waves
generate microphonic potentials in the epidermal head lines of Sepia and Lolliguncula. Journal
of Comparative Physiology A, 164: 1–5.
Budelmann, B. U., Schipp, R. and Boletzky, S. v. (1997). Cephalopoda. In Microscopic anatomy
of invertebrates. New York, Wiley-Liss, pp. 119–414.
Buresch, K. C., Gerlach, G. and Hanlon, R. T. (2006). Multiple genetic stocks of longfin squid
Loligo pealeii in the NW Atlantic: stocks segregate inshore in summer, but aggregate offshore
in winter. Marine Ecology Progress Series, 310: 263–270.
Campan, R. (1997). Tactic components in orientation. In Orientation and communication in
arthropods. Basel, Birkhäuser Verlag, pp. 1–40.
Carman, H. M. and Mactutus, C. F. (2001). Proximal versus distal cue utilization in spa-
tial navigation: the role of visual acuity? Neurobiology of Learning and Memory, 78: 332–
346.
Cartron, L., Darmaillacq, A. S., Jozet-Alves, C., Shashar, N. and Dickel, L. (2012). Cuttlefish rely
on both polarized light and landmarks for orientation. Animal Cognition, 15: 591–596.
Chen, C. S., Chiu T. S. and Haung, W. B. (2007). The spatial and temporal distribution patterns
of the Argentine short-finned squid, Illex argentinus, abundances in the Southwest Atlantic and
the effects of environmental influences. Zoological Studies, 46: 111–122.
Chen, X. J., Zhao, X. H. and Chen, Y. (2007). Influence of El Niño/La Niña on the western winter–
spring cohort of neon flying squid (Ommastrephes bartramii) in the northwestern Pacific Ocean.
ICES Journal of Marine Science, 64: 1152–1160.
Cheng, K. and Spetch, M. L. (1998). Mechanisms of landmark use in mammals and birds. In
Spatial representation in animals. Oxford, Oxford University Press, pp. 1–17.
Choi, K., Lee, C. I., Hwang, K., et al. (2008). Distribution and migration of Japanese common
squid, Todarodes pacificus, in the southwestern part of the East (Japan) Sea. Fisheries Research,
91: 281–290.
Collett, T. S., Dillmann, E., Giger, A. and Wehner, R. (1992). Visual landmarks and route-following
in desert ants. Journal of Comparative Physiology A, 170: 435–442.
Cook, S. B. (1971). A study of homing behavior in the limpet Siphonaria alternata. The Biological
Bulletin, 141: 449–457.
Cote, J., Clobert, J., Brodin, T., Fogarty, S. and Sih, A. (2010). Personality-dependent dispersal:
characterization, ontogeny and consequences for spatially structured populations. Philosophical
Transactions of the Royal Society of London. Series B, 365: 4065–4076.
Cronin, T. W., Shashar, N., Caldwell, R. L., et al. (2003). Polarization vision and its role in
biological signaling. Integrative and Comparative Biology, 43: 549–558.
Cummins, S. F., Boal, J. G., Buresch, K. C., et al. (2011). Extreme aggression in male squid
induced by a b-MSP-like pheromone. Current Biology, 21: 322–327.
Darmaillacq, A -S., Chichery, R. and Dickel, L. (2006). Food imprinting, new evidence from the
cuttlefish Sepia officinalis. Biology Letters, 2: 345–347.
Darmaillacq, A -S., Lesimple, C. and Dickel, L. (2008). Embryonic visual learning in the cuttlefish
Sepia officinalis. Animal Behaviour, 76: 131–134.
Davies, M. S. and Blackwell, J. (2007). Energy saving through trail following in a marine snail.
Proceedings of the Royal Society. B: Biological Sciences, 274: 1233–1236.
Dickel, L., Boal, J. G. and Budelmann, B. U. (2000). The effect of early experience on learning
and memory in cuttlefish. Developmental Psychobiology, 36: 101–110.
Dilly, P. N., Stephens, P. R. and Young, J. Z. (1975). Receptors in the statocyst of squids. Journal
of Physiology, 249: 59–61.
170 Christelle Jozet-Alves, Anne-Sophie Darmaillacq and Jean G. Boal

Dolins, F. L. and Mitchell, R. W. (2010). Spatial cognition, spatial perception, mapping the self
and space. Cambridge, Cambridge University Press.
Dunn, G. A. (1990). Conceptual problems with kinesis and taxis. In Biology of the Chemotactic
Response. Cambridge, Cambridge University Press, pp. 1–13.
Etienne, A. S., Maurer, R. and Séguinot, V. (1996). Path integration in mammals and its interaction
with visual landmarks. Journal of Experimental Biology, 199: 201–209.
Faraj, A. and Bez, N. (2007). Spatial considerations for the Dakhla stock of Octopus vulgaris:
indicators, patterns, and fisheries interactions. ICES Journal of Marine Science, 64: 1820–
1828.
Forsythe, J. W. and Hanlon, R. T. (1997). Foraging and associated behavior by Octopus cyanea
Gray, 1849 on a coral atoll, French Polynesia. Journal of Experimental Marine Biology and
Ecology, 209: 15–31.
Forsythe, J. W. and Toll, R. B. (1991). Clarification of the Western Atlantic Ocean pigmy octopus
complex: the identity and life history of Octopus joubini (Cephalopoda: Octopodinae). Bulletin
of Marine Science, 49: 88–97.
Franz, M. O. and Mallot, H. A. (2000). Biomimetic robot navigation. Robotics and Autonomous
Systems, 30: 133–153.
Gabe, S. H. (1975). Reproduction in the giant octopus of the North Pacific, Octopus dofleini
martini. Veliger, 18: 146–150.
Gallistel, C. R. (1993). The organization of learning. Cambridge, MIT Press.
Gaulin, S. J. C. and FitzGerald, R. W. (1986). Sex differences in spatial ability: an evolutionary
hypothesis and test. The American Naturalist, 127: 77–88.
Gaulin, S. J. C. and FitzGerald, R. W. (1989). Sexual selection for spatial-learning ability. Animal
Behaviour, 37: 322–331.
Gibson, B. M. and Shettleworth, S. J. (2005). Place versus response learning revisited: tests of
blocking on the radial maze. Behavioral Neuroscience, 119: 567–586.
Gilly, W. F., Markaida, U., Baxter, C. H., et al. (2006). Vertical and horizontal migrations by the
jumbo squid Dosidicus gigas revealed by electronic tagging. Marine Ecology Progress Series,
324: 1–17.
Gleadall, I. G. and Shashar, N. (2004). The octopus’s garden: the visual world of cephalopods. In
Complex worlds from simpler nervous systems. Cambridge, MIT Press, pp. 269–307.
Golledge, R. G. (1999). Wayfinding behavior: cognitive mapping and other spatial processes.
Baltimore, Johns Hopkins University Press.
Graziadei, P. (1964). Receptors in the sucker of the cuttlefish. Nature, 203: 384–386.
Graziadei, P. P. C. and Gagne, H. T. (1976). Sensory innervation in the rim of the octopus sucker.
Journal of Morphology, 150: 639–679.
Groeger, G., Cotton, P. A. and Williamson, R. (2005). Ontogenetic changes in the visual acuity
of Sepia officinalis measured using the optomotor response. Canadian Journal of Zoology, 83:
274–279.
Guibé, M., Boal, J. G. and Dickel, L. (2010). Early exposure to odors changes late visual prefer-
ences in cuttlefish. Developmental Psychobiology, 52: 833–837.
Hall, K. C. and Hanlon, R. T. (2002). Principal features of the mating system of a large spawning
aggregation of the giant Australian cuttlefish Sepia apama (Mollusca: Cephalopoda). Marine
Biology, 140: 533–545.
Hanlon, R. T. and Budelmann, B. U. (1987). Why cephalopods are probably not “deaf.” The
American Naturalist, 129: 312–317.
 Navigation in cephalopods 171

Hanlon, R. T. and Messenger, J. B. (1988). Adaptive coloration in young cuttlefish (Sepia offic-
inalis L.): the morphology and development of body patterns and their relation to behaviour.
Philosophical Transactions of the Royal Society of London. Series B, 320: 437–487.
Hanlon, R. T. and Messenger, J. B. (1996). Cephalopod behaviour. Cambridge, Cambridge Uni-
versity Press.
Hartwick, E. B., Breen, P. A. and Tulloch, L. (1978). A removal experiment with Octopus dofleini
(Wülker). Journal of the Fisheries Research Board of Canada, 35: 1492–1495.
Hartwick, E. B., Ambrose, R. F. and Robinson, S. M. C. (1984). Den utilization and the movements
of tagged Octopus dofleini. Marine Behaviour and Physiology, 11: 95–110.
Healy, S. D. (1998). Spatial representation in animals. Oxford, Oxford University Press.
Heinze, S. and Homberg, U. (2007). Maplike representation of celestial e-vector orientations in
the brain of an insect. Science, 315: 995–997.
Hiroyuki, T. (2001). Tracking the neon flying squid by the biotelemetry system in the central
North Pacific Ocean. Aquabiology, 23: 533–539.
Horváth, G. and Varjú, D. (2004). Polarized light in animal vision: polarization patterns in nature.
Berlin, Springer.
Hvorecny, L. M., Grudowski, J. L., Blakeslee, C. J., et al. (2007). Octopuses (Octopus bimacu-
loides) and cuttlefishes (Sepia pharaonis, S. officinalis) can conditionally discriminate. Animal
Cognition, 10: 449–459.
Ito, K. (2007). Studies on migration and causes of stock-size fluctuations in the northern Japanese
population of spear squid, Loligo bleekeri. Bulletin of Aomori Prefectural Fisheries Research
Center, 5: 11–75.
Jander, R., Daumer, K. and Waterman, T. H. (1963). Polarized light orientation by two Hawaiian
decapod cephalopods. Zeitschrift für vergleichende Physiologie, 46: 383–394.
Jones, C. M., Braithwaite, V. A. and Healy, S. D. (2003). The evolution of sex differences in spatial
ability. Behavioral Neuroscience, 117: 403–411.
Jozet-Alves, C., Modéran, J. and Dickel, L. (2008). Sex differences in spatial cognition in an
invertebrate: the cuttlefish. Proceedings of the Royal Society. B: Biological Sciences, 275:
2049–2054.
Kaib, M., Bruinsma, O. and Leuthold, R. H. (1982). Trail-following in termites: evidence for a
multicomponent system. Journal of Chemical Ecology, 8: 1193–1205.
Kaneko, N., Oshima, Y. and Ikeda, Y. (2006). Egg brooding behavior and embryonic development
of Octopus laqueus (Cephalopoda: Octopodidae). Molluscan Research, 26: 113–117.
Karson, M. A. (2003). Simultaneous discrimination learning and its neural correlates in the cuttle-
fish Sepia officinalis (Cephalopoda: Mollusca). Doctoral dissertation. East Lansing, Michigan
State University.
Karson, M. A., Boal, J. G. and Hanlon, R. T. (2003). Experimental evidence for spatial learning
in cuttlefish (Sepia officinalis). Journal of Comparative Psychology, 117: 149–155.
Kawabata, A., Yatsu, A., Ueno, Y., Suyama, S. and Kurita, Y. (2006). Spatial distribution of the
Japanese common squid, Todarodes pacificus, during its northward migration in the western
North Pacific Ocean. Fisheries Oceanography, 15: 113–124.
Kayes, R. J. (1974). The daily activity pattern of Octopus vulgaris in a natural habitat. Marine
Behaviour and Physiology, 2: 337–343.
Kidokoro, H., Goto, T., Nagasawa, T., et al. (2010). Impact of a climate regime shift on the
migration of Japanese common squid (Todarodes pacificus) in the Sea of Japan. ICES Journal
of Marine Science, 67: 1314–1322.
172 Christelle Jozet-Alves, Anne-Sophie Darmaillacq and Jean G. Boal

Kier, W. M., Messenger, J. B. and Miyan, J. A. (1985). Mechanoreceptors in the fins of the
cuttlefish, Sepia officinalis. Journal of Experimental Biology, 119: 369–373.
King, A. J., Adamo, S. A. and Hanlon, R. T. (2003). Squid egg mops provide sensory cues for
increased agonistic behaviour between male squid. Animal Behaviour, 66: 49–58.
Kiyofuji, H. and Saitoh, S. -I. (2004). Use of nighttime visible images to detect Japanese common
squid Todarodes pacificus fishing areas and potential migration routes in the Sea of Japan.
Marine Ecology Progress Series, 276: 173–186.
Komak, S., Boal, J. G., Dickel, L. and Budelmann, B. U. (2005). Behavioural responses of juvenile
cuttlefish (Sepia officinalis) to local water movements. Marine and Freshwater Behaviour and
Physiology, 38: 117–125.
Kraft, P., Evangelista, C., Dacke, M., Labhart, T. and Srinivasan, M. V. (2011). Honeybee nav-
igation: following routes using polarized-light cues. Philosophical Transactions of the Royal
Society B, 366: 703–708.
Layne, J. E., Barnes, W. J. P. and Duncan, L. M. J. (2003). Mechanisms of homing in the fiddler
crab Uca rapax. 2. Information sources and frame of reference for a path integration system.
Journal of Experimental Biology, 206: 4425–4442.
Lee, P. G. (1992). Chemotaxis by Octopus maya Voss et Solis in a Y-maze. Journal of Experimental
Marine Biology and Ecology, 153: 53–67.
Lerner, A., Sabbah, S., Erlick, C. and Shashar, N. (2011). Navigation by light polarization in clear
and turbid waters. Philosophical Transactions of the Royal Society of London. Series B, 366:
671–679.
Lohmann, K. J., Lohmann, C. M. F., Ehrhart, L.M., Bagley, D.A. and Swing, T. (2004). Animal
behaviour: geomagnetic map used in sea-turtle navigation. Nature, 428: 909–910.
Mackintosh, N. J. and Mackintosh, J. (1964). Performance of Octopus over a series of reversals of
a simultaneous discrimination. Animal Behaviour, 12: 321–324.
Marshall, N. J. and Cronin, T. W. (2011). Polarisation vision. Current Biology, 21: R101–105.
Marshall, N. J. and Messenger, J. B. (1996). Colour-blind camouflage. Nature, 382: 408–409.
Mather, J. A. (1988). Daytime activity of juvenile Octopus vulgaris in Bermuda. Malacologia, 29:
69–76.
Mather, J. A. (1991a). Navigation by spatial memory and use of visual landmarks in octopuses.
Journal of Comparative Physiology A, 168: 491–497.
Mather, J. A. (1991b). Foraging, feeding and prey remains in middens of juvenile Octopus vulgaris
(Mollusca: Cephalopoda). Journal of Zoology, 224: 27–39.
Mather, J. A. (1994). “Home” choice and modification by juvenile Octopus vulgaris (Mollusca:
Cephalopoda): specialized intelligence and tool use? Journal of Zoology, 233: 359–368.
Mather, J. A. (1995). Cognition in cephalopods. Advances in the Study of Behavior, 24: 317–353.
Mather, J. A. and Anderson, R. C. (1993). Personalities of octopuses (Octopus rubescens). Journal
of Comparative Psychology, 107: 336–340.
Mather, J. A. and O’Dor, R. K. (1991). Foraging strategies and predation risk shape the natural
history of juvenile Octopus vulgaris. Bulletin of Marine Science, 49: 256–269.
Mather, J. A., Resler, S. and Cosgrove, J. (1985). Activity and movement patterns of Octopus
dofleini. Marine Behavior and Physiology, 11: 301–314.
Mäthger, L. M., Barbosa, A., Miner, S. and Hanlon, R. T. (2006). Color blindness and contrast
perception in cuttlefish (Sepia officinalis) determined by a visual sensorimotor assay. Vision
Research, 46: 1746–1753.
Messenger, J. B. (1970). Optomotor responses and nystagmus in intact, blinded and statocystless
cuttlefish (Sepia officinalis L.). Journal of Experimental Biology, 53: 789–796.
 Navigation in cephalopods 173

Messenger, J. B. (1991). Photoreception and vision in Mollusks. In Evolution of the eye and visual
system. London, Macmillan, pp. 364–97.
Montgomery, J. C., Jeffs, A., Simpson, S. D., Meekan, M. and Tindle, C. (2006). Sound as an
orientation cue for the pelagic larvae of reef fishes and decapod crustaceans. Advances in
Marine Biology, 51: 143–196.
Mooney, T. A., Hanlon, R. T., Christensen-Dalsgaard, J., et al. (2010). Sound detection by the
longfin squid (Loligo pealeii) studied with auditory evoked potentials: sensitivity to low-
frequency particle motion and not pressure. Journal of Experimental Biology, 213: 3748–
3759.
Moore, P. A., Scholz, N. and Atema, J. (1991). Chemical orientation of lobsters, Homarus ameri-
canus, in turbulent odor plumes. Journal of Chemical Ecology, 17: 1293–1307.
Moreno, A., Dos Santos, A., Piatkowski, U., Santos, M. P. and Cabral, H. (2009). Distribution of
cephalopod paralarvae in relation to the regional oceanography of the western Iberia. Journal
of Plankton Research, 31: 73–91.
Moriyama, T. and Gunji, Y. -P. (1997). Autonomous learning in maze solution by Octopus.
Ethology, 103: 499–513.
Müller, M. and Wehner, R. (1988). Path integration in desert ants, Cataglyphis fortis. Proceedings
of the National Academy of Sciences of the United States of America, 85: 5287–5290.
Murata, M. and Nakamura, Y. (1998). Seasonal migration and diel vertical migration of the neon
flying squid, Ommastrephes bartramii in the North Pacific. In Contributed papers to interna-
tional symposium on large pelagic squids, July 18–19, 1996, for JAMARC’s 25th anniversary
of its foundation. Tokyo, JAMARC, pp. 13–30.
Nixon, M. and Young, J. Z. (2003). The brains and lives of cephalopods. Oxford, Oxford University
Press.
O’Dor, R. K., Hoar, J. A., Webber, D. M., et al. (1994). Squid (Loligo forbesi) performance and
metabolic rates in nature. Marine and Freshwater Behaviour and Physiology, 25: 163–177.
O’Dor, R. K., Adamo, S., Aitken, J.P., et al. (2002). Currents as environmental constraints on the
behavior, energetics and distribution of squid and cuttlefish. Bulletin of Marine Science, 71:
601–617.
O’Keefe, J. and Nadel, L. (1978). The hippocampus as a cognitive map. Oxford, Oxford University
Press.
Olson, R. J. and Young, J. W. (Eds.) (2007). The role of squid in open ocean ecosystems. Report
of the GLOBEC-CLIOTOP/PFRP workshop, 16–17 November 2006. GLOBEC Report 24: VI.
Honolulu, Hawaii, GLOBEC.
Olyott, L. J. H., Sauer, W. H. H. and Booth, A. J. (2007). Spatial patterns in the biology of the
chokka squid, Loligo reynaudii on the Agulhas Bank, South Africa. Reviews in Fish Biology
and Fisheries, 17: 159–172.
Onitsuka, G., Hirose, N., Miyahara, K., et al. (2010). Numerical simulation of the migration
and distribution of diamond squid (Thysanoteuthis rhombus) in the southwest Sea of Japan.
Fisheries Oceanography, 19: 63–75.
Oosthuizen, A., Jiwaji, M. and Shaw, P. (2004). Genetic analysis of the Octopus vulgaris population
on the coast of South Africa. South African Journal of Science, 100: 603–607.
Payne, N. L., Semmens, J. M. and Gillanders, B. M. (2011). Elemental uptake via immersion:
a mass-marking technique for the early life-history stages of cephalopods. Marine Ecology
Progress Series, 436: 169–176.
Pecl, G. T., Doubleday, Z. A., Danyushevsky, L., Gilbert, S. and Moltschaniwskyj, N. A. (2010).
Transgenerational marking of cephalopods with an enriched barium isotope: a promising tool
174 Christelle Jozet-Alves, Anne-Sophie Darmaillacq and Jean G. Boal

for empirically estimating post-hatching movement and population connectivity. ICES Journal
of Marine Science, 67: 1372–1380.
Pérez-Losada, M., Guerra, A., Carvalho, G. R., Sanjuan, A. and Shaw, P. W. (2002). Extensive
population subdivision of the cuttlefish Sepia officinalis (Mollusca: Cephalopoda) around the
Iberian Peninsula indicated by microsatellite DNA variation. Heredity, 89: 417–424.
Pérez-Losada, M., Nolte, M. J., Crandall, K. A. and Shaw, P. W. (2007). Testing hypotheses
of population structuring in the Northeast Atlantic Ocean and Mediterranean Sea using the
common cuttlefish Sepia officinalis. Molecular Ecology, 16: 2667–2679.
Pierce, G. J., Valavanis, V. D., Guerra, A., et al. (2008). A review of cephalopod-environment
interactions in European seas. Hydrobiologia, 612: 49–70.
Reale, D., Reader, S. M., Sol, D., McDougall, P. T. and Dingemanse, N. J. (2007). Integrating
animal temperament within ecology and evolution. Biological Reviews, 82: 291–318.
Reichow, D. and Smith, M. J. (2001). Microsatellites reveal high levels of gene flow among
populations of the California squid Loligo opalescens. Molecular Ecology, 10: 1101–1109.
Restle, F. (1957). Discrimination of cues in mazes: a resolution of the “place-vs.-response”
question. Psychological Review, 64: 217–228.
Rosa, R. and Seibel, B. A. (2010). Metabolic physiology of the Humboldt squid, Dosidicus
gigas: implications for vertical migration in a pronounced oxygen minimum zone. Progress in
Oceanography, 86: 72–80.
Saidel, W. M., Lettvin, J. Y. and McNichol, E. F. (1983). Processing of polarized light by squid
photoreceptors. Nature, 304: 534–536.
Sakaguchi, K. (2010). Migration of tagged Japanese common squid, Todarodes pacificus, in waters
around Hokkaido [Japan]. Scientific Reports of Hokkaido Fisheries Experimental Station, 77:
45–72.
Sanders F. K. and Young, J. Z. (1940). Learning and other functions of the higher nervous centres
of Sepia. Journal of Neurophysiology, 3: 501–526.
Sauer, W. H., Roberts, M. J., Lipinski, M. R., et al. (1997). Choreography of the squid’s “nuptial
dance.” The Biological Bulletin, 192: 203–207.
Save, E., Poucet, B. and Thinus-Blanc, C. (1998). Landmark use and the cognitive map in the rat.
In Spatial representation in animals. Oxford, Oxford University Press, pp. 119–132.
Schiller, P. H. (1949). Delayed detour response in the octopus. Journal of Comparative and
Physiological Psychology, 42: 220–225.
Schmajuk, N. A. and Thieme, A. D. (1992). Purposive behavior and cognitive mapping: a neural
network model. Biological Cybernetics, 67: 165–174.
Scholz, A. T., Horrall, R. M., Cooper, J. C. and Hasler, A. D. (1976). Imprinting to chemical cues:
the basis for home stream selection in salmon. Science, 192: 1247–1249.
Semmens, J. M., Pecl, G. T., Gillanders, B. M., et al. (2007). Approaches to resolving cephalopod
movement and migration patterns. Reviews in Fish Biology and Fisheries, 17: 401–423.
Shashar, N., Rutledge, P. S. and Cronin, T. W. (1996). Polarization vision in cuttlefish in a concealed
communication channel? Journal of Experimental Biology, 199: 2077–2084.
Shashar, N., Hagan, R., Boal, J. G. and Hanlon, R. T. (2000). Cuttlefish use polarization sensitivity
in predation on silvery fish. Vision Research, 40: 71–75.
Shaw, P. W., Pierce, G. J. and Boyle, P. R. (1999). Subtle population structuring within a highly
vagile marine invertebrate, the veined squid Loligo forbesi, demonstrated with microsatellite
DNA markers. Molecular Ecology, 8: 407–417.
Shea, E. K. and Vecchione, M. (2010). Ontogenic changes in diel vertical migration patterns com-
pared with known allometric changes in three mesopelagic squid species suggest an expanded
definition of a paralarva. ICES Journal of Marine Science, 67: 1436–1443.
 Navigation in cephalopods 175

Shettleworth, S. J. (2010). Cognition, evolution and behavior, 2nd edn. Oxford, Oxford University
Press.
Sims, D. W., Genner, M. J., Southward, A. J. and Hawkins, S. J. (2001). Timing of squid migration
reflects North Atlantic climate variability. Proceedings of the Royal Society. B: Biological
Sciences, 268: 2607–2611.
Sinn, D. L. and Moltschaniwskyj, N. A. (2005). Personality traits in dumpling squid (Euprymna
tasmanica): context-specific traits and their correlation with biological characteristics. Journal
of Comparative Physiology, 119: 99–110.
Sinn, D. L., Moltschaniwskyj, N. A., Wapstra, E. and Dall, S. R. X. (2010). Are behavioral syn-
dromes invariant? Spatiotemporal variation in shy/bold behavior in squid. Behavioral Ecology
and Sociobiology, 64: 693–702.
Stamps, J. A. (2007). Growth-mortality tradeoffs and “personality traits” in animals. Ecology
Letters, 10: 355–363.
Stark, K. E. (2008). Ecology of the arrow squid (Nototodarus gouldi) in Southeastern Australian
waters: a multi-scale investigation of spatial and temporal variability. PhD thesis, Tasmania,
University of Tasmania.
Sutherland, N. S., Mackintosh, N. J. and Mackintosh, J. (1963). Simultaneous discrimination
training of Octopus and transfer of discrimination along a continuum. Journal of Comparative
and Physiological Psychology, 56: 150–156.
Tansley, K. (1965). Vision in vertebrates. London, Chapman and Hall.
Tinbergen, N. (1972). The animal in its world. Cambridge, MA, Harvard University Press.
Tolimieri, N., Haine, O., Jeffs, A., McCauley, R. and Montgomery, J. (2004). Directional orienta-
tion of pomacentrid larvae to ambient reef sound. Coral Reefs, 23: 184–191.
Van Heukelem, W. F. (1966). Some aspects of the ecology and ethology of Octopus cyanea Gray.
MS thesis. Honolulu, University of Hawaii.
Villanueva, R. (1992). Deep-sea cephalopods of the north-western Mediterranean: indications of
up-slope ontogenetic migration in two bathybenthic species. Journal of Zoology, 227: 267–276.
Villanueva, R. and Norman, M. D. (2008). Biology of the planktonic stages of benthic octopuses.
Oceanography and Marine Biology, 46: 105–202.
Walderon, M. D., Nolt, K. J., Haas, R. E., et al. (2011). Distance chemoreception and the detection
of conspecifics in Octopus bimaculoides. Journal of Molluscan Studies, 77: 309–311.
Walker, J. J., Longo, N. and Bitterman, M. E. (1970). The octopus in the laboratory. Handling,
maintenance, training. Behavior Research Methods and Instrumentation, 2: 15–18.
Wang, J., Pierce, G. J., Boyle, P. R., et al. (2003). Spatial and temporal patterns of cuttlefish (Sepia
officinalis) abundance and environmental influences – a case study using trawl fishery data in
French Atlantic coastal, English Channel, and adjacent waters. ICES Journal of Marine Science,
60: 1149–1158.
Warburton, K. (1990). The use of local landmarks by foraging goldfish. Animal Behaviour, 40:
500–505.
Watanabe, H., Kubodera, T., Moku, M. and Kawaguchi, K. (2006). Diel vertical migration of
squid in the warm core ring and cold water masses in the transition region of the western North
Pacific. Marine Ecology Progress Series, 315: 187–197.
Waterman, T. H. (2006). Reviving a neglected celestial underwater polarization compass for
aquatic animals. Biological Reviews, 81: 111–115.
Wehner, R. (2003). Desert ant navigation: how miniature brains solve complex tasks. Journal of
Comparative Physiology A, 189: 579–588.
Wehner, R., Lehrer, M., Harvey, W. R. (1996). Navigation. Journal of Experimental Biology, 199:
1–261.
176 Christelle Jozet-Alves, Anne-Sophie Darmaillacq and Jean G. Boal

