Download the full version of the ebook at ebookfinal.

com

The Psychology of Eyewitness Identification 1st

Edition James M. Lampinen

https://ebookfinal.com/download/the-psychology-of-
eyewitness-identification-1st-edition-james-m-lampinen/

OR CLICK BUTTON

DOWNLOAD EBOOK

Download more ebook instantly today at https://ebookfinal.com

Instant digital products (PDF, ePub, MOBI) available
Download now and explore formats that suit you...

Protecting Children from Violence Evidence Based

Interventions 1st Edition James Michael Lampinen

https://ebookfinal.com/download/protecting-children-from-violence-
evidence-based-interventions-1st-edition-james-michael-lampinen/

ebookfinal.com

Raptors of the World Helm Identification Guides 1st

Edition James Ferguson-Lees

https://ebookfinal.com/download/raptors-of-the-world-helm-
identification-guides-1st-edition-james-ferguson-lees/

ebookfinal.com

The Handbook of Evolutionary Psychology 1st Edition David

M. Buss

https://ebookfinal.com/download/the-handbook-of-evolutionary-
psychology-1st-edition-david-m-buss/

ebookfinal.com

Positive Psychology of Love 1st Edition M. Hojjat

https://ebookfinal.com/download/positive-psychology-of-love-1st-
edition-m-hojjat/

ebookfinal.com
Psychology of Classroom Learning An Encyclopedia
Psychology of Classroom Learning 1st Edition Eric M.
Anderman
https://ebookfinal.com/download/psychology-of-classroom-learning-an-
encyclopedia-psychology-of-classroom-learning-1st-edition-eric-m-
anderman/
ebookfinal.com

Ecological Psychology in Context James Gibson Roger Barker

and the Legacy of William James s Radical Empiricism
Resources for Ecological Psychology 1st Edition Harry Heft
https://ebookfinal.com/download/ecological-psychology-in-context-
james-gibson-roger-barker-and-the-legacy-of-william-james-s-radical-
empiricism-resources-for-ecological-psychology-1st-edition-harry-heft/
ebookfinal.com

Emotions in the Field The Psychology and Anthropology of

Fieldwork Experience 1st Edition James Davies

https://ebookfinal.com/download/emotions-in-the-field-the-psychology-
and-anthropology-of-fieldwork-experience-1st-edition-james-davies/

ebookfinal.com

The Process of Research in Psychology 3rd Edition Dawn M.

Mcbride

https://ebookfinal.com/download/the-process-of-research-in-
psychology-3rd-edition-dawn-m-mcbride/

ebookfinal.com

Biological Psychology 11th Edition James W. Kalat

https://ebookfinal.com/download/biological-psychology-11th-edition-
james-w-kalat/

ebookfinal.com
The Psychology of
Eyewitness Identification
ESSAYS IN COGNITIVE PSYCHOLOGY

North American Editors:

Henry L. Roediger, III, Washington University in St. Louis
James R. Pomerantz, Rice University

European Editors:
Alan D. Baddeley, University of York
Vicki Bruce, University of Edinburgh
Jonathan Grainger, Université de Provence

Essays in Cognitive Psychology is designed to meet the need for rapid publica-
tion of brief volumes in cognitive psychology. Primary topics will include
perception, movement and action, attention, memory, mental represen-
tation, language and problem solving. Furthermore, the series seeks to
define cognitive psychology in its broadest sense, encompassing all topics
either informed by, or informing, the study of mental processes. As such,
it covers a wide range of subjects including computational approaches to
cognition, cognitive neuroscience, social cognition, and cognitive devel-
opment, as well as areas more traditionally defined as cognitive psychol-
ogy. Each volume in the series will make a conceptual contribution to the
topic by reviewing and synthesizing the existing research literature, by
advancing theory in the area, or by some combination of these missions.
The principal aim is that authors will provide an overview of their own
highly successful research program in an area. It is also expected that
volumes will, to some extent, include an assessment of current knowl-
edge and identification of possible future trends in research. Each book
will be a self-contained unit supplying the advanced reader with a well-
structured review of the work described and evaluated.

FORTHCOMING
Bahrick et al.: Maintenance of Knowledge
Butler & Kang: The Mnemonic Benefits of Retrieval Practice
Weaver: Flashbulb Memory

PUBLISHED
Schmidt: Extraordinary Memories of Exceptional Events
Lampinen, Neuschatz, & Cling: Psychology of Eyewitness Identification
Brown: The Tip of the Tongue State
Worthen & Hunt: Mnemonology: Mnemonics for the 21st Century
Surprenant & Neath: Principles of Memory
Kensinger: Emotional Memory Across the Lifespan
Millar: Space and Sense
Evans: Hypothetical Thinking
Gallo: Associative Illusions of Memory
Cowan: Working Memory Capacity
McNamara: Semantic Priming
Brown: The Déjà Vu Experience
Coventry & Garrod: Saying, Seeing, and Acting
Robertson: Space, Objects, Minds and Brains
Cornoldi & Vecchi: Visuo-Spatial Working Memory and Individual Differences
Sternberg et al.: The Creativity Conundrum
Poletiek: Hypothesis-Testing Behaviour
Garnham: Mental Models and the Interpretations of Anaphora

For updated information about published and forthcoming titles in the

Essays in Cognitive Psychology series, please visit: www.psypress.com/
essays
The Psychology of
Eyewitness Identification
JamEs mIchaEl lamPInEn, JEffrEy s. nEuschaTz,
and andrEw d. clIng

Psychology Press

New York London

Psychology Press Psychology Press
Taylor & Francis Group Taylor & Francis Group
711 Third Avenue 27 Church Road
New York, NY 10017 Hove, East Sussex BN3 2FA
© 2012 by Taylor & Francis Group, LLC
Psychology Press is an imprint of Taylor & Francis Group, an Informa business

Printed in the United States of America on acid-free paper

Version Date: 2011912

International Standard Book Number: 978-1-84872-883-7 (Hardback)

For permission to photocopy or use material electronically from this work, please access www.
copyright.com (http://www.copyright.com/) or contact the Copyright Clearance Center, Inc.
(CCC), 222 Rosewood Drive, Danvers, MA 01923, 978-750-8400. CCC is a not-for-profit organiza-
tion that provides licenses and registration for a variety of users. For organizations that have been
granted a photocopy license by the CCC, a separate system of payment has been arranged.

Trademark Notice: Product or corporate names may be trademarks or registered trademarks, and
are used only for identification and explanation without intent to infringe.

Library of Congress Cataloging‑in‑Publication Data

Lampinen, James M.
The psychology of eyewitness identification / James Michael Lampinen, Jeffrey
S. Neuschatz, Andrew D. Cling.
p. cm. -- (Essays in cognitive psychology)
Includes bibliographical references and index.
ISBN 978-1-84872-883-7 (hbk. : alk. paper)
1. Criminal investigation--Psychological aspects. 2. Mistaken
identity--Psychological aspects. 3. Eyewitness identification--Psychological
aspects. 4. Memory. 5. Forensic psychology. I. Neuschatz, Jeffrey S. II. Cling,
Andrew D. III. Title.

HV8073.5.L36 2012
363.25’8--dc23 2011036788

Visit the Taylor & Francis Web site at

http://www.taylorandfrancis.com
and the Psychology Press Web site at
http://www.psypress.com
CONTENTS

Acknowledgments ix

About the Authors xi

Chapter 1 Twenty-Seven Years 1

Chapter 2 Theoretical Approaches to Eyewitness Identification 11
Chapter 3 Estimator Variables 49
Chapter 4 System Variables 113
Chapter 5 Indicia of Reliability 167
Chapter 6 Field Studies of Eyewitness Identification 205
Chapter 7 Expert Testimony 231
Chapter 8 Conclusions and Thoughts 255
Chapter 9 Philosophical Afterword: Memory and Reasonable
Belief 263

References 287

Author Index 321

Subject Index 329

vii
ACKNOWLEDGMENTS

The authors would like to thank Amber Culbertson-Faegre, Blake Erickson,

Chris Peters, Lindsey Sweeney, and all the members of the Law and
Psychology Lab at the University of Arkansas, and Stacy Wetmore, Devin
Harker, and Will Exley from the University of Alabama in Huntsville, all
of whom helped proofread sections of the manuscript. We would also like
to thank Stephanie Drew from Psychology Press and Henry Roediger,
series editor, for their help and patience in the process of shepherding this
book through to the very end.

ix
ABOUT THE AUTHORS

James Michael Lampinen is a professor of psychology at the University of

Arkansas. He received his bachelor’s degree in psychology from Elmhurst
College and his PhD in cognitive psychology from Northwestern
University. He then spent 2 years as a postdoctoral fellow at the State
University of New York in Binghamton. Dr. Lampinen’s research focuses
on the subjective experience of false memories, mechanisms people use
to avoid false memories, and applications of memory research to legal
issues such as the accuracy of eyewitness identification and efforts to
recover missing or wanted individuals (i.e., prospective person memory).
Dr. Lampinen has published 45 book chapters and journal articles, has
edited two books, and is under contract for a book on missing persons
cases (Missing in America) and an introductory book on human memory
(Memory 101). Dr. Lampinen is a recipient of the University of Arkansas’
Outstanding Mentor Award and Fulbright College of Arts and Sciences’
Master Researcher Award. He lives in Fayetteville, Arkansas, with his
wife Stephanie.

Jeffrey S. Neuschatz received his BS at Roger Williams University in 1992

and his PhD at Binghamton University in 1999, and is associate professor
and chair of the Psychology Department at the University of Alabama in
Huntsville. Dr. Neuschatz is the coauthor of more than 30 scholarly works.
His work has appeared in leading academic journals including Law and
Human Behavior and Journal of Experimental Psychology: Learning, Memory,
and Cognition. He is the recipient of the University of Alabama–Huntsville
Foundation Award for Creative Achievement.

Andrew D. Cling received his AB at the University of Missouri at Columbia

in 1979 and his PhD at Vanderbilt University in 1987, and is professor and
chair of the Department of Philosophy at the University of Alabama in
Huntsville. Dr. Cling’s research in philosophy is currently focused on
questions in the theory of knowledge. In particular, he is interested in a
family of ancient skeptical paradoxes that seem to show that some of our
xi
xii About the Authors

core assumptions about having reasons for belief and good standards for
intellectual judgment are inconsistent. Dr. Cling is also engaged in inter-
disciplinary work on memory and eyewitness identification. Dr. Cling’s
papers have appeared in such journals as Philosophy and Phenomenological
Research, Synthese, Philosophical Studies, Philosophy of Science, and Philosophical
Psychology.
 1
C H A P T E R

Twenty-Seven Years

Twenty-seven years is an awfully long time. Twenty-seven years ago one

of us was still in graduate school (AC), one of us was just out of high
school (JL), and one of us had just had his bar mitzvah (JN). The Soviet
Union was still considered the United States’ chief nemesis, and Ronald
Reagan was president of the United States. Twenty-seven years ago IBM
had just released its XT personal computer, which came with a whopping
128 kB of memory and a floppy disk drive to boot. The World Wide Web
was not yet invented, and cell phones were in their infancy. There was no
Facebook. There was no Farmtown. There was no Wii.
In 27 years, a young adult, just out of high school, goes through a series
of milestones that may include going to college, starting one’s career,
forming new friendships, travel, courtship and marriage, parenthood,
and many other changes. One’s hair can fall out. Or it can turn gray. In
the United States, the average person lives 78 years (World Bank, 2010).
Twenty-seven years is more than one third of the typical person’s life and
close to one half of the typical person’s adult life.
For Michael Green, 27 years was the time of his longest nightmare
(Solano, 2010). Back in 1983, four men abducted and sexually assaulted a
woman in Houston. The men got away after a police chase, and investiga-
tors immediately started canvassing the area. They stopped males who
matched the general description of the perpetrators and showed them to
the victim to see if she could identify them. When Michael Green was
stopped and shown to the victim, shortly after the crime, she failed to
identify him as one of her attackers. But the police did not stop there. A
week later, the victim was shown a photographic lineup that included a
picture of Michael Green, as well as pictures of several other men. When
the victim saw the lineup, she identified Green, even though a week ear-
lier she had not. Later on, Green was put in a live lineup and shown to the
victim again. She identified him again. At trial she indicated that she was

1
2 The Psychology of Eyewitness Identification

absolutely certain that Green was her attacker. Based on her identification,
Green was convicted and sentenced to 75 years in prison.
There were a number of reasons to be concerned about the victim’s
identification in this case. For one thing, shortly after her attack, when her
memory was freshest, the victim failed to identify Green as her attacker.
Then, a week later, when her memory was presumably less fresh, she did
identify Green. Additionally, Green was shown to the victim multiple
times, once shortly after the assault, once in a photo lineup a week later,
then a bit later in a live lineup. This identification procedure is bound
to lead to problems, because it makes it obvious to the witness who the
police suspect of the crime. It may also result in an elevated feeling of
familiarity with the suspect’s face.
Despite problems with the identification procedure, the jury convicted
Green. Unfortunately, the jury was mistaken. In 2010, new DNA tests were
ordered in Green’s case, and he was exonerated. An innocent man, he had
spent 27 years of his life in prison. While in prison, his mother had died.
He was not able to go to her funeral. Michael Green will never get those
27 years back.
To date, there have been 258 DNA exoneration cases in the United States
(Innocence Project, 2010). That’s 258 and counting. In the average case, the
person exonerated spent 13 years in prison before being released. In 70%
of the cases, the person exonerated was a member of a racial or ethnic
minority group. Importantly, for the present book, a mistaken eyewitness
identification was a contributing cause in more than 75% of these wrong-
ful convictions (see Figure 1.1). Of the cases involving mistaken eyewit-
ness identifications, more than one third of cases involved two or more
witnesses making the same mistaken identification.
The 258 DNA exoneration cases likely represent the tip of the iceberg
when it comes to wrongful convictions. About three quarters of DNA
exoneration cases involve sexual assaults (Gross, Jacoby, Matheson,
Montgomery, & Patel, 2005). This is because biological evidence is more
likely to be available in sexual assault cases than in other types of cases.
As eyewitness researcher John Tuttle put it, “Unless the guy robbing
that 7-11 store gets pretty damned excited, he’s not going to be leaving
behind any biological fluids” (cited by Wells, n.d.-a). Without available
biological evidence, there may be no way for individuals convicted of
other types of crimes to establish their factual innocence. Given that
rapists make up only 10% of the state prison population in the United
States—and only one third of those cases involve sexual assaults by
strangers—it has been estimated that the number of undetected wrong-
ful convictions in the United States numbers in the thousands (Gross et
al., 2005).
 Twenty-Seven Years 3

77%

52%

23%
16%

Informants/
False Confessions Snitches

Faulty Forensic Science

Mistaken Eyewitness
Identifications

FIGURE 1.1 Contributing factors in the first 225 DNA exoneration

cases. Percentages sum to more than 100% because in some cases more
than one factor was involved. (Source: Based on data obtain from The
Innocence Project at the Benjamin N. Cardozo School of Law at Yeshiva
University, http://www.innocenceproject.org/understand/; clip art used
to create the image obtained from the Open Clipart Library, http://www.
openclipart.org; clip art image drawn by Johnny_automatic.)

