FROM THE EDITOR

ESTABLISHED 1845
Mind from Matter
In 2016 a panel of physicists, a cosmologist and a philosopher gathered at the American
Secrets of the Mind Museum of Natural History to discuss an idea seemingly befitting science fiction: Are we liv-
is published by the staff ing in a computer simulation? How exactly the flesh and blood of our brain is able to formu-
of Scientific American, late an aware, self-examining mind capable of critical thought remains a mystery. Perhaps
with project management by:
the answer eludes us because, the panel mused, we are the avatars of a higher species’ simu-
lation and simply unable to discover the truth. As intriguing a hypothesis as it is, neurosci-
Editor in Chief: Laura Helmuth
ence has learned enough about our consciousness to counter such a fantastical possibility.
Managing Editor: Curtis Brainard Newly mapped networks within the human brain show regions that fire in concert to cre-
Senior Editor, Collections: ate cognition (page 6). Zapping the brain with magnetic pulses while recording neural activ-
Andrea Gawrylewski ity might soon detect conscious thought, which could be especially useful for patients who
are awake but unable to communicate or respond to external stimuli (page 26). These discov-
Creative Director: Michael Mrak
eries chip away at the isolating experience of humanity and the idea that a person can never
Issue Designer: Lawrence R. Gendron truly know whether anyone but oneself is truly conscious (page 32).
Senior Graphics Editor: Jen Christiansen To some extent, we exist
Associate Graphics Editor: A
 manda Montañez in our own bubbles of sub-
jective experience. A grow-
Photography Editor: Monica Bradley
ing body of evidence sug-
Associate Photo Editor: Liz Tormes gests that perception is a
construction of the brain
Copy Director: Maria-Christina Keller
(page 38). Because the brain
Senior Copy Editors:
initiates some actions be­­
Angelique Rondeau, Aaron Shattuck
fore we become aware that
Managing Production Editor: Richard Hunt we have made a decision,
Prepress and Quality Manager: we might even deduce that
Silvia De Santis each of us is some kind of
Executive Assistant Supervisor: Maya Harty
biochemical puppet, but
experiments confirm that
we do indeed have free will
Acting President: Stephen Pincock
(page 44). And our cogni-
Executive Vice President: Michael Florek tion clearly results from
Vice President, Commercial: Andrew Douglas
high­ly evolved neural mech-
anisms, common to all of
Publisher and Vice President:
us, for making new memories (page 48), navigating social relationships (page 54) and recog-
Jeremy A. Abbate
nizing faces (page 60). Ultimately a shared sense of reality influences how we perceive our-
Associate Vice President,
Business Development: Diane McGarvey
selves and the formation of “in-groups” and “out-groups,” which can create social and politi-
cal division (page 68).
Marketing Director, Institutional Partnerships For all that, however, a lot happens outside of our awareness. Based on neuronal firing pat-
and Customer Development: Jessica Cole terns, algorithms can infer intended body movements in patients with paralysis of their limbs,
Programmatic Product Manager: Zoya Lysak illuminating the unconscious brain-body dynamic (page 74). During sleep, the brain makes
Director, Integrated Media: Matt Bondlow crucial gains in learning, memory and emotion processing (page 82 and page 88).
And there is much out there that can alter our reality, from drugs to disease. Potentially
Senior Marketing Manager:
Christopher Monello
a third of older COVID patients experience delirium, perhaps increasing their risk for demen-
tia later on (page 116). Psychiatrists now debate whether bouts of mania in which people report
Senior Product Manager: Ian Kelly
enhanced recall, increased empathy and spirituality might constitute a new category of men-
Senior Web Producer: Jessica Ramirez tal disorder (page 110). Ironically, rather than distorting it, substances such as Ecstasy and
Senior Commercial Operations Coordinator: “magic” mushrooms may actually help restore a rational view of the world (page 122).
Christine Kaelin Consciousness may come from an alien programmer, or perhaps it pervades our universe,
not just our brains but all things, as philosopher Philip Goff posits (page 124). Either way, our
Jorm Sangsorn/Getty Images (this page)

Custom Publishing Editor: Lisa Pallatroni

Head, Communications, USA: Rachel Scheer
remarkable ability to think about our own thoughts—meta-thoughts, as S  cientific American
Maria Corte (cover illustration);

columnist John Horgan calls them on page 96—places us at a tantalizing intersection of life
Press Manager: Sarah Hausman
and sentience. What we think matters to us regardless of how the mind arises from matter. 
Production Controller: Madelyn Keyes-Milch
Advertising Production Manager:  Andrea Gawrylewski
Michael Broomes Senior Editor, Collections, [email protected]

SCIENTIFICAMERICAN.COM  |  1
 SPECIAL EDITION

Volume 31, Number 1, Winter 2022

WHAT IS CONSCIOUSNESS?
6 How Matter Becomes Mind
The new discipline of network neuroscience yields
a picture of how mental activity arises from carefully
orchestrated interactions among different brain areas.
By Max Bertolero and Danielle S. Bassett
14 How to Make a Consciousness Meter
Zapping the brain with magnetic pulses while
measuring its electrical activity is proving to be a reliable
way to detect consciousness. B  y Christof Koch
20 Can Lab-Grown Brains
Become Conscious?
A handful of experiments are raising questions about
whether clumps of cells and disembodied brains could be
sentient and how scientists would know if they were.
By Sara Reardon
26 The Brain Electric
Electrodes that stimulate brain tissue reveal the
topography of conscious experience. B y Christof Koch
32 How Do I Know I’m Not the Only 4 what is consciousness ?
Conscious Being in the Universe?
The solipsism problem, also called the problem of other
minds, lurks at the heart of science, philosophy, religion,
the arts and the human condition. B  y John Horgan

HOW WE PERCEIVE THE WORLD

38 Our Inner Universes
Reality is constructed by the brain, and no two brains
are exactly alike. B
 y Anil K. Seth
44 Why We Have Free Will
Neurons fire in your head before you become aware that
you have made a decision. But this discovery does not
mean you are a “biochemical puppet.” B  y Eddy Nahmias
48 The Brain Learns in Unexpected Ways
Neuroscientists have discovered a set of unfamiliar
cellular mechanisms for making fresh memories.
By R. Douglas Fields
54 The Brain’s Social Road Maps
Neural circuits that track our whereabouts in space
and time may also play vital roles in determining how
we relate to other people.
By Matthew Schafer and Daniela Schiller
60 Face Values
Brain regions that process faces reveal deep insights
into the neural mechanisms of vision. B y Doris Y. Tsao
68 Radical Change
Uncertainty in the world threatens our sense of self.
To cope, people embrace populism. B y Michael A. Hogg
36 how we perceive the world

2  |  SCIENTIFIC AMERICAN  |  SPECIAL EDITION  |  WINTER 2022

 UNCONSCIOUS MIND
74 The Intention Machine
A new generation of brain-machine interfaces can deduce
what a person wants. B
 y Richard A. Andersen
82 Infectious Dreams
How the COVID-19 pandemic is changing
our sleeping lives. B
 y Tore Nielsen
88 Sleep Learning Gets Real
Experimental techniques demonstrate how to strengthen
memories when our brains are off-line. 
By Ken A. Paller and Delphine Oudiette
94 Answering Queries in Real Time
while Dreaming
Researchers demonstrate that during REM sleep,
people can hear—and respond to—simple questions
such as “What is eight minus six?” B
 y Diana Kwon
96 Can Science Illuminate Our Inner
Dark Matter?
Neither introspection nor brain scans can reveal
our deepest thoughts. B
 y John Horgan

ALTERED REALITY
102 A Disorder of Mind and Brain
unconscious mind 72
A mysterious condition once known as hysteria is
challenging the divide between psychiatry and neurology.
By Diana Kwon
110 The Undiscovered Illness
Hundreds of thousands of people experience mania
without ever getting depressed. Why does psychiatry
insist on calling them bipolar? B
 y Simon Makin
116 Could COVID Delirium
Bring On Dementia?
Delirium is very common on COVID wards. Researchers
are testing whether these temporary bouts of confusion
could bring on permanent cognitive decline.
By Carrie Arnold
122 A Psychedelic Renaissance
Psilocybin and MDMA represent a first wave of therapies
for psychiatric disorders that help patients by changing
the way they view reality. B
 y Danielle Schlosser and
Thomas R. Insel

DEPARTMENTS
1 FROM THE EDITOR
Mind from Matter
1 24 END NOTE
Does Consciousness Pervade the Universe?
altered reality 100 Philosopher Philip Goff answers questions about
“panpsychism.” B
 y Gareth Cook
Articles in this special issue are updated or adapted from previous issues of S cientific American a nd Nature and from ScientificAmerican.com.
Copyright © 2022 Scientific American, a division of Springer Nature America, Inc. All rights reserved. Scientific American Special (ISSN 1936-1513), Volume 31, Number 1, Winter 2022,
published by Scientific American, a division of Springer Nature America, Inc., 1 New York Plaza, Suite 4600, New York, N.Y. 10004-1562. Canadian BN No. 127387652RT;
TVQ1218059275 TQ0001. To purchase additional quantities: U.S., $13.95 each; elsewhere, $17.95 each. Send payment to Scientific American Back Issues, P.O. Box 3187, Harlan Iowa 51537. Inquiries:
fax 212-355-0408 or telephone 212-451-8415. Printed in U.S.A.

SCIENTIFICAMERICAN.COM  |  3
 What
Is
Conscıousness?

How Matter Becomes Mind page 6

How to Make a Consciousness Meter page 14

Can Lab-Grown Brains Become Conscious? page 20

The Brain Electric page 26

How Do I Know I’m Not the Only

Conscious Being in the Universe? page 32

Illustration by Maria Corte

SCIENTIFICAMERICAN.COM  |  5
 how
matter
becomes
mınd The new discipline of network neuroscience
yields a picture of  how mental activity arises
from carefully orchestrated interactions
among different brain areas
By Max Bertolero and Danielle S. Bassett

Illustration by Mark Ross Studios

6  |  SCIENTIFIC AMERICAN  |  SPECIAL EDITION  |  WINTER 2022

SCIENTIFICAMERICAN.COM  |  7
N etworks pervade our lives. Every day we use intricate networks
of roads, railways, maritime routes and skyways traversed by commercial
flights. They exist even beyond our immediate experience. Think of the
World Wide Web, the power grid and the universe, of which the Milky Way
is an infinitesimal node in a seemingly boundless network of galaxies. Few
such systems of interacting connections, however, match the complexity of
the one underneath our skull.
Neuroscience has gained a higher profile in recent years, as
many people have grown familiar with splashily colored imag-
es that show brain regions “lighting up” during a mental task.
There is, for instance, the temporal lobe, the area by your ear,
which is involved with memory, and the occipital lobe at the back
of your head, which dedicates itself to vision.
What has been missing from this account of human brain
function is how all these distinct regions interact to give rise to
To understand how n
THE MUSIC OF THE MIND
 etworks underlie our cognitive capabil-
ities, first consider the analogy of an orchestra playing a sympho-
ny. Until recently, neuroscientists have largely studied the func-
tioning of individual brain regions in isolation, the neural equiv-
alent of separate brass, percussion, string and woodwind sections.
In the brain, this stratification represents an approach that dates
back to Plato—quite simply, it entails carving nature at the joints
who we are. Our laboratory and others have borrowed a language and then studying the individual components that remain.
from a branch of mathematics called graph theory that allows Just as it is useful to understand how the amygdala helps to
us to parse, probe and predict complex interactions of the brain process emotions, it is similarly vital to grasp how a violin produc-
that bridge the seemingly vast gap between frenzied neural elec- es high-pitched sounds. Still, even a complete list of brain regions
trical activity and an array of cognitive tasks—sensing, remem- and their functions—vision, motor, emotion, and so on—does not
bering, making decisions, learning a new skill and initiating tell us how the brain really works. Nor does an inventory of instru-
movement. This new field of network neuroscience builds on ments provide a recipe for Beethoven’s E  roica symphony.
and reinforces the idea that certain regions of the brain carry Network neuroscientists have begun to tame these mysteries
out defined activities. In the most fundamental sense, what the by examining the way each brain region is embedded in a larger
brain is—and thus who we are as conscious beings—is, in fact, network of such regions and by mapping the connections be-
defined by a sprawling network of 100 billion neurons with at tween regions to study how each is embedded in the large, inte-
least 100 trillion connecting points, or synapses. grated network that is the brain. There are two major approach-
Network neuroscience seeks to capture this complexity. We es. First, examining structural connectivity captures the instru-
can now model the data supplied by brain imaging as a graph mentation of the brain’s orchestra. It is the physical means of
composed of nodes and edges. In a graph, nodes represent the creating the music, and the unique instrumentation of a given
units of the network, such as neurons or, in another context, air- musical work constrains what can be played. Instrumentation
ports. Edges serve as the connections between nodes—think of matters, but it is not the music itself. Put another way, just as a
one neuron intertwined with the next or contemplate airline collection of instruments is not music, an assemblage of wires
flight routes. In our work, the human brain is reduced to a graph does not represent brain function.
of roughly 300 nodes. Diverse areas can be linked together by Second, living brains are massive orchestras of neurons that
edges representing the brain’s structural connections: thick bun- fire together in quite specific patterns. We hear a brain’s music by
dles of tubular wires called white matter tracts that tie togeth- measuring the correlation between the activity of each pair of re-
er brain regions. This depiction of the brain as a unified network gions, indicating that they are working in concert. This measure
has already furnished a clearer picture of cognitive functioning, of joint activity is known as functional connectivity, and we collo-
along with the practical benefit of enabling better diagnoses and quially think of it as reflecting the music of the brain. If two re-
treatment of psychiatric disorders. As we glimpse ahead, an un- gions fire with the same time-varying fluctuations, they are con-
derstanding of brain networks may lead to a blueprint for im- sidered to be functionally connected. This music is just as impor-
proved artificial intelligence, new medicines and electrical-stim- tant as the decibels produced by a French horn or viola. The volume
ulation technology to alter malfunctioning neural circuitry in of the brain’s music can be thought of as the level of activity of elec-
depres­­sion—and perhaps even the development of genetic ther- trical signals buzzing about one brain area or another.
apies to treat mental illness. At any moment, though, some areas within the three-pound

8  |  SCIENTIFIC AMERICAN  |  SPECIAL EDITION  |  WINTER 2022

organ are more active than others. We have all heard the saying tal control module spans the frontal, parietal and temporal lobes.
that people use a small fraction of their brain capacity. In fact, It developed relatively recently on the timescale of evolution.
the entire brain is active at any point in time, but a given task The module is especially large in humans, relative to our closest
modulates the activity of only a portion of the brain from its base- primate ancestors. It is analogous to an orchestra conductor and
line level of activity. becomes active across a large number of cognitive tasks.
That arrangement does not mean that you fulfill only half of The frontoparietal module ensures that the brain’s multiple
your cognitive potential. In fact, if your entire brain were strongly modules function in unison. It is heavily involved in what is
active at the same time, it would be as if all the orchestra members called executive function, which encompasses the separate pro-
were playing as loudly as possible—and that scenario would create cesses of decision-making, short-term memory and cognitive
chaos, not enable communication. The deafening sound would not control. The last is the ability to develop complex strategies and
convey the emotional overtones present in a great musical piece. inhibit inappropriate behavior.
It is the pitch, rhythms, tempo and strategic pauses that commu- Another highly interconnected module is the salience mod-
nicate information, both during a symphony and inside your head. ule, which hooks up to the frontoparietal control module and
contributes to a range of behaviors related to attention and re-
MODULARITY sponding to novel stimuli. For example, take a look at two words:
J us t a s a n o r c h e s t ra c an be divided into groups of instru- blue and red. If you are asked to respond with the color of the
ments from different families, the brain can be separated into word, you will react much faster to the one set in red. The fron-
collections of nodes called modules—a description of localized toparietal and salience modules activate when responding to
networks. All brains are modular. Even the 302-neuron network the color green because you have to suppress a natural inclina-
of the nematode C  aenorhabditis elegans has a modular struc- tion to read the word as “blue.”
ture. Nodes within a module share stronger connections to one Finally, the default mode module spans the same lobes as the
another than to nodes in other modules. frontoparietal control network. It contains many hubs and is
Each module in the brain has a certain function, just as ev- linked to a variety of cognitive tasks, including introspective
ery family of instruments plays a role in the symphony. We re- thought, learning, memory retrieval, emotional processing, in-
cently performed an evaluation of a large number of indepen- ference of the mental state of others and even gambling. Criti-
dent studies—a meta-analysis—that included more than 10,000 cally, damage to these hub-rich modules disturbs functional con-
functional magnetic resonance imaging (fMRI) experiments of nections throughout the brain and causes widespread cognitive
subjects performing 83 different cognitive tasks and discovered difficulties, just as bad weather at a hub airport delays air traf-
that separate tasks map to different brain-network modules. fic all over the country.
There are modules occupied with attention, memory and intro-
spective thought. Other modules, we found, are dedicated to PERSONAL CONNECTIONS
hearing, motor movement and vision. A lt h o u g h o u r b ra i n s h
 ave certain basic network compo-
These sensory and motor cognitive processes involve single, nents—modules interconnected by hubs—each of us shows slight
contiguous modules, most of which are confined to one lobe of variations in the way our neural circuits are wired. Researchers
the brain. We also found that computations in modules do not have devoted intense scrutiny to this diversity. In an initial phase
spur more activity in other modules—a critical aspect of modu- of what is called the Human Connectome Project, 1,200 young
lar processing. Imagine a scenario in which every musician in people volunteered to participate in a study of brain-network ar-
an orchestra had to change the notes played every time anoth- chitecture, funded by the National Institutes of Health. (The fi-
er musician changed his or her notes. The orchestra would spi- nal goal of the project is to cover the entire life span.) Each indi-
ral out of control and would certainly not produce aesthetically vidual’s structural and functional connectivity networks were
pleasing sounds. Processing in the brain is similar—each mod- probed using fMRI. These data were supplemented by a cogni-
ule must be able to function mostly independently. Philosophers tive battery of testing and questionnaires to analyze 280 behav-
as early as Plato and as recent as Jerry Fodor have noted this ne- ioral and cognitive traits. Participants provided information about
cessity, and our research confirms it. how well they slept, how often they drank alcohol, their language
Even though brain modules are largely independent, a sym- and memory abilities, and their emotional states. Neuroscientists
phony requires that families of instruments be played in unison. from all over the world have been poring over this incredibly rich
Information generated by one module must eventually be inte- data set to learn how our brain networks encode who we are.
grated with other modules. Watching a movie with only a brain Using data from hundreds of participants in the Human Con-
module for vision—without access to the one for emotions— nectome Project, our lab and others have demonstrated that
would detract greatly from the experience. brain-connectivity patterns establish a “fingerprint” that distin-
For that reason, to complete many cognitive tasks, modules guishes each individual. People with strong functional connec-
must often work together. A short-term memory task—holding a tions among certain regions have an extensive vocabulary and
new phone number in your head—requires the cooperation of au- exhibit higher fluid intelligence—helpful for solving novel prob-
ditory, attention and memory-processing modules. To integrate lems—and are able to delay gratification. They tend to have more
and control the activity of multiple modules, the brain uses hubs— education and life satisfaction and better memory and attention.
nodes where connections from the brain’s different modules meet. Others with weaker functional connections among those same
Some key modules that control and integrate brain activity brain areas have lower fluid intelligence, histories of substance
are less circumspect than others in their doings. Their connec- use, poor sleep and a decreased capacity for concentration.
tions extend globally to multiple brain lobes. The frontoparie- Inspired by this research, we showed that the findings could

SCIENTIFICAMERICAN.COM  |  9
 Decoding 100 Trillion Messages Module 1 Node

The Milky Way h  as hundreds of billions of stars—just a fraction of the 100 trillion connections
in our brains that enable us to sense, think and act. To unravel this complexity, network
neuroscientists create a map, or “graph,” consisting of nodes linked by edges that fit into
Local hub
modules, which are tethered to one another with highly connected nodes called hubs.
Connector hub

From Modules to Hubs to Thoughts Module 2

Collections of nodes form modules that devote themselves to processing vision, attention and motor
behaviors, among other tasks ● A . Some of the nodes act as local hubs that link to other nodes in their own
module. A node that has many linkages to a lot of modules is known as a connector hub (the type
most commonly referenced in this article) ● B . Its diverse connections across the brain’s modules are
critical for many tasks, particularly complex behaviors ● C.

Brain Modules
Visual
Attention
Frontoparietal control
Somatic motor
Salience
B Connector hubs
Default with the strongest
links to multiple other
Limbic
modules appear in this
side view, colored to
A Seven key modules, indicate the seven
denoted by colors, pivotal brain modules.
spread across sometimes
disconnected areas
of the brain.

C A graph of the human brain’s nodes and edges shows the strongest connector hubs
represented as large circles. Each node’s color represents the module it belongs to.
Nodes can be visualized as repelling magnets with edges between nodes acting
as springs that hold them together. Tightly connected nodes cluster together.
Connector hubs occupy the center because they are well connected to all modules.

10  |  SCIENTIFIC AMERICAN  |  SPECIAL EDITION  |  WINTER 2022

 Putting It Together Task Brain Module
Modules for vision, attention and other cognitive Braille reading Strength of

Strongest
functions are dedicated to specific tasks, often re­p­ Visual tracking relationship
resented here by psychological tests. The most active Action observation
tasks rise to the top. The visual module, for instance,
Picture naming (silently)
is involved with naming, reading and observing. Many
tasks require multiple modules. For example, a mental Brightness perception
Visual
rotation task recruits both the visual and the attention Picture naming (out loud)
modules. Some modules are entrusted with more Silent reading
abstract tasks. The frontoparietal module engages Visual attention
in switching tasks or recalling lists. The default mode Drawing
module attends to subjective emotional states or passive
Controlling eye movement
listening when a person is at rest.
Mental rotation
Visual control Attention
Pointing
Writing
Imagined movement
Wisconsin Card Sorting Test (reasoning)
Counting
Tower of London (complex planning task)
n-back working memory task
Sternberg working memory task Frontoparietal
Task switching control
Word stem completion (out loud)
Free word list recall
Stroop task
Flanker response inhibition task
Detecting vibrations through touch
Finger tapping
Vocal rehearsing Somatic
Small hand movements motor
Whistling
Grasping
Isometric force
Awareness of need to urinate
Stimulation monitoring
Nonpainful electrical stimulation
Breath holding Salience
Word stem completion (silent)
Playing music
Imaging what others think
Categorizing emotional scenes
Passive listening
Lying
Pitch detection
Event recall (episodic memory) Default
Delayed gratification
Word generation (out loud)
Word meaning discrimination
Grammar
Face-emotion identification
Scent detection
Video games
Classical conditioning Limbic
Eating/drinking
Passive viewing
Monetary reward task

Graphics by Max Bertolero (brains and network diagram) and Jen Christiansen (task chart) SCIENTIFICAMERICAN.COM  |  11
be described by particular patterns tent strengthening of the structur-
among the hub connections. If al connections to hubs throughout
your brain network has strong the course of childhood is associ-
hubs with many connections ated with an increase in the segre-
across modules, it tends to have gation between modules and an
modules that are clearly segregat- augmentation in the efficiency
ed from one another, and you will with which young people perform
perform better on a range of tasks, executive tasks such as complex
from short-term memory to math- reasoning and self-regulation. We
ematics, language or social cogni- have also found that the segrega-
tion. Put simply, your thoughts, tion of modules from one another
feelings, quirks, flaws and mental is more rapid in children who have
strengths are all encoded by the a higher socioeconomic status,
specific or­­ganization of the brain highlighting the key impact of
as a unified, integrated network. In their environment.
sum, it is the music your brain Although changes in structural
plays that makes you y  ou. connectivity are slow, the reconfig-
The brain’s synchronized mod- uration of functional connections
ules both establish your identity can occur quickly, in a few seconds
and help to retain it over time. The or minutes. These rapid shifts are
musical compositions they play ap- instrumental for moving between
pear to always be similar. The like- tasks and for the massive amount
ness could be witnessed when par- of learning demanded even by a
ticipants in two other studies in the MULTITUDES o  f white matter connections in this scan single task. In a set of studies that
Human Connectome Project en- are used to model the brain’s physical pathways­— we published from 2011 to 2019, we
gaged in various tasks that in­­volved functional networks use these structural linkages found that networks with modules
short-term memory, recognition of to carry out an array of cognitive tasks. that can change readily turn up in
the emotions of others, gambling, individuals who have greater exec-
finger tapping, language, mathematics, social reasoning and a utive function and learning capacity.
self-induced “resting state” in which they let their mind wander. To better understand what was happening, we used publicly
Fascinatingly, the networks’ functional wiring has more sim- available data from a landmark study known as MyConnectome,
ilarities than expected across all these activities. Returning to in which Stanford University psychology professor Russell
our analogy, it is not as if the brain plays Beethoven when doing Poldrack personally underwent imaging and cognitive apprais-
math and Tupac when resting. The symphony in our head is the als three times a week for more than a year. Whereas modules
same musician playing the same musical genre. This consisten- are mostly autonomous and segregated, at times the brain will
cy derives from the fact that the brain’s physical pathways, or spontaneously reorganize its connections. This property, called
structural connections, place constraints on the routes over the functional network flexibility, lets a node with strong function-
brain’s integrated network that a neural signal can travel. And al connections within a module suddenly establish many con-
those pathways delineate how functional connections—the ones, nections to a different module, changing the flow of information
say, for math or language—can be configured. In the musical through the network. Using data from this study, we found that
metaphor, a bass drum cannot play the melodic line of a piano. the rerouting of a network’s connections changes from day to
Changes in the brain’s music inevitably occur, just as new ar- day in a manner that matches positive mood, arousal and fa-
rangements do for orchestral music. Physical connections un- tigue. In healthy individuals, such network flexibility correlates
dergo alterations over the course of months or years, whereas with better cognitive function.
functional connectivity shifts on the order of seconds, when a
person switches between one mental task and the next. DISSONANT NOTES
Danielle S. Bassett and Matthew Cieslak/University of Pennsylvania

Transformations in both structural and functional connectiv- The configuration o  f brain connections also re­­flects one’s
ity are important during adolescent brain development, when the mental health. Aberrant connectivity patterns accompany depres-
finishing touches of the brain’s wiring diagram are being refined. sion, schizophrenia, Alzheimer’s, Parkinson’s, autism spectrum
This period is of critical importance because the first signs of men- disorder, attention deficit disorder, dementia and epilepsy.
tal disorders often appear in adolescence or early adulthood. Most mental illnesses are not confined to one area of the brain.
One area our research relates to is understanding how brain The circuitry affected in schizophrenia extends quite widely
networks develop through childhood and adolescence and into across the entire organ. The so-called disconnectivity hypothe-
adulthood. These processes are driven by underlying physiolog- sis for schizophrenia holds that there is nothing abnormal about
ical changes, but they are also influenced by learning, exposure the individual modules. Instead the disarray relates to an over-
to new ideas and skills, an individual’s socioeconomic status and abundance of connections between modules.
other experiences. In a healthy brain, modules are mostly autonomous and seg-
Brain-network modules emerge very early in life, even in the regated, and the ability to bring about flexible changes in net-
womb, but their connectivity is refined as we grow up. Consis- work connections is beneficial for cognitive functioning—with-

12  |  SCIENTIFIC AMERICAN  |  SPECIAL EDITION  |  WINTER 2022

in certain limits. In our lab, we found that in the brains of peo- STATES OF MIND
ple with schizophrenia and their first-degree relatives, there is When Nobel Prize–winning physicist R  ichard Feynman died
an overabundance of flexibility in how networks reconfigure in 1988, his blackboard read, “What I cannot create, I do not un-
themselves. Auditory hallucinations might result when nodes derstand.” He created a beautiful aphorism, yet it misses a pivotal
unexpectedly switch links between speech and auditory mod- idea: it should be revised to “What I cannot create and control, I
ules. The uninvited mix can result in what seem to be the utter- do not understand.” Absent such control, we still know enough to
ings of voices in one’s head. enjoy a symphony, even if we do not qualify to be the conductor.
Like schizophrenia, major depressive disorder is not caused When it comes to the brain, we have a basic understanding of
by a single abnormal brain region. Three specific modules ap- its form and the importance of its network architecture. We know
pear to be affected in depression: the frontoparietal control, sa- that our brain determines who we are, but we are just beginning
lience and default mode modules. In fact, the symptoms of de- to understand how it all happens. To rephrase mathematician
pression—emotional disinhibition, altered sensitivity to emo- Pierre-Simon Laplace’s explanation of determinism and mechan-
tional events and rumination—map to these modules. ics and apply it to the brain, one’s present brain, and so one’s men-
As a result, normal communication among the three modules tal state, can be thought of as a compilation of past states that
becomes destabilized. Activities from module to module typical- can be used to predict the future. A neuroscientist who knew all
ly tug back and forth to balance the cognitive processing of sen- the principles of brain function and everything about someone’s
sory inputs with more introspective thoughts. In depression, brain could predict that person’s mental conditions—the future,
though, the default mode dominates, and the afflicted person as well as the past, would be present inside the person’s mind.
lapses into ruminative thought. The music of the brain thus be- This knowledge could be used to prevent pain and suffering,
comes in­­creasingly unbalanced, with one family of instruments given that many mental illnesses are associated with network
governing the symphony. These observations have broadened abnormalities. With enough engineering ingenuity, we may de-
our understanding of the network properties of depression to velop implanted devices that alter or even generate new brain
the extent that a connectivity pattern in a brain can allow us to networks or edit genomes to prevent the disorganized networks
diagnose certain subtypes of the disorder and determine which associated with mental disorders from occurring in the first
areas should be treated with electrical-stimulation technology. place. Such an achievement would enable us to treat diseases
and to restore brain function after stroke or injury and enhance
NETWORKS EVOLVE it in healthy individuals.
B e s i d e s s t u dy i n g d e v e l o p m e n t, n
 etwork neuroscientists Before those futuristic scenarios materialize, two major gaps
have begun to ask why brain networks have taken their present must be filled: we need to know more about how personal ge-
form over tens of thousands of years. The areas identified as netics, early-life development and environment determine one’s
hubs are also the locations in the human brain that have expand- brain’s structure and how that structure leads to functional ca-
ed the most during evolution, making them up to 30 times the pacities. Neuroscientists have some knowledge from the human
size they are in macaques. Larger brain hubs most likely permit genome about the structure that gives rise to functional net-
greater integration of processing across modules and so support works but still need to learn precisely how this process occurs.
more complex computations. It is as if evolution increased the We are starting to grasp the way brain networks develop and are
number of musicians in a section of the orchestra, fostering more shaped by the environment but are not close to explaining the
intricate melodies. entire complexity of this process. The brain’s wiring, its struc-
Another way neuroscientists have explored these questions tural connectivity, constrains how various modules interact with
is by creating computer-generated networks and subjecting them one another, but our knowledge remains limited. As we fill in
to evolutionary pressures. In our lab, we have probed the evolu- these gaps, chances improve for interventions to guide brain
tionary origins of hubs. This exercise started with a network in functioning into healthy trajectories.
which all edges were placed uniformly at random. Next, the net- What holds us back, for the moment, is our still blurry vision
work was rewired, mimicking natural selection to form segre- of the brain—it is as if we are outside the concert hall and have
gated modules and display a property known in network science seen only sketches of the instruments. Inside each brain region
as small-worldness, in which paths form to let distant network that neuroscientists study are millions of neurons firing every
nodes communicate with surprising ease. Thousands of such millisecond, and we are able just to indirectly measure their av-
networks then evolved, each of which ultimately contained hubs erage activity levels every second or so. Thus far we can rough-
strongly connected to multiple modules but also tightly inter- ly identify the human brain’s structural connections. Luckily, sci-
connected to one another, forming what is called a club. Noth- entists and engineers have taken steps to deliver ever clearer
ing in the selection process explicitly selected for a club of hubs— data that will enable a deeper look into perhaps the most com-
they simply emerged from this iterative process. plex network in the known universe: your brain. 
This simulation demonstrates that one potential solution to
evolving a brain capable of exchanging information among mod- Max Bertolero i s a research scientist at the University of Pennsylvania. He received
ules requires hubs with strong connections. Notably, real net- a doctorate in systems neuroscience from the University of California, Berkeley, and
works—brains, airports, power grids—also have durable, tight- undergraduate degrees in philosophy and psychology from Columbia University.
ly interconnected hubs, exactly as predicted by evolutionary ex-
periments. That observation does not mean evolution necessarily Danielle S. Bassett is an associate professor in the department of bioengineering at the
occurred in the same way as the simulation, but it shows a pos- University of Pennsylvania, where she studies networks in physical and biological systems.
sible means by which one of nature’s tricks might operate. In 2014 she became a MacArthur Fellow.

SCIENTIFICAMERICAN.COM  |  13
HOW TO MAKE A
CONSCIOUSNESS
CONSCIOUSNESS
METER Zapping the brain with magnetic
pulses while measuring its electrical
activity is proving to be a reliable
way to detect consciousness
By Christof Koch
Illustration by Ashley Mackenzie

14  |  SCIENTIFIC AMERICAN  |  SPECIAL EDITION  |  WINTER 2022

SCIENTIFICAMERICAN.COM  |  15
I h ave d i ed m a n y ti m es o v e r . E ve ry night w he n I lay dow n my w e ary se l f
to rest, my consciousness is extinguished. I experience nothing until I wake up inside
my sleeping body—in a dream disconnected from the external world. Or later con-
sciousness resurfaces in the morning on my return to the wakening world.
Daily life contains many such experi-
ences. In my childhood, I had an appen-
dectomy and was anesthetized—my con-
sciousness was switched off and, follow-
ing the surgery, was restored. A fading
memory from my teenage years places me
in the back seat of a Renault that is driv-
ing down a tree-lined avenue in North Af-
rica. Suddenly, the scenery changes
abruptly. I’m on the same street, seeing
medical or research settings to determine
whether a person is experiencing any-
thing at all. This ability to detect con-
sciousness could also help physicians and
family members make critical decisions,
such as withdrawal of life-sustaining
therapy, for tens of thousands of uncom-
municative patients.

RECORDING BRAIN WAVES

There are, of course, other ways to re-
cord brain activity besides the venerable
EEG. The most common tools measure
the dynamics of blood flowing inside the
brain with magnetic scanners or track the
magnetic field around the brain with
magnetoencephalography (MEG). Yet
these instruments, as well as more recent
techniques such as near-infrared spec-
troscopy, come with methodological and
things now from the ground up. The car C o n t e m p l at i n g t h e p
 ossibility of a practical issues that preclude their rou-
had hit a tree, ejecting me onto the cob- consciousness meter requires consider- tine clinical use for the time being.
blestones, and I lost consciousness. ation of the internal dynamics of our The EEG measures the tiny voltage fluc-
Many readers will have had similar mental life, activity that waxes and wanes tuations (10 to 100 microvolts) generated
recollections of consciousness lost and re­­ within fractions of a second, dictating the by electrical activity across the neocortex,
gained. We are used to the diurnal cycle of measuring of those fluctuating brain sig- the brain’s outer surface, which is respon-
waking, sleeping and dreaming. But that nals at a similar timescale. The most im- sible for perception, action, memory and
experience is not the same for everyone. portant physiological tool to infer con- thought. The main actors whose collective
For some patients with brain trauma, con- sciousness from probing the brain has electrical activity is thought to be responsi-
sciousness flees for days, weeks or longer. been, and continues to be, the electroen- ble for the EEG signals, via a mechanism
In practice, a clinician may have diffi- cephalogram (EEG). known as volume conduction, are cortical
culty establishing whether someone is The EEG was developed by German pyramidal neurons, named for their tetra-
quietly sleeping, anesthetized or severely psychiatrist Hans Berger, whose lifelong hedral shape. Contributions from deeper
brain-injured. Is a person lying with eyes quest was to uncover the link between ob- structures, such as the thalamus, have to be
open experiencing anything, no matter jective brain activity and subjective phe- inferred indirectly via their action on corti-
the content, or has the conscious mind nomena. He recorded the first ever brain cal cells. The technology relies on elec-
fled the body and left no one at home? waves of a patient in 1924 but, filled with trodes placed directly on the scalp—that is,
Ideally, a technology could be devised doubt, did not publish his findings until without the need for invasive surgery to
to serve as a form of consciousness meter 1929. The rest is history, as the EEG be- penetrate the skull. With the move toward
to answer these questions. At first, the came the foundational tool of an entire high-density EEG setups—with up to 256
idea of the equivalent of a blood pressure field of medicine called clinical neuro- electrodes—maps showing the distribu-
cuff for consciousness might seem ab­­ physiology, although Berger was never ac- tion of electrical activity across the brain
surd. But the development of several new corded any significant recognition in Nazi have become commonplace.
technologies has raised real prospects for Germany and hanged himself in 1941, de- Still, placing the electrodes with their
detectors that meet the criteria for con- spite being nominated for the Nobel Prize wet, conductive gel onto the scrubbed
sciousness meters—devices useful in several times. skin of the head is cumbersome, time-

16  |  SCIENTIFIC AMERICAN  |  SPECIAL EDITION  |  WINTER 2022

 consuming, and prone to errors if
the electrodes move, all of which
limits the technology. With today’s Zapping and Zipping
more sensitive dry electrodes, the In pursuit of a consciousness test, S
 ilvia Casarotto of the University of Milan and her colleagues
EEG is morphing from a clinical recruited 102 nonbrain-injured subjects and 48 still responsive and awake brain-injured pa­­
tool into a consumer de­­­­­­­vice that tients. Their brains were “zapped” with magnetic pulses (transcranial magnetic stimulation) in
can be used for biofeedback—al- both conscious and unconscious states, and brain activity was detected with an EEG and ana­
lowing athletes or do-it-yourself lyzed with a data-compression algorithm—and so it was said to be “zipped.” A value known
“brain hackers” to focus their as a perturbational complexity index (PCI) was calculated for the EEGs—and partici­pants were
thoughts or insomniacs to track, also interviewed about their state of mind. It was determined that a con­scious per­son exhibited
deepen and extend their naturally at least one value above 0.31 (PCI*), whereas unconscious subjects all had lower scores. Using
occurring sleep. this value, the zap-and-zip testing was then performed on patients with se­­vere disorders of
From the late 1940s onward, de- consciousness (results not shown), finding some individuals who appeared to be conscious.
tection of an “activated EEG” signal
was the surest sign of a ­conscious PCI PCI* cutoff (0.31)
subject. This state is characterized 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
by low-voltage, rapid up-and-down Unresponsive, Unconscious
fluctuating waves that are desyn-
chronized rather than in lockstep
across the skull. In general, as the Non-REM sleep
EEG shifts to lower frequencies,
consciousness is less likely to be Midazolam anesthesia
present. Yet there are enough excep- Xenon anesthesia
tions to this rule that it cannot serve
Propofol anesthesia
as a general basis to diagnose ab­­
Unresponsive, Conscious
sence or presence of consciousness
in a given individual. Thus, scien- REM sleep
tists and clinicians alike have cast Ketamine anesthesia
about for more reliable measures
Responsive, Conscious
and have now found one based on a
fundamental property of any con- Each dot shows
scious experience. the highest PCI
Nonbrain-injured subjects

for one subject

INTO THE NETHERWORLD

Before we come t o that, we should
consider why clinicians care about
detecting consciousness in two dis-
tinct groups of patients (pediatric
Wakefulness
patients represent a different chal-
Source: “Stratification of Unresponsive Patients by an Independently Validated Index of Brain Complexity,”

lenge that will not be addressed

here). The first consists of adults
with severe disorders of conscious-
ness following traumatic brain inju-
ry caused by gunshots, falls, acci-
by Silvia Casarotto et al., in A nnals of Neurology, Vol. 80, No. 5; November 2016

dents, and so on, infections of the

brain (encephalitis) or its surround-
ing protective layers (meningitis),
stroke, or drug or alcohol intoxica-
tion. After surviving the initial in-
sult, patients are stable but disabled Locked-in syndrome
and bedridden, unable to speak or
Brain-injured patients

Subcortical stroke
signal their thoughts and inten-
tions. With proper nursing care to
avoid bedsores and infections, these
Cortical stroke
patients can live for years.
In this first group, clinicians dis-
tinguish several subcategories. Pa-
Severe brain injury
tients in a vegetative state, which is
better described by the less pejora-

Graphic by Amanda Montañez SCIENTIFICAMERICAN.COM  |  17

tive term “unresponsive wakefulness syn- parts. Anesthesia eliminates pain and oth- The current most promising scientific
drome” (UWS), cycle in and out of sleep. er conscious experiences, prevents mobil- theory of consciousness, which encompass-
Yet setting up a bedside communication ity and stabilizes the autonomic nervous es both of these ideas, is Integrated Infor-
channel—“if you hear me, squeeze my system, which controls breathing and oth- mation Theory (IIT). Devised by Giu­lio
hand or move your eyes”—meets with fail- er functions, for hours at a time. Tononi, a psychiatrist and neuroscientist at
ure. UWS patients do swallow, yawn, and Patients “go under” with the expecta- the University of Wisconsin–Madison, IIT
open and move their eyes or head but not tion that they will not wake up during emphasizes the differentiated and integrat-
in a seemingly intentional manner. No surgery and that they will not have to con- ed aspect of any subjective experience and
willed actions are left—only brain stem tend with traumatic memories of intraop- postulates that the mechanism supporting
reflexes, activity that controls basic pro- erative experiences that could haunt them conscious experience in the human brain’s
cesses such as breathing, sleep-wake for the rest of their lives. Unfortunately, this neocortex must likewise incorporate these
transitions, heart rate, eye movements goal is not always met. Intraoperative recall, two attributes. To probe the extent to which
and pupillary re­­sponses. Terri Schiavo is or “awareness under anesthesia,” can occur these mechanisms are intact, Tononi, to-
a name many remember, a patient in in a small number of operations, estimated gether with a team that included neurolo-
Florida who, following cardiac arrest, was to be in the one-per-1,000 range, in particu- gist and neuroscientist Marcello Massimini,
resuscitated and lingered for 15 years in lar when patients are paralyzed during a now at the University of Milan in Italy, de-
UWS until her medically induced death in procedure by an anesthesiologist to facili- vised an EEG-based method back in the ear-
2005. UWS patients are a modern phe- tate intubation and prevent gross muscle ly 2000s. It provides a very crude approxi-
nomenon depending for their survival on movements. Given that millions of Ameri- mation of IIT’s formal calculus. The team
the infrastructure of 911, emergency heli- cans undergo surgical-level anesthesia ev- verified its basic soundness by correctly dis-
copters and advanced medical care. ery year, this tiny fraction translates into criminating between when six healthy vol-
There are more than 10,000 such individ- thousands of awakenings under anesthesia. unteers were conscious but quietly resting
uals in the U.S. alone, living in hospices or Existing EEG measures monitor depth with eyes closed and when they were deep-
nursing homes or at home. of anesthesia during an operation. Yet ly asleep and therefore unconscious.
Whereas behavioral evidence is com- none of the vast diversity of anesthetic The brain of a deep sleeper acts like a
patible with the notion that UWS patients agents work in a consistent manner across stunted, badly tuned bell. Whereas the
do not experience anything, it is impor- all patients, who range from neonates to initial amplitude of the EEG is larger than
tant to recall that “absence of evidence is birthing mothers, the very elderly or the when the subject is awake, its duration is
not evidence of absence” and to give the very sick. What is needed is a tool that can much shorter, and it does not reverberate
patients the benefit of doubt. There is a di- reliably track the presence of conscious- across the cortex to connected regions.
agnostic gray zone into which UWS pa- ness in individual subjects across a large While neurons remain active in deep sleep,
tients fall as to the question of whether spectrum of normal and pathological con- as evidenced by the strong response in a
their injured brains are capable of experi- ditions under both acute (anesthesia) and local brain region, integration has broken
encing pain, distress, anxiety, isolation, chronic conditions (the plight of neurolog- down. Little of the electrical activity found
quiet resignation, a full-blown stream of ically impaired patients). in an awake brain is present.
thought—or perhaps just nothing. Some Although distinguishing the brain’s re-
studies have suggested that 20 percent of THE NATURE OF sponse during a restful state from its re-
UWS patients are conscious and are there- CONSCIOUS EXPERIENCE sponse while sleeping may seem trivial,
fore misdiagnosed. To family and friends To detect consciousness, it is necessary the method can be extended to the more
who may care for their loved one for years, to consider two essential characters of any difficult task of discriminating among a
knowing whether anybody is mentally subjective experience, no matter how variety of brain states. Indeed, in subse-
there can make a dramatic difference. mundane or exalted. First, by definition, quent years, Tononi, Massimini and 17 ad-
The situation is less ambiguous for min- any experience is different from all other ditional doctors and brain scientists tested
imally conscious state (MCS) patients. Un­­ experiences. It is specific to the moment the procedure in many more subjects. A
able to speak, they can signal but often and place in which it occurs. Each one is paper summarizing this landmark study
only in a sparse, minimal and erratic fash- highly informative—take the unique visu- was published in 2016 in the peer-re-
ion, smiling or crying in appropriate emo- al richness associated with a mountain viewed literature.
tional situations, vocalizing or gesturing on hike in the Rockies or another in the Cas- The method perturbs the brain by send-
occasion, or tracking salient objects with cade Range. Now combine these recollec- ing one or two pulses of magnetic energy
their eyes. Here the assumption is that tions with other sensory modalities, such via an enclosed coil of wire held against the
these patients do experience something, as sounds and smells, emotions and mem- scalp, a method called transcranial mag-
however minimal, at least some of the time. ories. Each one is distinct in its own way. netic stimulation, or TMS. This technique
The need to monitor consciousness The second point is that each experience is induces a brief electric current in the un-
also arises in a second, totally different set seamless, integrated and holistic. You can- derlying cortical neurons, which, in turn,
of patients who have a normally function- not separate the iconic percept of black engage other neurons in a cascade that re-
ing brain—people who undergo invasive smoke arising from the burning Twin verberates inside the head before the elec-
surgery for the usual host of ills, such as Towers on a backdrop of blue sky into a trical surge dies out in a fraction of a sec-
injuries, removal of a cancerous growth, half experience of the North Tower and ond. Think of the brain as a large church
or repair of knees, hips and other body another half experience of the South Tower. bell and the TMS device as the clapper.