Wells, M. J. (1964). Detour experiments with octopuses. Journal of Experimental Biology, 41:
621–642.
Wells, M. J. (1978). Octopus: physiology and behaviour of an advanced invertebrate. London,
Chapman and Hall.
Willows, A. O. D. (1999). Shoreward orientation involving geomagnetic cues in the nudibranch
mollusc Tritonia diomedea. Marine and Freshwater Behaviour and Physiology, 32: 181–192.
Winklhofer, M. (2009). The physics of geomagnetic-field transduction in animals. IEEE Transac-
tions on Magnetics, 45: 5259–5265.
Yarnall, J. L. (1969). Aspects of the behaviour of Octopus cyanea Gray. Animal Behaviour, 17:
747–754.
Yatsu, A., Yamanaka, K.and Yamashiro, C. (1999). Tracking experiments of the jumbo flying squid,
Dosidicus gigas, with an ultrasonic telemetry system in the Eastern Pacific Ocean. Bulletin of
the National Research Institute of Far Seas Fisheries, 36: 55–60.
Young, J. Z. (1960). The statocysts of Octopus vulgaris. Proceedings of the Royal Society. B:
Biological Sciences, 152: 3–29.
Young, M. A., Kvitek, R. G., Iampietro, P. J., et al. (2011). Seafloor mapping and landscape
ecology analyses used to monitor variations in spawning site preference and benthic egg mop
abundance for the California market squid (Doryteuthis opalescens). Journal of Experimental
Marine Biology and Ecology, 407: 226–233.
Young, R. E. and Harman, R. F. (1988). “Larva,” “paralarva,” and “subadult” in cephalopod
terminology. Malacologia, 29: 201–207.
Zeidberg, L. D. and Robison, B. H. (2007). Invasive range expansion by the Humboldt squid,
Dosidicus gigas, in the eastern North Pacific. Proceedings of the National Academy of Sciences
of the United States of America, 104: 12948–12950.
Zumholz, K., Klügel, A., Hansteen, T. and Piatkowski, U. (2007). Statolith microchemistry traces
the environmental history of the boreoatlantic armhook squid Gonatus fabricii. Marine Ecology
Progress Series, 333: 195–204.
8 Camouflage in benthic cephalopods:
what does it teach us?
Noam Josef and Nadav Shashar

8.1 Brief historical review

Neither the scientist nor the curious diver can remain impassive after witnessing an
octopus camouflaging itself in a reef. Intriguing, beautiful and efficient camouflagers,
cephalopods can produce thousands of different patterns (Packard, 1972; Packard &
Hochberg, 1977; Hanlon, 1982; Moynihan, 1985; Hanlon & Messenger, 1988, 1996;
reviewed in Borrelli, Fiorito & Gherardi, 2006).
Cephalopods are an ancient group dating from the upper Cambrian. Although they
are mollusks, coleoid cephalopods present an extraordinary resemblance to teleosts in
morphology, physiology, ecology and behavior (Packard, 1972; Hanlon & Messenger,
1996). A variety of physiological structures, crafted by evolution, enables cephalopods
to change their body patterning continuously and rapidly, conferring on them the cryp-
tic, communicative and aesthetically appealing characteristics typical of this group.
The ability of octopuses, cuttlefish and squid to quickly and dramatically change their
appearance has thrilled scientists from antiquity (Aristotle 350 BCE; Darwin, 1870;
Parker, 1948). Like many past researchers who witnessed their amazing displays, move-
ments and behaviors, we, too, have been captivated. Throughout this chapter, we use the
word “cephalopod” to refer to the benthic coleoid cephalopod taxa. However, the reader
should bear in mind that many pelagic coleoids are also capable of remarkable changes
in body patterning that may be similarly important in camouflage (e.g. Bush, Robison
& Caldwell, 2009; Zylinski & Johnsen, 2011).

8.2 Definitions and nomenclature

Camouflage is the common term describing the capacity of an animal to disguise itself
and avoid detection by visual seekers. The term crypsis (Endler, 1981) includes behav-
ioral as well as morphological strategies to prevent detection. Although easy to under-
stand, the concepts of camouflage and concealment are difficult to define and quantify.
As many as 15 different mechanisms and definitions can be found in the literature. Some
mechanisms operate synergistically and, as such, it is likely that some phenomena would
Cephalopod Cognition, eds A.-S. Darmaillacq, L. Dickel and J. Mather. Published by Cambridge University
Press. 
C Cambridge 2014.
178 Noam Josef and Nadav Shashar

Table 8.1 Commonly used camouflage terminology (modified after Stevens & Merilaita, 2009)

Term Definition

Camouflage All concealment strategies, including prevention of detection

and recognition
Crypsis Initially preventing detection
(a) Background matching Appearance generally matches the color, lightness and pattern
of one (specialist) or several (compromise) background types
(b) Self-shadow concealment Elimination or concealment of shadows that are created by
directional light, occasionally achieved by countershading
(c) Obliterate shading Countershading leads to the obliteration of three-dimensional
form
(d) Disruptive coloration A set of markings that creates the appearance of false edges
and boundaries and hinders the detection or recognition of
either part or all of an object’s true outline or shape
(e) Flicker-fusion camouflage Markings such as stripes blur during motion to match the
color/lightness of the general background, preventing
detection of the animal when in motion
(f) Distractive markings Direct the “attention” or gaze of the receiver from traits that
would give away the animal (such as the outline)
Motion camouflage Movement in a fashion that decreases the probability of
movement detection
Motion dazzle Markings complicate receiver estimation of speed and
trajectory
Masquerade Resemblance to an uninteresting object, such as a leaf or a stick
Countershading Using dark dorsal coloration or light ventral colors or
(Thayer’s Law) photophores for concealment by matching overall
brightness. When viewed from above, the animal has the
backdrop of dark, deep waters; in the opposite perspective,
the animal is viewed against brighter downwelling light,
which ventral photoluminescent organs can match
Silvering Common in aquatic environments and where an animal’s body
is highly reflective making it difficult to detect when light
incidence is non-directional (such as due to strong scattering
by waterborne particles)
Transparency Animal’s body (or a part of it) is transparent, reducing
detection; like a hood

be best described by a combination of several terms. To clarify this somewhat confusing

use of terms and guide the reader, Table 8.1 contains the terms and definitions used in
this chapter.
With predator avoidance as an evolutionary driving force, camouflage has generated
some classic examples of selection (e.g. the peppered moth; Bacot, 1907). Therefore,
since predators tend to select conspicuous individuals during their search for prey, good
camouflage can be defined as the phenotypic appearance of the survivor. Nonetheless,
one should note that animals also use warning colors that are deliberately conspicuous
to their predators (e.g. Cronin, 2005).
 Camouflage in benthic cephalopods 179

Existent in amazing variety, camouflage can be found in distant phylogenetic

groups, from flatworms and insects to reptiles and mammals. But its manifestation in
cephalopods is exceptional not only in the speed with which they can match their visual
backgrounds, but also in the diversity of colors and patterns they can express.

8.3 Dynamic versus static camouflage

While most animals exhibit fixed or slowly changing, “static camouflage” patterns
(Stevens & Merilaita, 2011), the “dynamic camouflage” of cephalopods describes their
ability to rapidly change their patterns, color and even texture to match the background
on which they are found.
The avoidance of visual detection by predators or by prey has driven the creation
of outstanding adaptations. Examples of amazing camouflage can be found through-
out the marine environment, including markings to match color, pattern and brightness
of the background (e.g. stonefish, Synanceia sp.) and Disruptive coloration that disturbs
the visual appearance of the outline of the body (e.g. some marine isopods, Merilaita,
1998) as well as the capacity of some fish to resemble sea-grass leaves and the sand-like
skin patterns possessed by others (Figure 8.1a–d). Camouflage achieves even greater
efficiency, however, in the dynamic coloration found in some cephalopods (Hanlon &
Messenger, 1988) and chameleons (Stuart-Fox & Moussalli, 2008).
Either type of camouflage, dynamic or static, entails a particular set of strategies and
behaviors the camouflaged individual uses to decrease detection, including remaining
motionless (Eilam, 2005), selecting a suitable body position and changing its orientation
according the ambient light conditions (Barbosa, Allen, Mäthger & Hanlon, 2012). What
distinguishes dynamic from static camouflage, however, is the ability of animals with
the former to match and therefore occupy a wide range of distinct visual environments.
For example, imagine a flatfish (e.g. the Moses sole, Pardachirus mormoratus) with its
slow-changing Mottle pattern lying on the sandy sea bed, where it is well camouflaged
(Figure 8.1d). Viewed against a stony reef or a seagrass patch, however, the flatfish would
be vulnerable to visual detection. In contrast, picture the same scenario with the common
octopus (Octopus vulgaris). A change in its visual environment does not render the octo-
pus conspicuous, as it can remain cryptic against different backgrounds. The flexibility
and phenotypic plasticity of dynamic camouflage enable animals that possess it to forage
in a variety of habitats without increasing its risk of predation. Likewise, predators that
use static camouflage behavior when hunting are restricted to environments in which
they will both remain cryptic and also have a high probability of encountering prey.
Thus, the dynamically camouflaging animal can find the most advantageous location for
predation and adjust its body pattern according to its surroundings.
The function of dynamic coloration is not restricted to camouflage: inter- and intraspe-
cific communication via body pattern changes is common in cephalopods (Hanlon &
Messenger, 1996; Hall & Hanlon, 2002; Langridge, Broom & Osorio, 2007). Since
camouflaging and communicating have opposite functions but are carried out by the
same mechanism, a camouflaged individual should either signal within a certain range
180 Noam Josef and Nadav Shashar

Figure 8.1 Camouflage examples in marine animals. (a) A stone fish, Synanceia verrucosa
choosing a place to settle which matches its body pattern. (b) An isopod: displaying Disruptive
coloration. (c) A Posidonia pipefish, Syngnathus sp. using object resemblance in hiding between
seagrass leafs. (d) A Moses sole, Pardachirus mormoratus showing a static Mottle pattern; note
partial missmatch with such a static camouflage. See plate section for color version.

of color variations or avoid the simultaneous expression of communication behavior to

its conspecifics.
The main organs involved in the coleoid cephalopod’s camouflaging process are the
chromatophore, iridophore and leucophore cells and the complex nervous system that
controls them (Boycott, 1953; Hanlon & Messenger, 1996; reviewed in Messenger, 2001;
Chrachri & Williamson, 2004). Chromatophores are small, elastic saccules of pigment
that reflect mainly the long-wavelength section of visible spectrum (i.e. black, brown,
red, orange, etc.). During camouflaging behavior, they are expanded by the contraction
of up to 24 radially arranged muscle fibers pulling the pigment sac open. The relaxation
of these muscles causes the chromatophore wall to contract, thereby reducing the area
visually affected by the pigment sac (Florey & Kriebel, 1969; Messenger, 2001).
Below the chromatophores lie the iridophores, which form a distinct, bright layer.
Iridophore cells are themselves colorless, but they provide structural color (e.g. irides-
cence) via wave-length-specific reflectivity for short-wave coloration (e.g. blues and
purples) in cephalopods. Iridophores are also found in the wall of the ink sac, testis and
on the eye (Arnold, 1967).
 Camouflage in benthic cephalopods 181

Figure 8.2 Skin spikiness in (a) Octopus vulgaris, (b) Octopus cyanea and (c) Sepia prashadi.
Physical texture of the skin through the expression of papillae helps creating a Disruptive
appearance, breaking the general outline of the animal.

The flattened leucophore cells act as diffuse reflectors and are responsible for the
“white spots” seen in many cephalopods, produced by reflecting all wavelengths of
scattered light (Cloney & Brocco, 1983). Leucophores, iridophores and other structures
reflect the chromatic characteristics of the local light field, possibly allowing for partial
color matching by passive means (Hanlon & Messenger, 1996). This may explain how
cephalopods are able to match the colors around them without being able to perceive them
(Mäthger, Shashar & Hanlon, 2009). Most cephalopod species so far investigated possess
a single mid-wavelength visual pigment in their eyes, meaning they are effectively color-
blind (Messenger, 1977; Marshall & Messenger, 1996; reviewed in Hanlon & Messenger,
1996).
In addition to the ability to alter their body patterning, many cephalopods can change
the physical texture of their skin using papillae to give their skin appearances that
can range from completely smooth to very spiky (Hanlon & Messenger, 1988; Allen,
Mäthger, Barbosa & Hanlon, 2009). Combined with body posture, physical texture
can give the animal a unique appearance (often similar to the textural attributes of its
immediate surroundings), probably complicating shape-oriented searches by potential
predators (Figure 8.2a–c).
182 Noam Josef and Nadav Shashar

8.4 Neural physiology

Controlling a complex assembly of chromatic skin cells for single-unit functionality

requires a unique and precise neural control system. Electrophysiological, anatomical
and behavioral studies showed that the control system is hierarchically organized (Nixon
& Young, 2003). Hanlon and Messenger (1996) proposed a simplified outline of this
complex system:

Eye → Optic lobe → Lateral basal lobe → Chromatophore lobe → Skin.

Anyone who has observed an octopus camouflaging or a squid signaling is probably

familiar with the bilateral presentation of color on its skin (see Hanlon & Messenger,
1996 for examples). This is consistent with a bilateral organization of the system, in
which each side of the body is neurally controlled by the same side of the brain (and eye).
The nerve cells run, with or without limited synapses in the stellate ganglion, directly
to the chromatophore muscles (Sereni & Young, 1932). Such direct neural control has
several advantages:

1. Speed: entire body pattern can be altered in less than a second.

2. Continuity: as an animal moves, the environment is constantly changing around it.
This system has the potential to match body pattern, color and brightness to the
background at any given moment.
3. High resolution: the ability to present highly detailed expressions of patterns and
texture through the precise control of fine chromatic units.
4. Polyphenism: the ability to present a wide variety of different patterns (see ethograms
in Hanlon & Messenger, 1988, 1996; Hanlon, Forsythe & Joneschild, 1999).
5. Independent bilaterality: an individual may present a different pattern on each side of
its body (e.g. an agonistic display on one side and a camouflage pattern on the other).
6. Stability: independent neurological control means that other motor activities do not
interfere with the chromatic signals.

8.5 Body patterns

The speed, variability and complexity of cephalopod body patterning constitute an evo-
lutionary survival strategy that is nothing less than extraordinary. The term polyphenism
has been used to describe the use of multiple appearances by an individual (Mayr, 1963),
and it has been proposed that this adaptation may inhibit a predator’s ability to form
a search image (Curio, 1976). Most coleoid cephalopods can produce around 12 body
patterns (Borrelli, Fiorito & Gherardi, 2006) whose classification is at best difficult. Each
body pattern comprises a combination of chromatic, textural, postural and locomotor
components (Packard & Sanders, 1969; Packard & Hochberg, 1977; reviewed in Hanlon
& Messenger 1996). In an effort to classify the large number of patterns, Hanlon and
co-workers categorized body patterns into three main groups: Uniform, Mottled and
Disruptive (Hanlon & Messenger, 1988; Hanlon et al., 2009). The simplicity of this
 Camouflage in benthic cephalopods 183

classification has helped researchers study processes involved in camouflage and per-
ception using new methodologies (such as image analysis) while still supporting older
ethological research. However, although the findings of Hanlon and others are appealing
and persuasive, a degree of uncertainty remains as many body patterns do not fall clearly
into one of these categories. Furthermore, exploiting such a generalized classification
system may prevent us from noticing subtle variations in body-pattern responses to a
given stimulus.

8.6 Camouflage complexities

No simple task for an active cephalopod, maintaining camouflage appears to require

continual visual data streaming, a color-changing mechanism, texture assessment and
the ability to make predictions. For example, the immediate surroundings in which
octopuses live are often heterogeneous. Vegetation, light, terrain (whether corals, gravel
or sand are visible), color, texture, brightness and contrast are all subject to spatial and
temporal variation. Consequently, the body patterns needed for effective concealment are
equally diverse and the task of matching a specific background object is complex (Shohet,
Baddeley, Anderson, Kelman & Osorio, 2006). Indeed, a camouflaging cephalopod is
confronted with a set of problems and a variety of options with which to solve them:
Target species: Camouflage can be designed to hinder detection by a single species
or even a specific individual. This may be especially true when a predator must prevent
its detection by a specific prey. In some cases this may also involve avoiding raising
the alarm in other members of the prey’s group or in nearby species. The predator may
need to consider the specific line of sight and field of view of the prey, the capacity of
its visual system, its level of attention, and so on, to remain camouflaged. However, the
situation is different for the animal exploiting camouflage to avoid detection by predators.
Cephalopods are preyed upon by a wide range of animals such as fish, marine mammals
and birds. Each of these predators is equipped with its own sensory (visual) system, mode
of search and direction of approach. Therefore, a camouflaging cephalopod cannot limit
itself to a single strategy to prevent detection (e.g. using slow motion to fool motion
detectors), but instead it must avoid detection by any potential predator. Nonetheless,
cuttlefish and octopuses are known to ignore previously conceived threats once it has
been established that they are not under attack.
Orientation: Because a cephalopod cannot predict the position of potential predators,
its camouflage pattern needs to be effective from all orientations.
Point-of-view predicament: Since there is a difference in the position of the prey/
predator and the camouflaging animal, the latter has to match itself to a background seen
from a point of view it does not have (Figure 8.3). For example, if the octopus is located
mostly on the seabed while the predator is located in the water column, the octopus has
to match the scene as seen from above. This requires the camouflaging animal to match
its surroundings as they appear from above.
General versus specific resemblance: A cephalopod may either match the general
properties of its visual surroundings or it may mimic a specific object in its immediate
184 Noam Josef and Nadav Shashar

Figure 8.3 The point of view predicament: (a) Top-down view of a camouflaged Octopus vulgaris
among bright pebbles. The octopus is marked with an arrow. (b) The same scene mentioned in
(a) from the octopus’s point of view.

surroundings. Neither strategy has proven superior to the other; therefore, both are
widely used by cephalopods and in the greater animal kingdom.
Texture assessments: Imitating algae, coral, sand or rock, sometimes from a sub-
stantial distance from the objective, requires the cephalopod to visually assess the three-
dimensional texture of its background, possibly based on previous interactions with it
(Allen, Mäthger, Barbosa & Hanlon, 2009).
Color assessments: The physical lack of color perception promoted cephalopod
development of the creative solution (as yet unexplained) of background matching. Color
matching, which could be based as much on texture recognition as on the reflection of
the animal’s surroundings by its reflecting cells, remains one of the most intriguing
mysteries of cephalopod camouflage.
Matching feedback: The body pattern used by a cephalopod needs to match that
perceived by its visual system of its background. It would be useful (but not vital) for
the animal to be able to verify that indeed it is producing a matching pattern. No such
self-feedback mechanism has yet been reported (but see Hanlon & Messenger, 1988 and
Allen, Mäthger, Barbosa & Hanlon, 2009, for a putative feedback path).

8.7 Camouflage limitations

The defensive behaviors of many animals fall into three categories: freezing, fleeing
and fighting (Eilam, 2005). No single category alone provides an effective means of
defense, and combinations of defensive behaviors that may increase the survival rate of
the individual are limited (Eilam, 2005). Indeed, these categories set possible limitations
on camouflage, rendering the animal, under some conditions, unable to camouflage or
causing it to abandon camouflage altogether as a tactic.
Freezing is an important part in the behavior of hiding animals because most animals
are very sensitive to motion in their visual fields (Land & Nilsson, 2002). For example,
moving objects are often detected by the peripheral vision while a stable object goes
 Camouflage in benthic cephalopods 185

unnoticed. Once an animal has frozen in place, it can constantly monitor the distance
and activity of the target (predator or prey). If a frozen individual assumes it has been
detected nonetheless, it may execute a conspicuous display (possibly as a warning or a
declaration of recognition) followed by fleeing or fighting behavior (Eilam, 2005; see
Hanlon, Forsythe & Joneschild, 1999 – accompanying video http://hermes.mbl.edu/
mrc/hanlon/video.html for an example in octopus). Since an animal cannot freeze and
flee simultaneously, these defenses are mutually exclusive and are often controlled by
different motor systems (Canteras, 2002; Lamprea et al., 2002; Blanchard et al., 2003;
Brandao, Troncoso, Silva & Huston, 2003; Comoli, Ribeiro-Barbosa & Canteras, 2003;
Barros, Silva, Huston & Tomaz, 2004).
Movement is an obvious limitation for camouflage in animals in fleeing or chasing
mode. Due to the high sensitivity of most visual systems to movement and changes
in the image presented to the retina, it is likely that a fast moving animal cannot stay
camouflaged. Indeed fast moving, fleeing, chasing and attacking cephalopods lose their
cryptic appearance and often present a distinct pattern (Borrelli, Fiorito & Gherardi,
2006). However, cephalopods in motion are able to partially maintain crypsis while
moving slowly over areas of reef or sand (Hanlon, Forsythe & Joneschild, 1999; Mather
& Mather, 2004; Langridge, Broom & Osorio, 2007).

8.8 Communication and courtship

Communication requires that the signaler reveal its existence to other animals in its
vicinity. In the case of visual communication this often results in camouflage being
broken. Cephalopods have partially solved this limitation by two means. First, many taxa
employ directional displays in the form of bilaterally asymmetrical patterns, allowing
them to display a conspicuous pattern to potential mates on one side of the body
while maintaining a camouflage pattern on the other (Hanlon & Messenger, 1996;
Langridge, 2006). Second, by using discrete communication channels, signals are visible
to conspecifics but not to most potential predators (Shashar, Rutledge & Cronin, 1996).
Many cephalopod species can produce partly linearly polarized patterns on their body,
which can be seen only by polarization sensitive animals, such as themselves (Cronin,
Shashar & Wolff, 1995; Shashar, Rutledge & Cronin, 1996; Mäthger & Hanlon, 2006;
Talbot & Marshall, 2010; Temple et al., 2012). This may allow a communicating animal
to remain cryptic to polarization insensitive animals while communicating with its
conspecifics. Despite the research efforts that have been invested in cephalopod behavior,
however, the function of the polarization patterns is not yet fully understood (Boal et al.,
2004; Mäthger, Shashar & Hanlon, 2009).

