Master scientifically proven study techniques to become a better student and Topic Subtopic

more adept learner. Science Neuroscience & Psychology

The Learning Brain

The Learning Brain

“Pure intellectual stimulation that can be popped into
the [audio or video player] anytime.”
—Harvard Magazine
Course Guidebook
“Passionate, erudite, living legend lecturers. Academia’s
best lecturers are being captured on tape.”
Professor Thad A. Polk
—The Los Angeles Times
University of Michigan

“A serious force in American education.”

—The Wall Street Journal

Thad A. Polk is an Arthur F. Thurnau Professor in the

Department of Psychology at the University of Michigan.
He received an interdisciplinary Ph.D. in Computer Science
and Psychology from Carnegie Mellon University. Professor
Polk’s teaching—on topics ranging from the human mind
and brain to cognitive psychology—has been recognized
with numerous awards, including being named one of The
Princeton Review’s Best 300 Professors in the United States.
He is also a frequent visiting scientist at the Max Planck
Institute for Human Development in Berlin.

THE GREAT COURSES ®

Corporate Headquarters
4840 Westfields Boulevard, Suite 500
Chantilly, VA 20151-2299
USA
Guidebook

Phone: 1-800-832-2412
www.thegreatcourses.com
Professor Photo: © Jeff Mauritzen - inPhotograph.com.
Cover Image: © Sergey Nivens/Alamy.

Course No. 1569 © 2018 The Teaching Company. PB1569A

Published by
THE GREAT COURSES

Corporate Headquarters

4840 Westfields Boulevard | Suite 500 | Chantilly, Virginia | 20151‑2299

[phone] 1.800.832.2412 | [fax] 703.378.3819 | [ web] www.thegreatcourses.com

Copyright © The Teaching Company, 2018

Printed in the United States of America

This book is in copyright. All rights reserved. Without limiting the rights under copyright reserved
above, no part of this publication may be reproduced, stored in or introduced into a retrieval system,
or transmitted, in any form, or by any means (electronic, mechanical, photocopying, recording, or
otherwise), without the prior written permission of The Teaching Company.
 Thad A. Polk, PhD
Arthur F. Thurnau Professor of Psychology
University of Michigan

Thad A. Polk is an Arthur F. Thurnau Professor in the Department

of Psychology at the University of Michigan. He received a BA in
Mathematics from the University of Virginia and an interdisciplinary
PhD in Computer Science and Psychology from Carnegie Mellon
University. He received postdoctoral training in cognitive neuroscience
at the University of Pennsylvania before joining the faculty at the
University of Michigan.

Professor Polk’s research combines functional imaging of the human

brain with computational modeling and behavioral methods to
investigate the neural architecture underlying cognition. Some of
his major projects have investigated changes in the brain as we age,
contributions of nature versus nurture to neural organization, and
differences in the brains of smokers who quit compared with those
who do not. Professor Polk regularly collaborates with scientists at
the University of Texas at Dallas and at the Max Planck Institute
for Human Development in Berlin, where he is a frequent visiting
scientist. At the University of Michigan, he is an associate chair of
the Department of Psychology and the chair of the Health Sciences
and Behavioral Sciences Institutional Review Boards.

PROFESSOR BIOGRAPHY i
 Professor Polk regularly teaches large lecture courses as well as small
seminars on topics ranging from the human mind and brain, to
cognitive psychology, to computational modeling of cognition. His
teaching at the University of Michigan has been recognized with
numerous awards, including the Excellence in Education Award
from the College of Literature, Science, and the Arts and the Arthur
F. Thurnau Professorship, the university’s highest undergraduate
teaching award. He was also featured in the University of Michigan’s
Professors Reaching Out for Students (PROFS) lecture series and was
named to The Princeton Review’s list of the Best 300 Professors in
the United States.

Professor Polk’s other Great Courses are The Aging Brain and The
Addictive Brain. 

ii The Learning Brain

Acknowledgements
I’d like to express my heartfelt gratitude to Alyssa Skelton, who worked
tirelessly on almost every aspect of this course with me. She identified
potential topics to include, performed extensive literature reviews,
provided detailed feedback and suggestions on drafts of the lectures,
and proposed potential graphics to use throughout the course. The
course is significantly better than it would have been without her
many significant contributions. Thank you so much, Alyssa!

I’d also like to thank my lab manager, Holly Gagnon, who did a lot
of work tracking down articles and editing lectures as I worked on
the course. She played a particularly important role in the preparation
of the extensive bibliography. Thank you, Holly!

Finally, I’d like to thank my wife, Norma, and my 3 daughters—

Sarah, Rachel, and Lydia—for their patience and support as I spent
countless hours working on this course. I love you!

ACKNOWLEDGEMENTS iii
 Table of Contents
Introduction
Professor Biography . . . . . . . . . . . . . . . . . . . . . . . . . . i
Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . iii
Course Scope . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

Guides
1 Learning 101 . . . . . . . . . . . . . . . . . . . . . . . . . . 4

2 What Amnesia Teaches Us about Learning . . . . . 11

3 Conscious, Explicit Learning . . . . . . . . . . . . . . . 20

4 Episodic Memory and Eyewitness Testimony . . . 30

5 Semantic Memory . . . . . . . . . . . . . . . . . . . . . . 41

6 The Neural Basis of Explicit Learning . . . . . . . . . 50

7 Strategies for Effective Explicit Learning . . . . . . 60

8 Controversies in Explicit Learning Research . . . . 70

9 Unconscious, Implicit Learning . . . . . . . . . . . . . 80

10 The Psychology of Skill Learning . . . . . . . . . . . . 91

11 Language Acquisition . . . . . . . . . . . . . . . . . . 100

12 The Neural Basis of Implicit Learning . . . . . . . . 109

iv The Learning Brain

 13 Strategies for Effective Skill Learning . . . . . . . .120

14 Learning Bad Habits: Addiction . . . . . . . . . . . 130

15 Introduction to Working Memory . . . . . . . . . . 140

16 Components of Working Memory . . . . . . . . . . 149

17 The Neural Basis of Working Memory . . . . . . . 159

18 Training Your Working Memory . . . . . . . . . . . . 169

19 How Motivation Affects Learning . . . . . . . . . . 178

20 How Stress and Emotion Affect Learning . . . . . 188

21 How Sleep Affects Learning . . . . . . . . . . . . . . 199

22 How Aging Affects Learning . . . . . . . . . . . . . 210

23 Dyslexia and Other Learning Disabilities . . . . . . 220

24 Optimizing Your Learning . . . . . . . . . . . . . . . 231

Supplementary Material
Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 241
Image Credits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 249

TABLE OF CONTENTS v
vi The Learning Brain
The Learning Brain
One of our most important assets as human beings is our ability to
learn and adapt. As babies, we learn to crawl, stand, and eventually
walk. As toddlers, we learn to speak and understand our native
language. As we get older, we learn how to read and write; we learn
about history and literature, and about science and engineering.
We’re also constantly acquiring new skills, such as playing a sport or
mastering a musical instrument.

But how? This course will explore what’s going on in our minds
that allows us to learn so effortlessly and in so many ways. The
course will examine not only the psychological mechanisms, but also
the neural mechanisms, that are involved. You will discover that
your brain is equipped with multiple, powerful, but very different
learning mechanisms that depend on specific neural circuits and
that understanding how learning works can make you a more
effective learner.

The course begins by explaining what psychologists mean by learning

and how it relates to memory. It also outlines some of the major
themes in the science of learning that you will encounter in the course.

Then comes a detailed examination of explicit learning, which refers

to learning verbalizable, conscious information, such as the knowledge
you acquire in school and information about the personal episodes
that you experience every day. You will examine amnesia, which
dramatically impairs explicit learning and has taught us a great deal
about how normal learning and memory work. You will discover that
your memory for personal episodes tends to focus on the gist, or the
big picture, and often overlooks, or even misrepresents, many of the
details. Rather than thinking of your memory as an accurate video
of what happened, it’s probably better to think of it as a plausible, but

COURSE SCOPE 1
 somewhat unreliable, reconstruction of what might have happened.
Unfortunately, we usually can’t tell the difference between reality
and our reconstructions.

Next, the course will take you inside the skull to examine what’s
going on in your brain when you’re learning explicit information.
You will discover that one brain area, called the hippocampus, plays a
particularly important role, but that the information you learn tends
to get consolidated into other brain regions and therefore depends
less and less on the hippocampus as time goes by.

The course then changes gears to explore the fascinating world of

unconscious, implicit learning. You will discover that your brain
does at least as much learning behind the scenes as it does in your
conscious awareness. You will examine the fundamental mechanisms
of classical and operant conditioning that underlie your ability to
learn associations between stimuli in your environment and between
rewards and the behaviors that lead to those rewards. You will also
discover how people learn the amazing skills that they perform so
effortlessly, such as how athletes learn to play their sport and how
musicians learn to play their instrument. An entire lecture is devoted
to exploring how children learn their native language, which may
be the most amazing learning accomplishment of all. You will also
explore cutting-edge neuroscience research that is beginning to shed
light on how this kind of learning works at a cellular level.

Next, the course turns to working memory and explores the critical
role it plays in all complex thought. You will learn about leading
theories of how working memory works, at both a psychological
and a neural level. You will also explore recent research examining
whether it’s possible to expand working memory and perhaps even
increase intelligence.

2 The Learning Brain

The final section of the course will dive into some of the key
factors that influence learning and memory. You will discover that
motivation is at least as important as ability and that how you feel
can have a dramatic impact on what you learn. You will examine the
fascinating and crucial role that sleep plays in learning and explore
how learning changes as you age—as well as how it doesn’t. You
will also discover what’s going on when learning disabilities such as
dyslexia rear their head.

Throughout, the course will repeatedly identify specific ways that you
can apply what you’re learning to become a more effective learner in
your daily life. For example, you will discover study strategies that
are scientifically proven to improve learning and memory as well as
practice strategies that have been proven to help people learn skills
more effectively.

By the end, you will have a new understanding about how learning
works, how it is implemented in the brain, and how you can be a
more effective learner. 

COURSE SCOPE 3
 LECTURE 1
Learning 101

L  ARNING IS CRUCIAL TO OUR LIVES AS

E
HUMAN BEINGS. OUR ABILITY TO LEARN AND
adapt is one of our most important and
powerful assets. The goal of this course is to
help you understand how the truly amazing
learning mechanisms in your brain work so
that you can get the most out of them in
whatever kind of learning you’re doing in
your life. In this lecture, you will explore what
learning is and why it is so crucial to us as
human beings. You will also learn about how
the scientific study of learning got started as
well as some of the key themes that will be
addressed in the course. 

4 The Learning Brain

 Defining Learning and Memory
•• One fairly intuitive definition of learning is that it is changing
your behavior in response to previous experience. That’s not a bad
definition, but there are a few important points to keep in mind
that might lead us to refine it a little.

•• First, there are situations in which you learn something but there is
no opportunity to show it in your behavior. For example, suppose
that you learn some interesting but obscure fact about Berlin, such
as that a former airport there has been turned into a park. You
may have learned that fact and stored it in your memory, but it’s
quite possible that it will never influence your subsequent behavior,
unless you happen to visit Berlin or have a conversation about its
former airport.

The reason human beings can live and thrive in an incredible

range of different environments is because we can learn. More
than any other species on the planet, we can adapt and change our
behavior to better fit the needs of our environment.

Lecture 1 | Learning 101 5

 •• In other words, not everything you know is reflected in a change
in behavior, but that doesn’t mean you didn’t learn it.

•• Let’s revise our definition and say that learning is acquiring

knowledge from experience. Now, that new knowledge could
affect your behavior. For example, if someone asked you if you
knew that a former airport in Berlin had been turned into a park,
then your answer would be different after learning that fact. But
learning doesn’t always have to be reflected in your behavior.

•• But here is still at least one problem with our new definition. When
people think of knowledge, they typically think of information
that is consciously available and verbalizable, such as the fact that
birds have wings. We obviously do learn that kind of information,
but we also learn a lot of behavioral responses that are not
consciously available.

•• For example, if we sit in a room where a fan is blowing, we typically

get habituated to the sound of the fan and learn to ignore it. That
kind of learning has been studied extensively in lower organisms,
and although you could call it a kind of knowledge, doing so seems
a little strange.

•• So, let’s define learning as acquiring knowledge or behavioral

responses from experience. The “from experience” part is also
crucial. Learning might come from studying, or from being
taught, or just from living life, but it has to come from experience.
Behavioral responses that are genetically programmed, such as
instincts and reflexes, don’t count as being learned.

•• Psychologists typically consider memory to be the result or product

of learning. It is the record of learning that is stored in your mind.
Learning involves making physical changes in the brain that allow
information to be retrieved later on. And those changes constitute
the physical basis of memory.

6 The Learning Brain

 The Scientific Study of Learning
•• For most of human history, the study of learning was restricted
to philosophy. And even after the power of the scientific method
became clear in fields like chemistry and physics, it still wasn’t
applied to study the human mind for quite some time.

•• In fact, up until the late 19th century,

most scientists and philosophers Scientific advances
probably didn’t think that learning have radically changed
could be studied scientifically. our environment
Mental processes like learning and and require us to
memory were seen as fundamentally constantly learn about
different than the natural phenomena new technologies.
of chemistry and physics.

•• But that all changed in 1885, when Hermann Ebbinghaus

published his famous book Über das Gedächtnis, which means
“about memory” in English. Like all psychologists at the time,
Ebbinghaus was trained as a philosopher. But inspired by scientific
studies of human perception that had recently been published,
he decided to try applying scientific methods to study human
learning and memory. His work ultimately revolutionized the field
and helped establish psychology as a legitimate scientific field in
its own right.

•• Ebbinghaus used only a single human participant in his

experiments—himself. He wanted to study learning in a way that
wasn’t contaminated too much by what he already knew, so he
decided to try to learn lists of meaningless syllables. He studied a
list until he could repeat it back twice in a row without any errors.

•• Ebbinghaus plotted his performance as a function of the delay and

produced what are now called forgetting curves. His performance
got worse with longer delays. But the drop-off wasn’t continuous.

Lecture 1 | Learning 101 7

 In fact, he found that most of the forgetting happened during the
first hour or 2. Performance after a day wasn’t much worse than
performance after an hour, and performance after 2 days wasn’t
much worse than performance after 1 day.

•• Ebbinghaus also studied how much he learned as a function of

each repetition. And like the forgetting curve, he found that the
first repetition led to the most learning, the second repetition to
a little less, and so on.

•• Ebbinghaus’s findings have been confirmed in countless

experiments since then. But by far his biggest contribution was to
demonstrate that human learning and memory could be studied
scientifically in a way that other scientists could replicate and
extend. And the repercussions of that contribution are still being
felt to this day.

8 The Learning Brain

•• This course will explore those repercussions in depth. In particular,
you will repeatedly discover how the scientific study of learning
has made major advances in at least 3 major directions: It has
dramatically improved our ability to help people learn more
effectively, whether in the classroom or in real life; has led to much
better theories of the cognitive mechanisms involved in learning;
and is beginning to shed light on the neural mechanisms that
underlie learning.

1 Learning scientists have recently discovered that there is a lot

we can do to make our learning more effective—not only in
the classroom, but also in our daily lives. There are also ways to
optimize our learning so that we increase how much we learn.

2 Many people think of learning as a single, unitary process,

but nothing could be further from the truth. In the past few
decades, cognitive psychologists have discovered that human
beings are equipped with a variety of very different mechanisms
that are tailored for learning different kinds of information.
Specifically, short-term working memory is very different from
long-term memory. In fact, psychologists have discovered that
we even use different mechanisms for storing different kinds of
information within working memory and long-term memory.

Our capacity to learn has never been more important than

it is today. A few hundred years ago, most people could
do just fine without much formal education. But today’s
workforce requires people to learn new skills all the time.
For example, the US Bureau of Labor Statistics reported
that people born between 1957 and 1964 held an average
of 12 different jobs before the age of 50, and more than
half of those jobs were held after the age of 25.

Lecture 1 | Learning 101 9

 3 There are physical changes in our brain when we learn a new
piece of information. Our extraordinary ability to learn depends
largely on making millions of very small changes in the strength
of brain connections. Different types of learning that have been
hypothesized to depend on different psychological mechanisms
have been found to depend on different parts of the brain.

SUGGESTED READING

Anderson, Learning and Memory.

Bransford, Brown, and Cocking, How People Learn.
Schacter, Searching for Memory.

QUESTIONS TO CONSIDER

1 Many simple organisms (e.g., cockroaches and earthworms) thrive

without doing nearly as much learning as humans do. What do you
think allows them to do so?

2 Why do you think it was difficult for early scientists to recognize

that human learning could be studied scientifically?

10 The Learning Brain

 LECTURE 2
What Amnesia Teaches
Us about Learning

T  IS LECTURE TELLS THE STORY OF A MAN

H
WHO MAY HAVE TAUGHT US MORE ABOUT
learning and its neural substrates than
anyone in history: Henry Molaison. He wasn’t
a famous psychologist or neuroscientist, but
a neurological patient who suffered from
profound amnesia. Studies of Henry—or
patient HM, as he’s known in the scientific
literature—have radically changed our
understanding of how learning works and
how it’s implemented in the brain. In this
lecture, you will discover what those studies
have taught us about the learning brain. 

Lecture 2 | What Amnesia Teaches Us about Learning 11

 Explicit learning is very different from implicit learning. Likewise,
storing information in long-term memory depends on different
cognitive and neural mechanisms than does storing information in
working memory.
These facts have important implications for us if we want to
optimize our learning. In particular, we shouldn’t expect strategies
that optimize explicit learning to be effective when learning
implicit information.
For example, strategies that are very effective when learning US
history might not be the best strategies to use when learning to
play golf.
If we want to optimize our learning, then we need to tailor our
learning strategies depending on the kind of information that we’re
trying to learn.

A Case of Amnesia
•• Henry Molaison was born in 1926 and grew up near Hartford,
Connecticut. When he was around 7 years old, he was run over
by a bicycle, hit his head, and was unconscious for a few minutes.
Later, he started experiencing epileptic seizures. At first, the
seizures were manageable, but they grew worse over time.

•• By the time Henry was 27, he was having about 10 seizures a day,
and large doses of anticonvulsant medication weren’t helping. He
was therefore desperate to find an alternative treatment. That’s
when a renowned neurosurgeon named William Scoville suggested
surgery, which can often be an effective way to treat epilepsy when
medications don’t work.

12 The Learning Brain

•• Doctors put electrodes on Henry’s scalp, recorded his brain waves,
and waited for a seizure in the hopes that they would be able
to isolate the source. They found some brain wave abnormalities
in the temporal lobes during one of his seizures and decided to
perform an experimental surgery and remove some parts of his
temporal lobes.

•• The hope was that the surgery would help control Henry’s seizures,
and it did. After the surgery, he only had about 5 minor seizures a
month. Unfortunately, he also suffered a very serious side effect:
He became profoundly amnesiac. As a result, that surgery has
never been performed again.

•• Henry still remembered who he was and where he grew up.

He remembered his family and his childhood friends. He also
remembered many specific events from his childhood, including
the bicycle accident that may have led to his seizures.

When many people

think of amnesia, they
imagine a person who
no longer remembers
their family and
friends, or even who
they are and where
they grew up. But
that’s not the way it
is for the vast majority
of people who have
developed amnesia as a
result of brain damage.

Lecture 2 | What Amnesia Teaches Us about Learning 13

 •• But Henry did have trouble remembering events that occurred right
before the surgery. In fact, he remembered almost no events from
the 2 years preceding the surgery, and his memory for the 9 years
before that was also impaired. This kind of memory impairment
is typically called retrograde amnesia. The term “retrograde” refers
to when the forgotten events occurred. In retrograde amnesia, the
forgotten events occurred before the brain damage.

•• Henry’s retrograde amnesia exhibited a temporal gradient; that is,

his older childhood memories were intact, but later memories from
the years immediately preceding his surgery were impaired. This
pattern, sometimes called Ribot’s law (after the French psychologist
Théodule-Armand Ribot), is very common in amnesia.

•• But Henry’s most significant deficit wasn’t in remembering; it

was actually in learning. After his surgery, Henry’s ability to
learn new information was catastrophically impaired, even if that
information was encountered over and over again on an almost
daily basis.

•• This kind of impairment in learning new information and laying

down new memories is often called anterograde amnesia, in which
the forgotten events occurred after the brain injury. Henry’s
anterograde amnesia included difficulty learning traumatic
information. Even though people typically have stronger, more
vivid memories for traumatic, stressful events than for more
mundane events, Henry’s memories for traumatic events were
dramatically impaired.

Most patients with amnesia have

both a temporally graded retrograde
amnesia and an anterograde amnesia.

14 The Learning Brain

 Types of Learning and Memory
Preserved in Amnesia
•• Many other aspects of learning and memory were preserved in
Henry. And it was the specific pattern of spared versus impaired
functions that taught scientists the most about human learning.

•• One of the first discoveries was made by Brenda Milner, who

did most of the early research on Henry. She asked him to trace
drawings while looking at a mirror rather than directly at the paper
itself. Left and right were reversed, and so were up and down.

•• This task is difficult, and at first, Henry was bad at it. But after
practicing the task repeatedly for 3 days, he could trace the figure
quickly and accurately. Henry’s performance was changing as a
result of his experience—in other words, he was learning.

•• When Milner returned for the second and third day of training,
Henry didn’t remember having ever done the task before. He was
much better at it than he was the previous day, but his learning was
all unconscious. Apparently, learning a motor skill is fundamentally
different than learning a person’s name or other conscious facts.

•• And it’s not just motor skill learning that is preserved in amnesia.
For example, in 1980, Neal Cohen and Larry Squire investigated
the ability of amnesiacs to learn a perceptual skill in which they
had to learn to recognize unfamiliar visual stimuli. Specifically,
they asked them to read mirror-reversed text as fast as they could.
As in Milner’s study, the participants practiced the skill for 3
straight days. A group of nonamnesiac control subjects practiced
the same task.

Lecture 2 | What Amnesia Teaches Us about Learning 15

 •• The amnesiacs improved at reading mirror-reversed text the more
they practiced. In fact, they learned just as quickly as the normal
control subjects did. Furthermore, they were still good at it 3
months later. So, not only did the amnesiacs learn to perform this
skill, but they also had retained this newly acquired ability. And
like HM, the amnesiacs in this study learned this perceptual skill
even though they didn’t consciously remember having done the
task before.

•• Inspired by these early findings, several scientists looked for and

found other types of learning and memory that were preserved
in amnesia, including priming, which occurs when previous
exposure to a stimulus facilitates your processing of similar stimuli
in the future.

•• For example, suppose that you’re reading a magazine and see a

picture of a celebrity that you know, but it takes you a minute
to recognize the person. If you see a picture of the same celebrity
a few minutes later, you’ll probably recognize them faster than
you did the first time. Previous exposure to the celebrity’s picture
primes you and makes you a little faster to recognize that person
the next time.

Amnesiacs can learn perceptual and motor skills just like everyone
else. Although they may not consciously remember ever having
performed the tasks before, their performance continues to improve
with additional practice. Another way of putting this is in terms
of knowing how versus knowing that. Amnesiacs can learn how
to perform a skill, even when they don’t know that they’ve ever
practiced it before.

16 The Learning Brain

•• Priming is another type of unconscious learning. A study by
Carolyn Cave and Larry Squire found that amnesiacs exhibit
normal priming effects that last at least a week, even though their
conscious memory is dramatically impaired.

•• In 1995, John Gabrieli, along with a number of his colleagues,

investigated whether amnesiacs exhibit intact classical conditioning,
or Pavlovian conditioning, after Ivan Pavlov. In experiments with
his dogs, Pavlov measured how much they salivated. They salivated
a lot when Pavlov got their food out. But by accident, he noticed
that they also started salivating as soon as he entered the room,
even if he didn’t have any food with him.

•• Pavlov hypothesized that the dogs had learned that they were likely
to be fed soon after he came in, so they were anticipating the food
even before it arrived. He began to investigate the phenomenon
systematically and discovered that if he consistently paired the
sound of a bell with the presentation of food, then pretty soon the
dogs would learn an unconscious association between the bell and
the food. In particular, they would salivate whenever they heard
the bell, even if there wasn’t any food.

•• Classical conditioning is another type of learning. It’s another

example of behavior being changed as a result of previous
experience. And just like skill learning and priming, classical
conditioning is preserved in amnesia.

•• Together, these results radically changed our understanding of

learning and how it’s implemented in the brain. At the most
general level, they conclusively demonstrate that human beings
use multiple different learning systems, depending on what
they’re learning.

Lecture 2 | What Amnesia Teaches Us about Learning 17

 •• In particular, learning unconscious information is fundamentally
different than learning conscious information and even depends
on different parts of the brain. After all, amnesia clearly shows that
brain damage that dramatically impairs conscious memories can
leave unconscious memories intact.

•• Psychologists often refer to conscious memories as explicit and

unconscious memories as implicit.

•• Explicit memories, sometimes called declarative memories, are

memories that you can consciously bring to mind and describe
verbally. When most people think of learning and memory,
they’re thinking of explicit learning and memory, such as
remembering what you ate for lunch. These are the kinds of
memories that amnesiacs have difficulty learning.

•• Implicit memories are memories that you can’t consciously recall

but that nevertheless influence your subsequent behavior. For
example, your memory for how to ride a bike is an automatic,
implicit memory. The same is true for priming and conditioning;
amnesiacs typically don’t have trouble learning these kinds of
implicit memories.

•• Another aspect of learning and memory that was spared in Henry

was working memory, which refers to the ability to store, rehearse,
and manipulate information temporarily. For the purposes of this
course, working memory will be considered synonymous with
short-term memory. Working memory is a temporary memory
system that we actively use to help us solve problems and achieve
our goals.

•• Henry’s working memory was very much intact, despite his

profound deficit in storing explicit information into long-term
memory. He could temporarily store information, rehearse it, and
manipulate it in all the same ways that other people do. In fact,

18 The Learning Brain

 his intelligence as measured by traditional IQ tests was in the
normal range. He also performed normally on more formal tests
of working memory.

SUGGESTED READING

Corkin, Permanent Present Tense.

Dittrich, “The Brain That Couldn’t Remember.”
Squire, “The Legacy of Patient HM for Neuroscience.”
Wearing, Forever Today.

QUESTIONS TO CONSIDER

1 Do you think conscious, explicit learning or unconscious, implicit

learning is more important in our daily lives?

2 Why do you think we use different mechanisms to learn conscious,

explicit information and unconscious, implicit information? Why
would that division of labor be adaptive?

Lecture 2 | What Amnesia Teaches Us about Learning 19

 LECTURE 3
Conscious, Explicit Learning

L  ARNING EXPLICIT, CONSCIOUS INFORMATION

E
IS FUNDAMENTALLY DIFFERENT THAN
learning unconscious, implicit information.
And this fact has some important
implications for optimizing our learning. In
particular, strategies that are effective for
explicit learning might not be effective for
implicit learning, and vice versa. Therefore,
this course will separately address explicit
learning and implicit learning. The next few
lectures—including this one—will be devoted
to the psychology and neuroscience of
explicit learning. 

20 The Learning Brain

 Learning about Explicit Learning
•• In an experiment testing explicit learning, John Bransford
and Jeffery Franks started with the following 4 relatively
complicated sentences:

1 The ants in the kitchen ate the sweet jelly which was on the table.

2 The warm breeze blowing from the sea stirred the heavy
evening air.

3 The rock which rolled down the mountain crushed the tiny hut
at the edge of the woods.

4 The old man resting on the couch read the story in the newspaper.

•• They then used these complicated sentences to make up simpler

sentences that conveyed one or more of the key ideas from the
original sentence. For example, from the first sentence, they
created “The ants were in the kitchen” and “The jelly was on the
table,” which conveyed some of the ideas from the original sentence
but not all of them.

•• They also varied how many ideas from the original sentence these
simpler sentences conveyed. For example, the sentence “The jelly
was on the table” basically conveys a single idea, while the sentence
“The ants in the kitchen ate the jelly” conveys 2: that the ants were
in the kitchen and that they ate the jelly.

•• They presented 24 of these sentences to subjects and asked them

to answer some questions about them. Then, they tested their
memory about 5 minutes later. Once again, they presented
sentences that conveyed ideas from the original sentences and
varied how many ideas were included in each test sentence. Some

Lecture 3 | Conscious, Explicit Learning 21

 of these sentences really had been in the original list, but others
had not. And the subjects were asked to give a rating indicating
whether each sentence was from the original list or not.

•• Subjects gave almost exactly the same ratings for new sentences
as they did for old sentences. Essentially, they had no idea which
sentences were new and which were old. This doesn’t mean that
they didn’t learn anything, but what they learned was the ideas
from the original sentences, not the exact wording.

•• For example, when Bransford and Franks plotted the confidence

ratings against the number of ideas that each test sentence conveyed,
they found that the ratings increased systematically the more ideas
were included. Subjects gave ratings close to 4 for test sentences
that conveyed 4 ideas from the original sentences and tended to
give negative ratings for test sentences that only conveyed a single
idea. And that was true whether the test sentence was new or old.

•• This experiment showed that we learn and store ideas, not

verbatim information.

Visual versus Verbal Information

•• Another characteristic of long-term explicit learning is that we
remember visual information significantly better than verbal
information. And we can exploit this fact to help us remember
more information.

•• One of the first studies in this area was conducted by Roger

Shepard, who gave college students a deck of more than 500 index
cards that each had a single word written on it and asked them
to look through them all at their own pace in preparation for a

22 The Learning Brain

 subsequent memory test. Another group of students got a deck of
index cards containing sentences. A third group of subjects saw a
series of a few hundred pictures.

•• Next, all 3 groups were given a forced-choice recognition test, in

which they were shown a bunch of stimulus pairs and one stimulus
in each pair had been studied while the other had not. Then, the
participants had to guess which stimulus they had studied.

•• Shepard found that the participants remembered about 90% of

the words and sentences that they had studied, which is pretty
impressive in its own right. But he also found that their memory for
the pictures was much better. Most of the participants remembered
more than 98% of the pictures, and many of them didn’t make
any errors.

Many people have the mistaken

idea that we have some kind
of automatic recorder in
our heads that is constantly
storing a detailed
representation of whatever
we pay attention to and
that we can just rewind
and play back the recording
whenever we want to.
Instead, we’re constantly
processing the information that we
encounter and trying to pull out the key
ideas—the gist of the information. And it’s that gist that gets stored
into long-term memory, not the detailed, verbatim information.

Lecture 3 | Conscious, Explicit Learning 23

 •• He also tested their memory for the pictures after various delays.
Even a week later, people remembered about 90% of the pictures.
In other words, memory for pictures you saw a week ago is similar
to memory for words and sentences that you saw just a few minutes
ago. These findings suggest that our memory for pictures is better
than our memory for words and sentences.

•• Leo Standing has probably done the most work investigating how
much visual information we can remember, and his results suggest
that although we may forget a little of the visual information
we encode, there always seems to be more room to store new
information, particularly for pictures.

•• In one very ambitious study, he asked participants to try to

remember up to 10,000 pictures that were presented once each
for 5 seconds. Afterward, he gave them a forced-choice recognition
test. Participants correctly identified the picture they had seen
before more than 70% of the time. Even if you factor in cases
where they didn’t remember but just happened to guess correctly,
they still were able to remember more than 6500 pictures, each of
which had been presented a single time for about 5 seconds many
days earlier.

There are at least 3 lessons that can be drawn from studies of

visual memory.
1 Scientists have been unable to find any kind of capacity limit on
how much we can store.
2 We remember visual information significantly better than
verbal information.
3 We remember vivid, striking pictures better than we do ordinary
pictures.

24 The Learning Brain

•• Standing also compared memory for what he called vivid pictures
with memory for ordinary pictures. The vivid pictures contained
striking, interesting, or unusual features that were absent in the
ordinary pictures. For example, the vivid picture might be of a
crashed plane, while the ordinary picture would be a plane on a
runway. And vividness also improved memory. In fact, participants
made at least twice as many errors when trying to remember the
ordinary pictures compared with the vivid pictures.

Previous Knowledge
•• How does what you already know affect your ability to learn new
information? It turns out that your previous knowledge can have
a dramatic effect.

•• In a famous experiment conducted by John Bransford and Marcia

Johnson, participants listened to a paragraph and then their
memory for the ideas conveyed in the experiment was tested.
Here’s the paragraph:

The procedure is actually quite simple. First, you

arrange items into different groups. Of course, one
pile may be sufficient depending on how much there
is to do. If you have to go somewhere else due to
lack of facilities, that’s the next step. Otherwise,
you’re pretty well set. It’s important not to overdo
things. That is, it’s better to do too few things at
once than too many. In the short run, this may not
seem important, but complications can easily arise. A
mistake can be expensive as well. At first, the whole
procedure will seem complicated. Soon, however, it
will become just another facet of life. It’s difficult
to foresee any end to the necessity for this task in

Lecture 3 | Conscious, Explicit Learning 25

 the immediate future, but then one can never tell.
After the procedure is completed, one arranges the
materials into different groups again. Then, they can
be put into their appropriate places. Eventually, they’ll
be used once more, and the whole cycle will then have
to be repeated. However, that’s a part of life.

•• If you don’t know what that paragraph was about, then it probably
sounded pretty weird and may have been difficult to make sense
of. But what if you know that the paragraph was about washing
clothes? Once you have that prior knowledge, everything in the
paragraph makes sense.

•• Bransford and Johnson were interested in how people’s

understanding and memory were affected by this kind of knowledge.
They tested people’s memory under 3 different conditions:

1 Subjects listened to the paragraph without any prior context.

Most of them probably had no idea what it was about but did
their best to remember as much of it as they could.

2 Subjects were told that the paragraph was about washing clothes
before they heard it. They understood the appropriate context
when they were hearing the paragraph.

3 Subjects were also told that the paragraph was about washing
clothes, but they were only told after they had already heard
the paragraph. They did, however, know what the paragraph
was about before they had to try to remember it.

•• The researchers imposed a 2-minute delay to ensure that the

information wasn’t still in working memory. Then, all 3 groups
wrote down as much as they could remember. Finally, the
researchers counted how many of the key ideas each group had
written down.

26 The Learning Brain

•• They found that the subjects who were never told that the
paragraph was about washing clothes typically remembered about
3 of the main ideas. In contrast, the group that knew the paragraph
was about washing clothes before they heard it remembered about
6 of the main ideas on average. So, they remembered about twice
as much information.

•• One of the most interesting and surprising findings from this

study came from subjects in the third group, whose responses
looked a lot like the people who were never told that the paragraph
was about washing clothes. They also only remembered about 3 of
the main ideas on average, or about half as much information as
the people who knew the appropriate context from the beginning.

•• The standard interpretation of these results is that relevant

knowledge mainly helps you learn when you’re encoding new
information into memory, rather than when you’re trying to
retrieve it from memory. If you know that the paragraph is going
to be about laundry beforehand, then you can activate all your
existing knowledge about that topic and then connect the new
information to what you already know. You can build associations
or connections with your existing knowledge base and get the new
information integrated. Then, you can use those associations to
pull the new information back out of memory when you need it.

•• After encoding, it’s too late. If you didn’t know what the paragraph
was about when you were encoding it, then you wouldn’t be able
to connect the new information to everything you already know
about laundry. You couldn’t build those associations with your
existing knowledge. And if those associations didn’t get built
during encoding, then they won’t be available when you try to
retrieve the information from memory.

Lecture 3 | Conscious, Explicit Learning 27

 Effectively Learning Explicit Information
•• Memory strategies that exploit principles that help people learn
explicit information more effectively have been around for a long
time. Probably the most famous, and certainly one of the most
effective, is called the method of loci.

•• Imagine a building with many rooms that you have a strong

memory for, such as a house that you’ve lived in. The idea is to
create visual images for new information that you’re trying to learn
and then to store those visual images in the different rooms of the
building. The more vivid and striking you can make those images,
the better. For example, if you’re trying to remember a grocery list,
you might begin by trying to store the image of a cow being milked
in your living room and then storing one more striking image in
each subsequent room in the building.