□ The Science of Eyewitness Identification

The science of eyewitness memory has been around for more than a cen-
tury and in the past 30 years has produced a wealth of important findings
that can be used by social policy makers to improve the criminal justice
system. The focus of this book is on those variables that impact eyewit-
ness identification accuracy. We are interested in cases where police show
a witness one or more people and ask the witness to make an identifica-
tion. There are different procedures that police can use in order to see if a
witness can identify a suspect. For instance, in Michael Green’s case, the
first identification attempt involved a procedure called a showup. A showup
is an identification procedure in which a witness is shown a single suspect
and is asked if this is the person who committed the crime. Showups can
be conducted live, as in Michael Green’s case. These are sometimes called
corporeal showups. Showups can also be conducted by showing the witness
a photograph of the suspect. These are called photographic showups.
The second time the victim saw Michael Green, he was in a lineup. In a
lineup, a witness is shown multiple people (often six) and is asked to indi-
cate which individual, if any, committed the crime. A good lineup should
have a single suspect and a number of foils (commonly five; foils are also
4 The Psychology of Eyewitness Identification

sometimes called fillers or distracters). The suspect is the person who the
police think might have committed the crime. The foils are known to be
innocent, although they match the general description of the culprit. Foils
are used to protect innocent suspects. If the lineup procedures are fair (see
Chapters 2 and 4) and the suspect is innocent, a witness who makes a mis-
take is more likely to pick one of the foils than he or she is to pick the
suspect.
Lineups can be conducted live or with a set of photos. Live lineups are
sometimes called corporeal lineups, and photographic lineups are some-
times called photoarrays. In the United States, photoarrays are often made
up of six pictures, arranged in two rows of three photos. These are some-
times informally referred to as six packs. In the United States, photoar-
rays are far and away more common than corporeal lineups (Fulero &
Wrightsman, 2009). One reason for this is that it is typically much easier
to put together the lineup when all you need to do is find photographs of
foils who match the description of the culprit, as opposed to finding live
people who match the description of the culprit. Additionally, under U.S.
law, suspects in corporeal lineups have a right to counsel, whereas sus-
pects in photoarrays do not (United States v. Ash, 1973).
For the most part, research psychologists study eyewitness memory by
conducting experiments. In a typical eyewitness experiment, participants
view a mock crime that may be in the form of a videotaped re-creation or
a slide presentation, or may be staged live. In some experiments, partici-
pants are deceived into believing an actual crime has occurred (e.g., theft
of a computer). In other experiments, participants know all along that it
is just a simulation of a crime. Following the mock crime, there is usually
a retention interval. The retention interval is simply the time that passes
between the event and the attempted identification. Following the reten-
tion interval, the participants may be shown a lineup (or showup) and
asked to make a decision. For half the participants, the lineup includes
the person who committed the crime. This is called a target-present lineup.
Target-present lineups mimic the situation where the police have a sus-
pect, and the suspect is in fact guilty of the crime. For the other half of the
participants, the lineup does not include the person who committed the
crime. This is called a target-absent lineup. Target-absent lineups are meant
to mimic the situation where the police have a suspect, but the suspect is
in fact innocent of the crime. The advantage of using experimental tech-
niques such as these is that they allow researchers to examine how accu-
rate witnesses are under varying conditions. The researchers can measure
the number of correct identifications of the suspect in the target-present
lineup and the number of mistaken identifications of the suspect in the
target-absent lineup and see what factors influence accuracy.
To give just one example, consider a clever study by Pigott, Brigham,
and Bothwell (1990). They obtained the cooperation of bank managers
 Twenty-Seven Years 5

at a number of different banks to conduct the study. At each bank, a vol-

unteer entered the bank, went to the island where people fill out their
deposit slips, and then went up to a teller and tried to cash an obvi-
ously forged postal money order. The money order was a $10 money
order with an extra “1” written on it in ink so that it said $110. All the
tellers who were approached in this way refused to cash the money
order. The volunteer then acted irritated, took the money order back,
and left the bank. The tellers informed their supervisors, and a few
hours later a researcher posing as a plainclothes detective interviewed
the tellers and showed them photoarrays that either included the cul-
prit or did not include the culprit. The lineups had been constructed
for the researchers by an actual police detective. When the tellers were
shown the target-present lineup, the suspect was identified about half
of the time. When the tellers were shown the target-absent lineup, an
innocent person was mistakenly identified close to 40% of the time.
Keep in mind that the tellers in this experiment thought that a real
crime had occurred and that they were being interviewed by a real
police officer.
In the past 30 years, hundreds of studies of eyewitness identification
have been conducted. These studies shed light on the factors that can
influence eyewitness accuracy. More importantly, the studies have had an
impact on police practice, training, and policies (Wells et al., 2000). In the
present book, we will be discussing some of the most important variables
that have been identified by eyewitness researchers.

□ The Scope of the Problem

How big of a problem is mistaken eyewitness identification? There is no
way to know for sure. According to survey research, eyewitness evidence
plays a critical role in at least 77,000 criminal cases per year in the United
States (Wells et al., 1998). Archival studies of actual eyewitness identifi-
cation attempts have found that witnesses falsely identify foils around
20% of the time (Valentine, Pickering, & Darling, 2003). This number
likely underestimates the number of foil identifications, because police
sometimes record foil identifications by simply noting that the suspect
was not identified (Wells, n.d.-a). This means that at least one out of five
identification attempts results in the selection of an innocent person. If
real witnesses to real crimes make the mistake of picking innocent foils,
then real witnesses to real crimes undoubtedly also make the mistake of
picking innocent suspects. After all, from a witness’s point of view, an
innocent foil and an innocent suspect are functionally equivalent (nei-
ther is the culprit).
6 The Psychology of Eyewitness Identification

In an attempt to provide a more precise answer to the question of how

often witnesses identify innocent suspects, Levi (1998) derived an esti-
mate of the probability that a suspect is guilty, given that the suspect has
been chosen from a lineup, by using insights from probability theory (i.e.,
the Bayes theorem). First, consider a simplified case, where half of all line-
ups contain a guilty suspect and half of all lineups contain an innocent
suspect (i.e., in the real world, there are just as many target-present and
target-absent lineups). Statisticians would say the base rate for guilt is 50%
and would denote it by saying p(G) = .50. Assuming that base rate, Levi
pointed out that the probability of guilt, given that a suspect is chosen
from a lineup, can be given by the following simplified version of the
Bayes theorem:

p(C|TP)
p(G|C) =
p(C|TP) + p(C|TA)

In this equation, p(G|C) (is the probability that the person is guilty given
that they were chosen from a lineup, p(C|TP) is the probability of choos-
ing a suspect from a target-present lineup, and p(C|TA) the probability of
choosing a suspect from a target-absent lineup. Based on values taken from
a number of published peer-reviewed studies, Levi estimated the probabil-
ity of choosing the suspect from a target-present lineup was about 44% and
the probability of choosing the suspect from a fair target-absent lineup was
about 9.17%. So putting those values back into the equation, one gets the
following:

.44
p(G|C) = = .83
.44 + .0917

So if Levi’s assumptions are correct, about 83% of the suspects identified

from lineups are guilty and about 17% are innocent. That’s quite a star-
tling conclusion: Levi’s calculations suggest that 17% of the suspects iden-
tified from lineups are innocent!
This value, of course, is just an estimate and depends on whether the
right values were used for p(G), p(C|TP), and p(C|TA). Levi argued that
the estimate is probably a conservative estimate. First, Levi argued that
actual witnessing conditions are usually not as good as they are in lab-
oratory studies. Thus, the actual value of p(C|TP) is probably less than
44% and the value of p(C|TA) is probably more than 9.17%. Second, the
value given for p(C|TA) assumes that fair lineup procedures were being
used, but this is not always the case in real life (see Chapters 2 and 4).
Thus, p(C|TA) may be even higher. Levi argued that these considerations
 Twenty-Seven Years 7

suggest that the actual rate of mistaken suspect identifications might be

more than 17% in real-world lineups.
The estimate provided by Levi also depends heavily on the base rate
of guilt. The base rate indicates what percentage of all police lineups
actually includes a guilty suspect. You can see why base rates matter.
If the base rate of guilt was 100%, then all suspect identifications would
be correct identifications, even if witnesses made their decisions by roll-
ing a six-sided die or reading chicken entrails. If the base rate of guilt
was 0%, then all suspect identifications would be mistaken identifica-
tions. The basic principle is that, everything else being equal, the higher
the base rate for guilt, the more likely it is that a suspect who has been
identified by a witness is guilty. As p(G) increases, p(G|C) increases.
Figure 1.2 shows the effects of base rates on the probability that some-
one is guilty, given that they have been chosen from a lineup (based
on the estimates of p(C|TP) and p(C|TA) that Levi derived from prior
research). Levi argued that the base rate of guilt may be less than 50%,
because police may identify someone as a suspect simply because he or
she matches the description and was in the area, resulting in many inno-
cent suspects being put into lineups. That’s certainly what happened in
Michael Green’s case.

0.9
Probability of Correct or Mistaken ID

0.8

0.7

0.6

0.5
Mistaken ID
0.4 Correct ID
0.3

0.2

0.1

0
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
Base Rate Probability of Guilt

FIGURE 1.2 Conditional probability that suspect is guilty or innocent

given that suspect was identified by a witness. The conditional proba-
bilities are shown as a function of the base rate of guilt. Assumes values
of p(C|TP) and p(C|TA) derived from Levi (1998).
8 The Psychology of Eyewitness Identification

□ Plan for the Book

There is good reason to believe that there is a problem of mistaken iden-
tifications. We believe that the key to addressing this problem is to use
psychological theory and research to develop practices and approaches
that can be of service to the criminal justice system. In what follows, we
describe some of that research. In Chapter 2, we describe basic psycho-
logical theory that relates to eyewitness identification. These theories
cover the whole range of experimental psychology including perception,
memory, and social reasoning. We next describe, in Chapter 3, some of
the most important estimator variables that impact identification accuracy.
An estimator variable is a variable that impacts witness memory but that
the criminal justice system has no control over (e.g., how well lit the crime
scene was, or how far away the perpetrator was). We then describe a range
of important system variables that can have an impact on accuracy (Chapter
4). A system variable is a variable that impacts accuracy and that is under
the control of the criminal justice system (e.g., the instructions given to
the witness). Finally, we discuss what we refer to as indicia of reliability
(Chapter 5). By indicia of reliability, we mean variables that are correlated
with accuracy, even though they don’t directly cause the witness to be
accurate or inaccurate. For instance, on average, confident witnesses are
more accurate than less confident witnesses. However, confidence does
not cause the witness to be accurate. As we will see, factors like confidence
are related to accuracy, but the relationship is imperfect and can some-
times be misleading.
In Chapters 6 and 7, we will move to applications of the broad body of
knowledge developed by eyewitness researchers over the past 30 years. In
Chapter 6 we will describe studies that have examined eyewitness iden-
tification accuracy in real-world settings. We will find that, for the most
part, these studies of real-world witnesses tend to confirm the findings
of laboratory studies. In Chapter 7, we will discuss the role of eyewitness
experts as witnesses in the court system. We will find that the eyewitness
expert has an important role to play in informing the court system about
the factors that influence eyewitness memory.

□ A Final Word
Sometimes, when talking to reporters or lawyers, the topic of eyewitness
identification errors will come up. In such conversations, we sometimes
hear that eyewitness testimony is notoriously inaccurate. As we have
reviewed above, it is true that eyewitnesses can sometimes make grievous
 Twenty-Seven Years 9

errors, and it is important to be aware of this fact. But to state that eye-
witness testimony is error prone is to miss the point by means of over-
simplification. If all we had to say was that eyewitnesses make errors,
we wouldn’t need to write a book. A brief article would do. And there
wouldn’t be a need for over 2,000 articles and counting on the topic of
eyewitness testimony. An article or two would do.
The truth is more complex and nuanced than the simple conclusion
that eyewitness reports are inaccurate. What researchers have found over
the last three decades is that sometimes witnesses are accurate and some-
times they are inaccurate, and that there is a set of variables that can be
used to assess the likely accuracy of the witness (estimator variables) and
inform police practice (system variables). This is an extraordinarily posi-
tive message, because it suggests concrete steps that can be taken to help
make the criminal justice system work better. When the wrong person is
convicted of a crime, not only does an innocent person languish in prison,
but also the actual culprit is free to commit additional crimes. That’s why
it’s so important to get things right. This book is about how to get things
right.
 2
C H A P T E R

Theoretical Approaches to
Eyewitness Identification

Famed social psychologist Kurt Lewin (1952, p. 169) once commented,

“There is nothing so practical as a good theory.” Most psychologists who
do applied work understand well the interplay between application and
basic psychological theory. Not only does basic research in human cogni-
tion and social processes inform our understanding of the performance of
the eyewitness in choosing (or not choosing) someone from a lineup, but
the reverse is also true. The applied questions that eyewitness researchers
pose oftentimes provide new theoretical insights. A colleague in graduate
school once said to one of us (JL), “In applied research you get your theo-
retical insights for free.” That’s part of what makes research on the topic
of eyewitness identification so exciting and rewarding.
Another important fact about role theory in eyewitness research is
that it is multifaceted. The theoretical issues that touch on the psychol-
ogy of eyewitness identification run the gambit of human experimental
psychology: memory, social influence, decision making, emotion, person-
ality, and perception. All of these basic psychological mechanisms influ-
ence eyewitness identification. The current chapter provides background
for issues that we will be reviewing in the subsequent chapters. It by no
means covers all of the theoretical issues relevant to eyewitness identi-
fication. In our view, that would require a review of all of psychology.
Rather, in this chapter we review some of the most important theoreti-
cal perspectives that relate to the accuracy of eyewitness identification.
These ideas will be returned to again and again during the remainder of
this monograph. This chapter is broken up into three main sections. First
we discuss basic theory with regard to human memory that is useful for
understanding eyewitness identification. Second, we discuss basic pro-
cesses involved in human face perception and memory for faces. Third,

11
12 The Psychology of Eyewitness Identification

we discuss theoretical frameworks that have been specifically designed to

help us understand the nature of eyewitness identification.

□ Human Memory
You are sitting on an old wooden bench at a bus terminal waiting for the
8:45 a.m. bus that is going to take you to Binghamton, New York, from
Chicago. You are feeling a number of mixed emotions. On the one hand,
you are looking forward to arriving in Binghamton, where you plan to
spend a few days hanging out with old friends from graduate school. You
are traveling to a 15-year reunion. On the other hand, it’s going to be a
long bus ride, 700 miles one way, and you’re not looking forward to being
on a crowded, smelly bus for that length of time. You obsessively check
your coat pocket, making sure you haven’t misplaced your bus ticket. A
gentleman in a Chicago Bears knit cap approaches you, brandishes a sil-
ver gun, and politely asks you for your wallet. You comply, your heart
racing. The man strolls off, looking back over his shoulder only once, as
he makes his way out of the terminal. For a minute, you sit there in shock,
not knowing what to do. You reach for your coat pocket and take out your
cell phone to call 911. It looks like your trip is going to be delayed.

Basic Stages of Human Memory

The above example illustrates the type of situation that can happen in
which eyewitness identification is going to be crucial. Once the police
arrive, they will likely interview you. They will ask you what happened,
what was stolen, what the gentleman looked like, what he was wearing,
and a range of other questions. They might take you to the station and
have you look through some mugshots. They might have you talk to a
police artist to produce a composite of the thief. Or, if they catch someone
who matches the general description you provided, the police might pres-
ent you with a lineup or a showup (a one-person lineup).
What is interesting about all of this is that the robber is not there any
longer. He is long gone. Yet, when the police ask you what the robber
looked like, you are able to provide an answer. How does this occur? The
answer must surely be that stored in the synapses of your brain is a men-
tal representation of the event that just happened to you and what the
robber looked like (Barsalou, 2008). That’s what a memory is, a mental
representation of a past event. This memory representation is not an exact
or literal copy of what happened, and, in fact, it is well established that
memory representations are subject to distortion (Lampinen, Neuschatz,
 Theoretical Approaches to Eyewitness Identification 13