8.9 Quantifying camouflage

The challenge of measuring and quantifying camouflage is studied in a range of

fields, such as image analysis, light spectrometry and animal behavior (Gretzmacher,
186 Noam Josef and Nadav Shashar

Table 8.2 Image parameters used to determine cuttlefish body patterns while camouflaged

Species studied Parameters tested Citation

Sepia officinalis Contrast and object size Barbosa et al. (2008)

Sepia officinalis Edges and contrast Chiao, Kelman and Hanlon (2005)
Sepia officinalis Edge completion Zylinski, Osorio, and Shohet (2009);
Zylinski, Darmaillacq, and
Shashar (2012)
Sepia pharaonis Area of light objects in the substrate Chiao and Hanlon (2001b)
Sepia officinalis Contrast, intensity and configuration Chiao, Chubb, and Hanlon (2007)
of small light elements
Sepia pharaonis Size, contrast and number of white Chiao and Hanlon (2001a)
squares
Sepia officinalis Scaling Chiao, Chubb, Buresch, Siemann,
and Hanlon (2009)
Sepia officinalis Localized visual edges Kelman, Baddeley, Shohet and
Osorio (2007)
Sepia officinalis Granularity Chiao et al. (2010)
Sepia officinalis Relative size of objects in Hanlon and Messenger (1988)
comparison to self body size

Ruppert & Nyberg, 1998; Reynolds, 2011). The analysis of cryptic body patterns depends
first on how we define the camouflage patterns. Trying to do so, the intricate camouflage
phenomenon has been untangled into its constituent parts to figure out which environ-
mental characteristics are reflected in the animal’s body patterns (Table 8.2). For a number
of cephalopod species, ethograms (Hanlon, Watson & Barbosa, 2010; Mather, Griebel &
Byrne, 2010), which provide detailed catalogues of body patterns, postures and behav-
iors, have been devised. But lacking a numerical, measurable element, an ethogram is a
dry inventory of the appearances and behaviors exhibited by a species. Another approach
is to rank the strength of body pattern components used in camouflage responses as per-
ceived by human observations (e.g. Hanlon, Forsythe & Joneschild, 1999; Barbosa et al.,
2007). For example, Hanlon, Forsythe and Joneschild (1999) asked observers to rank
the brightness, color, shape and skin patterning of foraging octopuses. Scores were used
to quantify polyphenism and cryptic behavior and to assess camouflage. Although such
methods are useful in determining how cephalopods themselves perceive their visual
environment, an unbiased process that minimizes the human perspective is preferable
to assess how effective body patterns are against detection by predators.
In recent years, the quest for objective examination has led to the development of
automatic methods that rely on unbiased mathematical approaches (Zylinski, How,
Osorio, Hanlon & Marshall, 2011; Josef, Amodio, Fiorito & Shashar, 2012). As such,
they score the strength of individual components, elucidating which aspects of the pattern
are influenced by different visual parameters. But this approach, too, suffers the same
limitation as the subjective measures, in that it is suitable for exploring the animal’s
perception of its environment as opposed to evaluating the effectiveness of the pattern
used. Image, spectrum and video analyses have facilitated new approaches that are
 Camouflage in benthic cephalopods 187

superior to traditional methods in terms of speed, accuracy and objectivity. The downside
of such objective methods is that they reduce the phenomenon to variables, thereby losing
touch with the biological relevance of the patterns themselves. Indeed, measuring and
quantifying complex cryptic behavior requires a certain amount of modesty, as the
process we are trying to understand has been tested by millions of eyes of different
species and evolutionarily refined to its present manifestation.
Before moving on to discuss mathematical camouflage descriptors, here we review
some basic principles of image analysis. Digital images comprise thousands of pixels
organized as a matrix of intensity values usually ranging from 0 to 255. Camouflage
image analysis is based on examining mathematical descriptors of the pixels in areas
of the animal’s body and comparing them with similar descriptors of the background
against which the animal is camouflaging. Three overlaying matrixes (the red, green, and
blue [RGB] channels) form a color image in the RGB format. If an animal’s perception of
the visual world is based on a single color receptor in its eyes – as is the case with shallow
water cephalopods, equipped with a single, mid-wavelength photoreceptor – the animal
perceives its surroundings in grayscale due to the lack of another matrix channel with
which to compare the image. Therefore, most analyses simplify the problem by using
grayscale images derived from the green channel intensities. But which mathematical
descriptors best describe similarity or resemblance in textures and patterns? Numerous
studies in computer and machine vision have tackled similar questions (e.g. Ojala,
Pietikainen & Harwood, 1996; Lowe, 2004), usually focusing on identifying differences
between or within images. Although such studies provide useful tools applicable to the
study of animal camouflage, these tools often tend to be too computationally complex
for the task at hand. Among the image parameters that have been used to investigate
cephalopod camouflage patterns (partial list, Table 8.2), each has its own advantages and
limitations that cannot be neglected when analyzing the results. None of the mentioned
studies claim to describe the neurological processes behind cephalopod body patterning;
rather, they attempt to find the descriptors that can best quantify camouflage while
remaining independent of researcher bias.
In a recent study we used previously applied image analysis pattern descriptors to
investigate the camouflage patterns of Octopus cyanea (Josef, Amodio, Fiorito, &
Shashar, 2012). Shannon’s entropy, one of the most intuitive parameters to compre-
hend (and sometimes referred as a “measure of uncertainty”), was found to be a good
proxy for pattern similarity. Other methods, including average root mean square devi-
ation (RMSD), standard deviation (STD), fast Fourier transformation (FFT), cross-
correlation, two-dimensional (2D) cross-correlation, edge detection, Hough transforma-
tion, histogram correlation, blob analysis, spectral estimation and histogram normal-fit
comparisons, were all deemed inadequate for use as camouflage descriptors. In fact,
we found, despite the significant loss of information they entail, that simple measuring
methods were sufficient to quantify similarities between an octopus and its surroundings.
Indeed, our study shows empirically that it is possible to estimate similarity between
an octopus and different parts of its background even when using a relatively simple
parameter such as the entropy. But, because it is insensitive to the spatial organization
of pixels in the image, entropy cannot be used in all applications. Comparisons of our
188 Noam Josef and Nadav Shashar

Figure 8.4 Commonly used artificial patterns. (a) Checkerboard, 50% black and white pixels
eliminating color dominance effect. With relation to scale, it may evoke either Mottle or
Disruptive body patterns. (b) Various size white objects over a black background, examining the
effect of area and object size in eliciting patterning. (c) Holstein cow patterns composed of 50%
black and white pixels, mostly evoking Disruptive body patterns. (d) Fifty-percent gray,
representing neutral and uniform substrata.

results with other complex and well-established methods, however, demonstrated that
entropy has both the power and the accuracy required to function as a natural scene
descriptor.
The ongoing search for adequate analysis descriptors is a “hot” topic in camouflage
research. Such descriptors would enable us to clearly distinguish camouflage from
signaling, identify similarity zones in a given scene and quantify, possibly even grade, the
quality of camouflage (Zylinski, How, Osorio, Hanlon & Marshall, 2011; Josef, Amodio,
Fiorito & Shashar, 2012). Additionally, they could also lead to an automated and objective
method of assessing camouflage.

8.10 How do cephalopods decide what camouflage strategy to use?

Cephalopods must rapidly and continuously assess their surroundings to determine

which body pattern to use. Scientists, interested in understanding how this assessment
is made and what parameters of the visual scene are taken into consideration by the
animal, cannot definitively declare whether camouflage is a controlled cognitive process,
during which the animal is aware of what it is matching. Conversely, it is possible that
cephalopod camouflage is a reflexive response that accepts inputs from the visual, motor
and memory systems to produce the desired body pattern.
The complex and diverse habitats of benthic cephalopods make field research chal-
lenging and hinder investigations of the effects of specific visual features on camouflage
patterns. The use of artificial substrates in controlled settings became a popular tech-
nique among researchers, as it enabled specific visual cues used in camouflage to be
examined and, in so doing, isolating the relevant image variables, which were then
manipulated while the other parameters were static (Figure 8.4; Chiao & Hanlon, 2001a,
2001b; Barbosa, Florio, Chiao & Hanlon, 2004; Chiao, Kelman & Hanlon, 2005; Bar-
bosa et al., 2007; Chiao, Chubb & Hanlon, 2007; Kelman, Baddeley, Shohet & Osorio,
2007; Barbosa, Litman & Hanlon, 2008). Among the images tested, the checkerboard
pattern is an important tool that can be modified for contrast and scale and even used in
 Camouflage in benthic cephalopods 189

experiments to assess polarization and color sensitivity (Marshall & Messenger, 1996;
Grable, Shashar, Gilles, Chiao & Hanlon, 2002).

8.11 Camouflaged to resemble what?

A camouflaging cephalopod can take into account its local visual field in its entirety
and execute general background matching, or it can imitate specific, selected structures
or objects in its surroundings, a strategy that is known as masquerade. Alternative
strategies such as Disruptive coloration may also be employed, a further testament to the
animal’s versatile camouflaging abilities. Indeed, the cephalopod’s ability to dynamically
camouflage allows it to choose the best strategy for each situation.
If using an object-oriented strategy, an animal must quickly identify an object charac-
teristic of the surrounding, convincingly and accurately imitate it and remain motionless.
On the other hand, the strategy of general background matching requires the animal to
find the best way to describe and represent a relatively large area around it, while ensur-
ing that it is properly oriented vis-à-vis its background and the seeker’s field of view. A
further option could involve mimicking the general characteristics of a specific object
without attempting to copy it in detail, simplifying the computational requirements as all
object parameters are precisely duplicated, but using camouflage to generally “belong”
to the scene. Hanlon and Messenger (1988) alluded to this option in their discussion of
juvenile cuttlefish mimicking pebbles.
In a recent study we attempted to determine which of these approaches is used by free
ranging O. cyanea on a coral reef (Josef, Amodio, Fiorito & Shashar, 2012). Free-ranging
octopuses were photographed from above, and the green channel of the resulting images
was analyzed. First, an octopus mantle viewed from above was sampled and measured for
both entropy and rotational averaged (RA)-slopes (Table 8.3). Next, the same procedure
was carried out for the rest of the image on areas equal in size to the animal’s mantle. By
shifting the areas one pixel at a time, the entire image was examined. The differences in
measurements between the octopus and its background were calculated for each position
of the shifting area, and their difference values were assigned to the central pixel of the
area. Comparing the octopus to the overall image creates a similarity index map, in
which a low difference value signifies high similarity.
The similarity index is computed as follows (RA-slope example):
  
|OctopusRA–slope − SubSampleRA–slope |
RAfft slope similarity percentage = 1− ·100
Maximum DifferenceRA–slope
However, the RA-slope alone is not sufficient to define similarity because the slopes
of two images with different intensities can be highly similar. Therefore, we used the
RA-mean as an indication of intensity and only sections of the image with RA-mean
similarity greater than 90% were measured for differences in RA-slopes.
Superimposing the resulting similarity maps over the original images revealed regions
of animal and background similarity and others that were dissimilar (see example in
Figure 8.5). We found that rather than use body patterns that represented an average of
190 Noam Josef and Nadav Shashar

Table 8.3 Image analysis descriptors used in the cephalopod camouflage study

Pattern descriptor Purpose Citation

PIV (pixel intensity variance) Distinguishing camouflage from Zylinski, How, Osorio, Hanlon and
signaling patterns Marshall (2011)
RA-slope (rotational averaged Quantifying similarity between two Zylinski, How, Osorio, Hanlon and
fast Fourier transformation images Marshall (2011)
[FFT] slope)
Modified entropy Comparing complexity This paper
Contrast Comparing either the white square Chiao and Hanlon (2001a)
to the rest of the body or to the
background
Spectrum granularity Distinguishing between Barbosa et al. (2008)
Uniform/Stipple, Mottled and
Disruptive body patterns
Energy per pixel Quantifying the influence of Chiao, Chubb, Buresch, Siemann
substrate spatial scale and the and Hanlon (2009)
camouflager
Two-dimensional Fourier Analyzing spatial frequency Chiao, Kelman and Hanlon (2005)
transform information
Spectral reflectance Measuring the color match between Mäthger, Chiao, Barbosa and
animal and background Hanlon (2008)
Webber contrast Comparing the animal’s Disruptive Hanlon et al. (2009)
body patterns and its background
attributes

their immediate environment (as in general background matching), the octopuses tended
to imitate certain regions, behavior consistent with a masquerade strategy. Repeating this
analysis using entropy estimation produced the same results, namely, the same regions
of the images with high similarity to the octopuses were identified (Figure 8.5). The
observed octopus strategy was not to precisely imitate specific objects, but, rather, to use
typical traits of the relevant objects. In doing so, the octopuses reduced the likelihood of
being detected by inaccurately mimicking the imitated object. Of course we do not intend
to suggest that octopuses use mathematical parameters in their camouflaging decisions,
merely that these descriptors assist us in understanding which habitat characteristics are
being utilized in camouflage.

8.12 Concluding remarks

The dynamic camouflage of cephalopods attracts the interest of increasing numbers of

researchers across multiple fields, including behavioral biology, military organizations
and computer vision. The development of autonomous and increasingly sophisticated
methods to assess camouflage is providing us with the power of greater insight into which
elements of a visual scene are selected for matching. In turn, understanding which scene
characteristics are being used by animals to determine body patterns for camouflage
 (a) (c) (e)

(b) (d) (f)

Figure 8.5 Image analysis of camouflaged octopuses. (a,b) Cryptic Octopus cyanea and Octopus vulgaris respectively in their natural habitat. The white
square marks the sampled mantle for similarity comparison. (c,d) Computer-generated similarity map of the RA-slopes descriptor. For clarity reasons
only 90% similarity or higher is presented. (e,f) Computer-generated similarity map of the entropy descriptor. See plate section for color version.
192 Noam Josef and Nadav Shashar

helps improve image analysis processes. Future research is needed to understand the
neurological processes taking place, from the basics of pattern control to brain functions
and neurologically mapping the body patterns.

Acknowledgments

We thank Steve MacCusker, Chris Talbot, Amit Lerner, Omer Polak and Keren Levi for
their ongoing assistance, and Zvika Livnat, Ron Wolf and Ilan Ben-Tov who donated
some of their high resolution images. We are grateful for the support of Israeli Science
Foundation Grant No. 1081/10, the ASSEMBLE program and the Halperin and the
Schechter Foundations.

References

Allen, J. J., Mäthger, L. M., Barbosa, A. and Hanlon, R. T. (2009). Cuttlefish use visual cues to
control three-dimensional skin papillae for camouflage. Journal of Comparative Physiology A,
195: 547–555.
Arnold, J. M. (1967). Organellogenesis of the cephalopod iridophore: cytomembranes in devel-
opment. Journal of Ultrastructure Research, 20: 410–420.
Bacot, A. (1907). The melanic variety of the peppered moth. Nature, 77: 294–295.
Barbosa, A., Allen, J. J., Mäthger, L. M. and Hanlon, R. T. (2012). Cuttlefish use visual cues
to determine arm postures for camouflage. Proceedings of the Royal Society. B: Biological
Sciences, 279: 84–90.
Barbosa, A., Florio, C. F., Chiao, C. C. and Hanlon, R. T. (2004). Visual background features that
elicit mottled body patterns in cuttlefish Sepia officinalis. The Biological Bulletin, 207: 154.
Barbosa, A., Litman, L. and Hanlon, R. T. (2008). Changeable cuttlefish camouflage is influenced
by horizontal and vertical aspects of the visual background. Journal of Comparative Physiology
A, 194: 405–413.
Barbosa, A., Mäthger, L. M., Buresch, K. C., Kelly, J., Chubb, C., Chiao, C. C. and Hanlon, R. T.
(2008). Cuttlefish camouflage: The effects of substrate contrast and size in evoking Uniform,
Mottle or Disruptive body patterns. Vision Research, 48: 1242–1253.
Barbosa, A., Mäthger, L. M., Chubb, C., Florio, C., Chiao, C. C. and Hanlon, R. T. (2007).
Disruptive colouration in cuttlefish: a visual perception mechanism that regulates ontogenetic
adjustment of skin patterning. Journal of Experimental Biology, 210: 1139–1147.
Barros, M., Silva, M. A. D., Huston, J. P. and Tomaz, C. (2004). Multibehavioral analysis of
fear and anxiety before, during, and after experimentally induced predatory stress in Callithrix
penicillata. Pharmacology Biochemistry and Behavior, 78: 357–367.
Blanchard, D. C., Li, C. I., Hubbard, D., Markham, C. M., Yang, M., Takahashi, L. K. and
Blanchard, R. J. (2003). Dorsal premammillary nucleus differentially modulates defensive
behaviors induced by different threat stimuli in rats. Neuroscience Letters, 345: 145–148.
Boal, J. G., Shashar, N., Grable, M. M., Vaughan, K. H., Loew, E. R. and Hanlon. R. T. (2004).
Behavioral evidence for intraspecific signals with monochromatic and polarized light by cut-
tlefish. Behaviour, 141: 837–861.
Borrelli, L., Fiorito, G. and Gherardi, F. (2006). A catalogue of body pattering in Cephalopoda.
Florence, Italy: Firenze University Press.
 Camouflage in benthic cephalopods 193

Boycott, B. B. (1953). The chromatophore system of cephalopods. Proceedings of the Linnean

Society of London, 164: 235–240.
Brandao, M. L., Troncoso, A. C., Silva, M. A. D. and Huston, J. P. (2003). The relevance of
neuronal substrates of defense in the midbrain tectum to anxiety and stress: empirical and
conceptual considerations. European Journal of Pharmacology, 463: 225–233.
Bush, S. L., Robison, B. H. and Caldwell, R. L. (2009). Behaving in the dark: locomotor, chromatic,
postural and bioluminescent behaviors of the deep-sea squid Octopoteuthis deletron Young
1972. The Biological Bulletin, 216: 7–22.
Canteras, N. S. (2002). The medial hypothalamic defensive system: hodological organization and
functional implications. Pharmacology Biochemistry and Behavior, 71: 481–491.
Chiao, C. C., Chubb, C., Buresch, K. C., Barbosa, A., Allen, J. J., Mäthger, L. M. and Hanlon,
R. T. (2010). Mottle camouflage patterns in cuttlefish: quantitative characterization and visual
background stimuli that evoke them. Journal of Experimental Biology, 213: 187–199.
Chiao, C. C., Chubb, C., Buresch, K. C., Siemann, L. and Hanlon, R. T. (2009). The scaling effects
of substrate texture on camouflage patterning in cuttlefish. Vision Research, 49: 1647–1656.
Chiao, C. C., Chubb, C. and Hanlon, R. T. (2007). Interactive effects of size, contrast, intensity
and configuration of background objects in evoking Disruptive camouflage in cuttlefish. Vision
Research, 47: 2223–2235.
Chiao, C. C. and Hanlon, R. T. (2001a). Cuttlefish camouflage: visual perception of size contrast
and number of white squares on artificial checkerboard substrate initiates Disruptive coloration.
Journal of Experimental Biology, 204: 2119–2125.
Chiao, C. C. and Hanlon, R. T. (2001b). Cuttlefish cue visually on area – not shape or aspect
ratio – of light objects in the substrate to produce Disruptive body patterns for camouflage. The
Biological Bulletin, 201: 269–270.
Chiao, C. C., Kelman, E. J. and Hanlon, R. T. (2005). Disruptive body patterning of cuttlefish
(Sepia officinalis) requires visual information regarding edges and contrast of objects in natural
substrate backgrounds. The Biological Bulletin, 208: 7–11.
Chrachri, A. and Williamson, R. (2004). Cholinergic and glutamatergic spontaneous and evoked
excitatory postsynaptic currents in optic lobe neurons of cuttlefish, Sepia officinalis. Brain
Research, 1020: 178–187.
Cloney, R. A. and Brocco, S. L. (1983). Chromatophore organs, reflector cells, iridocytes and
leucophores in cephalopods. American Zoologist, 23: 581–592.
Comoli, E., Ribeiro-Barbosa, E. R. and Canteras, N. S. (2003). Predatory hunting and exposure
to a live predator induce opposite patterns of Fos immunoreactivity in the PAG. Behavioural
Brain Research, 138: 17–28.
Cronin, T. W. (2005). The visual ecology of predator–prey interactions. In Barbosa, P. and Castel-
laris, I. (eds.) Ecology of predator-prey interactions. New York, NY: Oxford University Press,
pp. 105–138.
Cronin, T. W., Shashar, N. and Wolff, L. (1995). Imaging technology reveals the polarized light
fields that exist in nature. Biophotonics International, 2: 38–41.
Curio, E. (1976). The ethology of predation. Berlin, Germany: Springer-Verlag.
Darwin, C. (1870). Note on the habits of the Pampas Woodpecker (Colaptes campestris). Pro-
ceedings of the Zoological Society of London, 38: 705–706.
Eilam, D. (2005). Die hard: a blend of freezing and fleeing as a dynamic defense – implications for
the control of defensive behavior. Neuroscience and Biobehavioral Reviews, 29: 1181–1191.
Endler, J. A. (1981). An overview of the relationships between mimicry and crypsis. Biology
Journal of the Linnean Society, 16: 25–31.
194 Noam Josef and Nadav Shashar

Florey, E. and Kriebel, M. E. (1969). Electrical and mechanical responses of chromatophore

muscle fibers of the squid, Loligo opalescens, to nerve stimulation and drugs. Zeitschrift fur
Vergleichende Physiologie, 65: 98–130.
Grable, M., Shashar, N., Gilles, N. L., Chiao, C. C. and Hanlon, R. T. (2002). Cuttlefish body
patterns as a behavioral assay to determine polarization perception. Biology Bulletin, 203:
232–234.
Gretzmacher, F. M., Ruppert, G. S. and Nyberg, S. (1998). Camouflage assessment considering
human perception data. Proceedings SPIE, Targets and Backgrounds: Characterization and
Representation IV, 3375: 58–67.
Hall, K. and Hanlon, R. (2002). Principal features of the mating system of a large spawning
aggregation of the giant Australian cuttlefish Sepia apama (Mollusca: Cephalopoda). Marine
Biology, 140: 533–545.
Hanlon, R. T. (1982). The functional organization of chromatophores and iridescent cells in the
body patterning of Loligo plei (Cephalopoda: Myopsida). Malacologia, 23: 89–119.
Hanlon, R. T., Chiao, C. C., Mäthger, L. M., Barbosa, A., Buresch, K. C. and Chubb, C. (2009).
Cephalopod dynamic camouflage: bridging the continuum between background matching and
Disruptive coloration. Philosophical Transactions of the Royal Society of London. Series B,
364: 429–437.
Hanlon, R. T., Forsythe, J. W. and Joneschild, D. E. (1999). Crypsis, conspicuousness, mimicry and
polyphenism as anti-predator defenses of foraging octopuses on Indo-Pacific coral reefs, with
a method of quantifying crypsis from video tapes. Biological Journal of the Linnean Society,
66: 1–22.
Hanlon, R. T. and Messenger, J. B. (1988). Adaptive coloration in young cuttlefish (Sepia offic-
inalis L.): the morphology and development of body patterns and their relation to behaviour.
Philosophical Transactions of the Royal Society B, 320: 437–487.
Hanlon, R. T. and Messenger, J. B. (eds.) (1996). Cephalopod behaviour. Cambridge: Cambridge
University Press.
Hanlon, R. T., Watson, A. C. and Barbosa, A. (2010). A “Mimic Octopus” in the Atlantic: flatfish
mimicry and camouflage by Macrotritopus defilippi. Biological Bulletin, 218: 15–24.
Josef, N., Amodio, P., Fiorito, G. and Shashar, N. (2012). Camouflaging in a complex environment
– octopuses use specific features of their surroundings for background matching. Public Library
of Science One, 7: e37579.
Kelman, E. J., Baddeley, R. J., Shohet, A. J. and Osorio, D. (2007). Expression of Disruptive
camouflage by the cuttlefish, Sepia officinalis. Proceedings of the Royal Society. B: Biological
Sciences, 274: 1369–1375.
Lamprea, M. R., Cardenas, F. P., Vianna, D. M., Castilho, V. M., Cruz-Morales, S. E. and Brandao,
M. L. (2002). The distribution of fos immunoreactivity in rat brain following freezing and
escape responses elicited by electrical stimulation of the inferior colliculus. Brain Research,
950: 186–194.
Land, M. F. and Nilsson, D. E. (2002). Animal eyes. Oxford, UK: Oxford University Press.
Langridge, K. V. (2006). Symmetrical crypsis and asymmetrical signaling in the cuttlefish Sepia
officinalis. Proceedings of the Royal Society. B: Biological Sciences, 273: 959–967.
Langridge, K. V., Broom, M. and Osorio, D. (2007). Selective signalling by cuttlefish to predators.
Current Biology, 17: R1044–R1045.
Lehmann, U. (1976). The ammonites: their life and their world. Translation 1981. Cambridge,
UK: Cambridge University Press.
Lowe, D. (2004). Distinctive image features from scale–invariant keypoints. International Journal
on Computer Vision, 60: 91–110.
 Camouflage in benthic cephalopods 195

Marshall, N. J. and Messenger, J. B. (1996). Colour-blind camouflage. Nature, 382: 408–409.