•• When you get to the grocery store, you can walk through the
building in your mind’s eye and conjure up the strange image that
you stored there. You remember the cow being milked in your
living room, so you buy milk. You continue mentally walking
through each room in the building, retrieving the associated image
and recalling the next item on the list.

•• This age-old method has been proven to work time and time
again—because it exploits the principles in this lecture: You are
storing visual information rather than verbal information; storing
vivid, striking visual images rather than standard, ordinary images;
and connecting the information that you’re trying to learn with
information you already know.

28 The Learning Brain

 SUGGESTED READING

Schacter, The Seven Sins of Memory.

Tulving, Elements of Episodic Memory.
——— , “Episodic Memory.”

QUESTIONS TO CONSIDER

1 Why do you think we remember visual information better than

verbal information?

2 What are some strategies you can employ in your daily life to
optimize your explicit learning?

Lecture 3 | Conscious, Explicit Learning 29

 LECTURE 4
Episodic Memory
and Eyewitness Testimony

T 
HIS LECTURE IS ABOUT HOW SOME OF THE
INFORMATION THAT WE LEARN IS ACTUALLY
information that we infer. We are constantly
drawing conclusions about the information
that we encounter, and some of the
information that we store during learning
may be our inferences rather than accurate
memories. And this has important real-world
implications for eyewitness testimony and for
how to interview witnesses. 

30 The Learning Brain

Evidence suggests that our memory is constructive, meaning that
we take fragments of memory and then put together, or construct,
our memory. The problem is that we can’t always tell the difference
between the true fragments and what our mind has added.

Making Inferences
•• We are constantly making inferences to construct a plausible
mental model of the situations we encounter. And we tend to
remember our mental simulation of events, including all the
inferences we made, rather than the original information itself.

•• And it turns out that we make these kinds of inferences all the
time. We do it when we’re encoding new information and when
we’re retrieving information from our memory. In an experiment
that demonstrates this fact very clearly, a group of subjects were
asked to try to remember the following story:

Carol Harris was a problem child from birth. She

was wild, stubborn, and violent. By the time Carol
turned eight, she was still unmanageable. Her parents
were very concerned about her mental health. There
was no good institution for her problem in her state.
Her parents finally decided to take some action. They
hired a private teacher for Carol.

•• A week after reading the paragraph, subjects were asked some

questions about it. But before the questions, half the subjects were
told that the story they had read last week was actually about Helen
Keller, and the other half of the subjects weren’t told that.

Lecture 4 | Episodic Memory and Eyewitness Testimony 31

 When we encounter new information, we tend to build up a
mental model of what’s going on in the situation. And constructing
such a mental model means filling in some of the details that
weren’t specified.
It’s that mental model that we learn and remember, rather than
the original information, or even just the gist of the original
information.

•• Then, all subjects were asked the same question: Do you remember
reading the sentence “She’s deaf, blind, and cannot speak?” That
sentence wasn’t actually in there, but the scientists found that
about half the group who were told that the story had been about
Helen Keller said that they remembered that sentence. In contrast,
only 5% of the control group said that.

•• The Helen Keller group was making an inference: That sentence

wasn’t in there, yet half of them said that it was. But when were they
making that inference? It had to be at the time of retrieval. After all,
there was no difference between the 2 groups of subjects until then.
Nothing was different during
encoding or the week before the
test. The difference between We make inferences all the
the 2 groups only happens right time and tend to remember
before the retrieval. the inferred information as
if it actually happened.

32 The Learning Brain

 Studies of Eyewitness Testimony
•• Our legal system often relies on the testimony of eyewitnesses.
But how reliable is eyewitness testimony? Some experiments have
explored this question.

•• One of the scientists who pioneered the study of eyewitness

testimony is Elizabeth Loftus, who designed a simple experimental
paradigm in which participants first encode some event—such as by
watching a video of a car accident, so they are now an eyewitness—
and their memory for that event is subsequently tested.

•• But before that happens, they may or may not receive some
misinformation—some subtle information that could potentially
distort their memory for the original event. Then, everybody’s
memory gets tested and the researcher determines whether the
people who were misinformed exhibit memory distortions relative
to the people who weren’t misinformed.

•• Loftus, along with John Palmer, performed one of the first

experiments like this, in which participants watched a video of
a car accident in which 2 cars collide. The 2 groups were then
asked a question about the video. The control group was asked
this question: How fast do you think those cars were going when
they hit each other? The misled group was asked a very similar
question with a minor tweak: How fast do you think those cars
were going when they smashed into each other?

•• Using the phrase “smashed into each other” suggests that it may
have been a more violent collision, while the phrase “hit each other”
doesn’t have the same strong connotation of a violent collision. But
that’s the only difference between the 2 groups.

Lecture 4 | Episodic Memory and Eyewitness Testimony 33

 •• Then, everyone from both groups is asked the same test question:
Did you see broken glass in the video? In reality, there was no
broken glass. But about 14% of the control subjects misremembered
seeing broken glass. In contrast, 32% of the subjects in the misled
group made the error. In other words, that one small change in
phrasing more than doubled the probability of a false memory for
broken glass.

•• This study demonstrates that some relatively subtle suggestive

information can have a dramatic influence on somebody’s
eyewitness testimony for an accident.

•• In another study, all participants watched a video of a Datsun, an

old brand of car, stopped at a yield sign in front of an intersection
when another car comes from behind the Datsun, passes it to enter
the intersection, and causes an accident.

•• The control group is asked this question: Did the car pass the
Datsun when it was stopped at the yield sign? The misled group
is asked a slightly different question: Did the car pass the Datsun
when it was stopped at the stop sign? After a delay, both groups
are shown a picture of the Datsun parked at a yield and another
picture of the Datsun parked at a stop sign. Then, they’re asked a
final question: Which one did you see?

•• About 75% of the control group

got it right and picked the picture
with the yield sign. But only 41% of
the misled group got the same question
right. Most of them thought they
remembered seeing a stop sign. And most
of them were completely unaware of having
been misled in any way. Nevertheless, the subtle
misinformation did affect their memory.

34 The Learning Brain

•• This is another example in which some
relatively subtle misinformation between
the time of encoding and the time of
retrieval is leading to false memories
and distorting eyewitness testimony.

•• These findings raise an important

question: What happened to the
original information? For example,
what happened to the information
about the yield sign in the people who
were getting it wrong?

•• One possibility is that the original information is sort of deleted

or overwritten by the misleading information. So, when you first
watch the video, you see the car at the yield sign. Then, there’s
some misinformation: Did the car pass the other car while it was
parked at the stop sign? Maybe at that point “stop” simply replaces
“yield” in memory. In that case, “yield” is no longer in there; it’s
been overwritten.

•• Another possibility is that “yield” is still in there, even though

“stop” has now been added. So, rather than overwriting the original
memory of the yield sign, you simply augment your memory with
the stop sign. In that case, both pieces of information are in your
memory simultaneously.

•• To distinguish these 2 alternatives, Michael McCloskey and

Maria Zaragoza conducted a famous experiment, in which all
participants see a sequence of pictures of a repairman who enters an
office, fixes a chair, and then finds and steals $20 and a calculator.
Critically, in one of the pictures, you can see that the repairman
has a hammer. The subjects are eyewitnesses to this crime.

Lecture 4 | Episodic Memory and Eyewitness Testimony 35

 •• Next, the participants read a description of what happened.
One of the narratives was misleading while another one was
not. Specifically, in the misleading narrative, the repairman was
described as having a screwdriver rather than a hammer. In the
other narrative, the repairman was described as having a tool—a
neutral description that is consistent with the original story and
isn’t misleading.

•• Finally, everyone’s memory is tested, but this time in 2 different

ways. Some of the people are asked if the repairman was
carrying a hammer or a screwdriver—directly contrasting the
original information (the hammer) with the misinformation (the
screwdriver). This is the way all the previous misinformation
experiments had been done.

•• Sure enough, McCloskey and Zaragoza found the traditional

misinformation effect: The people who were misled—who read the
narrative that mentioned a screwdriver—picked the screwdriver
more than the hammer. In fact, across 6 different experiments,
they only picked the hammer 37% of the time.

•• In contrast, the people who read the neutral description that was
not misleading correctly picked the hammer 72% of the time.
So, there was a whopping 35% difference in accuracy between
the 2 groups. This is the standard misinformation effect seen in
previous studies.

•• The critical question is why. Did

the people who fell prey to the When we learn new
misinformation effect and picked the information, we
screwdriver overwrite the hammer in tend to add it to the
their memory, or was their memory of information that we
the hammer still in there? already have stored.

36 The Learning Brain

•• To answer that question, McCloskey and Zaragoza tested another
set of subjects in a different way. Everything was kept the same as
the previous setup, but this time they changed the memory test:
Rather than asking people to choose between the hammer and the
screwdriver, they asked them whether the repairman was carrying
a hammer or a wrench. In other words, they’re not even given the
option to choose the screwdriver.

•• Subjects chose the hammer 72% of the time, which was very
similar to the people who read the neutral description. These
results suggest that “hammer” is still in their memory somewhere.
They didn’t overwrite it; instead, they augmented their memory.

Improving the Quality of

Eyewitness Testimony
•• Imagine that you’re walking
down the sidewalk and see
some bank robbers run out of
a bank, jump into a getaway
car, and speed away. The
police interview you to ask
you about what you saw. How
could they do that in such a way
that they’re more likely to get
accurate information and less likely
to get misinformation?

•• That question led psychologists to develop the cognitive interview.

Specifically, scientists who study memory came up with an
approach that police departments can use to try to elicit the most
accurate eyewitness testimony possible.

Lecture 4 | Episodic Memory and Eyewitness Testimony 37

 Imagine that you try to learn a list of words while you are
scuba diving or while you are standing on dry land. When you
try to retrieve the words, you will tend to remember the words
you studied underwater better if you go scuba diving again.
The surrounding context provides a whole bunch of cues that
got associated with the original memory. And if you reinstate
the same context, then all those cues can help you retrieve the
original information.

38 The Learning Brain

•• The cognitive interview is based on 3 key ideas: Reinstate the
original conditions as closely as possible, let the witness tell the
story without interruption before asking any questions, and when
you do ask questions, ask them in reverse chronological order.

1 To the extent that you can recreate, or reinstate, the same

conditions that were present when you encoded the original
memory, that should help you remember. Reinstating the
original conditions exploits a principle from cognitive
psychology called encoding specificity, which means that our
memories are tied quite specifically to the conditions that were
present when that information was learned.

2 The goal of letting the eyewitness tell his or her story without
interruption is to try to eliminate the possibility of introducing
a misinformation effect. A police officer asking an eyewitness
about a traffic accident he or she just saw might unconsciously
say something that could bias the witness testimony. By
avoiding asking any questions at first, there is no way the
officer’s questions can bias the eyewitness’s memory.

3 There will probably be a lot of details that the witness may

not have mentioned when he or she was first trying to tell the
story, so you need to ask questions. If you ask questions about
what happened first before asking about later events, then the
witness will construct a memory for the beginning of the event,
and some of the stuff that’s in that memory may not actually be
accurate. The problem is that anything that has been added will
then influence every subsequent answer that the witness gives.
While asking questions in reverse chronological order doesn’t
completely eliminate this problem, it does alleviate it, because
things that happened later can’t influence earlier events.

Lecture 4 | Episodic Memory and Eyewitness Testimony 39

 Learning about episodes and events is what psychologists often call
episodic memory.
Learning facts—relevant information about the world—is what
psychologists typically call semantic memory.

SUGGESTED READING

Fisher and Geiselman, Memory-Enhancing Techniques for

Investigative Interviewing.
Loftus, Eyewitness Testimony.

QUESTIONS TO CONSIDER

1 Do you think we would be better off or worse off if our memory

were more like an accurate video recording?

2 Is there anything you think we should change about the US legal

system in light of what you learned from this lecture?

40 The Learning Brain

 LECTURE 5
Semantic Memory

W  DON’T HAVE A SINGLE, UNITARY SYSTEM

E
IN OUR MIND THAT IS RESPONSIBLE FOR
learning. Instead, we have multiple brain
systems for learning different kinds
of information. For example, learning
explicit information depends on different
mechanisms—and brain circuits—than
learning implicit information. Psychologists
typically even distinguish between
different types of explicit, conscious
memory: episodic memory and semantic
memory. As you will learn in this lecture,
episodic memories are tied to a specific
time and place and are remembered from
a first-person perspective, while semantic
memory is our memory for facts—that is,
our general knowledge about the world. 

Lecture 5 | Semantic Memory 41

 Episodic versus Semantic Memory
•• Explicit memory refers to memory that is consciously accessible
and verbalizable. You know that 2 + 2 = 4; this is an example of
verbalizable, conscious, explicit memory.

•• Episodic memories are a type of explicit memory that refers to

memories for personal episodes in your life. For example, your
memory of eating breakfast today is an episodic memory.

•• Episodic memories all share some specific characteristics. First,

they are all tied to a specific time and place. Your memory of
eating breakfast this morning is tied to your kitchen table at about
8 a.m. In addition, episodic memories have a first-person, personal
perspective. These are memories that you can bring to mind, and
when you do, it’s as if you’re seeing that memory through your
own eyes again. Maybe you have a memory of walking your dog
this morning.

•• Episodic memories aren’t the only type of explicit memories. We

also have explicit memories for facts about the world. For example,
you remember that 2 + 2 = 4, and this is a piece of knowledge that
you’ve learned and stored in your memory.

•• These types of memories are not tied to a specific time and place
and are not remembered from a personal perspective. They are
just disembodied facts that you happen to know; they are part of
your general knowledge about the world.

•• Psychologists call this semantic memory. Semantic memories are

still explicit. You can still verbalize these facts and consciously
bring them to mind. But they’re not personal episodic memories
that are tied to a specific time and place.

42 The Learning Brain

 Categorization
•• How do we learn all this information and store it in such a way
that we can access it quickly and easily? Our semantic memory is
organized by category. For example, if you see a bunch of pictures
of robins, sparrows, and finches, you can quickly recognize that
they’re all birds. Even though they look different from each other,
you immediately realize that they all belong to the same category:
birds. And you have a lot of knowledge organized around that
category. For example, you know that birds lay eggs and that they
typically make nests and can fly.

•• Organizing our semantic memory around categories is very useful

because it helps us interact with the world efficiently. For example,
knowing that 2 different creatures that look a little bit different are
both birds allows us to make all kinds of useful inferences. We can
infer that they probably both fly, that they probably make nests,
that they probably lay eggs, and so on.

•• The reason categorization is so useful is because there are a lot of

similarities among the members of any given category. And those
similarities mean that we can treat them in similar ways.

•• If we can categorize something

as food, that can tell us whether Categorization is such a
we should eat it or not, which is fundamental ability that
very important for our survival. we typically take it for
Likewise, being able to categorize granted and don’t even
something as an enemy that might notice that we’re doing it.
be a threat is also very useful.

•• If categorization and category learning are so important and

fundamental, then we might expect to see lots of other species
doing it—even species that don’t seem all that smart. And sure
enough, we do see this. For example, Ramesh Bhatt, Edward

Lecture 5 | Semantic Memory 43

 Wasserman, and colleagues found evidence that pigeons learn to
categorize the objects they see and organize their world in terms
of those categories.

•• How do we categorize things? One natural idea would be that we

categorize based on how things look, or sound, or smell, or feel;
that is, maybe we categorize based on the perceptual features of
the objects that we encounter. Things that look alike get grouped
together, foods that taste similar get grouped together, and so on.

•• It turns out that our categories

are more sophisticated than that. Essentialism refers to
And this is even true in children. the idea that we can
For example, in an experiment categorize objects based
conducted by Susan Gelman on underlying features
and Ellen Markman, they found
that we can’t see but that
that children as young as 4 years
old can categorize objects based
are more essential to the
on essential underlying features nature of an object than
rather than just on superficial its perceptual features.
perceptual features.

•• The intuitive idea that many people initially have is that there are
some specific features that define a mental category. And if an
object has those features, then it’s guaranteed to be a member of the
category. If it doesn’t, then you know it’s not. So, these features are
both necessary and sufficient to determine category membership.

•• Defining features play a significant role in scientific classification

systems. For example, scientists often identify defining features
when classifying a biological organism into its family, genus,
and species.

44 The Learning Brain

•• But this is probably not the way our mind works. Our brains
probably don’t categorize things based on these kinds of defining
features that are necessary and sufficient to define a mental category.

•• To illustrate, consider the concept of a game. Basketball is a game,

and chess is a game. But can you come up with a set of defining
features that is guaranteed to tell you whether something is a game
or not? Those features need to guarantee that both tennis and
playing solitaire are in, while similar things, such as running a
marathon or swimming, are out. This is pretty difficult.

•• These kinds of issues have led psychologists to a view of

categorization that assumes that mental categories are not black
and white, with clearly defined boundaries; rather, mental
categories have gray areas, and some members of a category are
better examples of that category than others. And instead of
assuming the existence of defining features that are both necessary
and sufficient for category membership, psychologists instead view
features as being characteristic of a category—but not defining.

•• According to this view, you don’t just check off the defining
features and decide whether an object is guaranteed to be in or
out of the category. Rather, you ask how similar is it to other
members of the category.

Category boundaries aren’t black and

white. They’re fuzzy. And as you get
close to the edges of a category, people
will be unsure whether something is
in a category or not. For example,
although scientifically a tomato is
classified as a fruit, most people
think of it as a borderline case.

Lecture 5 | Semantic Memory 45

 •• And some category members should therefore be better examples
of a category than others. For example, robins seem like a really
good example of the bird category. A chicken is also a bird, but it
doesn’t seem like a very good example of a bird. This is because
chickens have fewer characteristic features of birds than robins do.
Robins fly, make nests, sing, and lay eggs. They have pretty much
all the characteristic features of birds. Chickens, on the other hand,
don’t fly, don’t make nests, and don’t sing.

Category Learning Systems

•• How do we learn these incredibly useful, often fuzzy categories
that underlie our semantic knowledge base? There is now a fair bit
of evidence that we can learn categories in more than one way—
that is, we have multiple category learning systems in our brains.

•• The distinction between the different category learning systems is

similar to the differences between explicit, conscious learning and
implicit, unconscious learning. Specifically, we can learn categories
consciously and explicitly using a kind of rule-based approach,
but we can also learn categories unconsciously and implicitly
over time by detecting subtle consistencies or regularities in our
environment. Which system we use depends a lot on the nature
of the categories we’re trying to learn.

•• For example, suppose that a child is trying to

learn what a triangle is. It’s pretty easy to try
out various conscious hypotheses that can
be explicitly verbalized. And once the
child hits on the idea that a triangle
is a closed figure with 3 straight sides,
then he or she can easily decide whether
other geometric figures are triangles or not.

46 The Learning Brain

•• That kind of category learning is a lot like hypothesis testing. You
come up with a specific, verbalizable hypothesis and then test it
to see if it correctly categorizes new examples. If it doesn’t, then
you immediately reject your previous hypothesis, come up with a
new one that fits with the examples you’ve already seen, and try
to test it. That’s rule-based category learning, and it can be very
fast and effective if the category being learned can be captured by
a straightforward verbal rule.

•• On the other hand, most of the natural categories that we encounter

in the world are fuzzy and can’t be captured by a simple verbal
rule. For example, children are constantly learning the meanings
of new words without ever being given an explicit definition. In
this case, there is no verbalizable rule that captures the category
that’s being learned, so we rely on our implicit learning systems
to gradually pick up on the subtle regularities that distinguish
members of those categories from members of other categories.

•• There is evidence that these 2 types of category learning depend

on different brain circuits. Edward Smith, Andrea Patalano, and
John Jonides performed one of the first experiments in support
of this hypothesis. They found that rule-based categorization
activated regions near the front of the brain while similarity-based
categorization activated regions farther back.

•• So, we use a rule-based explicit approach to learn categories

and also a more implicit approach that gradually picks up on
subtle regularities in the features that characterize members of
those categories.

•• While the explicit approach relies on generating verbalizable

rules and then testing them and replacing them with new rules if
they don’t work, 2 major classes of theories have been proposed
to explain how the implicit category learning system works: one
based on prototypes and the other based on exemplars.

Lecture 5 | Semantic Memory 47

 •• Prototype theories assume that we compute a kind of running
average for each of our categories. That running average is the
prototype for that category.

•• For example, the first time a child encounters a dog, he or she

stores a prototype representation of the dog category that is based
exclusively on that one dog. But when the child encounters a
second dog, he or she updates the prototype to try to capture the
similarities between the first 2 dogs. And every time the child
encounters a new dog, his or her prototype dog is getting modified
a little bit until it finally becomes a kind of an ideal dog.

•• Notice that the prototype dog doesn’t correspond to any particular

real dog that the child has seen before. It’s more like a running
average of all of them. And that prototype then forms the basis
for categorization. When the child encounters a new creature, he
or she compares it against the prototypes that he or she has stored
in memory and tries to find the best fit. Whatever matches best is
then used to determine the category of the new creature.

•• The other major class of category learning theories is based on

storing exemplars rather than prototypes. The specific items that
are stored in memory are referred to as exemplars, which just
means examples.

•• According to exemplar theories, we use our memory of specific

dogs as the basis for categorization. After encountering his or her
first few dogs, a child has a memory for each of those animals.
When the child encounters a new creature, he or she identifies the
memory that is the best match to the new creature. And if that
matching item from memory was a dog, then the child categorizes
the new creature as a dog.

48 The Learning Brain

•• It’s still an open question as to which of these 2 classes of theories is
closer to the truth because it has been very difficult to distinguish
them experimentally.

SUGGESTED READING

Markman, Categorization and Naming in Children: Problems

of Induction.
Markman and Ross, “Category Use and Category Learning.”
Rosch and Lloyd, Cognition and Categorization.

QUESTIONS TO CONSIDER

1 Based on the information that animals also form categories, do you

think the categories in a dog’s semantic memory would be similar
to those in a human’s?

2 Do you think explicit, rule-based category learning or unconscious,

implicit category learning is more important in real life?

Lecture 5 | Semantic Memory 49

 LECTURE 6
The Neural Basis of
Explicit Learning

S  FAR, THIS COURSE HAS FOCUSED ON THE

O
PSYCHOLOGY OF EXPLICIT LEARNING AND
memory. You have discovered that learning
explicit information is fundamentally
different than learning implicit information,
and that even within explicit learning,
learning and remembering episodic
information is distinguished from learning
and remembering semantic information. But
how does our brain do this? What are the
neural mechanisms that allow us to learn
new explicit information and store it for later
retrieval? Are there specific parts of the
brain that are dedicated to learning explicit
information, and if so, what are they? This
lecture will address these questions. 

50 The Learning Brain

 Brain Anatomy
•• Your brain is divided into 2
cerebral hemispheres: one
on the left and the other
on the right. Each cerebral
hemisphere is often divided
into 4 lobes: the left and
right frontal lobes at the front,
the occipital lobes at the back, the
parietal lobes at the top, and the temporal
lobes on the left and right behind your ears.

•• Directional terms are also critical when discussing brain anatomy:

Superior means toward the top and inferior means toward the
bottom, anterior means toward the front and posterior means
toward the back, and lateral means toward the side while medial
means toward the middle.

•• The 2 temporal lobes are on the bottom of the left and right side of
your brain. They are below, or inferior to, the parietal lobes and are
in front of, or anterior to, the occipital lobes. The medial temporal
lobes—the part of each temporal lobe toward the middle of the
brain, away from the ears—are the parts of the brain that were
surgically removed in Henry Molaison (the amnesiac patient you
learned about in lecture 2), and doing so led to a profound, but also
quite isolated, deficit in learning explicit, conscious information.

•• There are quite a few brain structures in the medial temporal

lobes. There are some cortical regions (the cortex is the layer of
tissue on the outside of the brain) in the medial temporal lobes,
and underneath those cortical regions are 2 major subcortical
structures: the hippocampus and the amygdala.

Lecture 6 | The Neural Basis of Explicit Learning 51

 •• Henry’s surgery removed most of this entire region in both
hemispheres. This raises a few questions: Do all of these structures
have equal importance in learning and memory, or is one of them
particularly crucial? In particular, could a smaller and more
restricted lesion also lead to impairments in learning and memory?

•• To answer such questions in humans is difficult, because we can’t

always see subtle brain damage on MRI or CT scans. We really
need to examine the brain at autopsy to get the full picture. But
that means we need to find amnesiac patients who are willing to
participate in experiments investigating their memory and who
offer to donate their brains for study after they die. Fortunately
for scientists, a number of very generous amnesiac patients have
done just that.

•• Stuart Zola-Morgan, Larry Squire, and David Amaral studied

the brains of several of these patients after their death and found
that damage that was restricted exclusively to the hippocampus
was sufficient to produce amnesia. These results imply that the
hippocampus is the key player in explicit learning and memory.

•• Subsequent work in animals has confirmed this hypothesis.

Specifically, animals with damage that is restricted to the
hippocampus also exhibit impairments in explicit memory. These
studies also helped refine our understanding of what explicit
memory is.

•• These studies involved carefully examining the characteristics

of learning and memory tasks that were impaired in animals
after medial temporal lobe damage. And these studies led to an
important insight: Learning tasks that depend on the medial
temporal lobes involve learning relationships among multiple
concepts and in a way that can be expressed in novel situations.

52 The Learning Brain

 Learning New Information
and Memory Storage
•• Learning new explicit information depends critically on the
hippocampus, and damage to this region leads to a severe anterograde
amnesia—that is, a difficulty in learning new information. But
amnesiacs don’t just suffer from anterograde amnesia; they also
typically suffer from a temporally graded retrograde amnesia. They
have difficulty remembering explicit information that they learned
a few days, weeks, or even years before their brain damage. But
their more remote childhood memories are typically intact. That’s
what is meant by a temporal gradient: The older memories are
preserved, but the more recent memories are impaired.

•• Results about the neural basis of explicit learning and memory

tell us that the medial temporal lobe structures, such as the
hippocampus, are not the ultimate site of memory storage. If
amnesiacs who have damage to these structures can still remember
their childhood, then those childhood memories obviously can’t
reside in the medial temporal lobes.

•• Instead, evidence suggests that they reside in the cerebral cortex.

Specifically, explicit memories seem to be stored in the same
cortical regions that were involved in processing the information
when it was first encountered.

•• The cerebral cortex performs many functions, and different parts

of the cerebral cortex perform different functions. For example,
your occipital lobe at the back of your brain processes visual
information. The parietal lobe at the top of your brain processes
spatial information and your sense of touch. Hearing is handled
in superior parts of your temporal lobe, and motor function is
controlled in posterior parts of the frontal lobe. And the list goes
on and on.

Lecture 6 | The Neural Basis of Explicit Learning 53

 •• When you process new information, many different parts of your
cerebral cortex are involved. For example, suppose that you see an
apple on a table, pick it up, and take a bite. Your occipital cortex
will process the visual information, and your frontal cortex will
control your movements to allow you to pick it up. The sound of
taking the bite will be processed in the auditory cortex in your
temporal lobe, and the taste of the apple will be processed in the
insula, which is under the frontal lobes.

•• It turns out that your memory of picking up and biting into that
apple depends on those same brain areas. When you remember
what the apple looked like, you’re activating the occipital cortex.
When you remember what biting it sounded like, you’re activating
the auditory cortex. When you remember what it tasted like, you’re
activating the insula—and so on.

Information Processing and

Memory Consolidation
•• Memories ultimately get stored outside the hippocampus and
medial temporal lobe structures. But if explicit memories get stored
outside the hippocampus, then why do amnesiacs experience
retrograde amnesia? In other words, why do they have trouble
retrieving memories of what happened before their brain damage?
After all, if those memories are stored outside the hippocampus,
then shouldn’t they be immune to hippocampal damage?

•• The explanation is based on memory consolidation. When we first

learn new explicit information, the memory depends critically on
the hippocampus. Even though the memory is stored in a variety of
different cortical regions, you need the hippocampus to reactivate
the appropriate patterns in each of those regions. But gradually

54 The Learning Brain

 over the course of days, months, and even years, the hippocampus
becomes less and less critical to the representation and retrieval
of the memory. The memory becomes consolidated in the cortex
and no longer depends on the hippocampus.

•• That’s why childhood memories are spared in amnesia while

memories from the days and weeks immediately before the brain
damage are not. The childhood memories have been consolidated
in the cortex and don’t need the hippocampus anymore. In contrast,
the more recent memories haven’t been consolidated yet. So, if the
hippocampus gets damaged, those memories get lost.

•• Why is our memory system organized this way? Why not just
consolidate the memory into the cortex right away? Why does
consolidation take so long? Isn’t that inefficient?

Many scientists think of the hippocampus

as a kind of organizer that gets all the
relevant cortical regions together to
regenerate the visual, auditory, and
tactile experiences associated with
a particular memory.
But over time, those cortical
regions begin to communicate
directly with each other
and no longer need the
hippocampus to get them
together. Once that happens,
the memory has become
consolidated, and hippocampal
damage will no longer affect it.

Lecture 6 | The Neural Basis of Explicit Learning 55

 •• A very interesting answer to these questions was proposed by
Jay McClelland, Bruce McNaughton, and Randy O’Reilly, who
argued that gradual consolidation of memories is actually helpful,
rather than harmful. In particular, gradual consolidation allows
us to incorporate new information into our memories without
overwriting existing information. To understand why, we need to
address how information gets processed in the brain.

•• Your brain is an extremely complex network of brain cells—called

neurons—and connections—called synapses. When a brain cell
fires, or gets activated, it sends signals through synapses to other
brain cells. Some of those signals are excitatory and tend to make
the next cell fire, while other signals are inhibitory and tend to
prevent the next cell from firing.

•• When you see an apple, a large population of neurons is firing

to represent the apple. And the pattern of activation across
those neurons can then trigger another pattern of activation
corresponding to another thought, and so on. Seeing the apple
can make you think about lunch, for example.

•• Storing a memory really means changing a bunch of synapses in

such a way that the neural activation pattern corresponding to the
new information will get activated at an appropriate time in the
future. By changing those synapses, you can also store memories.
And at a cellular level, that’s what learning does in the brain: It
changes synapses.

•• One synapse might get a little bit stronger and excite the next
cell a little more than it did before. Another synapse might get
weaker and excite the next cell a little less. By changing those
synaptic strengths in just the right way, the brain learns a long-
term memory. And when appropriate cues are encountered in the
future, they will trigger the neural activation pattern associated
with that memory. That’s memory retrieval.

56 The Learning Brain

•• But the problem is that we have to store lots of different memories
using the same synapses. And that can lead to significant
interference and memory problems, such as proactive interference,
which means that previous learning interferes with new learning,
and retroactive interference, which means that later learning
interferes with the retrieval of earlier memories.

•• So, we are left with a dilemma: We want to learn new things,

but we also want to be able to remember things we have learned
in the past. And new learning can interfere with that. This is
sometimes called the stability-plasticity dilemma. We want our
existing memories to remain stable, but we also want memory to
be plastic, or malleable, so that we can add new information to
what we already know.

•• According to McClelland, McNaughton, and O’Reilly, that’s why

slow, gradual consolidation of memories into the cortex is helpful
rather than harmful. The hippocampal learning system can
quickly learn new relationships and associations. And that makes
us plastic and flexible. But consolidation of those memories into
the cortex needs to be gradual and slow so that we don’t overwrite
the information that’s already stored there.

•• In particular, they argued that if we interleave the rehearsal

of previously stored information with the rehearsal of new
information, then we can make synaptic changes that incorporate
both memories simultaneously. But that requires regularly
rehearsing previously stored information.

•• There is actually evidence that we do this when we sleep. In a study

conducted by Daoyun Ji and Matthew Wilson, they trained rats to
run around a circular maze in alternating directions and recorded
from neurons in the hippocampus and the visual cortex. Different
cells fired at different times as the rats ran around the maze.

Lecture 6 | The Neural Basis of Explicit Learning 57

 •• But they kept recording from those cells after the rats went to sleep
and found that sequences of cell firings that occurred when the
rats were awake and running around the maze reappeared when
the rats were asleep. And that was true in both the hippocampus
and the visual cortex.

•• Furthermore, the firing sequences in the hippocampus and visual

cortex were coordinated: If the hippocampus sequence during sleep
corresponded to the sequence when the awake rat was running in
a clockwise direction, then the visual cortex sequence during the
same period of sleep also matched the awake clockwise sequence.

•• These results suggest that the memory of running around the

maze was being reactivated in the hippocampus and cortex
during sleep—which is exactly the kind of mechanism that could
consolidate memories in the cortex.

SUGGESTED READING

Burgess, Maguire, and O’Keefe, “The Human Hippocampus

and Spatial and Episodic Memory.”
Eichenbaum, Yonelinas, and Ranganath, “The Medial
Temporal Lobe and Recognition Memory.”
Hart and Kraut, Neural Basis of Semantic Memory.
McClelland, McNaughton, and Oreilly, “Why There Are
Complementary Learning Systems in the Hippocampus and
Neocortex.”
Redish, Beyond the Cognitive Map.
Squire, “Memory Systems of the Brain.”

58 The Learning Brain

 QUESTIONS TO CONSIDER

1 Most of the research on memory consolidation has focused on

episodic memory. Do you think semantic memories are likely
to be consolidated the same way? How would you test your
idea experimentally?

2 Damage to the hippocampus affects explicit learning and

memory. How do you think damage to cortical consolidation sites
would affect explicit learning and memory if the hippocampus
were intact?

Lecture 6 | The Neural Basis of Explicit Learning 59

 LECTURE 7
Strategies for Effective
Explicit Learning

T  E VAST MAJORITY OF STUDENTS HAVE

H
ADOPTED STUDY STRATEGIES BASED MORE
on trial and error than on solid scientific
evidence. The good news is that educational
and cognitive psychologists have studied this
topic and learned a lot about which study
techniques work and which don’t. In this
lecture, you will learn about methods that
have been tested in real experiments and that
have been demonstrated to improve learning
and retention of explicit information. 

60 The Learning Brain

 Ineffective Study Techniques
•• There are a few study techniques that many people swear by but
that research demonstrates aren’t the most effective.

•• Highlighting or underlining books and notes is one of the most

popular study techniques. Unfortunately, the evidence for the
usefulness of highlighting is mixed at best, at least the way it’s
normally done.

•• In one representative experiment, Robert Fowler and Ann Barker

asked 3 groups of undergraduate students to study scientific
articles about boredom and about city life in preparation for a test
that was given a week later. One group read the articles without
doing any highlighting and served as the control group. A second
group actively highlighted as much of
the text as they wanted to. And
a third group didn’t highlight
themselves but read the
articles that had already
been highlighted by the
second group.

•• All 3 groups came back

a week later and got to
study the same materials
from the week before for
10 minutes. The 2 highlight
groups studied the highlighted
articles while the control group
studied the unmarked article. Then, they all took a test on the
content of the articles. All 3 groups performed about the same.
Highlighting didn’t help. And a number of subsequent studies
have found similar results.

Lecture 7 | Strategies for Effective Explicit Learning 61

 •• In fact, one study found evidence that underlining can hurt later
test performance. Sarah Peterson asked students to study a history
chapter that they would later be tested on. One group of students
was asked to underline important text in the chapter and was able
to review their marked-up chapter before taking a test. Another
group also underlined important text in the chapter but reviewed a
clean copy of the chapter before the test. And a third group studied
the chapter without doing any underlining and also reviewed a
clean copy before the test.

•• About half the questions on the test were basic factual questions
about the material. On these questions, all 3 groups performed
about the same. Once again, marking some of the text as
particularly important—in this case by underlining—didn’t
help. The other half of the questions required students to draw
inferences based on what they had read. On these questions, the
students who had used underlining during both the initial study
period and the review period actually performed worse than the
other 2 groups.