18  |  SCIENTIFIC AMERICAN  |  SPECIAL EDITION  |  WINTER 2022

Once struck, a well-cast bell will ring at its
characteristic pitch for a considerable time,
A measurement value derived by
as does the brain. Its electrical activity is researchers enabled them to establish
a critical threshold—the minimum
monitored by a high-density EEG cap worn
by the patient. The EEG is averaged and dis-
played during the course of 200 TMS puls-
es, as if it were a movie unfolding in time.
degree of complex brain activity
In an awake brain, with intact connec- supporting consciousness.
tivity, this monitoring of different areas in
response to the probe shows a highly
complex pattern over much of the cortex, compassed two groups. One consisted of wakeful one. In the MCS group, consisting
activity that is neither totally predictable 102 volunteers with no known brain im- of patients with at least some signs of be-
nor completely random—and is emblem- pairment who experienced various con- havior be­­yond reflexive functions such as
atic of what is meant by “complex.” scious or unconscious states: quietly breathing, the method correctly assigned
The researchers estimate its complex- awake with eyes closed or dreaming dur- consciousness to 36 of 38 patients, misdi-
ity, the extent to which this response dif- ing REM sleep (the latter is also a con- agnosing the other two as unconscious. Of
fers across the cortex and across time, us- scious state the researchers assessed by the 43 UWS patients, in which communi-
ing a mathematical measure capturing its randomly awakening the sleepers during cation failed, 34 had a brain response
diversity. The technique itself is borrowed REM sleep and only including their EEGs whose complexity was less than that of
from computer science and is the basis of in the final results if they reported any anyone of the benchmark population
the popular “zip” compression algorithm dream experience immediately prior to when conscious, an expected result. That
for reducing the storage demand of imag- awakening). The EEGs were also as­­sessed is, the complexity of their EEG responses
es or movies, which is why the entire pro- under anesthesia using ketamine, a phar- was comparable to that of the benchmark
cedure of measuring consciousness is macological agent that disconnects and group when not detecting consciousness.
known in the trade as zap and zip. Ulti- dissociates the mind from the external Much more troubling, however, were
mately each person’s EEG response is world but does not extinguish conscious- the other nine patients who responded to
mapped onto a single number, the pertur- ness. (At a lower dose, ketamine is abused the TMS pulse with a complex pattern of
bational complexity index, or PCI. If the as a hallucinogenic drug, known as vita- electrical activity that lies above the
brain does not react to the magnetic re- min K.) The un­­conscious conditions for threshold. That is, the perturbational
sponse—say, because the cortical activity which EEG was measured during the complexity of their brains’ responses was
is suppressed or only wiggles minimally— study were deep sleep (reporting no expe- as high as in many conscious benchmark
PCI will be close to zero, whereas maxi- riences immediately prior to being awak- controls. These patients with high-com-
mal complexity yields a PCI of one. The ened) and surgical-level anesthesia using plexity cortical responses may experience
larger the PCI, the more diverse the three different agents (midazolam, xenon something yet are unable to communi-
brain’s response to the magnetic pulse. and propofol). The study also included 48 cate with the world and their loved ones.
brain-injured but responsive and awake As any successful experiment does, this
ZAP AND ZIP IN PATIENTS patients who were assessed while awake one is leading to additional clinical studies.
T he l ogic of the 2016 study, which in­­ as controls. How can the zap-and-zip method be im­­
volved patients from specialized clinics in The investigators found that con- proved to achieve 100 percent accuracy in
Belgium and Italy, is straightforward. In a sciousness could be inferred with com- minimally conscious patients? Could other
first step, zap and zip is applied to a con- plete accuracy in every single subject us- groups of patients, such as those with ca­­ta­
trol population to infer a cutoff value— ing the same PCI* value of 0.31. That is, in tonia or late-stage dementia, infants, or
tagged PCI*—above which consciousness every one of the 540 conditions tested young children, also be tested? Another
is thought to be present. In every case in across the 150 subjects, if the electrical re­­ question is whether other physiological or
which consciousness can be reliably estab- sponse was at or below this threshold, the behavioral measures can be developed to
lished in any one subject, the person’s PCI subject was unconscious. If above PCI*, corroborate the inference that some UWS
values should be greater than PCI*, and in the subject was conscious. Everyone in patients are conscious. Can the method be
every case in which the subject is uncon- the study, whether an uninjured volun- turned into a prognostic tool, inferring to
scious, PCI values should be below this teer or a brain-injured patient, received a what extent UWS patients are on the road
threshold. This procedure establishes PCI* correct classification. This achievement is to recovery? Those questions need to be
as a critical threshold—the minimum re­mark­able given the variability in gen- tackled moving forward. But in the inter-
measure of complex brain activity—sup- der, age, brain locations where the TMS im, let us celebrate a milestone in untan-
porting consciousness. Then, in a second pulses were applied, and medical and be- gling the ancient mind-body problem. 
step, this threshold is used to infer wheth- havioral conditions in the study cohort.
er consciousness is present in patients liv- The team then applied zap and zip with Christof Koch is chief scientist of MindScope at the Allen
ing in the gray zone, where more conven- this threshold value (of 0.31) to a distinct Institute for Brain Science and of the Tiny Blue Dot Foun­
tional measurements are insufficient. set of patients with severe disorders of dation, as well as author of The Feeling of Life Itself—Why
In the study, the benchmark popula- consciousness—those either in a minimal- Consciousness Is Widespread but Can’t Be Computed (MIT
tion used to calibrate the procedure en­­ ly conscious state or in an unresponsive Press, 2019). He is on Scientific American’s board of advisers.

SCIENTIFICAMERICAN.COM  |  19
20  |  SCIENTIFIC AMERICAN  |  SPECIAL EDITION  |  WINTER 2022
Can Lab-Grown Brains
Become Conscious?
A handful of experiments are raising questions about whether
clumps of cells and disembodied brains could be sentient and
how scientists would know if they were • By Sara Reardon
Illustration by Thomas Fuchs
SCIENTIFICAMERICAN.COM  |  21
I
n A ly s s o n M u o t r i ’ s l a b o rat o ry, h u n d r e d s o f m i n i at u r e h u ma n b ra i n s ,
the size of sesame seeds, float in petri dishes, sparking with electrical activity. These
tiny structures, known as brain organoids, are grown from hu­­man stem cells and have
become a fa­­miliar fixture in many labs that study the properties of the brain. Muotri, a
neuroscientist at the University of California, San Diego, has found some unusual ways
to de­­ploy his. He has connected organoids to walking robots, modified their genomes
with Neandertal genes, launched them into orbit onboard the International Space Sta-
tion and used them as models to develop more hu­­manlike artificial-intelligence sys-
tems. Like many scientists, Muotri temporarily pivoted to studying ­COVID, using brain
or­­gan­oids to test how drugs perform against the SARS-CoV-2 coronavirus.
But one experiment has drawn more scrutiny than the oth- to those used in animal research, to guide the humane use of
ers. In August 2019 Muotri’s group published a paper in C  ell brain organoids and other experiments that could achieve con-
Stem Cell reporting the creation of human brain organoids that sciousness. In June 2020 the U.S. National Academies of Scienc-
produced coordinated waves of activity resembling those seen es, Engineering and Medicine began a study with the aim of
in premature babies. The waves continued for months before outlining the potential legal and ethical issues associated with
the team shut the experiment down. brain organoids and human–animal chimeras.
This type of brain-wide, coordinated electrical activity is one The concerns over lab-grown brains have also highlighted a
of the properties of a conscious brain. The team’s finding led problem: neuroscientists have no agreed way to define and
ethicists and scientists to raise a host of moral and philosophi- measure consciousness. Without a working definition, ethicists
cal questions about whether organoids should be allowed to worry that it will be impossible to stop an experiment before it
reach this level of advanced development, whether “conscious” crosses a line.
organoids might be entitled to special treatment and rights not The current crop of experiments could force the issue. If sci-
afforded to other clumps of cells and the possibility that con- entists become convinced that an organoid has gained con-
sciousness could be created from scratch. sciousness, they might need to hurry up and agree on a theory of
The idea of bodiless, self-aware brains was already on the how that happened, says Anil K. Seth, a cognitive neuroscientist
minds of many neuroscientists and bioethicists. Just a few at the University of Sussex in England. But, he says, if one per-
months earlier a team at Yale University announced that it had son’s favored theory deems the organoid conscious, whereas
at least partially restored life to the brains of pigs that had been another’s does not, any confidence that consciousness has been
killed hours before. By removing the brains from the pigs’ skulls attained vanishes. “Confidence largely depends on what theory
and infusing them with a chemical cocktail, the researchers we believe in. It’s a circularity.”
revived the neurons’ cellular functions and their ability to
transmit electrical signals. SENTIENT STATES
Other experiments, such as efforts to add human neurons to C r e at i n g a c o n s c i o us s ys t e m m
 ight be a whole lot easier
mouse brains, have raised questions, with some scientists and than defining it. Researchers and clinicians define conscious-
ethicists arguing that these experiments should not be allowed. ness in many different ways for various purposes, but it is hard
The studies set the stage for a debate between those who to synthesize them into one neat operational definition that
want to avoid the creation of consciousness and those who see could be used to decide on the status of a lab-grown brain.
complex organoids as a means to study and test treatments for Physicians generally assess the level of consciousness in
human diseases. Muotri and many other neuroscientists think patients in a vegetative state on the basis of whether the person
that human brain organoids could be the key to understanding blinks or flinches in response to pain or other stimuli. Using
uniquely human conditions such as autism and schizophrenia, electroencephalogram (EEG) readings, for instance, research-
which are impossible to study in detail in mouse models. To ers can also measure how the brain responds when it is zapped
achieve this goal, Muotri says, he and others might need to with an electrical pulse. A conscious brain will display much
deliberately create consciousness. more complex, unpredictable electrical activity than one that is
Researchers have been calling for a set of guidelines, similar unconscious, which responds with simple, regular patterns.

22  |  SCIENTIFIC AMERICAN  |  SPECIAL EDITION  |  WINTER 2022

 IN DEVELOPING human brain organoids, preneuronal cells (red) turn into neurons (green), which wire up into networks (white).

But such tests might not adequately probe whether a person to be complex enough to be classed as conscious. And organoids
lacks consciousness. In brain-imaging studies of people who are cannot blink or recoil from a painful stimulus, so they would
in a coma or a vegetative state, scientists have shown that unre- not pass the clinical test for consciousness.
Muotri Lab/University of California, San Diego

sponsive individuals can display some brain activity reminis- In contrast, it is much more likely that an intact brain from a
cent of consciousness—such as activity in motor areas when recently killed pig has the necessary structures for conscious-
asked to think about walking. ness, as well as wiring created by memories and experiences the
In any case, standard medical tests for consciousness are dif- animal had while it was alive. “Thinking about a brain that has
ficult to apply to brain cells grown in dishes or to disembodied been filled with all this, it is hard to imagine that brain would be
animal brains. When Muotri suggested that his organoids’ fir- empty,” says Jeantine Lunshof, a philosopher and neuroethicist
ing patterns were just as complex as those seen in preterm at Harvard University. “What they can do in terms of thinking, I
infants, people were unsure what to make of that. Some re­­ don’t know, but it’s for sure not zero,” Lunshof says. Bringing a
search­ers do not consider the brain activity in a preterm infant dead brain back to a semblance of life, as the Yale team did,

SCIENTIFICAMERICAN.COM  |  23
might have the potential to restore a degree of consciousness, necessary to create complex EEG patterns. Still, Lancaster
although the scientists took pains to avoid this by using chemi- admits that for advanced organoids, it depends on the defini-
cal blocking agents that prevented brain-wide activity. tion. “If you thought a fly was conscious, it’s conceivable that an
Researchers agree that they need to take the possibilities organoid could be,” she says.
raised by these studies seriously. In October 2019 U.C.S.D. held Lancaster and most other researchers think that something
a conference of about a dozen neuroscientists and philoso- like a revitalized pig brain would be much more likely to achieve
phers, together with students and members of the public, with consciousness than an organoid. The team that did the work on
the intention of establishing and publishing an ethical frame- the pig brains, led by neuroscientist Nenad Sestan, was trying to
work for future experiments. But the paper was delayed for find new ways to revitalize organs, not to create consciousness.
months, partly because several of the authors could not agree The researchers were able to get individual neurons or groups of
on the basic requirements for consciousness. them to fire and were careful to try to avoid the creation of wide-
spread brain waves. Still, when Sestan’s team members saw what
INCREASINGLY COMPLEX looked like coordinated EEG activity in one of the brains, they
so far nobody h  as created consciousness in the lab, say sci- immediately halted the project. Even after a neurology specialist
entists and ethicists who study the issue. But they are asking confirmed that the pattern was not consistent with consciousness,
themselves what to watch out for and which theories of con- the group anesthetized the brains as a precautionary measure.
sciousness might be most relevant. According to an idea called Sestan also contacted the U.S. National Institutes of Health for
integrated information theory, for example, consciousness is guidance on how to proceed. The agency’s neuroethics panel,
a product of how densely neuronal networks are connected including Lunshof and Insoo Hyun, a bioethicist at Case Western
across the brain. The more neurons that interact with one University, assessed the work and agreed that Sestan should con-
another, the higher the degree of consciousness—a quantity tinue to anesthetize the brains. But the panel has not settled on
known as phi. If phi is greater than zero, the organism is con- more general regulations and does not routinely require a bioeth-
sidered conscious. ics assessment for organoid proposals, because its members think
Most animals reach this bar, according to the theory. Christof that consciousness is unlikely to arise. The nih has not arrived at
Koch, who serves on Scientific American’s board of advisers and a definition of consciousness, either. “It’s so flexible, everyone
is chief scientist of the MindScope Program at the Allen Institute claims their own meaning,” Hyun says. “If it’s not clear we’re talk-
for Brain Science, doubts any existing organoid could achieve ing about the same thing, it’s a big problem for discourse.”
this threshold but concedes that a more advanced one might.
Other competing theories of consciousness require sensory FUZZY DEFINITIONS
input or coordinated electrical patterns across multiple brain S o m e t h i n k i t i s f u t i l e t o even try to identify conscious-
regions. An idea known as global workspace theory, for in­­ ness in any sort of lab-maintained brain. “It’s just impossible to
stance, posits that the brain’s prefrontal cortex functions as a say meaningful things about what these bunches of brain cells
computer, processing sensory inputs and interpreting them to could think or perceive, given we don’t understand conscious-
form a sense of being. Because organoids do not have a pre- ness,” says Steven Laureys, a neurologist at the University of
frontal cortex and cannot receive input, they cannot become Liège in Belgium, who pioneered some of the imaging-based
conscious. “Without input and output, the neurons may be talk- measures of consciousness in people in a vegetative state. “We
ing with each other, but that doesn’t necessarily mean anything shouldn’t be too arrogant.” Further research should proceed
like human thought,” says Madeline Lancaster, a developmen- very carefully, he says.
tal bi­­ol­o­gist at the University of Cambridge. Laureys and others point out that the experience of an organ-
Connecting organoids to organs, however, could be a fairly oid is likely to be very different from that of a preterm infant, an
simple task. In 2019 Lancaster’s team grew human brain organ- adult human or a pig and would not be directly comparable.
oids next to a mouse spinal column and back muscle. When Furthermore, the structures in an organoid might be too small
nerves from the human organoid connected with the spinal col- to have their activity measured accurately, and similarities
umn, the muscles began to spontaneously contract. between the EEG patterns of organoids and of preterm baby
Most organoids are built to reproduce only one portion of brains could be coincidental. Other scientists who work on brain
the brain—the cortex. But if they develop long enough and with organoids also caution against making assumptions about the
the right kinds of growth factor, human stem cells spontane- link between activity patterns in organoids and consciousness.
ously re-create many different parts of the brain, which then “This system is not the human brain,” says Sergiu Pasca, a
begin coordinating their electrical activity. In a study published neuroscientist at Stanford University. “They’re made out of neu-
in 2017, molecular biologist Paola Arlotta of Harvard coaxed rons. Neurons have electrical activity, but we have to think care-
stem cells to develop into brain organoids composed of many fully about how to compare them.”
different cell types, including light-sensitive cells like those Muotri wants his organoid systems to be comparable, in at
found in the retina. When exposed to light, neurons in the least some ways, with human brains so that he can study
organoids began firing. But the fact that these cells were active human disorders and find treatments. His motivation is per-
does not mean the organoids could see and process visual infor- sonal: his teenage son has epilepsy and autism. “He struggles
mation, Arlotta says. It simply means that they could form the hard in life,” Muotri says. Brain organoids are a promising ave-
necessary circuits. nue because they recapitulate the earliest stages of brain wir-
Arlotta and Lancaster think their organoids are too primi- ing, which are impossible to study as a human embryo devel-
tive to be conscious because they lack the anatomical structures ops. But studying human brain disorders without a fully func-

24  |  SCIENTIFIC AMERICAN  |  SPECIAL EDITION  |  WINTER 2022

 NEUROSCIENTIST N
 enad Sestan used the BrainEx platform to restore neural activity in disembodied pig brains.

tioning brain, he says, is like studying a pancreas that does not Still, Muotri and others say they would welcome some strict-
produce insulin. “To get there, I need a brain organoid model er guidelines. These could include requiring scientists to justify
that really resembles a human brain. I might need an organoid the number of human brain organoids they use, to use them
that becomes conscious.” only for research that cannot be done in any other way, to
Muotri says he is agnostic about which definition to use to restrict the amount of pain that can be inflicted on them, and to
de­­cide whether an organoid reaches consciousness. At some dispose of them humanely.
point, he says, organoids might even be able to help researchers Having such advice in place ahead of time would help
answer questions about how brains produce conscious states. researchers weigh the costs and benefits of creating conscious
For instance, mathematician Gabriel Silva of U.C.S.D. is studying entities. And many researchers emphasize that such experi-
neural activity in Muotri’s organoids to develop an algorithm ments have the potential to yield important insights. “There are
that describes how the brain generates consciousness. The goal truly conscious people out there with neurological disorders
of his project, which is partially funded by Microsoft, is to create with no treatments,” Lancaster says. “If we did stop all of this
an artificial system that works like human consciousness. research because of the philosophical thought experiment,” she
At the moment, there are no regulations in the U.S. or in adds, “that would be very detrimental to actual human beings
Europe that would stop a researcher from creating conscious- who do need some new treatment.”
ness. The National Academies panel released a report in April Treatments could still, however, be tested in brain organoids
2021 outlining the latest research and what it views as appro- made using mouse stem cells or in regular animal models. Such
priate oversight. Members weighed in on questions such as experiments could also inform discussions about the ethical use
whether to obtain people’s consent to develop their cells into of human organoids. For instance, Hyun would like to see
brain organoids and how to study and dispose of organoids researchers compare the EEG patterns of mouse brain organ-
humanely. The International Society for Stem Cell Research has oids with those of living mice, which might indicate how well
also released organoid guidelines but is not addressing con- human organoids recapitulate the human brain.
sciousness, because it does not think the science is there yet. For his part, Muotri sees little difference between working
Hyun says that the nih neuroethics panel has not yet seen on a human organoid and using a lab mouse. “We work with
any proposals to create complex, conscious organoids that animal models that are conscious, and there are no problems,”
would necessitate new guidelines. And Muotri says he does not he says. “We need to move forward, and if it turns out they
know of anyone else deliberately trying to create conscious become conscious, to be honest, I don’t see it as a big deal.” 
Jesse Winter

organoids either, although a sufficiently complex organoid could,

by some definitions, reach that status accidentally. Sara Reardon i s a freelance reporter based in Bozeman, Mont.

SCIENTIFICAMERICAN.COM  |  25
 The
Brain
Electric
Electrodes that stimulate brain tissue reveal
the topography of conscious experience
By Christof Koch
Illustration by Zara Picken

Consider the following experiences:

●   ou’re headed toward a storm that’s a couple of miles away,
Y each one was evoked by directly stimulating the brain with an elec-
and you’ve got to get across a hill. You ask yourself: “How am trode. As American poet Walt Whitman intuited in his poem “I Sing
I going to get over that, through that?” the Body Electric,” these anecdotes illustrate the intimate relation-
●  You see little white dots on a black background, as if looking ship between the body and its animating soul. The brain and the
up at the stars at night. conscious mind are as inexorably linked as the two sides of a coin.
●  You look down at yourself lying in bed from above but see Recent clinical studies have uncovered some of the laws and
only your legs and lower trunk. regularities of conscious activity, findings that have occasionally
proved to be paradoxical. They show that brain areas involved in
These may seem like idiosyncratic events drawn from the vast conscious perception have little to do with thinking, planning
universe of perceptions, sensations, memories, thoughts and and other higher cognitive functions. Neuroengineers are now
dreams that make up our daily stream of consciousness. In fact, working to turn these insights into technologies to replace lost

SCIENTIFICAMERICAN.COM  |  27
 “My left foot “I see a star
Patient laughs shifted to the right in the top right
and says they and the sensation center. It was a blue
“felt well went all the way and silver star.”
internally.” up my calf.” “I feel
tingling in
“That my leg.”
felt good; quite “I just had the
erotic. I can’t even, urge to squeeze
um, I felt good. my fingers.
I can’t explain it. They just closed
Um, yeah, you’re by themselves.”
embarrassing me!”

“Pulsating feeling,
Patient mostly in the left
describes a negative chest but also in
emotional feeling, the left arm.”
seemingly localized
in the chest.

Patient
describes
a feeling
of fear.

Patient “Aroused,
says she calm ...
felt dizzy. sexually excited.”

Speech
arrest.

“Feels like
Patient cannot I’m going in a
Source: “Intrinsic Network Architecture Predicts the Effects Elicited by Intracranial Electrical Stimulation

repeat circle. Everything is

“If she comes, still and I’m moving.
I will go.” Makes me
feel sick.”
of the Human Brain,” by Kieran C. R. Fox et al., in N ature Human Behaviour, Vol. 4; July 2020

Patient
describes a
feeling of nervous
anticipation.

“It smells
funny.
Negative, like
nail polish.”

“You just
“For being in turned into
this temperature, somebody else. Your
just a little more face metamorphosed.
sensitive. Almost as if Your nose got saggy;
I was in a colder went to the left.
temperature.” Not pretty.”

28  |  SCIENTIFIC AMERICAN  |  SPECIAL EDITION  |  WINTER 2022

 cognitive function and, in the more distant future, to
enhance sensory, cognitive or memory capacities. For
Where Experiences example, a recent brain-machine interface provides
completely blind people with limited abilities to per-
Live in the Brain ceive light. These tools, however, also reveal the difficul-
ties of fully restoring sight or hearing. They underline
An atlas published in 2020 compiled the verbal reports of people with
even more the snags that stand in the way of sci-fi-like
epilepsy whose cortical areas were stimulated with electrodes during
enhancements that would enable access to the brain as
surgery. What they felt and perceived varied depending on which
if it were a computer storage drive.
brain region was stimulated. All of the 1,537 locations in these 67
patients where current was applied were mapped onto a digital brain
ANIMAL ELECTRICITY
model, a simplified version of which is depicted here. When stimulated
Nervous systems operate on the flow of electric cur-
at these sites, patients talked about their experiences.
rents through ultradense and hyperconnected networks
of switching elements. Countless physicians and scien-
tists have worked on this problem over the past two and
a half centuries, beginning with Italian physician Luigi
Colors of Cognition Galvani, who in the late 18th century connected a freshly
Stimulation was applied to varied killed frog to a long metal wire. By pointing the wire
“Just really couldn’t brain networks and regions. toward the sky during a thunderstorm, he made the frog’s
move (my fingers) too leg jump and twitch with each flash of lightning. Gal-
much; lost the motion. The Somatomotor vani’s investigations re­­veal­ed that nerve fibers transmit-
hand felt a little tight, but ted “animal electricity,” which is no different in kind from
the thumb was out the “atmospheric electricity” that Benjamin Franklin
of commission.” Visual
demonstrated with his kite experiments in Philadelphia
in 1752. In 1802 Galvani’s nephew Giovanni Aldini elec-
Dorsal attention trically stimulated the exposed brain of a decapitated
prisoner during a public event. A jaw quivered. An eye
Salience opened. The spectacle may have helped to inspire Mary
“I felt like my
Shelley to write the classic gothic novel Frankenstein.
arms were moving
but they weren’t. I felt Frontoparietal Subsequent animal studies demonstrated that excit-
side-to-side movements, ing particular brain regions triggered movements in spe-
like floating cific muscles and limbs. These investigations led to the
Limbic
in the air.” discovery of the motor cortex in the 1870s. In 1874 Amer-
ican physician Robert Bartholow performed the first
Default
direct brain stimulation of a conscious patient—a pio-
neering act clouded in ethical controversy because it
caused the patient pain and probably hastened her death.
Points of Stimulation
Intracranial electrical stimulation (iES) was re­­fined over
Researchers placed dozens of
electrodes onto the surface the following decades. It became part of the neurosur-
of the cerebral cortex accessed geon’s toolbox thanks to the ground-breaking work of
through an opening Wilder Penfield of the Montreal Neurological Institute,
in the skull. who between the 1930s and the 1950s used iES to map
cortical areas that process motor or sensory functions.
In some people with epilepsy, drugs fail to adequately
control the number or severity of seizures. Neurosurgery
becomes an option if those seizures originate in a delim-
ited neighborhood in the cortex—the outermost layer of
the brain involved in perception, motor control, speech,
reasoning, and so on—or in closely related structures,
such as the hippo­­campus. Uncontrolled hy­­per­ex­cit­a­bil­
ity starts because of local faulty wiring. It can grow and
eventually engulf the rest of the brain. How much tissue
to remove is a dilemma: cut too little, and seizures may
continue; cut too much, and the patient may lose the abil-
ity to speak, see or walk. Surgeons must avoid areas of
the cortex that are crucial for everyday behavior, such as
the primary auditory, visual, somatosensory and motor
cortices and the regions controlling understanding and
producing speech, areas known as the eloquent cortex.

Illustration by AXS Biomedical Animation Studio SCIENTIFICAMERICAN.COM  |  29

 iES is brought in as a means to look for tissue that to evoke these feelings. During sham stimulation (no
needs preserving. Neurosurgeons implant disk-shaped current applied), patients did not feel anything.
electrodes inside the skull, underneath the tough, leath- Although iES is safe and effective, it is also crude. The
erlike membrane known as the dura mater. Alterna- low-impedance electrodes are six to 10 square millime-
tively, they may insert needlelike electrodes into the ters in area and deliver up to 10 milliamperes of electric
brain’s gray matter to probe its function. Once the sur- current between adjacent electrodes—enough to modu-
geons have identified the focal point of the seizure and late the excitability of a million or more nerve cells. Still,
removed the electrodes, they cut or coagulate this tis- effects induced by iES can be quite localized. Responsive-
sue in a follow-up operation, and the patient usually ness can change from all to none within millimeters or
becomes seizure-free. across a sulcus (a groove on the cortical surface).
A different use for iES is chronic electrical stimula- The Parvizi team found that electrodes in the dedi-
tion, in which the electrodes are left permanently in cated sensory and motor areas were far more likely to be
place. Gentle pulses of current sent through the elec- responsive than those in areas of the cortex that process
trodes can control the tremors and rigidity of Parkinson’s higher cognitive functions. Half to two thirds of elec-
disease (a technique called deep-brain stimulation) or trodes above visual and tactile (somatosensory) cortex
areas triggered some conscious perception; in regions of
the lateral and anteromedial prefrontal cortex, which are
Patients reported electrode- involved with higher thought processes, at most one in

evoked experiences such as seeing

five electrodes did so. Put differently, electrodes in the
back of the cortex—in areas responsible for sensory expe-

distorted faces reminiscent riences—were more likely to be active than those toward
the front, which consists of regions of the cortex impor-
of paintings by Salvador Dalí. tant for cognitive activity such as thinking, planning,
moral reasoning, decision-making and intelligence.
Despite their importance for thinking, these re­­gions
reduce the incidence and severity of seizures. Pilot clin- have little to do with consciousness. Indeed, for the past
ical experiments are evaluating the use of such implanted century neurosurgeons have observed that so long as the
electrodes as a visual prosthetic device to enable people eloquent cortex is spared, massive regions of the pre-
with vision impairments to navigate and as a therapy for frontal cortex can be ablated without causing obvious
obsessive-compulsive disorder and depression. deficits in the daily stream of consciousness of these
patients. These regions of noneloquent cortex can mod-
HOT OR NOT ulate consciousness, but they are, by and large, not where
In July 2020 Nature Human Behaviour published an conscious experience appears to originate. That privilege
atlas highlighting locations across the cortex that, when belongs to more posterior regions—the parietal, tempo-
aroused with electrodes, evoked conscious experiences, ral and occipital lobes. Why the physical substrate of our
such as the storm and the disconnected body mentioned mental experiences should be in the back rather than in
earlier. Led by Josef Parvizi, a professor of neurology at the front of the brain remains a mystery.
the Stanford University School of Medicine, the clinical
team collected data from 67 people with epilepsy. The TO SEE OR NOT TO SEE
researchers recorded electrical activity from more than Applying iES to the visual cortex triggers optical sen-
1,500 sites in the cortex, primarily with subdural elec- sations known as phosphenes, brief flashes that resem-
trodes. They mapped the recordings from those sites to ble lightning striking a darkened plain. This observa-
spots on a digital brain model so they could compare data tion is the source of a long-standing dream of a pros-
from different brains (the pattern of ridges and valleys thetic device that restores some vision to people who
that give the organ the look of an oversized walnut dif- are blind. Millions worldwide live with deficits in both
fers from person to person). The team looked for “respon- eyes from retinitis pigmentosa, age-related macular
sive” electrodes that triggered some visual or tactile sen- degeneration, glaucoma, infection, cancer or trauma.
sation, muscle twitching or disrupted speech. If the Doctors, scientists and engineers started pursuing
patient did not feel anything when stimulated, that elec- visual prosthetics in the 1960s but have only recently
trode was marked as nonresponsive. been able to harness the appropriate technology to help
Patients reported a range of electrode-evoked subjec- blind people. One prominent example is a device known
tive experiences: briefly flashing points akin to stars of as Orion, developed by Second Sight Medical Products
light; distorted faces like those in the paintings of Sal- in Los Angeles. A tiny camera, mounted on glasses, con-
vador Dalí; bodily feelings such as tingling, tickling, verts images into pulses and transmits them wirelessly
burning, pulsing and so-called out-of-body experiences; to fire 60 electrodes sitting on the visual cortex. The
fear, unease, sexual arousal, merriment; the desire to handful of people who have had this experimental device
move a limb; the will to persevere in the face of some implanted into their brain perceive clouds of dots that
great but unrecognized challenge. Mere tickling of neu- allow them to navigate. “It’s still a blast every time I turn
ral tissue with a tiny bit of electric current was enough it on,” one study participant reports. “After seeing noth-

30  |  SCIENTIFIC AMERICAN  |  SPECIAL EDITION  |  WINTER 2022

ing to all of a sudden seeing little flickers of light move the fact that all of this requires trepanation of the skull.
around and figuring out that they mean something. It’s In general, turning scientific insights into actual thera-
just amazing to have some form of functional vision peutics is done in decades rather than in years. I am
again.” Orion significantly improves the quality of life for pretty confident that such enhancements will not occur
people who previously lived in complete darkness. It in my lifetime (I’m now 65).
enables them to safely cross the street or locate a door- The “easiest” hurdles to overcome on the way to such
way. But it does not allow them to regain the ability to a utopian (or perhaps dystopian) future are technologi-
recognize figures, shapes or letters. cal ones—reliably, quickly, and delicately reading and
A team at the University of California, Los Angeles, writing the brain electric. Neuralink’s device represents
and the Baylor College of Medicine led by neurosurgeon the best of currently available technology and will cer-
Daniel Yoshor recently did accomplish this feat, as tainly improve in future iterations. But we still have a
described in the journal Cell. They stimulated nearby long way to go before we can identify which of the 50,000
locations in the visual cortex to trigger phosphenes that or more neurons in any quinoa-sized bit of brain matter
appear close together, demonstrating that the external are involved in any given perception or action. Only when
visual environment is mapped in a regular fashion onto that happens will it be possible to limit electrical stimu-
the surface of the visual cortex. This observation has led lation to just those neurons and avoid stimulating nearby
to the erroneous belief that individual phosphenes are cells or output cables. That Parvizi and his colleagues
like pixels on a computer display—that is, if you were to failed to elicit conscious perceptions in more than half
simultaneously stimulate a series of points on the corti- of all stimulated sites shows we lack tools capable of reli-
cal surface in the shape of a cross, the subject should see ably eliciting any arbitrary sensation through electrical
points forming a cross. This does not happen, however. stimulation, let alone being able to evoke any highly
Stimulating more than one location yields unpredict- specific one.
able results. In one participant, simultaneous stimula- Even more challenging are surgical and regulatory
tion of five electrodes, each one associated with one dis- hurdles that demand that prosthetic devices can be rou-
crete phosphene, triggered the illumination of two large tinely and safely implanted by drilling through the hard
phosphenes that did not coalesce into a letter or any skull into the gray matter underneath while minimizing
other coherent form. If the researcher staggered acti- the risk of infections, bleeding and seizures. Furthermore,
vation of the electrodes in time, however, the subject the electronics has to function for years inside warm, wet
could identify shapes. The staggering reflected the delay and salty biological tissue—hardly an optimal operating
required to trace the shape of a letter, as if the researcher regime. You don’t want your prosthetic device to corrode
were outlining a letter into the hand of the subject or or freeze up in the equivalent of the blue screen of death.
onto a piece of paper. In this more dynamic manner, the For this reason, neural implants will remain a matter of
subject with the implant whose vision was blocked last resort for those with severe sensory or motor impair-
could identify a stimulus by tracing out a Z, N, V and W, ments. As neuroprosthetic devices move through clini-
rapidly distinguishing upward from downward motion cal trials, they will help people with visual impairments
or discriminating sequences of letters. see and paralyzed patients with spinal cord damage to
Seeing the shape of a single letter is not quite the text or to steer a wheelchair with their thoughts, like the
same as seeing a glorious sunset over Homer’s wine-dark mind-Pong-playing monkey. For everyone else, the ben-
sea, but it represents progress. Why staggering stimula- efits of highly invasive brain surgery are unlikely to out-
tion in time improves perception is not clear and reveals weigh the costs.
our ignorance concerning functioning cortical circuits. But the true Annapurna ahead involves understand-
ing how three pounds of excitable brain matter is respon-
WHAT LIES AHEAD sible for seeing, moving and suffering. Yes, the physical
Technological progress in so-called brain-ma­­chine substrate of heaven and hell is rooted in bioelectric sig-
interfaces is proceeding at a rapid pace. Elon Musk’s com- nals that obey natural laws. But that tells us precious lit-
pany Neuralink released in April 2021 an impressive tle about how a trillion electrical spiking signals each sec-
video showcasing a monkey playing the computer game ond, streaming over networks of tens of billions of het-
Pong without any controller. This was achieved with two erogeneous cells, constitute a sight, sound or emotion.
small chips implanted into the left and right motor cor- Intracranial brain stimulation highlights the daily
tices of the animal. Each chip has 1,024 hairlike elec- miracle of the brain’s water changing into the wine of
trodes that record the chattering of individual neurons. consciousness. The question remains, though: What is
Collectively they convey the monkey’s intention to it about the brain, the most complex piece of active mat-
quickly move the paddle up or down the screen to return ter in the known universe, that turns the activity of
the ball to the opposite side. Everything was done wire- 86 billion neurons into the feeling of life itself? 
lessly; no electronics or dangling wires were protruding
from the monkey’s head. Many assume that surgeons will Christof Koch is chief scientist of MindScope at the Allen Institute for
soon routinely replace or bypass faulty biological com- Brain Science and of the Tiny Blue Dot Foundation, as well as author of
ponents—defective eyes or ears, failing memories—with The Feeling of Life Itself—Why Consciousness Is Widespread but Can’t Be
superior electronic substitutes. Such optimism neglects Computed (MIT Press, 2019). He is on Scientific American’s board of advisers.