& Payne, 1998). Your accuracy when answering each of the questions that
the police ask you depends crucially on this memory representation, the
veridicality of the representation, your ability to access it, and your ability
to reason with that representation once it has been retrieved.
Memory psychologists make distinctions between three basic stages
of memory processing known as encoding, storage, and retrieval (Galotti,
2004). Encoding involves processing perceptual information (e.g., sounds,
sights, smells, and contextual information) into a mental code that can be
further processed. It is important to note that encoding is not a passive
process, but rather involves selection and elaboration (Weinstein & Mayer,
1986). Selection occurs because people have limited attentional capacity
(Kahneman, 1973). People are unable to process all of the environmen-
tal stimuli available at any given time. These capacity limitations require
that people selectively attend to some information and ignore other infor-
mation. Capacity is further limited by stress, pressure, and the need to
engage in other activities. Because the encoding stage is selective, some
information that may be relevant in later testimony is probably never
mentally encoded at all. For instance, you would not be able to accurately
answer a question about the culprit’s shoes if you never really took note
of his shoes.
In addition to being selective, the encoding stage of memory also involves
elaboration. Elaboration involves interpreting events, making inferences,
and drawing connections (Anderson, 1995). For instance, if you draw the
inference that the perpetrator was a homeless person, then your memory
for the perpetrator’s actions and appearance may be colored by stereotypes
you have about homeless people. This inference-making process is so com-
monplace that it typically happens automatically and without conscious
effort. It is important to note that the information that actually gets encoded
into memory is only that information that has been selected as well as elab-
orations that have been made on that information. Research indicates that
people often have difficulty later distinguishing between what was actually
experienced and the inferences that were made based on that information
(Hannigan & Reinitz, 2001, 2003; Kleider, Pezdek, Goldinger, & Kirk, 2008;
Lampinen, Faries, Neuschatz, & Toglia, 2000; Singer & Remillard, 2008).
This tendency to confuse inferences that one made with events as they actu-
ally happened makes memory subject to distortion (Chan & McDermott,
2006; Erskine, Markham, & Howie, 2001; García-Bajos & Migueles, 2003;
Greenberg, Westcott, & Bailey,1998; Lampinen et al., 1998; Roediger &
McDermott, 2000; Tuckey & Brewer, 2003).
The storage phase involves retaining information in memory after
encoding (Anderson, 1995). For instance, as a witness you would likely
encode a variety of details into your memory representation for the
above event. However, it is likely that several minutes would pass before
the police arrived to take your report. During that time period, the
14 The Psychology of Eyewitness Identification

information would need to be retained in your memory. After the initial

police report is taken, you might be called on to answer questions again
once a suspect had been caught. This may be days, weeks, or months
later. Before trial the prosecutor might review your testimony with you.
If the case goes to trial, you would be called on to answer questions
under oath. You might recall the event in other circumstances too. For
instance, you might recount the event in order to explain to your friends
and relatives why your trip to Binghamton had been interrupted. In
between each of these remindings, a representation of the details of
the event would have to be stored in your memory. Research indicates
that while in storage, memories are subject to distortion in the form
of postevent information (Lindsay, 1993; Loftus, 1991; Okado & Stark,
2005). Postevent information is information that a person encounters
after an event that is relevant to the event. For instance, a witness may
read newspaper accounts of the crime or may hear what other witnesses
said about the crime. Each of these encounters may provide details that
become incorporated into the witness’s memory for the event.
The final stage of memory is known as retrieval (Shiffrin & Atkinson,
1969). According to cue-dependent theories of memory, the act of
retrieval will occur successfully only to the extent that contextual details
that were available at the time of the original event are reactivated in
the person’s mind at the time he or she is trying to remember the event
(Reingold, 2002; Tulving & Thompson, 1973). In other words, you will
remember an event best when you are able to re-create in your mind the
full context of the original event that you are trying to remember. The
more contextual details, the better. The notion that retrieval depends
crucially on the quality of the contextual cues available at retrieval
provides some of the theoretical grounding for the cognitive interview
technique developed by Fisher and Geiselman (1992; see Chapter 4 for a
fuller description). Recommendations that have been developed by the
U.S. Department of Justice regarding the interviewing of eyewitnesses
explicitly recognize the fact that accurate memory depends on maxi-
mizing the number of retrieval cues available to the witness (Technical
Working Group, 1999).
The retrieval stage of memory is also a time when memories are sub-
ject to distortion. In fact, the very word retrieval is misleading. A more
apt word is reconstruction (Lampinen & Neuschatz, 2008). Consider the
description of remembering offered by Sir Frederic Bartlett (1932) in his
classic book Remembering: A Study in Experimental and Social Psychology:

Remembering is not a completely independent function, entirely

distinct from perceiving, imaging, or even from constructive think-
ing, but it has intimate relations with them all.… One’s memory of
an event reflects a blend of information contained in specific traces
 Theoretical Approaches to Eyewitness Identification 15

encoded at the time it occurred, plus inferences based on knowl-

edge, expectations, beliefs, and attitudes derived from other sources.

As Barlett’s description implies, when you remember an event,

you make use of a number of different types of information. You use
knowledge structures called scripts and schemas that provide informa-
tion about what typically happens during an event (Bower, Black, &
Turner, 1979). For instance, your script for going to a baseball game
might include knowledge that baseball games include a seventh-
inning stretch and that people often order hot dogs at baseball games.
Reconstruction also makes use of postevent information (Loftus, 1979).
For instance, after leaving a baseball game you may read about it in
the newspaper the next day, or may talk to a friend who was also at
the game. Memory reconstruction can also lead to borrowing specific
details from one event and incorporating them into one’s memory for
a related event (Odegard & Lampinen, 2004). For instance, you may
remember an obnoxious fan screaming obscenities at one baseball
game, when in fact the obnoxious fan screamed obscenities at a differ-
ent baseball game. In attempting to remember the baseball game, you
would reconstruct the event by combining information from all of these
sources. Thought of in this way, remembering is a dynamic decision-
making process that attempts to approximate the past by making use
of all available information stored in memory to create a version of the
past that is maximally coherent (see the Appendix for some philosophi-
cal implications of this view).
A good analogy that has been made by a number of different people is
that the act of reconstruction is a lot like a paleontologist trying to make
sense of a bunch of dinosaur bones, given a limited fossil record. Here is
how Daniel Schacter (1996) described the process:

For the paleontologist, the bone chips that are recovered on an arche-
ological dig and the dinosaur that is ultimately reconstructed from
them are not the same thing: the full blown dinosaur is constructed
by combining the bone chips with other available fragments, in
accordance with general knowledge of how the complete dinosaur
should appear. Similarly, for the rememberer, the engram (the stored
fragment of an episode) and the memory (the subjective experience
of recollecting a past event) are not the same thing. The stored frag-
ments contribute to the conscious experience of remembering, but
they are only part of it. (pp. 69–70)

So “retrieving” an event is actually “reconstructing” the event. Recent

research has shown that details from one event can be incorporated
into memories for another event, in a process called content borrowing, to
16 The Psychology of Eyewitness Identification

create vivid but false memories (Lampinen, Meier, Arnal, & Leding, 2005;
Lampinen, Ryals, & Smith, 2008). And, indeed, recent evidence from neu-
roscience has shown that memories can oftentimes be quite labile during
the retrieval stage (Lopez, 2000).

Varieties of Human Memory

In addition to the distinction between different functional stages of mem-
ory, memory psychologists also recognize a variety of different types of
memory. Perhaps the best known of these distinctions was made in a clas-
sic paper by Atkinson and Shiffrin in the late 1960s (Atkinson & Shiffrin,
1968). According to Atkinson and Shiffrin’s theory, representations of
events go through three separate memory buffers known as sensory mem-
ory, short-term memory, and long-term memory (see Figure 2.1). In the above
example, when the man approaches you with a gun, an image of the man’s
face is perceptually processed and stored in a memory buffer for a frac-
tion of a second. This memory buffer is known as sensory memory and
includes the basic perceptual details of the event. According to Atkinson
and Shiffrin, some information from the sensory memory buffer is then
transferred into short-term memory. Short-term memory stores informa-
tion for a few seconds unless it is actively rehearsed. One major limitation
on human cognition is that short-term memory has a limited capacity.
For instance, it is typically estimated that a person can hold around seven
items in short-term memory at a time (Miller, 1956). Because of this limi-
tation, only a fraction of the information in sensory storage actually gets
passed to short-term memory for detailed analysis. Finally, when infor-
mation in short-term memory is processed at a sufficient level of depth,
it can be passed on to long-term memory. Long-term memory contains
information from one’s entire life, and some psychologists even believe

Short Long
Sensory
Term Term
Memory
Memory Memory

FIGURE 2.1 Components of the Atkinson and Shiffrin (1968) model

of memory. (Source: Atkinson, R. C., & Shiffrin, R. M. (1968). Human
memory: A proposed system and its control processes. In K. W. Spence
& J. T. Spence (eds.), The psychology of learning and motivation (Vol. 1,
pp. 89–195). New York: Academic Press.)
 Theoretical Approaches to Eyewitness Identification 17

Central
Executive

Visuospatial Phonological
Sketchpad Loop

Episodic
Buffer

FIGURE 2.2 Components of the working memory model. (Source:

Based on figure originally presented by Baddeley, A. D. (2000). The
episodic buffer: A new component of working memory? Trends in
Cognitive Science, 4, 417–423.)

that every piece of information that has ever been passed on to long-
term memory is still stored in long-term memory (Shiffrin & Atkinson,
1969; Tulving, 1974; but see Loftus & Loftus, 1980). Long-term memory,
according to this theory, is subject to retrieval failure. That is, although
the relevant information may be present in long-term memory, without
the appropriate retrieval cues, the information may never be retrieved.
Many of the details in Atkinson and Shiffrin’s (1968) model have been
contested, but the model still provides a nice way of thinking about the
basic way human memory is organized. Subsequent research also elab-
orated on many of the ideas in the model. For instance, Baddeley and
Hitch (1974) described a model of working memory that in many ways
has replaced Atkinson and Shiffrin’s (1968) simpler idea of how short-
term memory works (see Figure 2.2). The Baddeley and Hitch (1974) model
made two important contributions to our understanding of memory. First,
at a functional level, Baddeley and Hitch (1974) argued that working mem-
ory was the place where problem solving and decision making occurred.
That is, working memory, as the name implies, was the place where the
actual work of reasoning with mental representations and memories hap-
pens. So when the police ask you which direction the perpetrator came
from, you might replay the scene in your mind’s eye. That replay of the
event would be occurring in working memory. The second contribution of
Baddeley and Hitch’s (1974) model is that it held that working memory is
actually made up of three components. One component of working mem-
ory processes primarily verbal information. This component is known
as the phonological loop (Baddeley, Gathercole, & Papagno, 1998). Another
component of working memory primarily processes visual information.
18 The Psychology of Eyewitness Identification

This component is known as the visuospatial sketchpad (Logie, 1995).

Finally, another component is involved in executive management of the
whole system, reasoning, and decision making. This component is called
the central executive (Baddeley, 1996). About 30 years after proposing the
original theory, Baddeley (2000) proposed another component of working
memory that he termed the episodic buffer. The episodic buffer is respon-
sible for combining information from the other buffers into a unified con-
scious experience.
When the attacker approaches you in the bus terminal, you register his
face in the visual spatial sketchpad. When he tells you to give him your
money or he’s going to shoot you, you process his words in the phonologi-
cal loop. The verbal and visual information are integrated into a unified
perceptual experience within the episodic buffer. When you decide on
your course of action (i.e., comply with his demands), you are likely using
the central executive.
Theorists have also made distinctions among different types of long-
term memory. The best known of these is the multiple memory system
proposed by Endel Tulving (Tulving, 1972, 1983, 1999). According to
Tulving, long-term memory can be thought of as being made up of three
partially overlapping systems. The first of these is procedural memory.
Procedural memory stores information about how to perform physical
and mental tasks such as swinging a golf club, doing long division, or tak-
ing your wallet out of your pocket to hand to a robber. Semantic memory
stores factual information that lacks details about where that information
was encountered. For instance, you might know that Columbus traveled
to America thinking he was en route to India, but it is unlikely that you
remember the specific event of learning that information. It is just a fact
that you know. Similarly, you know your name, but you almost certainly
don’t remember the specific event wherein you first learned your name.
Lastly, Tulving argued that there is a memory system known as episodic
memory. Episodic memory involves memories for specific events from
your life, such as the last movie you went to, the time you got robbed at
the bus station in Chicago, or the jerk yelling obscenities at the baseball
game. Tulving (2002) has referred to episodic memory as a kind of “mental
time travel” because it involves mentally re-creating the original event.
What the eyewitness is typically remembering are details about specific
events that occurred at a particular time and place. Consequently, almost
all research on eyewitness memory involves episodic memory.

Signal Detection Theory

In addition to making distinctions between different stages of memory
processing and different types of memory storage, an understanding of
 Theoretical Approaches to Eyewitness Identification 19

eyewitness memory requires a understanding of how psychologists think

about the basic processes involved in making memory judgments. Imagine
that the police approach you some time after the incident at the bus station
and show you a picture of an individual that they suspect may be the cul-
prit (i.e., showup; Steblay, Dysart, Fulero, & Lindsay, 2001). You are asked,
“Is this the person who robbed you?” How do you make the decision?
One important theory that can be useful in analyzing these situations is
signal detection theory (Parks, 1966). To illustrate signal detection theory,
imagine you decide to conduct the following experiment. You obtain a
sample of participants, and each participant is shown yearbook-style pho-
tographs of 50 people. Each picture is shown for 2 seconds, and you ask
the participants to try to remember the faces they are shown. Participants
then engage in an unrelated task for 10 minutes (known as the retention
interval). After the retention interval, you give them a test. The test is made
up of 100 pictures. Fifty of the pictures are of the people that had just
been studied (old pictures), and 50 pictures are of never-before-seen peo-
ple (new pictures). For each picture, you ask your participants to indicate
whether the picture is old or new. Tests like these are called old/new recogni-
tion tests.
Given the setup of your experiment, there are four possible outcomes.
You can see these outcomes in Figure 2.3.

• Participants can say “old” when an item really is old. These correct
responses are called hits by memory psychologists. In eyewitness
research, this is similar to a witness who correctly identifies the culprit.
• Participants can say “old” when an item is really new. These incor-
rect responses are called false alarms by memory psychologists. In
eyewitness research, this is similar to a witness who mistakenly
identifies an innocent person.
• Participants can say “new” when an item is really old. These incorrect
responses are called misses by memory psychologists. In eyewitness

Face was seen Face was not seen

Witness says, “I
HIT FALSE ALARM
saw that face.”
(FA)

Witness says, “I
CORRECT
didn’t see that MISS
REJECTION
face.”

FIGURE 2.3 Possible outcomes on an old/new recognition test.

20 The Psychology of Eyewitness Identification

research, this is similar to a witness who fails to identify the culprit

and the guilty person consequently goes free.
• Lastly participants can say “new” when an item really is new. These
correct responses are called correct rejections. In eyewitness research,
this is similar to a witness who correctly says the culprit is not there.

Typically, memory researchers report only the hit rate and the false alarm
rate, because the miss rate and correct rejection rate can be derived easily
from the hit and false alarm rates. For instance, if there were 50 previ-
ously studied pictures and the participant had 40 hits, then there must
have been 10 misses. And if there were 50 pictures that were not studied,
and the participant had 20 false alarms, there must have been 30 correct
rejections.
So what kinds of mental gyrations do people go through to decide
whether or not an item was studied? According to signal detection theory,
when an item such as a picture is shown on a recognition memory test, that
item is compared to representations stored in memory. If an item on the
memory test is very similar to one or more items stored in memory, it will
produce a strong subjective sense of familiarity or memory strength. If an item
on a memory test is not very similar to anything in memory, it will produce
a weaker feeling of familiarity or memory strength. It is important to note
that both old items and new items will produce some sense of familiarity
that can be thought of as varying along a continuum from weak to strong.
When considering the entire set of items, both the familiarity of the
new items and the familiarity of the old items will be normally distrib-
uted, with the old items being more familiar on average than the new
items, but with some overlap existing between the distributions. You can
see a graphical depiction of this situation in Figure 2.4.
The average familiarity of the old items will be greater than the aver-
age familiarity of the new items, but the trick is to decide for any par-
ticular item whether it is old or new. The better the person’s memory, the
further apart these two distributions will be, and the easier it will be for
the person to distinguish between the old and the new items. In signal
detection theory, the distance between the two distributions is known as
d´ (pronounced D prime). Technically, d´ is the distance between the distri-
butions measured in standard deviation units. Large values of d´ reflect
good memory (see Figure 2.5). Small values of d´ reflect poor memory.
A participant who is completely incapable of distinguishing old items
from new items would have a d´ score of 0. A participant who has a great
memory might have a d´ score of 1 or 2 or 3. If a participant had more false
alarms than hits, they would have a negative d´ score.
The measure of memory strength in signal detection theory (d´) is only
part of the story. According to signal detection theory, a participant’s task
on a recognition memory test is difficult and ambiguous. This is because
 0
0.1
0.2
0.3
0.4

0.05
0.15
0.25
0.35
0.45
0

0
0.1
0.2
0.3
0.4

0.1
0.2
0.3
0.4
0.05
0.15
0.25
0.35
0.45

0.05
0.15
0.25
0.35
0.45
–5 –5 –5
–4.78 –4.78 –4.78
–4.56 –4.56 –4.56
–4.34 –4.34 –4.34
–4.12 –4.12 –4.12
–3.9 –3.9 –3.9
–3.68 –3.68 –3.68
–3.46 –3.46 –3.46
–3.24 –3.24 –3.24

Old
Old

Old

New
–3.02
New

New
–3.02 –3.02
–2.8 –2.8 –2.8
–2.58 –2.58 –2.58
–2.36 –2.36 –2.36
–2.14 –2.14 –2.14
–1.92 –1.92 –1.92
–1.7 –1.7 –1.7
–1.48 –1.48 –1.48
–1.26 –1.26 –1.26
–1.04 –1.04 –1.04
–0.82 –0.82 –0.82
–0.6 –0.6 –0.6
–0.38 –0.38 –0.38
–0.16 –0.16 –0.16

FIGURE 2.4 Signal detection theory.