•• One theory about why highlighting or underlining isn’t more

helpful is that it restricts your attention to isolated facts and
leads you to overlook the bigger connections among ideas. That
might explain why students who underlined performed worse on
inference-based questions compared with students who studied
a clean copy of the chapter. By focusing on isolated facts, they
missed some of the bigger connections that were tested by the
inference-based questions.

•• Another problem with highlighting and underlining is that students

often don’t do a good job of distinguishing the truly central ideas
from more peripheral information. Some students hardly highlight
anything, but the more common problem is highlighting too much

62 The Learning Brain

 information. In fact, in the original highlighting study by Fowler
and Barker, students who highlighted more did worse on the test
than students who highlighted less.

•• It’s possible that highlighting is a good study technique for some

students if they can highlight a little more thoughtfully, but we
need more research to know for sure.

•• Another study technique that is associated with mixed results is

rereading. Rereading is extremely common among students. In
fact, it may be the most common study technique. And it does
have some benefits. For example, it seems to help with fill-in-the-
blank and short-answer testing. But there isn’t good evidence that
it actually improves comprehension or performance on inference-
based questions.

•• In addition, the benefits of rereading tail off very quickly. Although

rereading once has some benefits, rereading a second or third time
doesn’t seem to add anything—yet this is what many students do.

If you have read a chapter and then reread it, you will recognize
a lot of what you read. And that familiarity may fool you
into thinking that you really know what’s in the chapter.
Unfortunately, you may be in for a rude awakening if you’re then
tested on the chapter and have to actually recall the information.
This happens all the time in classrooms around the world. Students
who do poorly on an exam are often very surprised, because they
really thought they had mastered the material. And part of the reason
is probably because rereading gave them the illusion of mastery.

Lecture 7 | Strategies for Effective Explicit Learning 63

 •• An even more serious problem is that rereading can give students
the mistaken impression that they’ve mastered the material because
it seems familiar, so many students stop studying after rereading.
The problem is that recognizing information you’ve encountered
before is quite different from being able to recall that information.

•• A final reason that rereading probably isn’t the best study strategy
is very simple. Other techniques have been demonstrated to be
significantly better. And any time you spend rereading is time that
you’re not spending studying in these other, more effective ways.

Effective Study Techniques

•• A 2013 review paper by John Dunlosky, Daniel Willingham, and
colleagues evaluated study techniques from the scientific literature
in terms of 4 criteria: whether they work for different students,
for different types of material, regardless of how learning is tested,
and in different learning environments. Two techniques that they
categorized as moderately effective are generating explanations
and interleaving practice.

•• Generating explanations refers to deliberately trying to come

up with explanations for why concepts are true or for how those
concepts are related to what you already know. Instead of passively
reading an article or chapter, the idea is to constantly ask yourself
questions about the information.

•• For example, if a chemistry chapter claims that some compound is

more reactive than another, then ask yourself why that might be.
Rather than just passively encoding information, you’re actively
engaging with it and asking deeper questions about why things
are the way they are.

64 The Learning Brain

•• One of the first studies to investigate the effect of generating
explanations was conducted by Michael Pressley and colleagues.
A common interpretation of their results is that generating
explanations forces students to process the new information deeply
and to relate the new information to information they already have
stored in their long-term memory. And both types of processing
are associated with stronger and more durable memory.

•• The other study technique that Dunlosky’s review characterized

as moderately effective was interleaved practice.

•• Imagine that you’re taking an algebra course. You’re going to be

learning about a lot of different topics, such as isolating a variable
on one side of an equation and slopes and intercepts when graphing
lines. When a test rolls around, it will cover a bunch of different
topics, including isolating variables and slopes and intercepts. How
should you study? Should you spend a block of time studying
about isolating variables, then a block of time studying slopes and
intercepts, and so on? This is blocked practice. And this is the way
most algebra textbooks are organized.

•• An alternative approach is to mix up the topics as you study. For

example, if you have a whole bunch of algebra problems covering
a variety of different topics, then you could work on them in
a random order. First, you solve a problem involving slopes
and intercepts. Then, you solve a problem involving isolating a
variable. Then, you solve one involving factoring, then another
one involving isolating a variable, and then another one involving
slopes and intercepts. This is interleaved practice.

•• There is evidence that interleaved practice may improve learning.

For example, Doug Rohrer and Kelli Taylor compared blocked
versus interleaved practice with students who were learning how to

Lecture 7 | Strategies for Effective Explicit Learning 65

 figure out the volume of different geometric forms, such as spheres
and cones. One group of students learned using a blocked approach,
while the other group learned using an interleaved approach.

•• During the practice of the formulas in the experiment, the blocked

group actually performed better than the interleaved group. This
makes sense: If you need to figure out the volume of a sphere and
just solved another problem involving the volume of a sphere, then
you’re likely to remember the formula. But to investigate which
approach is better for long-term learning, both groups were given a
test a week later on all the geometric forms that they had studied,
and the interleaved group did more than 3 times better.

•• One theory about why interleaved practice is better than blocked

practice is that seeing different types of problems in immediate
succession makes it easier to notice the differences between them,
and being able to distinguish different types of problems makes
you better at choosing the appropriate solution strategy. Another
theory is that interleaving forces you to retrieve the information
from long-term memory, and practicing that kind of retrieval
makes you better at doing so when the test rolls around.

•• There are 2 other study techniques that were rated as even more
effective by Dunlosky’s review than generating explanations and
interleaving different topics when studying. The first is distributed
practice, which refers to spreading out your study over time with
breaks or lags between study sessions.

•• This is not what most students do. In fact, scientific investigations

of study schedules have found that most students exhibit
procrastination scallop—or cramming—which means that
students tend to procrastinate and do most of their studying
right before the exam and then take a break until ramping up
and studying right before the next exam. And it turns out that
cramming isn’t nearly as effective as distributed practice.

66 The Learning Brain

•• In one study, Harry Bahrick asked students to try to learn the
English translations of a bunch of Spanish words. All the students
studied the words during 6 sessions, but Bahrick varied the lag, or
spacing, between the study sessions. One group had all 6 sessions
back-to-back on the same day, kind of like cramming for an exam.
For another group, the 6 study sessions were all a day apart. And
for a third group, the study sessions were each a month apart.

•• The students who took a month-long break between each study

session forgot the meaning of a lot of the Spanish words they had
studied, so they had to relearn them. In contrast, the students who
had all 6 sessions back-to-back on the same day didn’t forget much
from one session to the next.

•• Then, Bahrick gave everyone a test 30 days after their last practice
session to see how much they remembered. The group that crammed
all their study into a single day remembered 68% of the Spanish
word translations. The group that had study sessions separated by
a day remembered 86% of the translations. And the group that
studied once a month remembered 95% of the translations.

•• The study technique that might

be the most effective of all is the Distributing study over
method of consistently testing time leads to stronger
yourself. The benefits of testing
have been shown in hundreds
and longer-lasting
of studies. In one such study, memory than cramming.
Andrew Butler asked students Students would be much
to study a few text passages. The better off studying all
students then restudied some of their subjects for a few
the passages and took a test on the minutes every day rather
other passages. A week later, they than always cramming
took a test that assessed how much for the next exam.
they had learned from the passages.

Lecture 7 | Strategies for Effective Explicit Learning 67

 When most people think
of testing, they think of
evaluation—of grades.
Perhaps as a result, many
students would prefer
not to have to take any
tests. But psychologists
have known for a long
time that testing is
actually one of the most
effective ways to learn.

•• Across 3 different experiments, the students consistently learned

more from the passages on which they had been previously tested,
compared with the passages that they had restudied. And that
wasn’t just true for fact-based questions; testing also led to improved
performance on more conceptual, inference-based questions.

•• Keep in mind that being tested on a passage meant that the students
didn’t get to restudy it, so they had actually studied those passages
for less time than the other passages, yet they learned more. This is
a very valuable lesson for students: Rather than rereading a chapter
they’ve already read, students would be much better off asking a
friend to test them on the material.

•• Another powerful strategy is creating flash cards that have

questions on one side and answers on the other. Then, students
can shuffle the cards and repeatedly test themselves on the material
until they’ve really mastered it.

68 The Learning Brain

•• Students can even take notes in a way that encourages self-testing.
One approach is to divide their notebooks into 2 columns: One
column is a little wider and is for taking the actual notes, while
the other column is for writing down key terms and questions
related to the adjacent notes. Then, students can cover up the notes
and test their knowledge of the key terms and questions that they
wrote in the second column. This is sometimes called the Cornell
note-taking system, proposed by a Cornell University education
professor in the 1950s.

SUGGESTED READING

Brown, Roediger, and McDaniel, Make It Stick.

Carey, “Forget What You Know about Good Study Habits.”
——— , How We Learn.
Didau, What If Everything You Knew about Education
Was Wrong?
Dunlosky, Rawson, Marsh, Nathan, and Willingham, “What
Works, What Doesn’t.”
Foer, Moonwalking with Einstein.
Pauk, How to Study in College.

QUESTIONS TO CONSIDER

1 Do you have any personal study techniques that work well for you
but that weren’t discussed in this lecture? If so, why do you think
they’re effective, based on what you’ve learned in the course so far?

2 Are there any specific ways that you can apply what you learned in
this lecture to make your learning more effective?

Lecture 7 | Strategies for Effective Explicit Learning 69

 LECTURE 8
Controversies in Explicit
Learning Research

T  IS LECTURE WILL ADDRESS A FEW

H
CONTROVERSIES IN THE FIELD OF HUMAN
learning and memory. These are topics
for which there seems to be a disconnect
between what the general public believes to
be proven fact and what science has actually
demonstrated. By the end of the lecture,
you should be in a much better position to
evaluate the validity of claims you run across
in the popular press. 

70 The Learning Brain

 Learning Styles
•• Are some students visual learners while others are auditory learners?
Are some students left-brained while others are right-brained? Do
some students learn best by collaborating with others, while other
students learn best on their own? In short, do different people have
different styles of learning? And if so, should teachers tailor their
instruction to the specific learning style of individual students?

•• This certainly seems like a reasonable idea. After all, people are
quite different in every way you can think of, so it makes sense to
think that people probably also learn in different ways. And if so,
then tailoring instruction to match someone’s learning style could
have real benefits.

•• And consistent with the intuition, the idea of learning styles has had
an enormous impact in education and in society more generally.
Testing to determine learning styles has been recommended by
several education organizations, and a number of textbooks in
educational psychology discuss the merits of the approach. The
idea of learning styles has also led to commercial products and
training programs that generate significant revenue.

•• But there is not yet solid scientific evidence to support the claim
that tailoring instruction to match learning styles works. It’s
possible that future research will demonstrate the value of assessing
learning styles and designing individualized instruction, but the
evidence to date has a number of problems.

1 There is not a lot of theoretical agreement in the field about

how learning styles should be categorized. In particular, many
different learning style models have been proposed. For example,
Frank Coffield, David Moseley, Elaine Hall, and Kathryn
Ecclestone performed a systematic review of the scientific
literature on learning styles and identified 71 different models;

Lecture 8 | Controversies in Explicit Learning Research 71

 an in-depth analysis of 13 models that they considered to be
the most influential revealed very little convergence across the
models. We really don’t know what the right way to categorize
learners is, and the field doesn’t seem to be making progress
toward an agreed-upon conceptual framework.

2 A lot of the work in the field is based on what students report that
they like rather than what actually improves their learning. Most
of the tests that have been designed to assess learning styles ask
students to indicate their preferred way to perform different tasks,
and based on those preferences, the student gets categorized into
a particular learning style. But the critical issue isn’t how much
students like a particular approach, but rather how much they
learn using that approach. Learning methods that are actually the
most effective might not be what students most prefer.

In 2008, the Association for Psychological Science charged

4 leading researchers on learning—Harold Pashler, Mark
McDaniel, Doug Rohrer, and Bob Bjork—with the task of
evaluating the evidence for considering learning styles in education.
They found plenty of studies showing that one type of instruction
produced better results than another type, but it tended to be better
for all the students.
They didn’t find any studies in which instruction type A was better
for learning style A than learning style B but instruction type B was
better for learning style B than learning style A.
In fact, they had a difficult time finding studies that had randomly
assigned students with different learning styles to different types of
instruction. But in the small number of such studies that they did
find, matching the type of instruction to the student never helped.

72 The Learning Brain

 3 There is very little solid evidence supporting the claim that
tailoring instruction to match student learning styles has a
significant positive impact. For example, John Hattie reviewed
hundreds of studies in an attempt to identify the factors
that most influenced learning. The effect of individualized
instruction was so small that it’s probably not a good use of
teacher’s time. Furthermore, very few of the studies that have
claimed evidence for learning styles have actually provided a
rigorous test of the idea.

The Mozart Effect

•• A second controversial topic in the study of learning is the so-
called Mozart effect. Does listening to Mozart make you smarter?
Does exposing children to the music of Mozart and other classical
composers help them do better in school? Should parents of
preschoolers play classical music at home to help their children
become better learners? Should pregnant mothers play Mozart to
their unborn babies to stimulate their brains?

•• Many people think that the answer to these questions is yes. Many
popular books tout the benefits of listening to classical music—
not only to improve cognitive function, but also to reduce stress
and heal the body. Entire companies have sprung up marketing
and selling products based on the assumed benefits of listening
to classical music.

•• Interest in the Mozart effect was triggered in large part by an

influential study that was published as a 1-page article in Nature in
1993. In the study, conducted by Frances Rauscher, Gordon Shaw,
and Katherine Ky, 36 college students listened to 10 minutes of a
composition by Mozart (Sonata for Two Pianos in D major), or to
10 minutes of relaxation instructions, or to 10 minutes of silence.

Lecture 8 | Controversies in Explicit Learning Research 73

 •• Immediately after the 10-minute listening period, the students
completed one of 3 abstract reasoning tasks from a famous IQ
test called the Stanford-Binet Intelligence Scale. The students did
significantly better on the tasks after listening to Mozart than
they did after listening to silence or to relaxation instructions.
The authors also calculated an IQ equivalence score based on task
performance that was about 8 to 9 points higher after listening to
Mozart compared to the other conditions.

•• This is the finding that has since been dubbed “the Mozart effect,”
although the scientists who conducted the study never used that
term in the original paper.

•• The same authors published another study 2 years later in which

they compared listening to Mozart with listening to the minimalist
composer Philip Glass. Once again, they found that performance
on a task from the Stanford-Binet IQ test was better after listening
to Mozart.

In 1997, a musician and former music critic named

Don Campbell published a popular book titled
The Mozart Effect, which argued that music,
particularly classical music—such as the
music of Mozart—had the power not only to
strengthen the mind, but also to heal the body
and unlock a person’s creative spirit.
Based on the book’s success, Campbell started
an online business selling CDs intended to
enhance children’s creativity and success in
school, and even to help treat dyslexia, autism,
and attention deficit hyperactivity disorder.
These CDs were among the best-selling classical
CDs for more than 3 years.
74 The Learning Brain
•• Bruce Rideout also published 3 replications in which listening
to a Mozart composition led to improved performance on a task
modeled after the task that Rauscher and colleagues had used.

•• The popular press latched onto these findings because they seem
to suggest that the simple act of listening to 10 minutes of Mozart
can increase your IQ by almost 10 points. This is certainly an
interesting finding, but there are 2 important limitations that often
get glossed over:

1 The tasks on which performance improved came from a single

subtest of the Stanford-Binet test—specifically, the spatial
subtest. The studies really showed that listening to Mozart
makes you better at a specific kind of spatial processing. They
definitely didn’t show that Mozart makes you smarter in general.

2 More importantly, the effects were very short-lived. Specifically,

the benefits of listening to the Mozart composition only lasted
about 10 to 15 minutes. After that, the spike in spatial processing
performance disappeared.

•• An accurate summary of the results would therefore be that

spending 10 minutes listening to Mozart’s Sonata for Two Pianos
in D major made students better at a specific spatial processing
task for the next 10 minutes or so. This is fairly different than the
claim that listening to Mozart makes people smarter. But it’s still
interesting in its own right. For example, if it were true, physicians
conducting surgeries that require complex spatial processing might
benefit from a presurgical music session. Because of its potential
importance, many scientists followed up on the original study in
an effort to understand it and extend it. Unfortunately, many of
them failed to find the same kinds of effects.

Lecture 8 | Controversies in Explicit Learning Research 75

 •• Scientists don’t yet know for sure why some studies have found the
effect while others haven’t, but it’s fair to say that the effect isn’t
particularly strong or robust. Furthermore, even if the effect were
demonstrated to be real, scientists agree that it is short-lived and
restricted to specific tests of spatial processing.

Repression and Recovery of

Traumatic Memories
•• A third controversy in the field of learning and memory is
the issue of whether traumatic memories can be repressed and
later recovered.

•• The idea that it is possible to repress a memory of a traumatic event

and then recover it decades later resonates with many people. The
basic idea is that memory repression is a kind of psychological
defense mechanism. Experiencing an extremely traumatic event,
such as witnessing a murder, leads the mind to bury that memory
as a coping mechanism. Then, later, perhaps when the victim is no
longer in a threatening environment, the repressed memory might
bubble to the surface when an appropriate cue serves as a reminder.

•• A key controversy in the field is whether such recovered memories

are authentic. Are they true, or are they actually false memories?
Note that claiming that a recovered memory is false does not imply
that the person is lying; he or she may adamantly believe that the
memory is true, even if it isn’t.

•• There are some reasons to believe that many recovered memories

might be authentic. First, there are many people who claim to
have recovered traumatic memories for events they had previously

76 The Learning Brain

 forgotten, and the vast majority seem very genuine and might
actually be events they prefer not to remember. Furthermore, many
people who experienced traumatic events as children do claim that
they don’t remember them.

•• On the other hand, there are also reasons to doubt the authenticity
of at least some recovered memories. First, a number of studies
have shown that people tend to remember emotionally arousing
information better than emotionally neutral information. For
example, many combat veterans want to forget what they witnessed
but are tormented by extremely vivid memories of the horrors they
experienced. If forgetting is a psychological defense mechanism,
then you might expect the opposite.

Linda Williams interviewed more than 100 women who had

experienced some kind of sexual abuse as children, as documented
by hospital records. More than 1/3 of them did not report any
memories of the incident that necessitated the hospital visit.
Maybe they did remember, but they just didn’t want to talk about
it. After all, doing so could be extremely painful. But it turns
out that many of these same women reported other very personal
and potentially painful information about their lives, including
details about their sexual activity, abortions, drug use, and
illegal behavior.
In fact, some women who failed to report any memory of the
incident that led to the hospital visit reported other incidents of
sexual abuse from their childhood. And if they’re willing to report
one experience of sexual abuse, then why would they be unwilling
to report another?

Lecture 8 | Controversies in Explicit Learning Research 77

 •• In addition, there have been a number of cases of recovered
memories that have been demonstrated to be false. In one case, a
woman remembered being abused by her father at home at the age
of 2, but she was being raised in prison by her mother, who was
serving a sentence at the time, so the girl actually had no contact
with the father.

•• There is evidence on both sides of this very controversial issue.

Some experts believe that the vast majority of recovered memories
are authentic and mostly accurate, while others believe that most,
if not all, such memories are confabulations. The truth is probably
somewhere in the middle.

•• Memory is both fallible and malleable, so it seems not only

plausible, but highly likely, that some recovered memories are not
accurate, even if the person has no doubt about their authenticity.
On the other hand, memory repression after a traumatic event
is quite commonly reported, and the evidence to date probably
does not warrant immediately dismissing all recovered memories
as fiction.

SUGGESTED READING

Lilienfeld, 50 Great Myths of Popular Psychology.

Loftus, “Planting Misinformation in the Human Mind.”
Loftus and Ketcham, The Myth of Repressed Memory.
Pashler, McDaniel, Rohrer, and Bjork, “Learning Styles.”
Terr, Unchained Memories.

78 The Learning Brain

 QUESTIONS TO CONSIDER

1 Although the evidence to date for incorporating learning styles in

education is weak, it’s possible that future studies will demonstrate
the value. Do you think they will? In other words, do you think
that people do have significantly different learning styles and that
tailoring instruction to those styles would make a big difference?

2 Why do you think listening to Mozart was associated with

improved spatial reasoning (even if it was short-lived and only in a
subset of studies)?

3 Do you think people can repress and subsequently recover an

authentic traumatic memory?

Lecture 8 | Controversies in Explicit Learning Research 79

 LECTURE 9
Unconscious, Implicit
Learning

I  ADDITION TO LEARNING CONSCIOUS, EXPLICIT

N
INFORMATION, WE DO A LOT OF UNCONSCIOUS,
implicit learning—the subject of the next 6
lectures of this course. There are 3 major
types of implicit learning: procedural learning,
nonassociative learning, and associative
learning. Although these types of learning lead
to changes in behavior, they don’t lead to new,
explicit knowledge that can be verbalized. But
they also differ from each other in important
ways. This lecture will focus on nonassociative
learning and associative learning. 

80 The Learning Brain

 Nonassociative Learning
•• Nonassociative learning refers to changes in behavior related
to a stimulus that do not involve associating that stimulus
with another stimulus or event. When repeated exposure to a
stimulus by itself changes your reaction to that stimulus, that’s
nonassociative learning.

•• A type of nonassociative implicit learning is habituation. We

habituate to stimuli all the time, and we’re typically unaware of
it. For example, you get habituated to the sound of a fan blowing.
Over time, your response to the sound gets smaller and smaller
until finally you don’t notice it at all.

•• Admittedly, this is a very simple kind of learning, but it’s still

learning. Your behavior is changing as a result of your previous
experience—in this case, your experience of being repeatedly
exposed to a stimulus. Essentially, you’re learning to ignore it.

•• But this kind of learning is quite different from explicit learning.

You’re not learning declarative, verbalizable information in this
case; rather, you’re implicitly learning to pay less and less attention
to a particular stimulus.

•• Also, notice that this is not associative learning. You’re not

associating the sound of the fan with something else in the world.
You’re just learning to ignore one particular sensory input.

•• The opposite can also happen; that is, rather than learning to ignore
a stimulus, you can learn to become more sensitive to it. This is
called sensitization, and it’s also a form of nonassociative learning.

Lecture 9 | Unconscious, Implicit Learning 81

 •• Imagine that you’re trying to watch
a movie at the theater, but
someone in the row behind you is
constantly whispering to a friend
throughout the movie. Rather
than getting used to the sound
and habituating to it, you might
actually become more and more
sensitive as the movie goes on.
You might perceive the whispers
as getting louder, even if they
aren’t.

•• This is an example of sensitization.

Previous experience with the
whispering makes you more and more
sensitive to it.

•• And just like habituation, sensitization is a type of nonassociative

memory. You’re not necessarily learning to associate the whispers
with something else; you’re just learning to become more sensitive
to it.

•• A related learning phenomenon is priming, which occurs when

previous exposure to a stimulus makes you faster, or more efficient,
at processing similar stimuli in the future. For example, suppose
you are asked to repeatedly say a few relatively difficult-to-
pronounce words out loud, such as “otorhinolaryngologist” and
“antidisestablishmentarianism.” The more you say the words,
you’ll probably get a little faster and more fluid. Saying them the
first few times “primes the pump” and makes the words come out
more fluidly and efficiently the next time.

82 The Learning Brain

 Associative Learning: Classical Conditioning
•• Associative learning includes both classical conditioning and
operant, or instrumental, conditioning.

•• The study of conditioning began by accident. The famous

physiologist Ivan Pavlov was studying the digestive system of dogs
and decided to measure how much saliva the animals produced
when they were given meat powder.

•• But he quickly discovered that the dogs started salivating as soon

as the experimenters entered the room, which made it impossible
for him to conduct the experiment he was planning. But being the
brilliant scientist that he was, Pavlov realized that this behavior
also demonstrated a clear type of learning.

•• Pavlov devoted much of the rest of his career to studying this kind
of learning, which has come to be known as classical conditioning.
It’s also sometimes called Pavlovian conditioning in honor of
its discoverer.

•• Classical conditioning is associative because it involves learning

new associations.

•• An unconditioned stimulus is an environmental stimulus that

automatically triggers some unconditioned response. For example,
giving a dog meat powder automatically triggers salivation, so in
Pavlov’s case, the meat powder is the unconditioned stimulus
and the salivation in response to the meat powder is the
unconditioned response.

•• These are called unconditioned, or unconditional, because they’re

not conditional on any learning. Salivating when given food doesn’t
depend on learning; it’s just an automatic, reflexive response.

Lecture 9 | Unconscious, Implicit Learning 83

 •• But then, a conditioned stimulus is presented with the
unconditioned stimulus. Critically, the conditioned stimulus
does not automatically trigger the response of interest; rather, the
response is conditional on learning.

•• The conditioned stimulus in Pavlov’s case was the sound of the

experimenters entering the room. That sound doesn’t automatically
trigger salivation the way that having food in your mouth does, but
the dogs eventually learned to associate that sound with getting
food, so after a few days, they started salivating in response to
that sound.

•• Such a response—that is conditional on learning—is called

the conditioned response. Salivation in response to food is an
unconditioned response, because that doesn’t depend on learning.

•• But when Pavlov’s dogs started salivating in response to the sound

of the experimenters, that was a conditioned response, because that
did depend on learning.

•• Why are these associations being learned? Many people assume

that our brains automatically associate events that happen together
in time and space.

•• According to that idea, the reason that the sound of the

experimenters got associated with the meat powder is because the
dogs were given the meat soon after they heard the experimenters.
In other words, the association was formed because of temporal
contiguity, because the sound and the meat powder occurred
together in time. But that idea is wrong.

•• A number of studies have shown that what really matters is

contingency, not contiguity. In particular, it’s not sufficient that the
conditioned stimulus occurs at the same time as the unconditioned

84 The Learning Brain

 stimulus. The conditioned stimulus actually needs to be a good
predictor of the unconditioned stimulus, and a better predictor
than other stimuli that may be present.

•• For example, imagine that the unconditioned stimulus does tend

to occur when the conditioned stimulus occurs, but it occurs just as
frequently when the conditioned stimulus isn’t present. In Pavlov’s
case, imagine that the dogs were just as likely to get meat powder
from a machine as they were to get it from the experimenters when
they entered the room.

•• In that case, the sound of the experimenters isn’t a great predictor

of the meat powder, even though it co-occurs with the meat
powder in time. And a number of experiments have
shown that conditioning doesn’t occur under
those circumstances.

•• The appearance of the conditioned stimulus

needs to mean that the unconditioned
stimulus is more likely to occur than before.
Co-occurrence alone is not sufficient.

•• Another important finding in the study

of classical conditioning is associative
bias. Put simply, not all associations
were created equal. Some associations
are quite natural while others are more
artificial, and humans and animals both
learn the natural associations more easily.

Lecture 9 | Unconscious, Implicit Learning 85

 •• Consider learned food aversions, for example. Many people have
developed aversions to particular foods after eating a bad batch.
Maybe you can no longer stand the sight of crab because you once
ate some spoiled crab that made you sick.

•• Learned food aversions are quite common, and they can be induced
in laboratory animals very easily. For example, if rats are given a
novel food to eat and are then injected with a drug that makes
them sick, they will quickly learn an aversion to the novel food.
And they learn this aversion much faster than Pavlov’s dogs learned
their association. In fact, animals can learn food aversion after a
single trial. And they can learn it even if they don’t get sick until
hours after eating the novel food.

•• Why is learning a food aversion so much easier than learning an

arbitrary association? Presumably, it’s because eating spoiled food
is a real-life risk that animals need to avoid in order to survive.
Therefore, all animals, including humans, are particularly sensitive
to learning those kinds of associations.

Associative Learning: Operant Conditioning

•• Operant, or instrumental, conditioning involves learning how to
behave in order to receive rewards and avoid punishments. Actions
that are reinforced in a certain context will tend to be performed
more often in that context, and actions that are punished in a
situation will tend to be avoided in similar situations.

•• Like classical conditioning, operant conditioning is also a type of

associative implicit learning. But in this case, a 3-way association
is being learned between the context, the action, and the reward.

86 The Learning Brain

•• People exploit the principles of operant conditioning every day
when they try to shape the behavior of others, especially children.
The teacher who praises a young student for correctly reading a
new word and the parent who punishes a child for not finishing
his or her chores are both relying on operant conditioning to try
to shape the child’s behavior.

•• The original research on operant conditioning was published in

1898 by Edward Thorndike, who put hungry cats inside what he
called a puzzle box and measured how long it took them to escape
to get to a dish of food that was just outside. Inside each box was
a device that would open the door to the box if pressed or pulled
in the right way.

•• Initially, the cats moved around relatively randomly, clawing at the

box and meowing. But eventually, they would accidentally hit the
unlatching device, the door would open, and they would be able
to escape and reach the food.

•• The critical test was what would happen if Thorndike kept putting
the cats back in the same box. And he found that they gradually
got faster and faster at escaping. After 5 or 6 trials, the cats learned
to go straight to the lever and push it, allowing them to escape
within just a few seconds.

•• Thorndike hypothesized that escaping the box and getting food

was reinforcing the behavior of pressing the lever relative to all the
other actions that the cat could perform inside the box. And with
each successful escape, that behavior got more and more reinforced
and therefore more likely to be performed.

•• Thorndike summarized this idea in his law of effect, which

basically states that we learn to do what gets rewarded and to
avoid doing what gets punished.

Lecture 9 | Unconscious, Implicit Learning 87

 •• Operant conditioning is typically classified into 4 major
types: positive reinforcement, negative reinforcement, positive
punishment, and negative punishment. As expected, reinforcement
means that something desirable is happening, while punishment
means that something undesirable is happening. But the terms
“positive” and “negative” don’t have their normal meaning of good
and bad in this context; rather, positive means that something
is being added, while negative means that something is being
taken away.

•• In positive reinforcement, something is being added that is

desirable. If taking a specific action leads to getting something
that you like, that would be positive reinforcement. An example
is giving children a treat when they finish their homework;
the children are getting something they like as a result of a
particular behavior, so that behavior is getting reinforced.

•• During negative reinforcement, on the other hand, taking

an action leads to something unpleasant being removed. For
example, suppose that the children’s reward for finishing their
homework is that they don’t have to do the dishes. The behavior
of finishing the homework is still being reinforced, but the
reward is based on removing something unpleasant.

•• Positive punishment refers to actively punishing a specific

behavior by adding something unpleasant. An example is
making a child do the dishes because he or she used vulgar
language. The hope would be that doing so would reduce the
frequency of that vulgar language in the future.

•• Negative punishment involves removing something pleasant

as a punishment for some behavior. Think of taking away
TV privileges because of the foul language in the hopes of
discouraging that behavior.

88 The Learning Brain

Learned Helplessness
The principles of operant conditioning have had an enormous
influence in the real world in a variety of fields. Many parents
apply these principles almost instinctively in the way they raise
their children, and teachers are trained to reinforce desirable
behaviors in the hopes of helping students succeed.
Operant conditioning has also been used to explain, and
potentially treat, many psychological and social problems,
including clinical depression, addiction, and weight management.
Some of this work was inspired by findings of learned
helplessness in animals. Some of the most famous work in this
area was done by Martin Seligman and Steven Maier.
•• One group of animals could avoid a shock by pushing a
lever, and they quickly learned to do so. This is an example of
negative reinforcement. But another group of animals were
not able to avoid the shock during the training phase.
•• Then, both groups of animals were put in a new situation in
which they could escape a shock by jumping over a barrier.
The animals who had been able to press a lever to avoid the
shock during the training phase also learned to jump over
the barrier. They tried different actions until they finally hit
on jumping over the barrier. And once that behavior was
reinforced, they quickly learned to do it on future trials.
•• But the other group of animals, who had not been able to
avoid the shock during the training phase, didn’t learn to jump
over the barrier, even though it would have worked. Rather,
they passively suffered through the shocks without trying
anything to avoid them.
The standard interpretation is that this second group of animals
had learned something harmful—that they couldn’t avoid shocks
no matter what they did. As a result, they didn’t even try to
avoid the shocks during the second phase of the experiment.
They had learned that they were helpless.

Lecture 9 | Unconscious, Implicit Learning 89

 SUGGESTED READING

Eichenbaum and Cohen, From Conditioning to

Conscious Recollection.
Reber, Implicit Learning and Tacit Knowledge.

QUESTIONS TO CONSIDER

1 Do you think it’s significant that implicit memories are

unconscious? Suppose that you became consciously aware of all
the implicit learning going on in your brain. How would that
affect you?

2 What are some examples of when you’ve been conditioned in

your life?

90 The Learning Brain

 LECTURE 10
The Psychology of
Skill Learning

L  ARNING ANY COMPLEX MOTOR SKILL IS

E
DIFFICULT. THE FIRST TIME YOU HIT A TENNIS
serve or golf ball, or tied your shoe,
it required all of your effort and
concentration—and you still probably didn’t
do very well. But with enough practice, you
eventually learn to perform the skill kind of
automatically. And when you do, it doesn’t
even require your full attention. How do you
go from being a complete novice at a skill
to getting pretty automated so that you can
perform other tasks at the same time? That’s
skill acquisition—the topic of this lecture. 

Lecture 10 | The Psychology of Skill Learning 91

 Skill Acquisition
•• A standard way that psychologists think about skill acquisition
is as converting explicit, declarative knowledge into an implicit,
procedural skill. How do we go from knowing that to knowing how?

•• Explicit, declarative knowledge is knowledge about a skill that you

can verbalize and talk about—declare. It’s the book knowledge and
verbalized instructions about how to perform a skill. But actually
doing a skill requires implicit, procedural memory. Just because
you can talk about how to do a skill, that doesn’t mean that you
can actually do it. Somehow you need to convert the declarative
knowledge into a procedural skill that you can actually execute.
And that takes practice and time.

•• Imagine that you are just starting to learn to play the piano and
that your first lesson is how to play a D major scale. The first time
you do this, you will be pretty slow. But the second time you do
it, you will probably be a little bit faster. And after about a week of
practice, you will be much faster. You’re not going to be a skilled
pianist after just a week of practice, but you will be dramatically
better than you were on the first day.

When learning a complex skill,

such as piano, tennis, or golf, the
most dramatic improvements
will be made in the first few
days and weeks of practice.

92 The Learning Brain

•• If you continue to practice,
then you’ll still continue Actually doing a skill requires
to improve. You’ll still implicit, procedural memory—
get faster at playing that what some people call muscle
musical scale. But the rate memory, even though it’s stored
of your improvement will in the brain, not in the muscles.
slow down. When you’re
practicing a complex skill,
you’ll improve a lot initially, but it becomes more and more difficult
to get better as you become more and more skilled. You reach a
plateau, and it takes more and more practice to make even small
improvements. This characteristic pattern that is observed when
people are learning new skills is called the power law of practice.

Stages of Skill Acquisition

•• Paul Fitts and Michael Posner came up with a very influential
theory that proposes that we go through 3 major stages over the
course of skill acquisition: the cognitive stage, the associative stage,
and the autonomous stage.

1 The cognitive stage is dominated by cognition—that is, by

thinking, or by explicit, declarative knowledge.

a Suppose that you’ve never played golf and that you take
a lesson. The golf pro will probably start out by giving
you some verbal instructions, such as “keep your left arm
straight when you swing.” Now you have a bunch of verbal
instructions that you will dutifully try to execute. In fact, you
may try to memorize some of the verbal instructions or even
write them down. And you may consciously and deliberately
rehearse the verbal, declarative knowledge that the pro told
you as you’re trying to hit a golf ball.

Lecture 10 | The Psychology of Skill Learning 93

 b One of the key characteristics of the cognitive stage is that
it’s cognitively demanding; that is, it requires all of your
attention. So, you can’t really do other things when you’re
in the cognitive stage. All of your attention has to be focused
on performing the task and rehearsing the facts that you’ve
committed to memory.

c Notice that there’s a big difference between knowing the

declarative knowledge associated with the skill and being
able to execute that skill. Having the declarative knowledge
is different from having the skill; you could have one and
not the other.