Mather, J. A., Griebel, U. and Byrne, R. A. (2010). Squid dances: an ethogram of postures
and actions of Sepioteuthis sepioidea squid with a muscular hydrostatic system. Marine and
Freshwater Behaviour and Physiology, 43: 45–61.
Mather, J. A. and Mather, D. L. (2004). Apparent movement in a visual display: the “passing
cloud” of Octopus cyanea (Mollusca: Cephalopoda). Journal of Zoology, 263: 89–94.
Mäthger, L. M, Chiao, C. C., Barbosa, A. and Hanlon, R. T. (2008). Color matching on natural
substrates in cuttlefish, Sepia officinalis. Journal of Comparative Physiology A, 194: 577–585.
Mäthger, L. M. and Hanlon, R. T. (2006). Anatomical basis for camouflaged polarized light
communication in squid. Biology Letters, 2: 494–496.
Mäthger, L. M., Shashar, N. and Hanlon, R. T. (2009). Do cephalopods communicate using
polarized light reflections from their skin. Journal of Experimental Biology, 212: 2133–
2140.
Mayr, E. (1963). Animal species and evolution. Cambridge, MA: Harvard University Press.
Merilaita, S. (1998). Crypsis through Disruptive coloration in an isopod. Proceedings of the Royal
Society. B: Biological Sciences, 265: 1059–1064.
Messenger, J. B. (1977). Evidence that octopus is color blind. Journal of Experimental Biology,
70: 49–55.
Messenger, J. B. (2001). Cephalopod chromatophores: neurobiology and natural history. Biologi-
cal Reviews, 76: 473–528.
Moynihan, M. (1985). Communication and non-communication in cephalopods. Bloomington,
IL: Indiana University Press.
Nixon, M. and Young, J. Z. (2003). The brains and lives of cephalopods. New York, NY: Oxford
University Press.
Ojala, T., Pietikainen, M. and Harwood, D. (1996). A comparative study of texture measures with
classification based on feature distributions. Pattern Recognition, 29: 51–59.
Packard, A. (1972). Cephalopods and fish: the limits of convergence. Biological Reviews, 47:
241–307.
Packard, A. and Hochberg, F. G. (1977). Skin patterning in octopus and other genera. Symposia
of the Zoological Society of London, 38: 191–231.
Packard, A. and Sanders, G. D. (1969). What the octopus shows to the world. Endeavour, 28:
92–99.
Parker, G. H. (1948). Animal colour changes and their neurohumours. Cambridge, UK: Cambridge
University Press.
Reynolds, C. (2011). Interactive evolution of camouflage. Artificial Life, 17: 123–136.
Sereni, E. and Young, J. Z. (1932). Nervous degeneration and regeneration in cephalopods.
Pubblicazione de la Stazione Zoologica di Napoli, 12: 173–208.
Shashar, N., Rutledge, P. S. and Cronin, T. W. (1996). Polarization vision in cuttlefish – a concealed
communication channel. Journal of Experimental Biology, 199: 2077–2084.
Shohet, A. J., Baddeley, R. J., Anderson, J. C., Kelman, E. J. and Osorio, D. (2006). Cuttlefish
responses to visual orientation of substrates, water flow and a model of motion camouflage.
Journal of Experimental Biology, 209: 4717–4723.
Stevens, M. and Merilaita, S. (2009). Animal camouflage: current issues and new perspectives.
Philosophical Transactions of the Royal Society B, 364: 423–427.
Stevens, M. and Merilaita, S. (2011). Animal camouflage. Cambridge, UK: Cambridge University
Press.
Stuart-Fox, D. and Moussalli, A. (2008). Selection for social signalling drives the evolution of
chameleon colour change. Public Library of Science One, Biology, 6: e25.
196 Noam Josef and Nadav Shashar

Talbot, C. M. and Marshall, N. J. (2010). Polarization sensitivity in two species of cuttlefish – Sepia
plangon (Gray 1849) and Sepia mestus (Gray 1849) – demonstrated with polarized optomotor
stimuli. Journal of Experimental Biology, 213: 3364–3370.
Temple, S. E., Pignatelli, V., Cook, T., How, M. J., Chiou, T.-H., Roberts, N. W. and Marshall, N.
J. (2012). High-resolution polarisation vision in a cuttlefish. Current Biology, 22: R121–R122.
Zylinski, S., Darmaillacq, A.-S. and Shashar, N. (2012). Visual interpolation for contour comple-
tion by the cuttlefish Sepia officinalis and its use in dynamic camouflage. Proceedings of the
Royal Society. B: Biological Sciences, 279: 2386–2390.
Zylinski, S., How, M. J., Osorio, D., Hanlon, R. T. and Marshall, N. J. (2011). To be seen or to
hide: visual characteristics of body patterns for camouflage and communication in the Australian
giant cuttlefish Sepia apama. The American Naturalist, 177: 681–690.
Zylinski, S. and Johnsen, S. (2011). Mesopelagic cephalopods switch between transparency and
pigmentation to optimize camouflage in the deep. Current Biology, 21: 1937–1941.
Zylinski, S., Osorio, D. and Shohet, A. J. (2009). Edge detection and texture classification by
cuttlefish. Journal of Vision, 9: 1–10.
9 Cuttlefish camouflage:
vision and cognition
Sarah Zylinski and Daniel Osorio

Because vision is used by different animals for such a variety of purposes, it is inconceivable that
all seeing animals use the same representations; each can confidently be expected to use one or
more representations that are nicely tailored to the owner’s purposes. (David Marr (1982))

9.1 Cephalopod camouflage and comparative cognition

Camouflage is directed at the vision of adversaries, with selection on both sides driving
an evolutionary arms race. Fossils suggest that Mesozoic coleoid cephalopods moved to
deep water to avoid predation and competition, with shell-loss an adaptation to overcome
depth limits imposed by hydrostatic pressure on gas spaces in shells. There followed a
secondary colonization of shallow waters where the soft-bodied, shell-less cephalopods
acquired novel defences against fish and reptile predators (Hanlon & Messenger, 1996;
Packard, 1972). A second hypothesis favours earlier colonization in the Devonian, when
the predators were early fishes such as chondrosteans and placoderms (Hanlon & Mes-
senger, 1996). Either way, fish can be described as the ‘designers of cephalopod skin’
(Packard, 1972), and we expect threat responses and camouflage patterns to be tuned
to vertebrate vision (Hanlon & Messenger, 1988; Langridge, Broom & Osorio, 2007).
These changes were probably linked to the more general evolution of large brains and
complex behaviours of modern cephalopods (Budelmann, 1994; Young, Vecchione &
Donovan, 1998).
Whatever its evolutionary origins we can agree that cephalopod camouflage works
well for at least one species of terrestrial primate, namely humans. Camouflage relies on
defeating figure-ground segregation to prevent the detection of the body as distinct from
the background (Cott, 1940; Stevens, 2007). Superficially, camouflage seems unprob-
lematic – we know a moth is camouflaged because we see that it is well matched to the
tree-bark it rests on – yet substantial neural computation is required to recover informa-
tion about objects from optical images. This chapter describes our current understanding
of cuttlefish (Sepia spp.) vision, how it is used to control camouflage, and the evidence
that this is a cognitive process.
This work generally falls in line with the traditions of vision science; first seeking to
identify the low level (i.e. spatially local) image parameters and features that are used by
Cephalopod Cognition, eds A.-S. Darmaillacq, L. Dickel and J. Mather. Published by Cambridge University
Press. 
C Cambridge 2014.
198 Sarah Zylinski and Daniel Osorio

the animals, and then asking how this information is processed by or represented in the
brain to make decisions – in this case to select a body pattern. How might this inform
our understanding of cognition?

9.2 Animal visual cognition

Comparative cognition is almost universally grounded in comparisons with human intel-

ligence (Byrne & Bates, 2006). We ask whether an animal has a mental map, con-
sciousness, theory of mind, declarative memory and so-forth. Often for animals without
language the definitions of these phenomena are unclear (Shettleworth, 2001). Instead
the chief experimental question is whether the animal’s behaviour can be described by
a behaviourist account (Real, 1993): if not it is inferred that the behaviour requires
cognition.
Unfortunately, as David Marr (1982) implied in the opening quotation, the approach
to animal cognition summarized above seems flawed: the theory is flimsy, and the
anthropocentric perspective is worrying. Human performance is not a gold standard for
cephalopod polarization vision or motor control. Humans are good at interacting with
objects, but do not have to control their body pattern. Cephalopods do things differently,
and we should be open to the possibility that this applies to mental processes (Temple
et al., 2012). The control of cuttlefish adaptive colouration for camouflage exemplifies
this point; its remarkable versatility and sensitivity to the visual environment offers a
unique approach to vision in an animal without language. We can examine how traditional
approaches to visual perception (Campbell & Robson, 1968; Cavanagh, 2011; Gibson,
1979; Marr, 1982; Marr & Hildreth, 1980) apply to these remarkable animals.
The difficulty in applying existing precepts to cephalopod behaviour is exemplified
by the apparent complexity of cuttlefish camouflage compared to what seems to be
necessary for effective concealment! Flatfish, such as plaice (Pleuronectes platessa)
and tropical flounders (Bothus ocellatus), control the level of expression of one to three
basic body patterns, which they can mix to produce – for the human eye – excellent
camouflage on a wide range of the sea-floor backgrounds (Kelman, Tiptus & Osorio
2006; Ramachandran et al., 1996). Similarly, it is convenient to classify cephalopod
camouflage into three basic types: Uniform, Mottle and Disruptive (Hanlon, 2007;
Hanlon & Messenger, 1988). However, cuttlefish almost certainly have substantially
more than three types of camouflage pattern, and more than three degrees of freedom
in the control of their appearance. The common cuttlefish Sepia officinalis controls
the expression of about 40 basic behavioural components more or less independently.
How their expression is coordinated, and how visual information is used to serve this
behaviour is a fascinating question in cephalopod biology and comparative cognition.
The nature of cognitive processes is central to understanding neural mechanisms
of behaviour. In her book Cognition, evolution, and behavior, Shettleworth (2010)
says: ‘Cognition, broadly defined, includes perception, learning, memory and decision
making, in short all ways in which animals take in information about the world through
 Cuttlefish camouflage: vision and cognition 199

the senses, process, retain and decide to act on it.’ Although pragmatic, the definition is
broad, and it is useful to define terms for this article. Perception, which is the ability to
make inferences about the world from sense data (Gregory, 1997), decision-making and
learning refer to behavioural capabilities, which should be distinguished from cognition,
which refers to underlying mechanisms, and is commonly understood to entail internal
representation (Tolman, 1948). It remains potentially confusing to refer to any type of
learning or object recognition as cognitive. In a classic cognitive model, Tolman (1948)
proposed that rats use a cognitive (i.e. internal) map to navigate a maze. Even though it is
widely agreed that behaviourist theory faces major difficulties in accounting for animal
learning (Mackintosh, 1974), the long-standing controversy about evidence for cognitive
maps exemplifies the difficulties in proving the existence of an internal representation in
any given experiment (Bennett, 1996; Burgess, 2006; Collett & Collett, 2002; Wang &
Spelke, 2002). Similarly there is scepticism about the need for representation, especially
in the field of artificial intelligence and robotics (Brooks, 1991).
Related to the distinction between cognitive and non-cognitive behaviour, vision sci-
ence has long recognized a distinction between low-level processes informed purely by
bottom-up, scene-based information such as edge detection, and high-level processes that
take account of global organization and involve long-range top-down, viewer-informed,
decision-based analyses that combine low-level retinal input with prior knowledge
to generate representation and are (probably) cognitive (Cavanagh, 2011). Cognitive
mechanisms are thought to be computationally demanding and, hence, costly for
neural processing, but they offer flexible processing and interpretation of sensory
information.
Marr’s (1982) emphasis on internal representations and computation has been espe-
cially influential and controversial. Marr proposed that vision is a multi-stage process
based on successive neural representations of the retinal image. In the first stage, local
(i.e. low-level) mechanisms, such as linear spatial filters, and edge and detectors, pro-
duce a primal sketch, which is elaborated by more global and contextual information
to produce a 2½-dimensional representation of a scene. Marr also made a well-known
distinction between the algorithmic, computational and hardware (i.e. physiological) lev-
els of understanding of the brain, which is especially relevant to comparisons between
remotely related species.
Marr contrasted his cognitive account with Gibson’s (1979) theory of direct perception.
Gibson largely rejected the need for representation, and instead emphasized how the
interaction of an actively moving agent with the visual world can constrain and simplify
visual computation. Subsequent work on animal and robotic vision has often favoured a
Gibsonian perspective; it emphasizes the interaction between the actively moving animal
and its visual world, and argues that Marr overstated the amount of computation necessary
and significance of representations for vision (Brooks, 1991; Krapp & Hengstenberg,
1997; Srinivasan & Venkatesh, 1997).
As they cannot verbally report their experience, animal experiments often rely on
binary-choice tests of behaviour. Such tests are poorly suited to investigating cognition,
because a simpler account is normally possible. The study of cuttlefish camouflage
200 Sarah Zylinski and Daniel Osorio

Figure 9.1 By placing animals on carefully designed stimuli and observing their body-pattern
responses we can gain unique insight into their perception of the visual world. Here, an
individual Sepia officinalis is settled in a test arena with a small chequerboard stimulus. The
animal uses a Mottle-type body pattern containing light and dark elements that are spatially
similar to those in the stimulus.

echoes broader debates on animal cognition, but the richness of the behaviour offers
new insights into how animals perceive complex images (Crook, Baddeley & Osorio,
2002). Recent work has moved from identifying low-level mechanisms towards more
cognitive accounts. Experiments set out to relate the choice of body pattern directly to
local image parameters and features, such as spatial frequency and edges (Figure 9.1);
yet, as we will see, it is often easier to conceive that the animals use an object-based
representation to select the right body pattern for the environment. Marr’s (1982) primal
sketch theory is important here because of its bottom-up approach to visual processing:
it shows how scene representation can be achieved in the absence of knowledge or
experience of the contents.
 Cuttlefish camouflage: vision and cognition 201

Figure 9.2 The bright blues and greens in the mating displays of the Australian giant cuttlefish
Sepia apama are primarily due to wavelength specific reflectors. The animals themselves are
most likely colour-blind. See plate section for colour version.

9.3 Expression of body patterns

The wealth of patterns used by cuttlefish is compelling evidence for the visual sophis-
tication of their potential predators and prey. This section outlines the basic biology of
cephalopod adaptive camouflage and goes on to outline how we can study this behaviour.

9.3.1 Skin pigmentation

Cephalopod skin is a remarkable organ, which compensates for the animals’ lack of
physical defences with control of both colour and physical texture (Hanlon, Forsythe
& Joneschild, 1999; Hanlon & Messenger, 1988, 1996). The skin is multi-layered:
chromatophores, which contain ommochrome pigments that produce blacks, browns,
reds and yellows, overlie white leucophores, while iridophores contain wavelength-
specific reflectors that result in vibrant blues and greens (Mäthger & Hanlon, 2007;
Messenger, 2001) (Figure 9.2). Chromatophore density varies considerably between
taxa, but on Octopus vulgaris mantle chromatophore density reaches 230 per mm2
(Messenger 2001). The chromatophores pigment sacs are opened by supporting muscles,
which are innervated directly from the brain, allowing rapid colour changes and precise
control of the visible area of pigment. In theory the animal could control these individual
202 Sarah Zylinski and Daniel Osorio

chromatophores like pixels on a computer monitor. Patterns could then be expressed in

response to the visual environment via a mechanism akin to an output of a retinotopic
map, with photoreceptor-to-chromatophore mapping of the background.
In fact the expression of chromatophores is coordinated, sometimes as a localized
group, and sometimes as more global patterns. These form the basic behavioural ele-
ments known as ‘chromatic components’ (Hanlon & Messenger, 1988; Packard &
Hochberg, 1977; Packard & Sanders, 1971). Chromatic components in turn tend to
be expressed, along with textural and postural components, in common suites to form
complete body patterns (Hanlon & Messenger, 1996). The suite of 50 or so behavioural
components identified in Sepia has obvious equivalents in other coleoid cephalopods,
but is unmatched in any other animal group (Kelman, Tiptus & Osorio, 2006; Stuart-
Fox & Moussalli, 2009; Stuart-Fox, Whiting & Moussalli, 2006). Measurement of their
expression is central to characterizing cephalopod visual behaviour. The potential for
independent expression of the behavioural components offers the cuttlefish an enormous
repertoire of body patterns, especially as the animal can precisely regulate the intensity
of each component. The extent to which the animals can realize this potential is poorly
known.
To facilitate understanding of the patterns that are typically expressed in response to a
set of basic visual stimuli, we will refer to the three main classes of body pattern in Han-
lon’s UMD (Uniform, Mottle, Disruptive; Figure 9.3) scheme (Hanlon, 2007). However,
the reader should bear in mind that the animal has enormous – and largely unexplored –
flexibility: each class contains many different body patterns distinguished by the rela-
tive expression of individual components and overall luminance. Equally, different body
pattern classes are often mixed (see Figure 9.3, below). How Sepia determines what
body pattern components to express in order to achieve effective camouflage, given
local visual cues, forms the crux of our research and the core of this chapter.

9.3.2 Principal component analysis and other descriptors of body patterns

It is readily apparent that to study cuttlefish adaptive camouflage requires a rigorous,
efficient and meaningful method to describe body patterns. Most methods deal with
the appearance in single images (e.g. Barbosa et al., 2008; Chiao, Chubb, Buresch &
Siemann, 2009; Shohet, Baddeley, Anderson & Osorio, 2007), but the patterns of change
over time are also of considerable interest (Langridge, Broom & Osorio, 2007; Zylinski,
Osorio & Shohet, 2009a). At present two main methods are used to describe body
patterns. One is to measure the spatial properties of the body pattern, much as one can
characterize a visual texture, in terms of the responses of a set of tuned spatial filters
(Allen et al., 2010; e.g. Barbosa et al., 2008; Chiao et al., 2010). Such measures can
be run on photographs of the animals in a semi-automated manner, and they can be
compared directly to measures of the background allowing a direct evaluation of the
quality of camouflage. A disadvantage is that they are normally one-dimensional; the
pattern is defined by a single parameter, and so does not represent the full range of
patterns that can be produced. An alternative and more labour-intensive method is to
score the level of expression of a large number (or all) of the body pattern components
 Cuttlefish camouflage: vision and cognition 203

(Kelman, Baddeley, Shohet & Osorio, 2007; Kelman, Osorio & Baddeley, 2008). This
method potentially allows the investigation of how camouflage patterns are fine-tuned in
response to visual stimuli. However, large multivariate data sets are unwieldy to analyse
and difficult to interpret, and as a result biological relevance and patterns within the data
can become lost. Principal component analysis (PCA) allows for such multivariate data
sets to be reduced with a minimum loss of information, effectively transforming data of
high dimensionality to one of fewer dimensions (Manly, 1994). PCA creates a set of
new independent orthogonal variables based on the degree of covariance between sets
of original variables. The first principal component (PC) is derived from an eigen-vector
that accounts for as much of the variance within the original data as possible, with each
following PC accounting for successively smaller proportions of the observed variance.
Ideally for further analysis, the significant variation of the data is adequately described
in the first few PCs (Jolliffe, 1986).
In our work on cuttlefish colouration a data set for PCA is produced by assessing the
strength of expression of up to 42 chromatic, textural and postural components, which
were identified by Hanlon and Messenger (1988). Each component is scored on a four-
point scale from not expressed (0) to strongly expressed (3) (e.g. Kelman, Baddeley,
Shohet & Osorio, 2007; Zylinski, Darmaillacq & Shashar, 2012). The number of com-
ponents ultimately included in the analysis varies as uninformative components can be
excluded. The data set of component scores is subsequently reduced by PCA. A decision
must next be made regarding the number of PCs retained for further analysis, commonly
determined from a scree plot and/or the Kaiser criterion (Jolliffe, 1986). Eigen-vectors
can be rotated to maximize the distance between PCs, but in practice unrotated PCs tend
to provide variables that are biologically relevant and easily interpretable – with rele-
vance to cuttlefish camouflage responses: in our experiments PCs 1 and 2 typically and
broadly equate to Disruptive and Mottle body patterns (Figure 9.3; and see Figure 9.4
later in the chapter). Consideration of the original body pattern components that con-
tribute to the retained PCs allows us to visualize which body pattern components tend
to be correlated in their expression (e.g. Kelman, Baddeley, Shohet & Osorio, 2007;
Kelman, Osorio & Baddeley, 2008; Zylinski, Osorio & Shohet, 2009b). The response
to test stimuli can then be assessed using the new PC scores through common statistical
tests.

9.4 Cuttlefish vision

Cephalopods have large, single chambered camera-type eyes (with the notable exception
of the nautiloids, which have a unique pin-hole eye) whose optics and overall structure
resembles that of fish, providing a striking example of convergence driven by a shared
ecology (Land & Nilsson, 2002; Packard, 1972). Similarities include eye size range,
lenses with a varying refractive index to minimize spherical aberration, variable pupils
and migrating screening pigments. There are also differences: cephalopods photorecep-
tors face towards the light (i.e. unlike ours they are the ‘right way around’, meaning
there is no blind spot); they have rhabdomeric rather than ciliary photoreceptors, and
204 Sarah Zylinski and Daniel Osorio

Figure 9.3 Demonstration of just some of the body pattern variation Sepia officinalis uses in
camouflage body-pattern responses to the visual environment. The top row shows the primary
characteristics of the main three classes of body pattern referred to in the text: Uniform (left),
Mottle (centre) and Disruptive (right). The remaining images show how individual chromatic
components patterns can be varied in their expression, along with the overall contrast of the body
pattern, to give a wide range of body patterns that cannot always be readily identified as
Uniform, Mottle or Disruptive. See plate section for colour version.
 Cuttlefish camouflage: vision and cognition 205

most species have only a single photopigment type (Marshall & Messenger, 1996). Dif-
ferences are also apparent in the location of early neuronal processing; while vertebrates
process a lot of visual information in the retina via the selective receptive fields of gan-
glion cells, the cephalopod retina lacks such cells and therefore does little processing.
It is believed that the types of processing done in the vertebrate retina occurs instead in
the large optic lobe (Williamson & Chrachri, 2004).
Studies of neural receptive fields from vertebrate retinal ganglion cells and primary
visual cortex, along with human psychophysical studies, show that the vertebrate visual
system first encodes the retinal image via local processes, operating on small regions
of the retinal image. The local operations in the low-level stage of vision have been
modelled as linear spatio-temporal filters, and as non-linear operations such as edge and
motion detectors and, perhaps, to local texture analyzers (Bruce, Green & Georgeson,
1996). These non-linear processes are sometimes identified as feature detectors. Apart
from local processing all stages of the visual pathway (but especially beyond the retina)
have long-range interactions and receive feedback from higher centres. Such processes
are probably crucial for figure-ground segregation: for example object borders can be
identified by intensity differences, colour, texture, motion and depth (Zipser, Lamme &
Schiller, 1996). Typically these processes which are sensitive to global context and are
thought to occur at (or involve via feedback) later stages of processing; for example, pre-
striate cortical areas, and they are classified as ‘high-level’. In addition to the integration
of multiple low-level cues this involves long-range interactions, contextual information,
memory and attention. High-level vision is conceived as involving representations of
the visual world and, hence, is cognitive (Cavanagh, 2011; Marr, 1982).
Findings on cephalopod spatial-vision fit into the foregoing scheme where low-level
local processing involving linear-spatial filtering and local-feature detection is followed
by higher-level (putatively cognitive) processes that integrate information from multiple
sources. The animals are sensitive to a similar set of low-level spatial-image parameters to
humans (Snowden, Thompson & Troscianko, 2006); namely mean reflectance, contrast,
spatial frequency, orientation, spatial phase and depth, but not colour. The way in
which the animals integrate this low level information to select camouflage then gives
evidence for the existence of some type of higher level internal representation about the
organization of the world into objects.
The next sections of this chapter outline experiments that support such conclusions,
and indicate how findings are important for higher order visual tasks such as object
detection and texture discrimination. We then go on to explore experiments that add to
these basic findings to tackle potentially more cognitive processes.

9.4.1 Camouflage and low-level vision

Unfortunately there are few electrophysiological studies of cephalopod vision (Saidel,
Shashar, Schmolesky & Hanlon, 2005), but we can study vision via body patterns.
The approach is exemplified by Marshall and Messenger’s (1996) demonstration that
cuttlefish are colour blind. They placed the animals on substrates of either uniformly
coloured gravel, gravel with stones varying in achromatic brightness (i.e. black and
206 Sarah Zylinski and Daniel Osorio

white) or gravel with stones whose reflectance spectra differed (they looked blue and
yellow to the human eye) but were predicted to be of equal brightness to the cuttlefish
photoreceptor. The animals behaved as if the latter background was uniform, consistent
with their being colour-blind, a finding that has since been confirmed by a more detailed
study (Mäthger, Barbosa, Miner & Hanlon, 2006).
About five years after Marshall and Messenger’s (1996) work on colour vision, Chiao,
Hanlon and their co-workers began a systematic study of responses to artificial back-
grounds by Sepia officinalis and the tropical species Sepia pharaonis (Chiao & Hanlon,
2001a, 2001b). A key innovation was the use of a simple chequerboard stimulus (see
below and Figure 9.1). The choice seems natural to those who are familiar with the use
of stimuli such as gratings in human psychophysics, but there is a concern that a 2D
chequerboard is such an impoverished stimulus compared to the visual environment of
the shallow sea-floor habitat that the animals would not display meaningful responses.
However, chequerboards have the advantages that they can be viewed from multiple
orientations (see Shohet, Baddeley, Anderson, Kelman & Osorio (2006) for responses
to gratings) and they contain completed ‘objects’ in the form of individual checks. In
fact as this and subsequent studies have shown the animals respond meaningfully to che-
querboards and other simple patterns with well-defined spatial characteristics (Zylinski
& Osorio, 2011).
Chiao and Hanlon (2001a) found that for pharaoh cuttlefish (S. pharaonis), whose
patterns closely resemble those of S. officinalis; both check size and achromatic contrast
affected the body patterns. They characterized the body pattern by scoring the expression
of a limited number of coarse-scale components that contribute to the ‘Disruptive’ body
pattern (Hanlon & Messenger 1988, 1996), of which the most obvious is the so-called
‘White Square’ component, henceforth referred to as WS (see Figure 9.3). In keeping
with what we might expect for camouflage, animals increased the expression of coarse,
higher-contrast components as the checks increased in size and contrast.
In human psychophysics grating thresholds can be related to the modulation trans-
fer function (MTF), which is a convenient measure of the optical quality of the eye
and of spatial resolution (Campbell & Robson, 1968). What do Chiao and Hanlon’s
(2001a, 2001b) results imply about the acuity and contrast discrimination thresholds of
S. pharaonis? A complication is that patterns used for smaller checks and lower contrast
stimuli are not discoverable by scoring the Disruptive components. Chequerboards of
intermediate sizes (relative to the WS) and contrasts result in the expression of com-
ponents typical of the so-called ‘Mottle’ body pattern (see Figure 9.1), as explored by
Barbosa and co-workers and by us (Barbosa et al., 2008; Zylinski, Osorio & Shohet,
2009b). As with human grating detection (Campbell & Robson, 1968) both contrast
and spatial scale affect visibility, thus very small checks at high contrast are treated as
a Uniform pattern, and elicit the Uniform body pattern, with the contrast thresholds
falling as dimensions increase, up to a maximum and low contrasts (Zylinski, Osorio &
Shohet, 2009b).
So can we say that these fine spatial and low contrast patterns, which elicit a Uniform
body pattern, are below the visual acuity and contrast discrimination thresholds of
the cuttlefish? Well possibly, but not conclusively, because this might be reflecting the
 Cuttlefish camouflage: vision and cognition 207

animals decision about the best choice of pattern rather than a psychophysical threshold
imposed by the eye or low-level neural mechanisms. In other words, the cuttlefish might
be able to see that a fine chequerboard stimulus is made up of individual small checks,
and the most effective match to this within its body pattern repertoire is the Uniform
pattern. We return to these experiments in the account of edge detection below.