0.06 0.06 0.06

Correct
0.28 0.28 0.28

Say “New”

d´
Say “New”

Say “New”
d´

d´
0.5 0.5 0.5

Rejections FAs
Misses
0.72 0.72 0.72
0.94 0.94 0.94
1.16 1.16 1.16
1.38 1.38 1.38
1.6 1.6 1.6
1.82 1.82 1.82

Hits
2.04 2.04 2.04
2.26 2.26 2.26

Say “Old”
Say “Old”

Say “Old”

2.48 2.48 2.48

2.7 2.7 2.7
2.92 2.92 2.92
3.14 3.14 3.14
3.36 3.36 3.36
3.58 3.58 3.58
3.8 3.8 3.8
4.02 4.02 4.02
4.24 4.24 4.24
4.46 4.46 4.46
4.68 4.68 4.68
4.9 4.9 4.9
Theoretical Approaches to Eyewitness Identification
21
22 The Psychology of Eyewitness Identification

Say “New” Say “Old”

0.45 d´ = 0.2

0.4

0.35

0.3

0.25 Old

New
0.2

0.15

0.1

0.05

0
–5
–4.77
–4.54
–4.31
–4.08
–3.85
–3.62
–3.39
–3.16
–2.93
–2.7
–2.47
–2.24
–2.01
–1.78
–1.55
–1.32
–1.09
–0.86
–0.63
–0.4
–0.17
0.06
0.29
0.52
0.75
0.98
1.21
1.44
1.67
1.9
2.13
2.36
2.59
2.82
3.05
3.28
3.51
3.74
3.97
4.2
4.43
4.66
4.89
Say “New” Say “Old”

d´ = 1
0.45

0.4

0.35

0.3

Old
0.25
New

0.2

0.15

0.1

0.05

0
–5
–4.78
–4.56
–4.34
–4.12
–3.9
–3.68
–3.46
–3.24
–3.02
–2.8
–2.58
–2.36
–2.14
–1.92
–1.7
–1.48
–1.26
–1.04
–0.82
–0.6
–0.38
–0.16
0.06
0.28
0.5
0.72
0.94
1.16
1.38
1.6
1.82
2.04
2.26
2.48
2.7
2.92
3.14
3.36
3.58
3.8
4.02
4.24
4.46
4.68
4.9

FIGURE 2.5 Graphs illustrating changes in d’ in signal detection the-

ory. The top figure illustrates poor memory and a low value for d’. The
bottom figure shows someone with better memory and a higher value
for d’.

the distributions for the two types of items (old items and new items) over-
lap. Overall, old items are more familiar than new items, but some new
items are more familiar than some old items. Because of this, participants
have to decide how much familiarity a face has to have before they are
 Theoretical Approaches to Eyewitness Identification 23

Say “New” Say “Old”

Say “New” Say “Old”
0.45 d´ = 1 0.45 d´ = 1

0.4 0.4
Old Old

0.35 New 0.35 New

0.3 0.3

0.25 0.25

0.2 0.2
zc = –1.55 zc = 0

0.15 0.15

0.1 0.1

0.05 0.05

0 0
–4.77
–4.54
–4.31
–4.08
–3.85
–3.62
–3.39
–3.16
–2.93
–2.47
–2.24
–2.01
–1.78
–1.55
–1.32
–1.09
–0.86
–0.63
–0.17

–4.77
–4.54
–4.31
–4.08
–3.85
–3.62
–3.39
–3.16
–2.93
–2.47
–2.24
–2.01
–1.78
–1.55
–1.32
–1.09
–0.86
–0.63
–0.17
–2.7

–0.4
0.06
0.29
0.52
0.75
0.98
1.21
1.44
1.67
2.13
2.36
2.59
2.82
3.05
3.28
3.51
3.74
3.97
4.43
4.66
4.89

–2.7

–0.4
0.06
0.29
0.52
0.75
0.98
1.21
1.44
1.67
2.13
2.36
2.59
2.82
3.05
3.28
3.51
3.74
3.97
4.43
4.66
4.89
1.9

4.2

1.9

4.2
–5

–5
Say “New” Say “Old” Say “New” Say “Old”

0.45 d´ = 1 0.45 d´ = 1

0.4 0.4
Old Old

0.35 New 0.35 New

0.3 0.3

0.25 0.25
zc = 1.22 zc = 2.82
0.2 0.2

0.15 0.15

0.1 0.1

0.05 0.05

0 0
–4.77
–4.54
–4.31
–4.08
–3.85
–3.62
–3.39
–3.16
–2.93
–2.47
–2.24
–2.01
–1.78
–1.55
–1.32
–1.09
–0.86
–0.63
–0.17

–4.77
–4.54
–4.31
–4.08
–3.85
–3.62
–3.39
–3.16
–2.93
–2.47
–2.24
–2.01
–1.78
–1.55
–1.32
–1.09
–0.86
–0.63
–0.17
–2.7

–0.4
0.06
0.29
0.52
0.75
0.98
1.21
1.44
1.67
2.13
2.36
2.59
2.82
3.05
3.28
3.51
3.74
3.97
4.43
4.66
4.89

–2.7

–0.4
0.06
0.29
0.52
0.75
0.98
1.21
1.44
1.67
2.13
2.36
2.59
2.82
3.05
3.28
3.51
3.74
3.97
4.43
4.66
4.89
1.9

4.2

1.9

4.2
–5

–5

FIGURE 2.6 Graphs illustrating changes in response criterion in signal

detection theory. Graph on top left illustrates a very liberal decision
criterion. Graph on bottom right illustrates a very conservative deci-
sion criterion.

willing to say it was studied. This is known as the participant’s response

criterion (sometimes this is called the response bias). The response criterion
is shown in the graphs as a straight horizontal line indicating the famil-
iarity that is needed before the participant will indicate that an item was
studied. Anything that is more familiar than the criterion will be called
“old.” Anything less familiar than the criterion will be called “new.”
The placement of the response criterion is up to the individual.
You can see the consequences of setting different response criteria in
Figure 2.6. Participants can decide to be very liberal, and respond “old”
even if the degree of familiarity is relatively low. This will increase
the proportion of hits, but will come at a cost of more false alarms. A
participant might decide to be very conservative and respond “old”
only when the degree of familiarity is very high. This will limit false
24 The Psychology of Eyewitness Identification

alarms but come at the cost of fewer hits. Or a participant may adopt
a neutral response criterion and split the difference between the two
types of errors.
According to signal detection theory, a particular participant’s perfor-
mance on an old/new recognition test is characterized both by how good
the participant’s memory is (i.e., d´) and by the response criterion they
decide to use.
Imagine you do the study where students study yearbook photos. After
the study phase and retention interval, you hypnotize half of the partici-
pants in an attempt to improve their memories (see Chapter 4). You then
give them an old/new recognition test. Imagine that the hit rate for stu-
dents who were not hypnotized is 60% and the hit rate for those who were
hypnotized is 80%. Considered in isolation, that sounds like the hypnosis
was a pretty big help. However, if you find out that the false alarm rate
in the no-hypnosis group is 5% and the false alarm rate in the hypnosis
group is 15%, you might change your mind. From the standpoint of signal
detection theory, we want to know if those differences reflect changes in
memory strength, changes in the response bias, or both. Luckily signal
detection theory makes it possible to estimate memory strength (d´) by
looking at the hit rates and false alarm rates combined. Details of the cal-
culations can be found in Snodgrass and Corwin (1998) as well as several
other references. In this example, we can calculate d´ for the hypnosis
group as 1.88, and we can calculate d´ for the not-hypnotized group as
1.90. So in this hypothetical example, hypnosis had no appreciable effect
on memory. It changed only the response criterion that participants were
setting. Participants set a more liberal response criterion when they were
hypnotized. This led to more hits and more false alarms in our hypotheti-
cal example.
Throughout chapters later in this monograph, we will make use of
insights from signal detection theory. It is important to note that memory
researchers sometimes debate some of the technical details underlying
signal detection theory (Ratcliff, Sheu, & Gronlund, 1992; Singer, 2009;
Yonelinas, 1994). Those debates are important, but are beyond the scope of
the current volume. For the current volume, the most important point is
that signal detection theory provides us with a way of estimating memory
strength and response criterion, with large values of d´ reflecting better
memory.

Dual Process Theories

Signal detection theory is a useful starting point for understanding how
eyewitness memory works. However, one limitation of signal detection
theory is that it fails to fully account for the variety of different memory
 Theoretical Approaches to Eyewitness Identification 25

experiences that people have. In signal detection theory, memory is seen

as being represented along a single dimension of memory strength or
familiarity. Many memory theorists believe that this is an oversimplifica-
tion (Diana, Reder, Arndt, & Park, 2006; Jacoby, 1991; Joordens & Hockley,
2000; Lampinen, Watkins, & Odegard, 2006; Mandler, 1980; Yonelinas,
1994). These theorists think it makes sense to distinguish among qualita-
tively different types of memory. The best known of these theories posit
the existence of two qualitatively distinct memory processes. Thus, these
theories are known as dual process theories.
One of the most influential dual process models was proposed by
George Mandler (1980). According to Mandler, recognition memory (such
as recognizing a face) can occur in two distinct ways: recall and familiar-
ity. Recall involves the conscious re-experiencing of the item in the context
in which it was originally experienced. For instance, a student once told
one of us (JML) that he could remember the exact wording of the lame joke
one of us (JML) told when introducing a concept in class. Mandler distin-
guished this type of recognition memory from what he called familiar-
ity. Familiarity involves a feeling or intuition that the item was previously
presented, in the absence of an exact memory for the context in which
the item was presented. Mandler gave the example of seeing someone on
a bus, knowing that you have seen the person before, but not being sure
exactly where you know them from (a common experience for most of us).
According to Mandler, feelings of familiarity occur because the individ-
ual features that make up the item are tightly bound together in memory.
Thus, when you see the person on the bus, you know that you have previ-
ously seen that exact combination of eyes, nose, mouth, facial shape, hair-
line, and so on, but you cannot recall exactly where you saw that person
before. Mandler called these types of associations intra-item associations,
that is, associations between the features that make up the item. In recall,
recognition of the item occurs because there is a tight connection between
the individual item and the context in which it was originally encoun-
tered. Thus, when you see the person’s face, you can also remember that
you saw the person at a Halloween party and that the person was dressed
as Humpty Dumpty and made a drunken fool of himself. Such connec-
tions between the item and the context in which the item occurred are
known as inter-item associations.
Mandler’s view of memory was influential and led to the development
of other important dual process models. One of the most influential was
Jacoby’s (1991) process dissociation model. According to Jacoby, remembering
an item in the context of its original occurrence can be thought of as recol-
lection, whereas having a feeling that an item was previously presented
devoid of the context can be thought of as familiarity. One important con-
cept that Jacoby introduced was the idea that familiarity, at least some
of the time, occurs because prior experience with an item makes the
26 The Psychology of Eyewitness Identification

perceptual processing of that item easier in the future (Jacoby & Dallas,
1981). Jacoby referred to this idea as perceptual fluency. Mental processing
of any individual item (e.g., a face, word, or song) requires some time and
effort from your perceptual system. However, when you are presented
with an item that you have previously experienced, it is a little easier for
your perceptual system to process the item on the second go around. You
are not consciously aware of this increased fluency of processing; how-
ever, subjectively the increased ease of processing gives rise to a subjec-
tive feeling of familiarity. Jacoby thus argued that recognition memory
sometimes involves an inference that we make based on how easy or dif-
ficult it is to perceptually process an item.
Another important contribution of process dissociation theory is that
it provides a technique for teasing apart the contributions of recollec-
tion and familiarity. The basic principle is that recollection allows you
to make a response that depends upon the context in which the item
was presented, whereas familiarity does not allow this kind of context-
dependent response. For instance, imagine I show you pictures of 100
people. Half of the people are holding guns (i.e., bad people), and half of
the people are holding kittens (i.e., nice people). Imagine that when you
return to the lab on the following day, I show you 200 pictures of people
who aren’t holding anything. One hundred of the pictures are pictures
of the people you saw the previous day. One hundred of the pictures are
of completely new people. As I show each picture, I ask you to say “yes”
only to people who were holding guns. Successful performance of this
task requires conscious recollection. That is, you have to remember not
just that a particular face was presented but also that it was presented in
the context of a gun.
Jacoby (1991) used this insight to create a formal way of measuring
recollection and familiarity called the process dissociation procedure. The
process dissociation procedure takes advantage of the logic of opposition.
Participants take two kinds of tests, called inclusion and exclusion tests.
Given the example we have been working with, the two tests might be set
up as follows.

• Inclusion test: The test is made up of 50 nice people, 50 bad people,

and 100 new people. You are asked to say “no” to any person whose
picture was not shown on the previous day. You are asked to say
“yes” to any person whose picture you saw the previous day, regard-
less of whether they were holding a gun or a kitten. Thus you should
include both people who were holding guns and people who were
holding kittens.
• Exclusion test: This test is made up of 50 good people, 50 bad people,
and 100 new people. You are asked to say “yes” to any person who
you saw holding a gun the previous day, and to say “no” to anybody
 Theoretical Approaches to Eyewitness Identification 27

who was holding a kitten or anybody whose picture wasn’t present

at all. Thus you should exclude people who were holding kittens but
include people who were holding guns.

The key measure is how often you say “yes” to the people who had been
holding kittens. Here’s why: If you are using conscious recollection, you
should respond differently on the inclusion test and the exclusion test to
the people who were holding kittens on the previous day. On the inclusion
test, you should say “yes” to people who were holding kittens, whereas
on the exclusion test, you should say “no” to people who were holding
kittens. Performing this task requires recollection, because recollection is
defined as remembering the context in which an item was presented.
In Figure 2.7, the Venn diagrams are drawn to illustrate the probabil-
ity of saying “yes” in response to a person you saw holding a kitten on
the previous day. In the inclusion condition, you should include a person
holding a kitten if he or she is familiar. You should also include a person
holding a kitten if you consciously recollect him or her holding the kitten.
Thus, both of the circles are shaded. In the exclusion condition, you should
include people if they seem familiar, but you should exclude them if you
can consciously recollect that they were holding a kitten. By inspecting
the Venn diagrams you can see that, according to the theory, the probabil-
ity of conscious recollection (R) is simply the difference between saying
“yes” to people who were holding kittens in the inclusion (I) and exclu-
sion (E) conditions:

R=I–E (2.1)

R R

F F

Inclusion (I) Exclusion (E)

– R = R
R

F F
F

FIGURE 2.7 Estimating recollection using the process dissociation

procedure.
28 The Psychology of Eyewitness Identification

R = R – R

F F F

Exclusion (E) Familiarity (F) Joint Probability of

R and F (RXF)

FIGURE 2.8 Estimating familiarity using the process dissociation pro-

cedure.

