From Helplessness to Hope: The Seminal Career of Martin Seligman

Maier, Peterson, & Schwartz

1
From Helplessness to Hope: The Seminal Career of Martin Seligman

Steven F. Maier, Christopher Peterson, and Barry Schwartz
University of Colorado, University of Michigan, and Swarthmore College


This book explores a specific field of psychological research, but it also
celebrates the profound contributions to this field of Martin E. P. Seligman.
Therefore, the book blends the history of this research enterprise and Seligman's
own intellectual history. This chapter reviews the modest origins of the
phenomenon of "learned helplessness" in the animal laboratory, its extensions to
human beings (especially those displaying dramatic failures of adaptation), and
its eventual emergence as "learned optimism." The remainder of the book
documents two major themes. First, the insights arising out of research on
learned helplessness have been extended to almost every domain of modern
psychology. And second Seligman has played a significant role in almost all of
these extensions. In fact, this book makes a fitting tribute to the man whose
fingerprints appear on every chapter.

Although the research discussed in this book focuses on optimism and
hope, the research story does not begin there. Rather, it begins with the opposite
end of the pole—helplessness. As will become apparent Seligman is now a
strong proponent of the development of a positive psychology, but the historic,
intellectual seeds of the view that underlies this new emphasis are very much in
negative psychology. The critical first step in thinking that made this
development possible was an appreciation of the negative consequences of the
inability to control important environmental events. It is this inability that
produces the learned helplessness phenomenon.

---

Maier, Peterson, & Schwartz

2

The history of research on learned helplessness and learned optimism as
well as Seligman's own involvement in these areas reflects a large element of
chance. Furthermore, the development of research in this area also illustrates
two other important lessons in how science actually proceeds. First, it is often
difficult to predict at the outset where research will lead. Work on learned
helplessness began in the animal laboratory and for several years was directed at
deep theoretical issues in the psychology of learning and not at depression,
academic achievement, and other significant human phenomena. And second,
the history of learned helplessness research demonstrates the continuity between
basic and applied research in the way that it has moved effortlessly between
fundamental issues in learning, cognition, and motivation on the one hand, and
attempts to deal with problems of human adaptation and obstacles to the
achievement of human potential on the other.

Learned Helplessness in Animals

Early Experiments


Learned helplessness research, and Seligman's own work, began in the
mid-1960s in the animal learning laboratory of Richard L.Solomon at the
University of Pennsylvania. At that time the focus in the Solomon laboratory
was on the rigorous testing of a new theory designed to explain the occurrence
of avoidance learning. In avoidance learning, some warning signal (e.g., a light)
precedes the onset of an aversive stimulus (e.g., a shock) by a short period of
time (e.g., 10 s). A response by the animal (e.g., jumping a hurdle) after the
aversive event has started escapes the aversive stimulus. And a response during
the warning signal avoids the aversive event.

---

Maier, Peterson, & Schwartz

3

Animals readily learn to make avoidance responses in such experiments,
and this fact created a significant theoretical puzzle. According to the dominant
theories of the day, for a response to be learned, some event had to occur that
reinforced it. The reinforcer for escape responses was obvious—termination of
shock. But what was the reinforcer for avoidance responses? This also seems
obvious—the absence of shock. But not so fast. If the absence of shock is a
reinforcer, then why doesn't it reinforce everything an animal does? After all,
before the experiment, the animal went through its life seeking food, grooming,
sleeping, exploring, and each of these behaviors was accompanied by the
absence of shock. While this sort of account is clearly absurd, it demonstrates
why calling the absence of shock a reinforcer is problematic. If the absence of
shock constitutes a reinforcer in the avoidance experiment, it must be because
the shock is otherwise expected. This account makes obvious sense. The animal
expects something (shock) to happen if it doesn't respond. So it responds,
thereby preventing the "expected" event. It is thus the absence of this expected
aversive event that is the reinforcer of avoidance.

For researchers and theorists of the day, dominated as they were by the
principles of behaviorism, the problem with this account was that a major aim of
their enterprise was to explain behavior without having to appeal to mental
entities like "expectations." Solomon and his students typified this enterprise
and developed a theory—two process theory—to do that job for avoidance
learning. The theory argued that fear becomes classically conditioned (Process
1) to the warning signal on the early trials before the animal has learned to jump
the hurdle. The warning signal and the shock are paired together in Pavlovian
fashion on those trials. Avoidance responses do not occur until later trials, and
when they do, they escape the fear-provoking warning signal (Process 2) and are
followed by a rapid reduction in conditioned fear. The theory thus maintained

---

Maier, Peterson, & Schwartz

4
that the animal does not really learn to "avoid" the aversive event. Rather, the
so-called "avoidance" response is really an escape response; the animal,
motivated by conditioned fear produced by the warning signal, escapes this fear.