SCIENTIFICAMERICAN.COM  |  31
OPINION

How Do I Know
I’m Not the
Only Conscious
Being in the
Universe?
The solipsism problem, also called the
problem of other minds, lurks at the heart
of science, philosophy, religion, the arts
and the human condition
By John Horgan
SCIENTIFICAMERICAN.COM  |  33
 I t is a central dilemma of human life—more urgent, arguably,
than the inevitability of suffering and death. I have been brooding
and ranting to my students about it for years. It surely troubles us
more than ever during this plague-ridden era. Philosophers call it
the problem of other minds. I prefer to call it the solipsism problem.
Solipsism, technically, is an extreme form of skepticism, at once
utterly illogical and irrefutable. It holds that you are the only con-
scious being in existence. The cosmos sprang into existence when
you became sentient, and it will vanish when you die. As crazy as this
proposition seems, it rests on a brute fact: each of us is sealed in an
impermeable prison cell of subjective awareness. Even our most inti-
mate exchanges might as well be carried out via Zoom.
You experience your own mind every waking sec-
ond, but you can only infer the existence of other
minds through indirect means. Other people seem to
possess conscious perceptions, emotions, memories,
intentions, just as you do, but you cannot be sure
they do. You can guess how the world looks to me
based on my behavior and utterances, including
these words you are reading, but you have no first-
hand access to my inner life. For all you know, I
might be a mindless bot.
Natural selection instilled in us the capacity for a
so-called theory of mind—a talent for intuiting oth-
ers’ emotions and intentions. But we have a counter-
tendency to deceive one another and to fear we are
that you are conscious, let alone a jellyfish, bot or
doorknob. As long as we lack what neuroscientist
Christof Koch has called a consciousness meter—a
device that can measure consciousness in the same
way that a thermometer measures temperature—the-
ories of consciousness will remain in the realm of
pure speculation.
But the solipsism problem is far more than a tech-
nical philosophical matter. It is a paranoid but under-
standable response to the feelings of solitude that lurk
within us all. Even if you reject solipsism as an intellec-
tual position, you sense it, emotionally, whenever you
feel estranged from others, whenever you confront the
awful truth that you can never know—really know—
being deceived. The ultimate deception would be another person, and no one can really know you.
pretending you are conscious when you are not. Religion is one response to the solipsism problem.
The solipsism problem thwarts efforts to explain Our ancestors dreamed up a supernatural entity who
consciousness. Scientists and philosophers have pro- bears witness to our innermost fears and desires. No
posed countless contradictory hypotheses about matter how lonesome we feel, how alienated from
what consciousness is and how it arises. Panpsychists our fellow humans, God is always there watching
Getty Images ( p receding pages)

contend that all creatures and even inanimate mat- over us. He sees our souls, our most secret selves, and
ter—even a single proton!—possess consciousness. He loves us anyway. Wouldn’t it be nice to think so?
Hard-core materialists insist, conversely (and per- The arts, too, can be seen as attempts to overcome
versely), that not even humans are all that conscious. the solipsism problem. The artist, musician, poet, nov-
The solipsism problem prevents us from verifying elist says, “This is how my life feels” or “This is how life
or falsifying these and other claims. I cannot be certain might feel for another person.” They help us imagine

34  |  SCIENTIFIC AMERICAN  |  SPECIAL EDITION  |  WINTER 2022

what it is like to be a Black woman trying to save her sions actually see reality clearly? According to the
children from slavery or a Jewish ad salesman wan- Buddhist doctrine of anatta, the self does not really
dering through Dublin, wondering whether his wife is exist. When you try to pin down your own essence, to
cheating on him. But to imagine is not to know. grasp it, it slips through your fingers.
Some of my favorite works of art dwell on the solip- We have devised methods for cultivating self-
sism problem. In I’m Thinking of Ending Things and knowledge and quelling our anxieties, such as medi-
earlier films, as well as his novel A  ntkind, Charlie tation and psychotherapy. But these practices strike
Kaufman depicts other people as projections of a dis- me as forms of self-brainwashing. When we meditate
turbed protagonist. Kaufman no doubt hopes to help or see a therapist, we are not solving the solipsism
us, and himself, overcome the solipsism problem by problem. We are merely training ourselves to ignore
venting his anxiety about it, but I find his dramatiza- it, to suppress the horror and despair that it triggers.
tions almost too evocative.
Love, ideally, gives us the illusion of
transcending the solipsism problem. You
feel you really know someone, from the
NATURAL SELECTION INSTILLED IN US THE
inside out, and they know you. In moments CAPACITY FOR A SO-CALLED THEORY OF
of ecstatic sexual communion or mundane
togetherness—while you’re eating pizza
MIND—A TALENT FOR INTUITING OTHERS’
and watching T  he Alienist, s ay—you fuse EMOTIONS AND INTENTIONS. BUT WE HAVE
with your beloved. The barrier between
you seems to vanish. A COUNTERTENDENCY TO DECEIVE ONE
Inevitably, however, your lover disap-
points, deceives, betrays you. Or, less dra-
ANOTHER AND TO FEAR WE ARE BEING DECEIVED.
matically, some subtle biocognitive shift
occurs. You look at her as she nibbles her pizza and We have also invented mythical places in which
think, Who, what, is this odd creature? The solipsism the solipsism problem vanishes. We transcend our
problem has reemerged, more painful and suffocating solitude and merge with others into a unified whole.
than ever. We call these places heaven, nirvana, the Singularity.
It gets worse. In addition to the problem of other But solipsism is a cave from which we cannot
minds, there is the problem of our own. As evolution- escape—except, perhaps, by pretending it does not
ary psychologist Robert Trivers points out, we deceive exist. Or, paradoxically, by confronting it, the way
ourselves at least as effectively as we deceive others. A Kaufman does. Knowing we are in the cave may be as
corollary of this dark truth is that we know ourselves close as we can get to escaping it.
even less than we know others. Conceivably, technology could deliver us from the
If a lion could talk, philosopher Ludwig Wittgen- solipsism problem. Koch proposes that we all get brain
stein said, we couldn’t understand it. The same is true, implants with Wi-Fi so we can meld minds through a
I suspect, of our own deepest selves. If you could eaves- kind of high-tech telepathy. Philosopher Colin McGinn
drop on your subconscious, you would hear nothing suggests a technique that involves “brain splicing,”
but grunts, growls and moans—or perhaps the high- transferring bits of your brain into mine, and vice versa.
pitched squeaks of raw machine-code data coursing But do we really want to escape the prison of our
through a channel. subjective selves? The archnemesis of S  tar Trek: The
For the mentally ill, solipsism can become terrify- Next Generation i s the Borg, a legion of tech-enhanced
ingly vivid. Victims of Capgras syndrome think that humanoids who have fused into one big meta-entity.
identical imposters have replaced their loved ones. If Borg members have lost their separation from one
you have Cotard’s delusion, also known as walking another and hence their individuality. When they
corpse syndrome, you become convinced that you are meet ordinary humans, they mutter in a scary mono-
dead. A much more common disorder is derealization, tone, “You will be assimilated. Resistance is futile.”
which makes everything—you, others, reality as a As hard as solitude can be for me to bear, I do not
whole—feel strange, phony, simulated. want to be assimilated. If solipsism haunts me, so does
Derealization plagued me throughout my youth. oneness, a unification so complete that it extinguishes
One episode was self-induced. Hanging out with my puny mortal self. Perhaps the best way to cope
friends in high school, I thought it would be fun to with the solipsism problem in this weird, lonely time
hyperventilate, hold my breath and let someone is to imagine a world in which it has vanished. 
squeeze my chest until I passed out. When I woke up, I
didn’t recognize my buddies. They were demons jeer- John Horgan d irects the Center for Science Writings at the Stevens
ing at me. For weeks after that horrifying sensation Institute of Technology. His books include The End of Science,
faded, everything still felt unreal, as if I were in a The End of War and Mind-Body Problems, a vailable for free at
dreadful movie. mindbodyproblems.com. For many years he wrote the popular blog
What if those afflicted with these alleged delu- Cross Check for Scientific American.

SCIENTIFICAMERICAN.COM  |  35
 How We
Perceive
the World
Our Inner Universes page 38

Why We Have Free Will page 4 4

The Brain Learns

in Unexpected Ways page 48

The Brain’s Social Road Maps page 5 4

Face Values page 60

Radical Change page 68

Illustration by Maria Corte

SCIENTIFICAMERICAN.COM  |  37
 OUR
INNER
UNIVERSES Reality is constructed by the brain,
and no two brains are exactly alike
By Anil K. Seth
Illustration by Brook VanDevelder

“We do not see things as they are, we see them as we are.”

—from Seduction of the Minotaur, by Anaïs Nin (1961)

38  |  SCIENTIFIC AMERICAN  |  SPECIAL EDITION  |  WINTER 2022

SCIENTIFICAMERICAN.COM  |  39
O
n the 10th of April 2019 Pope Francis, President Salva Kiir of South Sudan and
former rebel leader Riek Machar sat down together for dinner at the Vatican. They
ate in silence, the start of a two-day retreat aimed at reconciliation from a civil war
that had killed some 400,000 people since 2013. At about the same time in my
laboratory at the University of Sussex in England, Ph.D. student Alberto Mariola
was starting to work on an experiment in which volunteers experience being in a
room they believe is there but is not. In psychiatry clinics across the globe, people
arrive complaining that things no longer seem “real” to them, whether it is the world around
them or their own selves. In the fractured societies in which we live, what is real—and what is
not—seems to be increasingly up for grabs. Warring sides may experience and believe in different
realities. Perhaps eating together in silence can help because it offers a small slice of reality that
can be agreed on, a stable platform on which to build further understanding.
We need not look to war and psychosis to find radically differ- by the brain, for the brain. And if my brain is different from your
ent inner universes. In 2015 a badly exposed photograph of a dress brain, my reality may be different from yours, too.
tore across the Internet, dividing the world into those who saw
it as blue and black (me included) and those who saw it as white THE PREDICTIVE BRAIN
and gold (half my lab). Those who saw it one way were so con- In Plato ’ s Allegory of the Cave, prisoners are chained to a blank
vinced they were right—that the dress truly was blue and black wall all their lives, so that they see only the play of shadows cast
or white and gold—that they found it almost impossible to be- by objects passing by a fire behind them, and they give the shad-
lieve that others might perceive it differently. ows names because for them the shadows are what is real. A thou-
We all know that our perceptual systems are easy to fool. The sand years later, but still a thousand years ago, Arabian scholar
popularity of visual illusions is testament to this phenomenon. Ibn al-Haytham wrote that perception, in the here and now, de-
Things seem to be one way, and they are revealed to be another: pends on processes of “judgment and inference” rather than in-
two lines appear to be different lengths, but when measured they volving direct access to an objective reality. Hundreds of years lat-
are exactly the same; we see movement in an image we know to er again Immanuel Kant realized that the chaos of unrestricted
be still. The story usually told about illusions is that they exploit sensory data would always remain meaningless without being giv-
quirks in the circuitry of perception, so that what we perceive de- en structure by preexisting conceptions or “beliefs,” which for him
viates from what is there. Implicit in this story, however, is the included a priori frame­works such as space and time. Kant’s term
assumption that a properly functioning perceptual system will “nou­men­on” refers to a “thing in itself”—Ding an sich—an objec-
render to our consciousness things precisely as they are. tive reality that will always be inaccessible to human perception.
The deeper truth is that perception is never a di­­rect window Today these ideas have gained a new momentum through an
onto an objective reality. All our perceptions are active construc- influential collection of theories that turn on the idea that the
tions, brain-based best guesses at the nature of a world that is brain is a kind of prediction machine and that perception of the
forever obscured behind a sensory veil. Visual illusions are frac- world—and of the self within it—is a process of brain-based pre-
tures in the Matrix, fleeting glimpses into this deeper truth. diction about the causes of sensory signals.
Take, for example, the experience of color—say, the bright red These new theories are usually traced to German physicist
of the coffee mug on my desk. The mug really does seem to be and physiologist Hermann von Helmholtz, who in the late 19th
red: its redness seems as real as its roundness and its solidity. century proposed that perception is a process of unconscious in-
These features of my experience seem to be truly existent prop- ference. Toward the end of the 20th century Helmholtz’s notion
erties of the world, detected by our senses and revealed to our was taken up by cognitive scientists and artificial-intelligence re­­
mind through the complex mechanisms of perception. search­ers, who reformulated it in terms of what is now general-
Yet we have known since Isaac Newton that colors do not exist ly known as predictive coding or predictive processing.
out there in the world. Instead they are cooked up by the brain from The central idea of predictive perception is that the brain is
mixtures of different wavelengths of colorless electromagnetic ra- attempting to figure out what is out there in the world (or in here,
diation. Colors are a clever trick that evolution has hit on to help in the body) by continually making and updating best guesses
the brain keep track of surfaces under changing lighting conditions. about the causes of its sensory inputs. It forms these best guess-
And we humans can sense only a tiny slice of the full electromag- es by combining prior expectations or “beliefs” about the world,
netic spectrum, nestled between the lows of infrared and the highs together with incoming sensory data, in a way that takes into ac-
of ultraviolet. Every color we perceive, every part of the totality of count how reliable the sensory signals are. Scientists usually con-
each of our visual worlds, comes from this thin slice of reality. ceive of this process as a form of Bayesian inference, a framework
Just knowing this is enough to tell us that perceptual experi- that specifies how to update beliefs or best guesses with new data
ence cannot be a comprehensive representation of an external when both are laden with uncertainty.
objective world. It is both less than that and more than that. The In theories of predictive perception, the brain ap­­proximates
reality we experience—the way things s eem—is not a direct re- this kind of Bayesian inference by continually generating predic-
flection of what is actually out there. It is a clever construction tions about sensory signals and comparing these predictions with

40  |  SCIENTIFIC AMERICAN  |  SPECIAL EDITION  |  WINTER 2022

 the sensory signals that arrive at the trolled hallucination does not mean it
eyes and the ears (and the nose and is okay to jump in front of a bus. This
the fingertips, and all the other sen- bus has primary qualities of solidity
sory surfaces on the outside and in- and space occupancy that exist inde-
side of the body). The differences be- pendently of our perceptual machin-
tween predicted and actual sensory ery and that can do us injury. It is the
signals give rise to so-called predic- way in which the bus appears to us
tion errors, which are used by the that is a controlled hallucination, not
brain to update its predictions, ready- the bus itself.
ing it for the next round of sensory in-
puts. By striving to minimize sensory-­ TRIPPING IN THE LAB
prediction errors everywhere and all A growing body of evidence sup-
the time, the brain implements ap- ports the idea that perception is con-
proximate Bayesian in­­ference, and trolled hallucination, at least in its
the resulting Bayesian best guess is broad outlines. A 2015 study by Chris-
what we perceive. toph Teufel of Cardiff University in
To understand how dramatically Wales and his colleagues offers a
this perspective shifts our intuitions striking example. In this study, the
about the neurological basis of percep- ability to recognize so-called two-tone
tion, it is helpful to think in terms of images was evaluated in patients with
bottom-up and top-down directions of early-stage psychosis who were prone
signal flow in the brain. If we assume to hallucinations.
that perception is a direct window Take a look at the top photograph
onto an external reality, then it is nat- POORLY EXPOSED p  hotograph of a dress on page 43—a sample of a two-tone
ural to think that the content of per- appears blue and black to some people, image. Probably all you will see is a
ception is carried by bottom-up sig- white and gold to others. bunch of black-and-white splotches.
nals—those that flow from the senso- Now look at the image at the bottom
ry surfaces inward. Top-down signals might contextualize or finesse of that page. Then have another look at the first photo; it ought to
what is perceived, but nothing more. Call this the “how things seem” look rather different. Where previously there was a splotchy mess,
view because it seems as if the world is revealing itself to us direct- there are now distinct objects, and something is happening.
ly through our senses. What I find remarkable about this exercise is that in your sec-
The prediction machine scenario is very different. Here the ond examination of the top image, the sensory signals arriving at
heavy lifting of perception is performed by the top-down signals your eyes have not changed at all from the first time you saw it.
that convey perceptual predictions, with the bottom-up sensory All that has changed are your brain’s predictions about the causes
flow serving only to calibrate these predictions, keeping them of these sensory signals. You have acquired a new high-level per-
yoked, in some appropriate way, to their causes in the world. In ceptual expectation, and this changes what you consciously see.
this view, our perceptions come from the inside out just as much If you show people many two-tone images, each followed by
as, if not more than, from the outside in. Rather than being a pas- the full picture, they might subsequently be able to identify a good
sive registration of an external objective reality, perception proportion of two-tone images, though not all of them. In Teufel’s
emerges as a process of active construction—a controlled hallu- study, people with early-stage psychosis were better at recogniz-
cination, as it has come to be known. ing two-tone images after having seen the full image than were
Why controlled hallucination? People tend to think of hallu- healthy control subjects. In other words, being hallucination-
cination as a kind of false perception, in clear contrast to verid- prone went along with perceptual priors having a stronger effect
ical, true-to-reality, normal perception. The prediction machine on perception. This is exactly what would be expected if halluci-
view suggests instead a continuity between hallucination and nations in psychosis depended on an overweighting of perceptu-
normal perception. Both depend on an interaction be­­tween top- al priors so that they overwhelmed sensory prediction errors, un-
down, brain-based predictions and bottom-up sensory data, but mooring perceptual best guesses from their causes in the world.
during hallucinations, sensory signals no longer keep these Recent research has revealed more of this story. In a 2021
top-down predictions appropriately tied to their causes in the study, Biyu He of New York University and her colleagues had
world. What we call hallucination, then, is just a form of uncon- neurosurgical patients look at ambiguous images, such as a Neck-
trolled perception, just as normal perception is a controlled form er cube, that constantly flip between two different appearances
of hallucination. even though the sensory input remains the same. By analyzing
This view of perception does not mean that nothing is real. Writ- the signals recorded from within the patients’ brains, they dis-
ing in the 17th century, English philosopher John Locke made an covered that information flowed more strongly in a top-down,
influential distinction between “primary” and “secondary” quali- “inside-out” direction when the perceived appearance was con-
Swiked.tumblr.com

ties. Primary qualities of an object, such as solidity and oc­­­cu­pan­cy sistent with the patients’ biases, as would be expected if percep-
of space, exist independently of a perceiver. Secondary qualities, in tual predictions were strong in this case. And when the perceived
contrast, exist only in relation to a perceiver—color is a good exam- appearance was inconsistent with preexisting biases, infor­
ple. This distinction ex­­plains why conceiving of perception as con- mation flow was stronger in the bottom-up direction, suggest-

SCIENTIFICAMERICAN.COM  |  41
ing a “prediction error” signal. This is an exciting new develop- The basic idea is simple. We again prerecorded some panoram-
ment in mapping the brain basis of controlled hallucinations. ic video footage, this time of the interior of our VR lab rather than
In my lab we have taken a different approach to exploring the of an outside campus scene. People coming to the lab are invited to
nature of perception and hallucination. Rather than looking into sit on a stool in the middle of the room and to put on a VR headset
the brain directly, we decided to simulate the influence of over- that has a camera attached to the front. They are encouraged to look
active perceptual priors using a unique virtual-reality setup mas- around the room and to see the room as it actually is, via the cam-
terminded by our resident VR guru, Keisuke Suzuki. We call it, era. But at some point, without telling them, we switch the feed so
with tongue firmly in cheek, the “hallucination machine.” that the headset now displays not the live real-world scene but rath-
Using a 360-degree camera, we first recorded panoramic vid- er the prerecorded panoramic video. Most people in this situation
eo footage of a busy square in the University of Sussex campus on continue to experience what they are seeing as real even though it
a Tuesday at lunchtime. We then processed is now a fake prerecording. (This is actual-
the footage through an algorithm based on ly very tricky to pull off in practice—it re-
Google’s AI program DeepDream to gen-
erate a simulated hallucination. What hap-
OUR PERCEPTIONS quires careful color balancing and align-
ment to avoid people noticing any differ-
pens is that the algorithm takes a so-called COME FROM THE ence that would tip them off to the shift.)
neural network—one of the workhorses of
AI—and runs it backward. The network we INSIDE OUT JUST AS I find this result fascinating because it
shows that it is possible to have people ex-
used had been trained to recognize objects
in images, so if you run it backward, up-
MUCH AS, IF NOT perience an unreal environment as being
fully real. This demonstration alone opens
dating the network’s input instead of its MORE THAN, FROM new frontiers for VR research: we can test
output, the network effectively projects
what it “thinks” is there onto and into the
THE OUTSIDE IN. the limits of what people will experience,
and believe, to be real. It also allows us to
image. Its predictions overwhelm the sen- investigate how experiencing things as be-
sory inputs, tipping the balance of perceptual best guessing to- ing real can affect other aspects of perception. Right now we are
ward these predictions. Our particular network was good at clas- running an experiment to find out whether people are worse at
sifying different breeds of dogs, so the video became unusually detecting unexpected changes in the room when they believe that
suffused by dog presences. what they are experiencing is real. If things do turn out this way
Many people who have viewed the processed footage through (the study is still ongoing, despite being heavily delayed by a glob-
the VR headset have commented that the experience is rather al pandemic), that finding would support the idea that the percep-
reminiscent not of the hallucinations of psychosis but of the ex- tion of things as being real itself acts as a high-level prior that can
uberant phenomenology of psychedelic trips. substantively shape our perceptual best guesses, affecting the con-
More recently, we have been implementing the hallucination tents of what we perceive.
machine in different ways to simulate different kinds of altered
visual experience. By extending our algorithm to include two cou- THE REALITY OF REALITY
pled neural networks—a “discriminator network” much like the The idea that the world of our experience might not be real
one in our original study and a “generator” network that has been is an enduring trope of philosophy and science fiction, as well as
trained to reproduce (“generate”) its input image—we have been of late-night pub discussions. Neo in T  he Matrix takes the red
able to model different types of hallucination. For example, we pill, and Morpheus shows him how what he thought was real is
have modeled the complex hallucinatory experiences reported an elaborate simulation, while the real Neo lies prone in a human
by people with Parkinson’s disease and some forms of dementia; body farm, a brain-in-a-vat power source for a dystopian AI. Phi-
the patterned, geometric hallucinations that occur after the loss losopher Nick Bostrom of the University of Oxford has famously
of foveal vision, as happens in Charles Bonnet syndrome; and a argued, based largely on statistics, that we are likely to be living
range of psychedeliclike hallucinations. We hope that by under- inside a computer simulation created in a posthu­­­­man age. I dis-
standing hallucinations better, we will be able to understand nor- agree with this argument in part because it assumes that con-
mal experience better, too, because predictive perception is at the sciousness can be simulated—I do not think that this is a safe as-
root of all our perceptual experience. sumption—but it is thought-provoking nonetheless.
Although these chunky metaphysical topics are fun to chew
THE PERCEPTION OF REALITY on, they are probably impossible to resolve. Instead what we have
Although the hallucination machine is un­­doubt­ed­ly trippy, peo- been exploring throughout this article is the relation between ap-
ple who experience it are fully aware that what they are experienc- pearance and reality in our conscious perceptions, where part of
ing is not real. Indeed, despite rapid advances in VR technology and this appearance is the appearance of being real itself.
computer graphics, no current VR setup delivers an experience that The central idea here is that perception is a process of active in-
is sufficiently convincing to be indistinguishable from reality. terpretation geared toward adaptive interaction with the world
This is the challenge we took up when designing a new “sub- through the body rather than a re-creation of the world within the
stitutional reality” setup at Sussex—the one we were working on mind. The contents of our perceptual worlds are controlled hallu-
when Pope Francis convened the retreat with Salva Kiir and Riek cinations, brain-­based best guesses about the ultimately unknow-
Machar. Our aim was to create a system in which volunteers able causes of sensory signals. For most of us, most of the time, these
would experience an environment as being real—and believe it controlled hallucinations are experienced as real. As Canadian rap-
to be real—when in fact it was not real. per and science communicator Baba Brinkman suggested to me,

42  |  SCIENTIFIC AMERICAN  |  SPECIAL EDITION  |  WINTER 2022

 when we agree about our hallucinations,
maybe that is what we call reality.
But we do not always agree, and we do
not always experience things as real. People
with dissociative psy­­chiatric conditions
such as derealization or depersonalization
syndrome report that their perceptual
worlds, even their own selves, lack a sense
of reality. Some kinds of hallucination, var-
ious psychedelic hallucinations among
them, combine a sense of unreality with per-
ceptual vividness, as does lucid dreaming.
People with synesthesia consistently have
additional sensory experiences, such as per-
ceiving colors when viewing black letters,
which they recognize as not real. Even with
normal perception, if you look directly at
the sun you will experience the subsequent
retinal afterimage as not being real. There
are many such ways in which we experience
our perceptions as not fully real.
What this means to me is that the prop- TWO-TONE IMAGE looks like a mess of black-and-white splotches, until you see
erty of realness that attends most of our the full image (below).
perceptions should not be taken for grant-
ed. It is another aspect of the way our brain settles on its Bayesian tive, creative mechanisms of perception has unexpected social rel-
best guesses about its sensory causes. One might thus ask what evance. Perhaps once we can appreciate the diversity of experi-
purpose it serves. Perhaps the answer is that a perceptual best enced realities scattered among the billions of perceiving brains
guess that includes the property of being real is usually more fit on this planet, we will find new platforms on which to build a
for purpose—that is, better able to guide behavior—than one that shared understanding and a better future—whether between sides
does not. We will behave more appropriately with respect to a cof- in a civil war, followers of different political parties, or two people
fee cup, an approaching bus or our partner’s mental state when we sharing a house and faced with washing the dishes. 
experience it as really existing.
But there is a trade-off. As illustrated by the dress illusion, when Anil K. Seth is a professor of cognitive and computational neuroscience at the University
we experience things as being real, we are less able to appreciate of Sussex in England. His research focuses on the biological basis of consciousness, and he
that our perceptual worlds may differ from those of others. (A pop- is author of Being You—A New Science of Consciousness (Dutton, 2021).
ular explanation for the differing percep-
tions of the garment holds that people who
spend most of their waking hours in day-
light see it as white and gold; night owls,
who are mainly exposed to artificial light,
see it as blue and black.) And even if these
differences start out small, they can become
entrenched and reinforced as we proceed to
harvest information differently, selecting
sensory data that are best aligned with our
individual emerging models of the world
and then updating our perceptual models
based on these biased data. We are all famil-
iar with this process from the echo cham-
bers of social media and the newspapers we
choose to read. I am suggesting that the
Richard Armstrong/EyeEm/Getty Images

same principles apply also at a deeper lev-

el, underneath our sociopolitical beliefs,
right down to the fabric of our perceptual
realities. They may even apply to our per-
ception of being a self—the experience of be-
ing me or of being you—because the expe-
rience of being a self is itself a perception. PERCEPTUAL SHIFT: V  iewing this photograph changes what one consciously sees
This is why understanding the construc- in the two-tone image (above).

SCIENTIFICAMERICAN.COM  |  43
why we
have
free
wıll
Neurons fire in your head before you become aware
that you have made a decision. But this discovery
does not mean you are a “biochemical puppet”
By Eddy Nahmias
Illustration by Simon Prades

O
ne fall night I lay awake wondering how I should
begin this essay. I imagined a variety of ways I could write
the first sentence and the next and the one after that.
Then I thought about how I could tie those sentences to
the following paragraph and the rest of the article. The
pros and cons of each of those options circled back and
forth in my head, keeping me from drifting off to sleep.
As this was happening, neurons were buzzing away in my brain. Indeed, that
neural activity explains why I imagined these options, and it explains why I am
writing these very words. It also explains why I have free will.

SCIENTIFICAMERICAN.COM  |  45
 Increasingly, neuroscientists, psychologists and NOT SO FAST
pundits say that I am wrong. Invoking a number of I call those who contend that science shows that
widely cited neuroscientific studies, they claim that free will is an illusion “willusionists.” There are many
unconscious processes drove me to select the words I reasons to be wary of the willusionists’ arguments. First,
ultimately wrote. Their arguments suggest our con- neuroscience currently lacks the technical sophistica-
scious deliberation and decisions happen only after tion to determine whether neural activity underlying
neural gears below the level of our conscious aware- our imagining and evaluating of future options has any
ness have already determined what we will choose. And impact on which option we then carry out minutes,
they conclude that because “our brains make us do it”— hours or days later. Instead the research discussed by
choosing for us one option over another—free will is willusionists fails to clearly define the border between
nothing more than an illusion. conscious and unconscious actions.
The experiments most often cited to show that our Consider the Libet experiment. It began with study
brains take charge behind the scenes were carried out participants preparing consciously to make a series of
by the late Benjamin Libet in the 1980s at the Univer- repetitive and unplanned actions. When the experiment
sity of California, San Francisco. There he instructed began, they flexed their wrists when a desire arose spon-
study participants outfitted with electrodes on their taneously. The neural activity involved in the conscious
heads to flick their wrists whenever they felt like it. The planning presumably influenced the later unconscious
electrodes detected fluctuations in electrical activity initiation of movements, revealing an interaction
called readiness potentials that occurred about half a between conscious and unconscious brain activity.
second before people made the flicking motion. But Similarly, the 2011 Haynes study, in which people
participants became aware of their intentions to move randomly picked whether to add or subtract over the
only about a quarter of a second before the movement, course of many trials, fails to provide convincing evi-
leading to the conclusion that their brains had decid- dence against free will. Early brain activity that occurred
ed before they became aware of what had happened. four seconds before participants were aware of making
In essence, unconscious brain processes were in the a choice may be an indication of unconscious biases
driver’s seat. toward one choice or the other.
More recent studies using functional MRI have sug- But this early brain activity predicted a choice with
gested the unconscious roots of our decisions begin an accuracy only 10 percent better than random chance.
even earlier. In research published in 2013, neurosci- Brain activity cannot, in general, settle our choices four
entist John-Dylan Haynes of the Bern­stein Center for seconds before we act, because we can react to changes
Computational Neuroscience Berlin and his colleagues in our situation in less time than that. If we could not,
had volunteers decide whether to add or subtract two we would all have died in car crashes by now! Uncon-
numbers while in the fMRI scanner. They found pat- scious neural activity, however, can prepare us to take
terns of neural activity that were predictive of wheth- an action by cuing us to consciously monitor our actions
er subjects would choose to add or subtract that to let us adjust our behavior as it occurs.
occurred four seconds before those subjects were aware Willusionists also point to psychological research
of making the choice. showing that we have less conscious control over our
These studies—and others like them—have led to actions than we think. It is true that we are often influ-
sweeping pronouncements that free will is dead. “Our enced unknowingly by subtle features of our environ-
decisions are predetermined unconsciously a long time ment and by emotional or cognitive biases. Until we
before our consciousness kicks in,” Haynes comment- understand them, we are not free to try to counteract
ed to N
 ew Scientist, a dding that “it seems that the brain them. This is one reason I think we have less free will
is making the decision before the person.” Others share than many people tend to believe. But there is a big dif-
his opinion. Evolutionary biologist Jerry Coyne has ference between less and none at all.
written: “So it is with all of our  ... choices: not one of The Libet and Haynes research deals with choices
them results from a free and conscious decision on our that people make without conscious deliberation at the
part. There is no freedom of choice, no free will.” Neu- time of action. Everyone performs repetitive or habitu-
roscientist Sam Harris has concluded from these find- al behaviors, sometimes quite sophisticated ones that
ings that we are “biochemical puppets”: “If we were to do not require much thought because the behaviors have
detect [people’s] conscious choices on a brain scanner been learned. You put your key in the lock. A shortstop
seconds before they were aware of them ... this would dives for a ground ball. A pianist becomes immersed in
directly challenge their status as conscious agents in playing Beethoven’s M  oonlight Sonata.
control of their inner lives.” The reflexive turning of the key, the lunging for the
But does the research really show that all our con- ball, or the depressing of the white and black keys requires
scious deliberation and planning is just a by-product a particular type of mental processing. What I was doing
of unconscious brain activity, having no effect on what on that sleepless night—conscious consideration of alter-
we do later on? No, it does not. There are several rea- native options—is a wholly different activity from engag-
sons to think that those who insist that free will is a ing in practiced routines. A body of psychological research
mirage are misguided. shows that conscious, purposeful processing of our

46  |  SCIENTIFIC AMERICAN  |  SPECIAL EDITION  |  WINTER 2022

thoughts likely does make a difference in what we do. ris has suggested this scenario “would expose this feel-
This work indicates that intentions we formulate to ing [of free will] for what it is: an illusion.”
carry out specific tasks in particular circumstances— To see whether people’s belief in free will would be
what psychologists call “implementation intentions”— challenged by the knowledge that the brain is engaged in
increase the likelihood that we will complete the planned unconscious information processing that predicts behav-
behavior. A study performed by psychologist Peter Goll- ior, Jason Shepard of Life University in Georgia, Shane
witzer of New York University and his colleagues revealed Reuter, then at Washington University in St. Louis, and
that dieters who consciously formed an intention to I performed a series of experiments in which we present-
ignore thoughts about tempting foods whenever they ed people with detailed scenarios describing futuristic
came to mind then ate less of those foods than those diet- brain-imaging technology, as posited by Harris.
ers who simply set the goal to lose weight. Hundreds of students at Georgia State University par-
Psychologist Roy F. Baumeister, then at Florida State ticipated in the studies. They read about a woman named
University, and his colleagues demonstrated that con- Jill who, in the distant future, wore a brain-imaging cap
scious reasoning improves performance on logical and for a month. Using information from the brain scanner,
linguistic tasks and that it helps in learning from past neuroscientists predicted everything she thought and did,
mistakes and overriding impulsive behaviors. In addi- even when she tried to fool the system. The scenario con-
tion, the late Walter Mischel of Columbia University cluded that “these experiments confirm that all human
found that our ability to willfully distract ourselves from mental activity just i s b
 rain activity such that everything
a temptation is crucial for self-control. that any human thinks or does could be predicted ahead
Every one of us carries out actions every day that we of time based on their earlier brain activity.”
have consciously planned for ourselves. It is possible that More than 80  percent of the participants reported
the neural activity that carries out this planning has no that they believed that such future technology was pos-
effect on what we do or that it just concocts stories after sible, yet 87  per­cent of them responded that Jill still
the fact to explain to ourselves and others what we did. had free will. They were also asked whether the exis-
But that would make little evolutionary sense. The brain tence of such technology would indicate that individ-
makes up only 2 percent of the human body’s weight but uals lack free will. Roughly 75  percent disagreed. Fur-
consumes 20  percent of its energy. There would be ther results showed that a significant majority felt that
strong evolutionary pressure against neural processes as long as the technology did not allow people’s brains
that enable intricate conscious thought yet are irrele- to be manipulated such that their decisions could be
vant to our behavior. The brain circuits responsible for controlled by others, they would have free will and be
my imagining that this is the best way to write this essay morally responsible for their behavior.
are likely causing it to turn out this way. Most participants in the experiments seem to think
that the hypothetical brain scanner is just recording
FREE WILL IN THE BRAIN? the brain activity that is Jill’s conscious reasoning and
W i l lu s i o n i s t s , h
 owever, suggest brain processing consideration about what to decide. Rather than tak-
responsible for conscious thinking simply cannot count ing this to mean that Jill’s brain is making her do some-
as free will. They often say that people who believe in thing—and that she has no free will—they may just be
free will must be “dualists” who are convinced that the thinking that the brain scanner is simply detecting how
mind somehow exists as a nonphysical entity, separate free will works in the brain.
from the brain. “Free will is the idea that we make choic- Why, then, do willusionists believe the opposite? It
es and have thoughts independent of anything remote- may have to do with the current state of knowledge.
ly resembling a physical process,” wrote neuroscientist Until neuroscience is able to explain consciousness—
Read Montague in 2008. And Coyne has claimed that which will require a theory to explain how our mind is
“true ‘free will’ ... would require us to step outside of our neither reducible to nor distinct from the workings of
brain’s structure and modify how it works.” our brain—it is tempting to think, as the willusionists
It is true that some people think of free will in this seem to, that if the brain does it all, there is nothing
way. But there is no good reason to do so. Most philosoph- left for the conscious mind to do.
ical theories develop a view of free will that is consistent As neuroscience advances and imaging technology
with a scientific understanding of human nature. And improves, these developments should help reveal more
despite willusionists’ claims, studies suggest most peo- precisely how much conscious control we have and to
ple accept that we can have free will even if our mental what extent our actions are governed by processes
activity is carried out entirely by brain activity. If most beyond our control. Finding resolutions for these ques-
people are not committed to a dualist view about free tions about free will is important. Our legal system—and
will, then it is a mistake to tell them that free will is an the moral basis for many of our society’s institutions—
illusion based on the scientific view that dualism is false. requires a better understanding of when people are, and
One way to test people’s assumptions about free will are not, responsible for what they do. 
is to de­­scribe the possibility of brain-imaging technol-
ogy that would allow perfect prediction of actions based Eddy Nahmias i s a professor in the department of philosophy and the
on information about prior brain activity. In fact, Har- Neuroscience Institute at Georgia State University.

SCIENTIFICAMERICAN.COM  |  47
The Brain
Learns in
Unexpected
Ways
Neuroscientists Our concepts of how the two and a half pounds
of flabby flesh between our ears accomplish learning date to
have discovered Ivan Pavlov’s classic experiments, where he found that dogs
a set of unfamiliar could learn to salivate at the sound of a bell. In 1949 psycholo-
gist Donald Hebb adapted ­Pavlov’s “associative learning rule” to
cellular mechanisms explain how brain cells might acquire knowledge. Hebb pro-
posed that when two neurons fire together, sending off im­­pulses
for making simultaneously, the connections between them—the synapses—
grow stronger. When this happens, learning has taken place. In
fresh memories the dogs’ case, it would mean the brain now knows that the
sound of a bell is followed immediately by the presence of food.
By R. Douglas Fields This idea gave rise to an oft-quoted axiom: “Synapses that fire
together wire together.”
Illustration by Eva Vazquez The theory proved sound, and the molecular details of how
synapses change during learning have been described in depth.
But not everything we remember results from reward or punish-
ment, and in fact, most experiences are forgotten. Even when
synapses do fire together, they sometimes do not wire together.
What we retain depends on our emotional response to an expe-
rience, how novel it is, where and when the event occurred and
our level of attention and motivation during the event, and we

SCIENTIFICAMERICAN.COM  |  49
process these thoughts and feelings while asleep. A narrow fo- surprised to find changes in unexpected parts of the brain, in-
cus on the synapse has given us a mere stick-figure conception cluding regions that have no neurons or synapses­­—­areas known
of how learning and the memories it engenders work. as white matter.
It turns out that strengthening a synapse cannot produce a
memory on its own, except for the most elementary reflexes in DEEP LEARNING
simple circuits. Vast changes throughout the expanse of the Consciousness arises from the cerebral cortex, the three-milli-
brain are necessary to create a coherent memory. Whether you meter-thick outer layer of the human brain, so this gray matter
are recalling last night’s conversation with dinner guests or us- layer is where most researchers expected to find learning-­
ing an acquired skill such as riding a bike, the activity of mil- induced modifications. But below the surface layer, billions of
lions of neurons in many different regions of your brain must tightly packed bundles of axons (nerve fibers), much like tightly
become linked to produce a coherent memory that interweaves wound fibers under the leather skin of a baseball, connect neu-
emotions, sights, sounds, smells, event sequences and other rons in the gray matter into circuits.
stored experiences. Because learning encompasses so many ele- These fiber bundles are white because the axons are coated
ments of our experiences, it must incorporate different cellular with a fatty substance called myelin, which acts as electrical in-
mechanisms beyond the changes that occur in synapses. This sulation and boosts the speed of transmission by 50 to 100 times.
recognition has led to a search for new ways to understand how White matter injury and disease are important areas of research,
information is transmitted, processed and stored in the brain to but until recently little attention had been given in these inves-
bring about learning. In the past 10 years neuroscientists have tigations to a possible role of myelin in information processing
come to realize that the iconic “gray matter” that makes up the and learning.
brain’s outer surface—familiar from graphic illustrations found In the past 10 or so years studies have begun to find differ-
everywhere from textbooks to children’s cartoons—is not the ences in white matter in brain scans of experts with a variety of
only part of the organ involved in the inscription of a permanent skills, including people with high proficiency in reading and
record of facts and events for later recall and replay. It turns out arithmetic. Expert golfers and trained jugglers also show differ-
that areas below the deeply folded, gray-colored surface also ences in white matter compared with novices, and white matter
play a pivotal role in learning. In just the past few years a series volume has even been associated with IQ. If information pro-
of studies from my laboratory and others has elucidated these cessing and learning arise from the strengthening of synaptic
processes, which could point to new ways of treating psychiat- connections between neurons in gray matter, why does learning
ric and developmental disorders that occur when learning im- affect the brain’s subsurface cabling?
pairments arise. A possible answer began to emerge from cellular studies in
If synaptic changes alone do not suffice, what does happen my lab investigating how synapses—but also other brain areas—
inside your brain when you learn something new? Magnetic res- change during learning. The reason for looking beyond the syn-
onance imaging methods now enable researchers to see through apse was that most of the drugs we have for treating neurologi-
a person’s skull and examine the brain’s structure. In scrutiniz- cal and psychological disorders work by altering synaptic
ing MRI scans, investigators began to notice differences in the transmission, and there is a pressing need for more effective
brain structure of individuals with specific highly developed agents. The present focus on synaptic transmission might cost
skills. Musicians, for example, have thicker regions of auditory us opportunities for better treatments for dementia, depression,
cortex than nonmusicians. At first, researchers presumed that schizophrenia or post-traumatic stress disorder (PTSD).
these subtle differences must have predisposed clarinetists and In the early 1990s my lab at the National Institutes of Health
pianists to excel at their given skills. But subsequent research and others began to explore the possibility that glia might be
found that learning changes the structure of the brain. able to sense information flowing through neural networks and
The kind of learning that leads to alterations in brain tissue alter it to improve performance. Experimental evidence that has
is not limited to repetitive sensorimotor skills such as playing accumulated since then shows that all types of glial cells re-
a musical instrument. Neuroscientist Bogdan Draganski, cur- spond to neural activity and can modify information transmis-
rently at the University of Lausanne in Switzerland, and his col- sion in the brain. One of the most surprising of these new find-
leagues witnessed increases in the volume of gray matter in ings involves myelin.
medical students’ brains after they studied for an examination. Myelin insulation is formed by layers of cell membrane
Many different cellular changes could expand gray matter vol- wrapped around axons like electrical tape. In the brain and spinal
ume, including the birth of new neurons and of nonneuronal cord, octopus-shaped glial cells (oligodendrocytes) do the wrap-
cells called glia. Vascular changes and the sprouting and prun- ping. In the limbs and trunk, sausage-shaped glial cells (Schwann
ing of axons and dendrites that extend from the main body of a cells) perform the same task. Many oligodendrocytes grip an axon
neuron could also do the same. Remarkably, physical changes in and wrap layers of myelin around it in segments, like the stacked
the brain can happen much faster during learning than might hands of baseball players gripping a bat to determine which team
be expected. Yaniv Assaf of Tel Aviv University and his colleagues bats first. The tiny gap between two myelin segments exposes a
showed that 16 laps around a race track in a computerized video one-micron section of bare axon where ion channels that generate
game were enough to cause changes in new players’ hippo­ electrical impulses become concentrated. These spaces, known as
campal brain region. Structural alterations in the hippo­campus the nodes of Ranvier, act like bioelectric repeaters to relay an elec-
in these gamers make sense because this brain region is critical trical impulse from node to node down the axon. The speed of im-
for spatial learning for navigation. In other studies, Assaf and, pulse transmission increases as more layers of myelin are wrapped
separately, Heidi Johansen-Berg of the University of Oxford were around the axon, protecting it better against voltage loss. Also, as

50  |  SCIENTIFIC AMERICAN  |  SPECIAL EDITION  |  WINTER 2022

 Neuron

Astrocyte

Axon Insulation Oligodendrocyte

Neuroscience textbooks r ecount that the connecting forming perinodal
loop around axon
points between neurons—the synapses—undergo altera-
tions when learning takes place. But new research shows Inner
Node of tongue
that changes also occur in myelin, part of the white matter Ranvier
that forms a sheath around the long filaments (axons)
that stretch out from the cell body of a neuron.