2 The associative stage involves tweaking the skill, associating it

with different responses, and hopefully improving. It involves
figuring out what works and what doesn’t and using that
feedback to slowly get rid of actions that lead to errors.

a Suppose that you’ve been taking tennis lessons for a

while and you’re trying to improve your serve. You go
to the court and hit a bunch of serves. As you practice,
you begin to fine-tune your serve by making changes
that lead to better results.

b You’re associating tweaks with

outcomes, or results. And with
additional practice, you start to figure
out what tweaks work and what tweaks
don’t work. For that to happen, you need
to be getting feedback. That feedback could
come from another person, but it doesn’t have
to. You just need to be able to observe what works
and what doesn’t.

94 The Learning Brain

 3 The autonomous stage is the point at which the skill can be
performed really well without having to think about it.

a Think of the skilled golfer

who can hold a conversation Many skilled athletes
while hitting golf balls at the and musicians will tell
range. Unlike the cognitive you that verbalization
stage, which requires a lot of is the enemy in the
attention, the autonomous autonomous stage:
stage doesn’t. Performing When you get good at
the skill is no longer nearly something, you want to
as cognitively demanding. just let it happen; if you
In this stage, there is less think too much, you
dependence on verbalization
might actually mess up.
and declarative knowledge.

b Furthermore, declarative knowledge about the skill may

actually become less available the more skilled you get. Maybe
when you were taking lessons initially, you memorized all
these things about the golf swing and could even recite them
from memory. But once the skill gets automated, you may
actually forget some of the declarative knowledge that you
used to rely on.

c Getting feedback is crucial during

Skilled performance the associative stage of skill
in the autonomous acquisition, but this is not the
stage depends a lot on case during the autonomous stage.
proprioception, which Musicians and athletes at this stage
refers to one’s ability of skill acquisition would typically
to sense or perceive know what happened even if they
the movement and weren’t getting much feedback.
location of one’s own
body parts in space.

Lecture 10 | The Psychology of Skill Learning 95

 How Skill Acquisition Happens:
A Theory
•• When practicing, how do we move from the cognitive to the
associative stage and from the associative to the automatic stage?
In other words, how do we transform declarative knowledge—the
book knowledge and verbalized instructions about how to perform
a skill—into a procedural skill?

•• One of the most influential answers to this question was

developed by John Anderson, who proposed that the nature of
our representation of procedural skills is very different from our
representation of declarative knowledge. In particular, he argued
that we represent procedural knowledge using production rules, or
associations between some conditions and some actions.

•• You can think of a production rule as an if-then association: If the

conditions are satisfied, then perform the action. Critically, this
association is not explicit or conscious or declarative; rather, the
association is automatic, unconscious, and implicit.

•• When certain conditions are satisfied, then a certain association

fires automatically, and the actions associated with those conditions
are automatically and immediately executed. You don’t have to
consciously decide to take those actions. That’s what makes it
implicit and procedural, rather than declarative.

•• For example, when you walk, you’re not usually thinking about
how to walk; instead, it’s an automatic set of associations. That’s
what production rules are: automatic associations. Walking is not
conscious; it’s an implicit, unconscious, procedural skill.

96 The Learning Brain

•• The idea is that your ability to walk is stored as a big set of
automatic production rules—the automatic associations between
conditions and actions. And these production rules are stored in
a procedural memory system, which is completely separate from
the explicit, declarative memory system.

•• According to Anderson, the key question is how to convert

declarative knowledge into production rules. Suppose that you
read a book about how to hit a tennis serve. How do you convert
that declarative knowledge into some automatic associations that
are stored in procedural memory and that will fire automatically
when you want to hit the ball?

•• Anderson refers to the conversion process as knowledge

compilation, in which you compile declarative knowledge and
turn it into procedural knowledge. In computer science, a compiler
takes a high-level description of the program you want to run and
transforms it into an executable form. In this case, the high-level
description is in English rather than a programming language,
and the executable form is a set of production rules rather than a
computer’s machine code—but the basic idea is the same.

•• According to Anderson, as we’re learning a skill, we’re taking

a high-level declarative description of what we want to do and
converting it into a form that our motor system can actually
execute. How do we transform explicit, declarative knowledge into
implicit production rules? How do we compile our knowledge?
Anderson proposed that there are 2 major stages: proceduralization
and composition.

•• Proceduralization involves taking an individual piece of declarative

knowledge and converting that piece into a single production
rule—an automatic if-then association between conditions
and actions.

Lecture 10 | The Psychology of Skill Learning 97

 •• For example, suppose that you’re trying to teach a child how to
tie his or her shoelaces and you start by telling him or her what to
do: “First, cross that right lace underneath the left lace and then
form a loop with that right lace. Then, wrap the left lace around
the base of that loop and push it through and out.”

•• The first step is to turn those individual pieces of declarative

knowledge into individual production rules—that’s
proceduralization. Consider the first part of the instruction: Cross
the right lace underneath the left lace. What would the production
rule corresponding to that instruction look like?

•• The production rule is just an automatic association between a

set of conditions and a set of actions. The actions in this case
correspond to a set of motor commands that will grab the right
lace and slide it underneath the left lace. The conditions specify
when those actions should be performed, such as having the goal
of tying your shoes, looking at your shoes, and seeing that they’re
untied. Under those conditions, you should perform the action of
crossing the right lace underneath the left.

•• You proceduralize that first piece of declarative knowledge by

creating an automatic association between those conditions and
those actions: If you have the goal of tying your shoes, you’re
looking at your shoes, and you see that they’re untied, then execute
the motor commands to grab the right lace and slide it underneath
the left lace.

•• That association is implicit, automatic, and procedural. It’s not

declarative anymore. You’re not talking yourself through the
process; your hands are just doing it.

•• The next step is to proceduralize each of the other pieces of

declarative knowledge. So, if the next instruction was to form
the loop, then you would create an automatic production rule to

98 The Learning Brain

 perform that action. The conditions might specify that you’re tying
your shoes and you’ve already crossed the right lace under the left
one, and the actions would be motor commands to form the loop.

•• This process continues for each of the other instructions for tying
shoelaces. You create separate production rules corresponding to
each piece of declarative knowledge.

•• The second part of knowledge compilation is composition,

which involves combining separate production rules together
into a single, more complicated production rule—in other words,
grouping them together. When you get good at tying your shoes,
you don’t do it piece by piece; rather, you perform the whole skill
as an integrated, fluid set of motions. Once you’ve composed this
master production rule for shoe tying, it will handle the whole job
in one fell swoop.

SUGGESTED READING

Anderson, “Acquisition of Cognitive Skill.”

——— , Cognitive Skills and Their Acquisition.
Fitts and Posner, Human Performance.

QUESTIONS TO CONSIDER

1 Do you think the cognitive stage or the associative stage is more

important when acquiring a skill?

2 Psychologists typically treat learning a sport and learning an

instrument similarly (as examples of motor skill learning). Can you
think of any important differences between them?

Lecture 10 | The Psychology of Skill Learning 99

 LECTURE 11
Language Acquisition

T  IS LECTURE IS ABOUT ONE OF THE

H
LEARNING BRAIN’S GREATEST ACHIEVEMENTS:
learning how to speak and understand a
language. Natural languages are arguably the
most complicated of all human inventions.
They involve hundreds of thousands
of words that are formed according to
complex rules that often include hundreds of
exceptions. And the words themselves have
to be combined according to sophisticated
syntactic rules that the world’s top linguists
still don’t completely understand. Yet toddlers
across the world learn these amazingly
complicated inventions. 

100 The Learning Brain

If you start learning a language as an adult, it’s very unlikely that
you’ll ever get as good as someone who learned that language as
a child.
In general, people need to start acquiring a language before age
10 to 12 if they ever want to speak it like a native. Those early
childhood years are therefore often referred to as a critical, or
sensitive, period for language acquisition.
It’s not about how long you work on the language; it’s about when
you start acquiring it.

Lecture 11 | Language Acquisition 101

 The Problem of Language Acquisition
•• All of us comprehend and generate millions of different sentences,
and we do this effortlessly—and very quickly. This is true even
though many of the sentences that we produce and many of the
sentences that we hear are completely novel. We know more
than 100,000 words in our native language and are constantly
combining these words in novel ways without even thinking about
it. And this ability is unique to humans; no other species does this.

•• The problem of language acquisition—how challenging the task

of acquiring a natural language is—can be illustrated by walking
through some of the obstacles that babies have to overcome to
learn their native language.

•• First, they have to realize that not all sounds are language. They
need to distinguish language sounds—that is, speech—from all
the other sounds in the environment, such as the sounds of the
doorbell ringing and a dog barking. This is the first obstacle in
language acquisition: just figuring out what the target is. What
are the sounds that babies need to focus on to start learning a
specific language?

•• Let’s say that a baby has figured out that some of the weird sounds
that come out of his or her parents’ mouths correspond to a kind of
communication. And the baby begins focusing on them and trying
to figure them out. That’s a very far cry from actually understanding
what is being said, much less being able to produce language. And
before a baby can understand what it means, he or she needs to be
able to recognize the sounds that make up the words.

•• Every language has roughly 40 to 100 different language sounds

that get combined to form the words of the language. Just as we
combine 26 different letters in different ways to spell words, we
also combine a relatively small set of language sounds in different

102 The Learning Brain

 ways to pronounce words. These basic, atomic language sounds
are called phonemes. And we all need to recognize the phonemes
in someone’s speech before we can figure out what the person
is saying.

•• For example, when you hear the word “cat,” you need to somehow
chop this continuous stream of noises into component phonemes:
“cuh,” “ah,” and “tuh.” And even if you have done that, you also
need to figure out which phonemes are grouped together in the
same word and which phonemes are in different words. That’s
actually very difficult, because when people speak, all of the words
glide together. It’s just a continuous stream of phonemes, and it’s
not at all clear where the word boundaries are.

•• That’s another major problem that babies have to overcome: Once

they have figured out what the language sounds are and recognize
the individual phonemes, they still have to chop the sequence of
phonemes up into words.

English speakers have adopted

the convention of producing
the sound “apple” when
they want to refer to a
crisp, spherical fruit
that grows on trees and
is usually red. But to
a baby who is learning
English, the sound of
the word “apple” doesn’t
really provide any clues
about what the word means.

Lecture 11 | Language Acquisition 103

 •• Let’s say that they’ve done all that. They still need to figure out
what the words mean. And for the vast majority of words, the
sound doesn’t give any clue about its meaning. That’s because
the mapping from the sound of a word to its meaning is usually
completely arbitrary; it’s just a cultural convention that all the
speakers of a language adopt to allow them to communicate.

•• Worse yet, the same sound can mean different things, and different
sounds can mean the same thing. For example, the word “bank”
can be used to refer to a financial institution or to the side of a
river. Conversely, the words “car” and “automobile” refer to the
same thing. Such irregularities make the task of learning word
meanings very difficult.

•• But let’s say that you have distinguished language sounds from
other sounds, parsed language sounds into individual phonemes,
figured out where the word boundaries are, and even begun to
figure out what some of the words mean. You still have to figure
out the grammar of your language. How do you put different
words together into phrases? How do you combine phrases into
sentences in a grammatical way?

•• And that grammatical knowledge has to generalize. In other words,

you can’t just imitate and repeat back sentences you’ve heard in
the past; you have to learn general-purpose grammatical rules that
you can apply to novel sentences that you’ve never heard before.

•• Children are learning a lot about language in the first few years
of life. And they do all of this without a lot of feedback. In other
words, adults don’t usually correct their children’s grammar or
pronunciation. Psycholinguists who have tried to correct their own
children’s grammar found that providing explicit feedback when
the child makes a mistake doesn’t help him or her acquire language

104 The Learning Brain

 any faster than other children do. Apparently, children simply
correct themselves over time. They implicitly learn the rules of
their language despite the lack of feedback.

Genetic Wiring and Linguistic Universals

•• The task of acquiring a language is extremely difficult, yet all
normal human children acquire natural language—and they do so
despite the fact that they aren’t getting much feedback about how
they’re doing. All these considerations led scientists to propose that
human beings must in some sense be born to learn language. In
other words, language learning is genetically wired into the brains
of all normal human beings from birth.

•• Babies automatically work on figuring out how their language

works, what sounds correspond to what words, what those words
mean, and how those words get put together into grammatically
acceptable phrases and sentences.

•• The basic argument is that if we weren’t genetically wired to

acquire language, then there is no way that babies could do it.
The task is just too daunting.

•• Babies in different parts of the world have to learn different

languages. American babies are not specifically wired to learn
English, and French babies are not genetically wired to learn French.

•• The claim is that human beings are genetically wired to learn

whatever natural language they’re exposed to, whether that
language is English, French, or one of the roughly 7000 other
natural languages spoken in the world today.

Lecture 11 | Language Acquisition 105

 •• But that only works if all those languages share some common
principles that babies everywhere can expect and exploit. And it
turns out that all languages do share some commonalities.

•• For example, all natural languages are based on words. And those
words are based on combinations of specific sets of phonemes.
Those words are divided into specific parts of speech, such as
nouns and verbs. And those parts of speech help determine how
words are put together into phrases and sentences.

•• These kinds of common principles are sometimes called linguistic

universals, meaning that they’re shared by all the world’s natural
languages. Some languages tend to put the verb at the end of the
sentence while others tend to put it in the middle, but regardless
of the language, there will be a verb. And that linguistic universal
constrains the language acquisition problem and makes it
more tractable.

Language Development
•• What do we see when we actually observe children trying to
acquire language? Let’s start at the level of just distinguishing the
phonemes of a language.

•• Different languages use different phonemes. For example, English

has phonemes corresponding to the sounds “ull” and “err.” That’s
what distinguishes the word “light” from the word “right.” Japanese,
on the other hand, does not have these 2 separate phonemes; rather,
it has a single phoneme that’s somewhere in the middle. This is
why adult Japanese speakers often have a difficult time hearing
the difference between the word “light” and the word “right.” And
there are plenty of phonetic differences in other languages that
English speakers have a difficult time hearing.

106 The Learning Brain

•• Babies have to be prepared to learn any language, and it turns out that
they are able to hear the subtle phonetic distinctions that are present
in every language. Six-month-old infants can often hear phonetic
distinctions that adults actually have a difficult time noticing. For
example, a baby in Japan can tell the difference between “light”
and “right” even though that baby’s parents might not be able to.

•• Research has shown that babies can hear the phonetic distinctions
in any natural language. But as they get more and more exposure to
their native language, their speech perception gets specialized to that
language. And sounds that correspond to different phonemes in that
language begin to sound more distinctive, while different sounds
that correspond to the same phoneme begin to sound more similar.

•• Some of the major stages of language development that children

go through include the holophrastic stage, sometimes called the
1-word stage; the telegraphic stage, sometimes called the 2-word
stage; and the stage at which they start learning the syntax, or
rules, of their grammar.

•• In the holophrastic stage, children are producing 1-word

utterances, such as “mama” and “daddy.” One key feature
of this stage is evidence of both undergeneralization and
overgeneralization. Undergeneralization occurs when a child
uses a word in a more restrictive sense than is appropriate. For
example, the word “dog” could refer to millions of animals
around the world, but a child in this stage may use “dog” only
to refer to his or her own dog. Overgeneralization would be
using the word “dog” to refer to cats as well as dogs.

•• In the telegraphic stage, children can pretty much communicate

anything they want to very succinctly. Even though there is
limited linguistic ability, children in this stage can communicate
extremely effectively using simple 2-word utterances, such as
“dog bark” and “drink milk.”

Lecture 11 | Language Acquisition 107

 •• Another important milestone
is learning the complex If you track the correct
grammatical rules of the use of irregular past tense
language. To take just one forms (such as “went”) as
example, children need to a function of a child’s age,
learn how to form the past children get worse before
tense of verbs. In the past they get better at using
tense in English, there are irregular past tense forms
2 different kinds of verbs: as they get older.
regular verbs, where you form
the past tense by adding “-ed” The U-shaped acquisition
(Today you “laugh” while curve supports the
yesterday you “laughed”); hypothesis that children
and irregular verbs, where you are beginning to learn
don’t form the past tense by general-purpose rules and
adding “-ed” (Today you “go” apply them to other words.
while yesterday you “went”).

SUGGESTED READING

Clark, First Language Acquisition.

——— , Language in Children.
Pinker, The Language Instinct.

QUESTIONS TO CONSIDER

1 Do you think the way we learn a language is fundamentally

different than the way we learn other skills?

2 Do you think we learn a second language in the same way we learn

our native language? What are some similarities and differences?

108 The Learning Brain

 LECTURE 12
The Neural Basis of
Implicit Learning

T  IS LECTURE IS ABOUT THE NEURAL

H
MECHANISMS UNDERLYING IMPLICIT
learning. How is simple nonassociative
learning, such as habituation and
sensitization, actually implemented at a
neural level? How does the brain change
during classical and operant conditioning? 

Lecture 12 | The Neural Basis of Implicit Learning 109

 Neural Mechanisms Underlying
Nonassociative Learning
•• Many of the early insights in implicit memory came from studying
the sea snail Aplysia californica (pictured below), which exhibits
many simple implicit learning mechanisms. For example, it exhibits
habituation to being touched. Initially, it quickly withdraws its
gill when it’s touched, but if it’s touched repeatedly and gently,
it eventually learns to ignore the touch
and stops withdrawing its gill. It
also exhibits a simple form of
sensitization, in which its
gill withdrawal reflex gets
stronger over time under
the right circumstances.

•• What changes in the brain

when we learn? Is the newly
learned information stored
in new brain cells that grow
in response to the learning, or
are new connections created between
existing neurons? Is there a specific place in the brain that is critical
to implicit learning, similar to the way the hippocampus and
medial temporal lobes are critical for explicit learning?

•• Eric Kandel and his colleagues started working on these kinds of

questions by studying the gill withdrawal reflex in Aplysia. And
those studies eventually led to the Nobel Prize in Physiology.
They began by mapping out the exact neural circuit involved in
the gill withdrawal response. Because the nerve cells in Aplysia
are big and easily identifiable, they were able to stimulate and
record from individual neurons to see if they were involved in this
reflex response.

110 The Learning Brain

•• They discovered that the circuit was always the same. They could
identify the same neurons in different animals, and the neurons
were always connected in the same way. And they could easily
identify specific neurons and repeatedly return to study them, both
before and after learning, even in different animals.

•• How does this circuit change after learning? First consider

habituation, one of the simplest forms of implicit learning. Aplysia
has a gill that it uses for breathing and a siphon that it uses to expel
water. When the animal is at rest, the gill and siphon are often
extended outside of the body. But if the siphon is gently touched,
the animal will withdraw them in a defensive reaction—the gill
withdrawal reflex. But if you repeatedly touch the siphon and
nothing bad happens to the animal, it eventually gets used to it
and stops withdrawing.

•• Kandel and his colleagues carefully examined the neural circuit

of the gill withdrawal reflex before and after habituation to try to
figure out what had changed that might account for the change in
behavior. And at first, they didn’t see much. All the same neurons
were there after learning as had been there before. No neurons had
grown, and no neurons had been removed.

•• Furthermore, the number and location of connections between

the neurons appeared to be unchanged. The neurons that had
been connected before habituation were still connected after
habituation. And there weren’t any new connections that hadn’t
been there before.

•• So, why did the snails behave any differently? Kandel and his
colleagues discovered that the neurons in the circuit reduced
the strength of their signals. Specifically, they released less
neurotransmitter than they did before habituation.

Lecture 12 | The Neural Basis of Implicit Learning 111

 •• Neurotransmitters are chemicals that neurons use to communicate
with each other. Essentially, when a neuron wants to send a
signal to another neuron, it releases a bunch of neurotransmitter
molecules in the synapse, the very small space between one neuron
and the next. The neurotransmitter molecules then move across
the synapse and bind to special receptor molecules on the next
neuron and turn them on.

•• When those receptors get turned on, they might trigger the next
neuron to fire and send a neurotransmitter signal to another
neuron—and so on. That’s the way information gets sent through
the neural circuits in your brain. All of your perceptions, decisions,
and actions are based on neurons firing and triggering other
neurons to fire.

•• Before habituation, touching the siphon leads a bunch of neurons

to fire and release a lot of neurotransmitter molecules. First,
sensory neurons in the siphon fire in response to the touch itself.
These sensory neurons are connected to motor neurons that
can withdraw the gill, and when the sensory neurons fire, they
release a bunch of neurotransmitter molecules into these synapses.
The neurotransmitter molecules bind to receptors on the motor
neurons and turn them on. And that causes the motor neurons
to start firing vigorously and leads to the withdrawal of the gill.

•• But Kandel and his colleagues found that after repeated gentle
touching, the sensory neurons in the siphon started releasing
less and less neurotransmitter. As a result, the receptors on the
motor neurons weren’t turned on as much, and the gill withdrawal
response became smaller and smaller. That was the neural basis
of habituation: reduced neurotransmitter release. And similar
findings have been reported in many other species.

112 The Learning Brain

•• The snail’s habituation to being touched usually only lasts a few
minutes. So, if you wait 15 minutes and then touch the siphon
again, the snail will withdraw its gill just like it used to before
learning. But Kandel and his colleagues found that if they kept
touching the siphon repeatedly over the course of a few days, then
the habituation would last much longer. In that case, they could
touch the animal’s siphon more than 3 weeks later and it still
wouldn’t withdraw its gill very much. In other words, the animal
had learned a long-term implicit memory.

The human brain contains nearly 100 billion neurons, or nerve

cells. And because neurons tend to make about 1000 connections
each, the total number of connections is around 100 trillion.
To put that number in perspective, there are probably more
neural connections in your brain than there are stars in the
Milky Way galaxy.

Lecture 12 | The Neural Basis of Implicit Learning 113

 •• Craig Bailey and Mary Chen investigated what makes the memory
last longer by carefully counting the number of active zones on
neurons—that is, the specific regions at the end of the neurons
where they made synaptic connections with other neurons. They
did this both before and after habituation. And they found a
profound structural change.

•• After long-term habituation, Changing the strength

the animals had fewer than half
of connections between
as many active zones as before
habituation. Apparently, the
neurons is a fundamental
animals had pruned away a lot of the mechanism underlying
synaptic connections supporting learning in the brain.
the gill withdrawal feedback.

•• In addition to habituation, sensitization—in which repeated

exposure to a stimulus makes you respond more and more—is
another kind of nonassociative memory. Scientists also studied
this kind of learning in Aplysia. Although repeated, gentle
touching of the siphon leads to habituation, repeated shocks lead
to sensitization; that is, after being shocked repeatedly, the animal
will exhibit a larger-than-normal gill withdrawal response if it’s
touched afterward, even if now there is no shock. It’s as if the
animal is now very wary of any kind of touch.

Implicit learning occurs in the same neural circuits that control

the behavior that is changing. This is fairly different from
explicit learning, in which a specific region of the brain—the
hippocampus—is dedicated to that function specifically.

114 The Learning Brain

•• The same neural mechanisms that underlie habituation also
seem to be at work during sensitization, but in the opposite
direction: Whereas short-term habituation leads to a decrease
in neurotransmitter release, short-term sensitization leads to
an increase in neurotransmitter release, and while long-term
habituation is associated with a decrease in the number of synaptic
connections, long-term sensitization is associated with an increase
in the number of connections.

Neural Mechanisms Underlying

Associative Learning
•• Next, let’s consider associative learning, such as classical and
operant conditioning. What was changing in the brains of Pavlov’s
dogs when they learned to associate the sound of experimenters
with the reward of food?

•• One of the key ideas to come out of studies on conditioning is

the idea of reward prediction error. We tend to learn associations
between stimuli and rewards, or between behaviors and rewards,
when we fail to predict those rewards.

•• Consider the phenomenon of blocking: If you’ve already learned

an association between 2 stimuli, then that existing association
can block the learning of other associations. For example, if you’ve
already learned an association between seeing a red light and
hearing a fan, then you might not learn an association between a
green light that starts coming on later that also predicts the fan.
In that case, the green light does not become associated with the
fan, because the red light could already predict it. As a result, there
was no prediction error and no new learning.

Lecture 12 | The Neural Basis of Implicit Learning 115

 •• In one of the most exciting neuroscientific discoveries of the last 25
years, Wolfram Schultz found what now appears to be the neural
basis of this kind of prediction error. He was studying neurons in
a small part of the brain called the ventral tegmental area (VTA),
which is in the midbrain at the top of the brain stem.

•• Schultz had implanted electrodes in the VTA of some monkeys,

and he was recording the activity of dopamine neurons—which
use the neurotransmitter dopamine to send signals to other
neurons—in that brain area while the monkeys were performing
a learning task.

•• The monkeys were performing a basic conditioning task.

Specifically, if they pressed a lever after a light flashed, then they
would get a squirt of good-tasting juice, which served as a reward.
Schultz recorded neural activity in the VTA while the animals
learned to press a lever so that he could watch how VTA neurons
behaved before and after learning.

•• When the monkeys first started doing the task, Schultz noticed
that the VTA neurons tended to fire when the monkey got a juice
reward. He saw a big spike of neural activity as the VTA neurons
fired and released dopamine. But once the monkeys figured out
that they would get the juice as long as they pressed the lever
when the light came on, then the VTA neurons stopped firing in
response to the juice.

•• That’s exactly what you would expect a reward prediction error

to look like in the brain. These neurons fire a lot when they get a
reward that they weren’t expecting and don’t fire when they get a
reward that they were expecting. In other words, the neurons fire
in response to a reward prediction error.

116 The Learning Brain

•• After the monkeys had learned the task, the VTA dopamine neurons
started firing in response to the light rather than the juice. Like
Pavlov’s dogs, the VTA neurons in Schultz’s experiment initially
fired in response to the juice reward, but after conditioning, they
started firing in response to the stimulus that predicted the juice:
the flash of the light.

•• In addition to the midbrain dopamine system, another neural

mechanism that plays a critical role in associative learning, and
is also critical in many other types of learning, is long-term
potentiation (LTP).

•• In 1949, Donald Hebb wrote The Organization of Behavior, in

which he laid out a theory that tried to connect what psychologists
knew about behavior with what neuroscientists knew about
the brain. In particular, he proposed an idea that has come to
be known as Hebbian learning: If the activity of one neuron
repeatedly leads to the firing of another neuron, then the synaptic
connection between them will get stronger. Put simply, neurons
that fire together, wire together.

•• More than 20 years after Hebb originally proposed this idea, Tim
Bliss and Terje Lømo discovered empirical evidence that Hebb was
right. Specifically, they found that if they artificially stimulated
a neural circuit with high-frequency electrical activity, the
strength of the synaptic connections in that pathway got stronger.
Furthermore, they found that the strength, or potentiation, of
those synaptic connections lasted for days or even weeks, so they
referred to the phenomenon as long-term potentiation (LTP).

•• A few years later, Gary Lynch and his colleagues demonstrated that
it was also possible to induce the long-term depression (LTD) of
synaptic connections. In that case, the synaptic connections get
weaker rather than stronger.

Lecture 12 | The Neural Basis of Implicit Learning 117

 •• It seems likely that LTP and LTD play an important role in long-
term associative learning. But just because artificial stimulation
leads to synaptic changes, that doesn’t necessarily mean that these
mechanisms play a role in real-life learning.

•• To test that hypothesis, Richard Morris and his colleagues tested

rats’ ability to learn before and after being injected with a drug
that blocks LTP. Specifically, the rats had to perform the Morris
water maze task, which involves learning where a hidden platform
is in a pool of opaque water.

•• Normally, rats swim around until they stumble upon the platform
but then quickly learn its location and swim straight for it when
they’re put back in the pool. But after being injected with the
LTP-blocking drug, the rats were significantly impaired at learning
the platform’s location, suggesting that LTP is playing a role in
real-world learning.

•• LTP and LTD have been found in numerous parts of the brain,
but they have been most studied in the hippocampus—which is
crucial for explicit learning. These are basic neural mechanisms
that could play a role in both implicit and explicit learning.

SUGGESTED READING

Kandel, In Search of Memory.

Schultz, “Predictive Reward Signal of Dopamine Neurons.”
Squire and Kandel, Memory.

118 The Learning Brain

 QUESTIONS TO CONSIDER

1 What do you see as the major advantages and disadvantages

of studying learning in animals when trying to understand
human learning?

2 The neural mechanisms that underlie human learning are thought

to be quite similar to those that underlie animal learning. So, why
do you think humans learn so much more than animals do?

Lecture 12 | The Neural Basis of Implicit Learning 119

 LECTURE 13
Strategies for Effective
Skill Learning

H  W SHOULD YOU PRACTICE TO GET BETTER

O
AT A PARTICULAR SKILL? UNFORTUNATELY,
the approaches to training that are used
by most coaches are based largely on
intuition and trial and error rather than on
sound scientific evidence, and many of
our intuitions about effective practice may
actually be undermining our improvement.
This lecture will focus on 3 key ideas about
practice that have been scientifically
demonstrated to be effective in the
acquisition and long-term retention of skills:
space, challenge, and randomize (SCoRe). 

120 The Learning Brain

 A Misconception about Effective Practice
•• Many people believe that whatever method of practice leads to the
most immediate improvement must be the most effective method.
For example, the tennis player who repeatedly hits crosscourt
forehand after crosscourt forehand during practice may begin to
get in a groove with that shot and quickly improve his or her ability
to hit it consistently and accurately. Unfortunately, the confidence
that one feels during practice doesn’t always translate into better
performance during a match.

•• To understand why this is, it’s important to distinguish long-term

learning from short-term performance. In most cases, the goal of
practice is long-term learning—that is, improvement that lasts
and that will be apparent in any context. The tennis player on the
practice court would like to see a lasting improvement in his or her
overall game that’s apparent whether he or she plays on hardcourt
or clay and no matter what opponent he or she faces.

•• However, we often choose practice methods based not on how

much they improve our long-term learning, but rather based on
how much they improve our short-term performance—that is, our
ability to perform the skill in the context of the current practice
session. For example, it’s much easier to groove a tennis swing
when repeatedly hitting the same shot rather than intermixing
shots, so short-term performance during the practice session will
likely be better.

•• It’s only natural to assume that

if our short-term performance is Vince Lombardi famously
getting better, then our method said, “Practice doesn’t make
of practice must be paying off, perfect. Perfect practice
and our long-term learning will makes perfect.” But what
also be maximized. But it turns constitutes perfect practice?
out that our intuition is wrong.

Lecture 13 | Strategies for Effective Skill Learning 121

 In fact, training methods that are most effective for long-term
learning tend to introduce difficulties during practice that actually
make short-term performance worse.

•• Dominic Simon and Robert Bjork performed an experiment

that illustrates this point. They asked people to learn 3 different
sequences of key presses on the number pad of a keyboard. Half the
participants practiced one sequence over and over before moving on
to another sequence. The other half practiced one of the sequences
once and then practiced a different sequence once, and so on, in an
unpredictable, random order. Both groups returned the next day
and were tested again to assess their longer-term learning.

•• Predictably, during the practice session, people who got to practice

each sequence over and over in separate blocks did better than
people who practiced the sequences in a random order. The people
in the blocked condition also predicted that they would continue
to do well when tested the next day. But they were wrong. In fact,
the people who practiced the sequences in an unpredictable order
did much better the following day.

•• There are 2 important lessons we can draw from this study:

1 Long-term learning can be significantly improved by

introducing changes that actually make short-term performance
more difficult rather than easier. Bjork refers to these kinds of
changes as desirable difficulties.

2 Our intuitions about which training method will be most

effective for long-term learning are not reliable. Most people
assume that a training method that leads to rapid improvement
during training will also lead to lasting improvements. They
therefore gravitate toward such methods. Unfortunately, those
methods typically don’t actually maximize long-term learning.

122 The Learning Brain

•• If our intuitions about effective practice are wrong, then how
should we practice? The 3 most important principles of effective
practice can be summarized in the acronym SCoRe: space,
challenge, and randomize. Each of these ideas can be seen as a kind
of desirable difficulty; that is, each makes short-term performance
more difficult, but that short-term difficulty is actually desirable
because it leads to better long-term learning.

Space
•• To maximize long-term learning, practice should be spaced out,
or distributed, over time rather than crammed into a short period.

•• Carol-anne Moulton and colleagues demonstrated the advantages

of spaced practice in people training to be surgeons. Thirty-eight
surgical residents received 4 training sessions on a specific type
of microscopic surgery that required connecting very small blood
vessels together. Half the residents received all the training in
one day, which was the standard approach used in the residency
program. The other half of the residents received one training
session each week for 4 weeks. So, both groups received the same
amount of training, but one group’s training was spaced while the
other group’s training was not.

•• One month after the final training session, all of the residents were
asked to perform the surgery on a live rat while experts who did
not know about the experiment observed. The residents whose
training was spaced performed much better. They all completed
the surgery successfully, while some of the residents in the other
group did not. Furthermore, their surgically repaired blood vessels
were more stable, and the experts rated their performance as being
significantly better.

Lecture 13 | Strategies for Effective Skill Learning 123

 •• There are a few potential reasons why spacing helps. One idea is
that people are more engaged and attentive after a delay. If you’ve
just spent the past 2 hours practicing your golf putting, it might be
tough to remain focused for another hour of the same kind of work.
But if you take a break and space your practice, then you can often
return more energized and be more productive in your third hour.

•• Another reason spacing might help is that practicing a skill at

different times can make it less dependent on any specific practice
context. For example, suppose you’re hitting golf balls. In one
practice session, the wind is blowing into your face, and hitting
the ball lower leads to better results because it keeps the ball below
the wind. But then you take a break, the wind shifts, and now it’s
with you during your next practice session. Now hitting the ball
higher might lead to better results because the ball can ride the
wind. In this case, spacing your practice helped you learn to hit
shots in 2 different types of conditions, both of which you’re going
to encounter when playing real rounds of golf.

««Space your practice to maximize long-term

learning. For example, athletes benefit more
from practicing for 30 minutes every day than
from practicing for 4 hours once a week. «

Challenge
•• Research suggests that your practice will be more effective if
you deliberately challenge yourself. If you can identify specific
challenging areas to work on, set specific goals within those
areas that are difficult—but not impossible—to achieve, and

124 The Learning Brain

 systematically work to achieve those goals, then you’ll improve
faster than if you spend the same amount of time practicing
without challenging yourself.

•• One study that illustrates the benefits of deliberately challenging

yourself was conducted by Nina Keith and Anders Ericsson, who
recruited 60 university students who had several years of typing
experience. They gave them all a bunch of typing tasks, measured
how quickly they could perform simple perceptual and motor
tasks, and interviewed them to find out how they learned to type.
They reported 2 interesting results.

•• First, perceptual and motor processing speed didn’t make a

difference. You might expect the people who were really fast at
perceiving visual stimuli or who could press a button very rapidly
would have a significant advantage and would be the fastest typists,
but they weren’t. In fact, perceptual and motor processing speed
didn’t correlate with typing speed at all.

•• Second, they found that what really mattered was deliberate,

challenging training. Specifically, the fastest typists had 2 common
characteristics: They had taken a formal typing class that included
deliberate typing goals and challenged themselves to type as fast
as possible during their daily typing tasks. Neither characteristic
alone was sufficient to identify the fast typists.

««If you want to get better at any complex

skill, you need to be systematic and
deliberate in the way you practice. Rather
than practicing things that you’re already
good at, deliberately challenge yourself and
stretch your abilities when you practice. «

Lecture 13 | Strategies for Effective Skill Learning 125

 Randomize
•• Most complex skills involve multiple parts, each of which must
be mastered. For example, tennis requires being able to hit
forehands, backhands, volleys, and overheads. Random practice
refers to mixing the practice of different parts of a skill, rather than
practicing each part separately.

•• In the experiment on learning sequences of key presses on a number

pad, the group that practiced the sequences in an unpredictable order
was using a random practice schedule: After practicing one sequence,
they would practice a different one, in a random order. The other
group’s practice was blocked, rather than being random; that is, they
practiced a single sequence over and over before switching to another
sequence and practicing that one over and over.