9.4.2 Edge detection and spatial phase

Whereas detection thresholds for spatial patterns – whether gratings, chequerboards
or opticians test-charts – can generally be modelled by treating the visual pathway as
a linear spatial system (e.g. as a set of spatial filters), the detection of local features
such as motion and edges requires a non-linear mechanism (Bruce, Green & George-
son, 1996). Edges and objects turn out to be of considerable significance to cuttlefish
camouflage.
The experiments described above involved chequerboards with sharply defined edges
separating adjacent checks. However, Barbosa et al. (2008) found that varying check
contrast affects the choice of body pattern even for stimuli that cuttlefish can easily detect.
They found that reducing contrast did not cause the animals to resort to the Uniform
body pattern, as for a uniform background, but rather they display a Mottle pattern.
As contrast differences between objects decrease, edges become less distinct (the step-
difference becomes less) so these results suggest that object edginess might be important
in the animal’s choice between Disruptive and Mottle body patterns. Chiao, Kelman and
Hanlon (2005) explored a related question using manipulated photographs of a gravel
substrate that elicited the expression of Disruptive components. They found that, while
an un-manipulated photograph of the gravel resulted in the same levels of expression in
the measured components as the real gravel, when they removed the edge information
in the image using a low-pass (Gaussian) filter the expression of these components was
significantly reduced, even though information pertaining to object area and contrast
was retained. Mäthger et al. (2007) demonstrated the same phenomenon using real
substrates: by filling in the spaces around pebbles that elicited a Disruptive body pattern
with sand, thereby reducing the strength of the edge information (as demonstrated using
a computer-based edge detector), they showed that the animals reduced the expression
of Disruptive components.
Any spatial pattern can be completely represented by a Fourier transform, which
includes both spatial frequency and spatial phase information. Kelman, Baddeley,
Shohet & Osorio (2007) tested phase sensitivity by comparing responses to conven-
tional chequerboard patterns with those to patterns with identical spatial power-spectra
but randomized phase. These phase-randomized chequerboards look like normal che-
querboards without clearly defined edges. This treatment, like reduction of contrast,
essentially caused the animals to switch from the Disruptive components they had used
on the original chequerboards to a Mottle. It seems that cuttlefish, like humans, are sen-
sitive to phase. All of these findings point to the existence of specialized edge detectors
in cuttlefish vision, with the presence of edges favouring the expression of Disruptive
components. Edges are of course a characteristic of object boundaries. By comparison
208 Sarah Zylinski and Daniel Osorio

patterns lacking edges but with an otherwise comparable spatial frequency power-
spectrum favour Mottle body pattern components.
Can the relationship between the animal’s background and the pattern that is displayed
be summarized by the simple set of relations:
Objects → Disruptive, Texture → Mottle, Uniform → Uniform,
and can it be more rigorously defined? We have proposed (Zylinski, Osorio & Shohet,
2009b) that, while Disruptive patterns are used in response to edgy, low-contrast stimuli,
Mottle patterns are used where objects are small or edges are poorly defined (either
through the visual attributes of the objects or the physiological limitations of the animal’s
visual system). We suggested a model based on the ability or inability of the cuttlefish
to distinguish higher Fourier components within a given stimulus. Edges occur where
sinusoidal components of different frequencies are in phase at their zero-crossings
(Morrone & Burr, 1988). In a two-dimensional (2D) Fourier transform a square wave
can be approximated by a low-frequency fundamental sinusoid with the addition of
the third and fifth harmonics – these are in phase with the fundamental wave –and the
more sinusoidal terms added the more representative of the original signal it becomes.
According to this MTF + minimal edges model, if a stimulus is of high enough spatial
frequency that only the fundamental component is resolvable, or if the edge information
is not well defined so that higher harmonics are not in phase with the fundamental then
the cuttlefish will respond with a Mottle-type pattern. If, however, the stimulus contains
spatial frequency information low enough for higher harmonics to be resolved and
there is phase congruency between the fundamental and (at least) the third harmonic,
characterizing edges, then the animal will respond with Disruptive-type body patterns.
We demonstrated that, by independently modulating the spatial characteristics and the
contrast of checks in a chequerboard background and analyzing the resulting body-
pattern responses of cuttlefish, thresholds between body patterns could be distinguished
that were consistent with this model.

9.4.3 Objects and object completion

Human vision has powerful mechanisms for separating objects from their background
that combine both local and more global cues (Bruce, Green & Georgeson, 1996;
Zipser, Lamme & Schiller, 1996). The underlying mechanisms, especially where they
implicate long-range interactions, have attracted considerable attention in work on visual
processing in the cortex (Lamme & Roelfsema, 2000; Zipser, Lamme & Schiller, 1996);
for example, the studies of illusory contours in visual area 2 of primates by Peterhans
and co-workers (e.g. Peterhans & von der Heydt, 1989, 1991). It is, therefore, interesting
to ask how cuttlefish use local visual cues in figure-ground segregation.
Chiao and Hanlon (2001a, 2001b) suggested that the overall area of objects in the
visual environment is important in determining the cuttlefish’s body pattern. How might
the animals measure object area? As previously mentioned, human vision can identify
step edges that indicate object boundaries by local edge detectors (Marr & Hildreth, 1980;
Morrone & Burr, 1988). In a series of experiments designed specifically to investigate
 Cuttlefish camouflage: vision and cognition 209

the importance of edges in determining camouflage body patterns (Zylinski, Osorio &
Shohet, 2009c), we first asked: if the animals express a given body pattern in response
to a whole object (in this case a 2D white circle on a grey background), will just the
edges of these same objects provide sufficient information for the animals to respond
in the same way? The approach was to produce images where regions where defined by
edges but with no mean intensity difference between the figure and the ground; this is
done simply by high-pass filtering an image (Figure 9.4). It turns out that the responses
of the animals to the whole objects and to the high-pass outlines of the circles were not
statistically separable. By comparison if the contrast of a figure (such as light circle on
a dark background) is reduced, the animals suppress Disruptive components and use a
Uniform body pattern.
What does the finding that outlines appear to substitute for whole objects say about
the importance of object area as a cue for determining the animals’ choice of camouflage
pattern? An obvious interpretation is that, in the real world, objects without area do not
occur: edges are indicative of the presence of objects, which is exactly why they are
important. Bearing this in mind, we wondered if the presence of isolated edge segments,
or fragments, might be sufficient for the animal to infer the presence of an associated
object (also known as contour completion). Our rationale was that, in the real world,
objects that the cuttlefish wants to use in its decision about camouflage patterns might
be partially occluded by other objects in the three-dimensional (3D) environment, or
perhaps partially buried in finer substrate (e.g. white pebbles partly hidden from the
animal’s view by seaweed, or partly buried in darker sand), and correctly interpreting
these visual ambiguities may be the difference between being well camouflaged, and,
hence, protected from visual predation, and being conspicuous.
We tested half-circle, quarter-circle and eighth-circle high-pass filtered edge frag-
ments and found that indeed cuttlefish interpreted the presence of edges as evidence
for the presence of corresponding objects, as determined by the animal’s use of Disrup-
tive components being indistinguishable from those used in response to whole circles
(Figure 9.5a,b). Strikingly this was the case even for quarter-circle edge fragments, in
which the white area of each fragment was less than one would predict was needed
to elicit such as response (Figure 9.5c). Clearly the animal is choosing camouflage by
inferring the size of the objects in the background rather than direct measurement. The
extrapolation of object presence ceased when edge fragments were further reduced: the
animals responses were in keeping with those expected for individual small objects,
using Mottle-type body patterns, when edge fragments were eighth-circles.
We see from the experiments outlined above that, as in vertebrate and machine vision,
edges are very important to cuttlefish. Not only are they using edges to gain information
about objects in their environment, but also, additionally, they are translating this
information back into their body patterns in order to achieve effective camouflage.

9.4.4 Contour completion

The finding that Sepia can use isolated edge segments to infer the presence of whole
objects (Zylinski, Osorio & Shohet, 2009c) suggests a way to test a distinct set of visual
210 Sarah Zylinski and Daniel Osorio

PC 2

PC 1

Figure 9.4 Example of how scores from two principal components (PC 1 and PC 2) can be used to
visualize how different stimuli influence body patterning beyond simple classifications. The
example here is based on the experiment described in section 9.3.3, using high-pass filtered
‘objects’ and edge fragments to explore edge detection mechanisms in S. officinalis (see
Zylinski, Osorio & Shohet, 2009c for futher detail). Different patterned and shaded areas show
the range of PC scores for 20 cuttlefish responses to the stimuli shown below (presented as
multiple printed objects across the floor and walls of the test arena) and are presented as such to
allow the reader to easily see the separation and overlap in the responses to the various stimuli.
Cuttlefish images on the axes give examples of typical responses at extremes. For example,
high-contrast white objects on a black background (open circle) result in responses characterized
by low PC 1 scores and high PC 2 scores, equating to a Disruptive-type body-pattern response.
By contrast, quarter-edge segments (black-tinted circle) result in high PC 1 scores and low PC 2
scores, equating to more Mottle-type responses. Low PC 1 and PC 2 scores equate to a Uniform
body pattern, as seen in response to a homogenous grey stimulus (dark-grey circle).
 Cuttlefish camouflage: vision and cognition 211

(a) (b) (c) (d) (e) (f )

Figure 9.5 (a) Cuttlefish respond with body patterns containing blocky, high contrast Disruptive
components in response to white ‘objects’ (two-dimensional printed filled-in circles on a
mid-grey background) of an area 40–120% of the white square (WS) component. (b) The same
response is given to the outline of the circle alone. (c) Animals also give the same response when
edge fragments down to ¼ of the circle are presented. (d) When edges are reduced further the
animals change from the blocky Disruptive response and instead use a Mottle-type response. (e)
When edge fragments are clustered but anomalously configured the same Mottle-type response
is used. (f) When the same fragments are shown in the correct configuration then the cuttlefish
once again use Disruptive components. Adapted from Zylinski, Darmaillacq and Shashar (2012).

phenomena associated with the reconstruction of fragmented information of which best

known are modal and amodal completion. Modal completion refers to that which occurs
when a foreground object is viewed against a background with predominantly similar
intensity properties, such as the classic Kanisza Triangle (Pessoa, Thompson & Noë,
1998). In this case boundaries of the object are completed by the percept of illusory
contours accompanied by a concurrent brightness enhancement. Amodal completion
refers to the situation when a foreground figure appears to partially occlude a target
figure, but the visual system is nevertheless able to perceive the object in its entirety by
linking the visually fragmented parts of the target, completing it without a sensation of
the occurrence of the borders (Halko, Mingolla & Somers, 2008; Pessoa, Thompson &
Noë, 1998; Peterhans & von der Heydt, 1989).
Based on the knowledge gained from the edge detection experiment described above
(Zylinski, Osorio & Shohet, 2009c), Zylinski, Darmaillacq and Shashar (2012) took
the outline (edges) of an ‘object’ (a white circle) to which S. officinalis responded with
Disruptive components (Figure 9.5b) and disassembled it into edge fragments smaller
than those from which the animals infer the presence of larger objects (Figure 9.5d).
In other words, these fragments resulted in the use of a Mottle- or Uniform-type body
pattern when randomly scattered across a mid-grey background (printed 2D stimuli;
Figure 9.5d). However, when these same fragments were presented in a configuration
that was in keeping with the outline of the original circle, the animals responded with
Disruptive components (Figure 9.5f). In other words, the animals were able to recon-
struct the fragmented visual information and respond as if the complete object was
present when these fragment edges were configured thus. To ensure this response was
not a response to the clustering of high contrast objects (see section 9.3.5) the same
212 Sarah Zylinski and Daniel Osorio

fragments were tested in the same global positions but in an anomalous configuration:
the fragments were rotated on their axes. The cuttlefish responded to this stimulus
with non-Disruptive body patterns similar to the responses to the scattered fragments
(Figure 9.5e). This strengthens the interpretation of these results as a demonstration of
boundary completion.

9.4.5 Texture
We can consider visual texture a property of ‘stuff ’ in a scene, in contradistinction with
‘things’ such as lines and objects (Adelson & Bergen, 1991; Landy & Graham, 2004).
This might be analogous to the linguistic differences between the mass noun ‘water’
and the count noun ‘cuttlefish’. Another analogy is the separation of sketchable and
non-sketchable image regions, where non-sketchable is equivalent to texture.
Often boundaries between image regions are not formed by distinct intensity changes,
represented by step edges, but are instead separated by textural differences that have
no associated intensity differences. What makes one texture easily distinguishable from
another is not straightforward and is incompletely understood, even in human vision.
Texture discrimination has been the subject of much research for human and machine
vision (Julesz, 1981; Julesz & Schumer, 1981; Portilla & Simoncelli, 2000). In cuttlefish
visual perception and camouflage it seems that the Mottle body pattern is used largely in
response to image texture, just as the use of Disruptive components seems to be strongly
correlated with the presence of objects (within a given size range) with distinct edges.
Indeed, the pattern components themselves are much more ‘textural’ than the blocky,
distinct Disruptive components.
Chiao, Chubb, Buresch and Siemann (2009) used a set of well-defined artificial sub-
strates to take a closer look at the interaction between textural background features and
body-pattern responses. They used a random noise matrix, filtered at various spatial
scales, to keep the overall characteristics the same but changed the scale of the back-
ground attributes. They showed that, as with the closed objects created by chequerboards,
these random shapes (which still contained strong edge information) evoked a transition
from Uniform- to Mottle- to Disruptive-type patterns with increased scale (or relative
area of black and white in the background). Interestingly, they noted that some of the
backgrounds that were expected to elicit a Mottle response also contained Disruptive
components. They suggested that spatial scale is more important than defined objects
in determining which body pattern will be used, but given that in real visual scenes
substrates containing such defined edges and high contrast components will usually be
associated with distinct objects, it is not clear to what extent scale and object can be
decoupled. We suspect, based on the findings outlined above, that cuttlefish interpolate
and extrapolate information; the final outcome of the body pattern is likely based on
the integration of multiple cues of which scale is just one. For example, these random
noise patterns retain the equal ratios of black and white seen in chequerboards; although
it becomes harder to define, design and interpret more complex patterns, these seem to
be necessary to unravel the complete story of visual cues in cuttlefish camouflage.
We carried out experiments to attempt to explicitly test the responses of juvenile S.
officinalis to objects defined by local texture rather than by overall luminance (Zylinski,
 Cuttlefish camouflage: vision and cognition 213

PC 2

PC 1

Figure 9.6 Plotting scores of individual cuttlefish for two principal components (PC 1 and PC 2)
enables us to visualize how the animals use textural information. Patterned and tinted areas show
the range of responses for individual cuttlefish to the stimuli tested, as given below. Cuttlefish
images show the general body pattern characteristics associated with the high/low scores on each
axis. We see from this that the body-pattern response to objects formed as textural ‘clumps’ of
chequerboard (white circle) is distinct from that of continuous chequerboard (light grey circle)
and untextured objects (dark grey circle). Unclumped individual textural components (dotted
circle) result in a similar, but not identical, response as the continuous chequerboard.

Osorio & Shohet, 2009c). To do this we designed a mid-grey printed background

scattered randomly with ‘objects’ made up of circular patches of small black and white
checks (Figure 9.6). This meant the overall luminance of the objects was identical
to that of the background, but that the objects, unlike the background, had a distinct
texture. The size of the patch circles was predetermined to elicit a Disruptive-type
response when displayed as white objects on a mid-grey background. We also assayed
the body-pattern responses of cuttlefish to these small checks configured as a continuous
214 Sarah Zylinski and Daniel Osorio

chequerboard, and as individual checks scattered across the grey background rather
than clustered into objects (Figure 9.6). Interestingly, we found that these latter two
stimuli resulted in a similar (but not identical) Mottle-type response, with the scattered
individual squares resulting in a much stronger Mottle than the continuous chequer-
board.
Contrary to expectations based on the responses to the latter two stimuli described
above, the main test stimulus (patches of objects) resulted in responses consistently
containing high-contrast Disruptive components (PC 2 in this study), such as the White
Square and White Head Bar. In other words, the animals used a significantly different
response to the same small black and white squares depending on whether they were
scattered across the background or clustered into objects, suggesting perceptual group-
ing. However, the animals appeared to integrate aspects of the textural nature of the
patches, and expressed Mottle-type components (PC 1 in this study) in addition to the
Disruptive components. Such results can be compared to the responses to both white
circles and a mixture of black and white circles, which scored high on PC 2 but low
on PC 1 (i.e. similar strength of Disruptive components but fewer Mottle components).
These findings show that: (a) the cuttlefish cue in on global attributes of the visual
environment, not just absolute local features; and (b) texture, distinct from objects, lines
and features, drive Mottle-type body patterns and associated components.
The responses of animals to some stimuli tested by Chiao, Chubb and Hanlon (2007)
point to similar findings, but not explicitly so. For example, one experiment tested
something akin to the opposite to our test: they tested cuttlefish body-pattern responses
to square white ‘objects’ on a background composed of smaller textural components.
Although in their paper they interpret this in terms of size predominating over contrast, an
alternative interpretation would be that reducing the integrity of the edge information (in
terms of contrast difference) between the large white figure and the small chequerboard
background resulted in the suppression of Disruptive components. This is in keeping
with our findings that the loss of edge information results in the animal responding with
Mottle or Uniform patterns, even in the presence of objects typically associated with
Disruptive type patterns.

9.5 Making decisions about body patterns

Do cuttlefish make active choices about camouflage? Do they show preference for back-
grounds or substrates over others, and is this preference based on the visual attributes?
When faced with a background containing multiple options for camouflage, how do the
animals solve the dilemma of what they should camouflage to? Laboratory and field
studies have tackled these questions in several species.
Zylinski, How, Osorio, Hanlon and Marshall (2011) analysed the characteristics of
body patterns and backgrounds of images of the Australian giant cuttlefish, Sepia apama,
in the near-shore benthic habitats that form their breeding grounds. Here, males embark
in conspicuous agonistic kinetic displays (i.e. moving stripes) with other males and direct
similar displays to females (Figure 9.7a). When not involved in mating behaviours, both
 (a)

(b)

Figure 9.7 The giant Australian cuttlefish, Sepia apama, chooses simple backgrounds to display
against, which probably acts to increase signal efficacy, and more complex backgrounds to
camouflage against making visual search more difficult for potential predators. (a) A male
engaged in an agonistic display to another male. (b) A camouflaging individual uses chromatic
and textural components. (c) Plot of intensity variances (a measure of visual complexity) from
image analysis of the mantle patterns of signalling and camouflaging animals compared
background areas against which they were photographed, along with those for a random set of
habitats (control). Adapted from Zylinski, How, Osorio, Hanlon and Marshall (2011). See plate
section for colour version of (a) and (b).
216 Sarah Zylinski and Daniel Osorio

Camouflage Signalling Control

0.06

0.05
Pixel intensity variance

0.04

0.03

0.02

0.01

Animals Backgrounds Animals Backgrounds

(c)

Figure 9.7 (cont.)

sexes use camouflaged body patterns (Figure 9.7b). The study clearly showed that
there were differences not just between the spatial characteristics of the body patterns
an animal used during these different behaviours, but also in the visual characteris-
tics of the backgrounds used when camouflaging compared to when signalling. Back-
grounds used for camouflage typically have a significantly higher variance in inten-
sity, suggesting that they were more visually complex, while backgrounds used for
signalling against were more homogenous (Figure 9.7c). This suggests that the cut-
tlefish actively choose a simple background reducing noise to improve signal efficacy
and, conversely, a more visually complex area for camouflage in order to improve its
effectiveness.
Allen et al. (2010) investigated background choice and camouflage pattern use in S.
officinalis in a series of laboratory-based experiments. These experiments focused on
camouflage pattern choice in the presence of multiple visual background types. They
found that, within the small range of artificial and natural backgrounds used and the
relatively small area within which they were tested, when given a choice of substrates
designed to elicit different classes of body patterns the animals did not show a preference
to settle on a specific substrate (suggesting they have no preference for using a particular
class of body pattern). The only significant finding with regards to choice was that, in
the case of loose natural substrates, the cuttlefish showed a preference for substrates
into which they could bury over those substrates they could not. Buresch et al. (2011)
found that S. officinalis made choices as to which aspect of the visual environment
to match in order to obtain camouflage. They tested the body-pattern responses of
cuttlefish to 3D objects on a 2D substrate at varying levels of contrast. They found that
3D objects were salient for camouflage when they were of a higher contrast than the
substrate.
 Cuttlefish camouflage: vision and cognition 217

LOW LEVEL?

Edges
HIGHER LEVEL?
Knowledge

Texture
Area INTEGRATION

Shading
Contrast
?

Scale

Figure 9.8 A schematic diagram of how cuttlefish might integrate putative object-based low and
higher level (bottom-up) and viewer-based scene knowledge (top-down) information about the
local visual environment in order to make decisions about what body pattern to use.

9.6 Concluding remarks

Here we have discussed just those aspects of cuttlefish vision and camouflage for which
we have a reasonable level of understanding. Little is yet known about the mechanisms
used by cuttlefish in other aspects of vision that are crucial in vertebrate vision, such as
the processing of motion (Zylinski, Osorio & Shohet, 2009a) and the perception of depth
(Allen, Mäthger, Barbosa & Hanlon, 2009; Barbosa, Allen, Mäthger & Hanlon, 2012;
Buresch et al., 2011). Much evidence points to the conclusion that the body patterns used
by cuttlefish for camouflage are not hard-wired responses to specific cues (Figure 9.8).
By following the input to eye, to optic lobe, to brain, to the chromatophore output, we
encounter cognitive processes.
Cephalopods have large brains that have evolved independently of vertebrates and it
is, therefore, interesting to ask to what extent they share similar principles of operation.
At one level this can concern relatively low-level phenomena, such as the evidence that
cuttlefish have local edge detectors that resemble those thought to be used by humans
(Kelman, Baddeley, Shohet & Osorio, 2007; Kelman, Osorio & Baddeley, 2008; Morrone
& Burr, 1988; Zylinski, Osorio & Shohet, 2009b). On the other hand, cephalopods mostly
lack colour vision (Marshall & Messenger, 1996; Mäthger, Barbosa, Miner & Hanlon,
2006), but make much greater use of polarization (Temple et al., 2012), giving them
218 Sarah Zylinski and Daniel Osorio

a different phenomenal world from ours; we have hardly begun to understand their
cognitive worlds.

Acknowledgements

We would like to acknowledge support from Office of Naval Research MURI grant no.
N00014–09–1–1053 to S. Z.

References

Adelson, E. and Bergen, J. (1991). The plenoptic function and elements of early vision. In
Computational models of visual processing M. Landy and J. A. Movshon (Eds.). Cambridge,
MA: MIT Press.
Allen, J. J., Mäthger, L. M., Barbosa, A., Buresch, K. C., Sogin, E., Schwartz, J., Chubb, C.
and Hanlon, R. T. (2010). Cuttlefish dynamic camouflage: responses to substrate choice and
integration of multiple visual cues. Proceedings of the Royal Society. B: Biological Sciences,
277(1684): 1031–1039.
Allen, J. J., Mäthger, L. M., Barbosa, A. and Hanlon, R. T. (2009). Cuttlefish use visual cues to
control three-dimensional skin papillae for camouflage. Journal of Comparative Physiology A,
195(6): 547–555.
Barbosa, A., Allen, J. J., Mäthger, L. M. and Hanlon, R. T. (2012). Cuttlefish use visual cues
to determine arm postures for camouflage. Proceedings of the Royal Society. B: Biological
Sciences, 279: 84–90.
Barbosa, A., Mäthger, L. M., Buresch, K. C., Kelly, J., Chubb, C., Chiao, C.-C. and Hanlon, R.
T. (2008). Cuttlefish camouflage: the effects of substrate contrast and size in evoking Uniform,
Mottle or Disruptive body patterns. Vision Research, 48(10): 1242–1253.
Bennett, A. T. (1996). Do animals have cognitive maps? Journal of Experimental Biology, 199(1):
219–224.
Brooks, R. (1991). Intelligence without representation. Artificial Intelligence, 47: 139–159.
Bruce, V., Green, P. R. and Georgeson, M. A. (1996). Visual perception: physiology, psychology
and ecology (3rd edn). Hove, UK: Psychology Press.
Budelmann, B. U. (1994). Cephalopod sense organs, nerves and the brain: adaptations for high
performance and life style. Marine and Freshwater Behaviour and Physiology, 25(1–3): 13–33.
Buresch, K., Mäthger, L., Allen, J., Bennice, C., Smith, N., Schram, J., Chiao, C.-C., Chubb, C.
and Hanlon, R. (2011). The use of background matching vs. masquerade for camouflage in
cuttlefish Sepia officinalis. Vision Research, 51(23–24): 2362–2368.
Burgess, N. (2006). Spatial memory: how egocentric and allocentric combine. Trends in Cognitive
Sciences, 10(12): 551–557.
Byrne, R. W. and Bates, L. A. (2006). Why are animals cognitive? Current Biology, 16(12):
R445–448.
Campbell, F. W. and Robson, J. G. (1968). Application of Fourier analysis to the visibility of
gratings. Journal of Physiology, 197: 551–566.
Cavanagh, P. (2011). Visual cognition. Vision Research, 51: 408–416.
Chiao, C.-C., Chubb, C., Buresch, K. C., Barbosa, A., Allen, J. J., Mäthger, L. M. and Hanlon,
R. T. (2010). Mottle camouflage patterns in cuttlefish: quantitative characterization and
 Cuttlefish camouflage: vision and cognition 219

visual background stimuli that evoke them. Journal of Experimental Biology, 213(2): 187–
199.
Chiao, C.-C., Chubb, C., Buresch, K. C. and Siemann, L. (2009). The scaling effects of substrate
texture on camouflage patterning in cuttlefish. Vision Research, 49: 1647–1656.
Chiao, C.-C., Chubb, C. and Hanlon, R. T. (2007). Interactive effects of size, contrast, intensity
and configuration of background objects in evoking Disruptive camouflage in cuttlefish. Vision
Research, 47: 2223–2235.
Chiao, C.-C. and Hanlon, R. T. (2001a). Cuttlefish camouflage: visual perception of size, contrast
and number of white squares on artificial chequerboard substrata initiates Disruptive coloration.
Journal of Experimental Biology, 204(12): 2119–2125.
Chiao, C.-C. and Hanlon, R. T. (2001b). Cuttlefish cue visually on area- not shape or aspect
ratio -of light objects in the substrate to produce Disruptive body patterns for camouflage. The
Biological Bulletin, 201(2): 269–270.
Chiao, C.-C., Kelman, E. J. and Hanlon, R. T. (2005). Disruptive body patterning of cuttlefish
(Sepia officinalis) requires visual information regarding edges and contrast of objects in natural
substrate backgrounds. The Biological Bulletin, 208(1): 7–11.
Collett, T. S. and Collett, M. (2002). Memory use in insect visual navigation. Nature reviews.
Neuroscience, 3(7): 542–552.
Cott, H. B. (1940). Adaptive colouration in animals. London, UK: Methuen Publishing.
Crook, A. C., Baddeley, R. and Osorio, D. (2002). Identifying the structure in cuttlefish visual
signals. Philosophical Transactions of the Royal Society B, 357(1427): 1617–1624.
Gibson, J. J. (1979). The ecological approach to visual perception. Boston, MA: Houghton Mifflin.
Gregory, R. L. (1997). Knowledge in perception and illusion. Philosophical Transactions of the
Royal Society B, 352(1358): 1121–1127.
Halko, M. A., Mingolla, E. and Somers, D. C. (2008). Multiple mechanisms of illusory contour
perception. Journal of Vision, 8(11): 1–17.
Hanlon, R. T. (2007). Cephalopod dynamic camouflage. Current Biology, 17(11): R400–R404.
Hanlon, R. T., Forsythe, J. W. and Joneschild, D. E. (1999). Crypsis, conspicuousness, mimicry and
polyphenism as antipredator defences of foraging octopuses on Indo-Pacific coral reefs, with a
method of quantifying crypsis from video tapes. Biological Journal of the Linnean Society, 66:
1–22.
Hanlon, R. T. and Messenger, J. B. (1988). Adaptive coloration in young cuttlefish (Sepia offic-
inalis L) – the morphology and development of body patterns and their relation to behavior.
Philosophical Transactions of the Royal Society B, 320(1200): 437–487.
Hanlon, R. T. and Messenger, J. B. (1996). Cephalopod behaviour. New York, NY: Cambridge
University Press.
Jolliffe, I. T. (1986). Principal component analysis. New York, NY: Springer-Verlag.
Julesz, B. (1981). Textons, the elements of texture perception, and their interactions. Nature,
290(5802): 91–97.
Julesz, B. and Schumer, R. A. (1981). Early visual perception. Annual Review of Psychology, 32:
575–627.
Kelman, E. J., Baddeley, R. J., Shohet, A. J. and Osorio, D. (2007). Perception of visual texture
and the expression of Disruptive camouflage by the cuttlefish Sepia officinalis. Proceedings of
the Royal Society. B: Biological Sciences, 274: 1369–1375.
Kelman, E. J., Osorio, D. and Baddeley, R. J. (2008). A review of cuttlefish camouflage and object
recognition and evidence for depth perception. Journal of Experimental Biology, 211(11):
1757–1763.
220 Sarah Zylinski and Daniel Osorio