Worker Cells
Insulating sheaths made of fatty, white myelin control the rate
at which electrical signals travel along axons. Cells called Axon
oligodendrocytes loop around and wrap myelin on an axon—
and, in some cases, remove it. Small gaps in myelin (nodes of Outer
Ranvier) contain ion channels that generate electrical impulses. tongue
Another cell type, the perinodal astrocyte, stops the secretion
of the myelin-removing thrombin (not shown). Oligodendrocyte

Thickening of the Inner tongue of oligodendrocyte

Myelin Sheath expands and wraps around axon
Time Wrapping and
Unwrapping
a Neuron
Oligodendrocytes start
wrapping myelin around
Astrocyte axons in electrically active
neurons. The degree of mye­
Axon lination controls how fast a
signal travels along an axon,
with thicker sheaths producing
Thinning of the Outer tongue detaches Perinodal speedier transmissions. The
Myelin Sheath and withdraws back to astrocyte enzyme thrombin cuts the
the cell body stitches that bind myelin to
the axon, and the perinodal
astrocyte brings this process
to a halt to procure the desired
thickness. Varying myelin’s
depth ensures that dispersed
signals arrive at a neural relay
point at the same time, enhanc­
Stitches ing performance on a new task.

a node of Ranvier becomes squeezed more tightly by the adjoining How, then, does the transmission speed in every link in the
myelin segments, an electrical impulse is initiated more rapidly human brain get timed appropriately so that an impulse arrives
because it takes less time to charge the smaller amount of nodal just when needed? We know that electrical signals shuffle along
Source: “Treadmilling Model for Plasticity of the Myelin Sheath,” by R. Douglas Fields

membrane to the voltage that triggers ion channels to open and at the pace of a slow walk in some axons but blaze away at the
generate an impulse. speed of a race car in others. Signals from two axons that con-
and Dipankar J. Dutta, in Trends in Neurosciences, Vol. 42, No. 7; July 2019

Disorders that damage myelin, such as multiple sclerosis and verge on neurons that act as relay points will not arrive together
Guillain-Barré syndrome, can cause serious disability because unless the travel time from their input source is optimized to
neural impulse transmission fails when the insulation is dam- compensate for differences in the lengths of the two axons and
aged. But until recently, the idea that myelin might be modified the speed at which impulses travel along each link.
routinely by neural impulses was not widely accepted. And even Because myelin is the most effective means of speeding im-
if myelin structure changed, how and why would this improve pulse transmission, axon myelination promotes optimal infor-
performance and learning? mation transmission through a network. If oligodendrocytes
The explanation was hiding in plain sight. It loops back to sense and respond to the information traffic flowing through
the old maxim about neurons firing and wiring together. In any neural circuits, then myelin formation and the way it adjusts im-
complex information or transportation network, the time of ar- pulse-transmission speed could be controlled by feedback from
rival at network relay points is critical—think of missing a con- the axon. But how can myelinating glia detect neural impulses
nection because your flight arrives too late. flowing through axons?

Illustration by David Cheney SCIENTIFICAMERICAN.COM  |  51

 SIGNAL TRANSMISSION genetics to make neurons fire at the appropriate time. The mice
Over the past t wo de cades our research and that of other with impaired myelination then performed the learned task pro-
labs has succeeded in identifying many neurotransmitters and ficiently. Eventually less invasive forms of brain stimulation may
other signaling molecules that convey to glia the presence of elec- become effective therapy to treat neurological and psychologi-
trical activity in the axon to stimulate myelination. Our experi- cal disorders caused by disrupted myelination.
ments have shown that when a neuron fires, neurotransmitters Despite these recent advances, stimulation to increase axon
are released not only at synapses but also all along the axon. We myelination is not always enough to enable new learning, because
found that the “tentacles” of the octopuslike oligodendrocytes we cannot synchronize the arrival of spikes at critical relay points
probe bare sections of axons in search of neurotransmitters being in neural networks simply by making the impulses travel as rap-
released from axons firing. When a single tentacle touches an idly as possible. There must also be a way to slow the speed of im-
axon that is firing, it forms a “spot weld” contact, which enables pulses from inputs that arrive at those points too soon.
communication between the axon and the oligodendrocyte. The The myelin that has already formed on axons has to be thick-
oligodendrocyte begins to synthesize myelin at that spot and ened or thinned in a controlled way to speed or slow signal trans-
wrap it around the axon. mission. Prior to our findings, there was no known explanation for
When we gave oligodendrocytes in cell culture the choice of how the myelin sheath could be thinned to slow signals, aside
myelinating electrically active axons or ones treated with botuli- from disease damage. Our research revealed another type of glial
num toxin to prevent the release of neurotransmitters, the oligo- cell involved in these “plastic” nervous system changes.
dendrocytes opted for the electrically active Surrounding the node of Ranvier is a
axons over the silent ones by a factor of glial cell called an astrocyte. Astrocytes
eight to one. So it may be that as a person have many functions, but most neurosci-
learns to play “Für Elise” on the piano, bare entists have largely ignored them because
axons are wrapped with myelin or the vol- they do not communicate with other cells
ume of existing sheaths is increased in cir- through electrical impulses. Surprisingly,
cuits that are activated repetitively during research in the past decade has shown
practice, which speeds information flow that astrocytes positioned close to the
through brain networks. New myelin then synapse between two neurons can regu-
shows up on an MRI as changes in white late synaptic transmission during learn-
matter tracts in parts of the brain that are ing by releasing or taking up neurotrans-
necessary for musical performance. mitters there. But until fairly recently,
Several labs have verified that action myelin biologists tended to ignore the
potentials, signals coursing the length of unique type of astrocyte that contacts an
axons, stimulate myelination of these ex- OLIGODENDROCYTE (g  reen) axon at a node of Ranvier.
posed areas of neural wiring. In 2014 Mi- prepares to coat an axon (purple) What exactly do these so-called peri-
chelle Monje’s lab at Stanford University with myelin. nodal astrocytes do to thin the myelin
showed that optogenetic stimulation (us- sheath? Just as one would begin when re-
ing lasers to make neurons fire) increased modeling a garment, these cells assist in
myelination in the mouse brain. That same year William Richard- cutting the “seams.” The myelin sheath is attached to the axon by
son’s lab at University College London demonstrated that when a spiral junction flanking the node of Ranvier. Under an electron
the formation of new myelin is prevented, mice are slower to microscope these junctions appear as spirals of stitches between
learn how to run on a wheel with some of its rungs removed. In the axon and the myelin, and the threads that form each stitch are
studies where they used a confocal microscope to watch myelin composed of a complex of three cell adhesion molecules. Our anal-
form in live zebra fish, researchers in David Lyons’s lab at the Uni- ysis of the molecular composition of these stitch points showed
versity of Edinburgh and in Bruce Appel’s lab at the University of that one of these molecules, neurofascin 155, has a site that can be
Colorado Denver observed that when the release of small sacs cleaved by a specific enzyme, thrombin, to thin the myelin.
containing neurotransmitters from axons is inhibited, often the Thrombin is made by neurons, but it also can enter the brain
first few wraps of myelin slip off, and the oligodendrocyte aborts from the vascular system. As the myelin lifts off the axon, the
the entire process. amount of bare axon at the node of Ranvier increases. The outer
In 2018, working with our colleagues, including Daisuke Kato layer of myelin is attached to the axon adjacent to the perinodal
and others from various institutions in Japan, we showed how astrocytes. When the myelin is detached from the axon, the
R. Douglas Fields/National Institutes of Health and NICHD

myelin promotes learning by ensuring that various spiking elec- outer layer withdraws into an oligodendrocyte, thinning the
trical signals traveling along axons arrive at the same time in the sheath. Both widening of the nodal gap and thinning of the my-
motor cortex, the brain region that controls movement. Using ge- elin sheath slow the speed of impulse transmission.
netically modified mice with impaired myelination that had been We found that the enzyme’s snipping of these threads that
trained to pull a lever to receive a reward, we found that learning stitch myelin to the axon can be controlled by the perinodal astro-
this task increased myelination in the motor cortex. cyte’s release of an inhibitor of thrombin. We carried out experi-
By using electrodes to record neural impulses, we found that ments on genetically modified mice in which astrocytes released
action potentials were less synchronized in the motor cortices less of this thrombin inhibitor. When we looked at their neurons
of mice with faulty myelination. We then boosted the synchro- with an electron microscope, we could see that the myelin had
nization of spike arrivals in the motor cortex by using opto­ thinned and that the nodal gap had increased. By using electronic

52  |  SCIENTIFIC AMERICAN  |  SPECIAL EDITION  |  WINTER 2022

amplifiers to detect neural impulses and measure their speed of This conclusion is based on mathematical modeling of the
transmission, we found that after the myelin thickness decreased fundamental physics of wave propagation done by me, together
in this way, the speed of impulse transmission in the optic nerve with my nih colleagues Sinisa Pajevic and Peter Basser. In 2020
slowed by about 20  percent and the animals’ vision declined. We a study by Patrick Steadman and his colleagues in Paul Frank-
were able to reverse all these changes by injecting thrombin inhib- land’s lab at the University of Toronto provided convincing ex-
itors, which are approved for treating vascular disorders. perimental support for the idea. Using genetically modified mice
Our experiments support a new hypothesis: the myelin in which myelination could be temporarily halted, the research-
sheath’s changes in thickness represent a new form of nervous ers found that the ability to learn to fear an unsafe environment
system plasticity governed by the addition and subtraction of and to remember safe locations depends on the formation of new
myelin. Additional layers of myelin are not added to axons as myelin. Moreover, they found that in this type of learning, brain-
one would wrap tape around a wire, because this would tie the wave activity during sleep becomes coupled between the hippo­
legs of the oligodendrocytes in knots. Instead new insulation is campus and the prefrontal cortex. The prevention of new myelin
affixed through the construction of a new inner layer that spi- formation also weakened connections and resulted in a type of
rals around the axon like a snake below the overlying myelin. impaired recall often found in people who have difficulty associ-
Meanwhile the outer layer of myelin can be detached by the ating fear after a traumatic event with the appropriate context.
perinodal astrocyte to thin the sheath. The thickness of the my- Learning and performing any complex task involves the co-
elin sheath is not fixed; instead it reflects a dynamic balance be- ordinated operation of many different neurons in diverse brain
tween the addition of layers next to the axon and removal of the regions and requires that signals proceed through large neural
outer layer under control of the astrocyte. networks at an optimal speed. The myelin sheath is crucial for
optimal transmission, but people begin to lose myelin in the ce-
BRAINY WAVES rebral cortex in their senior years. This gradual degradation is
T he op timal timing o  f action potentials at relay points is one of the reasons for cognitive slowing and the increasing dif-
critical for strengthening synapses by adjusting their timing to ficulty of learning new things as we age.
allow them to fire together. But myelin plasticity can contribute Consider how transmission delays disrupt long-distance com-
to neural circuit function and learning in another way—by tun- munication by telephone. Similarly, lags in the brain can cause
ing the frequency of brain-wave oscillations. Not all neural ac- cognitive difficulties and disorganized thinking in individuals
tivity in the brain arises from sensory inputs. Much of it takes with psychological disorders such as schizophrenia. Indeed, dif-
place because of what goes on in the brain itself at both con- ferences in brain-wave oscillations are seen in many neurologi-
scious and unconscious levels. This self-generated activity con- cal and psychiatric disorders. Alzheimer’s disease, for instance,
sists of oscillating waves of different frequencies that sweep is associated with changes in white matter.
through the brain, just as the vibration of a car engine at a cer- Drugs that control myelin production could provide new ap-
tain speed will set different parts of the automobile rattling to- proaches to treating these problems. Indeed, Fei Wang and his
gether at resonant frequencies. colleagues at the Third Military Medical University in China, in
These brain waves, or oscillations, are believed to be a key collaboration with Jonah Chan of the University of California,
mechanism for coupling neurons across distant regions of the San Francisco, reported in 2020 that the myelination-promoting
brain, which may be important for sorting and transmitting neu- drug clemastine given to mice with a gene deletion that impairs
ral information. Oscillations, for example, tie together neural ac- development of oligodendrocytes improved learning tested in a
tivity in the prefrontal cortex, which provides contextual mean- water maze. Because myelination is influenced by many forms
ing, and in the hippo­camp­us (responsible for encoding spatial of neural activity, a number of techniques—for example, cogni-
information). This oscillatory coupling enables a person to tive training, neurofeedback and physical therapy—may be help-
quickly recognize a familiar face at work, but it also makes it more ful in treating age-related cognitive decline and other disorders.
difficult to identify the same co-worker in an unfamiliar place. A 2018 study of older adults by Jung-Hae Youn and his col-
More important, the various stages of sleep, critical for stor- leagues in South Korea indicated that 10 weeks of memory-
ing long-term memories, can be identified by brain waves oscil- training exercises increased re­­­call. Brain imaging before and af-
lating at different frequencies. Our experiences accumulated ter training revealed increased integrity of white matter tracts
during the day are replayed during sleep and sorted for storage connecting to the frontal lobe in the group of seniors who un-
or deletion based on how they relate to other memories and emo- dertook the memory-training sessions.
tions, which can mark them as potentially useful (or not) in the These novel concepts have begun to change the way we think
future. Appropriate brain-wave oscillations are believed to be about how the brain works as a system. Myelin, long considered
pivotal in this process of memory consolidation. But the speed of inert insulation on axons, is now seen as making a contribution
impulse transmission is critical in synchronizing brain waves. to learning by controlling the speed at which signals travel along
Just as two toddlers must precisely time their leg movements neural wiring. In venturing beyond the synapse, we are beginning
to drive the up-and-down motion of a teeter-totter, the transmis- to fill out the stick-figure skeleton of synaptic plasticity to create
sion delays between two populations of oscillating neurons a fuller picture of what happens in our brain when we learn. 
must be timed so that coupled neurons oscillate in synchrony
across long distances in the brain. Myelin plasticity is important R. Douglas Fields is a senior investigator at the National Institutes of Health’s Section
for brain waves because the proper conduction velocity is neces- on Nervous System Development and Plasticity. He is author of E lectric Brain: How the New
sary to sustain oscillations that couple two regions of the brain Science of Brainwaves Reads Minds, Tells Us How We Learn, and Helps Us Change for the Better
at the same frequency. (BenBella Books, 2020).

SCIENTIFICAMERICAN.COM  |  53
 THE
BRAIN’S
SOCIAL
ROAD
MAPS
Neural circuits that track our whereabouts
in space and time may also play vital roles in
determining how we relate to other people
By Matthew Schafer and Daniela Schiller
Illustration by Richard Borge

SCIENTIFICAMERICAN.COM  |  55
W e are of ten told that there are no shortcuts in life.
But the brain—even the brain of a rat—is wired in a way that
completely ignores this kind of advice. The organ, in fact, epito-
mizes a short­cut-finding machine.
The first indication that the brain has a knack for finding
alternative routes was described in 1948 by Edward Tolman of the
University of California, Berkeley. Tolman performed a curious
experiment in which a hungry rat ran across an unpainted circular table into a dark, narrow
c orridor. The rat turned left, then right, and then took another
right and scurried to the far end of a well-lit narrow strip,
where, finally, a cup of food awaited. There were no choices to
be made. The rat had to follow the one available winding path,
and so it did, time and time again, for four days.
On the fifth day, as the rat once again ran straight across the
route most likely to bring it to its goal. Quite simply, it must
have built a model of the environment.
Such model building or mapmaking extends to more than
physical space. Mental maps may exist at the core of many of
our most “human” capacities, including memory, imagination,
inferences, abstract reasoning and even the dynamics of social
table into the corridor, it hit a wall—the path was blocked. The interactions. Researchers have begun to explore whether men-
animal went back to the table and started looking for alterna- tal maps document how close or distant one individual is to
tives. Overnight, the circular table had turned into a sunburst another and where that individual resides in a group’s social
arena. Instead of one track, there were now 18 radial paths to hierarchy. How does the brain, in fact, create the maps that
explore, all branching off from the sides of the table. After ven- allow us to make our way about the world?
turing out a few inches on a few different paths, the rat finally
chose to run all the way down path number six, the one leading A SPATIAL MAP
directly to the food. The first hints o  f a neural basis for mental maps came in
Taking the path straight to the food cup without prior expe- the 1970s. While studying a brain region called the hippo­
rience may seem trivial, but from the perspective of behavioral campus in rodents, John O’Keefe of University College London,
psychologists at the time, the rat’s navigational accomplish- along with his student Jonathan Dostrovsky, discovered a par-
ment was a remarkable feat. The main school of animal learn- ticular class of neurons that becomes active when mice occupy
ing in that era believed that maze behavior in a rat is a matter of specific locations in their environment. Some of these neurons
simple stimulus-response associations. When stimuli in the fired when the animal was in one location, and others switched
environment reliably produce a successful response, neural on when it moved to the next spot on the path along which it
connections that represent this association get strengthened. traveled, as if the cells were specialized to track w here t he ani-
In this view, the brain operates like a telephone switchboard mal was in space. By linking sequences of these “place cells”
that maintains only reliable connections between incoming calls together, researchers were able to reconstruct an animal’s navi-
from our sense organs and outgoing messages to the muscles. But gational trajectory. Work over the intervening decades con-
the behavioral switchboard was unable to explain the ability to firmed the existence of place cells in other animals, including
correctly choose a shortcut right off the bat without having first humans, and clarified many of their properties. Along the way,
experienced that specific path. Shortcuts and many other intrigu- a host of cell types surfaced, each uniquely contributing to the
ing observations along these lines lent support to a rival school of brain’s encoding of spatial representations.
thought promulgated by theorists who believe that in the course In the nearby entorhinal cortex, a region connected to the
of learning, a map gets established in a rat’s brain. Tolman—a hippo­campus, a research team led by Edvard Moser and May-
proponent of that school—coined the term: the cognitive map. Britt Moser, former postdoctoral visiting fellows in O’Keefe’s
According to Tolman, the brain does more than just learn laboratory, discovered neurons highly similar to place cells.
the direct associations among stimuli. Indeed, such associa- These cells also fired when an animal was in specific locations.
tions are often brittle, rendered outdated by changes in the But unlike place cells, each of these newly discovered cells
environment. As psychologists have learned in the decades spiked in multiple, regular locations. When mapped onto the
since Tolman’s work, the brain also builds, stores and uses men- animal’s position, the activity patterns of these “grid cells”
tal maps. These models of the world enable us to navigate our resembled highly regular, equilateral triangles. Like a spatial
surroundings, despite complex, changing environments— metric, these cells fired when an animal passed over the vertices
affording the flexibility to use shortcuts or detours as needed. of the triangles. The discovery of these cell types sparked excite-
The hungry rat in Tolman’s experiment must have remembered ment because of the emerging picture of how the brain controls
the location of the food, inferred the angle to it and chosen the navigation. Place cells and grid cells could provide a means to

56  |  SCIENTIFIC AMERICAN  |  SPECIAL EDITION  |  WINTER 2022

 Giving Way to the Abstract
Maps simplify the world b  y reducing an overwhelming amount of sensory and cognitive data into a format that can be used for
navigating physical space, pointing to shortcuts and detours to reach a destination faster. The organization of such maps—built on
the activity of cells dedicated to tracking both space and time—scales in the abstraction of what they represent: from the recognition
of another individual along the way to even a complex space that denotes social power and closeness to others.

Increasing Abstraction
Physical space Social information in physical space Social space
More

Self
Power
y y

Less
x x Close Affiliation Distant

locate oneself in space and determine distance and direction. Shaping plans also occurs during sleep. Sequences of place
These navigational tools are crucial for building mental maps. cell activity can be reactivated during sleep to replay the past or
(O’Keefe and the Mosers received the 2014 Nobel Prize in Medi- simulate the future. Without the ability to simulate new behav-
cine or Physiology for their work on place and grid cells.) iors, we would have to explore a multitude of real-world options
A wide variety of information is useful for creating such a before deciding on what action to take. We would be constant
map, and the hippo­campus-entorhinal system encodes much of empiricists, able to act only on direct observations. Instead off-
it. Discovering the location of a physical goal is one example: as line simulations give us the ability to envision possibilities
an animal navigates toward an objective, some hippocampal neu- without directly experiencing them.
rons fire depending on the direction and distance to reach it. The
cells increase their firing rate as the animal approaches the goal. MENTAL TIME TRAVEL
Other cells also enter the picture. A dedicated population of T ime and space a re inextricably linked. It is difficult to talk
“reward” cells encodes reward locations across different envi- about time without borrowing a spatial metaphor: time “passes”
ronments, providing a signal to guide an animal’s navigation as we “move” through it. We look “forward” to the future and
(think of an “X” marking the spot of treasure on a pirate’s map). “back” on our past. The same hippocampal-entorhinal system
Other cells track speed and direction and in doing so act like tracks move­ment through time. Work done largely in the lab of
internal speedometers and compasses that compute an ani- the late Howard Eichenbaum of Boston University re­­vealed neu-
mal’s progress as it travels through the environment. Specific rons in the hippocampal-entorhinal system that encode the
cells that signal the locations of landmarks in the surroundings time course of an animal’s experience. Time cells fire at succes-
serve as references to correct errors in the animal’s trajectory. A sive moments but do not track time in a simple clocklike fash-
 ol. 100; October 24, 2018

map must also have edges: cells that fire more as the animal ion. Instead they mark temporal context—stretching or shrink-
approaches the map’s perimeter. ing their firing durations if the length of a task changes, for
For humans, the importance of such an abundance of cell types example. Some time cells encode space as well. In the brain, in
seems obvious: the brain is responsible for knowing the location fact, physical and temporal space may be bound together.
Source: “Navigating Social Space,” by Matthew Schafer and Daniela Schiller, in N euron, V

of home and work, walls and dead ends, a favorite shop or the cor- The discovery of the crucial importance of these brain areas
ner store. It is still a mystery as to how all of this information is in space and time was not totally surprising. Psychologists had
drawn together into a coherent map, but these cells appear to pro- long suspected it to be the case. In 1953 Henry Molaison under-
vide the parts list for the elements of neural mapmaking. went bilateral hippocampal resection surgery to reduce ex­­
This hippo­campal-entorhinal system is more than a map- treme, life-disrupting epileptic seizures. The surgery was success-
maker, though, and the maps are more than a way to locate one- ful at quelling the seizures. But Molaison—known for decades
self in space. These maps also are used for active planning. When only as H.M.—became one of the most renowned cases in the
a rat comes to a junction in a familiar maze, it will pause while history of the brain sciences.
place cell firing sequences that relate to the different options are Molaison could remember most experiences from before
activated, as if the animal is contemplating the choices. his surgery—people he knew and recollections from culture
Humans engage similar processes. Research in participants and politics. But his ability to form such explicit memories
navigating virtual environments while their brains were postsurgery was practically nonexistent. Even so, certain types
scanned with functional magnetic resonance imaging shows of learning and memory remained untouched: he could still
that the hippo­­campus becomes active in ways consistent with learn some new skills with enough practice. But his recollec-
spatial planning, such as considering and planning routes. tions of new people, facts and events were immediately lost.

Illustrations by Jen Christiansen SCIENTIFICAMERICAN.COM  |  57

 Cognitive Cartography
Locations that Is Physical a
 nd S
 ocial
prompt place
The brain forms the idea of friend or foe by stitching together
Hippocampus cell firing
(home of diverse social characteristics from memories that track one’s
place cells) whereabouts. The recollections, research suggests, can then
Locations that be used to place an individual within a social hierarchy that
Entorhinal cortex prompt grid elucidates, say, where one stands in relation to others.
(home of grid cells) cell firing

PLACE AND GRID CELLS MAKING THE LEAP TO SOCIAL MAPS

Place cells pinpoint the animal’s whereabouts, each cell firing when Go right at the corner and continue to your destination. Building
a particular spot on a mental map is reached. A grid cell activates a map of physical surroundings is the work of place and grid
when an animal passes over the vertices of triangles superimposed cells. But the brain may also use these cells for constructing maps
on a mental map. The triangles’ pattern of activation helps the animal for social milieus: locating an acquaintance who grows closer but
compute the direction and distance traveled along a route. loses power in a relationship.

Physical Navigation Social Navigation

y More

Self
Power

Less
x Close Distant
Affiliation

From observing Molaison, neuroscientists discerned that decision-making. Mapping allows relations to be inferred, even
the hippo­campus was essential in forming the episodic memo- when they have not been experienced. It also allows for mental
ries that record facts and events. Research on the role of the shortcuts that go beyond the purview of the spatial and tempo-

Sources: “Scientific Background: The Brain’s Navigational Place and Grid Cell System,” by Ole Kiehn and Hans Forssberg, with illustrations by Mattias Karlen.
hippo­campus in episodic memory exploded, largely in parallel ral domains. In fact, reasoning using abstract concepts may
to studies on its maplike functions. depend on some of these same neural foundations.
The discoveries about the roles of the hippo­campus and ento- In one example of this new line of work, researchers Alexan-
rhinal cortex in spatial navigation and episodic memory were dra Constantinescu, Jill O’Reilly and Timothy Behrens, all then
significant for at least a couple of reasons. The work in spatial at the University of Oxford, asked participants to learn associa-
navigation in rodents marked the first time that a higher-order tions of different symbols with images of “stick” birds with var-

 ol. 100; October 24, 2018

cognitive function—something beyond basic sensory process- ious neck and leg lengths. A bird with a long neck but short
es—mapped onto clear neural correlates. H.M. showed us that legs, for example, might be linked with the image of a bell,
there were multiple types of memory supported by at least par- whereas a bird with a short neck and long legs might be con-
tially different neural systems, with the hippo­campus playing nected to a teddy bear. These linkages created a two-dimension-

Nobelprize.org; “Navigating Social Space,” by Matthew Schafer and Daniela Schiller, in N euron, V
a central role in the formation and storage of new episodic mem- al association space. Despite neuroimaging being too crude to
ories. These discoveries hinted that mechanisms of spatial and detect actual grid cells in the human brain, imaging conducted
temporal navigation might underlie episodic memory. This syn- during the learned-association testing nonetheless revealed a
thesis is perhaps best ex­­plained by the theoretical construct pro- gridlike pattern of activation within the entorhinal cortex.
posed decades earlier by Tolman; both episodic memory and This finding builds on earlier work by Christian Doeller of
spatial navigation might reflect the brain’s formation and use of the Max Planck Institute for Human Cognitive and Brain Sci-
cognitive maps. ence in Leipzig, Germany, and Neil Burgess of University Col-
Maps are not accurate portraits of the world in all of its com- lege London that first showed an entorhinal gridlike represen-
plexity. Rather they are representations of relations—distances tation in humans navigating a virtual maze. For both physical
and directions between locations and what exists where. Maps and abstract relations, the gridlike organization is highly effi-
reduce a dizzying amount of real-world information into a sim- cient. It makes the linkages of places or concepts more predict-
ple, easily readable format that is useful for effective, flexible able, enhancing how quickly inferences can be made about
navigation. The cell types mentioned earlier (place cells, grid these relations. As in physical space, this organization of infor-
cells and border cells, among others) may piece together such mation allows for inferring shortcuts—relations between ideas
related elements into a mental map, which other brain regions or perhaps analogies, stereotypes and even some aspects of cre-
can then read out to guide “navigation,” amounting to adaptive ativity itself could depend on such inferences.

58  |  SCIENTIFIC AMERICAN  |  SPECIAL EDITION  |  WINTER 2022

 PEOPLE MAPS dimensions of power and affiliation. In each interaction, we
T h e p r o g r e s s i o n from the physical to the abstract carries drew a line or vector from the participant to the character. In
over into the way the brain represents social relationships. Var- this way, we charted the evolving relations as trajectories
ious bits of knowledge about another person are distilled into through social space and computed information about the
the concept of that individual. When we see a photograph of angles and lengths of the social vectors.
someone or hear or see that person’s name, the same hippocam- We searched for neural signals that tracked this information
pal cells will fire, regardless of the sensory details of the stimu- by correlating a participant’s brain activity with the angle and
lus (for example, the famous “Jennifer Aniston neuron” de­­ length of the vectors for each decision. Activity in the hippo­
scribed by Itzhak Fried of the University of California, Los campus tracked the angle of the characters to the participant.
Angeles, and his colleagues). These hippo­campal cells are The degree to which hippocampal activity captured these social
responsible for representing concepts of specific individuals. coordinates also reflected the participants’ self-reported social
Other hippocampal cells track the physical locations of oth- skills. These findings suggest that the hippo­campus monitors
ers and are called social place cells. In an experiment by David social dynamics as it does physical locations by encoding rela-
Omer of the Hebrew University of Jerusalem, Nachum Ula- tions be­­tween points in multidimensional space. Indeed, it may
novsky of the Weizmann Institute of Science in Rehovot, Israel, be that along any arbitrary dimension in which we can order
and their colleagues, bats observed other bats navigating a sim- information, whether physical or abstract, the hippo­campus-
ple maze to reach a reward. The task of an observer bat was to entorhinal system plays a part.
simply watch and learn from a navigating bat, enabling it to Many questions about the brain’s social maps still remain
subsequently navigate the same route to get the same reward. un­­answered. How does this system interact with other brain
When the observer bat watched, hippocampal cells fired corre- regions? For example, in our role-playing study, we found that
sponding to the location of the other bat. the posterior cingulate cortex, a region also involved in repre-
Neural circuitry within specific subregions of the hippo­ senting spatial information, tracked the length of social vec-
campus (in particular, areas called CA1 and CA2) contributes to tors—functioning in effect as a measuring stick of “social dis-
such social memories. Artificial stimulation or inactivation of tance.” Further, a gridlike signal was found in brain regions that
these hippocampal areas enhances or diminishes an animal’s are interconnected with and tend to co-activate with the hippo­
ability to recognize other animals. In humans, hippocampal campal-entorhinal system, suggesting they form a network of
injury often spares memory for specific, individual faces, but brain regions with common functional properties.
the relation between this cardinal identifier of another person As research accumulates, questions of clinical importance
and that individual’s behavior may be lost. That observation arise as well. Can flawed mapping processes explain psychiatric
suggests that the hippo­campus does not simply record a face or dysfunction? Another possibility is that insights garnered from
some other personal detail but rather ties together diverse this brain architecture could inform artificial-intelligence
social characteristics. development. Well-organized internal models of the world
Hippocampal activity also tracks social hierarchies: the de­­ might be key to building more intelligent machines.
mands of a boss and a co-worker, for instance, may be valued That the same mapping system may underlie navigation
differently and confer different social standings. Common met- through space and time, reasoning, memory, imagination and
aphors illustrate the spatial dimensions of a hierarchy: a person even social dynamics suggests that our ability to construct
may try to gain status to “climb the social ladder” or “look models of the world might be what makes us such adaptive
down” at someone below them. Other factors are also critical. learners. The world is full of both physical and abstract rela-
Biological relatedness, common group goals, the remembered tions. Road maps of city streets and mental maps of interrelat-
history of favors and slights—all determine social proximity or ed concepts help us make sense of the world by extracting,
distance. Human relationships can be conceived of as geomet- organizing and storing related information. A new coffee shop
ric coordinates in social space that are defined by the dimen- on a familiar street can be easily placed within an existing spa-
sions of hierarchy and affiliation. tial map. Fresh concepts can be related to older ideas. And a
Work in our lab has explored these ideas in recent years. Our new acquaintance can reshape our social space.
results suggest that, as with other spaces, the hippo­campus Maps let us simulate possibilities and make predictions, all
organizes social information into a map­like format. To test this within the safety of our own heads. The mental shortcuts we
hypothesis, we put individuals in a choose-your-own-adventure can so readily conjure up might have their basis in the same
game in which they interacted with cartoon characters and system that allows us to figure out a detour around a traffic jam.
made decisions while their brains were scanned. We have just begun to discover the varied properties and capac-
In the game, players had just moved to a new town and ities of this system. Mental maps do more than help us find
needed to interact with the fictional characters to secure a job shortcuts through physical space—they enable us to navigate
and a place to stay. Participants made decisions on how to deal life itself. 
with a given character. Players could request that others per-
form favors to demonstrate their power, or they could submit to Matthew Schafer is pursuing a doctorate in neuroscience at the Icahn School of Medicine
demands made on them. In a subsequent interaction, they at Mount Sinai, focusing on the neural mechanisms of social cognition in the human brain.
could decide whether or not to make a gesture of attachment—
giving a hug or remaining at a distance. Daniela Schilleris an associate professor of both neuroscience and psychiatry at the
Using these decisions, we plotted each character at certain Icahn School of Medicine at Mount Sinai. She researches the neural mechanisms
coordinates on a map representing their movement along the underlying emotional control needed to adapt to constantly changing environments.

SCIENTIFICAMERICAN.COM  |  59
 FACE
VALUES
Brain regions that process faces reveal deep insights
into the neural mechanisms of vision
By Doris Y. Tsao
Illustration by Brian Stauffer

When I was in high school, I learned one day about the density
of curves in an introductory course on calculus. A simple pair of differ-
ential equations that model the interactions of predators and prey can
give rise to an infinite number of closed curves—picture concentric cir-
cles, one nested within another, like a bull’s-eye. What is more, the den-
sity of these curves varies depending on their location.
This last fact seemed so strange to me. I could easily imagine a finite
set of curves coming close together or pulling apart. But how could an
infinity of curves be denser in one region and less dense in another?
I  soon learned that there are different types of infinity with paradoxical
qualities, such as Hilbert’s Hotel (where the rooms are always fully
booked but new guests can always be accommodated) and the Banach-
Tarski apple (which can be split into five pieces and rearranged to make
two apples with the same volume as the original). I spent hours poring
over these mathematical proofs. Ultimately they struck me as symbolic
magic of no real consequence, but the seed of interest had taken root.

SCIENTIFICAMERICAN.COM  |  61
 Later, as an undergraduate at the California Institute to identify areas activated by the perception of three-
of Technology, I learned about the experiments of Da- dimensionality in images. I decided to show pictures of
vid Hubel and Torsten Wiesel and their landmark dis- faces and other objects to a monkey. When I compared
covery of how a region in the brain called the primary activation in the monkey’s brain in response to faces
visual cortex extracts edges from the images relayed with activation for other objects, I found several areas
from the eyes. I realized that what had mys­tified me back that lit up selectively for faces in the temporal lobe (the
in high school was the act of trying to i magine different area underneath the temple)—specifically in a region
densities of infinity. Unlike the mathematical tricks I called the inferotemporal (IT) cortex. Charles Gross, a
had studied in high school, the edges that Hubel and pioneer in the field of object vision, had discovered face-
Wiesel described are processed by neurons, so they ac- selective neurons in the IT cortex of macaques in the
tually exist in the brain. I came to recognize that visual early 1970s. But he had reported that these cells were
neuroscience was a way to understand how this neural randomly scattered throughout the IT cortex. Our fMRI
activity gives rise to the conscious perception of a curve. results provided the first indication that face cells might
The sense of excitement this realization triggered is be concentrated in defined regions.
hard to describe. I believe at each stage in life one has a
duty. And the duty of a college student is to dream, to FACE PATCHES
find the thing that captures one’s heart and seems worth A f ter publ ish ing my work , I was invited to give a
devoting a whole life to. Indeed, this is the single most talk describing the fMRI study as a candidate for a fac-
important step in science—to find the right problem. I ulty position at Caltech, but I was not offered the job.
was captivated by the challenge of understanding vision Many people were skeptical of the value of fMRI, which
and embarked on a quest to learn how patterns of elec- measures local blood flow, the brain’s plumbing. They
trical activity in the brain are able to encode perceptions argued that showing increased blood flow to a brain
area when a subject is looking at faces falls
far short of clarifying what neurons in the
FACE PATCHES DO ACT AS AN ASSEMBLY LINE area are actually encoding because the rela-
tion between blood flow and electrical activ-
TO SOLVE ONE OF THE BIG CHALLENGES OF VISION: ity is unclear. Perhaps by chance these face
HOW TO RECOGNIZE THINGS AROUND US patches simply contained a slightly larger
number of neurons responsive to faces, like
DESPITE CHANGES IN THE WAY THEY LOOK. icebergs randomly clustered at sea.
Because I had done the imaging experi-
ment in a monkey, I could directly address
of visual objects—not just lines and curves but even ob- this concern by inserting an electrode into an fMRI-iden-
jects as hard to define as faces. Accomplishing this ob- tified face area and asking, What images drive single
jective required pinpointing the specific brain regions neurons in this region most strongly? I performed this
dedicated to facial recognition and deciphering their un- experiment together with Winrich Freiwald, then a post-
derlying neural code—the means by which a pattern of doctoral fellow in Margaret Livingstone’s laboratory at
electrical impulses allows us to identify people around us. Harvard, where I was a graduate student. We presented
The journey of discovery began in graduate school at faces and other objects to a monkey while amplifying
Harvard University, where I studied stereopsis, the the electrical activity of individual neurons recorded by
mechanism by which depth perception arises from dif- the electrode. To monitor responses in real time, we con-
ferences between the images in the two eyes. One day I verted the neurons’ electrical signals to an audio signal
came across a paper by neuroscientist Nancy Kanwish- that we could hear with a loudspeaker in the lab.
er, now at the Massachusetts Institute of Technology, This experiment revealed an astonishing result: al-
and her colleagues, reporting the discovery of an area most every single cell in the area identified through
in the human brain that responded much more strong- fMRI was dedicated to processing faces. I can recall the
ly to pictures of faces than to images of any other object excitement of our first recording, hearing the “pop” of
when a person was inside a functional magnetic reso- cell after cell responding strongly to faces and very lit-
nance imaging (fMRI) brain scanner. The paper seemed tle to other objects. We sensed we were on to something
bizarre. I was used to the brain being made of parts with important, a piece of cortex that could reveal the brain’s
names like “basal ganglia” and “orbitofrontal cortex” high-level code for visual objects. Marge remarked on
that had some vague purpose one could only begin to the face patches: “You’ve found a golden egg.”
fathom. The concept of an area specifically devoted to I also remember feeling surprised during that first
processing faces seemed all too comprehensible and experiment. I had expected the face area would contain
therefore impossible. Anyone could make a reasonable cells that responded selectively to specific individuals,
conjecture about the function of a face area—it should analogous to orientation-selective cells in the primary
probably represent all the different faces that we know visual cortex that each respond to a specific edge orien-
and something about their expression and gender. tation. In fact, a number of well-publicized studies had
As a graduate student, I had used fMRI on monkeys suggested that single neurons can be remarkably selec-

62  |  SCIENTIFIC AMERICAN  |  SPECIAL EDITION  |  WINTER 2022

 tive for the faces of familiar people—
responding, say, only to Jennifer
Aniston. Contrary to my expecta- Where Are the Face Detectors?
tion, each cell seemed to fire vigor-
ously for almost any face. A set of six nodes in the inferotemporal cortex of both brain hemispheres specializes in iden-
I plugged madly away at Photo- tifying faces. These “face patches” function as an assembly line:
shop during these early experi- in the middle lateral and middle fundus patches, one
ments and found that the cells re- neuron might become active when faces look
sponded not just to faces of humans straight ahead; another might turn on for a
and monkeys but even to highly face looking to the right. At the end of the
simplified cartoon faces. assembly line, in the anterior medial patch,
Observing this phenomenon, I varying views are stitched together. Neu-
decided to create cartoon faces with rons in this patch are active in response
19 different features that seemed to the face of a specific individual, no Middle
pertinent to defining the identity of matter if the view is from the front or fundus
a face, including inter-eye distance, side. Re­sponses from a face patch of one patch
face aspect ratio and mouth height, monkey are generated for faces but not
among other characteristics. We objects (red areas in  A  ) and for the same ●
then went on to alter these values— in­­dividual, such as the dark-haired man, Middle
lateral Anterior
the inter-eye distance, for instance, from varying angles (red areas in  B ). ●
patch Inferotemporal cortex medial patch
varied from almost cyclopean to just
inside the face boundary. Individu- Neuron activity level: Low High
al cells responded to most faces but Each column represents neuron activity for an image like the one shown above it
interestingly did not always exhibit A B
the exact same rate of firing with all Faces Not faces Left profile Front Back
faces. Instead there was a systemat-
ic variation in their response: when
Each row of data represents a different
cell within the anterior medial patch

we plotted the firing of cells for the

different cartoon features, we found
a pattern in which there was a min-
imal response to one feature ex-
treme—the smallest inter-eye dis-
tance, for instance—and a maximal
response to the opposite extreme—
the largest eye separation—with in-
termediate responses to feature
From “Functional Compartmentalization and Viewpoint Generalization within the Macaque Face-Processing System,” by

values in the middle. The re­­­sponse

as a function of the value for each
feature looked like a ramp, a line
Winrich A. Freiwald and Doris Y. Tsao, in Science, Vol. 330; November 5, 2010 (data grids and photo insets)

slanted up or down.
Once again, I was invited to give a job talk at Caltech. termine which c ontrast relations can be used to recog-
Returning, I had more to offer than just fMRI images. nize a face: they should be the ones that are immune to
With the addition of the new results from single-cell re- changes in lighting. For example, “left eye darker than
cordings, it was clear to everyone that these face patch- nose” is a useful feature for detecting a face because it
es were real and likely played an important role in fa- does not matter if a face is photographed with lighting
cial recognition. Furthermore, understanding their un- from above, left, right or below: the left eye is always
derlying neural processes seemed like an effective way darker than the nose (check for yourself ).
to gain traction on the general problem of how the brain From a theoretical standpoint, this idea provides a
represents visual objects. This time I was offered the job. simple, elegant computational mechanism for facial
recognition, and we wondered whether face cells might
CONTRAST IS KEY be using it. When we measured the response of cells to
At Caltech, m  y colleagues and I dug deeper into the faces in which different regions varied in brightness, we
question of how these cells detect faces. We took inspi- found that cells often had a significant preference for a
ration from a paper by Pawan Sinha, a vision and com- particular contrast feature in an image.
putational neuroscientist at M.I.T., that suggested fac- To our astonishment, almost all the cells were whol-
es could be discerned on the basis of specific contrast ly consistent in their contrast preferences—just a sin-
relations between different regions of the face—wheth- gle cell was found that preferred the opposite polarity.
er the forehead region is brighter than the mouth re- Moreover, the preferred features were precisely those
gion, for example. Sinha suggested a clever way to de- identified by Sinha as being invulnerable to lighting