•• Although the short-term performance of the group that used

random practice was worse than the other group, the next day
the random practice group performed much better than the other
group. So, if your goal is long-term learning rather than short-term
performance, random practice is better than blocked practice.

•• Many people find this result counterintuitive. After all, isn’t

it better to break down a complex skill into separate parts and
practice each part separately without interference from all the
other parts? Although that is a very common way to practice and
although it does improve short-term performance, random practice
is better for long-term learning.

•• One potential explanation for why is that when you actually have
to perform the real skill later, you’ll typically have to perform all
the subparts in an unpredictable order. Using random practice
therefore better prepares you for the real execution of the skill. For

126 The Learning Brain

 example, when playing in a real tennis match, you’ll be forced to
intermix forehands and backhands in an unpredictable order. And
it turns out that it’s better to practice that way, too. As athletes
often say, you should practice the way you play.

•• In addition to randomizing the order of practice, it’s also helpful to

randomize the conditions of your practice. If you always practice
under the same, predictable conditions, then your learning can
become strongly tied to those conditions. For example, if you
always hit golf balls at the same driving range, then you may
limit the generality of your learning. When it comes time to use
those skills in a real competition, you could find that you haven’t
mastered them as well as you thought you had.

•• Robert Kerr and Bernard Booth conducted a study that illustrates

the benefits of randomizing practice conditions. Two groups of
children practiced tossing beanbags at a target. Half the children
always tossed to a target that was 3 feet away, while the other half
of the children tossed to targets that varied in distance but were
never 3 feet away. After training, both groups of children tossed
beanbags at a 3-foot target.

•• Incredibly, the children who practiced at varying distances actually

performed better than the children who had trained exclusively at
the 3-foot distance. Clearly, randomizing your practice can be a
very powerful approach.

««As you try to learn new skills, mix up your

practice rather than working on an isolated part
of the skill for an extended period of time. «

Lecture 13 | Strategies for Effective Skill Learning 127

 Qualifications on SCoRe Recommendations
•• It’s important to recognize that not all difficulties are desirable.
In particular, the level of difficulty needs to match the ability of
the learner. There is nothing wrong with novice tennis players
practicing forehand after forehand while they try to learn the
basics of the grip and swing. Similarly, although it’s fine for an
intermediate golfer to practice hitting a large green from 100
yards away, that’s probably not challenging enough to stretch a
professional golfer. The point is that people at every skill level
need to introduce difficulties during practice that are challenging
but also manageable.

•• Spacing and randomizing can also be taken too far. For example,
spacing becomes ineffective if the delay between practice sessions
is so long that the learners have forgotten what they previously
learned. And although randomizing practice conditions is helpful,
there is no need to include conditions that you would never actually
encounter in real life. For example, although it’s helpful for a
basketball player to practice shooting from a variety of different
distances to the hoop, there is no need for the player to practice
shots from 200 feet away.

•• It’s important to realize that the goal is not always long-term

learning. Sometimes, the goal is short-term performance. For
example, the musician who is preparing for a recital tomorrow
is more worried about short-term performance than long-term
learning. And in this kind of case, spacing and random practice
may not be the ideal way to prepare.

128 The Learning Brain

 SUGGESTED READING

Ericsson and Pool, Peak.

Schmidt and Lee, Motor Control and Learning.
Schmidt and Wrisberg, Motor Learning and Performance.

QUESTIONS TO CONSIDER

1 What new skill would you like to acquire? How might you apply
what you learned in this lecture to help you acquire that new
skill faster?

2 Have you identified any effective practice strategies of your own

that were not discussed in this lecture?

Lecture 13 | Strategies for Effective Skill Learning 129

 LECTURE 14
Learning Bad Habits:
Addiction

A  QUIRING NEW KNOWLEDGE AND

C
DEVELOPING NEW SKILLS CAN BE BOTH
helpful and fulfilling, but in some cases—
specifically, addiction—our sophisticated
learning mechanisms can actually work
against us and lead us to learn habits that
harm us more than they help us. Addictive
drugs can hijack learning mechanisms so
that instead of learning valuable skills and
habits, such as speaking a language or
playing an instrument, the addict learns
extremely strong associations with drug
taking that turn that harmful behavior
into a habit that can be extremely difficult
to break. 

130 The Learning Brain

 Key Features of Addiction
•• Addiction refers to continuing to compulsively engage in a
behavior despite the fact that doing so has significant negative
consequences. Often the behavior is using an addictive substance,
such as nicotine, alcohol, or painkillers, and the consequences
could range from significant health problems, to broken
relationships, to losing a job.

•• Some people manage to use drugs without significant negative

consequences in their life. Think of the social drinker who never
becomes an alcoholic. That would be considered drug use, but
not drug abuse. And without abuse, it wouldn’t be considered a
true addiction.

Most people have either struggled with an

addiction or know someone who has. Maybe you
know a long-time smoker who has repeatedly tried
to quit but just hasn’t been able to kick the habit.
If you’ve never been addicted yourself, it can be
difficult to understand. Don’t these people see
how harmful their addiction is, not only to them,
but also to those around them? Don’t they have
any willpower? Why don’t they just quit?
The challenge of overcoming an addiction involves
both psychological and neural mechanisms.

131
 •• Although most addictions involve taking drugs, many scientists
believe that some other behaviors should also be considered
addictions. An example is pathological gambling, which involves
compulsive behavior and can have significant negative consequences.

•• Many scientists believe that people can also become addicted to

playing video games or surfing the internet. But for those behaviors
to be classified as addictions, the person would have to pursue
them obsessively despite truly significant negative consequences.

•• But obsessive behavior and negative consequences aren’t the only

symptoms of addictions. Addictions are also characterized by
dependence and by craving.

•• First, addicts depend on their drug or behavior to feel normal, both

psychologically and even physically. For example, as heroin addicts
continue to regularly take heroin, their body comes to expect the
drug and begins to make physical changes to compensate for the
drug and to neutralize its effects. As a result of these physical
changes, heroin users need to take more of the drug to have the
same effect that a smaller dose used to have. That’s called tolerance,
and it’s a clear sign of physical dependence.

•• Another clear sign of physical dependence is withdrawal. If

addicts stop taking their drug for a while, they will typically
start to feel unpleasant side effects that are the mirror opposite of
the positive feelings produced by the drug. The addict’s body is
expecting the drug to be present and has therefore compensated
in an effort to neutralize the drug’s effects, and those very
compensations can lead to symptoms of withdrawal if the drug
is withheld.

132 The Learning Brain

•• For example, alcohol is a sedative that tends to suppress neural
activity. In alcoholics, the body compensates for these effects by
exciting neural activity more than normal. This extra excitatory
activity means that when alcoholics drink alcohol, their neural
activity doesn’t look as suppressed as it would in nonalcoholics who
are drinking. That’s another aspect of alcoholics’ tolerance to alcohol.

•• But when alcoholics stop drinking, their body is still generating

more than normal neural activity, but now there’s no alcohol to
suppress it. Alcoholics therefore starts to feel withdrawal symptoms
such as anxiety and shakiness that are caused by too much neural
excitation. In severe cases, victims might even experience dangerous
seizures and delirium that require medical attention.

•• The same is true for other drugs. Opiate drugs, such as heroin
and prescription painkillers, tend to produce a sense of euphoria
and also cause constipation. So, when they’re withheld, the addict
might experience withdrawal symptoms such as agitation and
diarrhea—the mirror opposite of the effects produced by the drug.

•• Another key characteristic of addiction is craving. Drug addicts

experience extremely strong urges to obtain and use their drug of
choice. Of course, everyone experiences cravings now and then,
such as for a candy bar when checking out at the grocery store. You
might even give in to those kinds of cravings sometimes.

•• But the cravings of addicts are different. Addictive cravings are

so strong that they completely consume the addicts, making it
difficult to think about anything else. These are the kind of drives
that lead addicts to lie, beg, steal, or do almost anything else to
get their drug. They’re so obsessed with the need for the drug that
they don’t consider the long-term consequences in terms of school,
work, or family. Getting the drug is all that matters.

Lecture 14 | Learning Bad Habits: Addiction 133

 How Addiction Reflects Pathological Learning
•• One of the key features of associative implicit learning, such as
classical and operant conditioning, is that over time, unconditioned
stimuli can become strongly associated with a reward or behavior.
Pavlov’s dogs learned to associate getting fed with hearing the
sound of experimenters entering the room.

•• Another key feature is that learning in these situations depends on

making a prediction error. If you’ve already learned an association
between a stimulus and a reward and the reward arrives after the
stimulus just as you predicted, then there’s no prediction error.
You were already expecting the reward, so you don’t need to learn
something new.

•• However, if you get the reward when you weren’t expecting it,
then there is a reward prediction error—you failed to predict the
reward when it showed up. And that means you should try to learn
so that you can better predict rewards in the future. That’s when
you learn new associations between whatever stimuli are in the
environment and the reward you just received.

•• The activity of neurons in the ventral tegmental area (VTA) tends

to track this kind of reward prediction error. These neurons fire
when an unexpected reward arrives, but not when a reward shows
up that was predictable. Furthermore, these neurons start to fire
in response to nonrewarding stimuli that have been associated
with a future reward.

•• The VTA neurons release the neurotransmitter dopamine when

they fire. This signals that an unexpected reward has arrived or is
on the way; that is, it signals reward prediction error. It also signals
the need for new learning.

134 The Learning Brain

•• When the VTA neurons fire and release dopamine, the organism
strengthens associations between the reward and whatever
stimuli are present that predict that reward. And the next time it
encounters those same stimuli, it will tend to predict that a reward
is on the way.

•• In addition to signaling that an unexpected reward is on the way,

dopamine also seems to be important in motivating you to pursue
the reward. It makes you want it.

•• Some of the best studies on dopamine’s role in motivation use

genetic engineering, which involves manipulating the DNA of
organisms, especially mice, so that those organisms have specific
genetic properties.

•• For example, scientists might knock out a specific gene or change

the form of a gene. Scientists have used this approach to create
mice that tend to produce abnormally high levels of dopamine as
well as mice that produce abnormally low levels of dopamine. The
findings from these studies tell us a lot about the role of dopamine
in motivation.

•• For example, mice that are genetically engineered to produce high

levels of dopamine seem to have abnormally strong motivation to
get food. Conversely, mice that have been genetically engineered
to have abnormally low levels of dopamine show a complete lack
of motivation. They won’t even walk across their cage to get food
when they’re hungry. In fact, they actually have to be hand-fed
or they’ll starve.

•• How does all of this relate to addiction? All addictive drugs lead
to an artificially large burst of dopamine. And that large burst of
dopamine is what makes the drugs so addictive.

Lecture 14 | Learning Bad Habits: Addiction 135

 Dopamine plays a central role in motivation, in addition to its
role in signaling reward prediction error and the need for new
learning. And that combination of functions makes a lot of sense.
Imagine an animal that has just smelled a new source of food
that it didn’t previously know about. It would be helpful to that
animal’s survival if it were motivated to find that food and to
learn its location.
And that’s exactly what the dopamine system does. The smell of
food signals a reward prediction error. The animal had not been
predicting a reward, but once it smells the food, it realizes that its
prediction had been wrong: Food is actually likely on its way.
That reward prediction error is accompanied by
the release of dopamine, which triggers
new learning, which will help
the animal remember this
food source in the future;
and wanting, which
motivates the animal
to look for and find
the food.

•• For example, cocaine is a dopamine reuptake inhibitor, which

means that it prevents the brain from removing dopamine after
it’s been released by dopamine neurons. And that means that
whenever dopamine is released, its effect will last longer and
will therefore be magnified.

136 The Learning Brain

•• Other addictive drugs also lead to increased dopamine levels, but
in different ways.

•• Methamphetamine produces even higher levels of dopamine

than cocaine does, and it’s also even more addictive.

•• Nicotine binds to special receptors on the VTA dopamine

neurons and makes them fire more than normal, leading to
increased dopamine release.

•• Alcohol and opioids turn off neurons that normally inhibit

the VTA dopamine neurons, which means that the dopamine
neurons fire more than normal.

•• To understand why that large burst of dopamine makes drugs so

addictive, consider what happens when a heroin addict shoots up or
a chain smoker takes a puff: The drug induces an artificially large
burst of dopamine, and the brain interprets that dopamine burst
as meaning that an unexpected reward has arrived or soon will.

•• Notice that the reward for drug addicts actually isn’t unexpected.
If they are chronic users, they know very well what to expect from
taking their drug of choice. In fact, they’ve probably developed a
tolerance to their drug so that the rewarding effect isn’t as strong
as it used to be.

•• So, the dopamine burst in this case isn’t caused by an unexpected

or better-than-normal reward; it’s caused by the drug messing with
the brain’s machinery and artificially creating a dopamine burst.

•• But the brain doesn’t know that—so it behaves the way it always
does when it gets a dopamine burst. It interprets it as meaning
that an unexpected reward is on its way and does 2 things: First,
it triggers wanting or motivation. (This is not regular wanting, but
the pathological craving that addicts report feeling.)

Lecture 14 | Learning Bad Habits: Addiction 137

 •• The brain’s second response is to learn: Dopamine normally
signals a reward prediction error and the need for more learning.
Specifically, the brain learns associations between whatever
environmental cues happen to predict the reward and the behavior
that led to it. In this case, the behavior being reinforced is taking
the drug.

•• The environmental cues that predict the reward are anything in

the environment that tends to predict drug taking, such as the
sight of a pack of cigarettes for a chain smoker, the street corner
where the user buys heroin, or the people with whom the user
typically does drugs. Anything and everything that is reliably
associated with drug taking constitutes an environmental cue that
the brain will associate with that behavior.

•• Worse yet, the dopamine neurons will start to fire in response to

the cues that predict the reward, rather than the reward itself, so the
addicts’ dopamine neurons start firing whenever they encounter an
environmental cue that they’ve associated with drug taking. For
example, the chain smoker starts experiencing a dopamine burst
whenever he or she sees a cigarette. Anything in the environment
that the addict associates with drug taking eventually triggers a
dopamine burst itself.

•• For chronic drug users, almost everything in their daily

environment is associated with drug taking—and therefore trigger
craving and an overwhelming urge to seek out and use the drug.
So, it’s virtually impossible for the addict to avoid encountering
these cues and experiencing the extremely powerful cravings that
end up controlling their behavior. And every time the addict gives
in to their cravings and takes their drug, another dopamine burst
is released, and the associations are strengthened even more. It’s a
vicious cycle—one that can be extremely difficult to break.

138 The Learning Brain

 SUGGESTED READING

Erickson, The Science of Addiction.

Kuhar, The Addicted Brain.
Kuhn, Swartzwelder, Wilson, Wilson, and Foster, Buzzed.
Lewis, Memoirs of an Addicted Brain.

QUESTIONS TO CONSIDER

1 What do you see as the key similarities and differences between

addiction to a drug and addiction to a behavior, such as gambling?

2 Think of someone you know who is addicted to alcohol, smoking,

or some other drug or behavior. Does what you learned in this
lecture change your view of that person and his or her addiction in
any way? If so, how?

Lecture 14 | Learning Bad Habits: Addiction 139

 LECTURE 15
Introduction to
Working Memory

S  ARTING WITH THIS LECTURE, THE COURSE

T
WILL TURN TO YET ANOTHER ASPECT OF
learning and memory: working memory. In
this lecture, you will discover what working
memory is, the central role that it plays in
virtually all cognitive tasks, and the fact that
it depends on different brain systems than
long-term memory does. You will also be
introduced to a simple model of working
memory. In this lecture, you will learn about
the phonological loop component of this
model, while the next lecture will address the
other components of working memory as
well as some of the neural mechanisms that
have been discovered. 

140 The Learning Brain

 Working Memory and Cognition
•• Working memory is the cognitive system that allows you to
temporarily store information for short periods of time, maintain
that information over a delay, and process that information in the
service of whatever you’re currently trying to do.

•• Working memory is a kind of temporary, or short-term, memory.

In fact, psychologists often use the terms “short-term memory” and
“working memory” interchangeably to refer to the same cognitive
process. But working memory has become the preferred term
because it conveys the idea that the system is actively engaged in
working with information. It’s not just a passive storage system;
it’s actively used for processing information.

•• Another important characteristic is that working memory has a

much more limited capacity than long-term memory, which holds
hundreds of thousands of words, countless facts about the world,
and thousands upon thousands of personal episodes from your life.

•• The capacity of your long-term memory is virtually unlimited.

Our brains are finite, so there must be some upper limit on how
much information we can store in long-term memory, but nobody
seems to have reached it in our limited lifespans.

•• In contrast, the capacity of working memory is severely limited.

For example, suppose that you were asked to close your eyes and
add 625,488,561 and 37,290,417 in your head. It’s pretty unlikely
you’d be able to remember the numbers, much less add them
together. And that’s not because you can’t add big numbers; if you
were able to do the addition on paper, you’d probably find it pretty
easy. The problem is that the capacity of your working memory is
limited; you can only store so much information at a time.

Lecture 15 | Introduction to Working Memory 141

 •• Working memory plays a crucial role in virtually any cognitive
task you can think of. When you’re holding a conversation with
someone, you need to temporarily store what he or she is saying,
process that information, and retrieve it later in the conversation.
When you’re cooking dinner, doing your taxes, playing an
instrument, reading, writing,
or driving, you’re using your
working memory. Working memory is central
to thought. Any time you’re
•• People with greater working thinking, you’re using your
memory capacity tend to working memory.
perform better on a wide range
of cognitive tasks. They perform
better on tests of reading comprehension and reasoning, and they get
better grades and higher standardized test scores. In fact, working
memory capacity is one of the best predictors of intelligence, at
least as measured by IQ tests. This is further evidence that working
memory is important to cognition in general.

•• Working memory is important, but so are long-term explicit and

implicit memory. What’s the relationship between working and
long-term memory?

•• Human beings use many different memory systems. For example,

explicit memory is fundamentally different than implicit memory
and depends on different neural substrates. And even within
explicit memory, we distinguish between semantic memory for
facts and episodic memory for personal events that are tied to a
specific time and place.

•• Working memory is itself a distinct memory system. In fact,

some of the evidence for that claim comes from the amnesiac
patient Henry Molaison, whose explicit long-term memory was
profoundly impaired but whose working memory was normal.

142 The Learning Brain

•• This is an example of a single dissociation, in which damage to one
part of the brain causes a deficit in one process, but not another. In
this case, damage to the hippocampus and other medial temporal
lobe structures leads to impairments in long-term memory, but
not working memory.

•• You might think that evidence like that proves that long-term
memory and working memory depend on different underlying
systems. But there’s a potential alternative explanation. Suppose
that the long-term memory tasks are just more difficult than the
working memory tasks. Then, you might perform worse on the
long-term memory tasks even if they depend on the same neural
systems as the working memory tasks.

•• We can rule out this alternative explanation by finding another

brain-damaged patient who exhibits the opposite dissociation—a
patient whose long-term memory is normal, but whose working
memory is dramatically impaired. One such patient is referred to
by his initials, KF, to protect his anonymity.

•• KF had a pretty severe motorcycle accident that led to brain

damage. But the damage was not in the medial temporal lobes,
and his long-term memory was fine. But his working memory
was dramatically impaired. In particular, whereas most people
can remember 6 or 7 digits after a brief delay, KF could only
remember about 2.

•• KF’s problems were the opposite of Henry Molaison’s. This is a

double dissociation: 2 patients who exhibit single dissociations
but in opposite directions. One patient is impaired on task A
but not task B, while the other patient is impaired on task B but
not task A.

Lecture 15 | Introduction to Working Memory 143

 •• The great thing about double dissociations is that you can’t explain
them away based on a difference in task difficulty. For example,
suppose that working memory and long-term memory depend on
the same neural system and the only reason that Henry Molaison
is more impaired on long-term than working memory tasks is
because the long-term memory tasks are more difficult. In that
case, you should never see a patient like KF who exhibits the
opposite dissociation and is selectively impaired on the tasks that
are assumed to be easier. The only way to explain both dissociations
is if working memory and long-term memory depend on different
underlying systems.

The Phonological Loop

•• Alan Baddeley and Graham Hitch proposed a theory of working
memory—called the multicomponent theory—that has become
one of the most influential theories in all of cognitive psychology,
probably because it is relatively simple and yet manages to explain
a wide variety of empirical phenomena.

•• One of the key assumptions of this theory is that working memory

consists of multiple, separable components. So, just as working
memory should be distinguished from long-term memory and
explicit memory should be distinguished from implicit memory,
Baddeley and Hitch argued that working memory should be
subdivided into separate functional components.

•• Originally, they proposed 3 components: a phonological loop,

a visuospatial sketchpad, and a central executive. But in 2000,
Baddeley argued for the addition of an episodic buffer component.

144 The Learning Brain

•• The phonological loop is the component of working memory that
is responsible for the temporary storage and rehearsal of linguistic
information. For example, if you are asked to remember a few
random words, such as “ball,” “truck,” “mirror,” and “star,”
and then a few seconds later are asked to say those words, you
would be using your phonological loop to hang on to that
information temporarily.

•• You could imagine different ways of storing those words. You could
store images of what the words look like based on the shapes of the
component letters. You could also imagine storing representations
of the words’ meanings. But a key assumption of the phonological
loop is that you store the words in a sound-based format.

•• Phonemes are the atomic units of distinctive speech sounds in a

language, such as the “cuh,” “ah,” and “tuh” sounds in the word
“cat.” The assumption is that when we’re storing language in
working memory, we’re storing phonemes, or language sounds.
That’s why it’s called the phonological loop.

•• This phenomenon is sometimes called the phonological

similarity effect. When we have to temporarily store words that
are phonologically similar—in other words, that sound alike—
we typically do worse than if we’re trying to store words that
sound different.

•• But what’s really interesting is that we only have trouble with items
that sound alike, not with items that look alike or items that have
similar meanings.

•• For example, consider the word “bough,” as in the bough of a

tree. The spelling of the word is very similar to “cough,” “dough,”
and “through,” which all end in “-ough” and only differ in the
first letter or 2. Those words look very similar, but they sound
completely different.

Lecture 15 | Introduction to Working Memory 145

 •• People don’t have any trouble storing words like this in working
memory. So, visual similarity doesn’t cause the same problems that
phonological similarity causes.

•• Likewise, semantic similarity also doesn’t cause problems; that

is, people have no problem storing words that have similar
meanings in working memory. So, if you were given the list
“huge,” “long,” “tall,” “big,” and “wide,” you probably wouldn’t
have any problem holding on to those words. And the fact that
the words are semantically related might actually make it easier
to remember them.

•• So, phonological similarity impairs working memory, but visual

and semantic similarity don’t. This suggests that when we store
verbal information in working memory, we’re storing the sound
of the information, not what it looks like and not what it means.
And that’s exactly what the phonological loop model assumes.

•• The other key assumption of the phonological loop model

corresponds to the “loop” in the model. The model assumes
that people have access to an articulatory loop that they can use
to rehearse the sound-based representations stored in working
memory. Whatever language sounds you can say in about 2
seconds, you can rehearse and refresh using this loop.

•• The reason it’s called an articulatory loop is because the assumption

is that you articulate the information in working memory. In other
words, you say it to yourself—not necessarily out loud; you might
do it subvocally. But the assumption is that you are using some
of the same mechanisms that you use when you speak. You’re just
speaking to yourself. And doing that reactivates and refreshes the
sound-based representations that are in working memory so that
you can hang onto them a little bit longer.

146 The Learning Brain

•• The model also assumes that if you’re trying to store language-
based information in working memory but that information isn’t
already in a sound-based format, then you’ll use the articulatory
loop to say it to yourself and convert it into a sound-based format.

•• For example, suppose that you are asked to remember a few words
in your working memory, but instead of saying them out loud,
you are only allowed to read them on a piece of paper or on a
computer screen. The words are presented in a visual format, not
a phonological format.

•• The model assumes that you will use your articulatory loop to say
the words to yourself and convert them into a phonological format.
And that’s the format that they’re stored in. So, even though the
words were presented visually, they are stored phonologically.

•• These assumptions about the articulatory loop make a number of

testable predictions. And the phonological loop model correctly
predicts that similar-sounding items will be more difficult to
remember and that making people talk out loud will impair their
memory. It also correctly predicts that phonological similarity
effects will disappear when people talk out loud, but only if items
are presented visually.

•• Furthermore, based on the assumption that the articulatory

rehearsal loop can only hold about 2 seconds of speech, this model
predicts that people should be able to remember more short words
than long words. The reason is that you can say more short words
in a 2-second window than you can long words. And that means
you can rehearse more short words than long words, and every
rehearsal refreshes the item in working memory.

Lecture 15 | Introduction to Working Memory 147

 •• This finding even extends
to how fast people speak. If People in the northeastern
you measure people’s speed of US tend to speak faster
speech—that is, how many than people with a southern
words they typically produce drawl. Northeasterners
per second—you’ll find that it’s also tend to have a larger
a very good predictor of their working memory capacity.
working memory capacity.

•• It also suggests that working memory capacity might not be

innate, but can be influenced, and potentially even improved, by
experience. In fact, an entire future lecture will be dedicated to
this fascinating topic.

SUGGESTED READING

Baddeley, Working Memory, Thought, and Action.

Baddeley, Eysenck, and Anderson, Memory.
Gathercole and Alloway, Working Memory and Learning.

QUESTIONS TO CONSIDER

1 Some psychologists view working memory as the currently

activated part of long-term memory. What do you see as the
strengths and weaknesses of this theory?

2 Psychologists used to think that information had to be stored in

working memory before it could be stored in long-term memory.
But Patient KF had relatively normal long-term memory despite
very impaired working memory. Do you think KF’s profile
demonstrates that psychologists were wrong?

148 The Learning Brain

 LECTURE 16
Components of
Working Memory

I  THE PREVIOUS LECTURE, YOU WERE

N
INTRODUCED TO ONE OF THE MOST FAMOUS
theories of cognitive psychology—namely,
Alan Baddeley’s multicomponent theory of
working memory, which assumes that working
memory can be broken down into component
parts. One part is the phonological loop,
which you learned about in the previous
lecture. This simple model can explain many
phenomena related to working memory for
language-based information, including effects
of phonological similarity, word length, and
speed of speech. In this lecture, you will learn
about the other components of Baddeley’s
model: the visuospatial sketchpad, the central
executive, and the episodic buffer. 

Lecture 16 | Components of Working Memory 149

 The Visuospatial Sketchpad
•• The visuospatial sketchpad is a separate component of your working
memory system that’s devoted to temporarily storing visuospatial
information. So, just as the phonological loop is devoted to storing
language sounds, the visuospatial sketchpad is devoted to storing
visual and spatial information.

•• For example, look around wherever you are right now and store
a visual image of what your environment looks like. Now close
your eyes and bring that visuospatial image to mind. That’s
your visuospatial sketchpad at work, temporarily storing that
visuospatial information while your eyes are closed.

•• Information can get into your sketchpad in 2 different ways. One

way is by perception. When you looked around and formed an
image of your local environment, you were putting that information
into your visuospatial sketchpad using visual perception.

•• The other way is to retrieve information from long-term memory.

For example, if you close your eyes and bring to mind the image
of your best friend’s face or of the house you grew up in, then
you’re using your visuospatial sketchpad again. But this time, the
information in the sketchpad wasn’t directly perceived from the
environment; rather, it was retrieved from your long-term memory.

•• Why did Baddeley assume that working memory is divided up

into these separate components? Maybe we have a single, unitary
working memory system that we use to temporarily store any kind
of information, whether it’s phonological or visual or spatial. So
how do we know that the visuospatial sketchpad is different from
the phonological loop?

150 The Learning Brain

•• Some of best evidence came from an experiment conducted by Lee
Brooks, who asked people to perform 2 working memory tasks:
one that would tap visuospatial working memory and one that
would tap phonological working memory. And he asked them to
perform each task in 2 different ways. In one version of the tasks,
he asked people to respond verbally—that is, by speaking their
answers. In the other version, he asked people to respond spatially,
by pointing to their answers on a piece of paper.

•• His idea was that if we really do have separate working memory

systems for storing sounds versus sights, then using a vocal, sound-
based response might interfere with sound-based working memory,
whereas using a visuospatial, pointing response might interfere
with visuospatial working memory.

•• Essentially, if making the response requires using some of the

mechanisms that are being used during the working memory task,
then that might cause problems, because you’re trying to use those
mechanisms to do 2 different things at the same time.

•• In both tasks, you have to remember something in your working

memory and then walk through that representation, making
judgments about different parts of it. The difference is whether
you have to remember a visuospatial representation of what a letter
looks like or remember a phonological representation of what a
sentence sounds like.

•• Brooks asked his participants to perform both of these tasks in 2

different ways: You either respond vocally, by saying “yes” or “no”
out loud; or spatially, by pointing to the correct answers on a sheet
of paper with a bunch of “yes”/”no” pairs written on it.

•• Brooks found that the 2 response types had opposite effects on the
2 tasks. First, in the visuospatial working memory task, he found
that people are faster if they get to say “yes” or “no” out loud and

Lecture 16 | Components of Working Memory 151

 slower if they have to point to the “yes” or “no” on the paper. In this
case, there is an interference from having to point. Pointing—that
is, making a visuospatial response—interferes with visuospatial
working memory.

•• You might think that talking is just easier than pointing and that
people would always be faster when they get to say their responses.
But Brooks found the opposite pattern for the sentence task: People
were faster when they pointed to the response on a piece of paper
and slower when they spoke their responses out loud.

•• For the visuospatial working memory task, pointing interferes, but

for the phonological working memory task, speaking interferes.
That is a double dissociation, which means that the 2 tasks must
depend on different underlying systems. After all, if they depended
on the same system, then they should be affected in the same way,
not opposite ways.

•• And those 2 systems are visuospatial working memory and

phonological working memory. Brook’s experiment provides
evidence that we really should distinguish these 2 types of working
memory, just as Baddeley’s multicomponent theory assumes.

The Central Executive

•• Think about the central executive officer (CEO) in a company.
This person delegates responsibilities to other people, monitors
how things are running, and makes sure that everything’s getting
taken care of. And that’s exactly what Baddeley’s central executive
is assumed to do in our mind—delegate responsibilities to specific
cognitive processes, monitor how our mental processing is going
and whether we’re making progress on our goals, and make sure
we’re doing what we need to be doing.

152 The Learning Brain

•• Imagine reading a technical book or manual to learn some useful
information, such as how to program your universal remote
control. Doing so requires all kinds of cognitive processing. You
have to use vision to recognize the letters and words on the page,
semantic memory to retrieve the meaning of the words you read,
and language skills to put the words together and figure out the
ideas that are being conveyed.

•• Beyond those basic cognitive processes, you also have to control

those processes and decide what to do when. That’s the function of
executive control. It’s the system that tells vision to read a particular
word and that tells language to comprehend a particular sentence.
And if you get distracted and your mind starts to wander, executive
control is what notices and gets you back on track. It’s overseeing
and managing the other cognitive operations.

•• In the case of working memory, it might tell the phonological loop

to rehearse some words, or it might tell the visuospatial sketchpad
to remember the spatial sequence of buttons that you have to push
to program your universal remote control.

•• And there is evidence that our brain includes a system devoted

to this kind of central executive function, as Baddeley’s theory
assumes. Recall that our cerebral cortex is divided into 4 lobes: the
temporal lobe, the occipital lobe, the parietal lobe, and the frontal
lobe. The posterior part of the frontal cortex handles motor control,
but in front of that is a region called the prefrontal cortex—and
damage to this part of the brain often leads to impairments in
executive control.

•• In particular, neurological patients with damage to the prefrontal

cortex act as if their CEO isn’t working right; that is, they have a
difficult time controlling their behavior, focusing on the things
they want to focus on, and avoiding distraction.

Lecture 16 | Components of Working Memory 153

 It’s perfectly normal to
occasionally get distracted
by things going on around
you, such as text messages
or email, but patients with
prefrontal cortex damage
are so distractible that they
can’t complete relatively
simple tasks, such as cooking
a meal or navigating to a new
location. They have pathological
problems with organization and
planning. They can’t coordinate,
prioritize, or decide what to do first
and what to delay until later.

•• These patients also commonly exhibit perseveration, which is a

pathological form of persevering—sticking to something and
continuing to pursue it—in which patients continue with a
behavior even after it’s no longer appropriate.

•• One famous task that’s used to test for perseveration is the

Wisconsin card-sorting test, in which you are shown a sequence
of cards and asked to sort them. For example, if you’re sorting by
suit, then you need to put the spades in one pile and the hearts in
another pile. If you’re sorting by number, then you need to put the
7s in one pile and the 4s in another pile.

154 The Learning Brain

•• You are not told what you should be sorting by. But the test
administrator tells you whether you’re right or wrong as you sort
each card. Initially, you have to try a few things to see whether
you’re sorting correctly. But pretty quickly, you’ll get the hang of
it and have no problem sorting the cards.

•• But then, the administrator will unexpectedly change the sorting

rule without telling you. So, if you were sorting by suit, the
administrator will change it so that you have to sort by number
instead. Naturally, you will make a few mistakes, but pretty
quickly you’ll realize that the rule has changed, so you’ll try sorting
by number and get back on track.

•• But patients with damage to the prefrontal

cortex will keep sorting by suit; that is, they’ll
perseverate in their previous behavior even
though it’s no longer appropriate. This is
a problem with executive control. The
patients have a difficult time inhibiting
a behavior that was working well and
trying something else instead.

•• The prefrontal cortex looks

a lot like Baddeley’s central
executive component. Like
the CEO of a company, our
prefrontal cortex oversees and
controls our behavior, deciding
what to pay attention to and what
to ignore, what to do and what
to avoid doing, and when to store
information into working memory
and when to retrieve it.

Lecture 16 | Components of Working Memory 155

 The Episodic Buffer
•• Originally, the phonological loop, the visuospatial sketchpad, and
the central executive were the only 3 components in Baddeley’s
model of working memory. And although these components were
able to explain a variety of empirical phenomena, Baddeley began
noticing some weaknesses in the model’s explanatory power that
ultimately led him to add a fourth component, which he called
the episodic buffer.

•• Like long-term episodic memory, the episodic buffer is assumed

to store personal episodes in memory. These episodic memories
integrate multiple different types of information. They can
contain visual information, spatial information, and phonological
information from perception. Critically, these memories can also
incorporate information from long-term memory.

•• The episodic buffer is where all these different types of information

get put together into a unified, integrated representation.
Psychologists sometimes refer to this function as binding.

•• But there are also critical differences between the episodic buffer
and long-term episodic memory. First, the episodic buffer is short-
term, so information in the buffer must be rehearsed or it will
quickly fade away. In addition, the capacity of the episodic buffer
is assumed to be limited, whereas long-term episodic memory is
assumed to have almost unlimited capacity. In particular, the
episodic buffer is assumed to have a capacity of about 4 chunks
(or meaningful units) of information.

•• For example, if you are asked to remember a random sequence of

letters, such as K B J O D T, the letters themselves are meaningful
units, or chunks, and you can probably remember about 4 at a time
without doing any rehearsal.

156 The Learning Brain

•• But suppose you are asked to remember the letters FBI, CIA,
and MIT. Now the letters are grouped into 3 meaningful units,
or chunks. So, even though you were actually given a total of 9
letters, you can probably remember all of them because you can
easily recall the 3 chunks. And from those chunks, you can extract
all the component letters.

•• This is also the explanation for our superior memory for a

meaningful sentence compared with a random sequence of
unrelated words. In the case of the sentence, the component
words can be grouped together into meaningful chunks, and those
chunks can then be stored in the episodic buffer for later retrieval.
The unrelated words can’t be grouped in that way, so they aren’t
remembered as well.

•• And that finding has practical implications for us when we’re

trying to remember information in our working memory. If
we can find a way to group the information into meaningful
chunks, then we’ll be more likely to remember it than if we treat
every piece of information as independent and unrelated to all
the others.

SUGGESTED READING

Baddeley, Working Memory, Thought, and Action.