Solomon and his students attempted to test this explanation of avoidance
learning with what was called a "transfer of control" experiment, in which the
intention was to conduct straightforward classical conditioning of fear by pairing
a neutral stimulus (e.g., a light) with an aversive stimulus. Then, in a different
environment, avoidance learning would be conducted using some other
stimulus (e.g., a tone) as the warning signal. After the avoidance response was
well established and the animal was responding reliably to the tone, the crucial
third phase of the experiment would be conducted. The light would be turned
on during the avoidance procedure, and the question was whether the animal
would now perform the avoidance response, even though the light had never
been used as a warning signal in the avoidance apparatus.

This was a key prediction made by two-process theory: If "avoidance"
responding was really "escape" from a fear-provoking warning signal, then any
time you presented such a signal, it ought to trigger the already learned
avoidance response. However, when Russell Leaf and J. Bruce Overmier,
graduate students in the Solomon laboratory, set out to test the prediction, they
had difficulty in conducting the experiment. The problem was that when, after
classical conditioning of fear had been established, the animals were exposed to
an avoidance procedure, they often failed to learn to avoid shock. Indeed, they
often failed even to learn to escape shock (Overmier, 1968; Overmier and Leaf,
1965). This was quite surprising given that such tasks are typically learned
rapidly.

Because having learned to avoid shock was a precondition for testing this
key prediction, the prediction could not be tested. The solution to the problem,

---

Maier, Peterson, & Schwartz

5
as it turned out, was to reverse phases 1 and 2 and conduct the avoidance
training first and the classical conditioning second. This was indeed done, and
the testing of two-process theory proceeded successfully. It seemed that
somehow the prior occurrence of classical conditioning interfered with the
learning of the instrumental escape and avoidance responses.

For researchers committed to rigorous testing of two-process theory, this
peculiar, accidentally discovered, order effect was largely a methodological
nuisance. However, another graduate student in the Solomon laboratory
(Seligman), and a graduate student in Henry Gleitman's laboratory which was
right next door (Steven F. Maier), thought that the "nuisance" deserved study in
its own right and might even be more interesting than the theory that was being
tested. The question was what was it about the shock animals received during
clasical conditioning that interfered with subsequent learning.

It is the very defining feature of classical conditioning that the behavior
of the subject has no impact on the occurrence of the unconditioned stimulus
(the UCS) or its properties. Could this have been important? This question led
to what is arguably the single most important experiment in the entire literature
concerning helplessness and optimism. Overmier and Seligman (1967) first gave
animals a series of either escapable shocks—shocks that could be terminated by
a response—or exactly matched but inescapable shocks, as in classical
conditioning. The animals were later tested for escape and avoidance learning
in a different apparatus. It turned out that the animals that had initially received
escapable shock learned normally, while those that had initially received
physically identical inescapable shocks failed to learn. This demonstration was
quickly followed by experiments in which it was found that an experience of
escapable shock "immunized" animals so that a later exposure to inescapable

---

Maier, Peterson, & Schwartz

6
shock was without effect on later learning (Seligman and Maier, 1967), and ideas
about control, helplessness, and optimism were born.

Learned Helplessness Theory


Why should inescapable shock interfere with later learning? The process
of attempting to answer this question became a crossroads for Seligman and
Maier. Explanatory concepts existed within the behaviorist theories that
dominated the 1960s that could provide an "explanation" (e.g., Bracewell and
Black, 1974). However, the explanation seemed contorted and inelegant, and
seemed to trivialize the phenomenon. If one added to this a growing
disenchantment with the pinched, behaviorist theories of the time, as well as two
personalities that wanted to "push the envelope," it was over-determined that a
new theory would be developed.

Seligman and Maier reasoned that it must be something about what the
animal learned about inescapable shock that was critical, rather than the shock
per se, because inescapable and escapable shocks were physically identical, yet
had drastically different effects. What could the animals be learning? Seligman
and Maier together pondered this seemingly easy-to-answer question for
months, consulted scholars in various disciplines, and could not come up with a
meaningful answer within the context of learning theory. Presumably, the key
was that the shock was inescapable. But what did that really mean? How can
that fact about shock be learned? This question obviously does not seem
difficult now, and probably would not have seemed all that difficult back then,
to someone not fortunate enough to have been immersed in the learning theories
of the day. Hovever, Seligman and Maier were immersed in those theories. And
those theories emphasized what might be called "magic moments" of temporal

---

Maier, Peterson, & Schwartz

7
conjunction of conditioned stimulus and unconditioned stimulus, response and
reinforcer. The language of "control" and "lack of control" that seems so natural
today was completely absent in the 1960s and early 1970s. An organism could
never learn about lack of control if it were locked into processing the world as a
series of these magic moments.

Influenced by some revolutionary experiments and ideas coming from
another lab mate, Robert Rescorla (e.g., 1967), Seligman and Maier ultimately
reasoned that the animal must be learning that responding and shock
termination are independent. This required that organisms be sensitive to the
probability of an outcome (e.g., shock termination) given that they had made
some response, to the probability of the outcome given that they had not made
that response, and to the relation between these two probabilities. Act and
outcome were independent when these two probabilities were equal, and
Seligman and Maier argued that this is what the animal learns about inescapable
shock—that shock termination is independent of voluntary responses.