Kelman, E. J., Tiptus, P. and Osorio, D. (2006). Juvenile plaice (Pleuronectes platessa) produce
camouflage by flexibly combining two separate patterns. Journal of Experimental Biology, 209:
3288–3292.
Krapp, H. G. and Hengstenberg, R. (1997). Estimation of self-motion by optic flow processing in
single visual interneurons. Nature, 384: 463–466.
Lamme, V. A. F. and Roelfsema, P. R. (2000). The distinct modes of vision offered by feedforward
and recurrent processing. Trends in Neurosciences, 23: 571–579.
Land, M. F. and Nilsson, D.-E. (2002). Animal eyes. Oxford, UK: Oxford University Press.
Landy, M. S. and Graham, N. (2004). Visual perception of texture. In The visual neurosciences
L. M. Chalupa & J. S. Werner (Eds.). Cambridge, MA: MIT Press.
Langridge, K. V., Broom, M. and Osorio, D. (2007). Selective signalling by cuttlefish to predators.
Current Biology, 17(24): R1044–R1045.
Mackintosh, N. J. (1974). Psychology of animal learning. Oxford, UK: Academic Press.
Manly, B. F. J. (1994). Multivariate statistical methods – a primer (2nd edn). London, UK:
Chapman and Hall.
Marr, D. (1982). Vision: a computational investigation into the human representation and pro-
cessing of visual information. New York, NY: Henry Holt and Co.
Marr, D. and Hildreth, E. (1980). Theory of edge detection. Proceedings of the Royal Society. B:
Biological Sciences, 207(1167): 187–217.
Marshall, N. J. and Messenger, J. B. (1996). Colour-blind camouflage. Nature, 382: 408–409.
Mäthger, L. M., Barbosa, A., Miner, S. and Hanlon, R. T. (2006). Color blindness and contrast
perception in cuttlefish (Sepia officinalis) determined by a visual sensorimotor assay. Vision
Research, 46(11): 1746–1753.
Mäthger, L. M., Chiao, C. C., Barbosa, A., Buresch, K. C., Kaye, S. and Hanlon, R. T. (2007).
Disruptive coloration elicited on controlled natural substrates in cuttlefish, Sepia officinalis.
Journal of Experimental Biology, 210(15): 2657–2666.
Mäthger, L. M. and Hanlon, R. T. (2007). Malleable skin coloration in cephalopods: selective
reflectance, transmission and absorbance of light by chromatophores and iridophores. Cell and
Tissue Research, 329(1): 179–186.
Messenger, J. B. (2001). Cephalopod chromatophores: neurobiology and natural history. Biologi-
cal Reviews, 76: 473–528.
Morrone, M. C. and Burr, D. C. (1988). Feature detection in human vision: a phase-
dependent energy model. Proceedings of the Royal Society. B: Biological Sciences, 235: 221–
245.
Packard, A. (1972). Cephalopods and fish: the limits of convergence. Biological Reviews, 47(2):
241–307.
Packard, A. and Hochberg, F. G. (1977). Skin patterning in Octopus and other genera. Symposium
of the Zoological Society of London, 38: 191–231.
Packard, A. and Sanders, G. D. (1971). Body patterns of Octopus vulgaris and maturation of the
response to disturbance. Animal Behaviour, 19(4): 780–790.
Pessoa, L., Thompson, E. and Noë, A. (1998). Finding out about filling-in: a guide to perceptual
completion for visual science and the philosophy of perception. Behavioral and Brain Sciences,
21(6): 723–802.
Peterhans, E. and von der Heydt, R. (1989). Mechanisms of contour perception in monkey visual
cortex. II. Contours bridging gaps. Journal of Neuroscience, 9(5): 1749–1763.
Peterhans, E. and von der Heydt, R. (1991). Subjective contours-bridging the gap between psy-
chophysics and physiology. Trends in Neurosciences, 14(3): 112–119.
 Cuttlefish camouflage: vision and cognition 221

Portilla, J. and Simoncelli, E. P. (2000). A parametric texture model based on joint statistics of
complex wavelet coefficients. International Journal of Computer Vision, 40(1): 49–71.
Ramachandran, V. S., Tyler, C. W., Gregory, R. L., Rogers-Ramachandran, D., Duensing, S.,
Pillsbury, C. and Ramachandran, C. (1996). Rapid adaptive camouflage in tropical flounders.
Nature, 379: 815–818.
Real, L. A. (1993). Toward a cognitive ecology. Trends in Ecology and Evolution, 8(11): 413–
417.
Saidel, W. M., Shashar, N., Schmolesky, M. T. and Hanlon, R. T. (2005). Discriminative responses
of squid (Loligo pealeii) photoreceptors to polarized light. Comparative Biochemistry and
Physiology A, 142: 340–346.
Shettleworth, S. (2001). Animal cognition and animal behaviour. Animal Behaviour, 61(2): 277–
286.
Shettleworth, S. (2010). Cognition, evolution, and behavior (2nd edn). Oxford, UK: Oxford
University Press.
Shohet, A. J., Baddeley, R. J., Anderson, J. C., Kelman, E. J. and Osorio, D. (2006). Cuttlefish
responses to visual orientation, water flow and a model of motion camouflage. Journal of
Experimental Biology, 209: 4717–4723.
Shohet, A., Baddeley, R. J., Anderson, J. C. and Osorio, D. (2007). Cuttlefish camouflage: a
quantitative study of patterning. Biological Journal of the Linnean Society, 92(2): 335–345.
Snowden, R. J., Thompson, P. and Troscianko, T. (2006). Basic vision: an introduction to visual
perception. Oxford, UK: Oxford University Press.
Srinivasan, M. V. and Venkatesh, S. (1997). Living eyes to seeing machines. Oxford, UK: Oxford
University Press.
Stevens, M. (2007). Predator perception and the interrelation between different forms of protective
coloration. Proceedings of the Royal Society. B: Biological Sciences, 274: 1457–1464.
Stuart-Fox, D. and Moussalli, A. (2009). Camouflage, communication and thermoregulation:
lessons from colour changing organisms. Philosophical Transactions of the Royal Society B,
364(1516): 463–470.
Stuart-Fox, D., Whiting, M. J. and Moussalli, A. (2006). Camouflage and colour change: antipreda-
tor responses to bird and snake predators across multiple populations in a dwarf chameleon.
Biological Journal of the Linnean Society, 88: 437–446.
Temple, S. E., Pignatelli, V., Cook, T., How, M. J., Chiou, T. H., Roberts, N. W. and Marshall,
N. J. (2012). High-resolution polarisation vision in a cuttlefish. Current Biology, 22(4): R121–
R122.
Tolman, E. C. (1948). Cognitive maps in rats and men. Psychological Review, 55: 189–208.
Wang, R. F. and Spelke, E. S. (2002). Human spatial representation: insights from animals. Trends
in Cognitive Sciences, 6(9): 376–382.
Williamson, R. and Chrachri, A. (2004). Cephalopod neural networks. Neurosignals, 13(1–2):
87–98.
Young, R. E., Vecchione, M. and Donovan, D. T. (1998). The evolution of coleoid cephalopods
and their present biodiversity and ecology. South African Journal of Marine Science, 20(1):
393–420.
Zipser, K., Lamme, V. A. F. and Schiller, P. H. (1996). Contextual modulation in primary visual
cortex. Journal of Neuroscience, 16(22): 7376–7389.
Zylinski, S., Darmaillacq, A.-S. and Shashar, N. (2012). Visual interpolation for contour comple-
tion by the European cuttlefish (Sepia officinalis) and its use in dynamic camouflage. Proceed-
ings of the Royal Society. B: Biological Sciences, 279(1737): 2386–2390.
222 Sarah Zylinski and Daniel Osorio

Zylinski, S., How, M. J., Osorio, D., Hanlon, R. T. and Marshall, N. J. (2011). To be seen or to
hide: visual characteristics of body patterns for camouflage and communication in the Australian
giant cuttlefish Sepia apama. The American Naturalist, 177(5): 681–690.
Zylinski, S. and Osorio, D. (2011). What can camouflage tell us about non-human visual percep-
tion? A case study of multiple cue use in the cuttlefish Sepia officinalis. In Animal camouflage:
mechanisms and function M. Stevens & S. Merilaita (Eds.). Cambridge, UK: Cambridge Uni-
versity Press.
Zylinski, S., Osorio, D. and Shohet, A. J. (2009a). Cuttlefish camouflage: context-dependent body
pattern use during motion. Proceedings of the Royal Society. B: Biological Sciences, 276(1675):
3963–3969.
Zylinski, S., Osorio, D. and Shohet, A. J. (2009b). Edge detection and texture classification by
cuttlefish. Journal of Vision, 9(13): 1–10.
Zylinski, S., Osorio, D. and Shohet, A. J. (2009c). Perception of edges and visual texture in the
camouflage of the common cuttlefish, Sepia officinalis. Philosophical Transactions of the Royal
Society B, 364(1516): 439–448.
10 Visual cognition in deep-sea
cephalopods: what we don’t know
and why we don’t know it
Sarah Zylinski and Sönke Johnsen

10.1 The other cephalopods

A quick glance at the recent cephalopod literature, or even at the chapters of this book,
tells us that when we talk about cephalopod cognition we are really considering cognition
in a handful of genera. There can be no argument that studies of these animals have led to
remarkable results that have challenged the traditional view of invertebrate intelligence.
Yet when we consider that less than 10 species of cephalopod are commonly seen as the
focus of behavioural studies, let alone in studies specifically about cognition, it becomes
apparent that claims regarding the cognitive capabilities of cephalopods are generali-
zations drawn from work on a handful of genera. The majority of the 800 or so described
species of cephalopod do not share the neritic and near-shore benthic habitats of the
taxa with which we are most familiar; virtually unknown in terms of their behaviour
and ecology, these species inhabit a different world in the deep, dark waters of the open
ocean (Figure 10.1).
In this chapter, we introduce and discuss the neglected cephalopods of the deep sea,
many of which are not so distantly related to the species with which we are familiar, but
whose existence in the deep sea has little in common with the complex reefs and near-
shore habitats associated with taxa such as Octopus and Sepia. What effect might these
differences in ecology have on the cognitive abilities of deep-sea cephalopods compared
with their shallow water relatives? Chapter 9 of this volume (by Zylinski & Osorio)
explored what we can deduce about cuttlefish visual cognition from the body patterns
they use for camouflage. In this chapter we revisit the theme of making inferences about
visual perceptive and cognitive abilities via body patterning in the handful of epipelagic
and mesopelagic species that have been studied to date. Prior to that, we consider how
the cognitive needs of these animals might differ from those of near-shore or shallow
benthic environments. We explore what is known about the potential for visual cognition
in deep-sea cephalopods and discuss why we know so little. Readers might find this
chapter heavier on natural history and observation than the others in this book, but we
hope that it will serve as a reminder that most cephalopods are essentially unknown
beyond their descriptions.

Cephalopod Cognition, eds A.-S. Darmaillacq, L. Dickel and J. Mather. Published by Cambridge University
Press. 
C Cambridge 2014.
224 Sarah Zylinski and Sönke Johnsen

0 Neustonic
Epipelagic Onychoteuthis

Pterygioteuthis
Cranchia
400 EYES GET
Upper mesopelagic BIGGER
REGION OF
BIOLUMINESCENCE BECOMES COUNTER-
Octopoteuthis ILLUMINATION
THE DOMINANT SOURCE OF LIGHT
800
Japetella
Lower mesopelagic

1200

Bathyteuthis

1600
Bathypelagic EYES GET
SMALLER

2000

2400+
(Benthopelagic) Cirroteuthis (3000m+)

Figure 10.1 Approximate depths in metres of habitat zones for clear oceanic waters.
Benthopelagic realm is close to the ocean floor at any depth of the pelagic realm. Vertical lines
show approximate depth ranges for some of the species referred to in this chapter, taken from
daytime trawl data for adults (Arkhipkin, 1996; Arkhipkin & Nigmatullin, 1997; Roper &
Young, 1975).

10.2 Why do we know so little about the other 95%?

Although notoriously ‘moody’ (Talbot & Marshall, 2010) the coastal cephalopod taxa
with which we are familiar (e.g. Sepia and Octopus) can be hatched and/or kept in
the laboratory with some level of expertise and patience (e.g. Domingues, Dimarco,
Andrade & Lee, 2005; Forsythe, Lee & Walsh, 2002). This has led to the volume of well-
controlled visual behavioural research we have for these animals (e.g. Barbosa, Allen,
Mäthger & Hanlon, 2012; Chiao, Chubb, Buresch, Siemann & Hanlon, 2009; Mäthger,
Chiao, Barbosa & Hanlon, 2008; Shashar, Rutledge & Cronin, 1996; Shohet, Baddeley,
Anderson, Kelman & Osorio, 2006; Zylinski, Osorio & Shohet, 2009), supplemented
by field observations and experiments where data can be obtained with relative ease
with scuba (e.g. Hall & Hanlon, 2002; Zylinski, How, Osorio, Hanlon & Marshall,
2011) or even by snorkelling (e.g. Hanlon, Forsythe & Joneschild, 1999). Behavioural
observations, let alone controlled behavioural experiments, are few and far between for
 Visual cognition in deep-sea cephalopods 225

mesopelagic animals. Collecting animals from mesopelagic depths requires either the
use of specialized trawl nets or undersea vehicles (e.g. a remotely operated vehicle (ROV)
or manned submersible). ROVs and submersibles enable in situ observations of animals,
which have greatly improved our knowledge of interactions, behaviour and physiology
of deep-sea animals, as well as enabling the gentle capture of specimens (Hunt & Seibel,
2000; Robison, 2004; Vecchione & Roper, 1991). However, experimental behavioural
biology requires a rigorous framework beyond these natural history observations, and
there are some concerns about the influence of motor vibrations and viewing lights on
in situ behavioural observations (for example, animals living in perpetual near-darkness
are probably literally and permanently blinded by these lights). The collection of animals
using trawl nets has been aided by the use of thermally-insulated and light-tight closing
cod ends (Childress, Barnes, Quetin & Robison, 1978), which enables the recovery of
specimens in collection-depth temperatures and light conditions, and prevents excessive
mechanical damage. However, in both cases, there is no guarantee that the animals
required for a planned experiment will be caught. Weeks at sea can be spent in frustration
as each trawl returns to the surface without the target taxa for a specific study. Luck and
patience are both critical for the deep-sea biologist!
Trends in data from vertebrates and hymenopteran insects suggest that certain
lifestyles and the need to deal with specific tasks predispose the evolution of com-
plex behaviour. Ecological complexity is generally associated with cognition, be it the
need to forage for cache-specific foods (e.g. Clayton, Dally & Emery, 2007; Hills, 2006),
navigate using cues and landmarks (e.g. Collett & Collett, 2002) or interact with con-
specifics (e.g. Bergman, Beehner, Cheney & Seyfarth, 2003; Grosenick, Clement &
Fernald, 2007). The ecology of shallow-water, benthic species of cephalopods such as
members of Sepia and Octopus is such that we see clear evidence for complexity in their
life histories: octopuses move through complex rocky reef or coral habitats to forage for
food (Forsythe & Hanlon, 1997; Hanlon, Forsythe & Joneschild, 1999) and navigate to
and from their lairs (Mather, 1991); cuttlefish signal to conspecifics when hunting and
during courtship (Hall & Hanlon, 2002; Langridge, Broom & Osorio, 2007; Zylinski,
How, Osorio, Hanlon & Marshall, 2011). Species from both genera are well known for
the complex body patterns they employ for both signalling and camouflage (Hanlon &
Messenger, 1988; Packard & Sanders, 1971).
The mesopelagic habitat offers a different set of ecological requirements: animals
can pass their entire lives without coming in contact with any abiotic structure, with
hundreds (if not thousands) of metres of water above and below them. Underwater
light decreases exponentially with depth, and by 150 m in clear, oceanic water more
than 99% of surface light has been scattered or absorbed (Jerlov, 1976) (this said, it
should be noted that this can result in surprisingly bright conditions in which the human
eye can still function perfectly well). Many species live at greater depths where the
amount of downwelling light is so small that it is no longer useful for vision, and
here biological light (bioluminescence) is used in many capacities such as signalling,
hunting and camouflage (e.g. silhouette reduction via counterillumination) (Haddock,
Moline & Case, 2010). Temperatures here tend to be cold (typically between 0 and
6°C) and constant due to the lack of mixing with surface waters (Robison, 2004). The
226 Sarah Zylinski and Sönke Johnsen

Figure 10.2 The lower-mesopelagic squid Histeoteuthis has asymmetrical eyes and optic lobes,
with a large left eye for looking up for passing silhouettes of potential prey against the weak
downwelling light, and a small right eye orientated downwards, probably specialized for
detecting point-source bioluminescence. See plate section for colour version.

mesopelagic habitat is virtually globally continuous, with the same species found in
similar conditions around the world, with absolute depths changing with corresponding
differences in temperature, oxygen minimum zone depths and light levels.
In the vast three-dimensional habitat that is the deep-sea pelagic realm, the diversity
and quantity of animals decrease in parallel with depth. It is renowned for being a hard
place to find a mate, and many non-broadcast spawning deep-sea taxa have evolved
unusual reproductive methods, such as male parasitism of females in deep-sea angler-
fish (Marshall, 1958). Cephalopods also appear to use some quirky tactics to maximize
the chances of reproduction when solitary individuals meet. For example, males of the
mesopelagic squid Octopoteuthis apparently mate (implant sperm packages) indiscrim-
inately when they encounter both males and females of the same species, presumably
because sex differences are hard to determine in low-light conditions and the cost of
failing to mate given the opportunity is high (Hoving, Bush & Robison, 2012).
There is also a trend for relative eye size of many taxa to increase as light gets
dimmer from the surface waters to the lower-mesopelagic realm, where the downwelling
surface light can still be useful if the eye is sensitive enough (Figure 10.1). In the
bathypelagic zone (below 1000 m) the trend reverses, and eye size tends to decrease
as ambient light becomes not detectable. In general, eyes in the mesopelagic realm
are either designed to maximize the possible light capture and make use of the small
amounts of light available at the cost of reducing spatial acuity, or they are optimized
to detect point sources (to enable them to see flashes of bioluminescence) at the cost
of reduced sensitivity to the ambient light (Warrant & Locket, 2004). Aside from such
general trends, evolution has found varied and novel ways to obtain visual information
in dim light, and the cephalopods offer some of the most interesting of these. Take,
for example, the deep mesopelagic squid Histioteuthis (Figure 10.2), which has highly
 Visual cognition in deep-sea cephalopods 227

asymmetrical eyes and correspondingly asymmetrical optic lobes (Maddock & Young,
1987; Wentworth & Muntz, 1989). This animal has a large left eye, probably specialized
for gaining information from the dim downwelling light as the animal swims on its side
with this large eye facing upwards. Meanwhile, the smaller, downward-facing right eye
is probably specialized for detecting point-source bioluminescence in the darker waters
below.

10.3 Do mesopelagic cephalopods need to be visually cognitive?

Is there evidence to suggest that they are?

Much emphasis has been placed on the large brains and well-developed sense organs in
cephalopods as adaptations for their ‘high performance’ lifestyles (Budelmann, 1994,
1995). They are typically viewed as fast animals moving in complex environments,
actively predating and evading predation, and often loosely social. However, as with
many deep-sea taxa, the metabolic rate of deep-sea cephalopods is greatly reduced
compared to their shallow water counterparts, likely a response to relaxed pressure from
visual predation (Childress, 1995; Seibel & Carlini, 2001; Seibel, Thuesen, Childress
& Gorodezky, 1997). Large brains are metabolically costly to maintain, and there is
evidence in mammals that brain size will be reduced where lifestyle allows for it (Safi,
Seid & Dechmann, 2005).
Godfrey-Smith (2002) described aspects of environmental complexity as an attempt
to give a general functional explanation of cognition. He used the term cognition to
describe ‘a collection of capabilities which, in combination, allow organisms to achieve
certain kinds of coordination between their actions and the world’ that ‘has the function
of making possible patterns of behaviour which enable organisms to effectively deal
with complex patterns and conditions in their environments’. In their consideration of
body pattern richness and habitat complexity (see below), Hanlon and Messenger (1996)
describe the pelagic realm as a uniform habitat, and it seems reasonable to accept that the
open ocean contains less structural complexity than a near-shore reef habitat. However,
if we use habitat to include the wider physical and biological characteristics of a species’
interactions, then we must consider what habitat complexity means on an individual
species basis. Given the characteristics of the mesopelagic realm, is there sufficient
selective pressure for the evolution or maintenance of higher-level visual processes?
Complex behaviours and corresponding large brains tend to evolve in response to
the selective pressure of complex ecologies, including trophic and social interactions
(Lefebvre & Sol, 2008). Brain areas that receive direct input from sensory systems or
are responsible for complex cognitive function are under strong selective pressure, with
neural development often reflecting these behavioural adaptations and sensory special-
izations. For example, in primates both an increasing reliance on frugivory (indicating
complex foraging behaviour) and social group size (indicating social complexity) are
independently positively correlated with the size of the neocortex brain region. Research
from multiple vertebrate groups, for example primates (Barton & Harvey, 2000), cichlid
fish (Pollen et al., 2007) and bats (Safi & Dechmann, 2005), suggest that the cognitive
228 Sarah Zylinski and Sönke Johnsen

centres in the brain are under specific pressure according to mosaic theory, whereby
localized changes in functionally distinct regions are mediated by selection on a specific
set of behaviours according to a given species’ ecology (Barton & Harvey, 2000). In a
comparative analysis controlling for phylogeny, bat species foraging in complex habi-
tats, which must distinguish prey from background noise whilst avoiding obstacles, were
shown to have larger brain regions associated with hearing and memory than species
that hunting in open spaces (Safi & Dechmann, 2005).
Neuroecology is the study of the adaptive variation in cognition and the brain, often
by employing behavioural and anatomical techniques to examine the neural correlates
of cognition (Sherry, 2006). There are few studies directly investigating the correlation
between pelagic versus benthic ecologies and brain structure independent of phylogeny.
Yopak (2012) reviewed the neuroecology of cartilaginous fishes and found, irrespective of
phylogenetic history, larger brains with well-developed telencephala and highly foliated
cerebella occurred in reef-associated species, suggesting brain structures may have
developed in conjunction with enhanced cognitive capabilities. In contrast, deep-sea
species had relatively smaller brains, which the author suggested might be indicative of
specialization of non-visual senses.
Maddock and Young (1987) measured 30 optic and chromatophore lobes of the brains
of 63 species of cephalopod from all types of habitats, providing the most comprehensive
data on the neuroecology of cephalopods currently available. However, this study did
not control for phylogeny, meaning phylogenetic effects may confound the observed
differences and similarities between taxa. We can nonetheless identify some general
trends in the correlation between habitat and brain regions associated with vision and
body patterning (Figure 10.3). These data suggest that the mosaic theory proposed for
mammalian brains (Barton & Harvey, 2000), may be relevant for cephalopods; specific
brain regions apparently associated with particular ecologies have a greater volume than
would be accounted for by allometric scaling. In Figure 10.3, we show these data for 42
of these species for which complete records for mantle length, absolute brain volume,
optic-lobe volume and chromatophore-lobe volume were given.