Let’s say that when people are shown pictures of individuals they had pre-
viously seen holding kittens, they say “yes” to those pictures 60% of the
time in the inclusion condition and 20% of the time in the exclusion con-
dition. Then, according to the logic of opposition, people are consciously
recollecting the context of the pictures 40% of the time (60 − 20 = 40).
It is also possible to estimate familiarity. To do so, you need to know
a little probability theory and you need to apply a little algebra. First,
examine that portion of the Venn diagrams where the R circle and the F
circle overlap. That’s called the joint probability of R and F. Simply put, it’s
the probability of both recollection and familiarity occurring at the same
time. When two events are independent (i.e., they don’t influence each
other), the joint probability of two events is equal to the product of their
individual probabilities. In other words, the probability that both recollec-
tion and familiarity will occur at the same time is the probability of recol-
lection occurring multiplied by the probability of familiarity occurring.
The situation is shown in Figure 2.8.
The equation corresponding to the situation illustrated in Figure 2.8 is

E = F − RF (2.2)

When you solve for F, the equation becomes

F = E / (1 − R) (2.3)

Earlier we had said that the probability of saying “yes” in the exclusion
condition was 20% (.20), and we estimated recollection to be 40% (.40). So
our estimate of familiarity becomes

F = 0.20 / (1 − 0.40) = .3333 or 33.33% (2.4)

So in this hypothetical example, if you are shown a picture on a memory

test, there is a 33.33% chance that it will produce a feeling of familiarity
 Theoretical Approaches to Eyewitness Identification 29

in you that will be sufficient to allow you to recognize that it was on the
study list the previous day. And there is a 40% chance that you will con-
sciously recollect the context in which the face was presented.
The distinction between recollection and familiarity is an important
one for a couple of different reasons. First, it is typically important that a
witness not simply recognize a person but also be able to recognize the
context in which he or she saw that person. For instance, in a phenomenon
known as unconscious transference, an innocent person seen in one con-
text may mistakenly be remembered as having been seen committing the
crime (Phillips, Geiselman, Haghighi, & Lin, 1997; Ross, Ceci, Dunning, &
Toglia, 1994).1 Second, recollection is usually associated with higher levels
of subjective confidence than is familiarity (Yonelinas, 2001). Confidence
is one of the best predictors of jurors’ judgments of the credibility of a wit-
ness (Cutler, Penrod, & Dexter, 1990; Fox & Walters, 1986; Lindsay, Wells,
& Rumpel, 1981).
In many cases, researchers may not want to use the process dissocia-
tion procedure to estimate recollection. For one thing, one needs a large
number of observations to get estimates using the technique. For another,
the technique is controversial among memory psychologists (in particu-
lar, the assumption that recollection and familiarity are independent is a
point of contention) (Curran & Hintzman, 1995; Dodson & Johnson, 1996;
Graf & Komatsu, 1995; for replies, see Jacoby, 1997; Jacoby, Yonelinas, &
Jennings, 1997). Because of this, alternative methods of measuring rec-
ollection and familiarity are sometimes used. One of these methods is
called the remember/know technique (Gardiner, 1988; Gardiner & Java,
1990, 1991; Lampinen, Copeland, & Neuschatz, 2001; Lampinen, Faries,
Neuschatz, & Toglia, 2000; Neuschatz et al., 2004; Rajaram, 1993, 1998;
Tulving, 1985). In the remember/know technique, participants are asked
to indicate whether an item is old or new. If they say it is old, they are
asked whether they can recall specific details of the item’s previous pre-
sentation, or if they believe the item is old for some other reason (e.g.,
it just seems familiar). If a participant believes that he or she can recall
specific details of the item’s presentation, then he or she makes a remember
judgment. If a participant believes an item is old, but does not remember
anything specific about the item’s presentation, the participant makes a
know judgment. Thus remember judgments can be used to indicate con-
scious recollection, whereas know judgments are used to indicate feelings
of familiarity. Remember/know judgments provide an alternative way of
measuring recollection and familiarity based on participants’ self-reports.
Of course, self-report data are open to challenge, and so the best approach
is to come at the question from a variety of different perspectives and to
seek converging evidence (Lampinen, Neushatz, & Payne, 1998).
30 The Psychology of Eyewitness Identification

□ Theories of Face Perception and Memory

In the “Human Memory” section of this chapter, we described some basic
theoretical views of human memory that are useful for understanding
the applied issue of eyewitness identification. These theories concerned
human memory writ large. However, the focus of the present monograph
is specifically on the problem of eyewitness identification. Eyewitness
identification typically deals with seeing a face in one context and then
recognizing that face in a slightly different context. Because of this, it is
useful for someone interested in eyewitness identification to have some
basic knowledge of the processes involved in recognizing people’s faces.

Functional Accounts of Face Processing

Faces are complex and highly socially relevant stimuli (Bruce, 2002). Faces
are important because they convey information about gender, age, ethnic-
ity, health, emotion, and personal identity (Donath, 2001). Moreover, faces
are dynamic entities. They change as we move (O’Toole, Roark, & Abdi,
2002; Roark, Barrett, Spence, Abdi, & O’Toole, 2003) and change across
time as we age (Sadler, 1986; Taylor, 2001). Yet from one viewing to the
next, there are things about our faces that remain the same and allow for
identification. Theories of facial perception need to account for all of these
aspects of face perception.
A useful functional account of all of these components of the face-pro-
cessing problem was developed by Bruce and Young in 1986. According
to the Bruce and Young framework, face processing involves a set of
processing modules that operate on seven distinct kinds of facial codes.
The facial codes proposed in the framework are pictorial codes, struc-
tural codes, visually derived semantic codes, identity-specific seman-
tic codes, name codes, expression codes, and facial speech codes. The
pictorial code simply describes the person’s face as a picture. The picto-
rial code for a face is much like the visual representation of any sort of
object. Structural codes provide configural information about the person’s
face, such as the distance between the eyes or the relative symmetry of
the face as a whole. As we will see in this section, evidence indicates
that this sort of configural information is crucial for facial recognition.
Visually derived semantic codes include information about the person’s
characteristics, such as gender, age, attractiveness, and so on. Identity-
specific semantic information includes information specific to individuals,
such as what they do for a living or where they are typically encountered.
As the name implies, name codes provide information about the names
that are associated with particular faces. Expression codes represent the
 Theoretical Approaches to Eyewitness Identification 31

Expression View-centered
Analysis Descriptions

Structural
Facial Speech Encoding
Expression
Analysis Independent
Descriptions

Directed
Visual
Processing
Face Recognition
Units

Cognitive
Person Identity
Nodes

System

Name Generation

FIGURE 2.9 Bruce and Young’s (1986) theory of face perception.

(Source: Based on figure originally presented in Bruce, V., & Young, A.
(1986). Understanding face recognition. British Journal of Psychology, 77,
305–327.)

emotional expressions associated with people’s faces. Finally, Bruce and

Young noted that we also represent speech-related visual information in
what they called a facial speech code.
According to the framework, these seven types of codes are used by a
set of modules (shown in Figure 2.9) in order to make different types of
perceptual judgments about faces. In particular, for tasks in which rec-
ognition of the face is crucial (e.g., that was the person who robbed me at
the bus terminal), facial recognition begins by developing a viewpoint-
dependent representation of the person’s face. This viewpoint-dependent
representation is then converted to a more abstract, expression-indepen-
dent structural code. Once this structural code is developed, the infor-
mation is compared to information stored in face recognition units. Face
32 The Psychology of Eyewitness Identification

recognition units represent particular previously seen faces. These units

become active when the current face being judged has a similar structural
representation. When a particular face unit becomes sufficiently activated,
it may activate a particular person identity node containing identity-spe-
cific semantic information about the person we have just encountered.
Finally, the foregoing activation may also lead to the activation of a name
node allowing us to provide a name for the individual being identified.
The model proposed by Bruce and Young (1986) has considerable appeal
as a heuristic for thinking about the steps involved in face processing and
how facial recognition can fail.

Holistic Processing and Featural Processing of Faces

A great deal of evidence indicates that face processing involves both the
local identification of individual facial features and the processing of
higher order configural information (Bruce, Doyle, Dench, & Burton, 1991;
Cabezza & Kato, 2001; Farah, Wilson, Drain, & Tanaka, 1998). Configural
information refers to how individual parts of a face are related to one
another. This configural information allows for face recognition to occur
holistically rather than by analyzing each individual facial feature one at
a time. The terms configural and holistic are often used interchangeably,
but it is important to note that there are distinctions in the terms that
are important. Holistic processing basically refers to any type of processing
where the unit of analysis is the face as a whole. For instance, holistic pro-
cessing may involve noting the symmetry of the face (Chen, Kao, & Tyler,
2006), information about the relative locations of different parts of the
face (Burton, Bruce, & Johnsonton, 1990), or even characteristics such as
attractiveness, honesty, or kindness (Frowd, Bruce, & Hancock, 2008). The
term configural, on the other hand, refers more specifically to relational
information between the parts of the face. Configural processing often is
holistic, in that it may take into account the face as a whole, but may also
involve lower order relations such as how far apart the person’s nostrils
are. Such information is relational, and hence configural, but it does not
deal with the face as a whole.
A number of different types of experimental paradigms have been
used to examine the degree to which holistic information is used in facial
recognition. The classic demonstration of the importance of holistic infor-
mation in face processing is the inverted face effect (Diamond & Carey,
1986; Passarotti, Smith, DeLano, & Huang, 2007; Taylor, Batty, & Itier, 2004;
Sekunova & Barton, 2008; Yin, 1969). You can see what the materials for
these kinds of experiments look like in Figure 2.10. In these studies, par-
ticipants are presented with a target face that is either in an upright orien-
tation or an upside-down orientation. Following a brief delay, participants
 Theoretical Approaches to Eyewitness Identification 33

Flip over to see the real Jeff Neuschatz

FIGURE 2.10 Illustration of the inverted face paradigm.

make a recognition judgment of an upright or upside-down picture.

Research indicates that when making judgments about the upside-down
pictures, recognition is slower and less accurate. Studies have shown that
the decreased performance when faces are inverted occurs due to changes
in configural information. For instance, Leder and Bruce (2000) digitally
varied either the bushiness of the eyebrows (local feature) or the distance
between the eyes (configural feature). Changes in configural informa-
tion were harder for people to spot in upside-down pictures than were
changes in local feature information.
Another design which has been used to good effect to study holis-
tic processing is the composite face design (Cheung, Richler, Palmeri,
& Gauthier, 2008; Gauthier & Bukach, 2007; Goffaux & Rossion, 2006;
McKone, Kanwisher, & Duchaine, 2007). In this design, participants are
presented with pictures in which the top half of one person’s face is com-
bined with the bottom half of another person’s face. You can see sample
stimuli in Figure 2.11. A line separates the two halves, resulting in an illu-
sion that it is a single person’s face. In a typical experiment, participants
are shown composite faces for a brief period of time, followed by a visual
mask. Participants are then shown a composite face in which the entire
face matches, half the face matches (either the top half or the bottom half),
or none of the face matches. The participant’s task is to indicate if a specific
34 The Psychology of Eyewitness Identification

Aligned

Is the top half of the face the same?

Misaligned

Is the top half of the face the same?

FIGURE 2.11 Illustration of the composite face paradigm. In this exam-

ple, participants would be asked if the top halves of the composite faces
match. In all cases, the bottom halves of the face do not match.

half of the face matches. For instance, participants might be asked, “Does
the top half of the face match?” Imagine you are shown such a compos-
ite face and are then shown a new composite face in which the top half
matches and the bottom half does not match. You are asked to indicate
whether the top half of the face matches the picture you just saw. To the
extent that you are relying on holistic processing, this task should be diffi-
cult, because the face as a whole does not match, interfering with your abil-
ity to recognize that the individual features in the top half of the face do
match. Indeed, this is what researchers have typically found. Participants
performing this task show impaired performance when measured by
both accuracy and response time. The effect can further be demonstrated
by misaligning the two halves of the composite, as shown in Figure 2.11
(Goffaux & Rossion, 2006; Hole, 1994; Hole, George, & Dunsmore, 1999; Le
Grand, Mondloch, Maurer, & Brent, 2004; Michel, Rossion, Han, Chung, &
Caldara, 2006; Robbins & McKone, 2007; Young, Hellawell, & Hay, 1987).
When the two halves of the face are misaligned, it eliminates the holistic
information and the task of matching just the top half of the face becomes
pretty easy.
Evidence indicates that configural information greatly aids recognition
of faces. For instance, Davidoff and Donnelly (1990) presented participants
with intact faces or faces in which the location of the parts were scrambled
 Theoretical Approaches to Eyewitness Identification 35

and found that recognition of faces was impaired when the parts were
scrambled. Moreover, it has been found that providing verbal descrip-
tions of faces can sometimes impair recognition of those faces, a finding
termed the verbal overshadowing effect (Schooler & Engstler-Schooler, 1990).
Evidence indicates that part of the reason verbalization can impair face
recognition is because verbal descriptions typically focus on individual
facial features, thus impairing recognition of configural information
(Fallshore & Schooler, 1995).
Other studies have also been performed to assess the importance of
configural information. For instance, researchers have used computer
software programs to change the location of facial features by a small
number of pixels (Freire & Lee, 2001). Thus the new pictures are identi-
cal to the old pictures with regard to local featural information, but vary
slightly in their configural information. Research indicates that people are
quite good at picking up even small changes in configural information.
Taken together, all of this research suggests that both featural informa-
tion and configural information are used in face recognition. However,
configural information appears to be especially important. These results
will have important implications for our understanding of the ability of
witnesses to recognize perpetrators from lineups.

Neurobiology of Face Perception

In parallel with the important work being done on functional accounts
of human face processing, neuroscientists have made major advances in
understanding how the brain processes information about faces (Bruce
& Young, 2000; Grill-Spector, Knouf, & Kanwisher, 2004; Kanwisher,
McDermott, & Chun, 1997; Servos, Engel, Gati, & Menon, 1999; Xu, 2005).
Like all visual stimuli, information needed to recognize a face is first
processed in the retina, where the information is then sent to the lateral
geniculate nucleus of the thalamus and then to the primary visual cortex
in the occipital lobe of the cerebral cortex (Enns, 2004; Martin, 1994). The
primary visual cortex is known to have detailed maps of visual space in
which different locations in space are represented by particular groups of
neurons (known as cortical columns). Additionally, particular collections of
neurons (known as feature detectors) are known to respond preferentially
to particular orientations of lines (Hubel & Wiesel, 1959, 1962). Thus, at
its earliest stop in the visual cortex, a detailed mapping of facial features
is represented by collections of neurons that indicate the orientations of
lines that make up the person’s face.
From the primary visual cortex, visual information is sent to a number
of different submodules. However, neuroscientists have noted that there
appear to be two parallel pathways leaving the primary visual cortex
36 The Psychology of Eyewitness Identification

(Ettlinger, 1990; Goodale & Milner, 1992; Mishkin & Ungerleider, 1982).
One pathway sends information up toward the parietal lobe and repre-
sents information about object locations and about how the person can
interact with the object. This is the where/how pathway (also known as the
ventral route). The other pathway sends information toward the tempo-
ral lobe and represents information about object identity. This is the what
pathway (also known as the dorsal route). In particular, it is known that
an area known as the inferotemporal cortex is especially important for
object recognition, including the recognition of faces. In early studies, for
instance, Klover and Bucey (1938) found that if this area was destroyed in
monkeys, the monkeys lost the ability to recognize common objects. This
work was later extended by Gross, Roch-Mirranda, and Bender (1972).
They recorded the activity of particular groups of cells in the infero-
temporal cortex of monkeys. They found that there were some cells that
appeared to be specialized to recognize particular types of objects. For
instance, some cells responded only when the monkeys were shown a
silhouette of a hand. Other cells responded only to faces. There were even
cells that appeared to respond to the presentation of a toilet brush that the
monkeys frequently saw.
With regard to recognition of faces, it appears that a particular part
of the inferotemporal cortex known as the fusiform face area (FFA; see
Figure 2.12) is specialized for the recognition of faces (Kanwisher,
McDermott, & Chun, 1997). In studies with monkeys, the FFA shows acti-
vation only to intact faces and does not respond to jumbled faces or other

FFA

FIGURE 2.12 This drawing shows the medial surface of one cerebral
hemisphere with the approximate location of the fusiform face area
(FFA) indicated. The exact location of the FFA varies from individual to
individual but is typically located along the midfusiform gyrus. (Source:
Obtained at WikiCommons. Drawn by Mysid, based on the online ver-
sion of the 1918 Gray’s Anatomy.)
 Theoretical Approaches to Eyewitness Identification 37

complex stimuli (Haxby, Hoffman, & Gobbini, 2000). In fact, humans with
damage to the FFA show deficits in face recognition known as prosopag-
nosia (Barton, Press, Keenan, & O’Connor, 2002; Mayer & Rossion, 2007).
The evidence from neuroscience makes it clear that when witnesses are
processing a culprit’s face, this information processing is occurring in the
FFA as well as related cortical areas.

Face Space Theory

Before leaving the topic of research on face perception, one last theory
is important for our understanding of eyewitness identification. To
most people it seems like common sense that some faces are just easier
to remember than other faces. Some people just stand out in the crowd.
Such faces are referred to as distinctive (Metzger, 2006). But what exactly
is meant by saying that a face is distinctive? One important theory that
has been used to try to explain this concept is Valentine’s (1995) face space
theory. The concept behind face space theory is that individual faces that
one has encountered are stored in memory as exemplars. The exemplars
can be thought of mathematically as inhabiting a kind of multidimen-
sional space. Exemplars are stored around the prototypical face with faces
that are more distinct being further away from the prototype. These dis-
tinctive faces are easier to recognize because there are fewer similar faces
surrounding them in the face space and so there is a lower risk of mix-
ing up the face you saw with another similar face. Valentine’s theory has
been used in a number of different contexts, including in attempting to
explain the finding that members of other races are more difficult to rec-
ognize than members of one’s own race (Valentine, 1991). According to
Valentine’s theory, this occurs because the dimensions along which mem-
bers of other races vary are different from the dimensions along which
one’s own race varies (see Figure 2.13). Member of other races therefore
end up being clustered very close together, making it difficult to make
distinctions among the exemplars.