Illustration by Body Scientific (brain) SCIENTIFICAMERICAN.COM  |  63

 animal. Work by our group and oth-

Shape + Appearance = Face ers has found that a similar pattern

of multiple face patches spanning
Identifying the face patches was only a first step. It then became necessary to explore what hap- the IT cortex exists in humans and
pens in the neurons within each patch, setting off a search for the brain’s coding scheme for fac- other primates such as marmosets.
es. To derive quantitative measures for faces, the Tsao laboratory came up with 25 features for This observation of a stereotyped
shape and 25 for appearance that could be used by each neuron in a face patch—a 50-dimen- pattern suggested that the patches
sional face space. The shape features can be thought of as those defining the skeleton—how might constitute a kind of assembly
wide the head is or the distance between the eyes. The appearance features specify the face’s line for processing faces. If so, one
surface texture (complexion, eye or hair color, and so on). would expect the six patches to be
connected to one another and each
patch to serve a distinct function.
Shape: Described by the position (x,y coordinates) of feature landmarks (yellow dots) To explore the neural connec-
tions among patches, we electrical-
x
ly stimulated different patches with
tiny amounts of current—a tech-
nique called microstimulation—
while the monkey was inside an
y fMRI scanner. The goal was to find
out what other parts of the brain
light up when a particular face patch
is stimulated. We discovered that
Examples of variability Average shape whenever we stimulated one face
patch, the other patches would light
up, but the surrounding cortex
Appearance: Variations in luminosity of the image after first aligning it to match an average face shape would not, indicating that, indeed,
the face patches are strongly inter-
Luminosity range connected. Furthermore, we found
that each patch performs a different
function. We presented pictures of
25 people, each at eight different
head orientations, to monkeys and
recorded responses from cells in

From “The Code for Facial Identity in the Primate Brain,” by Le Chang and Doris Y. Tsao, in Cell, Vol. 169, No. 6; June 1, 2017 (f ace images)
three regions: the middle lateral and
middle fundus patches (ML/MF),
Examples of variability the anterior lateral patch (AL) and
the anterior medial patch (AM).
We found striking differences
among these three regions. In ML/
changes. The experiment thus confirmed that face cells MF, cells responded selectively to specific views. For ex-
use contrast relations to detect faces. ample, one cell might prefer faces looking straight
More broadly, the result confirmed that these cells ahead, whereas another might opt for faces looking to
truly were face cells. At talks, skeptics would ask, How the left. In AL, cells were less view-specific. One class of
do you know? You can’t test every possible stimulus. cells responded to faces looking up, down and straight
How can you be sure it’s a face cell and not a pomegran- ahead; another responded to faces looking to the left or
ate cell or a lawn mower cell? This result nailed it for right. In AM, cells responded to specific individuals re-
me. The precise match between the way cells reacted to gardless of whether the view of the face was frontal or
changes in contrast between different parts of the face in profile. Thus, at the end of the network in AM, view-
and Sinha’s computational prediction was uncanny. specific representations were successfully stitched into
Our initial experiments had revealed two nearby cor- a view-invariant one.
tical patches that lit up for faces. But after further scan- Apparently face patches do act as an assembly line
ning (with the help of a contrast agent that increased to solve one of the big challenges of vision: how to rec-
severalfold the robustness of the signal), it became clear ognize things around us despite changes in the way
that there are in fact six face patches in each of the they look. A car can have any make and color, appear
brain’s two hemispheres (making a dozen golden eggs at any viewing angle and distance, and be partially ob-
total). They are distributed along the entire length of the scured by closer objects such as trees or other cars. Rec-
temporal lobe. These six patches, moreover, are not ran- ognizing an object despite these visual transformations
domly scattered throughout the IT cortex. They are lo- is called the invariance problem, and it became clear
cated in similar locations across hemispheres in each to us that a major function of the face-patch network

64  |  SCIENTIFIC AMERICAN  |  SPECIAL EDITION  |  WINTER 2022 Graphics by Jen Christiansen
 is to overcome this impediment.
Given the great sensitivityo
cells in face patches to changes in
f
The Face Code, at Last
facial identity, one might expect Having 50 coordinates that describe shape and appearance allows for a description of neurons’
that altering these cells’ respons- firing in response to a particular face­—a description that functions as a code that can be visual-
es should modify an animal’s per- ized geometrically. In this code, each face cell receives inputs for a face in the form of the 50
ception of facial identity. Neuro- coordinates, or dimensions. The neuron then fires with a particular intensity in response to a
scientists Josef Parvizi and Kala- certain face (red outlines), along what is called the preferred axis. The intensity increases steadi-
nit Grill-Spector of Stanford ly (monotonically) along the preferred axis. Furthermore, the response is the same for every
University had electrically stimu- face on an axis at right angles to the preferred axis, even though those faces may look very
lated a face-patch area in human different. This axis model of facial coding differs from a previous exemplar model that suggests
subjects who had electrodes im- that each neuron fires with maximum intensity to a single most preferred face.
planted in their brains for the
purpose of identifying the source Axis Exemplar
Orthogonal axis
of epileptic seizures and found model model
that stimulation distorted the Preferred axis (new) (old)
subjects’ perception of a face.
We wondered whether we Spike in
would find the same effect in nerve activity
monkeys when we stimulated
their face patches. Would doing
so alter the perception only of
faces, or would it affect that of
other objects as well? The bound-
ary between a face and a nonface
object is fluid—one can see a face
in a cloud or an electrical outlet
if prompted. We wanted to use
electrical microstimulation as a
tool to delineate precisely what
constitutes a face for a face patch.
We trained monkeys to report
whether two sequentially pre-
sented faces were the same or dif-
From “The Code for Facial Identity in the Primate Brain,” by Le Chang and Doris Y. Tsao, in Cell, Vol. 169, No. 6; June 1, 2017 (f ace grid)

ferent. Consistent with the earli-

er results in humans, we found
that microstimulation of face
patches strongly distorted per-
ception so that the animal would
always signal two identical faces as being different. crack the neural code for facial identity. That means un-
Interestingly, microstimulation had no effect on the derstanding how individual neurons process faces at a
perception of many nonface objects, but it did signifi- level of detail that would let us predict a cell’s response
cantly affect responses to a few objects whose shape is to any given face or decode the identity of an arbitrary
consistent with a face—apples, for one. But why does face based only on neural activity.
this stimulation influence the perception of an apple? The central challenge was to figure out a way to de-
One possibility is that the face patches are typical- scribe faces quantitatively with high precision. Le
ly used to represent not just faces but also other round Chang, then a postdoc in my lab, had the brilliant in-
objects like apples. Another hypothesis is that face sight to adopt a technique from the field of computer
patches are not normally used to represent these ob- vision called the active appearance model. In this ap-
jects, but stimulation induces an apple to appear face- proach, a face has two sets of descriptors, one for shape
like. It remains unclear whether face patches are use- and another for appearance. Think of the shape fea-
ful for detecting any nonface objects. tures as those defined by the skeleton—how wide the
head is or the distance between the eyes. The appear-
CRACKING THE CODE ance features define the surface texture of the face
U n c o v e r i n g the organization of the face-patch sys- (complexion, eye or hair color, and so on).
tem and properties of the cells within was a major ac- To generate these shape and appearance descriptors
complishment. But my dream when we first began re- for faces, we started with a large database of face imag-
cording from face patches was to achieve something es. For each face, we placed a set of markers on key fea-
more. I had intuited that these cells would allow us to tures. The spatial locations of these markers described

SCIENTIFICAMERICAN.COM  |  65
 insight about why this is important.

Pictures Worth 205 Neurons If a face cell has ramp-shaped tuning

to different features, its response can
For a given face, w
 e can predict how a cell will respond by taking a weighted sum of all 50 face be roughly approximated by a sim-
coordinates. To predict what face the monkey saw from neuronal activity, this entire process can ple weighted sum of the facial fea-
be reversed: by knowing the response of 205 face cells, it is possible to predict the 50 coordinates tures, with weights determined by
defining the exact facial features—and make a highly accurate reconstruction of a given face. the slopes of the ramp-shaped tun-
ing functions. In other words:
Original Images from the Face Database
response of face cells = weight
matrix × 50 face features

We can then simply invert this

equation to convert it to a form that
lets us predict the face being shown
from face cell responses:
50 face features = (1/weight
matrix) × response of face cells

At first, this equation seemed im­­

possibly simple to us. To test it, we
used responses to all but one of the
2,000 faces to learn the weight ma-
trix and then tried to predict the 50
face features of the excluded face.
Astonishingly, the prediction turned
out to be almost indistinguishable
from the actual face.

A WIN-WIN BET
Corresponding Reconstructed Faces Based on Neuron Activity At a meeting in Ascona, Switzer-
land, I presented our findings on how
we could reconstruct faces using neu-
ral activity. After my talk, Rodrigo
the shape of the face. From these varied shapes, we cal- Quian Quiroga, who discovered the famous Jennifer Ani­
culated an average face. We then morphed each face im- ston cell in the human medial temporal lobe in 2005 and
age in the database so its key features exactly matched is now at the University of Leicester in England, asked
those of the average face. The resulting images consti- me how my cells related to his concept that single neu-
tuted the appearance of the faces independent of shape. rons react to the faces of specific people. The Jennifer An­
We then performed principal components analysis iston cell, also known as a grandmother cell, is a putative
independently on the shape and appearance descrip- type of neuron that switches on in response to the face of
tors across the entire set of faces. This is a mathemati- a recognizable person—a celebrity or a close relative.
cal technique that finds the dimensions that vary the I told Rodrigo I thought our cells could be the build-
most in a complex data set. ing blocks for his cells, without thinking very deeply
By taking the top 25 principal components for shape about how this would work. That night, sleepless from
and the top 25 for appearance, we created a 50-dimen- jet lag, I recognized a major difference between our face
sional face space. This space is similar to our familiar cells and his. I had described in my talk how our face cells
3-D space, but each point represents a face rather than computed their response to weighted sums of different
a spatial location, and it comprises much more than face features. In the middle of the night, I realized this
just three dimensions. For 3-D space, any point can be computation is the same as a mathematical operation
described by three coordinates (x,y,z). For a 50-D face known as the dot product, whose geometric representa-
space, any point can be described by 50 coordinates. tion is the projection of a vector onto an axis (like the sun
In our experiment, we randomly drew 2,000 faces and projecting the shadow of a flagpole onto the ground).
Doris Y. Tsao (f ace images)

presented them to a monkey while recording cells from Remembering my high school linear algebra, I real-
two face patches. We found that almost every cell showed ized this implied that we should be able to construct
graded responses—resembling a ramp slanting up or a  large “null space” of faces for each cell—a series of
down—to a subset of the 50 features, consistent with my faces of varying identity that lie on an axis perpendic-
earlier experiments with cartoon faces. But we had a new ular to the axis of projection. Moreover, all these faces

66  |  SCIENTIFIC AMERICAN  |  SPECIAL EDITION  |  WINTER 2022

would cause the cell to fire in exactly the same way. ject representation. It is possible that all of the IT cor-
And this, in turn, would suggest cells in face patch- tex might be organized along the same principles gov-
es are fundamentally different from grandmother cells. erning the face-patch system, with clusters of neurons
It would demolish the vague intuition everyone shared encoding different sets of axes to represent an object.
about face cells—that they should be tuned to spe- We are now conducting experiments to test this idea.
cific faces.
I was the first person in the meeting’s breakfast hall NEURAL ROSETTA STONE
at 5  a.m. the next morning and hoped to find Rodrigo so If you ever go t o the British Museum, you will see
I could tell him about this counterintuitive prediction. an amazing artifact, the Rosetta stone, on which the
Amazingly, when he finally showed up, he told me he had same decree of Memphis is engraved in three different
the exact same idea. So we made a bet, and Rodrigo al- languages: Egyptian hieroglyphics, Demotic and an-
lowed the terms to be framed in a way that would be win- cient Greek. Because philologists knew ancient Greek,
win for me. If each cell really turned out to have the same they could use the Rosetta stone to help decipher Egyp-
response to different faces, then I would send Rodrigo tian hieroglyphics and Demotic. Similarly, faces, face
an expensive bottle of wine. If, on the other hand, the patches and the IT cortex form a neural Rosetta stone—
prediction did not pan out, he would send me solace wine. one that is still being deciphered. By showing pictures
In search of an answer back in our lab at Caltech, Le of faces to monkeys, we discovered face patches and
Chang first mapped the preferred axis for a given cell learned how cells within these patches detect and iden-
using responses to the 2,000 faces. Then he generated, tify faces. In turn, understanding coding principles in
while still recording from the same cell, a range of fac- the face-patch network may one day lead to insight into
es that could all be placed on an axis perpendicular to the organization of the entire IT cortex, revealing the se-
the cell’s preferred axis. Remarkably, all these faces elic- cret to how object identity more generally is encoded.
ited exactly the same response in the cell. The next week Perhaps the IT cortex contains additional networks spe-
Rodrigo received an exquisite bottle of Cabernet. cialized for processing other types of objects—a whir-
The finding proved that face cells are not encoding ring factory with multiple assembly lines.
the identities of specific individuals in
the IT cortex. Instead they are perform-
ing an axis projection, a much more ab-
stract computation.
UNDERSTANDING CODING PRINCIPLES IN THE FACE-
An analogy can be made to color. Col- PATCH NETWORK MAY ONE DAY LEAD TO INSIGHT
ors can be coded by specific names, such
as periwinkle, celandine and azure. Alter- INTO THE ORGANIZATION OF THE ENTIRE INFERO­
natively, one can code colors by particular
combinations of three simple numbers
TEMPORAL CORTEX, REVEALING THE SECRET TO HOW
that represent the amount of red, green OBJECT IDENTITY MORE GENERALLY IS ENCODED.
and blue that make up that color. In the lat-
ter scheme, a color cell performing a pro-
jection onto the red axis would fire electrical impulses, or I also hope that knowing the code for facial identity
spikes, proportional to the amount of red in any color. can help fulfill my college dream of discovering how we
Such a cell would fire at the same intensity for a brown imagine curves. Now that we understand face patches,
or yellow color containing the same amount of red mixed we can begin to train animals to imagine faces and ex-
in with other colors. Face cells use the same scheme, but plore how neural activity is shaped by the purely inter-
instead of just three axes, there are 50. And instead of nal act of imagination. Lots of new questions arise. Does
each axis coding the amount of red, green or blue, each imagination reactivate the code for the imagined face
axis codes the amount of deviation of the shape or ap- in the face patches? Does it bring back even earlier rep-
pearance of any given face from an average face. resentations of contours and shading that provide in-
It would seem then that the Jennifer Aniston cells puts to the face-patch system? We now have the tools
do not exist, at least not in the IT cortex. But single neu- to probe these questions and better understand how
rons responding selectively to specific familiar individ- the brain sees objects, imagined or real.
uals may still be at work in a part of the brain that pro- Because almost all the brain’s core behaviors­—con-
cesses the output of face cells. Memory storage regions— sciousness, visual memory, decision-making, lan­­guage—
the hippocampus and surrounding areas—may contain require object interactions, a deep understanding of ob-
cells that help a person recognize someone from past ject perception will help us gain insight into the entire
experience, akin to the famed grandmother cells. brain, not just the visual cortex. We are only starting to
Facial recognition in the IT cortex thus rests on a set solve the enigma of the face. 
of about 50 numbers in total that represent the mea-
surement of a face along a set of axes. And the discov- Doris Y. Tsao is a professor of biology at the California In­­stitute of Technol­ogy
ery of this extremely simple code for face identity has and an investigator of the Howard Hughes Medical Institute. In 2018 she was
major implications for our understanding of visual ob- named a MacArthur Fellow.

SCIENTIFICAMERICAN.COM  |  67
 RADICAL
CHANGE Uncertainty in the world threatens our sense of self.
To cope, people embrace populism
By Michael A. Hogg
Illustration by Wesley Allsbrook

Human societies are constantly rearranging themselves, causing profound

disruptions in our social lives. The industrial revolution of the late 18th and early 19th
centuries fragmented communities as people moved for work, the decay of empires in
the early 20th century reconfigured nations and national identities, and the Great
Depression of the 1930s shattered people’s economic security and future prospects. But
we are now in what is perhaps a time of unprecedented uncertainty. The early 21st cen-
tury is characterized by rapid and overwhelming change: globalization, immigration,
technological revolution, unlimited access to information, sociopolitical volatility, the
automation of work and a warming climate.

SCIENTIFICAMERICAN.COM  |  69
 ferent from them, they seek homogeneity
and become intoxicated by the freedom to
access only information that confirms
who they are or who they would like to be.
As a result, global populism is on the rise.

SEEKING SOCIAL IDENTITY

One powerful source o  f identity re­­
sides in social groups. They can be highly
effective at reducing a person’s self-un­­
certainty—particularly if such groups are
distinctive and have members who share
a sense of interdependence.
Groups play this central role in an­­
choring who we are because they are so­­
cial categories, and research shows that
social categorization is ubiquitous. A per-
son categorizes others as either “in-group”
or “out-group” members. They assign the
group’s attributes and social standing to
those others, thereby constructing a sub-
jective world where groups are internally
homo­ geneous and the differences be­­
tween groups are exaggerated and polar-
ized in an ethnocentric manner. And be­­
cause we also categorize ourselves, we in­­
ternalize shared in-group-defining attri-
butes as part of who we are. To build
Social groups People need to have a firm sense of identity and of social identity, we psychologically surround ourselves
soothe individ­ their place in the world, and for many the pace and with those who are like us.
ual self-uncer­ magnitude of such change can be alienating. This is This psychological process that causes people to
tain­ty but also because our sense of self is a fundamental organizing identify with groups and behave as group members is
delineate “in-­ principle for our own perceptions, feelings, attitudes called social categorization. It anchors and crystalliz-
groups” and and actions. Typically it is anchored in our close inter- es our sense of self by assigning us an identity that
“out-groups,” personal relationships, such as with our friends, fami- prescribes how we should behave, what we should
which can in­­ ly and partners, and in the variety of social groups and think and how we should make sense of the world. It
crease populism. categories that we belong to and identify with—our also makes interaction predictable, allows us to antic-
nationality, religion, ethnicity, profession. It allows us ipate how people will treat and think about us, and
to predict with some confidence how others will view furnishes consensual identity confirmation: people
us and treat us. like us—the in-group members—validate who we are.
Imagine navigating all the situations and people This self-uncertainty social-identity dynamic is not
we encounter in day-to-day life while continually feel- in itself a bad thing. It enables the collective organiza-
ing uncertain about who we are, how to behave and tion that lies at the heart of human society. Human
how social interactions will unfold. We would feel dis- achievements that require the coordination of many
orientated, anxious, stressed, cognitively depleted, in the service of common goals cannot be achieved
and lacking agency and control. This self-uncertainty by individuals on their own. Yet this dynamic becomes
can, in fact, be experienced as an exciting challenge if a problem when the sense of self-uncertainty and
we feel we have the material, social and psychological identity threat is acute, enduring and all-encompass-
resources to resolve it. If we feel we do not have these ing. People then experience an overwhelming need for
resources, however, it can be experienced as a highly identity—and not just any identities but ones that are
aversive threat to us and our place in the world. well equipped to resolve those disorienting, even
Generally, self-uncertainty is a sensation that peo- scary, feelings.
ple are motivated to reduce. When people are increas-
ingly unsure about who they are and how they fit into REDUCING UNCERTAINTY
this rapidly changing landscape, it can be—and indeed THROUGH GROUP MEMBERSHIP
Pyrosky/Getty Images

has become—a real problem for society. People are sup- S o m e f e at u r e s of groups and social identities are
porting and enabling authoritarian leaders, flocking to especially well suited to reducing self-uncertainty.
ideologies and worldviews that promote and celebrate Most important, groups need to be polarized from
the myth of a glorious past. Fearful of those who are dif- other groups and have unambiguous boundaries that

70  |  SCIENTIFIC AMERICAN  |  SPECIAL EDITION  |  WINTER 2022

distinguish between those who are “in” and those who shown to prefer leaders who are assertive and author-
are “out.” Internally they need to be clearly structured, itarian, even autocratic, and who deliver a simple,
typically in a hierarchical way. These features make black-and-white, affirmational message about “who
the group cohesive and homogeneous, such that mem- we are” rather than a more open, nuanced and textured
bers are interdependent and of one mind in sharing a identity message.
common fate. Perhaps most troubling is that self-uncertainty can
Diversity and dissent reinstate uncertainty and are enable and build support for leaders who possess the
therefore avoided. When these facets do occur, individ- so-called Dark Triad personality attributes: Machia-
uals and the group as a whole react decisively and vellianism, narcissism and psychopathy. Self-uncer-
harshly, creating an atmosphere of suspicion that lays tainty, in other words, seems to fuel populism.
the ground for persecution of alleged deviants. It Another source of identity information is “people
breeds an opportunity for personal dislikes and vendet- like you” who you feel embody the group’s identity
tas to escalate under the guise of protecting cohesion. and see the world in the same way as you do. These
That members are accepted and trusted fully is can be people with whom you interact face-to-face or
important not only for the group but also for the as friends, or they can be sources of information such
members themselves. After all, they desperately want as radio and television channels, particularly news
to be included so that their identity is validated and outlets, that you watch. But nowadays these sources
their uncertainty thus reduced. Prospective and new are overwhelmingly information and influence nodes
members—and those who suspect they are viewed with on the Internet, such as Web sites, social media, Twit-
suspicion or are uncertain about whether they are fully ter feeds, podcasts, and so forth.
accepted—will go to extremes on behalf of the group to The Web is an ideal place to decrease the discom-
prove their membership credentials and loyalty. These fort of self-uncertainty because it provides nonstop
individuals are vulnerable to zealotry and radicaliza- access to unlimited information that is often cherry-
tion. Neo-Nazis and white supremacists
who publicly engage in violent acts of
terrorism and racial hatred are one ex­­ WHEN PEOPLE ARE INCREASINGLY UNSURE ABOUT
amp­le of this extremism.
The social identity embodied by such
WHO THEY ARE AND HOW THEY FIT INTO THIS RAPIDLY
groups also needs to be uncomplicated CHANGING LANDSCAPE, IT CAN BE—AND INDEED
so that it can be taken at face value as
“the truth.” Subtlety and nuance are
HAS BECOME—A REAL PROBLEM FOR SOCIETY.
anathema because they are an impedi-
ment to uncertainty reduction. Clarity on where the picked by individuals themselves and algorithms that
group stands allows its members to know how they do it discreetly. Therefore, people are accessing only
should think and feel—as well as behave. Such identi- identity-confirming information. Confirmation bias, a
ties are bolstered by having a strong ideology that powerful and universal human bias that is especially
identifies distasteful and morally bankrupt out-groups strong under uncertainty, separates information and
who can be demonized and cast in the role of “enemy.” identity universes that fragment and polarize society.
Conspiracy theories thrive in this environment Online, people can easily seek out groups that may not
because they establish these out-groups as agents of be readily available in their physical lives.
historical victimization by the in-group. The Internet further empowers confirmation bias
under uncertainty because people want to be sur-
HOW UNCERTAINTY rounded by those who think alike so that their identi-
BREEDS POPULISM ties and worldview are continuously confirmed. The
I f s e l f - u n c e r ta i n t y m
 otivates people to identify contours of “truth” then get mapped onto these self-
with a group and internalize that identity as a key part contained social-identity universes. In this scenario,
of who they are, they need to be confident that they there are no absolute truths and no motivation to seri-
know exactly what their group’s identity is. When you ously explore and incorporate alternative viewpoints
need what y  ou consider to be reliable and trusted be­­cause that would be kryptonite to social identity’s
sources of identity information, where do you turn? power to reduce self-uncertainty. This dynamic helps
The first port of call is whoever you believe is consensu- to explain why people dwell in increasingly homoge-
ally viewed by the group as its leadership—typically neous echo chambers that confirm their identity. 
someone whose leadership position is also formalized.
Recent research on how self-uncertainty affects Michael A. Hogg is a professor and chair of social psychology at Claremont
the type of leaders that individuals prefer paints a Graduate University and an honorary professor at the University of Kent
potentially alarming picture. People just need some- in England. He is a former president of the Society of Experimental Social
one to tell them what to do—and ideally those direc- Psychology, an editor in chief of the journal G
 roup Processes & Intergroup
tives are coming from someone whom they can trust Relations, a nd a fellow of numerous societies, including the Association for
as “one of us.” Self-uncertain people have also been Psychological Science.

SCIENTIFICAMERICAN.COM  |  71
 Unconscious
Mind
The Intention Machine page 74

Infectious Dreams page 82

Sleep Learning Gets Real page 88

Answering Queries
in Real Time while Dreaming page 94

Can Science Illuminate

Our Inner Dark Matter? page 96

Illustration by Maria Corte

72  |  SCIENTIFIC AMERICAN  |  SPECIAL EDITION  |  WINTER 2022

 THE
INTENTION
MACHINE
A new generation of brain-machine
interfaces can deduce what a person wants
By Richard A. Andersen

Illustration by Mark Ross

SCIENTIFICAMERICAN.COM  |  75
I g e t g o o s e b u m p s e v e ry t im e I s e e i t. A pa ra ly z e d v o lu n t e e r s i t s i n
a  wheelchair while controlling a computer or robotic limb just with his or her
thoughts—a demonstration of a brain-machine interface (BMI) in action.
That happened in my laboratory in 2013, when Erik Sorto, a victim of a gunshot
wound when he was 21 years old, used his thoughts alone to drink a beer without
help for the first time in more than 10 years. The BMI sent a neural message from a
high-level cortical area. An electromechanical appendage was then able to reach out
and grasp the bottle, raising it to Sorto’s lips before he took a sip. His drink came a year after
surgery to implant electrodes in his brain to control signals that govern the thoughts that trig-
ger motor movement. My lab colleagues and I watched in wonderment as he completed this
deceptively simple task that is, in reality, intricately complex.
Witnessing such a feat immediately raises the question of
how mere thoughts can control a mechanical prosthesis. We
move our limbs unthinkingly every day—and completing these
motions with ease is the goal of any sophisticated BMI. Neuro-
scientists, though, have tried for decades to decode neural sig-
age activity over centimeters of brain tissue, capturing the
activity of many millions of neurons rather than that from indi-
vidual neurons in a single circuit. Functional magnetic reso-
nance imaging (fMRI) is an indirect measurement that records
an increase in blood flow to an active region. It can image
nals that initiate movements to reach out and grab objects. Lim- smaller areas than EEG, but its resolution is still rather low.
ited success in reading these signals has spurred a search for new Changes in blood flow are slow, so fMRI cannot distinguish
ways to tap into the cacophony of electrical activity resonating as rapid changes in brain activity.
the brain’s 86 billion neurons communicate. A new generation of To overcome these limitations, ideally one would like to
BMIs now holds the promise of creating a seamless tie between record the activity of individual neurons. Observing changes in
brain and prosthesis by tapping with great precision into the the firing rate of large numbers of single neurons can provide
neural regions that formulate actions—whether the desired goal the most complete picture of what is happening in a specific
is grasping a cup or taking a step. brain region. In recent years arrays of tiny electrodes implanted
in the brain have begun to make this type of recording possible.
FROM BRAIN TO ROBOT The arrays now in use are four-by-four-millimeter flat surfaces
A B M I o p e r at e s by sending and receiving—“writing” and with 100 electrodes. Each electrode, measuring one  to 1.5  milli-
“reading”—messages to and from the brain. There are two major meters long, sticks out of the flat surface. The entire array,
classes of the interface technology. A “write-in” BMI generally which resembles a bed of nails, can record activity from 100 to
uses electrical stimulation to transmit a signal to neural tissue. 200 neurons.
Successful clinical applications of this technology are already The signals recorded by these electrodes move to “decoders”
in use. The cochlear prosthesis stimulates the auditory nerve to that use mathematical algorithms to translate varied patterns
enable deaf subjects to hear. Deep-brain stimulation of an area of single-neuron firing into a signal that initiates a particular
that controls motor activity, the basal ganglia, treats motor dis- movement, such as control of a robotic limb or a computer.
orders such as Parkinson’s disease and essential tremor. These read-out BMIs will assist patients who have sustained
Devices that stimulate the retina are currently in clinical trials brain in­­jury because of spinal cord lesions, stroke, multiple
to alleviate certain forms of blindness. sclerosis, amyotrophic lateral sclerosis and Duchenne muscu-
“Read-out” BMIs, in contrast, record neural activity and are lar dystrophy.
still at a developmental stage. The unique challenges of reading Our lab has concentrated on people with tetraplegia, who
neural signals need to be ad­­dress­ed before this next-generation are unable to move either their upper or lower limbs because of
technology reaches patients. Coarse read-out techniques al­­ upper spinal cord injuries. We make recordings from the cere-
ready exist. The electroencephalogram (EEG) records the aver- bral cortex, the approximately three-millimeter-thick surface of

76  |  SCIENTIFIC AMERICAN  |  SPECIAL EDITION  |  WINTER 2022

 the brain’s two large hemispheres. If spread flat, the
cortex of each hemisphere would measure about
80,000 square millimeters. The number of cortical
regions that specialize in controlling specific brain
functions has grown as more data have been collected
and is now estimated to encompass more than 180
areas. These locations process sensory information,
communicate to other brain regions involved with
cognition, make decisions or send commands to trig-
ger an action.
In short, a brain-machine interface can interact
with many areas of the cortex. Among them are the
pri­mary cortical areas, which detect sensory inputs,
such as the angle and intensity of light impinging on
the retina or the sensation triggered in a peripheral
nerve ending. Also targeted are the densely connected
association cortices between the primary areas that
are specialized for language, object recognition, emo-
tion and executive control of decision-making.
A handful of groups have begun to record popula-
tions of single neurons in people who are paralyzed,
allowing them to operate a prosthesis in the controlled
setting of a lab. Major hurdles still persist before a
patient can be outfitted with a neural prosthetic device
as easily as a heart pacemaker. My group is pursuing re­­
cord­ings from the association areas instead of the
motor cortex targeted by other labs. Doing so, we hope,
may provide greater speed and versatility in sensing the
firing of neural signals that convey one’s intentions.
The specific association area my lab has studied is
the posterior parietal cortex (PPC), where plans to ini-
tiate movements begin. In our work with nonhuman
primates, we found one subarea of the PPC, called the
lateral intraparietal cortex, that discerns intentions
to begin eye movements. Limb-movement processing
occurs elsewhere in the PPC. The parietal reach re­­
gion prepares arm movements. Also, Hideo Sakata,
then at the Nihon University School of Medicine in
Japan, and his colleagues found that the anterior
intraparietal area formulates grasping movements.
Recordings from nonhuman primates indicate that
the PPC provides several possible advantages for brain INTERFACE TECHNOLOGY, developed by Richard A. Ander­sen (left) and
control of robotics or a computer cursor. It controls his Caltech team, enabled Erik Sorto (right) to move a robotic arm.
both arms, whereas the motor cortex in each hemi-
sphere, the area targeted by other labs, activates pri-
marily the limb on the opposite side of the body. The PPC also arrays we planned to use in humans into healthy nonhuman
indicates the goal of a movement. When a nonhuman primate, primates. The monkeys then learned to control computer cur-
for instance, is visually cued to reach for an object, this brain sors or robotic limbs.
area switches on immediately, flagging the location of a desired We built a team of scientists, clinicians and rehabilitation
Lance Hayashida and California Institute of Technology

object. In contrast, the motor cortex sends a signal for the path professionals from the California Institute of Technology, the
the reaching movement should take. Knowing the goal of an University of Southern California, the University of California,
intended motor action lets the BMI decode it quickly, within a Los Angeles, the Rancho Los Amigos National Rehabilitation
couple of hundred milliseconds, whereas figuring out the trajec- Center, and Casa Colina Hospital and Centers for Healthcare.
tory signal from the motor cortex can take more than a second. The team received a go-ahead from the Food and Drug Admin-
istration and institutional review boards charged with judging
FROM LAB TO PATIENT the safety and ethics of the procedure in the labs, hospitals and
It was not easy t o go from experiments in lab animals to studies rehabilitation clinics involved.
of the PPC in humans. Fifteen years elapsed before we made A volunteer in this type of project is a true pioneer because
the first human implant. First, we inserted the same electrode he or she may or may not benefit. Participants ultimately join

SCIENTIFICAMERICAN.COM  |  77
to help users of the technology who will seek it out once it is
perfected for everyday use. The implant surgery for Sorto, our
first volunteer, took place in April 2013 and was performed by
neuro­surgeons Charles Liu and Brian Lee. The procedure
By Thought Alone
went flawlessly, but then came the wait for healing before we For 15 years neuroscientists have built brain-machine inter­
could test the device. faces (BMIs) that allow neural signals to move computer
My colleagues at nasa’s Jet Propulsion Laboratory, which cursors or operate prostheses. The technology has moved
built and launched the Mars rovers, talk about the seven min- forward slowly because translating the electrical firing of
utes of terror when a rover enters the planet’s atmosphere neurons into commands to play a video game or move a robot
before it lands. For me it was two weeks of trepidation, wonder- arm involves highly intricate processes.
ing whether the implant would work. We knew in nonhuman A group at the California Institute of Technology has tried
primates how similar areas of the brain functioned, but a to advance the neuroprosthetic field by tapping into high-level
human implant was testing uncharted waters. No one had ever neural processing—the intent to initiate an action—and then
tried to record from a population of human PPC neurons before. conveying the relevant electrical signals to a robotic arm.
During the first day of testing we detected neural activity, Instead of sending out signals from the motor cortex to move an
and by the end of the week there were signals from enough arm, as attempted by other laboratories, the Caltech researchers
neurons to begin to determine if Sorto could control a robot place electrodes in the posterior parietal cortex (PPC), which
limb. Some of the neurons varied their activity when Sorto trans­mits to a prosthesis the brain’s intent to act.
imagined rotating his hand. His first task consisted of turn- Decoding neural signals remains a challenge for neuro­
ing the robot hand to different orientations to shake hands scientists. But using BMI signals from the posterior parietal
with a graduate student. He was thrilled, as were we, be­­cause cortex, the top of the cognitive command chain, appears to
this accomplishment marked the first time since his injury he result in faster, more versatile control of prosthetic technology.
could interact with the world using the bodily movement of a
robotic arm.
People often ask how long it takes to learn to use a BMI. In
fact, the technology worked right out of the box. It was intui-
tive and easy to use the brain’s intention signals to control the
robotic arm. By imagining different actions, Sorto could
watch recordings of individual neurons from his cortex and
turn them on and off at will.
We ask participants at the beginning of a study what they
would like to achieve by controlling a robot. For Sorto, he
wanted to be able to drink a beer on his own rather than ask-
ing someone else for help. He was able to master this feat
about one year into the study. With the team co-led by re­­
search scientist Spencer Kellis of Caltech, which in­­clud­ed
roboticists from the Applied Physics Laboratory at Johns
Hopkins University, we melded Sorto’s intention signals with
the processing power furnished by machine vision and smart
robotic technology.
The vision algorithm analyzes inputs from video cameras,
and the smart robot combines the intent signal with com-
puter algorithms to initiate the movement of the robot arm.
Sorto achieved this goal after a year’s time with cheers and
shouts of joy from everyone present. In 2015 we published in
Science o ur first results on using intention signals from the
PPC to control neural prostheses.
Sorto is not the only user of our technology. Nancy Smith,
now in her fourth year in the study, be­­came tetraplegic from
an automobile accident about 10 years ago. She had been a
high school teacher of computer graphics and played piano
as a pastime. In our studies with lead team members Tyson
Aflalo of Caltech and Nader Pouratian of U.C.L.A., we found a
detailed representation of the individual digits of both hands
in Smith’s PPC. Using virtual reality, she could imagine and
move 10 fingers individually on left and right “avatar” hands
displayed on a computer screen. Using the imagined move-
ment of five fingers from one hand, Smith could play simple
melodies on a computer-generated piano keyboard.

78  |  SCIENTIFIC AMERICAN  |  SPECIAL EDITION  |  WINTER 2022 Illustration by AXS Biomedical Animation Studio
 The Andersen laboratory at Caltech has pursued develop­ment of BMIs that
“read out“ brain signals of an intent to take an action and send them to a robotic
arm that can pick up a glass and allow a tetraplegic patient to drink (1–6). The
BMI provides touch and limb-positioning feedback—“write-in” signals—to the
ARRAY 3 somatosensory cortex that simu­lates tactile sensations and allows for fine-level
Electrode adjustments to the prosthesis (6–9). The researchers are currently integrating
arrays read out the
intended movements
2 read-out and write-in capabilities to achieve a fully bidirectional BMI.

from the activity INTENTION

of PPC neurons. The PPC forms movement intentions that
Premotor areas
normally go to the premotor and then
the motor cortex. But with spinal cord Primary motor cortex (hand area)
injury, the motor cortex becomes
disconnected from
the muscles of the
body below
the injury.
Reach

Grasp (hand shape)

Saccade (rapid eye
movements)
C
PP

1
Primary somatosensory
INPUT cortex (hand area)
Signals from sensory and
memory areas of the
cerebral cortex all converge
on the PPC. 9
Primary Episodic memory ARRAY
visual cortex Electrical stimulation in
the somatosensory cortex
produces the sensations
of touch and position from
the robot hand.

4 NEURAL SIGNAL
PROCESSOR
Electronics decode the CONTROL 5 8 STIMULATOR
intention signals quickly COMPUTER The stimulator generates
and formulate commands The commands can be small electric currents
for the robotic arm. coupled with video or to the electrodes of
eye-movement signals
7 CONTROL the stimulation array.
to increase the precision COMPUTER
of the command. Sensors on the robot
fingers and hand
detect position and
touch data, which are
sent to a stimulator.

6
ACTION
The electronically processed brain
signals prod the prosthesis to pick
up a glass, bring it to the lips and hold
it steady, allowing a sip to be taken.

SCIENTIFICAMERICAN.COM  |  79
 HOW THE BRAIN REPRESENTS GOALS works, their middle layers developed gain fields, just as was the
W e w e r e t h r i l l e d in working with these patients to find case in the PPC experiments. By mixing signals for visual inputs
neurons tuned to processing signals related to one’s intentions. and eye positions within the same neurons, as few as nine neu-
The amount of information to be gleaned from just a few hun- rons could represent the entire visual field.
dred neurons turned out to be overwhelming. We could decode Recently this idea of mixed representations—populations of
a range of cognitive activity, including mental strategizing neurons responding to multiple variables (as with the gain
(imagined versus attempted motion), finger movements, deci- fields)—has attracted renewed attention. For instance, record-
sions about recalling visual stimuli, hand postures for grasping, ings from the prefrontal cortex show a mixing of two types of
ob­­served actions, action verbs such as “grasp” or “push,” and memory task and different visual objects.
visual and somatosensory perception. To our surprise, inserting This work, moreover, may have a direct bearing in explain-
a few tiny electrode arrays en­­abled us to decode much of what ing what is happening in the PPC. We discovered this when we
a person intends to do, as well as the sensory inputs that lead to asked Smith, using a set of written instructions, to perform
the formation of intentions. eight different combinations of a task. One of her undertakings
The question of how much information can be re­­cord­ed from required strategizing to imagine or attempt an action. Another
a small patch of brain tissue reminded me of a similar scientific necessitated using the right and left side of the body; a third
problem that I had encountered early in my career. During my entailed squeezing a hand or shrugging a shoulder. We found
postdoctoral training with the late Vernon Mountcastle at the that PPC neurons mixed all these variables—and the intermin-
Johns Hop­kins University School of Medicine, we examined how gling exhibited a specific pattern, unlike the random interac-
tions we and others had reported in
lab animal experiments.