Egner, The Wiley Handbook of Cognitive Control.

Lecture 16 | Components of Working Memory 157

 QUESTIONS TO CONSIDER

1 Which of the components of Baddeley’s working memory model do

you think is most important in real life?

2 Do you think it’s adaptive to have separate working memory

components devoted to storing phonological versus visuospatial
information? Why not just have a single, unified working
memory system?

158 The Learning Brain

 LECTURE 17
The Neural Basis of
Working Memory

Y  UR BRAIN CAN STORE INFORMATION INTO

O
WORKING MEMORY WHEN NECESSARY,
rehearse and refresh that information to
keep it active, and retrieve the information
when it needs it to make a decision or
choose an action. How does it do this?
Is there a specific part of the brain that’s
devoted to working memory? Is the
distinction between phonological working
memory and visuospatial working memory
reflected in brain organization? Are there
specific neurons that fire when you’re
holding information in working memory and
that don’t fire when you’re not? These are
the kinds of questions that will be addressed
in this lecture. 

Lecture 17 | The Neural Basis of Working Memory 159

 Studying Neural Mechanisms
of Working Memory
•• In the early 1970s, Joaquin Fuster and Garrett Alexander made
one of the first and most important discoveries about the neural
basis of working memory. They were recording from neurons in
a monkey’s brain while the animal performed a delayed response
task—specifically, the monkey watched the experimenters put a
piece of apple into one of 2 holes in front of the animal’s cage.

•• The holes were then covered, and a blind was lowered between the
animal and the holes. Then, there was a delay of anywhere from 15
seconds to a little more than a minute. After the delay, the blind
was removed, which allowed the animal to reach out through its
cage into one of the covered holes.

•• If it reached into the hole with the apple, then it got to eat the apple.
If it reached into the other hole, then the experimenters showed
the monkey where the apple really was but didn’t let the animal
have it. Monkeys are smart, and they quickly figured out the task
and began consistently reaching into the hole containing the apple.

•• Even though this task is simple, it very clearly requires working

memory. First, the holes are covered, so the monkey can’t see the
apple. In addition, the experimenters moved the apple around
randomly, so guessing wouldn’t work. The monkeys had to store
the location of the apple in working memory on every trial and
hang onto that information across the delay. And that’s exactly
what they did.

•• While the monkeys were doing this, Fuster and Alexander were
recording from neurons in their brain. And they found a bunch of
neurons that fired continuously during the delay period and then
stopped firing after the animal made its response.

160 The Learning Brain

•• Keep in mind that nothing was happening during the delay
period. The blind was down, and the monkey couldn’t even see
the holes, much less the apple. Yet a bunch of neurons were firing
like crazy. Fuster and Alexander hypothesized that the activity in
these neurons corresponded to the monkey’s working memory of
the apple’s location.

•• These neurons were in the prefrontal cortex. Recall that the cortex
is the thin layer of gray matter around the outside of the brain,
and the prefrontal cortex is the cortex in the front of the brain.

•• Fuster and Alexander’s results suggested that the prefrontal cortex

is critical for working memory. And that idea also makes sense
given Alan Baddeley’s working memory model and the central
executive component, which controls our attention and decides
to keep some information active while letting other information
fade away.

•• The part of the brain that is associated with the central executive
is the prefrontal cortex. According to Baddeley’s theory, we should
expect to see neural activity in the prefrontal cortex during working
memory tasks. And we do.

•• Subsequent research has begun to tease apart what these prefrontal

neurons do in more detail. One very important line of work was
conducted by Patricia Goldman-Rakic and colleagues. Like Fuster
and Alexander, Goldman-Rakic’s team recorded from prefrontal
neurons while monkeys performed a delayed response task.

•• But instead of asking monkeys to remember which of 2 holes an

apple was hidden in, they required the monkeys to remember the
location where a visual stimulus, such as a square, flashed on a
screen. The square could appear in any of 8 different locations
and, just like in Fuster and Alexander’s experiment, the monkey
had to remember that location across a delay.

Lecture 17 | The Neural Basis of Working Memory 161

 •• This task is a little more difficult, but the monkeys did eventually
learn to do it quite well. And like Fuster and Alexander’s study,
Goldman-Rakic’s team found a bunch of prefrontal neurons that
fired vigorously during the delay. But they also found that many
of the neurons were location-specific.

•• For example, one neuron was active whenever the monkey was
trying to remember that the square had been in the upper-right
corner of the display, and a different neuron was active when the
monkey was trying to remember that the square had been in the
lower-left corner. But the neural responses were consistent: The
same neurons always fired when the monkey was remembering
“upper-right,” and a different set was always active when it was
remembering “lower-left.”

•• In other words, these prefrontal neurons have very specific duties in

spatial working memory. They don’t all contribute to remembering
all locations. Instead, each of them is responsible for remembering
a specific location and doesn’t actually participate in remembering
other locations.

•• And it turns out that it’s not just the prefrontal cortex that’s
involved in working memory. Subsequent experiments have
demonstrated neural activity during the delay period in a number
of more posterior parts of the brain. And the posterior brain area
that is activated is typically the area that represents the kind of
information being remembered.

•• For example, we tend to represent spatial information in the

parietal cortex, which is toward the top of the back half of the
brain. Patients with damage to this part of the brain often exhibit
a deficit called visuospatial neglect, in which they neglect, or fail
to pay attention to, one side of space. They might not read the text
on the left side of a page, for example. Clearly, the parietal cortex is

162 The Learning Brain

 crucial for representing the space around us. And it turns out that
when we store spatial information in working memory, neurons
in that area of the brain are active during the delay.

•• Based on results like this, one popular hypothesis is that when

we need to store information in working memory, we activate the
same brain regions that we normally use to represent that kind
of information during perception. If we use the parietal cortex to
represent spatial information during perception, then we use that
same area to store spatial information in working memory. And
the same principle is thought to apply to auditory information,
linguistic information, and so on.

Applying Animal Results to Humans

•• Monkeys may be smart, but we’re smarter, and our working
memory abilities are also significantly better. So, it’s important
to test whether the results found in monkeys also apply to
human beings.

•• Around 1990, scientists developed some methods for measuring

human brain activity without having to crack open the skull.
These so-called neuroimaging methods have revolutionized the
study of the human mind and brain. They’ve also revolutionized
the study of the neural basis of working memory in human beings.

•• One of the first neuroimaging studies of working memory in

humans was conducted by John Jonides and Edward Smith, who
used a technique called positron-emission tomography (PET).
Using this technique, Jonides and Smith estimated brain activity
while human participants performed 2 different working memory
tasks: one that required people to remember verbal information
and another that required people to remember spatial information.

Lecture 17 | The Neural Basis of Working Memory 163

 •• By comparing the brain activity patterns in the 2 tasks, Jonides
and Smith could identify the neural correlates of working memory
in humans. They could also test whether remembering different
kinds of information activates different regions, both in posterior
areas and in the prefrontal cortex.

•• They found that verbal working memory activated an area in the

left prefrontal cortex and a posterior area in the left hemisphere
that is commonly associated with language processing. And that’s
exactly what you would expect based on the animal studies.

In a typical PET study,

participants are injected
with a very low dose of
a radioactive tracer,
which follows the blood
supply wherever it goes,
including into the brain.
Furthermore, more blood
gets sent to brain areas that
are particularly active, so
more of the tracer goes to active
brain areas than to brain areas
that are inactive.
Because the tracer is radioactive, it decays and leads to the emission
of particles called positrons. The emission of those positrons can
be detected by a special scanner that can determine where the
positrons came from in the brain. Putting this all together, a
picture of brain activity can be created based on positron emission.

164 The Learning Brain

•• In short, the results of this PET study suggest that the neural
mechanisms underlying working memory in human beings share
some important similarities with the mechanisms used by monkeys.

•• But are the prefrontal regions involved in working memory general-

purpose ones, or are different areas of the prefrontal cortex engaged
when people have to hold on to different kinds of information? To
address this question, Jonides and Smith performed a PET study
in which participants had to try to remember spatial information.

•• Once again, they found that working memory activated both

prefrontal and posterior brain regions. They also found that the
posterior brain regions were areas associated with representing
and processing spatial information. So, consistent with the animal
studies, posterior brain areas involved in temporarily storing
information were some of the same areas normally involved in
representing and processing that information.

•• Furthermore, the spatial working memory task activated a

completely different part of the prefrontal cortex compared with
the verbal working memory task. In fact, it was on the other side
of the brain entirely. The verbal working memory task activated
the left prefrontal cortex, while the spatial working memory task
activated the right prefrontal cortex.

•• This has important implications for our understanding of what

the prefrontal cortex is doing during working memory. Apparently,
there’s not a single prefrontal region that performs exactly the
same function regardless of what you’re trying to store in working
memory; instead, different parts of the prefrontal cortex participate
in working memory for different types of information.

•• One hypothesis for what these prefrontal regions are doing is the
idea that the prefrontal cortex provides top-down excitation of
posterior brain regions during working memory tasks.

Lecture 17 | The Neural Basis of Working Memory 165

 •• Consider what happens when you perceive an orange. Nerve cells
in your eyes send signals into the brain that cause other nerve cells
to fire. And those cells cause still other cells to fire, and so on. So,
the neural activity is driven by a stimulus coming from outside.
That’s called bottom-up processing.

•• But what if you’re not perceiving an orange anymore? What if

you’re just temporarily storing the image of an orange that you saw
recently but that was taken out of view? Now the nerve cells in
your eyes are not sending signals back into the brain and driving
a pattern of neural activity that corresponds to your representation
of the orange. What’s going to drive the neural activity and keep
it active even when you can no longer see the orange?

•• One natural possibility is that’s what the prefrontal cortex does: It

drives the neural activity in posterior brain areas and keeps them
active, even though there’s nothing coming in from perception.
And when neural activity is driven internally rather than being
driven by external stimuli, that’s called top-down processing.

•• The hypothesis that the prefrontal cortex provides top-down

excitation of posterior brain regions during working memory
tasks also makes sense of the finding that different regions of the
prefrontal cortex are involved in working memory for different
kinds of information.

•• Different parts of the prefrontal cortex are connected to different

parts of the posterior cortex. And if a prefrontal region is going to
provide top-down excitation of a posterior region, then it needs
to be connected to it.

166 The Learning Brain

•• This explains why different parts of the prefrontal cortex would
be activated when you’re storing different types of information.
The information is actually being stored in the posterior cortex, in
the same area that tends to represent and process that information
during normal perception. And because there is a specific part
of the prefrontal cortex that is connected to that part of the
posterior cortex, that’s the region that would provide top-down
excitation for it.

The Convergence of
Psychology and Neuroscience
In Alan Baddeley’s model of working memory, which
assumes that working memory can be subdivided into
different components, he argues that we should distinguish a
phonological loop from a visuospatial buffer and that both of
these are controlled by a central executive.
That model actually maps pretty naturally onto the neural
results addressed in this lecture. For example, Jonides and Smith
found that storing verbal information depends on completely
different brain circuits than storing spatial information. And that
makes perfect sense, assuming that we use the phonological
loop to store verbal information while we use the visuospatial
sketchpad to store spatial information.
Furthermore, a natural interpretation of the neural data is that
the prefrontal cortex provides top-down excitation of posterior
brain areas. That seems similar to the top-down control that
you would expect Baddeley’s central executive to exert. And
Baddeley assumes that the central executive component of
working memory is implemented in the prefrontal cortex.

Lecture 17 | The Neural Basis of Working Memory 167

 SUGGESTED READING

Fuster, The Prefrontal Cortex.

Osaka, Logie, and D’Esposito, The Cognitive Neuroscience of
Working Memory.
Roberts, Robbins, and Weiskrantz, The Prefrontal Cortex.

QUESTIONS TO CONSIDER

1 Baddeley’s psychological model of working memory maps onto

some of the neuroscientific findings in some ways. Can you think
of any ways in which it doesn’t?

2 Based on what you’ve learned, how do you think damage to the

frontal lobe would affect someone’s behavior?

168 The Learning Brain

 LECTURE 18
Training Your Working
Memory

S  ME SCIENTISTS BELIEVE THAT WITH

O
APPROPRIATE TRAINING, INTELLIGENCE CAN
be increased somewhat. Specifically, there
is some evidence that training someone’s
working memory can lead to increases in
fluid intelligence, which refers to the ability to
reason and solve new problems independent
of previously acquired knowledge. There
is even evidence that this kind of training
can help children with attention deficit
hyperactivity disorder (ADHD). But the claim
that training can change intelligence is highly
controversial, and many scientists don’t
believe it. 

Lecture 18 | Training Your Working Memory 169

 IMPROVING FLUID INTELLIGENCE

•• In 2008, a study was conducted by Susanne Jaeggi, Martin

Buschkuehl, John Jonides, and Walter Perrig that was inspired
by the work of Graeme Halford and his colleagues, who argued
that working memory and intelligence share a common capacity
constraint. In other words, one of the key factors influencing how
smart you are is how much information you can hold in your
working memory.

•• Evidence shows that working memory is very important to thought

in general. People with greater working memory capacity tend
to perform better on a wide range of cognitive tasks, such as
reading comprehension, reasoning, and standardized tests. In
addition, working memory capacity is one of the best predictors
of performance on IQ tests. And because the problems on IQ
tests tend to predict performance on a bunch of other cognitive
tasks, that suggests that working memory plays a central role in
cognition more generally.

•• Inspired by these kinds of findings, Jaeggi and her colleagues

wanted to see if training people’s working memory could increase
working memory capacity and thereby increase fluid intelligence.
They developed a working memory task for the training that
could be made increasingly difficult as people improved so that
it would continue to stretch the working memory capacity of the
participants and hopefully cause that capacity to grow.

•• The researchers found that the training on the working memory

task improved participants’ task performance. But the critical
question is whether training leads to improvements in fluid
intelligence. To answer that question, the researchers gave
participants a 10-minute IQ test both before and after training
and tested whether participants would be able to solve significantly

170 The Learning Brain

 more problems after training compared with before. Of course,
the problems that were used were different in the pretest and
the posttest.

•• Sure enough, participants solved significantly more problems

after the working memory training than they did initially. Before
training, participants were able to solve between 9 and 10 problems
within the 10-minute time limit. But after training, they were
solving 12 problems on average.

•• Although that finding is consistent with an increase in fluid

intelligence, that’s not the only possible explanation. For example,
maybe the problems on the second test were a little easier than
the problems on the first test. Or maybe taking the first test gave
the participants some practice doing IQ tests, and that’s why they
performed better on the second test. Neither of those explanations
has anything to do with the working memory training.

•• To address those kinds of concerns, the researchers also recruited a

control group of participants who didn’t do the working memory
training. But in every other way, they were treated the same as the
training group. They took the same IQ tests, and the amount of
time between the 2 tests was the same as the training group. But
instead of training on the working memory task, they just went
about their lives as normal.

•• On the initial IQ test, the control group’s performance was

virtually identical to that of the training group. They also solved
between 9 and 10 problems in the 10-minute time limit on average.
But on the second test, the performance of the 2 groups was very
different: Whereas the training group solved about 12 problems
on average, the control group only solved about 10.5 on average.

Lecture 18 | Training Your Working Memory 171

 •• That difference between the training group and the control group
can’t be attributed to a difference in how easy the tests were. After
all, the control group took the same tests. It also can’t be attributed
to practice on the first test, because the control group also got to
take the first test. And they performed just as well as the training
group did before the training.

•• The researchers therefore concluded that the superior performance

of the training group on the second IQ test was caused by the
training. Apparently, working memory training can improve
performance on IQ tests. And because IQ tests are designed
to measure fluid intelligence, the results suggest that working
memory training can increase fluid intelligence.

•• If working memory training really does increase fluid intelligence,

then you might expect that more working memory training would
lead to greater increases than less training would. And that’s
exactly what the researchers found: The more training people
got, the more their performance on the IQ test improved. These
results suggest that working memory training causes an increase
in performance on the IQ test.

Improving ADHD Symptoms

•• Another area in which working memory training has sparked a
lot of interest is attention deficit hyperactivity disorder (ADHD),
which is characterized by long-term problems in paying attention,
excessive activity, and difficulty with impulse control. It’s often
treated by some combination of psychotherapy and psychostimulant
medication, but there is a lot of interest in trying to find behavioral
interventions that might help. And some evidence suggests that
working memory training might be such an intervention.

172 The Learning Brain

•• In 2005, Torkel Klingberg and colleagues published a randomized
control trial providing some of the strongest evidence that working
memory training might help ADHD. They recruited 53 children
with ADHD between the ages of 7 and 12.

•• Half the children were randomly assigned to complete at least

20 days of computerized working memory training for about 40
minutes per day. The children had to do both visuospatial working
memory tasks and auditory working memory tasks, and the tasks
became more and more difficult as the children got better.

•• The other half of the children served as a control group. They also
performed the same working memory tasks for the same amount of
time, but the tasks were all easy and didn’t become more difficult
as the children improved. So, the only real difference between the
2 groups was in the intensity of the working memory training and
whether it stretched the children or not.

•• The scientists reported 2 interesting results: The children in

the training group improved more than the control group on 2
working memory tasks on which they weren’t trained, suggesting
that the training improved working memory ability more generally
and not just the tasks on which the children got practice; and the
children in the training group also improved more than the control
group on 2 tests that are typically impaired in ADHD.

ADHD is the most

commonly diagnosed
mental disorder
in children and
adolescents.

Lecture 18 | Training Your Working Memory 173

 •• Consistent with these lab results, the parents of the children in
the training group reported a significant reduction in ADHD
symptoms compared with the parents of children in the
control group.

There is evidence that appropriate training can increase working

memory capacity and even improve fluid intelligence, and those
effects might be related to changes in brain connectivity. However,
not all studies that have trained working memory have improved
fluid intelligence.
Unfortunately, this is also the case in the ADHD literature.
Although some studies find improvements in ADHD symptoms
after working memory training, other studies don’t.
Why this is the case is still a topic of considerable debate. But these
discrepancies will all get resolved eventually; over time, science gets
closer and closer to the truth.

How Does Working Memory

Training Affect the Brain?
•• What is changing in the brain during and after working memory
training? Todd Thompson, Michael Waskom, and John Gabrieli
investigated this question using functional MRI, a technique that
is similar to positron-emission tomography that can be used to
measure brain activity in humans.

174 The Learning Brain

•• The scientists randomly assigned some participants to train on
Jaeggi’s working memory task for 20 days. Other participants were
randomly assigned to train on a perceptual object tracking task that
is also demanding but that doesn’t emphasize working memory.

•• Both before and after training, all the subjects participated in a

functional MRI scan that estimated brain activity during Jaeggi’s
working memory task. The researchers then examined how that
activity changed as a function of the training.

•• Thompson and his colleagues found that the working memory

training led to 2 major changes. First, brain activity was reduced
after working memory training compared with before training.
This is now a very common finding in neuroimaging studies, and
the standard interpretation is that the more you practice a task, the
less hard the brain has to work to perform it. The result is lower
activation levels.

•• But this result isn’t specific to working memory training; you’ll

typically find reduced brain activation after practicing any task,
whether it involves working memory or not.

•• They also found another neural change that was specific to working
memory training: The connections between prefrontal regions and
posterior brain regions became stronger. How could strengthening
these connections help improve working memory performance?

•• As you learned in the previous lecture, a very popular theory about

how the prefrontal cortex interacts with posterior brain regions
during working memory argues that it provides top-down excitation
of those posterior regions. The idea is that when you’re actively
perceiving visual or auditory stimuli, the perceptual pathways from
the eyes and ears drive activity in posterior brain regions.

Lecture 18 | Training Your Working Memory 175

 •• But how do you keep the brain representations of those stimuli
active if you want to remember them after the stimuli disappear?
You need excitatory input coming from somewhere else. And that’s
what the prefrontal cortex provides, according to this theory.

•• But for that to work, you need good connections between the
prefrontal cortex and the posterior brain regions that it excites.
And sure enough, as those connections get stronger, working
memory performance improves.

Should you spend time and effort trying to train your

working memory?
There are plenty of brain games, computer games, and apps that
offer the opportunity to train your brain, and many of those
programs include working memory training. But in general, it is
not worth it to spend time engaged in such training.
There is some evidence that such training might help you, and
it might also be fun. But if your main goal is to improve your
cognitive abilities, there are better ways to spend your time.
In particular, evidence shows that
eating a good diet and pursuing
an active lifestyle with plenty
of physical exercise can help
your brain at least as much
as cognitive training—and
probably more.

176 The Learning Brain

•• But that theory makes another interesting prediction: People whose
connections strengthen the most during the training should also
exhibit the most improvement in working memory performance.
And that’s exactly what Thompson and his colleagues found. The
people whose prefrontal-posterior connections changed the most
also showed the most change in working memory performance.

•• Together, these results suggest that working memory training leads

to changes in brain connections. And those connectivity changes
contribute to improvements in fluid intelligence.

SUGGESTED READING

Jaušovec and Pahor, Increasing Intelligence.

Klingberg, The Overflowing Brain.
——— , “Training and Plasticity of Working Memory.”

QUESTIONS TO CONSIDER

1 Based on what you learned in this lecture, do you believe Jaeggi’s

claim that they managed to increase fluid intelligence in their
participants? Why or why not?

2 How would society be affected if we could improve people’s

intelligence, perhaps even dramatically? Would all the effects
be positive?

Lecture 18 | Training Your Working Memory 177

 LECTURE 19
How Motivation
Affects Learning

T  E NEXT 5 LECTURES ARE ABOUT FACTORS

H
THAT CAN HAVE A BIG IMPACT ON LEARNING
and memory. This lecture is about a factor
that may fly under the radar but can have
an enormous effect on learning: motivation.
Psychologists have done quite a bit of work
exploring the question of what influences
motivation and have identified at least 4
factors that play a critical role: self-efficacy,
perceived control, intrinsic motivation, and
perceived value. As you will learn in this
lecture, these factors can influence learning
in different ways. 

178 The Learning Brain

Academic Achievement of
American High School Students
Every 3 years, the Programme for International Student Assessment
tests 15-year-olds all over the world to assess their achievement in key
academic areas, such as reading, math, and science. In 2015, the US
was ranked 38th out of 71 countries in math, 24th in science, and 24th
in reading. And that’s despite the fact that America spends more per
student than all but 3 or 4 countries in the world.
Many people are quick to blame the schools or teachers for the
mediocre performance of US students. But one factor that is often
overlooked is how teen culture in America views education and how
motivated high school students are to work hard in school.
To investigate this issue, Laurence Steinberg led a large survey in which
about 20,000 American high school students were asked about their
motivation and engagement in school. Here are a few of the more
startling findings:
•• More than 1/3 of high school students said the primary way
they get through the day at school is by “goofing off with their
friends.”
•• Almost 90% of high school students said they had copied
homework from a friend in the last year.
•• American high school students spend about 4 hours per week
on homework. High school students in other cultures typically
spend 4 hours on homework every day.
•• Less than 20% of those surveyed said their friends think it’s
important to do well in school, and less than 25% regularly
discuss schoolwork with their friends.
•• Almost 20% of students said they don’t try as hard as they
can in school because they’re afraid of what their friends
would think.
Based on these kinds of results, Steinberg argued that the real crisis in
American education doesn’t lie in the schools themselves; rather, it lies
in a culture that undermines student motivation and engagement.

Lecture 19 | How Motivation Affects Learning 179

 Self-Efficacy
•• Self-efficacy refers to your confidence in your ability to achieve
some goal. If you’re confident in your ability to solve crossword
puzzles, then you would have high self-efficacy in that domain.
Conversely, if you have very little confidence in your crossword
puzzle ability, then your self-efficacy would be low.

•• Your self-efficacy is likely to be different in different domains. You

might be very confident in your ability to solve crosswords, but
not at all confident in your ability to solve calculus problems—
or vice versa. Everyone has a different profile of self-efficacy
across domains.

•• It’s important to note that your self-efficacy, or confidence, in

some domain can be different than your actual ability in that
domain. But the opposite can also be true; that is, your ability can
be much higher than your self-efficacy.

•• Self-efficacy has an enormous impact on learning. And there are a

few reasons. First, self-efficacy influences what people try to learn.
The basic idea is that if you don’t believe that you’re capable of
learning something, then you may not bother even trying. After
all, why devote the time and energy if you believe you’re going
to fail?

•• For example, imagine a high school student who believes that she
is bad at math. Maybe she had a bad experience in an earlier math
class, or maybe other people have told her that she’s not the type
of student who does well at math. But for whatever reason, she
has now come to believe that she is just not cut out to learn math.
As a result of that low self-efficacy, she decides not to try. Rather
than risking failure, she chooses to take other, less challenging
courses instead.

180 The Learning Brain

•• This happens all the time. A much lower percentage of women
than men pursue college majors in the STEM (science, technology,
engineering, math) fields. And this is not because women are any
less capable than men in these areas. But women who are just
as capable as men nevertheless often doubt their ability in these
domains more than men do; that is, their self-efficacy is lower even
though their ability isn’t. And if your self-efficacy is low, you may
decide not to pursue a goal that you could actually accomplish if
you set your mind to it.

•• Another way that self-efficacy influences learning is by affecting

persistence. Research shows that people with higher self-efficacy
tend to continue working toward their goals even in the face
of difficulties.

•• Suppose that you’re trying to learn a challenging skill, such as how

to play the piano. It’s going to take a serious commitment of time
and energy to get good at playing the piano, and you’re going to
face some obstacles that you’ll have to overcome
with hard work and persistence.

•• But if your self-efficacy is

low, you may mistakenly
conclude that you
just can’t do it and
that it’s not worth
continuing to work
at it. As a result,
many people with
low self-efficacy give
up even when they
probably could have
acquired the skill with a
little more work.

Lecture 19 | How Motivation Affects Learning 181

 Perceived Control
•• Another factor that can significantly influence motivation and
learning is perceived control, which refers to the extent to which
you believe that you are in control of how much you learn, as
opposed to your learning being dependent on forces outside
your control.

•• For example, imagine a student named Peter who does poorly on

an exam despite studying really hard for it. Given that he studied
hard, Peter might feel like the exam was simply unfair and that
his poor performance was outside his control.

Just as self-efficacy is about your belief in your ability

rather than your ability itself, perceived control
is about your belief in how much control
you have rather than the actual level of
control itself.
A number of studies have found
that perceived control can have a
substantial impact on motivation
and engagement in learning. For
example, Raymond Perry, along
with a number of his colleagues,
found that students who were high on
perceived control exerted more effort
in their studies than students whose
perceived control was low. They also
reported less boredom and anxiety during
school as well as greater motivation to try to do
well. And they also got better grades.

182 The Learning Brain

•• That would be an example of low perceived control. Peter attributes
his poor performance to an external factor over which he has no
control—namely, an unfair exam.

•• But now imagine another student named Holly who also studied
hard for the same exam and also performed poorly. But unlike
Peter, Holly assumes that if she had prepared in a different
way, then she could have done better. So, maybe she goes to the
instructor’s office hours to review her exam, figure out what she
missed, and identify more effective ways to study in the future.

•• That would be an example of high perceived control. Holly assumes

that she has some control over her performance and that if she
studies in a more effective way, she’ll do better on the next exam.

•• Note that in this situation, Peter and Holly perceive different

levels of control in exactly the same learning situation. They both
presumably have about the same level of control in this situation,
but their perception of their level of control is different.

•• High perceived control is closely related to the idea of a growth

mindset, an idea that has been studied extensively by Carol Dweck,
who has argued that it’s important to distinguish between what
she calls a fixed mindset and a growth mindset.

•• People with a fixed mindset view their talents and abilities as fixed
and unchangeable. The idea of a fixed mindset and the idea of low
perceived control are similar. If your poor performance is a result
of fixed abilities, then you have very little perceived control. And
from that point of view, putting a lot of energy into studying or
working hard doesn’t make a lot of sense. In fact, students with
a fixed mindset might view the need to work hard as a sign of
inferior intelligence.

Lecture 19 | How Motivation Affects Learning 183

 •• In contrast, people with a growth mindset view their abilities as
malleable, or changeable. A growth mindset is more consistent
with a high level of perceived control. If you think your abilities
can change with effort, then that means you have more control
over your performance, so poor performance would tend to lead
to more effort rather than less.

•• People with a growth mindset also tend to have different

goals than people with a fixed mindset. Specifically, the goals
of people with a growth mindset are typically more about
improving and ultimately mastering some challenging domain.
That’s sometimes called a mastery orientation. Conversely, the
goals of people with a fixed mindset are typically more about
performing well and demonstrating their innate ability. They have
a performance orientation.

•• There’s now quite a bit of evidence that having a growth mindset

is associated with significantly more learning and better academic
performance. For example, one study found that academic
performance in seventh graders with a growth mindset tended
to improve over the next few years of school, while the academic
performance of those with a fixed mindset stayed relatively constant.

•• Even more impressive, teaching students that intelligence is

malleable rather than fixed has been shown to improve subsequent
learning and academic performance.

•• Dweck has even conducted studies suggesting that it might actually

be counterproductive to praise children for their intelligence,
because doing so tends to emphasize the ability and might promote
a more fixed mindset. Instead, she suggests praising children for
the process and for their development and growth.

184 The Learning Brain

 Intrinsic Motivation
•• A third factor that has been found to significantly influence
learning is intrinsic motivation. When you’re intrinsically
motivated to do something, you do it because you want to. It’s
inherently interesting or enjoyable to you, so you do it for the fun
of it. In contrast, when you’re extrinsically motivated to perform
a task, you do it because it satisfies some external goal.

•• Intrinsic motivation and personal interest are associated with

greater learning across many studies. For example, fourth graders
remember more about assigned reading passages that they rate
as personally interesting compared with passages that they find
less interesting.

•• Likewise, college students who develop an interest in a particular

topic, such as computers, are more likely to study computer-related
information on their own and persevere longer on computer-
related assignments compared with students who have less interest
in the topic.

•• Surprisingly, adding external incentives for good performance can

often reduce intrinsic motivation and undermine learning. For
example, in one famous study, undergraduate psychology students
worked on an inherently interesting puzzle called the Soma cube. The
puzzle could be put together in a variety of different configurations,
and in each session, the students were timed while they tried to
reproduce a configuration that was shown on a piece of paper.

•• In one of the sessions, half the students were paid a dollar for each
configuration they were able to reproduce. But in the next session,
they went back to doing it without any monetary incentive. The
other half of the students never got any money and just did it for fun.

Lecture 19 | How Motivation Affects Learning 185

 •• Interestingly, the students who never got paid maintained their
interest in the puzzle and kept working on it, while the students
who had been paid lost interest after the money dried up.

•• The researchers argued that providing an external incentive—

money—had undermined the students’ intrinsic motivation. Now
rather than doing it for fun and interest, they started doing it for
the money. And when there was no more money, they were no
longer interested.

•• This effect poses a challenge for education: On one hand, teachers

need to provide tests and external assignments with grades that
serve as an incentive for students to work hard, but doing so could
undermine the students’ intrinsic motivation and interest and
paradoxically impair their learning. Providing appropriate external
incentives while nurturing a student’s intrinsic motivation is one
of the real challenges of effective teaching.

Perceived Control
•• The fourth factor that has been shown to have a significant
impact on learning is the value that students place on what they’re
learning. Do students care about what they’re learning? Do they
think that what they’re learning is important in some way? If so,
that can significantly increase the amount that they learn.

•• Jacquelynne Eccles has done some of the best work on the

importance of value in learning. For example, she and her
colleagues have demonstrated that student beliefs about the value
of mathematical knowledge are one of the best predictors of
whether those students will enroll in math classes or not. But once
the students are enrolled, their self-efficacy, perceived control, and
intrinsic interest are the best predictors of their actual performance.

186 The Learning Brain

 SUGGESTED READING

Bandura, Self-Efficacy.
Deci and Flaste, Why We Do What We Do.
Duckworth, Grit.
Dweck, Mindset.

QUESTIONS TO CONSIDER

1 There are 4 factors that can significantly influence motivation:

self-efficacy, perceived control, intrinsic motivation, and perceived
value. Which of these factors do you think has had the biggest
influence on your motivation to learn?

2 How could teachers apply the information in this lecture to help

unmotivated students?

Lecture 19 | How Motivation Affects Learning 187

 LECTURE 20
How Stress and Emotion
Affect Learning

H  W ARE LEARNING AND MEMORY

O
AFFECTED BY HOW WE FEEL? EXACTLY
how do our feelings affect our learning and
memory? What’s going on in the brain and
the body that might explain the effects of
stress and emotion on memory? This lecture
will address these types of questions. 

188 The Learning Brain

 Emotional Memories
•• Psychologists refer to the vivid memories of traumatic or emotional
events as flashbulb memories. For many people, their memory of
the 9/11 attacks are a good example: They vividly remember where
they were, who they were with, what they were doing, and many
other contextual details.

•• Our memories for emotionally arousing events tend to be more

vivid than our memories for more neutral events. For example,
Margaret Bradley, Peter Lang, and their colleagues recruited nearly
90 undergraduates and asked them to rate pictures based on how
emotionally arousing they were and how pleasant they were.

•• Immediately afterward, they asked the participants to describe as

many pictures as they could remember. They also called them a
year later and asked them to do the same thing again. And on both
tests, participants remembered significantly more of the pictures
that they rated as emotionally arousing.

•• There’s also evidence that the effect of emotion on memory actually

gets stronger over time. Stephen Kaplan demonstrated this effect
as early as the 1960s. He and a student named Lewis Kleinsmith
asked about 50 undergraduate students to try to remember 8
arbitrary word-number pairings. For example, they might have to
remember that the word “vomit” was paired with the number 4
while the word “swim” was paired with the number 7.

•• Some of the words, such as “vomit” and “rape,” were more

emotionally arousing than other, more neutral words, such as
“swim” and “dance.” And those differences had a dramatic impact
on people’s memory and how that memory changed over time.

Lecture 20 | How Stress and Emotion Affect Learning 189

 •• About 20 minutes after studying the pairs, people were able to
remember about 20% of the numbers associated with words that
were not emotionally arousing. The next day, that percentage
dropped to about 12%. And a week later, participants remembered
virtually none of the numbers associated with the neutral words.

•• But the researchers found the opposite for words that were highly
arousing: 20 minutes after studying the pairs, the participants
remembered about 20% of the numbers associated with words
that were emotionally arousing. That’s about the same as for the
neutral words. But the next day, their memory for those numbers
actually got better rather than worse. Now they remembered
almost 40% of the numbers. And after a week, their performance
was even better.

•• More recent studies have found similar results. As time passes, the
gap between emotional memories and neutral memories grows.
And that has led many scientists to hypothesize that emotional
arousal strengthens memory consolidation.

•• So, emotional memories are typically more vivid than neutral

memories. People create flashbulb memories that are extremely
detailed and that they remember for a long time. But are those
emotional memories accurate? Most people certainly assume that
they are, given their vividness and their attention to detail. And
if asked to rate their confidence in the accuracy of emotional
memories, people inevitably give very high confidence ratings.

•• But our memory often reflects a plausible reconstruction of what

we experienced, and it can be distorted in a variety of ways.
Furthermore, we’re typically unaware of these distortions in our
memory and assume that what we think we remember actually
happened. And it turns out that’s also true for emotional memories.

190 The Learning Brain

•• For example, consider the Challenger disaster. On January 28,
1986, the space shuttle broke apart 73 seconds into its flight,
killing all 7 crew members. Many people vividly remember where
they were and what they were doing when they first heard about
the tragedy.

•• Ulric Neisser and Nicole Harsch interviewed 44 students who

were in Neisser’s psychology class when the disaster happened.
He asked them to fill out a questionnaire about their memory
for the event in class the day after it happened and contacted
them again when they were seniors and asked them to fill out a
similar questionnaire.

•• The researchers compared the 2 reports on 7 key attributes,

including where they said that they were when they heard the
news, what they were doing, and who told them. Half the subjects
gave conflicting answers for at least 5 of the attributes, and a
quarter of them were wrong about all 7.