10.4 Body pattern repertoire in the open ocean and deep sea

For shallow-water cephalopods there is a clear link between vision and body patterning
via the expression of chromatic components; there is no doubt that the visual system
of species such as Sepia officinalis plays a vital role in making decisions about what
body pattern to use in order to achieve effective camouflage in a given environment (see
chapter 9, this volume, by Zylinski & Osorio). Hanlon and Messenger (1996) put forward
a hypothesis of ‘ecological correlates of body patterning’ (ECBP) that suggested a strong
correlation between habitat complexity and patterning richness (number of chromatic
components). They proposed that diurnal cephalopod species occurring in complex
habitats such as coral reefs or kelp environments have a richer array of body patterns
at their disposal than near-reef or ‘murky habitat’ species, with a positive correlation
between the two. With this correlation they extrapolated and predicted that species
 Visual cognition in deep-sea cephalopods 229

(a) 3
1000

1 Abraliopsis morrisi 15 Illex illecebrosus 29 Sandalops

2 Allotheuthis media 16 Joubiniteuthis portieri 30 Sepia officinalis
3 Architeuthis dux 17 Loligo forbesi 31 Sepiola rondeleti
30
4 Bathyteuthis 18 Lolliguncula brevis 32 Sepiotheuthis sepiodea
Brain volume (mm3)

100 40 5 Chiroteuthis veranyi 19 Lycoteuthis diadema 33 Spirula spirula

21
17 32
15 6 Cranchia scabra 20 Mastigoteuthis schmidti 34 Teuthowenia megalops
24 34
19 38 7 Ctenopteryx siculus 21 Megalocranchia 35 Cirroteuthis veranyi
14
41 11 8 Discoteuthis laciniosa 22 Neorossia caroli 36 Cirrothauma murrayi
6
9 9 Egea inermis 23 Neoteuthis sp. 37 Eledonella
8 2 10
20 18
29 10 Galiteuthis glacialis 24 Octopoteuthis danae 38 Grimpoteuthis
10 33 5 35 36 11 Gonatus fabricii 25 Onychoteuthis banksii 39 Japetella
37
31 7 22
12 12 Grimalditeuthis bomplandi 26 Pickfordiateuthis pulchella 40 Octopus vulgaris
13 28 42
4 16
23 25 13 Heteroteuthis dispar 27 Pterygioteuthis giardi 41 Opisthoteuthis
26 14 Histioteuthis miranda 28 Pyroteuthis margaritifera 42 Vitreledonella
39
27
Key:
1
10 100 1000 10000 Nerictic Lower mesopelagic Octopod
Mantle length (mm) Upper mesopelagic Bathypelagic Benthic

(b)
700
(c)
6
6
600 40
Chromatophore lobe as % of brain volume
Optic lobe as % of brain volume

500 12
10

19 34 4
11 25
400 23 7 5 32
8
27 17
1 14
28
300 26 13 2
15 37
42 9 24
333729 31 18 3217
39 25 31 2 30
200 22 21 12 11 24
18 41
30 22
4 26
20 3 1 10
27 21
16 7
100 40 28 2 15 34
6 19
35 9
38 13 42 5
36 39 3635 20 3
0 0
1 10 100 1000 1 4 16 23 33 29 8 41 14 38 100 1000
Brain volume (mm3)

(d)

Figure 10.3 Brain size and brain regions in a range of cephalopods from different habitats (data
taken from Maddock & Young, 1987). The 34 species of decapods plotted on each graph are
listed on the right in alphabetical order. A further eight species of octopods are then listed in
alphabetical order and are distinguished by a red marker ring. Numbers relate to the placement
on the plot. The colour of marker represents a common habitat type (or usual collection depth
range) for adults during daylight hours (the depth occurrence of many species changes with size
and life-stage, and many of the species listed undergo vertical migration so will be found in
shallower waters at night), as given in the key. (a) Absolute brain volume plotted against mantle
230 Sarah Zylinski and Sönke Johnsen

in the less-complex ‘open ocean’ habitat would have the lowest number of chromatic
components, but conceded: ‘we can only make educated guesses about species living
on the deeper continental shelf, the deep fore-reef, on bland substrates like mud or in
oceanic and deep-water habitats’.
Accepting that Hanlon and Messenger’s (1996) ECBP was presented as a first attempt
to develop the correlation between habitat and body pattern richness, it should be noted
that it overweighs the significance of this correlation, because it assumes phylogenetic
independence between the data points. In other words, the correlation between patterning
richness and habitat complexity may be an artifact of the fact that most of the species for
which data exists are closely related. For example, of the 12 coral reef, rocky reef or kelp
environment species included, four are from the genus Octopus, while all of the ‘social
squids near reefs’ are loliginids. It might, therefore, be that the common ancestor of
Octopus had a richer range of body patterns compared with the common shared ancestor
of Loligo and, therefore, the apparent trend between body pattern number and habitat
could be largely explained by relatedness rather than by habitat complexity. A more
thorough investigation using independent contrasts would correct for this (Felsenstein,
1985; Purvis & Rambaut, 1995; Seibel & Carlini, 2001). However, before this can prove
useful we need to greatly increase and improve our knowledge and expand our data
set with quality behavioural and physiological data, particularly for mesopelagic and
bathypelagic taxa. Furthermore, while headway is being made in resolving phylogenetic
relationships within the cephalopods, these are (not surprisingly) biased towards neritic
taxa at the species level.
An important consideration for future work assessing the correlation between habi-
tat and body pattern richness is defining sensible and applicable metrics for habitat
complexity and chromatic components/body pattern richness. As Packard and Sanders
(1971) pointed out: ‘if we ask “How many patterns are there in an octopus?” the best,
though hardly satisfactory, answer is, “There are as many patterns as can be recognized
by the classifier”’. The large research effort on the body patterning of Sepia compared to
any other group is unsurprising given the extraordinary camouflage capabilities of these
animals (Hanlon, 2007; Zylinski & Osorio, 2011), but it has the potential to overweigh

←
Figure 10.3 (cont.) length, showing a positive correlation between brain volume and body length,
although there is a tendency for shallow-water benthic taxa (namely Loligo, Sepia and Octopus)
to fall above the trend, and bathypelagic/bathybenthic species to fall below the trend, be they
octopod or decapod (note log scales). (b) Plotting optic lobe volume (given as a percentage of
brain volume, measured separately because of location and large size (d)) against absolute brain
volume shows that optic lobe volume varies greatly and is not correlated with overall
brain volume. Interestingly, many pelagic and mesopelagic species have comparatively larger
optic lobes than neritic/benthic species that are considered highly visual. (c) Plotting the
chromatophore lobe volume against brain volume shows chromatophore lobe volume varies
greatly between species. There is a trend for shallow-water species to have relatively larger
chromatophore lobes, with notable exceptions (e.g. the deep-water pelagic octopod Eledonella).
Many deep-sea species had no chromatophore lobes or lobes too small to be measured. (d) An
oegopsid squid paralarva showing eye stalks with the large optic lobe located behind the eye on
the stalk and separate from the brain. See plate section for colour version.
 Visual cognition in deep-sea cephalopods 231

the number of chromatic components used by Sepia. Additionally, much work on Sepia
has been carried out under controlled laboratory conditions (this said, we still have an
incomplete description of their body patterning; see chapter 9, this volume, by Zylinski
& Osorio). Conversely, observations in mesopelagic taxa (see below) are typically made
from video footage under ROV lights, resulting in problems discriminating true physi-
ological changes under artificial lighting, rapid changes in body position and unnatural
behaviours. Issues such as these make sensible comparisons of body pattern richness
between species difficult. At a more basic level, the ECBP as it stands considers only the
structural aspects of the tangible environment; as discussed above, a more typical defini-
tion of a species’ habitat includes all aspects of physical and biological conditions. While
habitat structural complexity might drive a diverse range of body patterns for camou-
flage, social interactions might drive a wider range of body patterns for communication.

10.5 Evidence for complex visual behaviours and body patterning

in mesopelagic cephalopods

10.5.1 Octopoteuthis deletron

Bush, Robison and Caldwell (2009) used ROV footage from Monterey Bay Aquarium
Research Institute’s (MBARI’s) video archive (which records the occurrence of taxa
filmed during the Institute’s ROV deployments) to analyze the behaviour of the large
pelagic squid, Octopoteuthis deletron. They analyzed observations of 76 (presumed)
individuals recorded at depths between 344 and 1841 m and produced an ethogram of
behavioural components that were displayed by more than one individual. These included
17 light and dark chromatophore-based components (see chapter 9, this volume, by
Zylinski & Osorio and Hanlon & Messenger, 1996), some of which are surprisingly
complex given the low-light environment, such as fine arm bands and ventral mantle
bands. In addition to these chromatic components, the authors described a further 42
postural and bioluminescent components, giving a total of 59 components. This number
is comparable to those for several neritic and reef squid for which ethograms exist,
for example 59 in the broadfin squid Sepioteuthis affinis (Mauris, 1989), 39 in Loligo
opalescens (Hunt, Zeidberg, Hamner & Robison, 2000) and 56 in Loligo pealeii (Hanlon,
Maxwell, Shashar, Loew & Boyle, 1999). Bush, Robison and Caldwell (2009) suggest
these behavioural components might serve to impede search image formation, disrupt
hydrodynamic signatures and camouflage via Disruptive colouration and illumination,
as well as being used in conspecific communication and offensive behaviour.

10.5.2 Dosidicus gigas

Following the study of Octopoteuthis by Bush, Robison and Caldwell (2009; see above),
L. Trueblood, B. H. Robison and B. A. Seibel (unpublished data) used footage from the
MBARI’s ROV video library to produce an ethogram of behaviour and body patterning
of the Humboldt squid Dosidicus gigas at depths greater than 150 m. They described
232 Sarah Zylinski and Sönke Johnsen

29 putative chromatic components as well as 15 postural components and 6 locomotory

components. The authors suggest that of the 29 chromatic components described, 15
contained body countershading elements, making camouflage an important role in body
pattern usage. Individual squid were observed to use chromatic component changes
more when aggregated than when solitary, suggesting that body patterning is used for
intraspecific communication in this species.

10.5.3 Taningia danae

Kubodera, Koyama and Mori (2007) used a mid-water high definition baited camera rig to
record footage of the large eight-armed squid Taningia danae ‘hunting’ for suspended
bait between 240 m (night) and 940 m (day). They did not report any body pattern
colouration changes, but did observe arm-tip bioluminescence emissions associated
with bait attacks. The authors suggest these might function to both blind prey and
provide illumination to accurately determine target distance.

10.5.4 Japetella heathi

The pelagic octopus Japetella is caught in deep day trawls, with smaller individuals occur-
ring in shallow waters (600–800 m) and mature adults in deeper waters (800 m+) (Roper
& Young, 1975). This sluggish creature looks like a gelatinous ball with small arms –
very different from the fast-moving, protean octopuses we are familiar with. However,
we recently found that not all aspects of this small octopus are as slow as they first appear.
These animals are rather transparent when left undisturbed (with reflective guts and eyes
that are often associated with transparency), but are capable of expanding a layer of
red–orange chromatophores over the mantle (Zylinski & Johnsen, 2011; Figure 10.4).
We found that, while shadows and passing objects did not elicit the expansion of chro-
matophores in Japetella, a sudden onset of blue light caused these red-orange chro-
matophores to be expanded. This is consistent with the hypothesis that these and other
cephalopods (e.g. Onychoteuthis banksii) can maintain camouflage under both ambi-
ent light conditions and bioluminescent searchlights. Interestingly, we found that they
tracked passing objects, an observation made possible by their prominent, swivelling
eyes. Sweeney, Haddock and Johnsen (2007) suggesting that Japetella may have poor
visual acuity compared with other measured pelagic cephalopods. However, at four
cycles per degree, this is comparable to the acuity of a rainbow trout, which seems rather
good for an animal assumed to have very little visual stimulation within its habitat, and
which appears to have limited abilities to evade or respond to visual threat compared
with the rapid responses typical of many cephalopods.

10.6 Bioluminescence: using and detecting

In their discussion of ecological correlates of body patterning Hanlon and Messenger

(1988) stated: ‘We might . . . predict that cephalopods with light organs inhabiting the
 Visual cognition in deep-sea cephalopods 233

(a)

10 mm

(b)

80
        
chromatophore cover (%)

60

40

20

0
0 10 20 30 40
time (seconds)
(c)
80
ns ns ***
chromatophore cover (%)

60

40

20

0
object shadow tactile blue light
(n = 11) (n = 11) (n = 12) (n = 3)
Figure 10.4 (a) The same individual Japetella heathi octopus in transparent mode (left) and
pigmented mode (right). (b) Responses of a single J. heathi to directed blue light. Yellow boxes
234 Sarah Zylinski and Sönke Johnsen

Figure 10.5 Ventral and ocular photophores on the ventral surface of an Abraliopsis squid. See
plate section for colour version.

oceanic mid-waters could signal with bioluminescent body patterns’. As the intensity
of the ambient downwelling light decreases, the occurrence and importance of biolumi-
nescence increases. At depths greater than around 800 m (the exact depth being defined
by local conditions) bioluminescence becomes the dominant source of light (Young,
1983), present in every major taxonomic group represented. Haddock, Moline and Case
(2010) discuss the functions, diversity and physiology of bioluminescence in an excel-
lent review, and we direct the interested reader to this. Here, we draw attention to the
fact that almost half of cephalopod genera (66 of 139) possess bioluminescent organs
(Hastings & Morin, 1991), and a vast majority of the taxa possessing such organs are
mesopelagic (Figure 10.5). A well-known exception to this is the near-shore sepiolid
squid Euprymna scolopes, which produces bioluminescence via the luminescent bacte-
rial symbiont Vibrio fischeri (Ruby, 1996). The apparent functions of bioluminescence
in cephalopods are varied, including sexual signalling, biological searchlights, counter-
illumination and hunting behaviour (Clarke, 1963; Haddock, Moline and Case, 2010;
Johnsen, Widder & Mobley, 2004; Johnsen, 2005; Young, 1978, 1983). Biolumines-
cence seems to have evolved multiple times in cephalopods, with some species utilizing

←
Figure 10.4 (cont.) and icon indicate onset and cessation of individual lighting ‘bouts’, consisting
of a flashing blue light at one flash per second. Most bouts lasted for 3 seconds and, therefore,
subjected the animal to three flashes. Chromatophores can be seen to expand seconds after initial
exposure. After continued exposure the animal ceased reacting to the light with chromatophore
responses, and instead displayed evasive behaviour such as swimming away from the
light-source and retraction of the head into the mantle. (c) Responses of J. heathi to four different
stimuli. Objects passed in front of the animals, and shadows passed overhead, failed to evoke a
significant increase in chromatophore expression. A tactile stimulus (touching the arms with a
blunt needle) resulted in the rapid expansion of chromatophores. By comparison, directed blue
light resulted in a rapid and strong expression of chromatophores. White circles = pre-stimulus,
black circles = post-stimulus. Error bars show standard deviation from mean. From Zylinski &
Johnsen 2011. See plate section for colour version.
 Visual cognition in deep-sea cephalopods 235

Figure 10.6 Counterilluminating photophores in the squid Pterygioteuthis. Here, during

shipboard experiments, the very low downwelling blue light was been turned off and the animal
photographed from below under dim red light.

symbiotic luminescent bacteria (e.g. Euprymna), while most deep-sea taxa have intrinsic
bioluminescence.
Counterillumination is an important function of bioluminescence in pelagic cephalo-
pods (Johnsen, Widder & Mobley, 2004; McFall-Ngai, 1990; Widder, 2010). Many
predators visually hunting in the mesopelagic have upward-pointing eyes positioned
to take advantage of the conspicuousness of the silhouettes against the downwelling
ambient light. Many animals get around this by being transparent, which provides the
advantage of being effective from all viewing angles. Yet this is not a complete solution
to camouflage in the pelagic (see section 10.5.4. on Japetella heathi above). Another
solution is to use counterilluminating photophores that obliterate a silhouette by emitting
light at the wavelength and intensity of the downwelling light from ventral photophores
(Figure 10.6). Many cephalopods have extra-ocular photosensitive vesicles, which are
presumed to play a role in determining the output of ventral photophores. Young (1978)
conducted shipboard experiments where they used opaque ‘shutters’ to block information
about the intensity of downwelling light to a species of Abraliopsis squid over the dorsal
photosensitive vesicles, the eyes, or both. They demonstrated that the intensity of the
ventral photophores was independently affected by covering both eyes and photosensitive
vesicles, but that the latter had a greater effect. It would be interesting indeed to know
how information regarding downwelling light and the intensity of the photophore output
is processed.
The deep sea is often considered to be a lightless environment, but biological light
provides a method for communication that has the potential to be information rich as well
as spatially and temporally complex. Matching the spectral and intensity characteristics
of the ambient downwelling light for counterillumination is non-trivial. These are two
potential ways in which bioluminescence could support complex visual processing in
the pelagic realm.
236 Sarah Zylinski and Sönke Johnsen

Figure 10.7 Transparency is a common form of camouflage in the mesopelagic realm. Here the
transparency of a chranchiid squid is demonstrated as it rests over a colour standard. See plate
section for colour version.

10.7 Concluding remarks

Gaping holes exist in our knowledge of the basic behaviour of most cephalopods, many
of which inhabit the vast, three-dimensional wilderness that is the deep sea. Gaining
insight to potential complex behaviours in mesopelagic taxa is difficult for numerous
reasons.
The need for a wide body pattern repertoire for camouflage via chromatic components
is reduced in the mesopelagic realm. Body patterning is probably less useful at depth for
two main reasons: (1) light is more monochromatic; and (2) the visual resolution of most
animals is far worse due to spatial summation. Although camouflage is still important, it
is more commonly achieved in deep-sea cephalopods by passive means, such as reflec-
tors, transparency (Figure 10.7) or overall red/black colouration. Counterillumination
via bioluminescence to obliterate silhouettes must match the ambient downwelling light
in order to be effective, and the means by which this is achieved is not well understood.
However, it seems unlikely it will require cognitive processes in the traditional sense.
Cognition is needed when adaptive responses depend on changing conditions. When an
environment is highly constant – such as the deep sea – there is no selective pressure
for cognitive responses, as fixed responses are sufficient. In addition the environment
is colder, darker and food-limited, meaning that visual cognition and the brainpower
needed to mediate it are unlikely to evolve.
 Visual cognition in deep-sea cephalopods 237

At present, attempts to correlate body pattern richness with habitat (or environmental)
complexity seem fraught with difficulties of assessing either in a comparable or repeat-
able way. Pelagic habitats from well-lit surface waters to deep, cold waters, although
indeed equally low in habitat (structural) complexity, perhaps offer more variable and
complex environmental conditions than is often appreciated. A metric of complexity
that accounts for interactions with conspecifics and predators, the level of usable ambi-
ent light, the degree of bioluminescence available for visual tasks and factors regard-
ing metabolic constraints should be developed to enable comparisons between pelagic
species. Similarly, phylogenetic relatedness may be a confounding factor in apparent
correlations between body pattern richness and habitat complexity, and between body
patterning and visual cognition.
Chapter 9 of this volume (by Zylinski & Osorio) introduced concepts about what
makes vision cognitive, and emphasized that internal representation is likely an important
determinant here. The lack of a body pattern output that directly relates to aspects of
the visual environment in mesopelagic taxa invalidates the use of this assay for evidence
of internal representation. Many aspects of the ecology and physiology of a majority
of mesopelagic cephalopod taxa lead us to doubt the need or presence of cognitive
vision. However, ROV footage of some species suggests complex behaviours and a
surprisingly large body pattern repertoire. Furthermore, brain measurements show a
large investment in optic lobes (coupled with large eyes) and learning centres (e.g. the
vertical lobe system) in some taxa, which tantalize us with the potential for complex
visual processing. Finally, the complex species-specific photophore patterns found in
many mesopelagic squids provide another avenue for visual communication that is so
far poorly understood.
Chapter 9 of this volume (by Zylinski & Osorio) highlights similarities between visual
processing in the neritic cuttlefish S. officinalis and vertebrates (including humans). The
visual environment in the well-lit shallow-water benthic environment is not dissimilar
to our own visual world in many respects, and the body patterns used by S. officinalis
for camouflage are designed primarily to defeat shallow-water vertebrate (including
mammalian) predators. It is, therefore, perhaps unsurprising that evolution has provided
similar solutions to shared problems. In contrast, it is harder to imagine the visual tasks
facing cephalopods and their predators in a low-light world where biological light might
be the most important source of illumination. We hope time and research effort will
shed light on this. As new technologies allow us better access than ever to the world’s
largest and least understood habitat we will surely be treated to many more fascinating
cephalopod stories. Whether visual cognition is ultimately an interesting avenue of
research to pursue for the majority of these taxa, any behavioural data obtained can only
be an asset given the current level of knowledge.

Acknowledgements

We would like to acknowledge support from Office of Naval Research MURI grant no.
N00014–09–1–1053.
238 Sarah Zylinski and Sönke Johnsen

References

Arkhipkin, A. (1996). Age and growth of planktonic squids Cranchia scabra and Liocranchia
reinhardti (Cephalopoda, Cranchiidae) in epipelagic waters of the central-east Atlantic. Journal
of Plankton Research, 18: 1675–1683.
Arkhipkin, A. I. and Nigmatullin, C. M. (1997). Ecology of the oceanic squid Onychoteuthis
banksi and the relationship between the genera Onychoteuthis and Chaunoteuthis (Cephalopoda:
Onychoteuthidae). Journal of the Marine Biological Association of the United Kingdom, 77:
839–869.
Barbosa, A., Allen, J. J., Mäthger, L. M. and Hanlon, R. T. (2012). Cuttlefish use visual cues
to determine arm postures for camouflage. Proceedings of the Royal Society. B: Biological
Sciences, 279: 84–90.
Barton, R. A. and Harvey, P. H. (2000). Mosaic evolution of brain structure in mammals. Nature,
405: 1055–1058.
Bergman, T. J., Beehner, J. C., Cheney, D. L. and Seyfarth, R. M. (2003). Hierarchical classification
by rank and kinship in baboons. Science, 302: 1234–1236.
Budelmann, B. U. (1994). Cephalopod sense organs, nerves and the brain: adaptations for
high performance and life style. Marine and Freshwater Behaviour and Physiology, 25: 13–
33.
Budelmann, B. U. (1995). The cephalopod nervous system: what evolution has made of the
molluscan design. In The Nervous Systems of Invertebrates: An Evolutionary and Comparative
Approach O. Breidbach and W. Kutsch (Eds.). Basel, Switzerland: Birkhäuser Verlag.
Bush, S. L., Robison, B. H. and Caldwell, R. L. (2009). Behaving in the dark: locomotor, chromatic,
postural, and bioluminescent behaviors of the deep-sea squid Octopoteuthis deletron Young
1972. The Biological Bulletin, 216: 7–22.
Chiao, C.-C., Chubb, C., Buresch, K., Siemann, L. and Hanlon, R. T. (2009). The scaling
effects of substrate texture on camouflage patterning in cuttlefish. Vision Research, 49: 1647–
1656.
Childress, J. J. (1995). Are there physiological and biochemical adaptations of metabolism in
deep-sea animals? Trends in Ecology and Evolution, 10: 30–36.
Childress, J. J., Barnes, A. T., Quetin, L. B. and Robison, B. H. (1978). Thermally protecting cod
ends for the recovery of living deep-sea animals. Deep Sea Research, 25: 419–422.
Clarke, W. (1963). Function of bioluminescence in mesopelagic organisms. Nature, 198: 1244–
1246.
Clayton, N. S., Dally, J. M. and Emery, N. J. (2007). Social cognition by food-caching corvids. The
Western Scrub-Jay as a natural psychologist. Philosophical Transactions of the Royal Society
of London. Series B, 362: 507–522.
Collett, T. S. and Collett, M. (2002). Memory use in insect visual navigation. Nature Reviews
Neuroscience, 3: 542–552.
Domingues, P. M., Dimarco, F. P., Andrade, J. P. and Lee, P. G. (2005). Effect of artificial diets on
growth, survival and condition of adult cuttlefish, Sepia officinalis Linnaeus, 1758. Aquaculture
International, 13: 423–440.
Felsenstein, J. (1985). Phylogenies and the comparative method. The American Naturalist, 125:
1–15.
Forsythe, J. W. and Hanlon, R. T. (1997). Foraging and associated behavior by Octopus cyanea
Gray, 1849 on a coral atoll, French Polynesia. Journal of Experimental Marine Biology and
Ecology, 209: 15–31.
 Visual cognition in deep-sea cephalopods 239

Forsythe, J. W., Lee, P. and Walsh, L. (2002). The effects of crowding on growth of the European
cuttlefish, Sepia officinalis Linnaeus, 1758 reared at two temperatures. Journal of Experimental
Marine Biology and Ecology, 269: 173–185.
Godfrey-Smith, P. (2002). Environmental complexity and the evolution of cognition. In The
Evolution of Intelligence R. J. Sternberg and J. C. Kaufman (Eds.). New York, NY: Psychology
Press.
Grosenick, L., Clement, T. S. and Fernald, R. D. (2007). Fish can infer social rank by observation
alone. Nature, 445: 429–432.
Haddock, S. H. D., Moline, M. A. and Case, J. F. (2010). Bioluminescence in the sea. Annual
Review of Marine Science, 2: 443–493.
Hall, K. C. and Hanlon, R. T. (2002). Principal features of the mating system of a large spawning
aggregation of the giant Australian cuttlefish Sepia apama (Mollusca: Cephalopoda). Marine
Biology, 140: 533–545.
Hanlon, R. T. (2007). Cephalopod dynamic camouflage. Current Biology, 17: R400–R404.
Hanlon, R. T., Forsythe, J. W. and Joneschild, D. E. (1999). Crypsis, conspicuousness, mimicry and
polyphenism as antipredator defences of foraging octopuses on Indo-Pacific coral reefs, with a
method of quantifying crypsis from video tapes. Biological Journal of the Linnean Society, 66:
1–22.
Hanlon, R. T., Maxwell, M. R., Shashar, N., Loew, E. R. and Boyle, K. L. (1999). An ethogram
of body patterning behavior in the biomedically and commercially valuable squid Loligo pealei
off Cape Cod, Massachusetts. The Biological Bulletin, 197: 49–62.
Hanlon, R. T. and Messenger, J. B. (1988). Adaptive coloration in young cuttlefish (Sepia offic-
inalis L) – the morphology and development of body patterns and their relation to behavior.
Philosophical Transactions of the Royal Society of London. Series B, 320: 437–487.
Hanlon, R. T. and Messenger, J. B. (1996). Cephalopod Behaviour. New York, NY: Cambridge
University Press.
Hastings, J. W. and Morin, J. G. (1991). Bioluminescence. In Neural and Integrative Animal
Physiology, 4th edn, C. L. Prosser (Ed.). New York, NY: Wiley-Liss.
Hills, T. T. (2006). Animal foraging and the evolution of goal-directed cognition. Cognitive
Science, 30: 3–41.
Hoving, H. J. T., Bush, S. L. and Robison, B. H. (2012). A shot in the dark: same-sex sexual
behaviour in a deep-sea squid. Biology Letters, 8: 287–290.
Hunt, J. C. and Seibel, B. A. (2000). Life history of Gonatus onyx (Cephalopoda: Teuthoidea):
ontogenetic changes in habitat, behavior and physiology. Marine Biology, 136: 543–552.
Hunt, J. C., Zeidberg, L. D., Hamner, W. M. and Robison, B. H. (2000). The behaviour
of Loligo opalescens (Mollusca: Cephalopoda) as observed by a remotely operated vehi-
cle (ROV). Journal of the Marine Biological Association of the United Kingdom, 80: 873–
883.
Jerlov, N. G. (1976). Marine Optics. Amsterdam, the Netherlands: Elsevier.
Johnsen, S. (2005). The red and the black: bioluminescence and the color of animals in the deep
sea. Integrative and Comparative Biology, 45: 234–246.
Johnsen, S., Widder, E. A. and Mobley, C. D. (2004). Propagation and perception of biolumines-
cence: factors affecting counterillumination as a cryptic strategy. The Biological Bulletin, 207:
1–16.
Kubodera, T., Koyama, Y. and Mori, K. (2007). Observations of wild hunting behaviour and
bioluminescence of a large deep-sea, eight-armed squid, Taningia danae. Proceedings of the
Royal Society. B: Biological Sciences, 274: 1029–1034.
240 Sarah Zylinski and Sönke Johnsen