It was not long before Seligman and Maier realized that the comparison of
these two probabilities defined a dimension that could be called "behavioral
control over environmental events." Learning about this dimension—the
"computing" of probabilities—is quite far removed from the "magic moments" of
earlier theories. However, this still did not explain why animals exposed to
inescapable shock later fail to learn to escape. Seligman and Maier argued that
the learning that shock termination is independent of behavior has two major
consequences. First, this learning interferes with the subsequent formation of
associations between the escape response and shock termination. Second, this
learning undermines the motivation to attempt to escape. This entire set of
conjectures was first published in a chapter by Maier, Seligman, and Solomon
(1968) and was collectively called the learned helplessness hypothesis.

---

Maier, Peterson, & Schwartz

8

Research from 1970-1985


Animal research on learned helplessness in the next 10-15 years went in
two different directions. The first one focused on the behavioral phenomenon
itself and revolved around its generality and limits. Was the interference with
escape learning produced by inescapable shock restricted to escaping shock, or
would the organism also fail to escape other aversive events? Would an
inescapable event other than electric shock produce the same phenomenon? Did
uncontrollable aversive events affect aspects of behavior other than escape
learning? Did uncontrollable positive events produce analogous outcomes?
How long did the effects persist? What was the range of species that showed
helplessness phenomena? Could helplessness be demonstrated in humans?
Questions such as these were addressed by a growing number of investigators,
and answers to these questions indicated that the phenomenon was quite robust
and general (see Maier and Seligman, 1976).

The second direction concentrated on theory testing. The learned
helplessness theory initially not only met with great resistance but also
generated quite a controversy. This should be no surprise since the assumptions
about the nature of the learning process made by the theory were opposed to the
ideas that were then dominant. In addition to criticizing the ideas involved in
the theory of learned helplessness, opponents suggested alternative
explanations of the basic interference with escape learning produced by
inescapable shock. There were two categories of alternative theories. One
category was behavioral. As a class, these theories argued that exposure to
inescapable shock taught organisms some response that interfered with the one
they were later required to learn. The second category was neurochemical,

---

Maier, Peterson, & Schwartz

9
derived from some pioneering work by Jay Weiss (Weiss, 1968; Weiss, Stone, &
Harrell, 1970). These theories argued that inescapable shock depleted a
neurotransmitter, typically norepinephrine, that was necessary for the mediation
of movement. Therefore, helplessness was not the result of an interference with
learning per se, but rather it was the result of neurochemically based movement
impairment. What the behavioral and neurochemical accounts have in common
is an appeal to peripheral (movement based) rather than central (learning based)
mechanisms to explain interference.

The idea that the learned helplessness phenomenon could be explained
by the learning of interfering motor responses was relatively easy to disprove
(Maier, 1970). However, the neurochemical depletion and movement deficit
ideas were more challenging. It became clear that animals that had been
exposed to uncontrollable aversive events did later move less in the presence of
aversive events than did other animals. However, this could be readily
explained by both theories. Reduced motivation to escape consequent to
learning uncontrollability, as well as depletion of transmitters required for
movement, both would predict reduced movement.

The difference between the views thus came to be focused on whether
there was a true interference with associative processes, as well as a reduction in
movement, following exposure to uncontrollable aversive events. The difficulty
was that the learning tasks used in learned helplessness experiments
confounded poor learning with reduced movement. That is, all the tasks that
had been used required active motor output (e.g., jumping over the hurdle) as
the index of learning. A series of experiments attempted to resolve this issue by
assessing learning in tasks in which there was either no correlation (Jackson,
Alexander, & Maier, 1980), or even a negative correlation between learning and
movement (Minor, Jackson, & Maier, 1984). In the latter category of study, the

---

Maier, Peterson, & Schwartz

10
behavior needed to escape was the withholding of an active motor response, and
animals previously exposed to uncontrollable stressors continued to emit this
active motor response, thereby failing to learn to escape. Here, failure to learn
was reflected in greater movement rather than reduced movement.

Synthesis


Despite this research, there still were numerous difficulties for learned
helplessness theory. First, even at the level of psychological theory, learned
helplessness was vague concerning the mechanism by which uncontrollable
stressors produce later associative interference. Exactly what was interfered
with? Second, despite the existence of a true associative interference, movement
per se was nevertheless often still reduced. Third, learned helplessness theory
had no satisfactory explanation for many of the behavioral effects of
uncontrollable stressors that occurred in addition to interference with escape
learning. Why should uncontrollable stressors reduce aggressiveness, interfere
with maternal behavior, exaggerate fear conditioning, reduce food and water
intake, and increase responsiveness to opiate drugs such as morphine? The list
of consequences of uncontrollable stressors is long, and reduced incentive to
escape and associative interference can not explain all, or even most of them.

Purely behavioral research continues to make progress on the nature of
the alterations in associative processes that are produced by exposure to
uncontrollable stressors. The bulk of the evidence suggests that exposure to
uncontrollable stressors produces an attentional shift away from "internal,"
response-produced cues and toward external cues (Lee & Maier, 1988). This
might suggest that uncontrollable stressors produce a change in learning style,
not a deficit per se. Indeed, it might be expected that uncontrollably stressed

---