Conceptions of the Lineup Task

The theoretical accounts described thus far are basic theories of memory
and perception. At a general level, it is useful for anyone interested in the
science of eyewitness identification to have a basic understanding of these
theories. However, as research on eyewitness identification heated up in
the 1970s and 1980s, specific theoretical frameworks were developed that
were specifically applied to the task of making identifications in lineups
Exploring the Variety of Random
Documents with Different Content
language of Goethe, as he depicted himself passing through a similar
phase:
Zwei Seelen wohnen, ach, in meiner Brust.

One soul leaned to sloth, sensual and selfish indulgence: one, with
larger horizon, was for temperance, vigour, and Imperial duty. The
curious feature of this critical stage in the fortunes of Rome is that the
two tendencies are developed within the same family, and the Stoic
yields to the Sybarite. Annia Galeria Faustina was born of the same
parents as the father of Marcus Aurelius, and was reared in the same
atmosphere of old Roman virtue, or manliness, as the word signifies.
The great-grandfather of Marcus Aurelius was Annius Verus, a Senator
of great merit and of Spanish extraction. His son Annius Verus was
twice consul, and both his sons in turn—the father and uncle of Marcus
Aurelius—were promoted to the consulate. Everything we know of the
family suggests a fine and sober patrician type, and confirms the
beautiful picture of it given us by Marcus Aurelius in his “Meditations.”
The one element of possible weakness in the ancestry of the
Faustinas and of Commodus is in the mother of Annia Galeria Faustina.
Annius Verus had married Rupilia Faustina. Her family is obscure, and,
though one must hesitate to trace to her this strain of weakness and
vice on such slender grounds, one is disposed to believe that she was
married for her beauty, and brought into that strong family the tainted
germ which ripened in more than one of her descendants. It may,
however, very well be that the strength of the stock was decaying—
Marcus Aurelius himself was delicate—and its later descendants
succumbed to the evil influences about them. A genealogical table will
show how the fate of Rome hung on this family for more than a
generation:—
FAUSTINA THE ELDER
BUST IN THE LOUVRE
 Annius Verus (twice consul)
and Rupilia Faustina
|
+--------------------+----------------------------+
| | |
Annius Libo Annius Verus (consul) Annia Galeria Faustina
(consul) (marries Domitia Calvilla) (marries Antoninus Pius
| |
+-----------------+ ------+-------
| |
Annia Cornificia Marcus Aurelius Annia Faustina
(marries Annia Faustina) |
| |
+--------------+---------------+
|
Commodus

Faustina had inherited her mother’s beauty, and was reared in a

very conscientious home. It was the home in which Marcus Aurelius
learned his first lessons in virtue, as his father died early, and all the
chroniclers speak of it with great respect. We know very little about
her, however, until she becomes Empress, and, as she died three years
afterwards, we have not much concern with her. She is believed to
have married somewhat late for a Roman girl, in or about her sixteenth
year (120). Titus Aurelius Antoninus was then in his thirty-fourth year,
a tall, graceful, and handsome man, of quiet and captivating manners,
good cultivation, fine character, and a face of great dignity and
sweetness. He was of good family, and was advancing rapidly in the
public service. Shortly after the marriage he became consul, and he
remained in Rome in one or other civic capacity until 128 or 129. He
was very wealthy and greatly esteemed.
One of the chroniclers has charged her with light behaviour, and,
as this is the only period in which we can plausibly entertain it, we may
regard the charge for a moment. The book of Dio’s history for the reign
of Antoninus Pius is lost, so that neither he nor his commentators
throw any light on Faustina. Aurelius Victor and Eutropius say nothing
of her character. The one hostile witness is “Julius Capitolinus,” the
anonymous writer of the fourth century who provides the sketch of the
life of Antoninus Pius in the “Historia Augusta.” He says (c. 3): “Many
things are said of his wife’s excessive freedom and looseness of life,
which he had painfully to overlook.” Serviez enlarges on this with his
usual license. But as he makes Faustina the sister of Ælius Verus, and
says that she neglected the education of her children, which is also
untrue, we may ignore him.
It is now more customary to reject this charge against the elder
Faustina, on the ground that the single witness is a light anecdotist of
the fourth century. Moreover, when the tutor Fronto wrote a glowing
panegyric of Faustina after her death, Antoninus Pius answered that it
was even more true than eloquent, and swore that he “would rather
live with her at Gyaros [a barren island, to which criminals were
deported] than in a palace without her.” Nevertheless, we must leave
the question open. Antoninus Pius was not a puritan. When the
Emperor Julian introduces him before the gods, in his charming contest
of the Emperors for the highest praise (“The Cæsars”), he calls him “a
moderate man, not indeed in love-affairs, but in the administration of
the Empire.” Faustina was probably charming enough to merit his
sincere lament. But as Capitolinus mingles truth and untruth with a
very light hand, and the relevant book of Dio is wanting, we cannot
decide the issue.
In the year 128 or 129 Antoninus was appointed Proconsul of Asia,
and he and Faustina went to Smyrna. The elder of their two daughters
died about the same time. An amusing incident in connexion with their
arrival is narrated by Philostratus in his “Lives of the Sophists.” The
Proconsul at once occupied the finest house in Smyrna, the home of
the teacher Polemo, who was absent. Polemo was the idol of Smyrna,
and was proportionately conceited. He drew youths from all parts to his
school, and had won much favour from Hadrian for the city. He
travelled in a superb Phrygian chariot, and his mules had silver
trappings; and when some grumblers had hinted that he had diverted
to his own pocket some of Hadrian’s subsidies, he had pompously
written to the Emperor: “Polemo has given me an account of money
given by you to him.” This conceited sophist reached his house in the
middle of the night, and found the Proconsul and Faustina abed there.
He promptly turned them out, and roundly abused them. Years
afterwards, when the genial Antoninus was Emperor, and Polemo came
to the palace, he said laughingly to an attendant: “See that Polemo has
a chamber in the palace, and that no one turns him out.” Later an
actor came from Smyrna to complain that Polemo, the autocrat, had
turned him out of the theatre. “At what hour?” asked the Emperor
gravely. It was at midday. “That is nothing; he turned me out at
midnight,” said the Emperor.
The amiability and solid work of Antoninus must have won Polemo,
as Hadrian is reported to have said in his will that it was he who
advised the adoption of Antoninus. But the East generally so much
appreciated the Proconsul that, when he returned to Rome, he stood
very high in the favour of Hadrian. We again lose sight of Faustina until
he becomes Emperor, and then there are one or two brief references to
her before she dies in 141. At his accession he refused the greater part
of the money (aurum coronarium) which was due to him, by custom,
from the provinces, and drew very liberally on his private fortune for
paying the great expenses entailed. Faustina naturally demurred.
“Foolish woman,” he is said to have answered, “when we obtained the
Empire we lost what we previously possessed.” The only other
reference is contained in a letter of the younger Faustina to Marcus
Aurelius: “In the defection of Celsus my mother exhorted Antoninus to
be concerned first about his own family.” We know nothing of this
revolt. Apparently Antoninus, like Marcus Aurelius, was disposed to be
dangerously lenient. The final reference to Faustina is that she died in
the third year of his reign (141), and was deeply mourned by him.
Nominated “Augusta” in life, she was deified at death, and Antoninus
built in her honour the beautiful temple of which traces are still seen in
Rome. He also instituted in her honour a fresh charity for orphans, the
“Puellæ Faustinianæ,” and ordered that gold and silver statues of her
should be borne in the processions.
This sincere tribute of the Emperor tells at least of a great affection
and esteem, but the literary references to Faustina are too meagre and
disputable to bring her clearly before us. The busts that are believed to
represent her do not, unfortunately, assist us much. In the Capitoline
Museum at Rome is one that may depict her in her twenties or earlier.
It has a round and tranquil face, not devoid of strength, but more
directly suggesting an even and sober character. Another bust, in the
Vatican Museum, shows the same features at a later age; but a third,
in the same Museum, has not so pleasant an expression. The oval face
is hard and querulous. The loose lips droop at the ends; the large eyes,
prominent cheekbones, and strong chin have an expression that is very
far from tender or spiritual. The bust that is attributed to her in the
British Museum is between the two. The elder Faustina remains in
obscurity, and we pass to her more notorious daughter and successor.
For twenty years after the death of Faustina there was no Empress
of Rome. Antoninus, who was in his fifty-fifth year, refused to marry
again, and took a concubine—an arrangement recognized in Roman
law and practice, in which marriage had several degrees. It was an era
of general peace and great prosperity. The group of Stoic lawyers that
the Emperor gathered about him humanely moderated the rigour of
the laws, medical service was supplied to the poor in the towns, the
school-system was further endowed, and works of mercy continued to
multiply. The armies usually rested—and, it is to be feared, rusted—the
treasury was again filled, the Empire was happy and prosperous. In the
year 161 the cheerful, benevolent Antoninus passed away, and the two
men whom Hadrian had compelled him to adopt came to their joint
reign. With them are introduced two new Empresses of no little
interest.
The two boys whom Hadrian had lightly designated as the heirs to
the throne after Antoninus were Annius Verus, or Verissimus, as
Hadrian genially called him on account of his precocious gravity and
piety, and Lucius Verus, son of Hadrian’s dissolute companion. Annius
was a great favourite of the Emperor. He received office in his sixth
year, and donned the philosopher’s cloak in his twelfth. He was the pet
of his grandfather’s palace, but so serious in his Stoicism that his
mother had difficulty in persuading him to sleep in a bed instead of on
the floor. In his sixteenth year Hadrian gave him the manly toga, and
betrothed him to the daughter of Lucius Verus. In his eighteenth year
he was “terrified” to hear that he had been chosen for the succession,
and must go to live in the palace. Then Hadrian died, and Antoninus
adopted him.
 Gibbon has greatly praised Antoninus for preferring the welfare of
the State to the interest of his family in this adoption. It is true that, as
we know from coins, Antoninus and Faustina had had two sons, as well
as two daughters, but they must have died before the year 138. Dio
expressly says that Hadrian ordered Antoninus to adopt the two youths
“because he had no male children at the time.” His boys, like his elder
daughter, must have died before that time; and indeed we have no
further mention of them. But if this particular grace cannot be allowed
to Antoninus, we must admire his careful control of their education and
his discriminating guidance of their fortunes. The best masters in Rome
instructed each of them, and it was only the deep-rooted difference in
their constitutions—the moral strength of the one and weakness of the
other—that led them to diverge so widely. The vigilant eye of the
Emperor observed the dissimilarity of promise. He left Lucius Verus out
of the way of promotion, and destined Marcus for the great
advancement.
No sooner was Antoninus on the throne than he approached
Marcus, through Faustina, with a proposal of marriage with his
daughter. She had been promised by Hadrian to young Lucius Verus,
and Marcus was to marry Ceionia. The Emperor proposed to cancel
these contracts, and marry the younger Faustina to the young Stoic. It
would be extremely interesting if we could penetrate the feelings of the
young princess at the time. The later busts of her suggest a pretty,
round-faced girl, probably in her early teens, with small eyes and a
lively temperament. The grim and austere young scholar would not
attract her, and one can imagine her feelings when he asked time to
consider whether he would accept the hand of the Emperor’s charming
daughter. Marcus philosophically weighed the proposal in his mind until
the time he asked had expired, and then he consented to betrothal. He
was appointed Cæsar and consul designate, and given the palace of
Tiberius for a dwelling. A bust that we have of him, in the Capitol
Museum, represents him about this time—a face of singular beauty and
refinement framed in a mass of short curly hair.
Their marriage—a superb ceremony—did not take place until about
seven years later (145), a circumstance which we may regard as a
further philosophic error. During the years of waiting, and during most
of the reign of Antoninus, Marcus was absorbed in study. He was
penetrated with the aphorism of Plato, that the State would be happy
whose prince was a philosopher. What the effect was on Faustina we
may be in a better position to say later. Her mother had died in 141,
her womanhood was fully born, and the eye of her father had an
Empire to survey. At the death of Antoninus the throne was at once
offered to Marcus. In his last moments Antoninus had ordered the
golden statue of Fortune, which he kept in his chamber, to be conveyed
to Marcus. From a sense of duty he, unluckily for Rome, associated
Lucius Verus with him in the Empire. Somewhat delicate himself, he
relied on Verus for such work abroad as was immediately necessary,
and continued to frequent the schools.
His peaceful studies were quickly interrupted. Fatal floods and
scarcity of food disturbed the capital; the eastern frontier was again
aflame, and the German frontier was threatened. Marcus sent Verus to
take command in the East, after betrothing him to his daughter Lucilla,
held off the northern barbarians with bribes and diplomacy, and
worked hard for the relief of Rome. For a time his policy seemed to
triumph. The Germans were pacified, and the eastern peoples
repressed. Verus, indeed, advanced no farther than the voluptuous
palaces of Antioch and the licentious groves of Daphne. Once only
during the campaign did he quit the luxury of Antioch. He heard that
Marcus was coming East with his daughter Lucilla, and hastened to
meet him otherwhere than in garrulous Antioch. Marcus did not leave
Italy, however, and Verus wedded Lucilla, and returned to his perfumed
vices. Happily, there was in the East a Roman general of the old stamp,
Avidius Cassius, a strong and blunt man, disdainful of luxury. He lashed
the debauched troops into a state of discipline, pacified the East, and
let Verus return to Rome to enjoy his triumph.
Here begin the stories that have gathered about the memory of the
younger Faustina, and have persuaded many a writer that, as one of
the authorities says, she became a second Messalina. If we are to
believe the “Augustan History,” she behaved with the most abominable
license throughout her whole married life. Four Roman nobles are
specifically named as notorious lovers of the Empress, and she is
charged with general license. One of the four was named Tertullus, and
it is said that one day, when Marcus was in the theatre, an actor made
flagrant reference to this liaison. Asked for the name of a certain lover,
he said three times (ter), “Tullus, Tullus, Tullus.” It is added that
Marcus—who might very well miss a point in the theatre, as he read
and wrote letters there—was quite aware of the liaison, because he
one day surprised Faustina at breakfast with Tertullus. The Empress is
further charged with adultery with the voluptuous colleague of her
husband, and with wantoning among actors, gladiators, sailors, and
others of the baser sort.
The more sober writers on Faustina have generally been unwilling
to admit this debauchery. Duruy rejects the stories altogether, Merivale
recommends reserve, and Renan thinks that “careful research has
reduced to very small proportions the accusations which scandal was
pleased to bring against the wife of Marcus Aurelius.” It seems to me
that we can only come to the same conclusion as we did in regard to
Messalina; we must regard particular legends with reserve, but must
conclude that the general opinion of Faustina at the time, which the
stories embody, must have had a serious basis. Some of the stories put
on record by Capitolinus in the “Augustan History” are palpably false.
One runs that she confessed to Marcus her passion for a certain
gladiator, and that Marcus was directed by the Chaldæan sages, whom
he consulted, to kill the man and bathe the Empress in his blood. Her
passion was cured, but her next child was the brutal Commodus. This
story is so gross—I do not reproduce all the details—that the writer
does not insist on it, but he continues: “Still, as her conduct with the
gladiators is well known, Commodus probably was the son of a
gladiator.” Now the tutor of the princes, Fronto, remarks in one of his
letters, and the surviving busts bear him out, that Commodus had a
striking likeness to Marcus Aurelius. I may add that Commodus was
born in the year of the Emperor’s accession, when such conduct is
incredible.
 FAUSTINA THE YOUNGER
BUST (REPUTED) IN THE BRITISH MUSEUM