INSERTING A FEW TINY ELECTRODE

Activity of populations of neurons
for strategizing and for controlling
ARRAYS INTO THE BRAIN ENABLED each body side tends to overlap. If a
neuron fires to initiate the movement
US TO DECODE MUCH OF WHAT of the left hand, it will most likely also

A PERSON INTENDS TO DO. respond for an attempted righthand

movement, whereas neuron groups
that control the shoulder and hand are
more separated. We refer to this type
visual space is represented in the PPC of monkeys. Our eyes are of representation as partially mixed selectivity. We have since
like cameras, with the photosensitive retinas signaling the loca- found similarities in partially mixed representations that seem
tion of visual stimuli imaged on them—the entire image is re­­ to make up a semantics of movement. The activity of cells tuned
ferred to as a retinotopic map. Neurons respond to limited for the same action type tends to overlap. A neuron that re­­
regions of the retina, referred to as their receptive fields. In other sponds to videos of a person grasping an object will also likely
ways, processing visual perception is different than a video cam- become active when a person reads the word “grasp.” But cells
era recording. When a video camera moves around, the recorded responding to an action such as pushing tend to get separated
image also shifts, but when we move our eyes the world seems into their own group. In general, partially mixed coding ap­­
stable. The retinotopic image coming from the eyes must be con- pears to underlie computations that are similar (movements of
verted into a visual representation of space that takes into the left hand are similar to those of the right). It also separates
account where the eyes are looking so that as they move, the those that exhibit varying forms of neural processing (move-
world does not appear as if it were sliding around. ment of the shoulder differs from movement of the hand).
The PPC is a key processing center for high-order visual Mixed and partially mixed coding have been found in cer-
space representation. For a person to reach and grab an object, tain parts of the association cortex—and new studies must
the brain needs to take into account where the eyes are looking explore whether they appear in other lo­­ca­tions that govern lan-
to pick it up. PPC lesions in humans produce inaccurate reach- guage, object recognition and executive control. Additionally,
ing. In Mountcastle’s lab, we found individual PPC neurons had we would like to know whether the primary sensory or motor
receptive fields that registered parts of a scene. The same cells cortical re­­gions use a similar partially mixed structure.
also carried eye-position information. The two signals inter- Current studies indicate that, at least in the somatosensory
acted by multiplying the visual response by the position of the cortex, neurons do not respond to visual stimuli or the inten-
eyes in the head—the product of which is called a gain field. tion to make a movement but do respond to somatosensory
I continued to pursue this problem of understanding the stimuli and to the imagined execution of movements. Thus,
brain’s representation of space when I took my first faculty there is direct evidence that variables seen in the human PPC
position at the Salk Institute for Biological Studies, right across are not found in the primary somatosensory cortex, although it
the street from the University of California, San Diego. Working is still possible that partially mixed selectivity may exist in both
with David Zipser, a U.C.S.D. theoretical neuroscientist devel- areas but for different sets of variables.
oping neural networks, we reported in N  ature o
 n a computa- Another near-future goal is to find out how much learning
tional model of a neural network that combined retinotopic new tasks can affect the performance of the volunteers using
locations with gaze direction to make maps of space that are the prosthesis. If learning readily takes place, any area of the
invariant to eye movements. During training of the neural net- brain might then be im­­plant­ed and trained for any conceivable

80  |  SCIENTIFIC AMERICAN  |  SPECIAL EDITION  |  WINTER 2022

BMI task. For instance, an implant in the primary visual cortex FUTURE CHALLENGES
could learn to control motor tasks. But if learning is more a ma jo r f u t u r e c h a l l e n g e is to develop better electrodes
restricted, an implant would be needed in a motor area to per- for sending and receiving neural signals. We have found that
form motor tasks. Early results suggest this latter possibility, current implants continue to function for a relatively lengthy
and an implant may have to be placed in the area that has been five years. But better electrodes would ideally push the longev-
previously identified as controlling particular neural functions. ity of these systems even further and increase the number of
neurons that can be recorded from them. Another priority—an
WRITING IN SENSATIONS increase in the lengths of the electrodes’ tiny spikes—would
A BMI m ust do m ore t han just receive and process brain sig- help access areas located within folds of the cortex.
nals—it must also send feedback from a prosthesis to the brain. Flexible electrodes, which move with the slight jostling of
When we reach to pick up an object, visual feedback helps to the brain—from changes in blood pressure or the routine
direct the hand to the target. The positioning of the hand breathing cycle—will also allow for more stable recordings.
depends on the shape of the object to be grasped. If the hand Existing electrodes require recalibrating the decoder because
does not receive touch and limb-positioning signals once it the stiff electrodes change position with respect to neurons
begins to manipulate the object, performance de­­grades quickly. from day to day; researchers would ultimately like to follow the
Finding a way to correct this deficit is critical for our volun- activity of identical neurons over weeks and months.
teers with spinal cord lesions, who cannot move their body The implants need to be miniaturized, operate on low
below the injury. They also do not perceive the tactile sensa- power (to avoid heating the brain), and function wirelessly so
tions or positioning of their body that are essential to fluid no cables are needed to connect the de­­vice to brain tissue. All
movement. An ideal neural prosthesis, then, must compensate current BMI technology needs to be implanted with a surgical
through bidirectional signaling: it must transmit the inten- procedure. But one day, we hope, recording and stimulation
tions of the volunteer but also detect the touch and positioning interfaces will be developed that can receive and send signals
information arriving from sensors on a robotic limb. less invasively but with high precision. One step in this direc-
Robert Gaunt and his colleagues at the University of Pitts- tion is our recent finding in nonhuman primates that ultra-
burgh have addressed this issue by im­­plant­ing mi­­croelectrode sound re­­corded changes in blood volume linked to neural
arrays in the somatosensory cortex of a te­t­­­­­raplegic person— activity can be used for BMIs. Because the skull is an impedi-
where inputs from the limbs process feelings of touch. Gaunt’s ment to ultrasound, a small ultrasound-transparent window
lab sent small electric currents through the microelectrodes, would still be needed to replace a bit of the skull, but this sur-
and the subject re­­ported sensations from parts of the surface gery would be far less invasive than implanting microelec-
of the hand. trode arrays that require opening the dura mater, the strong
We have also used similar implants in the arm re­gion of the layer surrounding and protecting the brain, and directly in­­
somatosensory cortex. To our pleasant surprise, our subject, serting electrodes into the cortex.
FG, reported natural sensations such as squeezing, tapping and BMIs, of course, are aimed at assisting people with paralysis.
vibrations on the skin, known as cutaneous sensations. He also Yet science-fiction books, movies and the media have focused
perceived the feeling that the limb was moving—a sensation re­­ on the use of the technology for enhancement, conferring
fer­red to as proprioception. These experiments show that sub- “superhuman” abilities that might allow a person to react faster,
jects who have lost limb sensation can regain it through BMIs certainly an advantage for many motor tasks, or directly send
that have write-in perceptions. The next step is to provide a and receive information from the cortex, much like having a
rich variety of somatosensory feedback sensations to improve small cell phone implanted in the brain. But en­­hance­ment is
robotic manual dexterity under brain control. Toward this goal, still very much in the realm of science fiction and will be
the Pittsburgh group has recently shown that stimulation of achieved only when noninvasive technologies are developed
the primary somatosensory cortex improves the time to grasp that can operate at or near the precision of current microelec-
objects with a robot limb, compared with standard visual feed- trode array technology.
back only. Also, we would like to know if subjects detect a sense Finally, I would like to convey the satisfaction of doing basic
of “embodiment,” in which the robot limb appears to become research and making it available to pa­­tients. Fundamental sci-
part of their body. ence is necessary to both ad­­vance knowledge and develop
As these clinical studies show us, both writing in and read- medical therapies. To be able to then transfer these discoveries
ing out cortical signals, provide insight into the degree of reor- into a clinical setting brings the research endeavor to its ulti-
ganization of the cerebral cortex after neurological injury. mate realization. A scientist is left with an undeniable feeling
Numerous studies have reported a high degree of reorganiza- of personal fulfillment in sharing with patients their de­­light at
tion, but until recently there has been little focus on the funda- being able to move a robotic limb to interact again with the
mental structure that remains intact. BMI studies show that physical world. 
tetraplegic subjects can quickly use the motor and the PPC
cortex to control assistive devices, and stimulation of the so­­ Richard A. Andersen is James G. Boswell Professor of Neuroscience and
matosensory cortex produces sensations in deinnervated areas the Tianqiao and Chrissy Chen Brain-Machine Interface Center Leadership chair
that are similar to what would be expected for intact individu- and center director at the California Institute of Technology. He studies the neural
als. These results demonstrate considerable stability of the mechanisms of sight, hearing, balance, touch and action, as well as the development
adult cortex even after severe injury and in spite of injury- of neural prostheses. Andersen is a member of the National Academy of Sciences and
induced plasticity. the National Academy of Medicine.

SCIENTIFICAMERICAN.COM  |  81
INFECTIOUS
DREAMS

How the COVID-19 pandemic is changing our sleeping lives

By Tore Nielsen
Illustrations by Goñi Montes

SCIENTIFICAMERICAN.COM  |  83
F
o r m an y o f us, li v i ng in a ­C OVI D - 1 9 worl d fe e ls as if w e have b e e n
thrown into an alternative reality. We live day and night inside the same walls.
We fear touching groceries that arrive at our doorstep. If we venture into town,
we wear masks, and we get anxious if we pass someone who is not wearing one.
We have trouble discerning faces. It’s like living in a dream.
COVID has altered our dream worlds, too: how much we dream, how many of
our dreams we remember and the nature of our dreams themselves. In early
2020, when stay-at-home directives were put in place widely, society quite unexpectedly expe-
rienced what I am calling a dream surge: a global increase in the reporting of vivid, bizarre
dreams, many of which are concerned with coronavirus and social distancing. Terms such as
coronavirus dreams, lockdown dreams and ­COVID nightmares emerged on social media. By
April of that year, social and mainstream media outlets had begun broadcasting the message:
the world is dreaming about ­COVID.
Although widespread changes in dreaming had been reported 12 and older responded, almost all in North America. Results of
in the U.S. following extraordinary events such as the 9/11 attacks these inquiries, published in BMJ Open in December 2020, doc-
in 2001 and the 1989 San Francisco earthquake, a surge of this ument the precipitous surge, the striking variety of dreams and
magnitude had never been documented. This upwelling of dreams many related mental health effects.
is the first to occur globally and the first to happen in the era of Bulkeley’s three-day poll revealed that in March 2020, 29 per-
social media, which makes dreams readily accessible for immedi- cent of Americans recalled more dreams than usual. Solomono-
ate study. As a dream “event,” the pandemic is unprecedented. va and Robillard found that 37  percent of people had pandemic
But what kind of phenomenon is this, exactly? Why was it dreams, many marked by themes of insufficiently completing
happening with such vigor? To find out, Deirdre Barrett, an tasks (such as losing control of a vehicle) and being threatened
assistant professor at Harvard Medical School and editor in by others. Many online posts from the time reflect these findings.
chief of the journal D reaming, i nitiated a COVID dreams survey One person, whose Twitter handle is @monicaluhar, reported,
online in the week of March 22, 2020. Erin and Grace Gravley, “Had a dream about returning as a sub teacher in the fall, unpre-
San Francisco Bay Area artists, launched IDreamof­Covid.com, pared. Students were having a difficult time practicing social
a site archiving and illustrating pandemic dreams. The Twitter distancing, and teachers couldn’t stagger classes or have one-on-
account @CovidDreams began operation. Kelly Bulkeley, a psy- one meetings.” And @therealbeecarey said, “My phone had a
chologist of religion and director of the Sleep and Dream Data- virus and was posting so many random pictures from my cam-
base, followed with a YouGov survey of 2,477 American adults. era roll to instagram and my anxiety was at an all time high.”
And my former doctoral student Elizaveta Solomonova, now a More recent studies found qualitative changes in dream
postdoctoral fellow at McGill University, along with Rébecca emotions and concerns about health. Dream reports from Bra-
Robillard of the Royal’s Institute of Mental Health Research in zilian adults in social isolation had high proportions of words
Ottawa and others, launched a survey to which 968 people aged related to anger, sadness, contamination and cleanliness. Text

84  |  SCIENTIFIC AMERICAN  |  SPECIAL EDITION  |  WINTER 2022

mining of accounts of 810 Finnish dreams showed that most weeks of insufficient rest, they were more likely to wake up at
word clusters were laden with anxiousness; 55  percent were night and remember more dreams.
about the pandemic directly (lack of regard for social distanc-
ing, elderly people in trouble), and these emotions were more DREAM FUNCTIONS OVERWHELMED
prevalent among people who felt increased stress during the The subject mat ter o  f many ­COVID dreams directly or meta-
day. A study of 100 nurses conscripted to treat C
­ OVID patients phorically reflects fears about contagion and the challenges of
in Wuhan, China, revealed that 45  percent experienced night- social distancing. Even in normal times, we dream more about
mares—twice the lifetime rate among Chinese psychiatric out- novel experiences. For instance, people enrolled in programs to
patients and many times higher than that among the 5 percent rapidly learn French dream more about French. Replaying frag-
or so of the general population who have nightmare disorder. ments of experiences is one example of a functional role that
It seems clear that some basic biological and social dynam- researchers widely ascribe to REM sleep and dreaming: it helps us
ics may have played a role in this unprecedented opening of the solve problems. Other roles include consolidating the prior day’s
oneiric floodgates. At least three factors may have triggered or events into longer-lasting memories, fitting those events into an
sustained the dream surge: disrupted sleep schedules augment- ongoing narrative of our lives and helping us regulate emotions.
ing the amount of rapid eye movement (REM) sleep and there- Researchers have documented countless cases of dreams
fore dreaming; threats of contagion and social distancing tax- assisting in creative achievement. Empirical studies also show
ing dreaming’s capacity to regulate emotions; and social and that REM sleep aids in problem-solving that requires access to
mainstream media amplifying the public’s reaction to the surge. wide-ranging memory associations, which may explain why so
many dreams in the 2020 surge involve creative or strange
MORE REM SLEEP, MORE DREAMS attempts to deal with a C ­ OVID problem. One survey respon-
O n e o b v i o us e x p l a nat i o n for the surge is that sleep pat- dent said, “I was looking for a kind of cream that would either
terns changed abruptly when lockdowns took effect. Early pub- prevent or cure ­Covid-19. I got my hands on the last bottle.”
lications demonstrate elevated levels of insomnia in China’s Two other widely claimed dream functions are extinguishing
population, especially among frontline workers. In contrast, fearful memories and simulating social situations. They are relat-
stay-at-home orders, which removed long commutes to work, ed to emotion regulation and help to explain why pandemic
im­­proved sleep for many people. Respondents in China report- threats and social distancing challenges appear so often in surge
ed an average increase of 46 minutes in bed and an extra 34 dreams. Many dreams reported in the media include fearful reac-
minutes in total sleep time. Some 54  percent of people in Fin- tions to infection, finances and social distancing: “I tested positive
land said they slept more after lockdown. Overall, from March for pregnancy and covid  ... now I’m stressed.” Threats may take
13 to 27, 2020, time asleep in the U.S. increased almost 20  per- the form of metaphoric imagery such as tsunamis or aliens; zom-
cent nationwide, and states with the longest commute times, bies are common. Images of insects, spiders and other small crea-
such as Maryland and New Jersey, showed the largest increases. tures are also widely represented: “My foot was covered in ants
Longer slumber leads to more dreams; people in sleep labo- and 5-6 black widows were imbedded in the bottom of my foot.”
ratories who are allowed to snooze for more than 9.5 hours One way to understand direct and metaphoric imagery is to
recall more dreams than when sleeping a typical eight hours. consider that dreams express an individual’s core concerns, draw-
Sleeping longer also proportionally increases REM sleep, which ing on memories that are similar in emotional tone but different
is when the most vivid and emotional dreams occur. in subject matter. This contextualization is clear in post-traumat-
Relaxed schedules may also have caused dreaming to occur ic nightmares, in which a person’s reaction to a trauma, such as
later than usual in the morning, when REM sleep is more prev- terror during an assault, is depicted as terror in the face of, for
alent and intense and, thus, dreams are more bizarre. Dream example, a natural disaster such as a tsunami. The late Ernest
tweets reflect these qualities: “I was taking care of a newborn Hartmann, a Boston-area dream and nightmare research pioneer
girl that had COVID. . . it was so vivid and real.” Increased who studied dreams after the 9/11 attacks, stipulated that such
dreaming during late-morning REM intervals results from the contextualization best helps people adapt when it weaves togeth-
convergence of several processes. Sleep itself cycles through er old and new experiences. Successful integration produces a
deep and light stages about every 90 minutes, but pressure for more stable memory system that is resilient to future traumas.
REM sleep gradually increases as the need for deep, recupera- Metaphorical images can be part of a constructive effort to
tive sleep is progressively satisfied. Meanwhile a circadian pro- make sense of disruptive events. A related process is the extin-
cess that is tightly linked to our 24-hour core body temperature guishing of fear by the creation of new “safety memories.” These
rhythm gives an abrupt boost to REM sleep propensity late in possibilities, which I and others have investigated, reflect the
the sleep period and stays elevated through the morning. fact that memories of fearful events are almost never replayed
After the pandemic began, many people did sleep longer and in their entirety during dreaming. Instead elements of a memo-
later. In China, average weekly bedtime was delayed by 26 min- ry appear piecemeal, as if the original memory has been re­­
utes but wake-up time by 72 minutes. These values were 41 and duced to basic units. These elements recombine with newer
73 minutes in Italy and 30 and 42 minutes among U.S. universi- memories and cognitions to create contexts in which meta-
ty students. And without commutes, many people were freer to phors and other unusual juxtapositions of imagery seem incon-
linger in bed, remembering their dreams. Some early birds may gruous or incompatible with waking life—and, more important,
have turned into night owls, who typically have more REM are incompatible with feelings of fear. This creative dreaming
sleep and more frequent nightmares. And as people eliminated produces safety imagery that supersedes and inhibits the origi-
whatever sleep debts they may have accrued over days or even nal fear memory, helping to assuage distress over time.

SCIENTIFICAMERICAN.COM  |  85
 This mechanism can break down after severe trauma, how- were circulated widely online, feeding a narrative of pandemic
ever. When this happens, nightmares arise in which the fearful dreams that went viral, influencing people to recall their dreams,
memory is replayed realistically; the creative recombining of notice ­COVID themes and share them. This narrative may have
memory elements is thwarted. The pandemic’s ultimate impact even induced people to dream more about the pandemic.
on a person’s dreams will vary with whether or how severely Evidence suggests that mainstream media reporting probably
they are traumatized and how resilient they are. did not trigger the surge but may have temporarily amplified its
A second class of theories—also still speculative—may explain scope. The Bulkeley and Solomonova-Robillard polls corroborat-
social distancing themes, which permeated IDreamofCovid.com ed a clear groundswell in dream tweeting during March 2020,
reports. Emotions in these dreams range from surprise to dis- before the first media stories about such dreams appeared; indeed,
comfort to stress to nightmarish horror. Tweets located by the­ the earliest stories cited various tweet threads as their sources.
@CovidDreams account illustrate how incompatible dream sce- Once stories emerged, more surges in dream reporting
narios are with social distancing—so incompatible that they through early April 2020 were detected by I­ DreamofCovid.com
often trigger a rare moment of self-awareness and awakening: and @CovidDreams. The format of most early stories almost
“We were celebrating something by having a party. And I woke guaranteed amplification: they typically described some salient
myself up because something wasn’t right because we’re social dream themes observed in a survey and provided a link directing
distancing and not supposed to be having parties.” readers to participate in the same survey. In addition, 56 percent
These theories focus on dreaming’s social simulation function. of articles during the first week of stories featured interviews
The view that dreaming is a neural simulation of reality, analo- with the same Harvard dream scientist, which may have influ-
gous to virtual reality, is now widely accepted, and the notion enced readers to dream about themes repeated by her.
that the simulation of social life is an essential biological function The surge began to decline steadily in late April 2020, as did
is emerging. In 2000 Anne Germain, now CEO of sleep medicine the number of mainstream media articles, suggesting that any
start-up Noctem, and I proposed that images of characters inter- echo-chamber effect had run its course. The final nature of the
acting with the self in dreams could be basic to how dreaming surge remains to be seen. Until C ­ OVID vaccines or treatments
evolved, reflecting attachment relationships essential to the sur- are fully distributed and with waves of future infections or new
vival of prehistoric groups. The strong interpersonal bonds reit- viral variants possible, threats of disease and social distancing
erated during dreaming contribute to stronger group structures are likely to persist. Might the pandemic have produced a lasting
that help to organize defenses against predators and cooperation increase in humanity’s recall of dreams? Could pandemic con-
in problem-solving. Such dreams would still have adaptive value cerns become permanently woven into dream content? And if so,
today because family and group cohesion remain essential to will such alterations help or hinder people’s long-term adjust-
health and survival. It may be the case that an individual’s con- ments to our postpandemic futures?
cerns about other people are fine-tuned while they are in the sim- Therapists may need to step in to help certain people. The
ulated presence of those people. Important social relationships survey information considered in this article does not delve
and conflicts are portrayed realistically during dreaming. into nightmares in detail. But some health-care workers who
Other investigators, such as cognitive neuroscientist Antti saw relentless suffering later themselves suffered with recur-
Revonsuo of the University of Turku in Finland and the Universi- rent nightmares. And some patients who endured the ICU for
ty of Skövde in Sweden, have since proposed additional social days or weeks suffered from horrific nightmares during that
functions for dreaming: facilitating social perception (Who is time, which may in part have been the result of medications
around me?), social mind reading (What are they thinking?) and and sleep deprivation induced by around-the-clock hospital
the practice of social bonding skills. Another theory advanced by procedures and interminable monitor noises and alarms. These
psychology professor Mark Blagrove of Swansea University in survivors will need expert help to regain normal sleep. Thank-
Wales further postulates that by sharing dreams, people enhance fully, specialized techniques are highly effective.
empathy toward others. The range of dream functions is likely to People who are not traumatized but still a little freaked out
keep expanding as we learn more about the brain circuits under- about their C­ OVID dreams also have options. New technologies
lying social cognition and the roles REM sleep plays in memory such as targeted memory reactivation are providing individuals
for emotional stimuli, human faces and reactions to social exclu- with more control over their dream narratives. For example,
sion. Because social distancing is, in effect, an experiment in learning how to practice lucid dreaming—becoming aware that
social isolation at a level never before seen—and is likely antago- you are dreaming—aided by targeted memory reactivation or
nistic to human evolution—a clash with deep-rooted dream other methods could help transform worrisome pandemic
mechanisms should be evident on a massive scale. And because dreams into more pleasant, maybe even useful, dreams. Simply
social distancing disrupts normal relationships so profoundly— observing and reporting pandemic dreams seems to positively
causing many of us to spend excessive time with some people and impact mental health, as Natália Mota of the Federal University
no time with others—social simulations in dreams may play a of Rio Grande do Norte in Natal, Brazil, found in her studies.
crucial role in helping families, groups, even societies deal with Short of therapy, we can give ourselves permission to ease
sudden, widespread social adaptation. up and to enjoy banking those surplus hours of sleep. Dreams
can be vexing, but they are also impressionable, malleable and
THE ECHO CHAMBER OF SOCIAL MEDIA at times inspirational. 
There is one basic question about pandemic dreams that we
would like to nail down: whether the dream surge was amplified Tore Nielsen is a professor of psychiatry at the Université de Montréal and director
by the media. It is quite possible that early posts of a few dreams of the Dream and Nightmare Laboratory there.

86  |  SCIENTIFIC AMERICAN  |  SPECIAL EDITION  |  WINTER 2022

SCIENTIFICAMERICAN.COM  |  87
88  |  SCIENTIFIC AMERICAN  |  SPECIAL EDITION  |  WINTER 2022
SLEEP
LEARNING
GETS REAL
Experimental techniques demonstrate how to
strengthen memories when our brains are off-line
By Ken A. Paller and Delphine Oudiette
Photograph by Hannah Whitaker

SCIENTIFICAMERICAN.COM  |  89
I n Aldous Huxley ’ s Brave New World, a boy memorizes each word of a  lecture
in English, a language he does not speak. The learning happens as the boy sleeps within
earshot of a radio broadcast of the lecture. On awakening, he is able to recite the entire
lecture. Based on this discovery, the totalitarian authorities of Huxley’s dystopian world adapt
the method to shape the unconscious minds of all their citizens.
Sleep learning turns up throughout literature, pop
culture and ancient lore. Take Dexter, the lead charac-
ter in the animated television series D  exter’s Labora-
tory. In one episode, Dexter squanders his time for
homework, so instead he invents a contraption for
learning to speak French overnight. He wakes up the
next day unable to speak anything but French. The
dling with sleep had begun. In 1927 New Yorker Alois
B. Saliger invented an “Automatic Time-Controlled
Suggestion Machine,” which he marketed as the “Psy-
choPhone,” to allow a recorded message to be replayed
during the night. The setup seemed to evoke Huxley’s
imagined technology except that the user, rather than
the state, could select the message to be played.
idea of sleep learning isn’t just a modern invention. It Saliger’s invention was followed, in the 1930s and
also appears within a centuries-old mind-training 1940s, by studies documenting ostensible examples of
practice of Tibetan Buddhists; a message whispered sleep learning. A 1942 paper by Lawrence LeShan,
during sleep was intended to help a monk recognize then at the College of William  & Mary, detailed an
the events in his dreams as illusory. experiment in which the researcher visited a summer
Everyone knows we learn better when we are well camp where many of the boys had the habit of biting
rested. Most people, however, dismiss the notion of their fingernails. In a room where 20 such boys slept,
sleep learning out of hand. Yet a set of new neurosci- LeShan used a portable phonograph to play a voice
entific findings complicates this picture by showing repeating the sentence “My fingernails taste terribly
that a critical part of learning occurs during sleep: bitter.” The string of words recurred 300 times each
recently formed memories resurface during the night, night, beginning 150 minutes after the onset of sleep.
and this playback can help reinforce them, allowing at The experiment continued for 54 consecutive nights.
least a few to be remembered for a lifetime. During the last two weeks of camp, the phonograph
Some studies have even explored whether sleep broke, so the intrepid LeShan delivered the sentence
might be manipulated to enhance learning. They himself. Eight of the 20 boys stopped biting their nails,
reveal that sleep’s program for making daytime mem- whereas none of 20 others who slept without exposure
ories stronger can be boosted using sounds and odors. to the recording did so. These early efforts did not use
Results in rodents have even demonstrated a primi- physiological monitoring to verify that the boys were
tive form of memory implantation: using electrical really asleep, though, so the results remain suspect.
stimulation while animals slept, researchers taught The whole field took a severe hit in 1956, when two
them where they should go in their enclosures on scientists at RAND Corporation used electroencepha-
awakening. Huxley’s imagined version of sleep educa- lography (EEG) to record brain activity while 96 ques-
tion, in which entire texts are absorbed verbatim dur- tions and answers were read to sleeping study partici-
ing the night, is still relegated to the pages of his 1932 pants. (One example: “In what kind of store did Ulys­
classic. But experiments now indicate that it is possi- ses S. Grant work before the war?” Answer: “A hard­ware
ble to tinker with memories while a person is im­­ store.”) The next day correct answers were re­­called only
mersed in the depths of slumber, creating the basis for for information presented when sleepers showed signs
a new science of sleep learning. of awakening. These results led to a shift in the field
that persisted for 50 years, as researchers began to lose
THE PSYCHOPHONE faith in sleep learning as a viable phenomenon: partic-
For these techniques t o work, scientists have to ipants in these experiments appeared to learn only if
explore how information can be absorbed when con- they were not really sleeping when information was
sciousness is seemingly on a well-deserved break. presented to them.
Around the time that Huxley was writing B  rave New Most scientists during this time tended to avoid the
World, serious explorations into the possibility of med- topic of sleep learning, although a few researchers did

90  |  SCIENTIFIC AMERICAN  |  SPECIAL EDITION  |  WINTER 2022

plug away at asking whether sleep assists in remember- at one designated spot on a computer screen and a
ing new information. One typical study protocol probed teakettle at another. At the same time, they would
whether overnight sleep deprivation affected recall the hear a corresponding sound (a meow for the cat, a
day after learning something new. Another asked whistle for the kettle, and so on).
whether remembering was better after a nap than after After this learning phase, participants took a nap
an equal period of time spent awake. in a comfortable place in our lab. We monitored EEG
Various confounding factors can interfere with such recordings from electrodes placed on the head to veri-
studies. For example, the stress of sleep deprivation can fy that each individual was soundly asleep. These
harm cognitive functions, which then decreases memo- recordings provided intriguing data on the synchro-
ry recall. Eventually cognitive neuroscientists began to nized activity of networks of neurons in the brain’s
tackle these challenges by bringing together evidence outer layer, the cerebral cortex, that are relevant for
from multiple research methods. A sub­stant­ive founda- memory reactivation [see box on next page]. When we
tion of evidence gradually accrued to confirm that sleep detected slow-wave sleep, we played the meow, whis-
is a means of reviving memories ac­quired during the tle and other sounds associated with a subset of the
day, reopening the relation between sleep and memory objects from the learning phase. Sounds were present-
as a legitimate area of scientific study. ed softly, not much louder than background noise, so
Many researchers who took up the challenge focused the sleeper did not awaken.
on rapid eye movement (REM) sleep, the period when On awakening, people remembered locations cued
dreams are the most frequent and vivid. The guiding during sleep better than locations that had not been
assumption held that the brain’s
nighttime processing of memories
would be tied to dreaming, but
clear-cut data did not materialize.
Future programs for sleep learning might
In 1983 two noted scientists— help in preserving memories, speeding
Graeme Mitchison and Francis
Crick, neither a psychologist—went acquisition of new knowledge, or
so far as to speculate that REM
sleep was for forgetting. In a similar
even changing bad habits such as smoking.
vein, Giulio Tononi and Chiara
Cirel­li, both at the University of Wisconsin–Madison, cued during sleep. Whether sounds or odors served as
proposed that sleep could be the time for weakening cues in these experiments, they apparently triggered
connections among brain cells, making it easier for new the reactivation of spatial memories and so reduced
information to be acquired the following day. forgetting.
Instead of REM, some investigators focused their At first, the auditory procedures we used were high-
attention on slow-wave sleep (SWS), a period of deep ly controversial. The received wisdom among sleep
slumber without rapid eye movements. In 2007 Björn researchers held that sensory circuits in the cortex are
Rasch, then at the University of Lübeck in Germany, largely switched off during sleep, except for the sense
and his colleagues prepared people for a sleep experi- of smell. We were not swayed by this orthodox view.
ment by re­­quiring them to learn the locations of a set Instead we followed our hunch that the repeated play-
of objects while simultaneously smelling the odor of a ing of soft sounds might influence the sleeping brain
rose. Later, in their beds in the laboratory, sleeping and produce changes in recently stored memories.
study participants again encountered the same odor Indeed, the same memory benefits were also found
as electrical recordings confirmed one sleep stage or in many subsequent studies. A technique called func-
another. The odor activated the hippocampus, a brain tional magnetic resonance imaging highlighted which
area critical for learning to navigate one’s surround- brain areas take part in TMR, and EEG results brought
ings and for storing the new knowledge gained. On out the importance of specific brain oscillations. Two
awakening, participants recalled locations more accu- papers published in 2018—one by Scott Cairney of the
rately—but only following cueing from odors that ema- University of York in England and his colleagues, the
nated during the course of slow-wave (not REM) sleep. other by James Antony of Princeton University and
his colleagues—linked an oscillation, the sleep spindle,
TARGETED MEMORY REACTIVATION with the memory benefits of TMR.
I n 2 0 0 9 o u r l a b e x t e n d e d this methodology by Besides boosting spatial memory, these procedures
using sounds instead of odors. We found that sounds have also helped improve recall in other settings. TMR
played during SWS could improve recall for individu- can assist in mastery of playing a keyboard melody and
al objects of our choosing (instead of the recall of an learning new vocabulary or grammatical rules. The
entire collection of objects, as was the case in the odor technique can also help with simpler types of learning,
study). In our procedure—termed targeted memory such as adjustments in one’s body image. In condition-
re­­activation, or TMR—we first taught people the lo­­ ing experiments, TMR alters prior learning of an auto-
cations of 50 objects. They might learn to place a cat matic reaction to a stimulus caused by an earlier pairing

SCIENTIFICAMERICAN.COM  |  91
 The Maestros of Slumber
A complex symphony of neural activity governs the connection between sleep and memory
Brain rhythms provide clues t o how sleep helps to store memories ing of connections among neurons to strengthen memory stor-
for later retrieval. One type of neural signal, called a slow wave, age. A dialogue between the hippo­camp­us and the cortex involv-
cycling from 0.5 to four times a second, orchestrates the activity ing all these brain rhythms triggers a set of complex network
of neurons in the cerebral cortex. Each slow oscillation consists interactions. Through this process, known as consolidation, new
of a “down” phase, when neurons are silent, and an “up” phase, information can become integrated with existing memories. The
when they resume activity. This timing pattern helps to reinforce intertwining of memories, moreover, enables the gist of recent
recently formed memories by ensuring that multiple cortical experiences to be extracted to make sense of a complex world.
regions remain in an up state at the same time. Memory difficulties can arise when this neural dialogue be­­
The up phase can coincide with sleep spindles, brief increases comes im­­paired. Individuals with major damage centered in the
of a rhythm of 12 to 15 cycles per second. Spindles originate in hippo­campus or parts of the thalamus may develop a profound
the thalamus, which serves as a crossroads for information that amnesia. Without the expected interactions with these brain
is transmitted to virtually all parts of the cerebral cortex. Spindles regions during both sleep and waking, the cortex cannot store
have a rhythm of their own, recurring at approximately five-sec- mental records of facts and events known as declarative memo-
ond intervals. They coordinate the activity of sharp-wave ripples ries. In addition, a milder form of memory disorder may result
in the hippocampus. Ripples, for their part, are concurrent with when memory processing during sleep is seriously disrupted.
the replay of mem­­ories. Slow waves, all the while, assume the role Deciphering the physiological orchestration of the sleeping
of orchestra conductor: their measured oscillations in the cortex brain is prompting various new strategies for enhancing the brain’s
coordinate the pacing for sleep spindles and sharp-wave ripples. natural rhythms—including stimulation with slow electrical oscil-
The intricate coupling of these oscilla- lations, sounds or gentle motion. These methods echo humans’
tions underlies not only memory natural inclinations to take advantage of a lullaby’s rhythm or the
reactivation but also the alter- rocking of a cradle to lull a baby to sleep.  —K .A.P. and D.O.

A Symphony in Two Movements

1 hour Dramatic differences characterize two key sleep phases. The slow waves of deep sleep dominate the early part of
0 e 2h the night. During slow-wave sleep, some memories are spontaneously reactivated. Interventions that promote
Tim
this pro­cess can ensure that memories are retained. Rapid eye movement (REM) sleep prevails in the latter
Generalized Sleep Cycle part of a night’s slumber, but how it interacts with memory remains controversial.
Awake (orange)
Non-REM light sleep (green)
Slow-wave sleep (blue)
Harmonizing Brain Waves
REM sleep (yellow) Brain oscillations during sleep appear to play a role in strengthening new memories.
4h A key event is the “up” phase of a slow oscillation that coordinates the activity of other
brain rhythms. The ascending part of a slow oscillation in the cortex synchronizes with
sleep spin­dles in the thalamus. The spindles coordinate the activity of sharp-wave
ripples in the hippocampus. Ripples tend to coincide with a spindle trough.
5h
Electrical Activity
8h 6h in the Brain
7h
Slow waves
Cerebral cortex Slow-wave up
phase corresponds
with spindle

Spindles
Thalamus

Spindle trough coincides

with ripple activity
Sharp-wave ripples
Hippocampus

Illustration by Mesa Schumacher

of that stimulus with an electric shock. Ongoing studies could acquire new verbal knowledge, but this was evi-
are examining still other types of recall, such as associ- dent only through subtle nonconscious means. More
ating names with faces when first meeting new people. recently, we showed episodic learning during sleep
As the technology evolves, TMR should be tested to with full recollection of the learning. In a multi-institu-
see whether it could help to treat various disorders, tional collaboration by researchers in France, Germany,
reduce addictions or speed recovery from illness. Our the Netherlands and the U.S., we used a variant of the
lab, together with Northwestern University neurolo- TMR method to encourage lucid dreaming—a state in
gist Marc Slutzky, is currently testing a novel rehabili- which people realize they are dreaming while remain-
tation procedure for recovering arm-movement abili- ing in the midst of the dream. We then showed that
ties after stroke. Cue sounds are incorporated as part people could understand softly spoken questions from
of the therapy and are replayed during sleep to try to within these dreams and produce correct answers by
accelerate relearning of lost movements. The pros- signaling with their eyes, their respiration or subtle
pects appear promising because TMR can alter simi- twitches in their facial muscles. Sometimes people in
lar forms of motor learning in healthy individuals. these experiments woke up able to recollect parts of
their dreamtime Q and A. These rare occurrences con-
WHAT ABOUT LEARNING FRENCH? vincingly document full-blown learning experienced
T he demonstrated abilit y to reinforce memories entirely during sleep.
raises the question of whether new information can The boundaries of sleep hacking may continue to
be loaded into a person’s brain after falling asleep, expand, but this research has established that a nor-
a technique that calls forth the ethical specter of mal component of learning continues nocturnally off-
mind control invoked by Brave New World. I s it going line. Sleep is needed not just to help people stay alert
too far, though, to imagine that memories can be cre- and rejuvenated but also to reinforce memories ini-
ated surreptitiously? tially acquired while they were awake. We still need to
Although the orthodox response to such conjectures learn much more about off-line memory processing.
has for many years been an unqualified no, studies by Further work must ascertain how sleep helps learning
Anat Arzi, then at the Weizmann Institute of Science in and which brain mechanisms are engaged to preserve
Rehovot, Israel, and her colleagues demonstrated the the most valuable memories. It is also essential to find
creation of relatively simple memories using odors. In out more about the perils of poor or inadequate sleep
one experiment, the re­­search­ers succeeded in diminish- that might be affected by various forms of life stress,
ing the desire for tobacco in smokers who were keen to certain diseases or the experience of growing older.
quit. When asleep, study participants were exposed to A study led by Carmen Westerberg, then at North-
two odors, cigarette smoke and rotten fish. During the western, points in the de­­­­sired direction. Westerberg
next week, those who had smelled the mix of both odors tested patients with the memory dysfunction that
lit up 30  percent less, having apparently been condi- often precedes Alzheimer’s disease—amnestic mild
tioned to associate smoking with the aversive fish odor. cognitive impairment. The results documented a link
Acquiring a more complex memory is not as easy, between poor sleep and reduced ability to remember
but evidence from the past decade holds tantalizing information after an intervening overnight delay.
promise. Karim Benchenane of the French National All of this knowledge might help in creating pro-
Center for Scientific Re­­search (CNRS) and his col- grams of sleep learning to preserve memories, to speed
leagues have shown how to literally change the mind— the acquisition of new knowledge, or even to change
of a mouse. When they began their work, Benchenane bad habits such as smoking. Looking still further ahead,
and his team knew that when a mouse explores a new scientists might also explore whether we can gain con-
environment, neurons called place cells fire as the ani- trol over our dreams, which could lead to the prospect
mal traverses specific parts of an enclosure. These of nightmare therapies, sleep-based problem-solving
same neurons discharge again during sleep as the and perhaps even recreational dream travel. In a cul-
memory is apparently replayed. ture that already offers wrist-based sleep trackers and
The researchers stimulated the re­­ward system of mail-order gene tests, we can begin to contemplate new
the mouse brain (the medial forebrain bundle) precise- ways to convert daily downtime into a productive
ly when place cells became spontaneously active while endeavor—for some, a chilling prospect, and for others,
the animal was asleep. Amazingly, mice subsequently another welcome opportunity for self-improvement. 
spent more time at the locations that corresponded to
the stimulated place cells, heading there directly after Ken A. Paller is a professor of psychology, holds the James Padilla Chair in Arts
they woke up. More ex­­per­i­ments still need to disentan- & Sciences, and directs the cognitive neuroscience program at Northwestern
gle whether fully formed false memories were implant- University. His research on targeted memory reactivation was funded by the
ed in the mice during sleep or whether they were auto- U.S. National Science Foundation.
matically guided to those spots by a process of condi-
tioning, without more knowledge about why they were Delphine Oudiette is a tenured researcher for the French National Institute for
drawn to those locations. Health and Medical Research (INSERM) at the Brain & Spine Institute and at the
In 2019 Swiss researchers reported that sleepers sleep disorder department located at Pitié-Salpêtrière Hospital, both in Paris.