•• Nevertheless, when asked to rate their confidence in their memory

on a scale of 1 to 5, the average rating was 4.17. In fact, 13 of the
44 participants rated their confidence as 5 out of 5, and 3 of those
13 people had actually misremembered all 7 key attributes.

•• Clearly, although emotional memories are vivid and long-

lasting, they are subject to the same kinds of distortions as other
memories are.

•• Most of the time, it’s presumably adaptive to form vivid memories

for emotionally arousing information. Remembering a traumatic
event such as a car accident or getting robbed might help you
avoid similar situations in the future. But in some cases, traumatic
memories can become maladaptive and intrusive.

Lecture 20 | How Stress and Emotion Affect Learning 191

 •• For example, victims of post-traumatic stress disorder (PTSD)
often experience intrusive and recurrent flashbacks to previous
traumatic events. And those symptoms can persist for an extended
period of time, significantly interfering with their daily life.

When most people think of PTSD, they

think of war veterans whose memories
of traumatic battles have become
consuming and debilitating.
But soldiers aren’t the only
ones who experience PTSD.
Experiences like the sudden,
unexpected death of a loved
one or being physically or
sexually assaulted can also
cause PTSD.
It’s normal to feel afraid
during and after traumatic
events, and almost everyone
will experience a variety
of unpleasant reactions for
a period of time. But most
people recover from those
symptoms naturally within a
month or so. Victims of PTSD
don’t. They continue to feel extremely
stressed and frightened long after the
disturbing event, even when they’re no
longer in any danger.

192 The Learning Brain

 How Feelings Affect Learning and Memory
•• In a hypothesis proposed by J. A. Easterbrook called cue utilization
theory, he argued that emotional arousal tends to restrict our
utilization of environmental cues. Put simply, the more emotionally
aroused we get, the more we restrict our focus of attention to only
the most relevant environmental features. And focusing attention
can often improve learning.

•• One of the best pieces of evidence for cue utilization theory is

the weapon focus effect, which is the finding that when people
encounter someone with a weapon, they tend to focus most of their
attention on the weapon itself. As a result, they often remember
a lot about the weapon, but not so much about the perpetrator or
other features of the event.

•• In one study, participants were presented with a sequence of pictures

of a customer going through a line at a fast-food restaurant. Half of
the participants saw some pictures in which the customer pointed
a gun at the cashier, who then handed over some money. The
other half saw the customer hand the cashier a check, and then
the cashier also handed over some money. The pictures for the 2
groups were identical except for the change from a gun to a check.

•• The researchers found that participants in the weapon condition

spent more time looking at the gun, while participants in the
check condition spent more time looking at other features, such as
the customer’s appearance. Sure enough, the people in the check
condition remembered the customer better than the people in the
gun condition—just as cue utilization theory would predict.

•• This result also demonstrates that stress and emotional arousal

don’t always improve every aspect of memory. In this case, the
stressful gun condition actually produced worse memory for the

Lecture 20 | How Stress and Emotion Affect Learning 193

 customer than the unstressful check condition did. It all depends
on what you pay attention to.

•• Stress tends to make you narrow your focus of attention to the most
central, important information. And memory for that information
is often improved. But memory for other information is typically
impaired, precisely because you ignored it.

•• Short-term, acute emotions and stress tend to make memories

stronger and more vivid, although they don’t protect memory
from subsequent distortion. Does long-term, chronic stress also
strengthen memory? To answer that question, Victoria Luine,
along with Bruce McEwen and 2 other colleagues, measured
memory performance of rats that had been exposed to chronic
stress for 3 weeks and of rats that had not been stressed.

•• The animals were placed in a radial arm maze, in which 8 narrow

tracks, called arms, extend outward from a central platform. At
the end of each arm is a small well containing a peanut. The rat is
placed in the middle and has to walk out to the end of each arm
to get all 8 peanuts.

•• The rat wouldn’t want to walk down the same arm twice because
the peanut won’t be there anymore, so that would be considered
a mistake, while walking down an arm that the rat hasn’t visited
before would be considered a correct answer.

•• Luine and her colleagues compared the memory performance

of rats who had been stressed out for 3 weeks with the memory
performance of rats who hadn’t and found that stress impaired
memory performance. The chronically stressed rats made more
mistakes and made them earlier than the control rats.

•• While short-term stress often strengthens memory, long-term,

chronic stress seems to undermine it.

194 The Learning Brain

 The Biological Basis of Stress and Emotion
•• What’s going on in the body and the brain when we experience
intense emotions? And how does it affect learning and memory?

•• When we’re stressed or experience intense emotions, we activate

a biological system called the HPA axis, which includes the
hypothalamus, the pituitary gland, and the adrenal glands. When
the HPA axis is activated, the end result is the release of stress
hormones in the bloodstream, which in turn trigger a fight-or-
flight response: You get a jolt of adrenaline, your mouth goes dry,
your pupils dilate, your sensations are heightened, and your heart
starts to race.

•• Given that short-term stress often strengthens memory, it’s natural

to ask whether it’s the stress hormones that produce that effect.
To investigate that question, Larry Cahill and James McGaugh,
along with 2 other colleagues, used a drug called propranolol to
block the effect of the stress hormones. Then, they examined the
effects on memory.

•• Stress hormones work by attaching to special receptor molecules

throughout the body and turning them on, triggering the fight-or-
flight response. Propranolol attaches to the same receptors that the
stress hormones attach to, but it prevents the stress hormones from
turning the receptors on. So, if the activity of stress hormones is
critical to enhancing emotional memory, then emotional memory
should not be enhanced in people who take propranolol.

•• Cahill and his colleagues tested this idea by presenting people with
2 stories to remember. One was an emotional story about a boy
walking with his mom, getting hit by a car, and being rushed to
the hospital to be treated for life-threatening injuries. The other
was a neutral story in which the boy and his mom walk to the
hospital and observe a simulated emergency response.

Lecture 20 | How Stress and Emotion Affect Learning 195

 •• Some people took propranolol before listening to the story while
other people took a placebo. A week later, everyone’s memory was
tested, and the results in the placebo group were compared to the
results in the drug group.

•• The placebo group showed the standard emotional enhancement

effect. People in this group had a significantly stronger memory
for the emotional story compared with the neutral story. But the
drug group didn’t. Their memory for the emotional story was no
stronger than their memory for the neutral story. These results
suggest that the release of stress hormones is what causes emotional
memories to be vivid and strong.

•• There is also evidence that chronic exposure to stress hormones

over time can lead to changes in the brain that undermine
learning and memory. For example, certain asthma and anti-
inflammatory medications increase stress hormone levels, and
patients taking these medications sometimes exhibit learning and
memory deficits.

•• Recall that classical conditioning is the type of learning that

Ivan Pavlov studied in his dogs. After repeatedly presenting a
conditioned stimulus—a bell—with an unconditioned stimulus—
food—Pavlov’s dogs learned an association between the bell and
the food and would therefore salivate as soon as they heard the bell.

Hormones play an important role in how stress and emotion

affect memory.
There are specific areas of the brain that play a crucial role in
learning and remembering emotional information, and a lot of the
evidence has come from studies of classical fear conditioning.

196 The Learning Brain

•• Fear conditioning is very similar, except that the unconditioned
stimulus is something unpleasant, such as a shock, instead of food.
For example, you might condition a rat so that it associates the
sound of a tone with getting a shock. After learning, the rat will
exhibit a stereotypical fear response when it hears the tone. It will
freeze in place, and its breathing and heart rate will increase.

•• A number of studies have examined the neural basis of fear

conditioning and have repeatedly identified a neural structure
called the amygdala as being particularly important. The amygdala
is an almond-shaped brain region just in front of the hippocampus
in the medial temporal lobes. It is known to play an important
role in the processing of emotion, although its role in fear and fear
conditioning has been studied the most.

•• For example, rats and other animals that have lesions to the
amygdala don’t exhibit normal fear conditioning. After such
lesions, these animals do not learn the standard association between
a conditioned stimulus, such as a tone, and a fearful conditioned
stimulus, such as a shock.

•• Furthermore, administering drugs that block long-term

potentiation—the long-term strengthening of synaptic
connections that is thought to play a crucial role in many types
of learning—in the amygdala also prevent fear conditioning.
These studies demonstrate that the amygdala plays a crucial role
in emotional learning.

Many of the effects that emotion and stress

AMYGDALA
have on learning and memory seem to
depend on your amygdala.

Lecture 20 | How Stress and Emotion Affect Learning 197

 •• Other studies have shown that it’s also critical for the consolidation
of emotional memories. For example, injecting stress hormones
into the amygdala immediately after learning has been shown
to produce strong, longer-lasting memories. Conversely, blocking
the effect of stress hormones in the amygdala impairs this kind
of consolidation.

•• Similar results have also been observed in human beings who

have damage to the amygdala as a result of stroke or brain injury.
For example, these patients don’t show the normal narrowing of
attention that emotion tends to produce in people without brain
damage. They also don’t exhibit the same memory advantage for
emotional information compared with neutral information that
most people show, either initially or after a delay.

SUGGESTED READING

McEwen and Lasley, The End of Stress as We Know It.

Sapolsky, Why Zebras Don’t Get Ulcers.

QUESTIONS TO CONSIDER

1 Do you have any flashbulb memories for particularly emotional

events from your past? Do you think those memories are more
accurate than other memories, or just more vivid?

2 Can you think of any ways to reduce the level of chronic stress
in your life (and thereby potentially improve your learning and
memory)?

198 The Learning Brain

 LECTURE 21
How Sleep Affects Learning

M  NY PEOPLE VIEW SLEEP AS AN

A
ANNOYANCE AND A WASTE OF TIME, BUT
we can’t function as well when we don’t get
enough sleep. Our ability to think clearly,
to react quickly, and to focus our attention
all decline when we’re sleep deprived. In
addition to sleep’s role in helping us be
alert and focused, there is now a great deal
of evidence that sleep plays a crucial role in
learning and memory—which is the subject
of this lecture. 

Sleep deprivation has been implicated in a number of major tragedies,

including the space shuttle Challenger explosion, the Exxon Valdez
oil spill, and maybe even the Chernobyl nuclear disaster.

Lecture 21 | How Sleep Affects Learning 199

 The Behavioral Effects of Sleep on Learning
•• A number of studies have found a correlation between poor sleep
and impaired learning as measured by academic performance. For
example, one study found that the average grade point average
of high school students who slept less than 6 hours a night was
around a B−, whereas the average for students who slept 9 hours
or more was around a B+.

•• These types of results are compelling, but they’re also correlational.

And correlation does not imply causation: Just because poor
sleep is correlated with poor academic performance, that doesn’t
necessarily mean that poor sleep is causing the poor performance.
For example, maybe poor grades cause poor sleep rather than the
other way around, or maybe some other factor is causing both poor
grades and poor sleep.

We human beings spend roughly a third of our lives sleeping.

That means that a 60-year-old has spent about 20 years asleep.

200 The Learning Brain

•• To address these kinds of concerns, a number of studies have
randomly assigned some people to a sleep deprivation condition
and other people to a non–sleep deprived condition and then
assessed learning in the 2 groups. These studies have also found
that sleep deprivation impairs learning.

•• For example, in one study, conducted at the Stanford Center

for Sleep Sciences and Medicine, adolescents had their memory
tested after a normal night of sleep and after a night without
sleep. Specifically, they were presented with 25 four-letter
words and later were asked to try to remember as many as they
could. After a normal night of sleep, they remembered about
12 words on average. But after sleep deprivation, they only
remembered between 8 and 9. So, their performance was about
30% worse.

•• Studies like this confirm that learning

and memory are better after a good night’s A study of college
rest. After all, many things are better after
freshmen found
a good night’s rest; you’re more alert, and
you’re better at paying attention.
that good sleep
habits were a
•• You also learn and remember information better predictor
better before a good night’s rest. If you of GPA than any
study some information and then sleep other variable that
on it, the sleep itself often strengthens was considered,
and consolidates your memory so that including exercise,
you’ll be able to remember it better in diet, mood, stress,
the future. This effect applies to both social support, and
explicit, declarative memory as well as religious habits.
implicit, procedural learning.

Lecture 21 | How Sleep Affects Learning 201

 Strategies for Improving
the Quality of Sleep
By adopting some or all of the following strategies, the
quality of your sleep is likely to improve. And as your sleep
improves, so will your learning and memory.
•• Try to get enough sleep. Most adults should be getting
at least 7 hours of sleep at night, and preferably 8 or 9.
Different people need different amounts, but ideally you
should be going to bed early enough that your body
wakes up on its own before the alarm clock goes off.
•• Avoid caffeine, nicotine, and alcohol for a few hours
before you go to bed. All these drugs have been shown
to disrupt the quality of sleep—so has heavy food, so try
to avoid eating a big meal close to bedtime.
•• Tailor your bedroom for a good night’s sleep. Keep it
as peaceful and clean as possible so that it feels like
a relaxing oasis. Invest in a comfortable mattress and
pillow and get curtains that will block out the light as
effectively as possible. If it’s not quiet, consider a fan
or white noise machine. And keep it cool. We sleep
best when the air temperature is between 60° and 67°
Fahrenheit.
•• Establish a regular bedtime routine. Try to go to bed at
roughly the same time every day, even on weekends.
You might also adopt some relaxing bedtime habits,
such as reading or taking a bath, that your body will
begin to associate with going to bed and falling asleep.
And once you go to bed, try to avoid watching TV or
using electronics, which can make it more difficult to
fall asleep.
•• Get some exercise during the day. Take a walk or go for
a bike ride. As little as 10 minutes of aerobic exercise can
significantly improve the quality of your sleep.

202 The Learning Brain

 What Happens in the Brain while We Sleep
•• A common belief is that the brain turns off, or deactivates, when
we go to sleep. In fact, until the 1950s, that was a common
assumption among sleep scientists. But in 1953, Eugene Aserinsky
and Nathaniel Kleitman made a discovery that changed sleep
science forever.

•• They used electroencephalography (EEG) to record people’s

brain waves while they slept. And they found that, far from being
inactive, the brain goes through predictable cycles of unique
activity patterns when we sleep. In fact, at one point during these
cycles, the brain is just as active as it is when we’re awake.

•• Aserinsky also noticed that people move their eyes behind their
closed eyelids during these periods of significant brain activity,
but not during other periods of sleep. He therefore referred to the
highly active periods as rapid eye movement (REM) sleep and to
the other periods of sleep as non-REM sleep.

•• Scientists now know that there are many fundamental differences

between REM sleep and non-REM sleep. For example, if you
wake people up during REM sleep, they are much more likely to
report that they were dreaming, and quite often those dreams are
vivid and bizarre. But if you wake them up from non-REM sleep,
they’re much less likely to say they were dreaming.

•• REM sleep is also associated with a kind

of paralysis that prevents the limbs from Sleepwalking is
moving. Many scientists believe that this is associated with
the body’s way to prevent us from acting out non-REM sleep
our dreams while we’re sleeping. rather than
REM sleep.

Lecture 21 | How Sleep Affects Learning 203

 •• During normal sleep, we will typically cycle between non-REM
and REM sleep every 1.5 to 2 hours. We first enter a light non-
REM sleep, from which we can easily be awakened. Then, we
move into deeper stages of non-REM sleep, during which large
populations of neurons fire in unison 1 to 4 times per second.
This produces strong, low-frequency brain waves and is referred
to as slow-wave sleep.

•• During slow-wave sleep, our core temperature drops, and it’s more
difficult to wake up. We typically spend 20 to 40 minutes in this
kind of deep sleep before transitioning back into lighter non-REM
sleep for a few minutes and then finally into REM sleep.

•• The first period of REM sleep is usually pretty brief, lasting only
1 to 5 minutes. Then, we transition back into non-REM sleep,
and the cycle begins again. Overall, we spend about a quarter of
our total sleep time in REM sleep, about half in light non-REM
sleep, and another quarter in deep, slow-wave non-REM sleep, but
the proportion changes throughout the night.

•• With each new cycle, the REM periods tend to get a little bit
longer, while the deep, slow-wave periods tend to get shorter and
may disappear entirely in the last few cycles. As a result, the first
few cycles are dominated by deep, slow-wave sleep, while later
cycles are dominated by REM and light non-REM sleep.

•• The fact that the first few hours of sleep are qualitatively different
from the last few hours has led some scientists to hypothesize
that different types of sleep help to consolidate different types
of learning. Specifically, they have hypothesized that slow-wave
sleep is more important in the consolidation of explicit, declarative
memories, while REM sleep is more important in the consolidation
of implicit, procedural memories.

204 The Learning Brain

•• But why should sleep help consolidate any memories, whether
they’re explicit or implicit? One idea is that sleep protects new
memories. So, if you learn something new and then sleep on it,
then somehow the memory becomes less susceptible to interference
or forgetting. According to that view, sleep doesn’t change the
memory in any way; it just passively stabilizes it.

•• An alternative idea is that sleep actively changes memories and

generalizes them so that they can be used and remembered in a
wider variety of settings. Recent evidence is favoring this more
active view of sleep’s role in consolidation.

Reactivating Memories while We Sleep

•• There’s quite a bit of evidence that we reactivate our memories
while we sleep. And although it’s tempting to think that memory
reactivation would always correspond to a kind of dreaming and
might therefore occur during REM sleep, that’s not actually true.
Reactivation of declarative memories is most associated with non-
REM, slow-wave sleep.

•• But does this reactivation actually cause the memory to be

strengthened or consolidated? Remember, correlation does not
imply causation. To infer causality, we need to experimentally
manipulate reactivation and observe an effect on memory strength.
But how could we influence whether someone reactivates a memory
while they’re sleeping?

•• Björn Rasch, Jan Born, and a few colleagues did just that in one
of the most creative sleep experiments, in which human volunteers
learned an object location task before going to sleep for the
evening. Specifically, they were presented with an array of cards

Lecture 21 | How Sleep Affects Learning 205

 with pictures on them and had to remember where matching cards
were located in the array. After learning, they went to sleep, and
then their memory was tested the next morning.

•• But while they were learning the locations of the cards, they were
also presented with a smell—specifically, the smell of roses. Smells
can be very strong memory cues, and in this case, the scientists
used the smell of roses to try to reactivate memory of the card
location task during sleep. So, they monitored the participants’
brain waves while they were sleeping and then presented the smell
of roses during different periods of sleep.

•• Amazingly, presenting the smell of roses during non-REM, slow-

wave sleep improved their memory performance the next day. And
the effect was quite specific: Presenting the smell when the subjects
were still awake didn’t help their retention. Likewise, presenting
the smell during REM sleep didn’t improve their performance.
Finally, presenting the smell during slow-wave sleep only helped
people who had smelled the roses when they were learning the
task; it didn’t make any difference in people who hadn’t smelled
anything during learning.

•• This was not just a correlation. This experiment demonstrated that

smelling roses during slow-wave sleep caused improved memory
the next day. And a natural explanation is that the smell evoked
the memory of the cards during sleep and that reactivating that
memory strengthened it.

•• In a separate experiment, the scientists

used functional neuroimaging to examine
brain activity when people were exposed
to smells that had been presented
during previous learning. They
measured neural activity during
both slow-wave sleep and when

206 The Learning Brain

 people were awake, and they found that presenting the smell of
roses during slow-wave sleep led to activation in the hippocampus.
In fact, hippocampus activity was stronger during sleep than it was
when participants were awake.

•• The hippocampus is strongly associated with explicit, declarative

learning and memory. So, the results suggest that the smell was
indeed triggering reactivation of the previously learned memories
during slow-wave sleep. It also strengthened those memories the
next day.

Theories of Sleep’s Role in

Memory Consolidation
•• Two of the most popular theories about how the role of sleep
in memory consolidation works at a neural level are the system
consolidation theory and the synaptic homeostasis theory.

•• Explicit memories might depend on the hippocampus initially but

gradually get consolidated into the cerebral cortex over time. We
want to be able to learn new things quickly, but we don’t want the
new learning to interfere with our previous memories. We want
our memories to be plastic so that we can learn new things, but
we also want them to be stable.

•• The brain’s solution is to have 2 separate memory systems: one

for fast, immediate learning and another for slower learning
that integrates new information with old. The hippocampus is
assumed to be the fast learning system that can quickly learn new
relationships and associations. And that allows us to be plastic
and flexible.

Lecture 21 | How Sleep Affects Learning 207

 •• But consolidation of those memories into the cortex is slower and
more gradual, interleaving the new memories with older existing
memories. That allows us to add new information without
overwriting older information, but it requires the hippocampus
to reactivate new memories so that they can be incorporated
into the existing memory networks without overwriting what’s
already there.

•• And that’s where sleep comes in, according to the system

consolidation theory. During periods of slow-wave sleep, new
memories are repeatedly reactivated, and those reactivations allow
those memories to be gradually incorporated into the cerebral
cortex without overwriting previous memories.

•• According to the other major theory, synaptic homeostasis,

learning new information when you’re awake tends to increase
the strength of synapses between neurons, and sleep serves to
downscale synaptic strength back to a more standard level.

•• But wouldn’t downscaling synaptic strengths simply undermine

the previous learning and cause you to forget it? The answer is
no, and here’s why: Learning tends to strengthen a relatively small
proportion of synapses and leaves most others unchanged. But
downscaling during sleep is assumed to apply to all the synapses
in a neural circuit. As a result, the strong synapses aren’t affected
very much, while the weak synapses get significantly reduced.
Because the strong synapses are presumably the most reliable,
while the weak synapses are noisier, the overall effect is to improve
the quality of the memory representation.

•• The synaptic homeostasis theory and the system consolidation

theory are not mutually exclusive. In fact, many sleep scientists
believe both mechanisms play a role in the consolidation of
memories during sleep.

208 The Learning Brain

•• First, reactivation of memories during slow-wave sleep leads to the
strengthening of synapses in the cerebral cortex and the integration
of the new memory with preexisting memories. That’s the system
consolidation part. But then those synapses are downscaled to
maintain a homeostatic equilibrium and to improve the signal-to-
noise ratio in the memory. That’s the synaptic homeostasis part.

•• It’s even possible that different periods of sleep are associated

with the 2 functions, with slow-wave, non-REM sleep being
more important for system consolidation and REM sleep being
more important for synaptic homeostasis. Perhaps that’s even
the reason we have different types of sleep, although at present
that’s speculation.

SUGGESTED READING

Hobson, Dreaming.
Stickgold and Walker, The Neuroscience of Sleep.
Walker, Why We Sleep.

QUESTIONS TO CONSIDER

1 How significantly is your learning affected by your sleeping habits?

Do you think that you learn better after a good night’s sleep?

2 Are there any specific steps you can take to improve the quality of
your sleep?

Lecture 21 | How Sleep Affects Learning 209

 LECTURE 22
How Aging Affects Learning

T  E COMMON BELIEF THAT AGING IS

H
INEVITABLY ASSOCIATED WITH A
significant deterioration in mental abilities,
including learning and memory, is wrong. In
the absence of disease, such as Alzheimer’s,
the mental impairments associated with
aging are typically restricted to a few specific
cognitive processes, and most other aspects
of mental life don’t decline much. This lecture
will explore what scientists are discovering
about the effects of aging on learning and
memory—as well as what we can do about
it. As you will discover, there are scientifically
proven steps we can take to help us keep our
mental abilities sharp as we get older. 

210 The Learning Brain

How Learning and Memory Change as We Age
•• Aging has dramatically different effects on fluid intelligence
compared to crystallized intelligence.

•• Recall that fluid intelligence refers to cognitive abilities that

are relatively independent of what you know. Can you think
logically, identify patterns, and solve novel problems quickly
and effectively? Brain teasers and puzzles that depend more on
your ability to think in creative ways than on how much you
know emphasize fluid processing ability.

•• Recall that crystallized intelligence refers to cognitive abilities

that depend critically on your knowledge, experience, and
acquired skills. For example, crossword puzzles place significant
demands on crystallized intelligence because they depend a lot
on how much you know about the world.

•• Around the year 2000, Denise Park and her team gave people tests
that tap into fluid processing ability—such as asking people to try
to remember random words or line drawings or testing how fast
they could perform tasks that didn’t depend on knowledge—and
generated plots of people’s performance as a function of their age.

•• These same declines were seen in episodic memory and

working memory.

•• Recall that episodic memory refers to long-term, explicit memory

for personal episodes that you have experienced. Episodic
memories are remembered from a first-person perspective, as if
you’re reliving them through your own eyes. They’re also tied
to a specific time and place. Unfortunately, Park found that
our ability to store and retrieve episodic memories tends to get
a little bit worse with each passing decade.

Lecture 22 | How Aging Affects Learning 211

 •• Recall that working memory is our ability to store information
for up to a minute or so and then retrieve that information
when we need it later. For example, if you have to add some
multidigit numbers in your head, you have to constantly store
the results of each column and then retrieve that information
later to finish the solution. Park found that working memory
also tends to decline with age.

•• It’s important to keep in mind that subtle age-related declines in

episodic and working memory are normal. Many older people
are worried about Alzheimer’s disease, so when they notice that
their ability to learn and remember isn’t quite as good as it used
to be, they naturally worry that it means they’re going to develop
dementia. But occasional lapses in memory are normal as we age
and are caused by completely different mechanisms than the much
more severe symptoms of Alzheimer’s disease.

•• Furthermore, quite a few studies demonstrate that given enough

time, older people can perform fluid processing tasks just as well
as younger people. In fact, Timothy Salthouse analyzed age-related
declines in fluid intelligence as a function of processing speed and

212 The Learning Brain

 found that once he controlled for how fast people were, the age
differences vanished. So, it’s probably more about speed than it
is about ability.

•• It’s also important to note that there are substantial differences

between different people. Some people experience a lot of learning
and memory problems as they get older, but other people experience
very few. Most scientific studies of aging average across people of a
similar age, and when this is done, the average amount of decline
is found. But keep in mind that roughly half the people in the
sample are doing better than the average.

•• So, storing information in, and retrieving information from, both

episodic memory and working memory often (but not always)
decline a little as we get older. But it turns out that there are other
types of learning and memory that don’t typically decline with
age—and may even improve. For example, consider crystallized
intelligence, which depends critically on your knowledge,
experience, and acquired skills.

•• Park’s group also tested this kind of processing ability, using tasks
that tested things like world knowledge and vocabulary. The
results on these tasks looked very different than the results for
fluid intelligence. Here, the older groups tended to perform better
than the younger groups.

•• Semantic memory, or your knowledge base of everything you

know about the world, is a type of long-term, explicit memory,
just like episodic memory; that is, you can consciously bring to
mind things you know from your semantic memory, and you can
verbalize those memories and talk about them.

•• But unlike episodic memory, semantic memories are not

remembered from a first-person perspective and are not tied to a
specific time and place. And in the absence of disease, our semantic

Lecture 22 | How Aging Affects Learning 213

 memory seems to improve with age. Our vocabulary gets bigger;
our knowledge of the world expands. As a result, our crystallized
intelligence remains stable or even improves.

•• In addition to crystallized intelligence being preserved with age,

your memory for skills and habits is also pretty stable as you get
older. Knowing how to read, how to speak a language, and how
to play an instrument are examples of procedural memory. And
studies have repeatedly found that procedural memory doesn’t
normally decline as we get older.

How the Brain Changes as We Age

•• One important line of work in the area of how our brain changes
as we age has been conducted by Naftali Raz and his colleagues,
who examined MRI scans from the brains of a large number of
healthy adults, ranging from 20 to 80 years old. Using the MRI
scans, they painstakingly traced specific brain structures in each
individual participant and measured the volume. Then, they
plotted the volume of different structures as a function of the age
of the participant. Doing so allowed them to determine whether a
given brain region shrinks with age or whether its volume remains
relatively constant or even grows.

•• Raz’s team found 2 main brain regions that tend to shrink as we

age: the hippocampus and the prefrontal cortex.

•• Recall that the hippocampus is the region that is most associated

with amnesia (remember Henry Molaison, or patient HM).
When Raz and his colleagues plotted the size of the hippocampus
as a function of age, they found that the hippocampus tends to
shrink a little bit with each passing decade. And that was true

214 The Learning Brain

 both when comparing old people to young people and when
comparing volume estimates in the same individual 5 years later.
This study suggests that one reason why episodic memory tends
to decline with age may be that the hippocampus is shrinking.

•• Recall that the prefrontal cortex is kind of like the CEO of the
brain, controlling what we pay attention to, managing goals, and
delegating responsibilities to other brain processes. In particular,
the prefrontal cortex is hypothesized to play a central role in
working memory. When Raz and his colleagues plotted prefrontal
cortex volume versus age, they found that the prefrontal cortex
tends to get a little bit smaller as we get older. If the prefrontal
cortex is shrinking as we get older and working memory depends
on the prefrontal cortex, then we can understand why working
memory might tend to get worse as we age.

Lecture 22 | How Aging Affects Learning 215

 Preserving Learning and Memory Abilities
•• There are steps that we can take to help our brains age a little more
gracefully and perhaps preserve our learning and memory abilities
as long as possible. Four strategies that have fairly strong scientific
support are eating right, staying active, maintaining relationships,
and reducing stress. You can remember these 4 strategies with
the acronym EARS, where E stands for eating right, A stands for
activity, R stands for relationships, and S stands for stress reduction.

1 Eating right. There is quite a bit of evidence that eating a

healthy diet with fewer calories can extend lifespan. For
example, scientists have investigated the effects of eating a
Mediterranean diet, consisting of a lot of fruits and vegetables,
olive oil, fish, nuts, and legumes; whole grains; and moderate
amounts of alcohol. One study found that the incidence of
heart attack, stroke, and death were significantly lower in
people in the Mediterranean diet groups than people in the
control group and that eating a Mediterranean diet improved
cognitive functions.

Studies have found that eating a Mediterranean diet reduces your

risk of developing Alzheimer’s disease, maybe by as much as 50%.

216 The Learning Brain

A number of studies have found
that learning a second language
is associated with better cognition
in later life, and others have
found that learning a complex new
hobby, such as digital photography or
quilting, also has cognitive benefits.

2 Activity. In a study of the effects of

physical activity on the mind and brain,
Kirk Erickson and Arthur Kramer recruited more
than 120 older people and randomly assigned half of them to a
year of aerobic exercise, while the other half did stretching and
toning but didn’t participate in aerobic exercise. As expected
with normal aging, brain scans of the hippocampus looked
a little smaller in the control group after a year than it did
before the experiment. But in the other group, both the left
and the right hippocampus actually got a little larger after a
year of aerobic exercise. Furthermore, changes in the size of
the hippocampus were associated with changes in memory
performance. Vigorous mental activity can also help us age
more gracefully.

3 Relationships. There is a growing body of evidence that good

social relationships can help us stay sharp as we age. In a famous
study, one group of older participants worked with 5 to 7 other
participants on a long-term project that involved team-based
problem-solving, while another group of older participants
continued life as usual without that regular social interaction.
Scientists found that the people who had been assigned to
the social interaction group performed better on all tests of
cognitive performance, including processing speed, working
memory, reasoning, and spatial processing.

Lecture 22 | How Aging Affects Learning 217

 4 Stress reduction. Recall that when we experience stress, a
biological system called the HPA axis is activated, triggering a
fight-or-flight response: You get a jolt of adrenaline, your mouth
goes dry, your pupils dilate, your sensations are heightened,
and your heart starts to race. This response is very helpful if
you’re dealing with a life-threatening situation, but many people
chronically activate the HPA axis and the associated fight-or-
flight response from daily stress, and this can lead to disease
and actually increase the speed of aging.

It’s not always easy to reduce stress, especially if you’re

constantly faced with deadlines or other obligations.
But one method that has been scientifically
demonstrated to reduce stress is meditation. In
particular, meditation has been found to lower
heart rate, reduce blood pressure, and lower levels of
bad cholesterol.
It also turns out that 3 of the most effective
approaches to reducing stress are the
same methods that help us age more
gracefully: eating right, staying
active, and maintaining social
relationships.

218 The Learning Brain

 SUGGESTED READING

Buettner, The Blue Zone.

Park and Schwarz, Cognitive Aging.
Salthouse, Major Issues in Cognitive Aging.

QUESTIONS TO CONSIDER

1 Have you noticed any changes in your cognitive abilities as you

age? If so, what changes have the biggest impact on your daily life?

2 Considering the acronym EARS, are there any changes you’d like
to make in your life that might help you age more gracefully?

Lecture 22 | How Aging Affects Learning 219

 LECTURE 23
Dyslexia and Other
Learning Disabilities

T  IS LECTURE WILL ADDRESS A TOPIC THAT

H
IS CRITICALLY IMPORTANT BUT THAT IS
surrounded by many misconceptions—
learning disabilities. Most definitions
of the term “learning disability” arouse
controversy, but the education field tends
to use the term “specific learning disorder”
(SLD). The word “specific” makes clear that
the learning problem is typically limited to
a specific domain, such as learning to read
or write, and would not extend to learning
other things, such as how to ride a bike
or tie a shoelace. This lecture will focus
on dyslexia, including the cognitive and
neural mechanisms that underlie it and how
it’s treated. 

220 The Learning Brain

 Specific Learning Disorders
•• The key feature of a specific learning disorder (SLD) is an unusual
difficulty in learning some academic domain that cannot be
explained by the child’s intellect, educational opportunities,
sensory abilities, motivation, or emotional state.

•• There is a mistaken belief that SLDs reflect low intellect or a

lack of motivation. In fact, the learning problem is present
despite normal intelligence. It’s a specific problem in learning a
particular academic topic and doesn’t imply that the child will
have intellectual difficulties anywhere else.

•• And perhaps because of that discrepancy between their intellect

and academic achievement, many people assume that children
with an SLD must be lazy. After all, if it’s not a problem with their
intellect, or their vision, or their home life, or some other obvious
factor, then it’s natural to attribute the problem to a lack of effort.

•• And although repeated failure in an academic subject can

eventually lead to a kind of learned helplessness and to giving up,
that’s not usually the cause of the problem initially. In fact, many
children with an SLD struggle in their learning despite working
very hard, maybe even harder than most other children.

•• The implication is that SLDs are caused by something

biological in the student’s brain rather than something in their
external environment.

•• Another common misconception is that there’s a clear dividing

line between children with SLDs and children without SLDs. The
truth is that SLDs exist on a continuum, and the threshold for
what constitutes an SLD is actually somewhat arbitrary.

Lecture 23 | Dyslexia and Other Learning Disabilities 221

 •• In some cases, the diagnosis will
As a child, Steven be easy. If a child has a very severe
Spielberg had a very problem learning to read that is
difficult time with completely out of line with his
reading, and to this day, or her intellectual ability, then it
reading still requires a lot might be a very obvious diagnosis.
of time and effort and is Conversely, you’ll also find many
much more difficult than children whose learning seems
it is for most other adults. totally normal and who obviously
don’t have an SLD.
But his job requires
him to do quite a bit •• But the problem is you’ll find
of reading, especially children at every point between
of scripts. Spielberg is those extremes. You’ll find children
now one of the most who struggle a little more than
successful movie directors average but for whom the problem
in history, so he has is actually pretty negligible. You’ll
obviously learned to also find children with a problem
that could be considered minor.
cope with his disability
You’ll find others for whom the
extremely well. problem is moderate and still
SLDs can come in a others for whom it’s significant.
number of different
forms. Some children •• So, where do you draw the line
have difficulty learning when the line you draw will
to write, which is be somewhat arbitrary? Many
scientists actually choose not to
often referred to as
draw a line. Instead, they study
dysgraphia. Children reading ability or math ability—
with dysgraphia have or whatever it is—as a continuous
problems producing variable. Then, they try to explain
legible handwriting and why some children have a more or
often make an unusual less severe disability, rather than
number of spelling errors. just treating it as a binary, black-
or-white outcome.

222 The Learning Brain

•• One final misconception is the idea that learning disabilities are
a phase that children go through and that they’ll grow out of. In
fact, the scientific evidence shows that SLDs tend to persist into
adulthood, which makes sense given their biological origin. But
that doesn’t mean that people with SLDs can’t adapt and learn to
cope with them. They can.