Other parts of the legend are just as vulnerable. Thus it is said that
Faustina poisoned Verus when he boasted to his wife of his relations
with her. He died a very natural death, as we shall see later. On the
other hand, Dio, who lived shortly afterwards, and had no dislike for
scandal, knows nothing whatever about this looseness on the part of
the Empress, and there is nothing in Eutropius or Aurelius Victor. The
only other writer who, in a general way, accuses Faustina of
dissoluteness is the Emperor Julian (“Cæsars,” c. 28). We are therefore
in a dilemma, and must not too readily speak of Faustina as a second
Messalina. The quiet assumption of her guilt in Julian, and the fact that
the stories in the “Augustan History” are professedly taken from Marius
Maximus, an historical writer not far removed from her time, imply a
very general belief in her guilt. In one place Capitolinus says (c. 23)
that the Emperor “cleared her by his letters” of the charge of loose
behaviour with actors, and in another represents him as saying, when
he is urged to divorce her on account of her vices: “If we send away
the wife, we must give up her dowry,” though the Empire could hardly
be called Faustina’s dowry. In a third place, however, Capitolinus leaves
it open whether Marcus “was ignorant of, or ignored,” his wife’s
misconduct. For many writers, in fact, the attitude of Marcus is
decisive. If such things had been done he must have known, and, with
such knowledge, he could not have spoken so highly of his wife in his
“Meditations,” and would not have dared to set up, in her memory, an
altar on which the maidens of Rome should offer sacrifice before
marriage.
The scale, in truth, is somewhat evenly balanced, yet one cannot
easily conceive that the heavy charges of Marius Maximus and the
deliberate verdict of Julian had no foundation. Whether from weakness,
or from an excess of casuistry, Marcus Aurelius lacked decision or
penetration in such matters. He married his daughter to a profligate,
whom he afterwards deified, and he committed the Empire to a son
who had given early promise of vice. His grave and ascetic ways
probably repelled the gay and beautiful woman whom he had
diplomatically married, and she seems to have sought relief. None of
the busts, medallions, or coins, which more or less convey an image of
her to us, suggest character or culture, but rather a weak control and a
sensuous temper. From her Commodus derived the enfeebled will that
put him at the mercy of his more dissolute courtiers, and the sensuality
that made his short reign an indescribable debauch. Much as we
should like to relieve Marcus Aurelius of the shame of having begotten
such a monster, we must admit his parentage, and cast what blame
there is on the mother.
 In this unsatisfactory haze we must leave the conduct of the
Empress during the years in which her husband wrought for the safety
of the Empire, bequeathed his austere reflections to later ages, or
contemplated the golden images of his teachers in his lararium. The
triumphant return of Verus was quickly followed by years of gravest
anxiety. In the pestilential East the legions had absorbed the germs of
plague, had strewn them along their route, and had now disseminated
them throughout Rome. Thousands of victims, rich and poor,
succumbed to the subtle malady. Marcus vainly summoned the
ministers of every religion and the medical men of all schools, and
sacrificed those obscure Christians on whom popular anger was ever
ready to visit a calamity. His trouble increased when it was announced
that the fierce Marcomanni of the north had burst into the Empire, and
were driving the Romans before them. With great energy he mustered
the demoralized legions in the north, and set out with Verus against
the enemy. In the middle of the war (168) Verus, who had repeatedly
tried to return to the comfort of the capital, died. He had an apoplectic
fit on the journey, and we may ignore the various suggestions that
either Lucilla, or Faustina, or Marcus put an end to his useless career.
Marcus continued for several years the task of settling the frontier
tribes. It seems that Faustina went with him on these arduous
campaigns, though whether we may see in the circumstance any merit
on her part, or a device of the Emperor to control her conduct, it is
impossible to say. She at least earned a title—“Mother of the Camps”
and “Mother of the Legions”—which is found on few coins of the
Empresses. It is probable that her disorders belonged to an earlier
date, before and in the early part of the Emperor’s reign. It is chiefly at
Gaeta, the pretty bay on the coast where many Romans had villas, that
Capitolinus places her familiarity with gladiators and sailors. Possibly
the sobriety of her later years was accepted by her husband as an
expiation, and held to justify his eulogy of her.
Those later years were full of trouble and anxiety. Not only did two
of their children die, and their daughter Lucilla become the widow of a
notorious profligate, but the gods seemed to have entered upon a
contest with the virtue of Marcus Aurelius. A great earthquake shook
the East, the plague left a blackened trail over the Empire and infested
the camps, and other disasters were crowded into a few years. The
treasury ran short, and Marcus was obliged to put up the Imperial
treasures at auction to obtain funds for carrying on the war. His one
consolation was that the Eastern frontier was tranquil, yet in the year
175 a messenger came to announce that his great general, Avidius
Cassius, was in revolt, and claimed the Empire.
Verus, who must have felt the scorn of the stronger man, had
warned Marcus years before that Cassius was dangerous, but the
actual revolt is persistently connected in the chronicles with Faustina.
Cassius had ambition, and had only been prevented by his father in
earlier years from rising against Antoninus Pius. In 174 or 175, it is said
by Dio, he received a message from Faustina, proposing that, in the
event of Marcus dying, he should marry her, and occupy the throne.
Shortly after this a false message reached him that Marcus was dead,
and he at once announced to the legions that he assumed the Empire.
The message was quickly contradicted, but Cassius thought it too late
to retire, and he prepared for a struggle. Marcus sadly moved towards
the East. Before he had gone far, however, he learned that the soldiers,
who hated Cassius for his rigour, had put him to death.
The position of Faustina is once more in grave ambiguity. The
writer on Cassius in the “Historia Augusta” gives the rumour implicating
her, but rejects it. Unfortunately, his rejection is in this case no more
weighty than his acceptance in others. He admits that his source,
Marius Maximus, believes Faustina guilty, and ascribes it to “a wish to
defame” the Empress. Except that the hatred of Commodus at Rome
may have for some time been extended to the woman who had borne
him, there is no clear reason why Maximus should calumniate Faustina.
Dio, who lives very close to the time, gives it as a positive fact that
Faustina secretly urged Cassius to marry her, and occupy the throne, if
Marcus died. We may concur in the verdict of most of the writers on
the matter. Marcus was ailing, delicate, and overburdened with work. It
seemed to Faustina that he would not live long, and, as Commodus
was a callow and unpromising youth, and by no means sure of
succession, she sought an arrangement by which she should remain on
the throne if her husband died.
 It is not generally felt that there was anything gravely
reprehensible in this, but a secret negotiation of such a character does
not present her to us in an attractive light. Her subsequent zeal for the
punishment of Cassius and his friends is equally unpleasant, even if we
recall that she had no intention of raising him against the Emperor
while he lived. Several letters which passed between Marcus and
Faustina have been preserved in the “Historia Augusta,” from Marius
Maximus, and there seems to be little ground to doubt their
genuineness. They suggest that Marcus was in the habit of consulting
with Faustina on matters of grave importance. “Come up to the Alban
Mount,” he writes her, after telling of the sedition, “and by the favour
of the gods, we will discuss the affair in safety.” Faustina replies:

“I will set out to-morrow for the Alban Mount, as you

command, but I at once implore you, if you love your children,
to visit these rebels with the utmost severity. The soldiers and
their leaders have fallen into evil ways, and they will crush us if
we do not coerce them.”

In another letter she presses him again:

“My mother Faustina urged your father [by adoption] Pius,

at the time of the secession of Celsus, to feel first for his own
family.... You see how young Commodus is, and our son-in-law
Pompeianus is older and is abroad. Do not spare men who have
not spared you, and would not spare me and the children if
they won.”

A later letter of Marcus tells that he has read her exhortation in his
villa at Formiæ (on the Gulf of Gaeta). By that time he has heard that
Cassius is dead, and he will hear of no further revenge on his family.
He will spare his wife and children, and beg the Senate to be moderate
in punishing the accomplices, because “there is nothing that so much
commends the Emperor of Rome to the nations as clemency.” We
know, in fact, that he treated the family of Cassius with great
generosity.
 The Emperor and Empress then went to the East to complete the
work of pacification. In the course of the voyage, in a little village at
the foot of Mount Taurus, Faustina met her end in the year 175. As a
matter of course she was placed among the gods, but Marcus was not
content with the customary honouring of her memory. He gave the
village the name of Faustinopolis, founded a fresh charity with the title
of “Puellæ Faustinianæ,” and built a beautiful temple at Rome, which,
when he died a few years later, was dedicated in their joint names by
the Senate. As if to obliterate all the rumours about her infidelity, he
went on to ask extraordinary honours for her of the Senate. He set up
a special altar, with a silver statue of her, in the temple of Venus, and
directed that maidens about to marry should offer sacrifice on it; and
he had a golden statue of her placed on her seat in the theatre
whenever he attended its performances.
Dio gives two versions of the death of Faustina which were current
in his time. Some said that she died of gout, from which she suffered;
others held that she put an end to her life in fear lest her complicity
with Cassius should be discovered by Marcus in the East. The second
theory is superfluous. The natural cause of death seems adequate
enough, nor would she be in any serious danger if Marcus heard that
Cassius had made her the pretext of his rebellion. Her chief misdeeds
were to live after her. Frivolous, and probably licentious, in her early
married life, she seems to have settled in sober ways when she
became Empress, but we find no influence of hers in the ordering of
affairs. Had she only reared healthy children to succeed her husband,
she might have contributed worthily to the mighty task of supporting
the shaken Empire. Instead, she gave to the Empire Lucilla and
Commodus, her two surviving children, and it fell into a fresh
degradation.
 CHAPTER XI

THE WIVES OF THE SYBARITES

A
S Marcus Aurelius and Lucius Verus had been equal in Imperial
power, and both were married, we have one more Empress
to regard before we pass on to the wives of Commodus;
and the account we have already given of Verus will justify us in
relegating her to this distinct chapter. Verus had married Lucilla, the
eldest daughter of Marcus and Faustina; but the ambiguous repute
of her mother will warn us not to expect a painful spectacle of vice
in alliance with lofty virtue. Lucilla carries a step further the unhappy
disposition which we have suspected in her grandmother, and more
palpably detected in her mother. By her union with Lucius Verus vice
was once more decked with the Imperial purple and justified in the
eyes of Rome. We may briefly consider Lucilla as Empress before we
follow her lamentable career under the reign of her brother.

Lucilla was born in the first year of the married life of Marcus
and Faustina. Marcus was then a pale and thin-blooded scholar,
Faustina in the full warmth and sensuousness of young womanhood,
and it was not unnatural that the child should inherit the temper of
her mother without the spiritual restraint of her sire. She was
educated with the greatest care, and was betrothed to Verus in her
sixteenth year. Presumably by the will of her father, and certainly
with the full assent of Verus, she remained two further years in the
palace, while Verus wore out his strength in the dissipations of
Antioch. Marcus heard of his conduct, and sent out Lucilla to marry
him; as if a union with a young woman of seventeen or eighteen
would be apt to have a sobering influence on a man of Verus’s habits
and parentage. Verus met her at Ephesus, married her there with
great pomp, and returned with her to his pleasures at Antioch.
They came to Rome at the peace of 166, and Marcus could not
fail to learn in full the character of the man to whom he had
entrusted his daughter and half his power. The villa which Verus
occupied in the Clodian Way was the most notorious house of
debauch in Rome. It swarmed with the dancing-girls, boys, Eastern
slaves, musicians, conjurors, etc., that Verus had brought from the
East. One room was fitted up as a popular tavern, and we must
leave under the veil of a dead language the abominations that were
perpetrated there. One can only repeat such comparatively decent
details as that Verus would have gladiators to fight in his house
during dinner, and prolong the carouse until his slaves had to bear
away his stupefied form on his couch; or that, on other occasions,
he would emulate the early feats of Nero, and revel at nights in the
wine-shops and brothels of the popular quarter. One night he gave a
superbly furnished banquet, and at the close, in a drunken fit,
presented to his guests the costly plate, and even the litters, with
silver-harnessed mules, in which they were taken home.
Marcus made several futile attempts to brace him by a campaign
in the north, and must have been sincerely relieved when he at last
paid, by a premature death, the price of his excesses. Lucilla had
then been Empress for eleven years. As she is barely noticed in the
chronicles, we are left to imagine the effect on her of living through
her early womanhood in such a palace as that of Verus. Probably
disgust saved her very largely from the taint. Verus’s sister Fabia
lived with them, and was generally believed to be intimate with her
brother. She at least usurped the place of Lucilla in authority, and
the Empress must have been as much relieved as her father when
Verus died. He was rumoured to have been poisoned by Lucilla
because of his relations with Fabia; by Faustina, for betraying his
relations with her; and by Marcus, to rid the Empire of his
sottishness. But an apoplectic fit would be so natural a crown to
such a career that we can dispense with so much poison.
 Lucilla was then married by Marcus to an elderly and worthy
Senator, Claudius Pompeianus. She and her mother strongly
resented the marriage, and demanded a younger and more
attractive husband; but the Emperor was unusually firm. Unhappily,
his firmness was misplaced, for the austerity or age of Pompeianus
effected what the profligacy of Verus had failed to do, and Lucilla fell
into vicious ways. We may conjecture that this did not happen until
after her father’s death. Marcus had returned to the war against the
Marcomanni, and, after three years of great exertion and sacrifice,
was within sight of victory when death carried him off. He had not
married again, in spite of Fabia’s efforts to win him. In the fashion
approved even by philosophers, he took a concubine to his bed, and
virtuously refused to put a stepmother over his children. At his death
a new Empress comes upon the scene, and, as Lucilla still retained
her Imperial dignities and privileges, we shall have to consider them
in an unamiable conjunction.
The last and most fatal blunder of Marcus Aurelius was to leave
the Empire in the very uncertain hands of his son Commodus. War
had drained the treasury; plague, famine, and sloth had thinned and
weakened the population; vice had again been enthroned for all to
admire and imitate; the lusty barbarians were thundering at its
gates. A new Vespasian or Trajan was needed to restore its vigour, if
such a restoration were possible. Yet Marcus persuaded himself that
the pretty youth, with bright eyes and curly golden hair, who played
at soldiering in his suite in Germany, could bear this enormous
burden. Herodian, whose history of the Emperors now opens for us,
tells us that Marcus was really concerned on the matter as he lay in
his last illness. There were disquieting stories about the character of
Commodus. It was said that in his twelfth year he had, at
Centumcellæ (Civita Vecchia), ordered the bath-attendant to be
thrown into the furnace because the water was not hot enough. On
another occasion Marcus had driven away certain corrupting
attendants, but had recalled them at the petulant tears of his son.
They were with him in Pannonia. We may at least assume that even
the fond eye of a father must have discerned the weakness of
character which, in the course of a year or two, would let Commodus
sink to indescribable depths. Marcus, however, trustful to the end in
the sublime truths of his philosophy, was content to summon
Commodus to his tent, make a pretty speech to him in the presence
of his counsellors, and hand over to him the reins of government.
For a time Commodus remained in the camp, and let the elders
govern. Before long the lighter courtiers hint that it is more
comfortable in Rome, and he talks of going. The elders frown, and
Pompeianus lectures him. He bows submissively, but it is not long
before he decides to go. Numbers of officers discover a similar call to
the capital, and a gay cavalcade sets out. Rome is enchanted, and
goes out miles along the road to meet Commodus, and strews
flowers and laurel in his path, and enthuses over his handsome face
and the curly hair that shines like gold in the sun. It was the coming
of Caligula and Nero over again. The Roman people—quantum
mutatus ab illo!—had come to appreciate a pretty face, and a
prospect of endless games, immeasurably more than the security of
the frontier.
When Commodus had set out with his father for Germany, he
had been married—“hastily married,” the chronicle says—to a lady as
young and thoughtless as himself. Crispina was a very beautiful girl,
and of distinguished family. Her father, Bruttius Præsens, was a
Senator of great merit. It seems that she accompanied Commodus
to the camp, and returned with him to Rome. In his train were the
evil counsellors whom Marcus had banished and recalled. Their hour
had come.
For three years Commodus enjoyed the pleasures which they
provided or invented for him, and left the administration in the
capable hands of his father’s servants. Possibly this was the highest
virtue Marcus had expected of him. But the ambition of his
confidants steadily grew, and a bitter feud in the palace now came
to a head and gave them an opportunity. Crispina and Lucilla were
violently opposed to each other. The Imperial title of Lucilla paled
beside that of the wife of the ruling Emperor. The fire which had
been borne before her when she went abroad now passed to
Crispina, and she had to yield precedence in the palace and the
theatre. Crispina, on the other hand, resented the familiarity of
Commodus with his sister, and would hardly be ignorant of the
interpretation that was generally put on it. The adherents of the
palace were thus divided into two parties, and the Empresses fought
for the monopoly of Commodus’s favour. At last Lucilla despaired of
gaining her end through Commodus, and resolved to have him
murdered.
There is no room for doubt that the daughter of Faustina and
Marcus Aurelius was an abandoned woman. Dio declares that she
was “no better than Commodus.” We may trust that this is an
exaggeration, but the other authorities speak of the looseness of her
conduct, and are emphatically agreed that she inspired the plot to
murder her brother. No one doubts that her purpose was to recover
supreme power. The inferences and impressions we draw from
Imperial portraits are not very substantial, but it is interesting that
the statue of Lucilla, which we have, suggests just the type of
woman that the historians represent her to have been. It is the
figure of a full-bodied woman, of strong and imperious temper,
sensual to the limit of grossness. In her the beauty of her mother,
instead of being enhanced by the purity of her father, is blighted by
a general expression of coarseness and self-assertion.
Her criminal design was gradually imparted to her lovers. Among
these was a young noble named Quadratus, whom she soon fired
with a sense of her grievances, and a conspiracy was framed. The
actual assassination was undertaken by her stepson, Claudius
Pompeianus. Herodian says that his name was Quintianus, and he
may have had this name in addition. Dio gives a confused and
contradictory account—he describes Pompeianus as married to
Lucilla’s daughter, whereas Lucilla was married to his father, and he
says that she was intimate with him, yet hated him and wished to
destroy him—but, as he lived in Rome at the time, we must accept
the substance of his story. The young Senator Pompeianus was an
intimate friend of Commodus, and only an infatuation for Lucilla
could have drawn him into the plot. He spoiled it, and ruined the
conspirators, by his melodramatic display. As Commodus entered the
amphitheatre, he rushed upon him with a drawn sword. But he
announced his purpose by crying out: “The Senate sends thee this
sword,” and the guards arrested him.
The plot gave Commodus an opportunity to make a bloody
clearance of those who hampered his plans, and caused him to
regard the Senate with dark suspicion. The male conspirators were
executed, and Lucilla was banished to Capreæ. But Crispina had no
triumph by the removal of her rival. She had herself been tainted in
that atmosphere of vice, and was detected in one of her liaisons by
Commodus. She was banished to Capreæ, and there both she and
Lucilla were put to death.
 LUCILLA
BUST IN THE NATIONAL MUSEUM, ROME