SCIENTIFICAMERICAN.COM  |  93
 Answering Queries
in Real Time
while Dreaming
Researchers demonstrate that during REM sleep,
people can hear—and respond to—simple questions such as
“What is eight minus six?”
By Diana Kwon

D
reams are full of possibilities; believed that it was going to be possible— the French group asked its participant—
by drifting into the world beyond but until we actually demonstrated it, we a person with narcolepsy who had expert
our waking realities, we can visit weren’t sure.” lucid-dreaming abilities—to answer yes-or-
magical lands, travel through time and in- For this study, Paller and his colleagues no questions with facial muscle contrac-
teract with long-lost family and friends. recruited volunteers who said they remem- tions rather than eye movements.
The notion of communicating in real time bered at least one dream per week and pro- Across the four studies, there were a to-
with someone outside of our dream­ vided them with guidance on how to lucid tal of 36 participants and 158 trials during
scapes, however, sounds like science fic- dream. They were also trained to respond which the researchers could verify lucid
tion. A recent study demonstrates that, to to simple math problems by moving their dreaming and attempted to establish con-
some extent, this seeming fantasy can be eyes back and forth—for example, the cor- tact. Answers were considered correct if
made real. rect response to “eight minus six” would be three of four raters were in agreement on
Scientists already knew that one-way to move your eyes to the left and right whether the responses, sometimes very
contact is attainable. Previous studies twice. While the participants slept, elec- subtle movements, were accurate. Correct
have demonstrated that people can pro- trodes attached to their faces picked up responses were given in 18 percent of trials;
cess external cues, such as sounds and their eye movements, and electroencepha- another 18  percent were classified as am-
smells, while asleep. There is also evi- lography (EEG)—a method of monitoring biguous because raters could not come to a
dence that people are able to send mes- brain activity—kept track of what stage of consensus about whether participants gave
sages in the other direction: lucid dream- sleep they were in. a correct response or whether they had re-
ers—those who can become aware they As Paller’s team was conducting these sponded at all. Incorrect responses were
are in a dream—can be trained to signal, experiments, the researchers discovered given in 3  percent of the trials. Overall
using eye movements, that they are in the three groups in Germany, France and the there was no response in 60 percent.
midst of a dream. Netherlands who were trying to accom- One of the co-authors, Karen Konkoly,
Two-way communication, however, is plish the same thing. Instead of competing, a graduate student in Paller’s lab, specu-
more complex. It requires a person who is the groups decided to collaborate. They lates that participants failed to respond in
asleep to actually understand what they carried out similar experiments, though 60 percent of the trials because they sim-
hear from the outside and think about it with slightly different methods of answer- ply did not perceive the incoming com-
logically enough to generate an answer, ing questions and receiving responses. The munication. In those cases, they rarely re-
explains Ken Paller, a cognitive neuro­ German group, for example, transmitted ported any incorporation of the questions
scientist at Northwestern University. “We its math problems using Morse code, and into their dreams after waking up. But

94  |  SCIENTIFIC AMERICAN  |  SPECIAL EDITION  |  WINTER 2022

 she adds that it is also possible that dream- The findings “challenge our ideas “I really liked this study,” says Chris-
ers perceived the inputs but paid little at- about what sleep is,” says Benjamin Baird, tine Blume, a sleep scientist at the Center
tention and forgot before awakening. The a researcher who studies dreams at the for Chronobiology in Basel, Switzerland,
proportion of people who respond could University of Wisconsin–Madison and who was not involved in this work. “The
potentially be improved with more train- was not involved in this study. Sleep has ex­­tent to which information can be pro-
ing or by presenting questions when indi- classically been defined as unresponsive- cessed and responded to surprised me.”
viduals are in specific sleeping brain states, ness to external environmental stimuli— But she adds that it is important to keep
Konkoly says. and that feature is still typically part of in mind that the findings relate specifical-
After establishing successful two-way the definition today, Baird explains. “This ly to lucid dreaming, which is a special
communication, participants were woken work pushes us to think carefully—re- type of dreaming that not many people
up and asked to recount their dreams. In think, maybe—about some of those fun- are able to experience.
most cases, they could remember receiving damental definitions about the nature of Blume notes that even with lucid
the experimenters’ questions while asleep; sleep itself and what’s possible in sleep.” dreamers, in most trials, the researchers
in some cases, the questions appeared to This kind of two-way communication were not able to establish communication.
be coming from outside the dream, where- with dreamers could be used as a tool to Therefore, how applicable this technique
as other times they were integrated into the better study dreams, according to Paller. would be to learning or creativity remains
dream. (One participant reported that the In particular, he says, the observation that an open question, she says.
lights in their dream started flickering, the responses some people gave during Paller and his colleagues are now ex-
which they were able to recognize as the dreams did not match their reports after ploring what other types of questions can
Morse-coded math problem.) waking provides evidence that such real- be asked during sleep, as well as other
There were instances, however, when time techniques will help researchers ways of receiving messages from sleepers,
people either did not recall the interactions get more accurate accounts of dreams— such as sniffing. “We are hopefully going
or had a distorted account. For example, and address whether dreams play a use- to get better at doing this kind of experi-
there were trials in which individuals an- ful role in processes such as memory. ment,” he observes. “Then [we can] ask
swered a math problem correctly while Paller and his colleagues also suggest this new questions about what’s happening
Simon Kadula/Alamy

asleep but did not remember the question technique could be used by people to en- during dreams.” 
correctly after waking up. These findings hance problem-solving and creativity, by
were published in February 2021 in Cur- providing a new way to process content in Diana Kwon is a freelance journalist who covers health and
rent Biology. their dreams. the life sciences. She is based in Berlin.

SCIENTIFICAMERICAN.COM  |  95
 OPINION

CAN
SCIENCE
ILLUMINATE
OUR
INNER
DARK
MATTER?
Neither introspection nor brain scans
can reveal our deepest thoughts
By John Horgan

what ’ s going on in your head right now? how about ...

now? Or  . . . now? Answering this question is harder than
you might think. As soon as you pay attention to your
thoughts, you alter them, as surely as you alter an electron’s
course by looking at it. You can’t describe your thoughts the
way you describe, say, the room in which you are reading,
Maria Maglionico/Eye Em/Getty Images

which remains stolidly unaffected by your scrutiny.

William James draws attention to this paradox in “The Stream of Thought,” a
section of T
 he Principles of Psychology. T
 rying to examine your thoughts through
“introspective analysis,” he writes, is like studying snowflakes by catching them
in your “warm hand,” “seizing a spinning top to catch its motion” or “trying to
turn up the gas quickly enough to see how the darkness looks.”

96  |  SCIENTIFIC AMERICAN  |  SPECIAL EDITION  |  WINTER 2022

SCIENTIFICAMERICAN.COM  |  97
 I’ve been brooding over these sorts of puzzles a lot ogists can reduce the human mind to a mental equiva-
lately, even more than usual. In 2020 I released a book lent of atoms as physicists have done with matter.
entitled P  ay Attention: Sex, Death, and Science, a With some effort, I can direct my thoughts, focus
stream-of-thought account of a day in my life. Or rath- them, but they often seem to have a will of their own.
er, in the life of my fictional alter ego, Eamon Toole, a They swerve this way or that for reasons obscure to me,
“divorced science writer and professor struggling to re- a tendency that Buddhists disparage as “monkey mind.”
main rational while buffeted by fears and desires.” As When we do notice a thought and reflect on it—perhaps
soon as I send a book out into the world, I compulsive- to convey it to ourselves or to others—we instantly trans-
ly think of things I should have put in it. form it, turning it into a different, higher-order thought.
Also, my ongoing attempt to learn quantum me- Call it a meta-thought, a thought about a thought.
chanics has mystified my world, inner and outer. So, I’d Meta-thoughts—the thoughts I express to myself
like to offer a few thoughts (second thoughts? after- and to others through writing and speech—are my
thoughts?) about thoughts, the most inescapable and bread and butter. I make my living off them. But they
maddeningly elusive features of our existence. constitute an infinitesimal fraction of my thoughts. The
vast majority are unformed, incoherent, inexpressible,
META-THOUGHTS AND and they come and go without my dwelling on them.
THOUGHTLESS THOUGHTS You might call them thoughtless thoughts. Thoughtless
A note on terms. J ames coined the phrases “stream thoughts are what course through your head when no
of thought” and “stream of consciousness” and some- one is watching you, not even you.
times used them interchangeably, but I distinguish
thoughts from consciousness in the following way: CONSCIOUSNESS WITHOUT THOUGHT?
Thoughts are the c ontents of consciousness, including I once tried to teach m  editation to a class of
fears, fantasies, recollections, realizations, delibera- stressed-out freshmen. I told them to close their eyes,
tions, decisions and all the other flora of subjective ex- still their minds and stop thinking. After 10 minutes of
perience. If consciousness is the medium, thoughts are silence, I asked how many had succeeded in thinking
the message. of nothing. To my surprise, about half raised their
I also like the easy self-referentiality of “thoughts hands. I didn’t believe them. Even freshmen always
about thoughts,” which captures a deep truth about us. have thoughts, whether or not they notice them.
We are what Douglas Hofstadter calls self-generating Is thoughtless consciousness possible? Yes, accord-
“strange loops,” akin to M. C. Escher’s famous drawing ing to religious scholar Robert Forman, a veteran med-
itator. He claims that he and others
have achieved “pure conscious-
ness,” a mystical state devoid of any
THOUGHTS ARE “CONTINUOUS,” THEY “FLOW,” THEY specific thoughts. You are conscious

KEEP COMING EVEN WHEN WE PAY NO ATTENTION but not conscious of anything. Con-
sciousness without content strikes
TO THEM, AND THEY KEEP CHANGING; NO THOUGHT me as a contradiction, an oxymo-
ron, like a book without words or
IS PRECISELY LIKE ANOTHER. a film without images. And how
would you k  now you’re in a state of
pure consciousness? How would
of two hands drawing each other. (Who draws the draw- you remember it? Even Forman admits that states of
er?) Thoughts spring from thoughts and—in ways still pure consciousness, if they exist, are rare.
beyond our ken—from our brains, which contain rough- Meditation is touted as a route to knowledge of your
ly 100 billion neurons linked by one quadrillion synaps- deepest self, your innermost thoughts. I’ve had delight-
es, each of which processes an average of 10 electro- ful experiences meditating, especially on a silent retreat
chemical signals, or action potentials, every second. in 2018. But meditation and other contemplative tech-
If you equate action potentials with the operations niques are designed to control and suppress thoughts
of a computer, as many neuroscientists do, then the rather than to understand them. Meditation is self-
brain carries out 10  quadrillion operations in a typical brainwashing aimed at taming your monkey mind. I
second. That approaches the speed of the world’s fast- don’t want to tame my monkey mind; I want to study
est supercomputers, and the brain may perform expo- it, to comprehend its antics.
nentially more calculations via processes other than ac- Although we may not notice them and may even
tion potentials. The result of all this activity is that brains deny their existence, thoughtless thoughts are always
churn out thoughts as ceaselessly as hearts pump blood. there, underpinned by our brains’ incessant chatter.
As James puts it, thoughts are “continuous,” they Without thoughtless thoughts, we would lack meta-
“flow,” they keep coming even when we pay no atten- thoughts. Thoughtless thoughts are the dark matter of
tion to them, and they keep changing; no thought is pre- our minds, giving shape via hidden mechanisms to what
cisely like another. James thus doubts whether psychol- is observable, visible, luminous in our inner cosmos.

98  |  SCIENTIFIC AMERICAN  |  SPECIAL EDITION  |  WINTER 2022

THE LIMITS OF STREAM-OF-THOUGHT FICTION James defended his gobbledygookian work. “One great
C a n w e s t u dy t h o u g h t l e s s thoughts, the mind’s part of every human existence,” he told a friend, “is
dark matter, given that simple introspection doesn’t passed in a state which cannot be rendered sensible by
work? Some neuroscientists predict that external brain- the use of wideawake language, cutanddry grammar
scanning devices, such as MRIs, or arrays of implanted and goahead plot.” But even Finnegans Wake, a n unri-
electrodes will soon allow us to read minds. But this valed imagining of mental dark matter, consists entire-
feat would require cracking the neural code, the set of ly of Joyce’s hyperconscious, insanely erudite meta-
rules or algorithms that turn neural
activity into mental activity—that
is, thoughts. The neural code is the
enigma at the core of the mind-body WILL SCIENCE EVER DISCOVER A FINAL THEORY
problem. The more we investigate
it, the more intractable it seems. OF THE MIND? ONE THAT SOLVES THE MIND-BODY
I try to describe my thoughts in
my book P  ay Attention (original
PROBLEM AND MAKES US FULLY TRANSPARENT
title: W hat Is It Like to Be a Science TO OURSELVES?
Writer?). The book is based on jour-
nals in which I wrote down what I
did, saw, said, heard and thought over the course of a thoughts. And what about all the thoughts that cannot
typical day, as I commuted to the university where I be captured by words?
work, talked to a freshman humanities class (about
James’s “Stream of Thought”), jawed with colleagues HIDDEN-VARIABLE THEORIES OF THE MIND
over lunch (about Thomas Kuhn’s views of “truth”) and A final point: I see analogies between efforts to under-
spent the evening with “Emily,” my girlfriend. stand thoughts and the quantum realm. I alluded to one
In the first draft of my book, to make my thoughts correlation above: observing particles alters them, as does
seem more raw and real, I expressed them as sentence observing thoughts. Here’s another: some physicists, dis-
fragments running into each other, with little punctu- satisfied with probabilistic quantum accounts of electrons
ation. An editor who read this draft described it as and photons, seek to explain their behavior in terms of
“sludge.” Even I found that draft hard to read. So I “hidden variables” that follow deterministic rules.
cheated. I rewrote the book with more or less coherent Mind scientists, similarly, have proposed hidden-
sentences with conventional grammar and punctua- variable paradigms of the mind. Psychoanalysis holds
tion. I also added contextual information for readers, that our conscious minds are yanked this way and that
information that I wouldn’t have actually thought about by deep-rooted lusts and aversions. Evolutionary psy-
because I just knew it implicitly. chology traces our emotions and actions to instincts em-
To justify these moves in the direction of readabili- bedded in our ancestors by natural selection. Cognitive
ty, I could point out that the book’s narrator is an ex- science postulates that our thoughts stem from compu-
tremely self-conscious science writer trying to make his tations carried out by our neural machinery and are as
private thoughts public. He is in a sense performing his far removed from our conscious thoughts as the machine
thoughts, first for himself and then for readers. But that code of your smartphone is from the icons on its screen.
means my book consists of meta-thoughts. It isn’t an Although each of these paradigms has appealing fea-
accurate depiction of my thoughtless thoughts, which tures, each finally falls short, as do all theories of the
remain veiled from me. mind. Will science ever discover a final theory of the
Nobody depicts thoughts in all their raw weirdness mind? One that solves the mind-body problem and
as vividly as James Joyce. In U  lysses, Joyce plops us in- makes us fully transparent to ourselves? That reveals
side the heads of Stephen Dedalus, a teacher and aspir- the hidden variables underpinning and linking our
ing writer and an avatar for Joyce as a young man; Leo- meta-thoughts and thoughtless thoughts?
pold Bloom, a nerdy, genial ad salesman; Bloom’s vo- I doubt it. Physicists can’t grok the behavior of a sin-
luptuous wife, Molly; and other characters living in gle electron that is identical to every other electron. So
Dublin in the early 20th century. We see, feel, remem- what hope do we have of capturing the thought pass-
ber what they see, feel, remember. ing through your head right  . . . now, a thought unlike
But Joyce’s notoriously difficult masterpiece isn’t en- any that you, let alone anyone else, have ever had or will
tirely stream of thought. If it were, it would be far more have? And if we can’t grasp a single thought, which
difficult. To help orient us, give us a little context, Joyce melts the instant we grasp it, how can we possibly un-
occasionally shifts his point of view from inside char- derstand ourselves? Think about t hat. 
acters’ heads to outside, that is, from a first-person to a
third-person perspective. John Horgan d irects the Center for Science Writings at the Stevens Institute
Joyce’s final opus, F innegans Wake, w  hich I “read” of Technology. His books include T he End of Science, The End of War and Mind-
in college, makes no concessions to readability. Even Body Problems, a vailable for free at mindbodyproblems.com. For many years
Joyce’s admirers complained about its opacity, but he wrote the popular blog Cross Check for Scientific American.

SCIENTIFICAMERICAN.COM  |  99
Altered Reality
A Disorder of Mind and Brain page 102
The Undiscovered Illness page 110
Could COVID Delirium Bring On Dementia? page 116
A Psychedelic Renaissance page 122

Illustration by Maria Corte

SCIENTIFICAMERICAN.COM  |  101
 A

DISORDER
MIND
OF

BRAIN
AND

A mysterious condition once known as hysteria

is challenging the divide between
psychiatry and neurology
By Diana Kwon
Illustration by Vanessa Branchi
SCIENTIFICAMERICAN.COM  |  103
 such seemingly inexplicable illness can be found in
anyone and across the life span.
Generations of scientists have tried to understand

It all
these bizarre conditions, which have historically been
given diverse names, such as hysteria, conversion dis-
order or psychosomatic illness. These labels have,
however, long imposed particular explanations for
what many researchers now regard as a complex ill-

began
ness at the interface of psychiatry and neurology.
Some are still in use today, but the newest name for
these conditions, functional neurological disorder
(FND), is deliberately neutral, simply denoting a prob-
lem in the functioning of the nervous system.

with a
Patients with FND have long struggled to obtain
adequate care. They have been accused of feigning or
imagining symptoms, painfully but often fruitlessly
probed for childhood trauma and dismissed by doc-
tors who did not know how to treat someone who,

cough.
based on all the usual tests, appeared to be healthy.
“For many, many years physicians have underesti-
mated the prevalence of these disorders and the
human toll it takes,” says Kathrin LaFaver, a neurolo-
gist who specializes in movement disorders at Sara-
toga Hospital in New York State. “These people have
Four years ago Tracey McNiven, a Scottish woman really fallen [in the gap] between the fields of neurol-
in her mid-30s, caught a bad chest infection that left ogy and psychiatry.”
her with a persistent cough that refused to subside, Over the past decade or so, however, using tech-
even after medication. A few months later strange niques such as functional magnetic resonance imag-
symptoms started to appear. McNiven noticed numb- ing (fMRI), researchers have begun to understand
ness spreading through her legs and began to feel that what happens in the brains of patients with this enig-
their movement was out of her control. When she matic illness. And by applying new models of how the
walked, she felt like a marionette, with someone else brain works, they are gaining a better understanding
pulling the strings. Over the course of two weeks the of how the condition arises and how it may be treated.
odd loss of sensation progressively worsened. Then,
one evening at home, McNiven’s legs collapsed be­­ ENIGMATIC ILLNESSES
neath her. “I was lying there, and I felt like I couldn’t More than 3 , 0 0 0 years ago Mursili II, king of the
breathe,” she recalls. “I couldn’t feel below my waist.” Hittites, was caught in a terrifying thunderstorm. The
McNiven’s mother rushed her to the hospital where experience left him with a temporary speech impedi-
she remained for more than half a year. ment that went away—only to return several years lat-
During her first few weeks in the hospital, McNiv- er, after the monarch woke from a nightmare about
en endured a barrage of tests as doctors tried to un- the incident. His subjects attributed their king’s curi-
cover the cause of her symptoms. It could be a pro- ous ailment to the wrath of the Storm God, one of the
gressive neurodegenerative condition such as motor most important deities of the ancient civilization.
neuron d ­ isease, they thought. Or maybe it was multi- When modern-day scholars revisited the documents
ple sclerosis, a disease in which the body’s own im­­ detailing the event, they interpreted it as functional
mune cells attack the nervous system. Bafflingly, how- aphonia (the inability to speak).
ever, the brain scans, blood tests, spinal taps and ev- Like the Hittites, people throughout history have
erything else came back normal. turned to the supernatural—gods, witchcraft and de­­
McNiven’s predicament is not uncommon. Accord- mon­ic possession—to explain illnesses that today
ing to one of the most comprehensive assessments of would likely be diagnosed as FND. According to some
neurology clinics to date, roughly a third of patients historical interpretations, the first scientific attempt
have neurological symptoms that are deemed to be ei- to account for them emerged around 400  b.c.e., when
ther partially or entirely unexplained. These may in- Greek physicians, including Hippocrates, coined the
clude tremor, seizures, blindness, deafness, pain, pa- term “hysteria” to describe a wide collection of ail-
ralysis and coma and can parallel those of almost any ments, among them paralysis, headaches, dizziness
neurological disease. In some patients, such complica- and pain, in the belief that they were caused by the
tions can persist for years or even decades; some peo- uterus (hystera, i n Greek) wandering about the body.
ple require wheelchairs or cannot get out of bed. Al- Hysteria had its heyday in the 19th century, when
though women are more often diagnosed than men, it moved from the womb to the brain. Among several

104  |  SCIENTIFIC AMERICAN  |  SPECIAL EDITION  |  WINTER 2022

 physicians who studied it was Jean-Martin Charcot, from the realm of neurology to that of psychiatry and
regarded by many as the “father of neurology.” At the became the dogma for much of the 20th century. Per-
Sal­pêt­ri­ère Hospital in Paris, he painstakingly de­­tail­ haps coincidentally, as Freud’s influence in psychiatry
ed the symptoms of patients with hysteria and, after faded over the decades, so did cases of conversion dis-
they died, conducted autopsies on their brains. Al- order—to the point where some viewed it as a bygone
though Charcot was unable to identify any structural ailment of the Victorian era.
aberrations in those subjects, he was convinced that A century later a new generation of investigators
the impairments he saw were associated with unob- has turned its attention to this condition. Careful ob-
servable, fluctuating changes in the brain, which he servation of patients indicates that despite the drop in
called “dynamic or functional lesions.” diagnoses in the latter half of the 1900s, these disor-
Charcot also discovered that, contrary to common ders have not disappeared. And new research reveals
belief, male hysteria was not rare; instead it often that the condition encompasses both neurology and
went undetected. He highlighted, for example, cases psychiatry. In 2013 some physicians, concerned that
of hysteria among workers at a national railway com- the term “conversion disorder” was not widely ac­­cept­
pany that had seemingly emerged after minor physi- ed by patients and perhaps incorrectly pointed to psy-
cal injuries. His work popularized the study of hyste- chology as an exclusive driver for the condition, lob-
ria, inspiring several researchers, including Joseph bied for a change—causing FND to be included as an
Babinski, Pierre Janet and Sigmund Freud, to investi- alternative name for the ailment in the fifth edition of
gate it as well. the D
 iagnostic and Statistical Manual of Mental Dis-
Unlike Charcot, however, these men viewed the orders ( D
 SM-5).
condition as a disorder of the mind rather than the
brain. Freud proposed that it arose when repressed TELLTALE SIGNS
trauma from childhood abuse or other disturbing O n a c o o l , s u n n y a f t e r n o o n in February 2020,
events was transformed into physical symptoms; ac­­ I watched neurologist Jon Stone of the University of
cord­ing­ly, he called it conversion disorder. That view Edinburgh consult with first-time patients at the
and label cemented the displacement of the disorder Anne Rowling Regenerative Neurology Clinic, where

The Brain-Body Connection

Patients with physical symptoms (such as paralysis) but no appar- between areas involved in controlling movement and regions that
ent injuries may have functional neurological disorder (FND). affect attention and emotion, as shown in the cutaway. These
Neuroimaging has revealed subtle abnormalities in several brain linkages suggest a possible mechanism for the ailment. Activity
regions and networks. Studies find, for example, that functional in circuits associated with a sense of agency, such as the tempo-
connectivity—meaning correlations in activity—is heightened roparietal junction and its connections, may also be altered.

Motor cortex Motor cortex

Research Agenda,” by Susannah Pick et al., in J ournal of Neurology. Neurosurgery and Psychiatry, Vol. 90; 2019
Source: “Emotional Processing in Functional Neurological Disorder: A Review, Biopsychosocial Model and

Right hemispher Hypothalamus Left hemisp

here
Heightened
e
functional connectivity
Dorsal anterior Thalamus between brain areas is
cingulate cortex
apparent in structures
connected with
Perigenual
arrows
anterior
cingulate cortex
Dorsal anterior
cingulate cortex
Anterior
insular cortex Anterior
insular cortex
Perigenual
anterior
cingulate cortex

Amygdala Brain regions that are part of the salience

network (involved in finding and focusing on
Substantia nigra and attention-worthy information)
ventral tegmental area
Brain regions that are part of the limbic network
Hippocampus (one of the primary circuits that controls emotions)

Illustration by Body Scientific SCIENTIFICAMERICAN.COM  |  105

 his team specializes in diagnosing and treating FND. ceive the same level of sympathy, attention or care as
Stone listened carefully as people described when, those with so-called organic neurological illnesses.
where and how their symptoms started. He collected Stone decided to dig deeper. During his doctoral
detailed information about their medical and person- studies at the University of Edinburgh, he met Alan
al histories and conducted a neurological examina- Carson, who was training to become a psychiatrist and
tion. Then, like a detective, he pieced these details to- shared his interest. In 2002 the pair began to assess
gether to make a diagnosis. the scale of the problem by following the referrals to
In recent years FND has gone from a diagnosis of four neurology centers in Scotland over a period of 15
exclusion—a label doctors reserved for patients whose months. Their examination, which included more than
conditions defied all other explanations—to one made 3,700 patients, revealed that 1,144—close to a third—
had neurological symptoms deemed as
partially or completely medically unex-
CLOSE TO A THIRD OF PATIENTS plained. Of those, only four ended up be-

REFERRED TO NEUROLOGICAL ing diagnosed with another neurological

issue 18 months after their initial consul-
CLINICS IN SCOTLAND tation. This work demonstrated how wide­
spread these disorders were.
HAD SYMPTOMS THAT WERE Eventually Stone and Carson joined

MEDICALLY UNEXPLAINED. forces with Mark Hallett, a neurologist at

the National Institute of Neurological
Disorders and Stroke in the U.S., who had
after identifying distinct signs and symptoms. These also been taking strides to advance the field. In addi-
resemble those of other neurological disorders but tion to conducting his own research, Hallett had begun
possess identifiable differences. One example is mobilizing a group of scientists and physicians who
Hoover’s sign, in which weakness in a limb is tempo- could contribute to the study of FND. The community
rarily corrected when the patient’s attention is direct- gradually ballooned from several dozens of attendees
ed elsewhere. Another is tremor entrainment: when at a small workshop to a full-blown society for FND,
patients with a functional tremor in one arm are asked which was founded by Hallett, Carson and Stone and
to start shaking the other at a regular rhythm, the af- was inaugurated in 2019.
fected hand will start to shake with the same rhythm
as the other. This effect does not occur in people with PREDICTIONS GONE AWRY
neurodegenerative conditions such as Parkinson’s A y e a r a f t e r l a n d i n g in the hospital because of
disease. Clear signs of functional seizures include her FND, McNiven was referred to a psychologist. At
tightly shut eyes, rapid breathing and shaking that first, she did not think she needed to be there—her
lasts for several minutes—features rarely seen during symptoms had been improving with physiotherapy.
epileptic attacks. After several sessions of psychotherapy, however, Mc-
Such indicators have been known to doctors for de- Niven made a shocking discovery: she had blocked the
cades—Hoover’s sign, for one, was observed by physi- memory of certain key events in her childhood.
cian Charles Franklin Hoover in the 19th century. In Among those forgotten experiences were years of
the past, physicians would hide such signs from pa- physical abuse at the hands of a family friend.
tients, Stone explains. But he shows them to patients Although she had initially been reluctant to consid-
to help them understand the nature of their condition er the role of psychological factors in her illness, Mc-
and notes that physicians are increasingly taking up Niven now says that she thinks they do contribute. “I
this practice. Seeing such clues can help a person grasp don’t think it’s just purely down to that,” she tells me.
a condition that Stone likes to describe as a “software “But it certainly does have a big impact.” Many of those
problem, not a hardware problem” in the brain. who study FND today would agree—with caveats. Un-
Stone first came across these disorders as a junior like Freud, who focused on the role of repressed trauma,
doctor in the early 1990s. He found himself fascinated researchers now recognize that myriad factors are asso-
by them; having grown up with a stutter meant that ciated with this condition. These include predisposing
he had experienced being unable to control his own factors, such as adverse experiences during childhood,
body. And he was disturbed by how those with FND, a previous physical injury or mood and anxiety disor-
as he prefers to call the condition, were being treated. ders; triggers such as physical injury or a stressful life
The common attitude among medical professionals event; and maintaining factors, such as a lack of access
was that the symptoms were not real—at least not in to proper treatment or a patient’s responses to and be-
the same way as those seen in multiple sclerosis or liefs about the condition. The leading framework for
stroke, for example. Many physicians were concerned thinking about FND, the so-called biopsychosocial
that they would either fail to identify the true cause of model, takes all these factors into account.
a patient’s illness or be fooled by someone faking their As yet, there is no single, widely accepted explana-
symptoms. As a result, patients with FND did not re­­ tion for how these influences come together to create

106  |  SCIENTIFIC AMERICAN  |  SPECIAL EDITION  |  WINTER 2022

 FND, but some scientists have suggested that the
malady involves arguably one of the most funda-
mental functions of the brain: predictive process-
ing. Championed by neuroscientist Karl Friston of
University College London, predictive processing
posits that the brain is constantly making and eval-
uating predictions by comparing the data generat-
ed from our sensory organs to internal models
built from previous experience. When mismatches
occur between inferences and reality, the brain ei-
ther updates its existing models or sends com-
mands back down to the body to act in ways that
align with our expectations. For instance, if you
want to walk, but your leg is not moving, the brain
will generate a prediction error that can be re-
solved if you move that leg. In this way, Friston and
others propose, predictions underlie everything
the brain does, from perception to movement to
decision-making.
Neurologist Mark Edwards of St. George’s Uni-
versity Hospital in London and his colleagues have
suggested that with FND, this predictive machin- Resignation Syndrome
ery goes awry, so that patients develop abnormal
inferences of how their body should feel or func- Starting in the 1990s, reports of a mysterious illness began to emerge
tion. One of the biggest drivers of this anomaly, ac- from Sweden. Children started showing up in emergency rooms in a
cording to Edwards, is excessive focus on one’s own comalike state—immobile, mute, unable to eat or drink, and unrespon-
body. This heightened attention can be attributed sive, even to pain—with no identifiable medical cause, and they re­­
to a variety of factors, including an existing physi- mained in this state for months, sometimes years. The patients had some
cal illness, mood and anxiety disorders, or child- things in common: they were from families of refugees, most of whom
hood abuse. When a person experiences a trigger- belonged to ethnic or religious minority groups from former Soviet or
ing event—say, an injury to a limb or a panic at- Yugoslav states. And in many, the trigger for their illness appeared to be
tack—this heightened attentiveness may drive the rejection of an application for asylum.
one’s brain to develop altered predictions about the Hundreds of cases of the unexplainable illness had been reported by
body. In some cases, a past experience, such as ex- the early 2000s. As the number of afflicted children rose, the nature of
posure to sickness in the family, might also help the illness became a subject of intense debate. Some opined that the
shape these expectations. children were faking or that parents were inducing the condition to
Consider someone who falls and badly sprains a obtain a residence permit—suggestions that, despite sparking outrage
leg, resulting in a temporary loss of mobility in that among both clinicians and the public, continue to circulate today. In 2014
limb. In most people, the brain’s predictions about the Swedish National Board of Health recognized it as a novel condition,
the injured leg’s ability to move would get updated Uppgivenhetssyndrom ( “resignation syndrome”). Others said it was a
once mobility returns. This person, however, has a manifestation of a known illness, such as severe depression, catatonia or
tendency toward mild anxiety that amplifies the conversion disorder.
levels of subconscious attention they pay to their Karl Sallin, a pediatrician at Karolinska University Hospital in Sweden,
body and has been sensitive to health events since and his colleagues have proposed that the condition is a culture-bound
the sudden death of a parent. These predisposing functional neurological disorder. They suggest that resignation syndrome
factors magnify the sensations associated with inju- arises when factors such as prior psychological or physical trauma, loss
ry; in consequence, the internal model of the immo- of hope that asylum will be granted and fear of being deported combine
bile leg persists even after the limb regains its func- with culturally specific beliefs to subconsciously prescribe how the body
tion, leading to functional paralysis. (In some ways, should respond in the face of extreme external stress. Sallin notes that
this is the opposite of what happens in people who the apparent specificity of this condition, which is limited to certain refu-
experience phantom limb syndrome. Those people gee communities in Sweden, suggests that the illness is influenced
are unable to update the prediction error that oc- strongly by beliefs prevalent in a particular group. (Reports of a similar
curs when an expected sensation in a missing limb condition among refugees waiting for asylum in Australia on the tiny
is not met with actual sensory feedback.) island of Nauru have also emerged, however.) As yet, experts do not
The hypothesis that predictive processing is al- agree on what these disorders are.
tered in FND patients has now been tested in a The number of cases of resignation syndrome has decreased since
Magnus Wennman

handful of experiments. In a 2014 study, for exam- the peak in the early 2000s. Even so, hundreds of children have been
ple, Edwards and his team used a task called force diagnosed with the condition in recent years.  —D.K.
matching, in which a robotic device presses down

SCIENTIFICAMERICAN.COM  |  107
 on a finger and people are asked to match the force set of people with FND, the emotional circuitry might
with their own hand. Healthy people tend to overesti- be hijacking the motor system, Perez explains.
mate the force required by their own hand because Perez’s team has also found that some risk factors
the brain’s expectations “cancel out” some of its force may map onto these circuits. In a study published in
(a similar explanation applies to why you cannot tick- 2020, his group reported that the magnitude of the
le yourself ). People with FND, on the other hand, coupling of the motor regions with the limbic and sa-
were abnormally accurate, indicating that the internal lience areas of the brain positively correlated with the
prediction system was functioning differently. Even degree to which patients experienced physical abuse
so, much more evidence is needed to prove that this during childhood. Perez emphasizes, however, that
mechanism provides a correct and sufficient explana- this will probably be relevant only to the subset of pa-
tion for the condition. tients in whom trauma is present: in his study, a sig-
nificant proportion of patients did
not report any childhood physical
USING NONINVASIVE PROBES, abuse. Still, he notes that these find-

RESEARCHERS ARE FINDING SUBTLE ings point to how a risk factor such
as trauma could alter brain circuits
DIFFERENCES IN THE BRAINS OF in people who develop FND.
Scientists are also investigating
INDIVIDUALS WITH FUNCTIONAL how factors such as stress alter

NEUROLOGICAL DISORDER. brain circuits in FND. Neurologist

Selma Aybek of the University of
Bern says that although not all pa-
PROBING THE BRAIN tients have a history of trauma or stress, they may pos-
L i k e C h a r c o t, c ontemporary investigators of FND sess differences in their biological stress response.
have been examining the brains of patients to find Her group has found that, compared with healthy in-
changes associated with the condition. Modern scien- dividuals, FND patients have higher levels of the
tists, however, no longer have to wait to conduct an stress markers cortisol and amylase and report being
autopsy to peer into their subjects’ skulls. Using tech- more stressed after taking part in a stressful task.
niques such as fMRI, researchers have begun to reveal Building on these findings, her team is using neuro­
there are indeed differences in the brains of individu- imag­ing to examine whether there is an association
als with FND. “We’re beginning to identify the dy­­nam­ between stress-related regions and agency-related re-
ic lesion that Charcot was looking for,” says David gions in FND patients’ brains.
Perez, a neurologist-psychiatrist at Massachusetts Thus, a picture of the pathophysiology of FND pa-
General Hospital. tients is slowly emerging. But most of this work has
With fMRI, researchers have identified distinct been conducted in patients with motor symptoms,
patterns of activity in brain areas such as the temporo- which means that sensory symptoms such as altered
parietal junction—associated with a sense of agency— vision have yet to be explored. Many of these studies
in those with FND, compared with those asked to also have had small sample sizes, so findings will need
mimic the same symptoms. These findings help to to be validated in larger trials, says Valerie Voon, a
confirm that unlike conditions such as factitious dis- neuro­psychiatrist at the University of Cambridge,
order (a severe form of which is known as Munchau- who collaborated with Hallett on several ground-
sen syndrome), in which patients deliberately act out breaking projects. How these neuroimaging findings
other illnesses, symptoms in individuals with FND are fit with the predictive-processing model also remains
out of their control. an open question. It is plausible, Perez notes, that
Another significant discovery from neuroimaging many of the areas identified so far may be the circuit-
is that people with FND have enhanced connectivity ry through which the altered predictions arise.
between the motor-control regions and two brain net-
works involved in emotional processing: the salience A BRIGHTER FUTURE
network, responsible for detecting and focusing on at- In the summer a fter her second year of teacher’s col-
tention-worthy information, and the limbic network, lege in Scotland, a 19-year-old woman named Rachael
one of the primary systems controlling emotion. In a Troup was rushed to the hospital with what appeared to
2010 study, for example, Hallett’s group reported be a stroke. Brain scans showed that she did not have a
heightened linkages between the amygdala, a key re- stroke, however, and tests for other neurological diseas-
gion in the limbic system, and the supplementary mo- es came back normal. Eventually Troup was diagnosed
tor area, which is responsible for preparing to initiate with FND. But when she started treatment, it was ex-
movements. Others, such as Perez, have shown hyper- cruciating. Neither her doctors nor her physiothera-
connectivity between motor regions and salience net- pists seemed to know much about how to treat her con-
work areas such as the insula and the anterior cingu- dition, and the exercises they made her do hurt more
late. These observations suggest that, at least in a sub- than they helped. “I was in pain constantly,” she says.

108  |  SCIENTIFIC AMERICAN  |  SPECIAL EDITION  |  WINTER 2022

 After a few months Troup decided to stop going to ter their initial diagnosis. On top of that, patients still
physiotherapy. At the time the entire right side of her experience high levels of stigma and have trouble ac-
body was barely functioning, and she was using a cessing treatment, La­Faver says. “I still think we have
wheelchair for mobility. After being admitted to the a long way to go.”
hospital several more times for strokelike attacks, Fortunately, the situation is changing. As research-
however, Troup met Stone’s team and was provided ers’ interest in FND surged over the past decade, so did
with FND-tailored care. It involved a form of physio- the number of FND clinics around the world. Patients
therapy that employs techniques such as distraction are speaking up as well. In 2012, for example, the inter-
to shift the spotlight of attention away from the affect- national charity FND Hope was established with the
ed limbs while engaging in exercises to help restore aim of raising awareness and empowering patients.
normal control. Still, debates linger—and are reflected in an ongo-
For FND patients, shifting focus away from af­­ ing tussle over the name of the illness. In what frac-
fected limbs is often a crucial part of physiotherapy tion of patients are psychological factors present, for
be­­­­cause, as Edwards’s predictive-processing model sug­­ example? Do symptoms primarily arise through con-
gests, attention is critical to the generation of symp- version of stressors, or are other explanations also
toms. With attention deployed elsewhere, the brain’s necessary? One meta-analysis found that reports of
abnormal expectations about movement are unable to stressors in FND patients vary between 23 and 86 per-
take hold. Stone and his colleagues are part of an on- cent in different studies. W. Curt La­France, Jr., a neu-
going U.K.-wide, randomized controlled clinical trial rologist-psychiatrist at Brown University, says that it
testing this type of specialized physiotherapy for func- takes more time to identify such stressors than some
tional motor disorders (a subset of FND affecting doctors can devote to an individual patient—which
movement). In addition to retraining movement, the may account for this enormous spread. In his clinic
treatment includes educating patients about how and in the scientific literature, he has consistently
such symptoms could arise and the physical and psy- seen evidence of psychological stressors having con-
chological factors that may underlie it. verted into physical symptoms, and, accordingly, he
To expand the tool kit of interventions for FND, re- supports use of the term “conversion disorder.” Par-
searchers are also testing other alternatives. Another tially re­­­flecting such views, the older name remained
large clinical trial with more than 300 patients as- when FND was added to the D  SM-5, b
 ut the need to
sessed the efficacy of cognitive-behavioral therapy identify psychological factors for a diagnosis was
(CBT)—a type of goal-oriented intervention focused dropped. That excision also remains contentious.
on changing disruptive patterns of thinking or behav- One thing is clear, however: because the condition
ior—for functional seizures. The findings, published lies at the crossroads of neurology and psychiatry, in-
in June 2021 in L  ancet Psychiatry, suggest that CBT sights from both fields will be necessary to solve the
may not reduce seizure frequency in all patients. puzzle. This blurring of the line between mental and
At King’s College London, neuropsychiatrist Tim physical illness is a growing trend. Scientists now un-
Nicholson and his team are examining a noninvasive derstand, for example, that stress—a psychological
method of exciting the brain known as transcranial factor—can predispose people to Alzheimer’s disease
magnetic stimulation (TMS) as a potential interven- and that in­­flam­ma­tion—a physical factor—may give
tion for FND. His group completed a feasibility study, rise to depression. In addition, traditional neurologi-
and the results were promising enough for them to cal diseases such as epilepsy and stroke are often asso-
initiate a larger pilot clinical trial. There are compet- ciated with mood and behavioral disturbances. “The
ing explanations for why TMS might work. It induces brain doesn’t separate into neurology and psychiatry,”
a brief muscle twitch that could kickstart the relearn- Perez says. “We need a new science of brain and mind
ing of movement; stimulating brain areas altered in that really encapsulates that brain health equals men-
FND might help re­­store function, or it may have a tal health and physical health.”
placebo effect. LaFaver, working with colleagues at Over the past few years McNiven has frequently
Northwestern University, has examined the use of used a wheelchair because of her FND. But with the
meditation and mindfulness practice, which, she says, help of both physiotherapy and psychotherapy, she
patients have anecdotally reported as helpful for has started to recover. Her symptoms are not gone—
maintaining the benefits of treatment. she still experiences a lack of sensation in her legs, al-
Psychological treatments such as CBT currently re- tered vision and pain—and some days are worse than
main among the first-line interventions for people others. “I constantly feel like I’m fighting against my
with FND, according to Perez. There is a pressing need body,” McNiven says. But she hopes to make a full re-
for a range of effective treatments, however: the prog- covery. “You’ve got to keep that positive attitude to
nosis remains poor. It is still relatively uncommon for keep fighting through it. There’s always hope you can
FND patients to completely regain function, and re- get there with this ­condition.” 
lapses occur often. According to a 2014 meta-analysis
of 24 studies, on average 40  percent of patients re- Diana Kwon is a freelance journalist who covers health and the life sciences.
ported similar or worsened symptoms seven years af- She is based in Berlin.