•• Another common form of SLD is dyscalculia, which refers to a

difficulty learning math. Students with dyscalculia have much
more difficulty learning and using basic arithmetic operations
than other children their age.

•• Another common form of learning disability is called specific

language impairment (SLI). Children with SLI exhibit delayed
language development in the absence of hearing loss or other
developmental delays. SLI is also called developmental language
disorder, language delay, or developmental dysphasia.

Dyslexia
•• Perhaps the most common SLD, and also the best known, is
dyslexia. In fact, some reports estimate that as many as 80% of
children with a learning disability suffer from dyslexia. Dyslexia
has also received more attention from scientists than any other
learning disability, and a great deal has been learned about what’s
going on.

•• Dyslexia is characterized by a difficulty with recognizing or

decoding written words that cannot be attributed to poor vision,
low intelligence, or lack of education. These problems make it very
difficult for children with dyslexia to read.

Lecture 23 | Dyslexia and Other Learning Disabilities 223

 •• But reading involves more than just decoding written words;
it also requires the ability to comprehend what sequences of
decoded words mean. And although some classification systems
include impairments in either process, most scientists define
dyslexia as a problem with written word decoding—not language
comprehension. For example, most dyslexics can understand
spoken language just fine; their problem is in figuring out what
words the letters on the page represent.

•• While the key symptom of dyslexia is an unusual difficulty in

recognizing individual written words, both in terms of speed and
accuracy, there are many other associated symptoms. Children
with dyslexia typically have trouble sounding out words. They will
often mispronounce words and confuse one word for another that
has a similar spelling. They may even complain of feeling sick or
dizzy when they read.

Scientists and educators

refer to the difficulty
of learning to read as
developmental dyslexia.
This is distinguished from
acquired dyslexias, which
can arise from brain damage
in adulthood. The shorter
term, “dyslexia,” will be used
to refer to developmental
dyslexia in this lecture.

224 The Learning Brain

Some dyslexics report feeling as if letters within words jump around
when they’re trying to read. If you want to experience what this is
like, visit Victor Widell’s web-based simulation:
http://geon.github.io/programming/2016/03/03/dsxyliea.

•• Because reading comprehension depends on fast and accurate word

recognition, comprehension also naturally suffers. Although most
dyslexic children can and do eventually learn to read, they typically
continue to read more slowly than other people, even as adults.

•• Dyslexia is usually first noticed in school, but unfortunately it is

often overlooked and attributed to laziness or stupidity instead. In
fact, dyslexic children often mistakenly conclude that their reading
problems mean that they must be dumb, and problems with self-
esteem are extremely common.

•• Like most other learning disabilities, dyslexia represents the end

of a continuum, so categorizing someone as dyslexic is not always
clear-cut. Perhaps as a result, estimates of how common dyslexia
is vary. But most scientists believe that somewhere between 5%
and 10% of school-aged children could be considered dyslexic.

Dyslexia’s Cognitive Mechanisms

•• In the 1920s, those who studied dyslexia assumed that it was a
visual problem, and children who suffered from dyslexia were
often prescribed some sort of visual training as a treatment. The
belief that dyslexia is a visual deficit persists to this day among
some people in the general population, but it’s wrong, at least in
most cases.

Lecture 23 | Dyslexia and Other Learning Disabilities 225

 •• The evidence suggests that the underlying problem in dyslexia is
usually in the processing of phonology, which refers to the basic,
atomic sounds from which all words are composed. Those sounds
are called phonemes, and they differ from language to language.
For example, there are a total of 44 different phonemes in English,
and every word in the language is made up of a subset of those
sounds in a specific order.

•• Being able to break down a spoken word into its component

phonemes becomes less important as you get skilled at recognizing
hundreds of thousands of words at a glance, but it’s a crucial skill
when you’re learning to read.

•• When you’re first starting to read, written words are just unfamiliar
strings of letters, and you have to go through a sequence of steps
to figure out what word those letters correspond to. And a critical
step in that process is breaking down the sound of the word into
its component phonemes.

•• Most children with dyslexia exhibit a deficit in phonological

awareness; that is, they’re unaware of the phonological structure
of the words they hear. They can’t figure out that “buh” is the first
sound in “bat” and that “aa” is the first sound in “after.” And that
lack of phonological awareness makes it very difficult to sound out
written words and figure out what they are.

Dyslexia’s Neural Mechanisms

•• Neuroimaging techniques, such as positron-emission tomography
and functional MRI, have made it possible to estimate neural
activity in human beings while they’re performing different

226 The Learning Brain

 cognitive tasks. These methods have revolutionized the study
of many cognitive functions, including the study of reading
and dyslexia.

•• Reading involves a variety of cognitive processes. It involves

phonological mechanisms to process sounds as well as other
linguistic mechanisms to process the grammar and the meaning
of what is being read. And the mechanisms involved in processing
phonology, grammar, and meaning are also needed for spoken
language, so they’re not unique to reading.

•• But reading also requires visual mechanisms to recognize the

letters and words, and those mechanisms aren’t required for spoken
language. They are unique to reading.

•• Neuroimaging studies of reading find neural activity in brain

networks involved in spoken language and in networks involved
in visual recognition. Specifically, when people are speaking or
listening to spoken language, they tend to activate 2 major areas in
the left hemisphere that are named after the famous neurologists
who first identified them.

•• Broca’s area is on the left side

of your brain, toward the
front and the bottom,
a few inches behind
your left eye. Broca’s
area is named after
the French neurologist
Paul Broca, and it’s
critically involved in speech
production and in processing
grammar and phonology.

Lecture 23 | Dyslexia and Other Learning Disabilities 227

 •• Wernicke’s area, named after the German neurologist Carl
Wernicke, is also on the left side of the brain a few inches above
your left ear. Evidence suggests that Wernicke’s area and nearby
brain regions are crucially involved in processing the meaning
of speech as well as phonology.

•• Both of these brain regions are activated during reading, too, even
though you’re not normally hearing or speaking when you read.
But in addition to these regions, reading also activates regions in
the visual cortex, such as the visual word form area, which is on
the bottom surface of the left hemisphere of brain, toward the
back. It’s critically important in the visual recognition of letters
and written words.

•• Numerous neuroimaging studies of dyslexics have repeatedly

found that dyslexics exhibit less neural activity in Wernicke’s area
and in the visual cortex than normal readers do. Some studies
have also reported the same thing in Broca’s area, although those
results are not as consistent. Furthermore, structural neuroimaging
studies have found that these same brain regions are physically
smaller in dyslexics compared with normal readers.

•• Perhaps most interesting, a number of studies have found evidence

that the connections between these brain areas don’t work as
efficiently as they should in dyslexia. In particular, the visual
areas that recognize words and the language areas that process
phonology and meaning don’t seem to be able to communicate
with each other as effectively as they need to.

•• Obviously, that kind of communication is critical to reading. After

all, those are presumably the same connections that the brain uses
to associate the sight of a word with both its sound and its meaning.
So, if those connections aren’t working as well as they should, then
it makes sense that reading would be disrupted. And it is.

228 The Learning Brain

What can be done to help a dyslexic
child learn to read?
Schools in the US are legally obligated
to implement an Individualized Education
Program for any child diagnosed with dyslexia.
The child will first be tested to identify specific areas where
remediation is needed, and then a tailored education plan is
developed to address those needs and maximize the child’s
chance of academic success.
In the case of dyslexia, that plan might include intense training
in recognizing the phonemes that make up words and in
mapping those phonemes onto letters. Then, the child might
get substantial practice at sounding out words to build up
reading speed and accuracy. Another common approach is to
try to build up a vocabulary of frequent words that the child
learns to recognize relatively quickly and accurately. Finally, the
plan might include training on comprehending the meaning of
sentences and paragraphs that have been read.
Although this kind of intense, individualized training can
help, it’s not a cure, and reading problems tend to persist
into adulthood.
Nevertheless, most dyslexics can significantly improve their
reading, especially if the problem is detected early. In particular,
children who get help with their reading starting in kindergarten
or first grade can often improve enough to succeed in
elementary school and high school, while children who start
getting help later have a more difficult time and may not be able
to catch up.

Lecture 23 | Dyslexia and Other Learning Disabilities 229

 SUGGESTED READING

Fletcher, Learning Disabilities.

Pennington, Diagnosing Learning Disorders.
Shaywitz, Overcoming Dyslexia.

QUESTIONS TO CONSIDER

1 What do you think is the most common misconception about

dyslexia in the general population?

2 Many elementary school teachers aren’t experts on learning

disabilities and can’t devote substantial amounts of time to
individual students. How do you think the US educational system
should handle this problem?

230 The Learning Brain

 LECTURE 24
Optimizing Your Learning

T  IS FINAL LECTURE WILL DRAW ON WHAT

H
YOU’VE LEARNED IN THIS COURSE AND
offer 5 suggestions about what you can do
to learn more effectively. By implementing
some or all of these suggestions, you
will hopefully be able to get even more
out of your efforts to learn and grow in
the future. 

Learning has a lot more to do with what the student does than
with what the teacher does. As the great psychologist Herbert
Simon put it, “Learning results from what the student does and
thinks, and only from what the student does and thinks. The
teacher can advance learning only by influencing what the student
does to learn.”

Lecture 24 | Optimizing Your Learning 231

 1. Adopt a positive mindset.
•• Suppose that you want to learn something complex and
challenging, such as a foreign language or a musical instrument.
Learning something so complex would be very rewarding, yet few
people actually try. There are undoubtedly several reasons why
this is the case, but one thing that stops many people is simply the
belief that they can’t do it—that it’s just too difficult and beyond
their ability.

•• The truth is that the human brain is the most amazing,

sophisticated, and powerful mechanism in the universe—and you
have one! Your brain is always learning, whether you’re trying to
learn or not. It has already learned a bewildering array of complex
motor skills, including how to stand up, walk, and run. It has
already learned to read, write, and do math. And perhaps most
amazingly, it has already mastered a natural language, which is
arguably more complicated than anything else you would ever
want to learn.

•• Believing that you can learn something challenging is the first—

and, in some ways, the most important—step. Without that belief,
you might not even try. You’d also be likely to give up at the first
sign of difficulty.

•• In addition, it’s important to recognize that your learning is your

responsibility. Your learning is up to you.

•• What do you do when you encounter an obstacle to your learning?

The easy way out would be to say there’s nothing you can do. For
example, you could blame the teacher of a course and attribute
the problem to poor instruction. But if you do that, then you’re
not going to learn the material.

232 The Learning Brain

•• The alternative approach is to ask yourself what you can do to
overcome the obstacle. You could try to find another resource, such
as a book or website, that might help you learn the information.
Or you could discuss the material with a friend who’s an expert on
the topic. These are examples of taking control of your learning.

•• And this is how the most effective learners keep learning. They
don’t let obstacles get in the way; instead, they find ways to
overcome difficulties and continue their learning.

•• You need to adopt a positive mindset whenever you’re trying to

learn something new. You need to see challenges and obstacles as
opportunities to grow. After all, it’s when we stretch ourselves and
get outside of our comfort zone that we grow the most. And that
kind of positive attitude—referred to as a growth mindset—is
something we can get better and better at.

2. Be strategic and deliberate

about your learning.
•• Not all approaches to learning are equal. When you’re trying to
learn a new skill, deliberate, challenging practice is more effective
than mindlessly practicing the skill. If we want to maximize our
learning, then we should be thoughtful about how we spend our
time and try to do things that will maximize our learning.

•• To do that in practice, we can set specific goals for learning. Limit

yourself to 2 or 3 goals at a time; any more than that could quickly
become overwhelming and ultimately discouraging. You want
your goals to be challenging so that they will stretch you and
force you to learn. But they should also be specific and achievable
with concerted effort.

Lecture 24 | Optimizing Your Learning 233

 •• Setting challenging, but manageable, learning goals for yourself
can be very effective. You might want to write your goals down and
establish a target date by which you’d like to try to accomplish your
goals. You could even identify a reward that you’ll give yourself
if you accomplish your goal by the deadline. That kind of simple
positive feedback can make you even more eager to accomplish
the next goal on your list.

•• Another important way to make Cramming leads to less

your learning more strategic is to long-term learning than
identify holes in what you want spacing the same amount
to learn and then systematically of studying over time.
fill them in. Although this
makes a lot of sense, it’s not Randomizing and
always our natural tendency. interleaving what you
After all, it’s tempting to work work on is better than
on what we’re already good learning about one topic
at. Doing so provides positive and then switching to
reinforcement and gives us a another topic in blocks.
sense of accomplishment and
satisfaction. But it also limits Relating new information
how much we learn and grow. to information you already
know can also help you
•• Furthermore, difficulties during learn it more effectively.
practice are often desirable when
it comes to long-term learning,
even though they may hurt short-term performance. Strategies such
as spacing out your learning over time, deliberately challenging
yourself with tasks that are just outside your current ability, and
interleaving learning about different topics will all tend to make
learning more difficult. But they also tend to produce the best
results in the long run.

234 The Learning Brain

 3. Learn actively rather than passively.
•• We learn better when we are actively engaged in processing
information rather than passively encoding it. For example,
rereading and highlighting—which are more passive study
strategies—aren’t nearly as effective as generating explanations
and especially testing yourself, which are more active strategies.

•• You may have experienced the difference between active learning

and passive learning in the work environment. As many employers
will tell you, there’s a huge difference between so-called book
learning and real-world, on-the-job training.

•• But it’s not just the work world where you can see the advantages
of active learning. Imagine trying to learn algebra without doing
any practice problems. It just wouldn’t work. You can read the
textbook over and over, but until you’ve solved a bunch of actual
problems, you’ll never master algebra.

•• How can we be more active in our learning so that we can get the
most out of it? One idea is to intentionally ask questions while
we’re learning. Suppose that you want to learn about gardening,
so you’re reading a book on the topic. One excellent way to get
the most out of your reading is to try to come up with a set of key
questions that the book answers.

•• For example, imagine that you have to make up a test on the

material in that book. What questions would you include? This
exercise can be extremely helpful. For one thing, it forces you to
pay careful attention to what you’re reading and to actively engage
with it. It also requires you to identify the key ideas so that you
can ask questions that tap into the important concepts rather than
the trivial details.

Lecture 24 | Optimizing Your Learning 235

 •• Another way to make your learning more active and less passive is
to test yourself. Evidence shows that taking tests on material you’ve
studied is one of the best ways to learn that material. In particular,
testing yourself is much better than rereading or highlighting. One
reason may be because testing yourself gives you practice retrieving
information from memory, and that’s ultimately what you want to
be able to do in the future.

•• But where are you going to get the test? You’ll occasionally
run across a book that includes some questions, but that’s the
exception, not the rule. But if you’ve adopted the previous
suggestion and made up a quiz on the material you’re trying to
learn, then you’re all set! Just wait until the next day or week and
then try taking the quiz you made up. The exercise of taking that
test will help consolidate what you’ve learned. And if there are
questions you can’t answer, then that tells you exactly what you
might want to review.

•• A final practical way to make your learning more active and less
passive is to try to teach others what you’re learning. There’s no
better way to expose holes in your understanding than to try to
explain something to other people.

•• Trying to teach also forces you to engage deeply with the

information. You have to figure out what information you need
to convey, and you need to come up with a coherent story line in
which each idea naturally follows from the previous ideas.

•• Preparing to teach information also has a very strong motivational

component. Basically, you don’t want to look stupid. And that
desire can motivate you to go over the material carefully and
double-check the accuracy of what you’re planning to say.

236 The Learning Brain

 4. Identify good sources of information
that will challenge you.
•• Computer scientists have a saying: Garbage in, garbage out. It
means that a computer program is only as good as the data that
you give it as input. If the program is correct and you give it reliable
data as input, then you should get reliable answers as output. But if
you give it unreliable, garbage data as input, then the answers you
get out will also be garbage—even if the program itself is correct.

•• The same goes for the neural network inside your skull. If you train
your brain on information that is unreliable or biased, it will learn
generalizations that may be unreliable or flat-out wrong.

•• So, if you want to optimize your learning, then it’s crucial to give
your neural network information that is as reliable and unbiased
as possible. But how can we do that?

•• Consider the source. Who is providing the information, and

what are his or her qualifications? Is he or she actually trained
in the field that you’re trying to learn about, or is he or she just
a blogger with an opinion?

•• Consider the evidence. Does the source provide actual evidence

in support of their claims, or is it really just opinion? If there is
evidence, can you confirm its authenticity, and does it come from
credible sources? A quick Google search of an empirical claim
can often identify its source and authenticity (or lack thereof).

•• Be skeptical. There’s a lot of misinformation out there,

particularly on the internet. So, it pays to be skeptical and to
try to verify information using other sources before trusting
a source.

Lecture 24 | Optimizing Your Learning 237

 •• Seek out different points of view—particularly ones that are
different from your own and that will challenge you. Deeply
engaging with alternative points of view will stretch you and
help you learn significantly more than you could from a source
you already agree with.

Human beings have a natural tendency to seek out information that

is consistent with their current beliefs and to disregard information
that isn’t. But if you really want to learn and grow, you’re better off
learning about ideas that are different than your own.

5. Stay active, eat right,

and get enough sleep.
•• If you want to be the best learner you can be, then you need your
brain to be functioning as well as it possibly can. And the same
things that keep your body healthy also tend to keep your brain
healthy. Your body provides your brain with the nutrients and
oxygen that it needs to thrive, so if your body—particularly your
heart—is healthy, then your brain usually will be, too.

•• There are many things we can do to keep our brains healthy, but
let’s focus on 3 that scientific studies have identified as particularly
important: staying active, eating right, and getting enough sleep.

•• One way to stay active is by exercising your brain with

challenging, engaging intellectual stimulation. Think of your
brain like a muscle that needs exercise to stay strong and healthy;
the more you challenge yourself mentally, the healthier your

238 The Learning Brain

 brain will be. Research shows that activities such as mastering
a challenging hobby can improve brain function. There’s even
stronger evidence that staying active physically can also improve
brain function; it is perhaps the best way to keep your mind
sharp as you age.

•• Study after study has found that your diet not only influences
your health and longevity, but it also can have a dramatic impact
on your mental functions. For example, eating a relatively low-
calorie Mediterranean diet has not only been found to improve
heart health, but also to improve cognitive function and even
significantly reduce the risk of Alzheimer’s disease.

•• Sleep plays an important role in learning and memory, and good

sleep hygiene can help keep us mentally sharp. Try to get 7 to
8 hours of sleep a night. Try to go to bed and get up at about
the same time every day so that your body gets into a standard
cycle. Avoid alcohol or coffee for a few hours before bedtime.
Try to make your bedroom a peaceful haven that is as dark
and quiet as possible. Implementing just a few of these ideas
could help improve your sleep, which in turn could improve
your brain functions.

SUGGESTED READING

Duckworth, Grit.
Dweck, Mindset.
Ericsson and Pool, Peak.
Ratey and Hagerman, Spark.
Waldrop, “The Science of Teaching Science.”

Lecture 24 | Optimizing Your Learning 239

 QUESTIONS TO CONSIDER

1 Which of the 5 suggestions discussed in this lecture resonated the

most for you? Are there specific steps you could take to put the
suggestions into practice?

2 Are there any strategies that weren’t in this lecture but that you’ve
found helpful in optimizing your learning and memory?

240 The Learning Brain

Bibliography
Anderson, John R. “Acquisition of Cognitive Skill.” Psychological
Review 89, no. 4 (1982): 369–406. doi: 10.1037//0033-295x.89.4.369.

——— . Cognitive Skills and Their Acquisition. Hillsdale, NJ: L.

Earlbaum Associates, 1981.

——— . Learning and Memory: An Integrated Approach. 2nd ed. New

York: Wiley, 2000.

Baddeley, Alan D. Working Memory, Thought, and Action. Oxford

Psychology Series. New York: Oxford University Press, 2007.

Baddeley, Alan D., Michael W. Eysenck, and Mike Anderson.

Memory. New York: Psychology Press, 2009.

Bandura, Albert. Self-Efficacy: The Exercise of Control. New York: W.

H. Freeman, 1997.

Bransford, John, Ann L. Brown, and Rodney R. Cocking. How

People Learn: Brain, Mind, Experience, and School. Washington, DC:
National Academy Press, 2000.

Brown, Peter C., Henry L. Roediger, and Mark A. McDaniel. Make

It Stick: The Science of Successful Learning. Cambridge, MA: Belknap
Press, 2014.

Buettner, Dan. The Blue Zone: Lessons for Living Longer from the People
Who’ve Lived the Longest. Washington, DC: National Geographic,
2008.

Bibliography 241
 Burgess, N., E. A. Maguire, and J. O’Keefe. “The Human
Hippocampus and Spatial and Episodic Memory.” Neuron 35, no. 4
(2002): 625–641. doi: 10.1016/s0896-6273(02)00830-9.

Carey, Benedict. “Forget What You Know about Good Study Habits.”
The New York Times, September 16, 2010. http://www.nytimes.
com/2010/09/07/health/views/07mind.html.

——— . How We Learn: The Surprising Truth about When, Where,

and Why It Happens. New York: Random House, 2015.

Clark, Eve V. First Language Acquisition. 3rd ed. Cambridge University

Press, 2016.

——— . Language in Children. New York: Routledge, 2016.

Corkin, Suzanne. Permanent Present Tense: The Unforgettable Life of

the Amnesic Patient, H.M. New York: Basic Books, 2013.

Deci, E. L., and R. Flaste. Why We Do What We Do: Understanding

Self-Motivation. New York: Penguin Books, 1996.

Didau, David. What If Everything You Knew about Education Was

Wrong? Bancyfelin, Carmarthen, UK: Crown House Publishing,
2015.

Dittrich, Luke. “The Brain That Couldn’t Remember.” The New

York Times, August 3, 2016. https://www.nytimes.com/2016/08/07/
magazine/the-brain-that-couldnt-remember.html.

Duckworth, Angela. Grit: The Power of Passion and Perseverance. New

York: Scribner, 2016.

242 The Learning Brain

Dunlosky, J., K. A. Rawson, E. J. Marsh, M. J. Nathan, and D.
T. Willingham. “What Works, What Doesn’t.” Scientific American
Mind 24, no. 4 (2013): 46–53.

Dweck, Carol S. Mindset: The New Psychology of Success. New York:

Random House, 2006.

Egner, Tobias. The Wiley Handbook of Cognitive Control. Chichester,

West Sussex, UK: John Wiley & Sons, 2017.

Eichenbaum, Howard, Andy P. Yonelinas, and Charan Ranganath.

“The Medial Temporal Lobe and Recognition Memory.” Annual
Review of Neuroscience 30 (2007): 123–152.

Eichenbaum, Howard, and Neal J. Cohen. From Conditioning to

Conscious Recollection: Memory Systems of the Brain. Oxford Psychology
Series. New York: Oxford University Press, 2001.

Erickson, Carlton K. The Science of Addiction: From Neurobiology to

Treatment. New York: W. W. Norton & Company, 2007.

Ericsson, Anders, and Robert Pool. Peak: Secrets from the New Science
of Expertise. Boston: Houghton Mifflin Harcourt, 2016.

Fisher, Ronald P., and R. Edward Geiselman. Memory-Enhancing

Techniques for Investigative Interviewing: The Cognitive Interview.
Springfield, IL: Thomas Publishing Company, 1992.

Fitts, Paul Morris, and Michael I. Posner. Human Performance.

Belmont, CA: Brooks/Cole Publishing Company, 1967.

Fletcher, Jack. Learning Disabilities: From Identification to Intervention.

New York: Guilford Press, 2007.

Bibliography 243
 Foer, Joshua. Moonwalking with Einstein: The Art and Science of
Remembering Everything. New York: Penguin Press, 2011.

Fuster, Joaquin M. The Prefrontal Cortex. 5th ed. Boston: Elsevier/

AP, 2015.

Gathercole, Susan, and Tracy Packiam Alloway. Working Memory and

Learning: A Practical Guide for Teachers. London: SAGE Publications
Inc., 2008.

Hart, J., and M. A. Kraut. Neural Basis of Semantic Memory.

Cambridge: Cambridge University Press, 2007.

Hobson, J. Allan. Dreaming: A Very Short Introduction. Oxford:

Oxford University Press, 2005.

Jaušovec, Norbert, and Anja Pahor. Increasing Intelligence. London:

Academic Press, 2017.

Kandel, Eric R. In Search of Memory: The Emergence of a New Science

of Mind. New York: W. W. Norton & Company, 2006.

Klingberg, Torkel. The Overflowing Brain: Information Overload and

the Limits of Working Memory. New York: Oxford University Press,
2009.

——— . “Training and Plasticity of Working Memory.” Trends

in Cognitive Sciences 14, no. 7 (2010): 317–324. doi: 10.1016/j.
tics.2010.05.002.

Kuhar, Michael J. The Addicted Brain: Why We Abuse Drugs, Alcohol,

and Nicotine. Upper Saddle River, NJ: FT Press, 2011.

244 The Learning Brain

Kuhn, Cynthia, Scott Swartzwelder, Wilkie Wilson, Leigh Heather
Wilson, and Jeremy Foster. Buzzed: The Straight Facts about the Most
Used and Abused Drugs from Alcohol to Ecstasy. Rev. ed. New York:
W. W. Norton & Company, 2014.

Lewis, Marc. Memoirs of an Addicted Brain: A Neuroscientist Examines

His Former Life on Drugs. New York: PublicAffairs, 2012.

Lilienfeld, Scott O. 50 Great Myths of Popular Psychology: Shattering

Widespread Misconceptions about Human Behavior. Malden, MA:
Wiley-Blackwell, 2010.

Loftus, Elizabeth F. Eyewitness Testimony. Cambridge, MA: Harvard

University Press, 1979.

——— . “Planting Misinformation in the Human Mind: A 30-Year

Investigation of the Malleability of Memory.” Learning & Memory 12,
no. 4 (2005): 361–366. doi: 10.1101/lm.94705.

Loftus, Elizabeth F., and Katherine Ketcham. The Myth of Repressed

Memory: False Memories and Allegations of Sexual Abuse. New York:
St. Martin’s Press, 1994.

Markman, A. B., and B. H. Ross. “Category Use and Category

Learning.” Psychological Bulletin 129, no. 4 (2003): 592–613. doi:
10.1037/0033-2909.129.4.592.

Markman, Ellen M. Categorization and Naming in Children: Problems

of Induction. Cambridge, MA: MIT Press, 1989.

McClelland, J. L., B. L. McNaughton, and R. C. Oreilly. “Why There

Are Complementary Learning Systems in the Hippocampus and
Neocortex: Insights from the Successes and Failures of Connectionist
Models of Learning and Memory.” Psychological Review 102, no. 3
(1995): 419–457. doi: 10.1037/0033-295x.102.3.419.

Bibliography 245
 McEwen, Bruce S., and Elizabeth Norton Lasley. The End of Stress as
We Know It. Washington, DC: Joseph Henry Press, 2002.

Osaka, Naoyuki, Robert H. Logie, and Mark D’Esposito. The

Cognitive Neuroscience of Working Memory. New York: Oxford
University Press, 2007.

Park, Denise C., and Norbert Schwarz. Cognitive Aging: A Primer.

Philadelphia: Psychology Press, 2000.

Pashler, Harold, Mark McDaniel, Doug Rohrer, and Robert Bjork.

“Learning Styles: Concepts and Evidence.” Psychological Science
in the Public Interest 9, no. 3 (2008): 105–19. doi: 10.1111/j.1539-
6053.2009.01038.x.

Pauk, Walter. How to Study in College. New York: Houghton Mifflin,

1962.

Pennington, Bruce Franklin. Diagnosing Learning Disorders:

A Neuropsychological Framework. 2nd ed. New York: Guilford Press,
2009.

Pinker, Steven. The Language Instinct: How the Mind Creates Language.
New York: HarperPerennial ModernClassics, 2007.

Ratey, John J., and Eric Hagerman. Spark: The Revolutionary New
Science of Exercise and the Brain. New York: Little, Brown and
Company, 2008.

Reber, Arthur S. Implicit Learning and Tacit Knowledge: An Essay

on the Cognitive Unconscious. New York; Oxford [England]: Oxford
University Press; Clarendon Press, 1993.

Redish, A. David. Beyond the Cognitive Map: From Place Cells to

Episodic Memory. Cambridge, MA: MIT Press, 1999.

246 The Learning Brain

Roberts, A. C., Trevor W. Robbins, and Lawrence Weiskrantz. The
Prefrontal Cortex: Executive and Cognitive Functions. New York:
Oxford University Press, 1998.

Rosch, Eleanor, and Barbara B. Lloyd. Cognition and Categorization.

Hillsdale, NJ: L. Erlbaum Associates, 1978.

Salthouse, Timothy A. Major Issues in Cognitive Aging. New York:

Oxford University Press, 2010.

Sapolsky, Robert M. Why Zebras Don’t Get Ulcers. New York: Holt
Paperbacks, 2004.

Schacter, Daniel L. Searching for Memory: The Brain, the Mind, and
the Past. New York: BasicBooks, 1996.

——— . The Seven Sins of Memory: How the Mind Forgets and
Remembers. Boston: Houghton Mifflin, 2001.

Schmidt, Richard A., and Craig A. Wrisberg. Motor Learning and

Performance. 3rd ed. Champaign, IL: Human Kinetics, 2004.

Schmidt, Richard A., and Timothy Donald Lee. Motor Control and
Learning: A Behavioral Emphasis. 3rd ed. Champaign, IL: Human
Kinetics, 1999.

Schultz, W. “Predictive Reward Signal of Dopamine Neurons.”

Journal of Neurophysiology 80, no. 1 (1998): 1–27.

Shaywitz, Sally E. Overcoming Dyslexia: A New and Complete Science-

Based Program for Reading Problems at Any Level. Unabridged ed.
New York: Alfred Knopf, 2003.

Bibliography 247
 Squire, Larry R. “Memory Systems of the Brain: A Brief History and
Current Perspective.” Neurobiology of Learning and Memory 82, no.
3 (2004): 171–177. doi: 10.1016/j.nlm.2004.06.005.

——— . “The Legacy of Patient HM for Neuroscience.” Neuron 61,

no. 1 (2009): 6–9. doi: 10.1016/j.neuron.2008.12.023.

Squire, Larry R., and Eric R. Kandel. Memory: From Mind to

Molecules. 2nd ed. Greenwood Village, CO: Roberts & Company,
2009.

Stickgold, R., and Matthew P. Walker. The Neuroscience of Sleep.

Boston: Academic Press/Elsevier, 2009.

Terr, Lenore. Unchained Memories: True Stories of Traumatic Memories,

Lost and Found. New York: Basic Books, 1994.

Tulving, Endel. Elements of Episodic Memory. Oxford

[Oxfordshire]; New York: Clarendon Press; Oxford University Press,
1983.

——— . “Episodic Memory: From Mind to Brain.” Annual

Review of Psychology 53 (2002): 1–25. doi: 10.1146/annurev.
psych.53.100901.135114.

Waldrop, M. M. “The Science of Teaching Science.” Nature 523,

no. 7560 (2015): 272–274.

Walker, Matthew. Why We Sleep: Unlocking the Power of Sleep and

Dreams. New York: Scribner, 2017.

Wearing, Deborah. Forever Today. New York: Doubleday, 2005.

248 The Learning Brain

Image Credits
titles . . . . . . . . . . . . . . . . . . . . . . © TrishaMcmillan/iStock/Thinkstock
readings . . . . . . . . . . . . . . . . . . . . . . . . © annuker/iStock/Thinkstock
5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . © ISO/iStock/Thinkstock
5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . © Smitt/iStock/Thinkstock
8 . . . . . . . . . . . . . . . . . . . . . . . . . . The Teaching Company Collection
13 . . . . . . . . . . . . . . . . . . . © KatarzynaBialasiewicz/iStock/Thinkstock
23 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . © Chepko/iStock/Thinkstock
34 . . . . . . . . . . . . . . . . . . . . . . © johnnylemonseed/iStock/Thinkstock
35 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . © zager/iStock/Thinkstock
37 . . . . . . . . . . . . . . . . . . . © KatarzynaBialasiewicz/iStock/Thinkstock
38 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . © zabelin/iStock/Thinkstock
45 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . © moodboard/Thinkstock
46 . . . . . . . . . . . . . . . . . . . . . . . . © Michael Burrell/iStock/Thinkstock
51 . . . . . . . . . . . . . . . . . . . . . . . . © shopplaywood/iStock/Thinkstock
55 . . . . . . . . . . . . . . . . . . . . . . . . . . . © Dorling Kindersley/Thinkstock
61 . . . . . . . . . . . . . . . . . . . . . . . . . . © robophobic/iStock/Thinkstock
68 . . . . . . . . . . . . . . . . . . © monkeybusinessimages/iStock/Thinkstock
74 . . . . . . . . . . . . . . . . . . . . . . . . . . . . © Rohappy/iStock/Thinkstock
82 . . . . . . . . . . . . . . . . . . . . . . . . . . . © imtmphoto/iStock/Thinkstock
85 . . . . . . . . . . . . . . . . . . . . . . . . . © HannamariaH/iStock/Thinkstock
92 . . . . . . . . . . . . . . . . . . . . . . . . . . © Sementinov/iStock/Thinkstock
94 . . . . . . . . . . . . . . . . . . . . . . © Wavebreakmedia/iStock/Thinkstock
101 . . . . . . . . . . . . . . . . . . . © KatarzynaBialasiewicz/iStock/Thinkstock
103 . . . . . . . . . . . . . . . . . . . . . . . . . . . . © anna1311/iStock/Thinkstock
110 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . © hlsnow/iStock/Thinkstock
113 . . . . . . . . . . . . . . . . . . . . . . . . . . © den-belitsky/iStock/Thinkstock
131 . . . . . . . . . . . . . . . . . . . . . . . . © BrianAJackson/iStock/Thinkstock
136 . . . . . . . . . . . . . . . . . . . . . . . . . . . © cynoclub/iStock/Thinkstock
154 . . . . . . . . . . . . . . . . . . . . . . . . © Art-Of-Photo/iStock/Thinkstock
155 . . . . . . . . . . . . . . . . . . . . . . . . . . © Zoonar RF/Zoonar/Thinkstock
164 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . © wenht/iStock/Thinkstock
173 . . . . . . . . . . . . . . . . . . . . . © Wavebreakmedia/iStock/Thinkstock

Image Credits 249

 176 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . © HASLOO/Thinkstock
181 . . . . . . . . . . . . . . . . . . . . . . . . . . © DGLimages/iStock/Thinkstock
182 . . . . . . . . . . . . . . . . . . . . . . . . . © Tomwang112/iStock/Thinkstock
192 . . . . . . . . . . . . . . . . . . . . . © JOHNGOMEZPIX/iStock/Thinkstock
197 . . . . . . . . . . . . . . . . . . . . . . . . . . . © jambojam/iStock/Thinkstock
200 . . . . . . . . . . . . . . . . . . . . . . . . © sematadesign/iStock/Thinkstock
206 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . © LiuSol/iStock/Thinkstock
212 . . . . . . . . . . . . . . . . . . . . . . . . The Teaching Company Collection
215 . . . . . . . . . . . . . . . . . . . . . . . . The Teaching Company Collection
216 . . . . . . . . . . . . . . . . . . . . . . . . . . . . © kobeza/iStock/Thinkstock
217 . . . . . . . . . . . . . . . . . © monkeybusinessimages/iStock/Thinkstock
218 . . . . . . . . . . . . . . . . . . . . . © Jupiterimages/Stockbyte/Thinkstock
224 . . . . . . . . . . . . . . . . . . . . . . . . . . . . © kovalvs/iStock/Thinkstock
227 . . . . . . . . . . . . . . . . . . . . . . . . The Teaching Company Collection
229 . . . . . . . . . . . . . . . . . . . . . . . . . . . . © Mogala/iStock/Thinkstock

250 The Learning Brain