Langridge, K. V., Broom, M. and Osorio, D. (2007). Selective signalling by cuttlefish to predators.
Current Biology, 17: R1044–R1045.
Lefebvre, L. and Sol, D. (2008). Brains, lifestyles and cognition: are there general trends? Brain,
Behavior and Evolution, 72: 135–144.
Maddock, L. and Young, J. Z. (1987). Quantitative differences among the brains of cephalopods.
Journal of Zoology, 212: 739–767.
Marshall, N. B. (1958). Aspects of Deep Sea Biology, 2nd edn. London, UK: Huchinson.
Mather, J. A. (1991). Navigation by spatial memory and use of visual landmarks in octopuses.
Journal of Comparative Physiology A, 168: 491–497.
Mäthger, L. M., Chiao, C.-C., Barbosa, A. and Hanlon, R. T. (2008). Color matching on natural
substrates in cuttlefish, Sepia officinalis. Journal of Comparative Physiology, A, 194: 577–585.
Mauris, E. (1989). Colour patterns and body postures related to prey capture in Sepiola affinis
(Mollusca: Cephalopoda). Marine Behaviour and Physiology, 14: 189–200.
McFall-Ngai, M. J. (1990). Crypsis in the pelagic environment. American Zoologist, 30: 175–188.
Packard, A. and Sanders, G. D. (1971). Body patterns of Octopus vulgaris and maturation of the
response to disturbance. Animal Behaviour, 19: 780–790.
Pollen, A. A., Dobberfuhl, A. P., Scace, J., Igulu, M. M., Renn, S. C. P., Shumway, C. A. and
Hofmann, H. A. (2007). Environmental complexity and social organization sculpt the brain in
Lake Tanganyikan cichlid fish. Brain, Behavior and Evolution, 70: 21–39.
Purvis, A. and Rambaut, A. (1995). Comparative analysis by independent contrasts (CAIC): an
Apple Macintosh application for analysing comparative data. Computer Applications in the
Biosciences, 11: 247–251.
Robison, B. H. (2004). Deep pelagic biology. Journal of Experimental Marine Biology and
Ecology, 300: 253–272.
Roper, C. F. E. and Young, R. E. (1975). Vertical Distribution of Pelagic Cephalopods. Washington,
DC, USA: Smithsonian Institution Press.
Ruby, E. G. (1996). Lessons from a cooperative, bacterial-animal association: the Vibrio fischeri-
Euprymna scolopes light organ symbiosis. Annual Review of Microbiology, 50: 591–624.
Safi, K. and Dechmann, D. K. N. (2005). Adaptation of brain regions to habitat complexity:
a comparative analysis in bats (Chiroptera). Proceedings of the Royal Society. B: Biological
Sciences, 272: 179–186.
Safi, K., Seid, M. A. and Dechmann, D. K. N. (2005). Bigger is not always better: when brains get
smaller. Biology Letters, 1: 283–286.
Seibel, B. A. and Carlini, D. B. (2001). Metabolism of pelagic cephalopods as a function of habitat
depth: a reanalysis using phylogenetically independent contrasts. The Biological Bulletin, 201:
1–5.
Seibel, B. A., Thuesen, E. V., Childress, J. J. and Gorodezky, L. A. (1997). Decline in pelagic
cephalopod metabolism with habitat depth reflects differences in locomotory efficiency. The
Biological Bulletin, 192: 262–278.
Shashar, N., Rutledge, P. S. and Cronin, T. W. (1996). Polarization vision in cuttlefish – a concealed
communication channel? Journal of Experimental Biology, 199: 2077–2084.
Sherry, D. F. (2006). Neuroecology. Annual Review of Psychology, 57: 167–197.
Shohet, A. J., Baddeley, R. J., Anderson, J. C., Kelman, E. J. and Osorio, D. (2006). Cuttlefish
responses to visual orientation, water flow and a model of motion camouflage. Journal of
Experimental Biology, 209: 4717–4723.
Sweeney, A. M., Haddock, S. H. D. and Johnsen, S. (2007). Comparative visual acuity of coleoid
cephalopods. Integrative and Comparative Biology, 47: 808–814.
 Visual cognition in deep-sea cephalopods 241

Talbot, C. M. and Marshall, N. J. (2010). Polarization sensitivity in two species of cuttlefish, Sepia
plangon (Gray 1849) and Sepia mestus (Gray 1849), demonstrated with polarized optomotor
stimuli. Journal of Experimental Biology, 213: 3364–3370.
Vecchione, M. and Roper, C. F. E. (1991). Cephalopods observed from submersibles in the western
north Atlantic. Bulletin of Marine Science, 49: 433–445.
Warrant, E. J. and Locket, N. A. (2004). Vision in the deep sea. Biological Reviews, 79: 671–
712.
Wentworth, S. L. and Muntz, W. R. A. (1989). Asymmetries in the sense organs and central nervous
system of the squid Histioteuthis. Journal of the Zoological Society, London, 219: 607–619.
Widder, E. A. (2010). Bioluminescence in the ocean: origins of biological, chemical, and ecological
diversity. Science, 328: 704–708.
Yopak, K. E. (2012). Neuroecology of cartilaginous fishes: the functional implications of brain
scaling. Journal of Fish Biology, 80: 1968–2023.
Young, R. (1978). Vertical distribution and photosensitive vesicles of pelagic cephalopods from
Hawaiian waters. Fishery Bulletin, 76: 583–616.
Young, R. (1983). Oceanic bioluminescence: an overview of general functions. Bulletin of Marine
Science, 33: 829–845.
Zylinski, S., How, M. J., Osorio, D., Hanlon, R. T. and Marshall, N. J. (2011). To be seen or to
hide: visual characteristics of body patterns for camouflage and communication in the Australian
giant cuttlefish Sepia apama. The American Naturalist, 177: 681–690.
Zylinski, S. and Johnsen, S. (2011). Mesopelagic cephalopods switch between transparency and
pigmentation to optimize camouflage in the deep. Current Biology, 21: 1937–1941.
Zylinski, S. and Osorio, D. (2011). What can camouflage tell us about non-human visual percep-
tion? A case study of multiple cue use in the cuttlefish Sepia officinalis. In Animal Camouflage:
Mechanisms and Function M. Stevens and S. Merilaita (Eds.). Cambridge, UK: Cambridge
University Press.
Zylinski, S., Osorio, D. and Shohet, A. J. (2009). Perception of edges and visual texture in the
camouflage of the common cuttlefish, Sepia officinalis. Philosophical Transactions of the Royal
Society of London. Series B, 364: 439–448.
Index of species

Abraliopsis 235 Loligo (Doryteuthis) 151, 152, 153, 154,

ants 162, 164 230
Aplysia 37 forbesi 157, 160
Astrea 128 opalescens 157, 231
pealeii 157, 231
Balistes 139 vulgaris 155
blue jays 133 Lolliguncula 153, 160
Bothus 198
Macrotritopus defilippi 141
cartilaginous fishes 228 Mithrax 127
chameleons 179 moray eel 139
chitons 133 Mytilus 136
cichlid fish 227
crabs 162 Nautilus 38, 43
pompilius 31
decapod crustaceans 157
Doryteuthis (Loligo) Octopus 57, 72, 95, 96, 154, 179, 225,
opalescens 158 230
Dosidicus gigas 156, 160, 231 bimaculatus 126, 127, 137, 159, 161
briareus 159
Eledone 152 cyanea 127, 128, 132, 134, 140, 141, 142, 159,
Enteroctopus dofleini 62, 65, 126, 127, 128, 130, 162, 187, 189
133, 136, 154 dierythraeus 130, 136
Euprymna 156 dofleini 158
morsei 157 insularis 126, 130, 132, 141
scolopes 142, 234 joubini 138, 154, 155, 161
rubescens 111, 125, 135, 138, 154, 161, 163,
fish 179 165
Fissurella 127 teheulechus 127
vulgaris 62, 65, 126, 129, 130, 132, 133, 134,
Goldfish 163 135, 136, 137, 138, 139, 141, 151, 155, 157,
Gonatus fabricii 155 159, 162, 165
Octopoteuthis 226
Hapalogaster mertensii 128, 133 deletron 231
Hermissendra 37 Ommastrephes bartramii 160
Histioteuthis 226
Pachygrapsus 127
Illex argentinus 156 Pardachirus mormoratus 179
isopods 179 Pinna 133
plaice 132
Japetella heathi 232 Pleuronectes 198
Pomacentrid 157
Lima (Ctenoides) 127, 135 primates 227
limpets 162 Protothaca 136
 Index of species 243

Sepia 151, 152, 197, 202, 225, 230 Taningia danae 232
apama 158, 160, 161 Tegula
officinalis 3, 7, 10, 23, 45, 135, 152, 155, 157, aureocincta 127
158, 159, 160, 161, 163, 164, 166, 198, 206, eiseni 127
211, 212, 216, 230, 238 termites 162
pharaonis 162, 206 Todarodes 154
Sepioteuthis Tritonia diomedea 153
affinis 231
australis 160
Venerupis 136
lessoniana 157
Vibrio fischeri 234
squirrels 138
Strombus 130, 136
Synanceia 179 Watasenia scintillans 152
Index

availability 127, 128 optic lobe 19, 151, 205, 237

plexiform region 37
behaviour transection 87
behavioural components 202 two brains 117
behavioural lateralization 14 vertical lobe 8, 20, 73, 100, 135
behavioural plasticity 16
bioluminescence 225 Chemical perception 152
arm-tip bioluminescence 232 chemical trail 162
bioluminescent components 231 chemotactile search 128, 133
bioluminescent organs 234 food imprinting 9, 14
body patterns 198, 202 food–odor reward 39
body pattern components 202 olfactory lobe 37
body pattern repertoire 237 olfactory organs 34
components 186, 231 rhinophores 34, 39
counterillumination 235 chipping 136
countershading 141 cognition 94, 125, 198, 205, 227
diffuse reflectors 181 cognitive functions 72
Disruptive body pattern 16, 206 cognitive map 135, 199
general background matching 189 cognitive process 188
iridophore 180, 201 cognitive steps 136, 141
leucophore 180, 201 cognitive system 3
masquerade strategy 190 comparative study of cognition 31
mimicry 141 decision-making 132, 136, 199
mottle 206 defined objects 212
patterning richness 230 direct perception 199
pigment 180 feature detectors 205
polarized patterns 185 figure-ground segregation 208
spatial properties of the body patterns home range 136
202 information processing 95
static camouflage 179 innate preference 7
structural color 180 integrate visual and taste 74
uniform body pattern 206 perception 199
brain structures 17 perceptual categorization 10
amacrine cells 74, 78 perceptual grouping 214
brachial ganglion 101 prenatal cognitive activity 3
brachial plexus 101, 118 representation of space 105
brain maturation 17 secondary tactics 142
brain size 227 solving problems 136
lateralization 15 somatotopic map 112, 115
cerebral lateralization 12, 15 spatial map 135
local brachial modules 102 coleoid expansion 48
motor centres 19 complexity
motor control circuitry 100 complex ecologies 227
olfactory lobe 37 complex environments 134
 Index 245

convergence 37, 203 imaginary animals 97

convergent selection pressures 96 interaction of genes and experience 14
cross-modal effect 12 iridophore 180, 201

deep sea lateralization 15

deep-sea cephalopods 223 bilaterality 182
deep-water niches 48 bilaterally asymmetrical patterns 185
development cerebral lateralization 12, 15
brain maturation 17 side-turning preference 12
developmental neuroethology 4 visual lateralization 14
developmentally conserved sequence learning 128, 136, 139, 141
5 associative learning 7, 46
imprinting 9, 158 biphasic learning curve 41
innate preference 7 classical associative condition 38
interaction of genes and experience conditional discrimination 161
14 continuities in learning 23
post-embryonic maturation 19 food imprinting 9, 14
prenatal cognitive activity 3 food–odor reward 39
dispersal 153, 155 forgetting 134
divergence 32 habituation 6
drilling 136 imprinting 9, 158
learn to detour 165
ecological complexity 225 learning how to learn 133
egg capsule 3 memory 99
energy content 128 long-term memory 7, 41
enrichment 21, 67 memory space 46
ethogram 186, 231 short term memory 7, 41, 109
evolutionary arms race 197 mutually reinforcing activation 110
exploration 42, 58, 131, 134, 161 observational learning 74
curiosity 66 preference 127, 128
extrinisic exploration 58 reflect response 188
representations 205
foraging 125 route memory 43
chemotactile search 128, 133 rules 97
foraging strategies 125 sensitive period 10
generalist 128 spatial memory 134, 161
generalize 10 spatial scale 212
giving-up time 132 spatial strategy 166
home range 136 leucophore 180, 201
range size 165
rate-maximizing foragers 130 mechanism 95
secondary tactics 142 mechanosensory perception 34
specialist 133, 138 lateral line analogue 152
time minimizer 130, 137 statocysts 152
win-switch 134, 136 touch 153
forgetting 134 vibration 35
Fourier vibration cues 157
components 208 memory 99
transform 207 long-term memory 7, 41
memory space 46
habitat complexity 230 short-term memory 7, 41, 109
habituation 6 migration 156
hatching 19 mimicry 141
hatching period 154 movement
hatchlings 4, 9 fleeing 185
horizontal 156 freezing 184
hormonal system 95 locomotory components 232
246 Index

movement (cont.) high-level processes 199

migration 156 interaction levels 62
motor interneurons 99, 103
motor centres 19 low-level processes 199
motor control circuitry 100 synaptic plasticity 95, 99
motor sequence 163 activity-dependent synaptic plasticity 80, 83
pattern of movement 139 tetanization 87
position 179 modal completion 211
positive phototaxis 155 mutually reinforcing activation 110
postural 231 recurrent 108
postural components 202 representation 96
pulling 136 representation of space 105
sand digging 16 self-organizational processes 81
side-turning preference 12 somatotopic map 112, 115
strength 186 spatial map 135
tentacle extension 34, 40 neuroecology 228
vertical migrations 33, 156 neurotransmitter and chemical systems 95
water jetting 62 5-HT 85
Web Over 132 acetylcholine 22, 78
catecholamines 22
navigate 42, 150, 158 glutamate 76
celestial cues 157 hexamethonium 79
chemical trail 162 neuromodulators 84
detour 135, 165 neurotransmitters 21
learn to detour 165 NMDA 81
exploration oxytocin 23
curiosity 66 serotonin 22, 84
intrinsic exploration 59
landmark 43, 135 Octo-munculus 115
displacing landmarks 163 orientation 150, 179
visual landmarks 163 paralarvae 155
mazes 161
intra-maze cues 44 perception 199
object-oriented strategy 189 perceptual categorization 10
orientation 150, 179 perceptual grouping 214
path integration 162 personality 127, 155
piloting 150, 162 phenomenal world 218
representation of space 105 play
route memory 43 criteria for play 61
search function of play 66
saltatory search 126, 131 recapitulation theory of play 66
search image 131, 132, 140, 141, 142, types of play 61
182 polyphenism 182
spatial predation 125
spatial memory 134, 161 evolutionary arms race 197
spatial scale 212 generalist 128
spatial strategy 166 predator–prey arms race 135
tracking 160 prey 12
trail following 135 prey selectivity 128, 130
neural activities risk of predation 126, 130
discrete communication channels 185 specialist 133, 138
distributed control network 113 target species 183
distribution 155 Web Over 132
feed-forward 80 principal component analysis 203
feedback 108
feedback loop 95 salivary papilla 136
global attributes 214 sand digging 16
 Index 247

search contrast differences 207

saltatory search 126, 131 eyes 151
search image 141 eye-use preference 13
selection 135 pinhole eye 33
k-selected lifestyle 32 relative eye size 226
semelparous 136, 157 feature detectors 205
sex difference 166 figure-ground segregation 208
social interactions 231 fragment edges 211
stability 182 image analysis 187
suckers 104 luminance 12
sucker ganglia 101 optic lobe 19, 151, 205, 237
photophores 235
theory 96 point of view 183
timing polarization 151
biphasic learning curve 41 polarized light 164
flexibility of timing 139 polarized patterns 185
giving-up time 132 positive phototaxis 155
hatching period 154 proximate visual cues 163
post-embryonic maturation 19 search image 131, 132, 140, 141, 142, 182
sensitive period 10 similarity 187
time minimizer 130, 137 simple background 216
stimulus detection 5
vertical migration 33, 156 textural 202
visual perception texture assessments 184
amodal completion 211 texture discrimination 212
apparent movement 129 transparency 236
background matching 142 vertebrate vision 197
bioluminescence 225 viewing 135
arm-tip bioluminescence 232 visual acuity 232
bioluminescent components 231 visual landmarks 163
bioluminescent organs 234 visual lateralization 14
boundary completion 212 visual template 142
color assessments 184 visually complex 216
color-blind 181, 206
continuity 182 water currents 157
contour completion 209 window 10
Figure 1.1 Sepia officinalis embryonic stages according to Lemaire (1970) from left to right:
(A) stage 23, (B) stage 25 and (C) stage 30 (copyright L. Bonnaud and A.-S. Darmaillacq).

Figure 4.12 A tentative model for the octopus learning and memory system. See explanation in
text.
Figure 6.2 An octopus (Octopus insularis) during chemotactile search by many arms in the
second phase of foraging (Photograph courtesy of T. S. Leite.)
Figure 6.3 An octopus (Octopus insularis) accompanied by scavenging fish as it forages on the
rocky ground.
Figure 7.1 Cuttlefish Sepia officinalis. The picture has been taken in the English Channel off
Granville (Manche, France). The ocular globe is particularly voluminous. The eye is
characterized by the w-shaped papillae. (Copyright, F. Sichel).
Figure 7.2 Giant Pacific Octopus, Enteroctopus dofleini, paralarvae. They do not have the same
way of life as adults, which are generally benthic. Photo courtesy of J. Kocian.
Figure 7.3 Octopus cyanea in its den. Note the stones surrounding the den entrance. (Copyright,
A.-S. Darmaillacq).
Figure 8.1 Camouflage examples in marine animals. (a) A stone fish, Synanceia verrucosa
choosing a place to settle which matches its body pattern. (b) An isopod: displaying disruptive
coloration. (c) A Posidonia pipefish, Syngnathus sp. using object resemblance in hiding between
seagrass leafs. (d) A Moses sole, Pardachirus mormoratus showing a static Mottle pattern; note
partial missmatch with such a static camouflage.
 (a) (c) (e)

(b) (d) (f)

Figure 8.5 Image analysis of camouflaged octopuses. (a,b) Cryptic Octopus cyanea and Octopus vulgaris respectively in their natural habitat. The white
square marks the sampled mantle for similarity comparison. (c,d) Computer-generated similarity map of the RA-slopes descriptor. For clarity reasons
only 90% similarity or higher is presented. (e,f ) Computer-generated similarity map of the entropy descriptor.
Figure 9.2 The bright blues and greens in the mating displays of the Australian giant cuttlefish
Sepia apama are primarily due to wavelength specific reflectors. The animals themselves are
most likely colour-blind.
Figure 9.3 Demonstration of just some of the body pattern variation Sepia officinalis uses in
camouflage body pattern responses to the visual environment. The top row shows the primary
characteristics of the main three classes of body pattern referred to in the text: Uniform (left),
Mottle (centre) and Disruptive (right). The remaining images show how individual chromatic
components patterns can be varied in their expression, along with the overall contrast of the body
pattern, to give a wide range of body patterns that cannot always be readily identified as
Uniform, Mottle or Disruptive.
 (a)

(b)

Figure 9.7 The giant Australian cuttlefish, Sepia apama, chooses simple backgrounds to display
against, which probably acts to increase signal efficacy, and more complex backgrounds to
camouflage against making visual search more difficult for potential predators. (a) A male
engaged in an agonistic display to another male. (b) A camouflaging individual uses chromatic
and textural components. Adapted from Zylinski, How, Osorio, Hanlon and Marshall (2011).
Figure 10.2 The lower-mesopelagic squid Histeoteuthis has asymmetrical eyes and optic lobes,
with a large left eye for looking up for passing silhouettes of potential prey against the weak
downwelling light, and a small right eye orientated downwards, probably specialized for
detecting point-source bioluminescence.
(a) 3
1000

1 Abraliopsis morrisi 15 Illex illecebrosus 29 Sandalops

2 Allotheuthis media 16 Joubiniteuthis portieri 30 Sepia officinalis
3 Architeuthis dux 17 Loligo forbesi 31 Sepiola rondeleti
30
4 Bathyteuthis 18 Lolliguncula brevis 32 Sepiotheuthis sepiodea
Brain volume (mm3)

100 40 5 Chiroteuthis veranyi 19 Lycoteuthis diadema 33 Spirula spirula

21
17 32
15 6 Cranchia scabra 20 Mastigoteuthis schmidti 34 Teuthowenia megalops
24 34
19 38 7 Ctenopteryx siculus 21 Megalocranchia 35 Cirroteuthis veranyi
14
41 11 8 Discoteuthis laciniosa 22 Neorossia caroli 36 Cirrothauma murrayi
6
9 9 Egea inermis 23 Neoteuthis sp. 37 Eledonella
8 2 10
20 18
29 10 Galiteuthis glacialis 24 Octopoteuthis danae 38 Grimpoteuthis
10 33 5 35 36 11 Gonatus fabricii 25 Onychoteuthis banksii 39 Japetella
37
31 7 22
12 12 Grimalditeuthis bomplandi 26 Pickfordiateuthis pulchella 40 Octopus vulgaris
13 28 4 42
23 16 25 13 Heteroteuthis dispar 27 Pterygioteuthis giardi 41 Opisthoteuthis
26 14 Histioteuthis miranda 28 Pyroteuthis margaritifera 42 Vitreledonella
39
27
Key:
1
10 100 1000 10000 Nerictic Lower mesopelagic Octopod
Mantle length (mm) Upper mesopelagic Bathypelagic Benthic

(b)
700
(c)
6
6
600 40

Chromatophopre lobe as % of brain volume

Optic lobe as % of brain volume

500 12
10

19 34 4
11 25
400 23 7 5 32
8
27 17
1 14
28
300 26 13 2
15 37
42 9 24
333729 31 18 3217
39 25 31 2 30
200 22 21 12 11 24
18 41
30 22
4 26
20 3 1 10
27 21
16 7
100 40 28 2 15 34
6 19
35 9
38 13 42 5
36 39 3635 20 3
0 0
1 10 100 1000 1 4 16 23 33 29 8 41 14 38 100 1000
Brain volume (mm3)

(d)

Figure 10.3 Brain size and brain regions in a range of cephalopods from different habitats (data taken from
Maddock & Young, 1987). The 34 species of decapods plotted on each graph are listed on the right in
alphabetical order. A further eight species of octopods are then listed in alphabetical order and are
distinguished by a red marker ring. Numbers relate to the placement on the plot. The colour of marker
represents a common habitat type (or usual collection depth range) for adults during daylight hours (the
depth occurrence of many species changes with size and life-stage, and many of the species listed undergo
vertical migration so will be found in shallower waters at night), as given in the key. (a) Absolute brain
volume plotted against mantle length, showing a positive correlation between brain volume and body length,
although there is a tendency for shallow-water benthic taxa (namely Loligo, Sepia and Octopus) to fall above
the trend, and bathypelagic/bathybenthic species to fall below the trend, be they octopod or decapod (note log
scales). (b) Plotting optic lobe volume (given as a percentage of brain volume, measured separately because
of location and large size (d)) against absolute brain volume shows that optic lobe volume varies greatly and
is not correlated with overall brain volume. Interestingly, many pelagic and mesopelagic species have
comparatively larger optic lobes than neritic/benthic species that are considered highly visual. (c) Plotting the
chromatophore lobe volume against brain volume shows chromatophore lobe volume varies greatly between
species. There is a trend for shallow-water species to have relatively larger chromatophore lobes, with
notable exceptions (e.g. the deep-water pelagic octopod Eledonella). Many deep-sea species had no
chromatophore lobes or lobes too small to be measured. (d) An oegopsid squid paralarva showing eye stalks
with the large optic lobe located behind the eye on the stalk and separate from the brain.
(a)

10 mm

(b)
80
chromatophore cover (%)

60

40

20

0
0 10 20 30 40
time (seconds)

(c) 80
ns ns ***
chromatophore cover (%)

60

40

20

0
object shadow tactile blue light
(n = 11) (n = 11) (n = 12) (n = 3)
Figure 10.4 (a) The same individual Japetella heathi octopus in transparent mode (left) and pigmented mode
(right). (b) Responses of a single J. heathi to directed blue light. Yellow boxes and icon indicate onset and
cessation of individual lighting ‘bouts’, consisting of a flashing blue light at one flash per second. Most bouts
lasted for 3 seconds and, therefore, subjected the animal to three flashes. Chromatophores can be seen to
expand seconds after initial exposure. After continued exposure the animal ceased reacting to the light with
chromatophore responses, and instead displayed evasive behaviour such as swimming away from the
light-source and retraction of the head into the mantle. (c) Responses of J. heathi to four different stimuli.
Objects passed in front of the animals, and shadows passed overhead, failed to evoke a significant increase in
chromatophore expression. A tactile stimulus (touching the arms with a blunt needle) resulted in the rapid
expansion of chromatophores. By comparison, directed blue light resulted in a rapid and strong expression of
chromatophores. White circles = pre-stimulus, black circles = post-stimulus. Error bars show standard
deviation from mean. From Zylinski & Johnson 2011.
Figure 10.5 Ventral and ocular photophores on the ventral surface of an Abraliopsis squid.
Figure 10.7 Transparency is a common form of camouflage in the mesopelagic realm. Here the
transparency of a chranchiid squid is demonstrated as it rests over a colour standard.