SCIENTIFICAMERICAN.COM  |  109
110  |  SCIENTIFIC AMERICAN  |  SPECIAL EDITION  |  WINTER 2022
 ILLNESS
UNDISCOVERE
Hundreds of thousands
of people experience mania
without ever getting
depressed. Why does
psychiatry insist on calling
them bipolar?
By Simon Makin

Photography by The Voorhes

HE

SCIENTIFICAMERICAN.COM  |  111
 placed to advance this debate. In 2016 he published a self-study
in the journal Psychosis c ataloguing his symptoms, which in-
clude enhanced recall, increased empathy and spiritual experi-
ences. He has suffered some ill effects, including severe fa­­tigue,

IN OCTOBER
confusion and behavior that caused concern among friends and
colleagues: he once burst into tears while delivering a lecture.
But his professional training has helped him control his im-
pulses and avoid delusional thinking. On balance, he believes

1997,
that his madness, as he calls it, has enriched rather than dam-
aged his life. “I’m aware my case may be atypical,” Ho says. “Pre-
at the age of 58, David Ho had cisely for this reason, it challenges prevailing psychiatric beliefs
an unusual experience while that fail to acknowledge the positive value of mental disorders.”

listening to a recording of Bach. “I began to A MODERN ILLNESS

dance and pretended to conduct,” he says. Credit for the modern concep t o  f bipolar disorder usu-
“And as I practiced, instead of following the ally goes to 19th-century French psychiatrist Jean-Pierre Falret,
who called it folie circulaire, o
 r “circular insanity,” for its peri-
music, I felt as if I were creating it. I entered ods of pathologically elevated and depressed moods, usually
into a state of selfless oblivion, like a trance. separated by symptom-free periods of varying length. This idea
My mind exploded. Flashes of insight rained became gospel in the early 20th century, when a father of mod-
ern psychiatry, Emil Kraepelin, proposed a historically signifi-
down, and I saw beauty everywhere, in faces, cant hypothesis.
living things and the cosmos. I became disin- At the time, psychiatry drew a distinction between so-called
hibited, spontaneous, liberated.” reactive psychoses, which were seen as a response to outside
events, and endogenous psychoses, which were innate. Kraepe-
Ho was in the grips of his first episode of mania. His descrip- lin divided all endogenous psychoses into two broad classes:
tion sounds like an enviable burst of creative energy, but the symp- dementia praecox—now known as schizophrenia—and manic-
toms of mania can also include inflated self-esteem, grandiosity, depressive insanity, now known as bipolar disorder. Endoge-
racing thoughts, extreme talkativeness, decreased need for sleep, nous depression was therefore classed as a form of manic-
increased activity or agitation, reckless behavior, delusions and depressive insanity. All mania also fell under the same rubric
other psychotic events. Severe episodes can impair day-to-day because mania was thought never to be a reaction to outside
functions, sometimes enough to require hospitalization. events. There were dissenters, notably the renowned German
Perhaps the most surprising thing about such cases is that neurologist Carl Wernicke, who held that mania was related to
in the eyes of the psychiatric profession, mania does not exist hyperactivity of neural firing and depression to decreased neu-
as a distinct and unalloyed condition. Mania is usually known ral activity. But Kraepelin’s idea dominated and persists in to-
as the upside of bipolar disorder. For most people, it occurs day’s diagnostic system.
alongside periods of depression, the downside. But Ho, who The question of what to include under the umbrella of bipo-
has had at least 20 manic episodes since 1997, has never suf- lar disorder reignited in 1966. In separate investigations, psy-
fered from depression. Thousands of people in the U.S. share chiatrists Carlo Perris of Umeå University in Sweden and Jules
that experience. Un­­like those who experience only depression, Angst of the University of Zurich in Switzerland each studied
however, patients with mania alone are lumped with those who some 300 patients with either true bipolar disorder or depres-
have bipolar disorder. This puts psychiatry in the strange posi- sion alone and more than 2,000 of their close relatives.
tion of claiming that depression by itself is different from de- Both researchers found that relatives of the bipolar patients
pression accompanied by mania but that mania by itself is not. had more mood disorders than those of patients with depres-
Most psychiatrists agree unipolar mania exists, but there is sion alone. They also discovered that although bipolar illness
debate about whether it differs sufficiently from bipolar disor- was common in the relatives of bipolar patients, it was no
der in important enough ways to warrant a distinct diagnosis. more common in relatives of depressed patients than in the
Central to that debate is the tension in psychiatry between general population. These findings, Perris and Angst argued,
fewer, broader categories and more numerous, tightly defined suggested that bipolar disorder and depression were geneti-
ones. But the missing diagnosis may have consequences for pa- cally different conditions.
tients: some studies suggest that people with unipolar mania As a consequence, when the third edition of the D  iagnostic
may respond differently to certain treatments. If, as some re- and Statistical Manual of Mental Disorders, o rD
 SM, a ppeared
searchers believe, uni­polar mania and bipolar disorder differ in in 1980, it included major depressive disorder as a condition
their underlying biology, classifying mania separately could distinct from bipolar disorder. Perris and Angst’s studies fo-
speed the development of new treatments that are more per- cused only on depression and did not address mania. “There
sonalized and effective. But because unipolar mania is far less weren’t enough cases of pure mania to do anything reasonable,”
common than bipolar disorder, research into the condition has Angst says.
been both scant and equivocal. Whether unipolar mania should have its own diagnosis is
As both a patient and a clinical psychologist, Ho is well complicated by bipolar disorder’s clinical diversity. The manic

112  |  SCIENTIFIC AMERICAN  |  SPECIAL EDITION  |  WINTER 2022

and depressive phases vary in severity and the extent that one or symptoms necessary for a case to qualify as mania. Another is the
the other dominates. The pattern of episodes varies unpredict- frequency of episodes. Some studies include anybody who has
ably and from patient to patient. Mixed states, in­­volving aspects had at least one episode of mania with no history of depression,
of opposite mood extremes simultaneously, sometimes occur, whereas others require three or four. Still others stipulate a mini-
too. Indeed, many psychiatrists argue that mood disorders are mum number of years of illness. These differences have led to
best thought of as lying on a spectrum, ranging from major de- widely disparate prevalence estimates for unipolar mania, rang-
pression through various bipolar presentations to pure mania. ing from 1.1 to 65.3 percent of patients with bipolar disorder.
Most of the studies completed so far also have methodologi-
IN SEARCH OF A SUBTYPE cal problems. The bulk are retrospective, in which researchers
T he variabilit y of symp toms, along with findings from simply ask participants to recount past experiences—a process
large psychiatric genetics studies that implicate numerous bio- known to underestimate depression, perhaps inflating esti-
logical factors, suggests that bipolar disorder includes a range mates of pure mania. Prospective studies that follow patients for
of subtly different conditions. “One reason we still have limited years and include periodic assessments are better. “What you re-
understanding of bipolar disorder after 50 years of intense re- ally want is someone who’s lived their whole life, had multiple
search is that it’s treated as one entity, and it’s clearly not,” says episodes of mania, and never had depression,” Young says. “The
psychiatrist Paul Grof of the University of Toronto. first lady I saw like this died in her late 60s and had her first ep-
The resistance to subtyping may be isode at 21, which is getting on for 50
the result in part of changes in research years, so that’s very convincing.”
funding over the past few decades, as
the pharmaceutical industry has taken
MANIA IS USUALLY One of the longest prospective stud-
ies, led by David Solomon, now profes-
over progressively more psychiatric re­­ KNOWN AS THE sor emeritus at Brown University, be-
search from universities, Grof says. Drug
companies generally just want to know UPSIDE OF BIPOLAR gan in 1978 and was published in
2003. It began as a study of 229 bipo-
if a new drug is better than a placebo,
and the larger the pa­­tient group, the
DISORDER. FOR MOST lar patients, 27 of whom had mania
with no history of depression. The in-
greater the likelihood of finding a sig- PEOPLE, IT OCCURS vestigators followed those 27 patients
nificant difference. Subdividing bipolar
disorder into smaller populations would ALONGSIDE PERIODS for up to 20 years; seven of them re-
mained free of depression throughout
complicate these ef­­forts. The industry
also prefers to study diagnoses recog-
OF DEPRESSION, the period. The results suggest that of
the original 229 patients, 3 percent had
nized by the Food and Drug Admin­ THE DOWNSIDE. unipolar mania. Solomon does not ad-
istration—and unipolar mania is not vocate the creation of a separate diag-
on its list. nosis for uni­polar mania unless future
Institutional inertia can also come into play. Every rewrite of research establishes differences in genesis, prognosis or treat-
the D iagnostic and Statistical Manual of Mental Disorders is a ment response. But if the rate reported in the study held for the
laborious process. Each edition is based on the previous one, and general population, the number of people with unipolar mania
any change must be backed by fresh evidence, with papers sub- in the U.S. would be around 100,000—and there would be hun-
mitted to committees justifying the decision. The last edition, dreds of thousands more worldwide.
DSM-5, was published in 2013, and in the view of the committee The stories of people with unipolar mania help to explain
tasked with reviewing mood disorders, unipolar mania was cov- why some researchers are convinced that the disorder is a sepa-
ered by the bipolar diagnosis known as BP-I, which is mania rate entity. Lindsey, a ski coach from Portland, Me., is one such
with or without associated depression. “There was very limited case. She was 18 when she had her first experience of mania.
discussion as to whether mania should be separate be­­cause the Eighteen years later she has never been depressed, yet she still
onset and course of illness weren’t seen as that different from has a diagnosis of bipolar disorder. “I’m the happiest person I
BP-I,” says psychiatrist Trisha Suppes of Stanford University, who know,” she says. “I never accepted my diagnosis.” As a result, she
was a member of the DSM-5 work group for mood disorders. rejected treatment and continued to have episodes. She has been
The lack of a separate diagnosis may be making evidence hospitalized five times and has landed in jail more than once.
harder to gather. The standardized clinical interview used un- Lindsey’s episodes start with euphoria but can spiral into
der the DSM to make diagnoses for research studies has no cat- delusions and difficulty speaking. While manic, she feels no fa-
egory for unipolar mania, meaning investigations of the condi- tigue, hunger or pain. One such episode, in her late 20s, began
tion would have to rely on ad hoc techniques that might not on a hike in New Mexico when she was overcome by a vision
align with those used in other studies. Unipolar mania is thus that the world was coming to an end. Lindsey called her father,
at the hub of a catch-22: the absence of a diagnosis is an imped- who flew out to meet her and drive her home to Maine. “She
iment to research, and the paucity of research makes the cre- had medication,” her father says. “She just wasn’t taking it.”
ation of a diagnosis less likely. Early in the morning on an overnight stop in Nashville, Lindsey
In studies that do occur, the lack of a formal designation for started playing the piano in the hotel lobby. An employee called
unipolar mania makes it difficult to compare results. “A major the police, and Lindsey fled in the car.
problem is definitions,” says Allan Young, a psychiatrist at King’s In the adventure that followed, she deliberately got lost, bur-
College London. One source of disagreement is the severity of ied her possessions near a railroad track and abandoned the car.

SCIENTIFICAMERICAN.COM  |  113
She then hopped a freight train, got off in the
middle of rural Tennessee, climbed out of a
Psychic Fuel for
rock-walled valley and wandered into a chapel,
where the pastor was able to glean enough in­­
the Creative Brain
formation to contact her family. Shortly after re- The mad genius may be more than a cliché
suming the drive home, Lindsey ran away from
Of all the tropes of artists and mental afflictions, the most enduring is the one
her father at a highway rest stop and started
of a genius in the throes of mania. Iconic figures ranging from William Blake
picking daisies in a fenced-off electrical area.
to Ernest Hemingway to Kurt Cobain were known or believed to have bipolar
The police were called again, and although the
disorder. The association has intuitive appeal: the euphoria, abundant energy
officer urged her to leave with her father, she in-
and racing thoughts of mania are credible fuel for creativity.
sisted on being arrested.
Scientific evidence for the association has mostly been inconclusive. Much
In her cell, a guard pepper-sprayed her, and
of the data comes from historical sources, and most accounts are anecdotal.
she ended up in the office of the jail’s counselor.
Modern investigative techniques have revealed surprisingly little about what
Lindsey was barely able to speak at this point,
happens in the brain during mania, partly because brain imaging requires minimal
but she wrote “unipolar” repeatedly on a black-
head movement, so scanning someone in a floridly manic state is a challenge.
board. The counselor then read Lindsey a de-
As a dynamic process involving the interplay of multiple brain networks, creativity
scription of mania. She credits this encounter
is also difficult to research.
as the moment she accepted the need to take
But comparing findings from research into bipolar disorder with certain
medication. The counselor gave her Zyprexa
studies of creativity reveals hints of a link: cognitive “disinhibition” seems to be
(olanzapine), an antipsychotic. She recovered
a feature of both the creative state described as being in the “flow” and altered
and takes it to this day, though not without res-
brain circuits in bipolar disorder.
ervations. “My medication is like a dose of sad-
Brain-imaging studies have found reduced activity in a part of the prefrontal
ness, hunger, fatigue and pain,” she says. Lind-
cortex that helps to regulate emotion, which may be linked to impaired impulse
sey was euphoric throughout her ordeal, even
control and extremes of mood in people with bipolar disorder. (The prefrontal
while being pepper-sprayed. Only the people
cortex is the brain’s “orchestra conductor” responsible for directing various
around her suffer. “I feel like I’ve been blessed
mental processes.) Some of these studies have also found diminished activity
with this illness that makes me so happy,” she
in an area involved in suppressing the kind of spontaneous thought that appears
says, “but I feel selfish because of how it affects
to well up from the unconscious, seemingly out of nowhere.
my family.”
These results are reminiscent of a 2008 study of improvising jazz musicians
Lindsey married Andy, a journalist, in 2015,
and a 2012 study of freestyling rappers, conducted by the team of speech
not long after he witnessed her last hospital-
neuro­scientist Allen Braun, then at the National Institutes of Health, which
ization. “It made the relationship stronger in
found reduced activity in the part of the prefrontal cortex that inhibits
the end,” he says. “I got to see her as she clawed
spontaneous cognition. The researchers also found an increase in activity
her way back to sanity. It was impressive.” The
in a section of the prefrontal cortex that is part of the so-called default mode
most im­­ portant factor in her treatment is
network, which revs up when a person is not focusing on a task but rather
whether a physician accepts that she is not bi-
polar. “When that’s ignored, she no longer
trusts that person,” Andy says.

IT ALL GETS REAL convinced the geographical differences are genetic in origin and
A c u r i o us q u i r k i n the tale of this neglected disorder is that that unipolar mania therefore represents a distinct condition.
prevalence estimates vary worldwide and are consistently Getting to the bottom of this question will require large,
higher in non-Western countries. After qualifying in South Af- multicultural international studies. In the meantime, scientists
rica in 1997, psychiatrist Christoffel Grobler worked in an inpa- are trying to compensate for a shortage of data. One reason
tient unit in Ireland, where his bipolar patients were mostly in most early studies failed to find differences between mania and
depressed states. When Grobler re­­turned to South Africa in bipolar disorder may be that they are so slight that they can be
2009, he noticed the opposite pattern: his patients were mostly reliably detected only in large samples. Now in his 90s, Angst
in manic states. To investigate, in 2010 he and his colleagues addressed this problem by consolidating data from nine epide-
interviewed 103 bipolar patients in three hospitals, using a miological studies conducted in the U.S., Germany, Switzerland,
standard diagnostic questionnaire. They found that 32 percent Brazil and Holland. That study, published online in November
of patients qualified as unipolar, defined as having at least five 2018 in Bipolar Disorders, found that people with unipolar ma-
manic episodes over four or more years. “When I present this nia were more likely to be male but less likely to have attempted
at conferences, people come up and say, ‘We see this all the suicide or to suffer from anxiety, drug use and eating disorders.
time,’ ” Grobler says. Angst and his colleagues claim these findings suggest unipolar
Regional variations are tricky to interpret because cultural mania “should be established as a separate diagnosis.”
differences come into play: depression is more likely to be con- Some of these findings align with three reviews of research
sidered part of normal life in Africa, for example. The quality and on unipolar mania published in the past seven years. All three
procedures of health-care systems differ, and other causes, in- found that unipolar mania is less likely to co-occur with anxiety
cluding infection or intoxication, may be a factor. But Grobler is (which often accompanies depression) but more likely to come

114  |  SCIENTIFIC AMERICAN  |  SPECIAL EDITION  |  WINTER 2022

 whether the majority of their episodes were
is imagining things or ruminating on the past. They believe what they observed
manic or depressive and then created a further
reflects relaxation of focused attention and control, making way for a creative
division of patients whose manic episodes ac-
thought process in which inspiration bubbles up from the unconscious. Other
counted for at least 80 percent of the total. A
studies have found reduced thickness of certain cortical regions in both cre­
smaller percentage of patients who had at
ative and bipolar brains, which may be linked to altered brain activity and
least a majority of manic episodes responded
disinhibited cognition.
to lithium than among patients who had more
Another element in the thinking patterns of creative and manic people is
depressive episodes, and this difference was
the ability to make mental connections that elude others. Neuroscientist Nancy
greater for patients whose mania put them in
Andreasen of the University of Iowa has found that creative people show
the 80 percent group. Most telling, when those
greater activity in the so-called association cortices, which are regions tasked
with unipolar mania were excluded from this
with linking related elements of cognition. These brain areas are not devoted
analysis, these differences disappeared, sug-
to processing specific sensory or motor functions but instead engage with tasks
gesting the treatment difference relates to uni-
such as tying together a written word with its sound and meaning. Andreasen
polar mania and not to dominant manic polar-
believes creative ideas probably happen when these types of associations occur
ity and thus implying that unipolar mania is
freely in the brain during unconscious mental states, when thoughts become
its own entity.
momentarily disorganized—not unlike psychotic states of mania.
This observation resonates with clinical psychologist David Ho, who has
THE WAY AHEAD
experienced racing thoughts and extraordinarily enhanced recall during manic
T hose who are opposed to a separate di-
episodes, letting him write without inhibition or self-doubt. “With repression
agnosis sometimes point out that the mania in
vanished, my mind functioned with holistic oneness,” he says. “Creative
unipolar mania is indistinguishable from that
ideas rained down faster than I could cope.” Researchers do not know if the
in bipolar disorder. But the same is true of
association cortices are more active in mania, but all these findings suggest
depression, and many studies have found dif-
that at key moments of the creative process, our thought processes flow more
ferences in the brains of people with major de-
freely, with novel combinations of sights, sounds, memories, meanings and
pression compared with those of individuals
feelings producing insight and originality in creative work akin perhaps to what
with bipolar disorder. Future work that com-
happens during mania.
pares brains of people with unipolar mania and
Of course, mental illness is neither necessary nor sufficient for creative
bipolar disorder might be just as revealing.
talent, and severe manic episodes most likely are too debilitating for any kind
Biological and brain-imaging studies of un-
of sustained activity. But researchers have found that family members of people
ipolar mania are rare. But one from several de-
with bipolar disorder also tend to be more creative than average, supporting
cades ago gives clues to differences in physiol-
the idea that mild manifestations of the disorder may furnish cognitive benefits.
ogy. A 1992 CT scan study led by Sukdeb
It is important not to romanticize conditions that mainly cause suffering,
Mukherjee of the Medical College of Georgia
but evidence that mania can enhance creativity in some people may help
at Augusta University found that unipolar ma-
reduce the stigma of a diagnosis. “It is possible to retain a measure of madness
nia patients had smaller third ventricles (one
in dignified living,” Ho says, “and of dignity even in a state of madness.”  —S.M.
of four interconnected cavities in the brain
that let cerebrospinal fluid flow) than bipolar
patients did.
This result is intriguing because subse-
with psychotic symptoms. Unipolar mania also seems to confer quent studies found that bipolar patients who experienced
less social impairment and involve fewer recurrences and bet- multiple episodes have larger ventricles than people who are
ter remission rates than bipolar disorder. experiencing their first episode or healthy control subjects, a
Perhaps most important, people with unipolar mania show hint that en­­larged ventricles may be linked to pathology. The
subtle differences in their re­­sponse to drugs administered as implication that unipolar mania may not cause as much dam-
part of preventive treatment. Three studies found that patients age in the brain tallies with the better outcomes associated
with unipolar mania re­­spond less well to lithium (a mood stabi- with the condition.
lizer and first option for bipolar) than those with true bipolar Creating a separate diagnosis for unipolar mania remains
disorder do. controversial. An interim step would be to recognize it as an of-
One of these studies, published in 2012 by Olcay Yazici and ficial subtype of bipolar disorder. Such a move might encourage
Sibel Çakir, then both at Istanbul University, also examined the research and raise awareness among clinicians. “There’s a mys-
question of whether unipolar mania is merely bipolar disorder tery here we don’t understand: Why do some people get mania
weighted to the manic end of the spectrum—so-called domi- and then depression, whereas others stay unipolar manic?”
nant manic polarity. They divided 121 patients into two groups, Suppes asks. “It’s deserving of more research than it’s gotten so
34 with unipolar mania and 87 with classic bipolar disorder. As far.” Further investigation might also benefit patients who do
the earlier studies found, the unipolar group had a lower re- not identify with other labels. Lindsey pleads, “The most im-
sponse rate to lithium, and their response to another frontline portant thing my doctor could do for me is say, ‘I’m sorry, we
bipolar treatment, the anticonvulsant Depakote (divalproex so- were wrong—you’re not bipolar, you’re ­unipolar.’ ” 
dium), was no different.
The researchers next grouped all 121 patients according to Simon Makin is a freelance science writer based in London.

SCIENTIFICAMERICAN.COM  |  115
 COULD
COVID
DELIRIUM
BRING ON
DEMENTIA? Delirium is very common
on COVID wards.
Researchers are testing
whether these
temporary bouts of confusion
could bring on permanent
cognitive decline
By Carrie Arnold

Illustration by Fatinha Ramos

116  |  SCIENTIFIC AMERICAN  |  SPECIAL EDITION  |  WINTER 2022

SCIENTIFICAMERICAN.COM  |  117
 I n h e r jo b a s a p h y s i c i a n at t h e B o s t o n M e d i c a l C e n t e r ,
Sondra Crosby treated some of the first people in her region to get
COVID. So when she began feeling sick in April 2020, Crosby wasn’t
surprised to learn that she, too, had been infected. At first her symp-
toms felt like those of a bad cold, but by the next day she was too sick
to get out of bed. She struggled to eat and depended on her husband
to bring her sports drinks and fever-reducing medicine. Then she lost
track of time completely.
For five days Crosby lay in a confused haze, unable to remember the sim-
plest things, such as how to turn on her phone or what her address was. She
began hallucinating, seeing lizards on her walls and smelling a repugnant rep-
tilian odor. Only later did Crosby realize that she had had delirium, the formal
medical term for her abrupt, severe disorientation.
“I didn’t really start processing it until later when
I started to come out of it,” she says. “I didn’t have
the presence of mind to think that I was anything
more than just sick and dehydrated.”
has sparked physicians’ interest in the condition,
says Sharon Inouye, a geriatrician at the Marcus In-
stitute for Aging and Harvard Medical School, who
has studied delirium for more than 30 years.
Physicians treating people hospitalized with As clinicians face the immediate realities of con-
­COVID report that a large number experience delir- fusion and agitation on their wards, Inouye and oth-
ium and that the condition disproportionately af- er researchers are concerned about the future. In
fects older adults. An Aparil 2020 study in Stras- the past decade long-term studies have revealed
bourg, France, found that 65  percent of people who that a single episode of delirium can increase the
were severely ill with the novel coronavirus had risk of developing dementia years later and acceler-
acute confusion—a symptom of delirium. Data pre- ate rates of cognitive decline in those who already
sented at the 2020 meeting of the American College have the condition. The reverse is also true: having
of Chest Physicians by scientists at the Vanderbilt dementia makes someone more likely to develop
University Medical Center showed that 55  percent delirium. A set of simple steps, such as ensuring
of the 2,000 people they tracked who were treated a family member is present to help people orient
for ­COVID in intensive care units (ICUs) around the themselves, can reduce the incidence of delirium by
world had developed delirium. These numbers are 40 percent, but doctors struggle to follow that advice
much higher than doctors are used to: usually about in COVID wards.
one third of people who are critically ill develop de- But the links between delirium and dementia
lirium, according to a 2015 meta-analysis [see “How have been difficult to untangle: researchers need to
Common Is Delirium?” in box on page 121]. follow patients for years to get results. The surge in
Delirium is so common in COVID that some re- people with delirium produced by the pandemic has
searchers have proposed making the condition one focused attention on the condition and provided sci-
of the disease’s diagnostic criteria. The pandemic entists with a unique opportunity to follow patients

118  |  SCIENTIFIC AMERICAN  |  SPECIAL EDITION  |  WINTER 2022

 and determine whether and how delirium might af- physical and mental health. To Inouye, the connec- VISITS FROM
fect long-term cognition. Researchers have launched tion between her patients’ delirium and their poor relatives are a
several studies to explore the long-term neurocogni- prognosis was obvious. When she confessed her sus- source of comfort
tive impacts of C ­ OVID, including dementia, and picions to her bosses, however, they just shrugged. for people with
Inouye and others hope that this work will allow re- Their attitude, Inouye says, was that delirium was delirium, a com-
searchers to explore the links between the two con- just one of those things that happened. mon symptom
ditions in real time. “Why is it okay for older adults to come in the of ­COVID, but
If the pandemic can be said to have a silver lin- hospital and lose their minds?” Inouye asked. An- many hospitals had
ing, Inouye says, it has been to spur interest in how swering this question, she says, would be “an uphill strict no-visitor
delirium can lead to dementia—and vice versa. What battle my entire career.” policies at the start
is more, says Catherine Price, a neuropsychologist at Shortly after, she began a two-year fellowship to of the pandemic.
the University of Florida, the spread of C
­ OVID “has study the condition in depth. Her work showed that
highlighted the blurring of the lines be­­tween deliri- delirium occurs when several stressors converge. Pre­­
um and dementia, especially with more older adults existing vulnerabilities such as chronic disease or
in our populace.” cognitive impairment can combine with precipitat-
ing factors, including surgery, anesthesia or over-
Morteza Nikoubazl/NurPhoto/Getty Images

NEGLECTED CONDITION whelming infection, to cause a sudden onset of con-

I nouye ’ s interest in delirium began when she fusion, disorientation and attention difficulties, es-
landed her first job as an internal medicine physi- pecially in older adults.
cian at a Veterans Administration hospital in Con- “Delirium easily occurs when the brain is unable
necticut in 1985. In her first month there, she treat- to compensate for a stressful situation,” explains Tino
ed more than 40 people for a variety of conditions. Emanuele Poloni, a neurologist at the Golgi Cenci
Six of them developed delirium during their stay; Foundation near Milan, Italy. Researchers think that
none seemed to return to their previous level of the underlying biological causes are inflammation and

SCIENTIFICAMERICAN.COM  |  119
 an imbalance in neurotransmitters—chemical messen- management of delirium from a pharmaceutical
gers such as dopamine and acetylcholine. stand­point,” Price says.
Inouye’s mounting clinical experience has taught Scientists have developed three hypotheses to ex-
her that regardless of what precipitates delirium, plain how delirium might provoke dementia. One
around 70  percent of those with symptoms eventu- line of thinking holds that an accumulation of toxic
ally recover completely. In the 30  percent who do cellular trash in the brain could cause short-term
not, however, an episode of delirium predicts a delirium and lead to longer-term damage. The body
downward spiral over a period of months that leads usually clears this molecular rubbish by way of the
to profound cognitive impairment or even to symp- bloodstream and the glymphatic system, which is a
toms of dementia. network of channels filled with cerebrospinal fluid.
More formal studies have reinforced the link, to Damage to vessels from an acute episode of delirium
varying degrees. Inouye investigated a group of 560 could persist and trigger dementia, or a brain that
people aged 70 or older who had undergone surgery, experiences delirium could become more prone to
and she saw that cognitive decline over the subse- vascular problems in the future.
quent 36 months was three times faster in those The second suspect is inflammation, which often
who developed delirium than in those who did not troubles people who are hospitalized for infections,
have the condition. A 2020 meta-analysis of 23 stud- respiratory distress or cardiovascular disease. Sur-
ies showed that delirium during a hospital stay was gery and severe infections can cause a buildup of cel-
associated with 2.3 times greater odds of developing lular detritus in the brain, which triggers more in­­
dementia. And work by a team of Brazilian scien- flam­ma­ tion. This short-term, all-hands-on-deck re­­
tists showed that in a group of 309 people with an action safeguards the brain because it clears the
harmful debris, and the inflam-
mation ultimately dies down.
“WE DON’T UNDERSTAND THE MECHANISMS That is not the case for those who
develop delirium, Inouye says.
OF DELIRIUM AT ALL—WE REALLY DON’T. AND Persistent inflammation can trig-

THERE IS NO SUCCESSFUL MANAGEMENT OF ger an acute episode of delirium

and cause neurons and associat-
DELIRIUM FROM A PHARMACEUTICAL STANDPOINT.”  ed cells, such as astrocytes and
mi­­cro­glia, to deteriorate, leading
 —CATHERINE PRICE UNIVERSITY OF FLORIDA to cognitive damage.
The third idea is what is re-
ferred to as the threshold hypoth-
average age of 78 years, 32  percent of those who de- esis. Someone with dementia (even in the earliest
veloped delirium in the hospital progressed to hav- stages) has fewer connections between neurons and
ing dementia, compared with just 16  percent of can show damage to the insulation that wraps them
those who did not become delirious [see “Delirium and helps to convey signals, known as white matter.
and Cognitive Decline” in box on opposite page]. This loss strips the neurological reserves that help
What is more, the longer a person is delirious, the the person to cope with inflammation or infection,
greater their risk of subsequent cognitive impair- throwing them over the edge not just into delirium
ment, according to a 2013 study by psychologist James but into a more advanced dementia.
Jackson of Vanderbilt and his colleagues. Work by In- Even though the genesis of delirium and its mo-
ouye, Jackson and other researchers found that the lecular connections to dementia remain unknown,
reverse was also true: even after controlling for age, Inouye has managed to find a way to cut rates of de-
existing dementia symptoms increased the chances lirium in hospitals. She created a program of simple
of developing delirium. strategies known as HELP (Hospital Elder Life Pro-
gram), which focus on reducing sedation even dur-
CAUSING CONFUSION ing mechanical ventilation, paying close attention
Scientists still do not agree o n whether the to nutrition and hydration, and ensuring the pres-
link between delirium and dementia is strong only in ence of family members to help reassure and orient
those who would have developed dementia anyway or patients. A 2015 meta-analysis showed that these
whether delirium increases the risk of cognitive de- steps reduced delirium by around 40  percent. Hos-
cline even in individuals who are not predisposed to pitals around the U.S. began instituting these sim-
it. Nor can they say precisely what it is about delirium ple protocols. Then C ­ OVID struck and made them
that may provoke dementia. If researchers could iden- all but impossible.
tify these connections, then perhaps they could pre-
vent delirium from escalating into dementia. DEMENTIA SURGE
“We don’t understand the mechanisms of deliri- As Crosby endured c oronavirus-induced delirium
um at all—we really don’t. And there is no successful in her Boston bedroom, Poloni was treating delirious

120  |  SCIENTIFIC AMERICAN  |  SPECIAL EDITION  |  WINTER 2022

 people with C ­ OVID in Lombardy—Italy’s
ground zero for the coronavirus. Many of
Poloni’s pa­­tients already had dementia, and How Common Is Delirium?
like many physicians, he was watching for Typically one third of people who are critically ill will have an episode
common symptoms of respiratory infec- of delirium; for COVID-19, the proportion rises to more than half.
tions such as fever, cough and difficulty
breathing. But some of his patients did not
show those signs at all. Instead they mostly 31.8%
became “dull and sleepy,” Poloni says. Oth-
ers be­­came restless and agitated—all signs
of de­­lir­i­um. It was so prominent that Poloni Typical COVID-19
argued that delirium should be added to
55%
the virus’s diagnostic criteria. Inouye has
made that argument, too, and it is sup-
ported by a study she published in October
2020 showing that 28 percent of older adults
with ­COVID have delirium when they pres-
ent to the emergency department. The Cost of Delirium?
The high numbers of people who devel- In the U.S., the annual health-care costs for delirium are higher than those
oped delirium immediately made Inouye, for many other conditions common in older adults.
 ol. 59, Suppl. 2; November 2011 (“Delirium and Cognitive Decline”)

Price and other researchers worry that the

pandemic could lead to a surge in demen- Cardiovascular disease
Discharge Dementia: Results from a Cohort of Older Adults without Baseline Cognitive Impairment,” by F. B. Garcez et al., in A ge and Ageing, Vol. 48, No. 6; November 2019 (“The Cost of Delirium?”);
Nature; Sources: “Outcome of Delirium in Critically Ill Patients: Systematic Review and Meta-analysis,” by J.I.F. Salluh et al., in BMJ, Vol. 350; 2015 (“How Common Is Delirium?”); “Delirium and Post-

tia cases in the coming decades, on top of U.S. $258 billion

the increase in cases as a result of aging
populations [see “The Cost of Delirium?” Delirium
in box on this page]. “Is there going to be $152 billion
an increase in dementia from people who
Diabetes
had ­COVID during adulthood or midlife?”
asks Natalie Tronson, a neuropsychologist $92 billion
at the University of Michigan. “What hap-
Nonfatal falls
pens over the next decades as the popula-
tion ages more?” $19 billion
In a concerted effort to find answers,
“The Importance of Delirium: Economic and Societal Costs,” by D. L. Leslie and S. K. Inouye, in J ournal of the American Geriatrics Society. V

Hip fractures
institutes around the world have funded a
$7 billion
variety of studies into the long-term cogni-
tive effects of ­COVID, some of which look
at delirium. One such study in the U.S.
tracked people who were treated in the Delirium and Cognitive Decline
hospital for ­COVID, many of whom devel- People who experience delirium during hospitalization are at increased risk of
oped delirium during their stay, and mea- cognitive decline after discharge, according to a study of 309 patients in Brazil.
sured their cognitive and psychiatric func-
Dementia after no delirium
tion. An international study is planned
to measure the prevalence of delirium 16%
in patients with ­COVID in ICUs, as well
Dementia after delirium
as identifying factors that predict long-
term outcomes. 32%
A separate study in Germany and the
U.K. is also tracking neurocognitive out-
comes in people with C ­ OVID to determine
how delirium affects brain function months later. ic interest in the delirium-dementia connection—
Another research project led by a team at Vanderbilt and provide some insight.
is looking for an alternative to commonly used “It’s going to be, I think, a little bit frightening
sedatives such as benzodiazepines, which are known and a little bit enlightening about how illness affects
to increase delirium. The researchers are testing a dementia risk but also about what other lifestyle
sedative called dexmedetomidine to evaluate wheth- and genetic protective factors can influence risk as
er it is a safer option for people who are hospitalized well,” Tronson says. “We’re learning quickly, but
with ­COVID. there are still a lot of black boxes.” 
Inouye and Tronson hope that the funding of
these long-term studies will lead to ongoing scientif- Carrie Arnold is an independent public health reporter based in Virginia.

SCIENTIFICAMERICAN.COM  |  121
OPINION

A Psychedelic Renaissance
Psilocybin and MDMA represent a first wave
of therapies for psychiatric disorders
that help patients by changing the way
they view reality
By Danielle Schlosser and Thomas R. Insel

T
hrough a May 2 02 1 article he where these drugs were lost to science,
w rote in the Independent, 
 we although they were still used in recre-
learned about Steve Shorney, who ational and religious settings.
lived with depression for most of his life New studies have reignited the hope
despite years of psychotherapy, medica- that psychedelics could be powerful medi-
tion, yoga and many other attempts at cines for mental disorders. In a New Eng-
holistic treatments. With his decision to land Journal of Medicine report, two dos-
enroll in a psilocybin clinical trial at es of the chemical from the psilocybin
Imperial College London, his life “radical- mushroom appeared as effective as six
ly changed.” Psilocybin was different from weeks of escitalopram (Lexapro), a stan-
every other treatment or experience he dard antidepressant, for people with long- PSILOCYBIN MUSHROOMS
had had. As he recalled in the I ndepen- standing moderate-to-severe major de­­
dent a rticle, “I had seen an alternative pressive disorder (MDD). On many of the
reality, another way of being, and knew secondary measures of depression, such tive, effective and safe treatments. Fewer
beyond anything I’d known before that as remission of symptoms, psilocybin ap­­ than half of people with these disorders
day that life is extraordinary. And in that peared better than the standard treat- respond to medications or psychothera-
moment I felt happier, more alive, and ment, with 57 percent of subjects showing py, and about a third of MDD patients
more Me than I imagined was possible.” remission on psilocybin versus 28 percent have so-called treatment-refractory de­­
The use of psychedelics, especially psi- in remission with escitalopram, although pression that fails to respond at all. Dis-
locybin and MDMA (also known as Ecsta- the au­­thors caution that more research covering new approaches to treat mental
sy or Molly), is undergoing a renaissance. will be needed to confirm these results. In health conditions is critical. If psychedel-
More than five decades ago psychedelics a study reported in  Nature Medicine, ics prove to be effective and safe for these
were an active area of research, with more MDMA was found to be more effective disorders, they could be transformative
than 40,000 patients receiving LSD or psi- than placebo for people with severe post- in two interesting ways.
Roger Cremers/Bloomberg/Getty Images

locybin for alcoholism, anxiety or depres- traumatic stress disorder (PTSD). After First, these medications appear to be
sion. While we do not have rigorous clini- three sessions with MDMA, 67 percent no effective after acute administration. Most
cal trials from that time, the use of these longer met criteria for their diagnosis, psychiatric drugs, such as escitalopram for
drugs garnered both scientific and public and 33  percent showed complete remis- MDD, need to be taken for weeks, months
interest, with Hollywood celebrities such sion, relative to 32  and 5  percent, respec- or years to be effective. Presumably, they
as Cary Grant promoting their use. But tively, after receiving a placebo. control symptoms but do not alter the dis-
the War on Drugs, beginning during the Patients with disorders such as MDD order. When effective, psychedelics appear
Nixon years, led to a long, dark period and PTSD can certainly use more innova- to confer long-term effects, sometimes

122  |  SCIENTIFIC AMERICAN  |  SPECIAL EDITION  |  WINTER 2022

after a single ad­­ministration, suggesting primary care physicians. This combina- were classified as Schedule I compounds:
that they are not simply symptom-reduc- tion is especially promising, given that drugs with “high potential for abuse and
ing but disease-modifying. Moreover, some evidence suggests that psychedelics no currently accepted medical use.” New
some results from previous studies sug- enhance neuroplasticity, thereby opening science suggests that we may need to
gest that the benefits are long-lived, al­­ up a “critical window” of time to develop revisit this classification. 
though the duration of effects in these re­­ a healthier mindset.
cent studies remains to be determined. These new studies with psilocybin and Danielle Schlosser i s a psychologist and senior vice
Second, the studies thus far have fo­­ MDMA focus on only two of the many psy- president of clinical innovation at Compass Pathways,
cused on psychedelic-assisted psycho- chedelic compounds with potential for which is conducting clinical trials of psilocybin. In her
therapy—not just the drug but the experi- medical use. Could this class of drugs, prior role, she was in charge of vision and strategy for
ence in the context of structured, time- which includes LSD, mescaline and many the behavioral health portfolio at Verily Life Sciences,
limited psychotherapy. We know that lesser-known or yet to be discovered chem- an Alphabet company, and she remains on the faculty
psychotherapy can be helpful for most icals, revive drug development for psychia- at the University of California, San Francisco.
mental disorders, yet therapy is rarely try? Perhaps, but we need to be mindful Thomas R. Insel is a psychiatrist and neuroscientist
combined with medication. Psychedelics that in a previous era, overexuberance who served as director of the National Institute of Mental
remind us of the potential for combining about psychedelics led to a backlash with Health from 2002 to 2015. He is an adviser to Compass
medication and therapy, a practice that is little scientific research and no translation Pathways, as well as to several digital mental health
infrequent when most antidepressants to medical use. companies, and author of Healing: Our Path from Mental
and antianxiety drugs are prescribed by It has been 50 years since psychedelics Illness to Mental Health (Penguin, 2022).

SCIENTIFICAMERICAN.COM  |  123
 END NOTE

Does
Consciousness
Pervade
the Universe?
A philosopher answers questions
about “panpsychism”
By Gareth Cook

One of science’s most c hallenging prob-

lems is a question that can be stated
easily: Where does consciousness come
from? In his most recent book, G  alileo’s
Error: Foundations for a New Science
of  Consciousness, p
 hilosopher Philip tence. I simply mean “experience”: plea- that is, in terms of its intrinsic nature—
Goff considers a radical perspective: sure, pain, visual or auditory experience. is constituted of forms of consciousness.
What if consciousness is not something Human beings have a very rich and
special that the brain does but in­­stead complex experience; horses less so, mice Do you foresee a scenario in which
is a quality inherent to all matter? It is less so again. As we move to simpler panpsychism can be tested?
a theory known as panpsychism. He forms of life, we find simpler forms of You can’t look inside an electron to see
an­­swer­ed questions from former long- experience. Perhaps at some point the whether or not it is conscious, just as
time Mind Matters editor Gareth Cook. light switches off, and consciousness dis- you can’t look inside someone’s head
appears. But it’s at least coherent to sup- and see their feelings and experiences.
 or the complete interview, see www.
F pose that this continuum of conscious- We know that consciousness exists only
ScientificAmerican.com/article/ ness carries on into inorganic matter, because we are conscious.
does-­­consciousness-pervade-the-universe/ with fundamental particles having Neuroscientists correlate certain kinds
unimaginably simple forms of experience. of brain activity with certain kinds of
Can you explain, in simple experience. We now know which kinds of
terms, what you mean What does panpsychism seek brain activity are associated with feelings
by panpsychism? to bring to physics? of hunger, pleasure, pain, and so on. This
In our standard view of things, conscious- Philosophers of science have realized is really important information, but what
ness exists only in the brains of highly that physical science, for all its richness, we ultimately want from a science of con-
evolved organisms, and hence it exists is confined to telling us about the behav- sciousness is an explanation of those cor-
only in a tiny part of the universe and ior o
 f matter, what it does. Physics tells relations. Why is a particular feeling cor-
only in very recent history. According to us, for example, that matter has mass related with a particular pattern of brain
panpsychism, consciousness pervades and charge. These properties are com- activity? As soon as you start to answer
the universe and is a fundamental feature pletely defined in terms of behavior— this question, you move beyond what can
of it. This doesn’t mean that literally things like attraction, repulsion, resis- be, strictly speaking, tested, simply
everything is conscious. The basic commit- tance to acceleration. Physics tells us because consciousness is unobservable.
ment is that the fundamental constitu- absolutely nothing about what philoso- We have to turn to philosophy.
ents of reality—perhaps electrons and phers like to call the intrinsic nature o
f Science gives us correlations between
quarks—have incredibly simple forms matter: what matter is in and of itself. brain activity and experience. We then
of  experience, and the very complex Consciousness, for the panpsychist, have to find the philosophical theory
experience of the human or animal brain is the intrinsic nature of matter. There’s that best explains those correlations.
is somehow derived from the experience nothing supernatural or spiritual, but In my view, the only theory that holds up
of the brain’s most basic parts. matter can be described from two per- to scrutiny is panpsychism. 
Getty Images

I should clarify that by “conscious- spectives. Physical science describes

ness,” I don’t mean self-awareness or matter from the outside in terms of its Philip Goff s pecializes in the philosophy of mind
the capacity to reflect on one’s own exis- behavior. But matter from the “inside”— and metaphysics at Durham University in England.

124  |  SCIENTIFIC AMERICAN  |  SPECIAL EDITION  |  WINTER 2022