The conspiracy took place in the year 182, the third year of
Commodus’s reign. The remaining ten years of his life it would be
more agreeable to leave in the untranslatable language of the
chroniclers, but he virtually shared his throne with a woman of a
singular and interesting type, and we must include her in the gallery
of wives of the Emperors. Among the property of the wealthy young
conspirator, Quadratus, which was at once confiscated, was a very
handsome and engaging concubine of the name of Marcia. The
concubinatus was, as I have said, a legal and recognized union in
Rome, and we must not regard these women, who enter our
chronicle in that capacity, in quite the same light as the mistresses of
later Christian princes. They were sometimes of moderately good
family, though they seem generally to have belonged to the class of
emancipated slaves, and were included in the man’s property. Marcia
was of the latter class. Probably an orphan at an early age, she was
brought up by a eunuch, and sold by him to Quadratus. At the
dispersal of his property, or even during his life, she attracted the
notice of Commodus, and was transferred to the populous harem of
his three hundred concubines.
A few years later (185) an event occurred that greatly increased
her growing power over the Emperor. The chief favourite of
Commodus was a low-born and despicable courtier named Perennis,
who encouraged the Emperor to pursue his morbid sensual
impulses, while he himself accumulated wealth and power. He
flattered and indulged every fancy of his besotted master, and
controlled all the resources of the State in his own interest. He was
commander of the guards, and seems to have at length conceived
an ambition to displace Commodus. One day, when Commodus
presided at the games, which he very liberally provided, before an
immense crowd, a mild-looking man—said to be a philosopher—
rushed into the centre of the stage and roared out a warning to the
Emperor that Perennis was acquiring wealth and aiming at the
throne. The prefect had him burned alive, and escaped the
Emperor’s suspicion; but the end was nearer than he expected. A
regiment of fifteen hundred men from the legions of Britain marched
into Rome, demanded the head of Perennis, and forced Commodus
to recognize and punish the faults of his minister.
From that time Marcia occupies the place of prima inter pares in
the harem of Commodus. A good deal of research has been
expended on this leading concubine of the Emperor, because there
was a tradition in early Christian literature that she favoured and
 14
protected, if she did not herself belong to, the new religion. It was
said that she sent the eunuch, who had reared her, to liberate the
repressed Christians of Sardinia, and the peace which they enjoyed
at Rome during the reign of Commodus is attributed to her
influence. But if Marcia had ever belonged to the austere sect of the
early Christians, we must, for its credit, entirely dissociate her from it
in her Imperial days. She seems to have been to the brutal
Commodus what Cæsonia had been to the equally licentious
Caligula. She dressed willingly as an Amazon, and is actually
represented on the coins, with Commodus, in the helmet of a female
warrior. If we may put any trust in that meagre portrait of her, she
seems to have been of much the same type as Cæsonia: a
handsome, strong, vulgar woman, owing her influence to her
masculine robustness.
For seven years she occupied, without a quarrel, the chief place
in a palace in which all the orgies of Caligula, Nero, and Verus were
concentrated. At her persuasion Commodus changed the name of
Rome to “the Colony of Commodus.” One might almost suspect her
of genial irony in thus removing the venerable name from the
Imperial city during the years when it was degraded by Commodus.
Evil as the practices of Caligula and Nero had been, they were
surpassed by the insanities and obscenities of the son of Marcus
Aurelius. We must leave the veil over the life that was witnessed in
the palace during those ten years; but the crimes of Commodus
were not confined to the wild indulgence of his unbridled appetites.
The company of gladiators and the daily pleasure of killing degraded
him to the character of a mere butcher. He forced the priests of
orgiastic Eastern cults to perform on themselves the mutilations
which their ritual described; he beat them with the emblem of
Anubis which he carried in their processions. On one occasion he
had all the citizens of Rome with some infirmity of the feet gathered
in one place, and more or less dressed as dragons. Then the Roman
Hercules—as Commodus loved to be called—fell upon them with a
club, and killed numbers of them. This and other stories of his
indescribable lust and cruelty are told by an historian who saw
Commodus daily.
In the year 189 Marcia obtained even greater power over her
insane lover. The place of Perennis had been at once occupied by
another of the Emperor’s despicable courtiers, Cleander, a Phrygian
slave who had risen, by base means, to be the first minister of the
Empire. Like his predecessor, he encouraged Commodus to wallow in
his vices, while he took advantage of his insanity to enrich himself.
The highest positions in the State were sold by him, and men could
even purchase from him the right to take vengeance on their
enemies, or the privilege not to be executed for their wealth. The
treasury was again diminishing, and noble blood poured out freely to
refresh it. A great pestilence swept over Italy, exacting thousands of
victims daily in Rome alone. A terrible famine succeeded it. The
people, observing that the avaricious minister was endeavouring to
make a corner in corn, now broke into rebellion and pressed to the
palace of the Emperor.
Commodus was enjoying himself at the beautiful palace of the
Quintilians in the suburbs, which he had obtained by murder, when
the crowd surged up to the gates. Cleander turned the cavalry upon
the people, but the infantry sided with them, and they returned in a
storm of anger to the palace. None of his ministers dare approach
the room in which Commodus wantoned with his companions, but
his sister Fadilla and Marcia broke in with the news that his life was
in danger. Some writers say that it was Fadilla who informed him,
some that it was Marcia. We may suppose that both of them
endeavoured to awake him. The voluptuous coward at once
sacrificed Cleander to the crowd, and returned to his vices.
Marcia had now the leading influence over Commodus, and
Rome sank lower and lower. The butcheries of the amphitheatre
were his chief concern. He consorted daily with the gladiators, killed
vast numbers of beasts in the arena, and even fought with men who
had meekly to submit to be slain by him. Numbers of distinguished
or wealthy Romans were put to death on the most frivolous pretexts,
yet the Senators were compelled to view and applaud his daily
slaughters with such cries as: “Thou conquerest the world, O brave
Amazonian.” Dio, who sat among the Senators, tells us that one day
Commodus made a grotesque attempt to intimidate them. He had
just killed an ostrich, and came toward them with the head in one
hand and the bloody sword in the other. He grinned and wagged his
head, without saying a word, as he approached them, as if
intimating that it would be their turn next. Dio says that his
appearance was so ludicrous that he had hastily to pluck a leaf of
laurel, and chew it, to prevent him from laughing. We nearly missed
the writing of one of the most valuable histories of the period.
The “Golden Age,” as the Senate was compelled to describe this
appalling decade, came to a close through a fresh excess on the part
of Commodus Pius, as he was now styled. They had reached the last
day of the year 192, and were preparing for the great festivities of
the morrow. Commodus informed Marcia that he would spend the
night in the house of the gladiators, and issue from it on the morrow
at their head. He ordered his chamberlain Eclectus and his
commander of the guard Lætus to make the necessary preparation.
Marcia and the officers were horrified at his proposal, and besought
him to abandon it. After reading the disgusting details of his career
in the “Historia Augusta”—even if we make allowance for
exaggeration—one has some difficulty in realizing their indignation.
Apparently, however, this proposal to identify himself so intimately
with the degraded caste of public gladiators was regarded by them
as something of an entirely different nature from the filth and
obscenity of his practices in the palace, and they boldly opposed
him. He angrily shook them off, and put their names on his
condemned list. The “Augustan History,” recalling a story we have
heard before, introduces an element of romance into the adventure.
It makes Commodus tie the tablet to his bed, and go to sleep, when
the tablet is playfully removed by one of his jewel-decked boys, and
delivered accidentally into the hands of Marcia.
It is better to follow the version of Dio, who was in Rome at the
time. The two officers and Marcia, realizing that they had incurred
his anger, discussed the matter, and decided to assassinate him.
Marcia was directed to poison him. She put the poison in the meat
he ate, but its effect was spoiled by the quantity of wine he had
drunk, and it caused him to vomit. He became suspicious and
threatening, and went to the bath. They then hastily took into their
confidence his powerful and athletic bath-attendant, Narcissus, and
he entered and strangled the Emperor.
One reads with something like amazement that the successful
conspirators, instead of gladly announcing that they had rid Rome of
such a brute and tyrant, deliberated anxiously how they should
proceed. So blind was the attachment of the troops to their
paymaster, and of the common citizens to any generous provider of
games, that they concealed the deed. Commodus had himself fought
735 times in the public amphitheatre, and on those performances
alone had spent 200,000,000 drachmas. The temper of the
demoralized people and soldiers was uncertain, and they decided to
put the Empire at once in the hands of a strong soldier.
In the romantic story of the accession of the various Empresses
of Rome there are few cases so dramatic as that which introduces
the next Empress in the series. There was living in Rome at the time
an experienced commander, in his sixtieth year, of the name of
Pertinax. His father had kept a kind of tavern in a village of Liguria.
The son had obtained some education, and rapidly climbed the
ladder of promotion. He had married Flavia Titiana, the
accomplished daughter of a very wealthy and distinguished Senator.
Himself enamoured of Cornificia, the sister of Marcus Aurelius, he
had overlooked the vivacity of his wife, and she had at one time
attracted comment by her open regard for a musician. At the time of
the murder of Commodus, Pertinax was Prefect of Rome. He retired
to bed on that last night of the year 192 with no suspicion of the
great events that were happening in the Domus Vectiliana, to which,
it seems, Commodus had gone.
In the middle of the night he was awakened with the message
that the captain of the Prætorian Guards wished to see him. He
calmly said that he had for some time expected to be executed by
Commodus, and he continued to lie, in quiet dignity, when Lætus
entered to tell him that they offered him the Empire. He begged
Lætus to abandon his unseemly joke, and carry out his orders. He
was at last convinced that Commodus was dead, and, through the
darkness of the stormy winter night, they made their way to the
camp. They announced to the guards that Commodus had died of
apoplexy, and that Pertinax was submitted to be chosen by them as
Emperor. The soldiers listened with no enthusiasm. Under the license
of the reign of Commodus they had been permitted to take the most
extraordinary liberties, and they dreaded the accession of a
commander. The news had, however, spread by this time through
the city. People crowded into the torch-lit streets, and poured out
toward the camp, hailing the name of Pertinax and execrating that of
Commodus. A promise of 3,000 denarii to each man overcame the
last opposition of the Guards, and they coldly consented to the
choice. In the Senate, too, there was hesitation. “We see behind
you,” said the consul Falco, “the ministers of Commodus’s crimes,
Lætus and Marcia.” Pertinax himself, indeed, was still very reluctant;
but the Senate urged the Imperial power upon him, and the new
year dawned at Rome upon a people angrily scattering the statues
and memorials of Commodus, and expressing a wild rejoicing over
the advent of its new ruler.
Titiana never bore the title of Augusta, and we may dismiss very
briefly her few months of residence in the palace. The Senate
offered the title of Augusta to Titiana, and that of Cæsar to their
son, but Pertinax refused both. “Let the boy earn it,” he said of his
son; and Dio says that he kept the title from his wife, either because
of the insecurity of his position, or “because he would not let his
lascivious consort stain the name of Augusta.” Titiana was evidently
not the kind of woman to co-operate with Pertinax in his reforms,
and she probably shared the disdain with which her friends regarded
his ways. Although he at once began to undo the evil wrought by
Commodus—to banish the informers, regulate the taxes, and purify
the administration of justice—he alienated the Romans by passing to
an extreme of sobriety. The palace he purified in very summary
fashion. He had the whole apparatus of Commodus’s luxury sold by
auction, and Rome looked on with delight as the three hundred
pretty boys and three hundred choice concubines, the gold and silver
plate, the precious vases and silks and chariots and wonderful
machines of the Sybarite were exposed to their view. But Pertinax
carried his economy too far. Patricians told with contempt that he
would put half a lettuce on the Imperial board, and would make a
hare last three days; the people missed the unceasing stimulation of
the amphitheatre; the soldiers chafed at the discipline he sought to
enforce. Within three months of his remarkable accession to power
Pertinax was assassinated by the Guards, and Titiana fell back into
the obscurity from which she had momentarily emerged.
Another Empress of a day, and one that came to the throne
under no less romantic circumstances, claims our attention for a
moment before we pass on to a more imposing figure.
It was on the 28th of March, 193, that the soldiers brutally
assassinated Pertinax. On the rumour of trouble Pertinax had sent
his father-in-law, Sulpicianus, to secure tranquillity in the camp. As
he lingered there the soldiers returned with the dripping head of the
Emperor, and he recognized that the throne was vacant. With a
callousness that is almost incredible, but is fully attested, he at once
made an offer of money to the soldiers for the Imperial power. It
occurred to some of the soldiers that a higher bid might be secured,
and they announced from the rampart of their camp, in which they
had enclosed themselves, that the throne was, virtually, on sale. In
particular, they sent word to one of the wealthiest citizens, Didius
Julianus, and invited him to make an offer. Whether or no it be true
that he yielded to the vanity of his wife and daughter—he does not
seem to have needed pressure—Julianus went to the camp, and
made a higher offer than that of Sulpicianus.
It was the early evening, and a crowd had gathered to witness
the appalling spectacle of the sale of the Empire. Julianus pointed
out that his rival was the father-in-law of the man they had killed,
and might be expected to have some design of revenge. The soldiers
admitted Julianus by a ladder, and the two Senators made bids
against each other, the soldiers on the wall announcing their offers.
At length Julianus made an offer equal to more than £200 to each
soldier, and he was greeted as Emperor. Under the close guard of
the soldiers he was conducted, amid an angry people, to the Senate,
and forced upon the Senators. They then concluded their bargain by
conducting him to the palace, and the vain old man had time to
reflect on the extraordinary situation he had suddenly reached. His
wife, Manlia Scantilla, and daughter, Didia Clara, joined him “in fear
and concern” (the “Historia Augusta” says), and he finished the day
with a prolonged entertainment.
His wife and daughter were decorated with the title of Augusta
on the morrow, but they soon found that Julianus had squandered
his comfortable wealth on a dangerous bauble. Not only did the
Roman people jeer at him whenever he appeared, but the news
soon came that the distant legions were aflame with anger, and
were about to march on Rome to wrest the Empire from him.
Presently he heard that the commander of the troops in Pannonia
had begun his march at the head of a formidable army. Julianus first
had him declared a public enemy, and sent men to assassinate him;
then he offered to share the Empire with him. Severus and his
hardened troops passed relentlessly over the Alps, and proceeded
along the plains of Italy. Julianus stung the demoralized soldiers who
had sold him the Empire into some pretence of resistance, threw up
earthworks in the suburbs, endeavoured to train his elephants for
the fight, and, as a last resort, fortified the palace. But his
effeminate troops quailed before the seasoned legions from
Germany, and, when Severus reached Rome, Julianus found himself
deserted. The Senate decreed his death, and he was beheaded in
the palace which he had enjoyed, at the price of his fortune and his
life, for sixty-six days. And the two broken-hearted Augustæ laid
down their dignity, and bore the body of Didius Julianus to the tomb
of his ancestors.