THE THEORY OF

COGNITIVE DISSONANCE:
A CURRENT PERSPECTIVE’

Elliot Aronson
DEPARTMENT O F PSYCHOLOGY
UNIVERSITY O F TEXAS
AUSTIN TEXAS .

I . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
A . The Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
B . Dissonance Following a Decision ............................... 3
C . Dissonance Resulting from Effort ............................ 4
D . Insufficient Justification . . . . . . . . . . . . . . . . ........... 4
I 1 . What 1s Psychological Inconsistency? ................................ 5
I11 . Methodological Problems ...................................... 8
The Problem of Alternative Explanations ......................... 8
IV . The “Nothing But” Critique .................................... 12
A . Dissonance or Conflict? .................................... 12
B . Dissonance or Self-judgment? . . ...................... 14
V . The Multiple Mode Problem ................................... 16
A . Consistency with Other Events . . . . ..................... 17
B . Commitment and Volition ...................................... 18
VI . Dissonance Theory and Reward-Incentive Theory . . . . . . . . . . . . . . . . . . 20
A . Not Which but When ...................................... 20
B. Commitment and Complexity ............................... 22
VII . The “Underlying Cognition” Problem ............................ 24
VIII . The Importance of the Self-Concept and Other Expectancies ......... 26
IX . Man Cannot Live by Consonance Alone . . . . . ................. 30
X . Epilogue . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
References . . . . ........ ........................ 32

‘Slightly revised from a chapter entitled “Dissonance Theory: Progress and Prob-
..
lems. in The Cognitive Consistency Theories: a Source Book. edited by R . Abelson. E.
. .
Aronson. W McGuire. T. Newcomb M . Rosenberg. and P . Tannenbaum; Chicago:
Rand McNally. 1968; reprinted by permission of the author. editors. and publisher . This
paper was prepared while the author’s research was being supported by the National
Institute of Mental Health. grant MH 12357.

1
2 ELLIOT ARONSON

I. Introduction
As a formal statement, Festinger’s theory of cognitive dissonance
(1957) is quite primitive; it lacks the elegance and precision that are
commonly associated with scientific theorizing. Yet its impact has been
great. As McGuire has observed in his recent survey in the Annual
Review of Psychology ( 1 966, p. 492), “Over the past three years, dis-
sonance theory continued to generate more research and more hostility
than any other one approach.” We will allude to the “hostility” part of
this statement from time to time throughout this article; but first, let us
discuss the research.
The research has been as diverse as it has been plentiful; its range
extends from maze running in rats (Lawrence and Festinger, 1962) to
the development of values in children (Aronson and Carlsmith, 1963);
from the hunger of college sophomores (Brehm et al., 1964) to the
proselytizing behavior of religious zealots (Festinger et al., 1956). For
descriptive summaries of dissonance experiments, the reader is referred
to Festinger (1957); Festinger and Aronson (1960); Brehm and Cohen
( 1 962); Festinger and Bramel ( 1962); Festinger and Freedman ( 1964).
The proliferation of research testing and extending dissonance
theory results for the most part from the generality and simplicity of
the theory. Although it has been applied primarily in social psychologi-
cal settings, it is not limited to social psychological phenomena such as
interpersonal relations or feelings toward a communicator and his
communication. Rather, its domain is in the widest of places-the skull
of an individual organism.2

A. THETHEORY
The core notion of the theory is extremely simple: Dissonance is a
negative drive state which occurs whenever an individual simultane-
ously holds two cognitions (ideas, beliefs, opinions) which are psycho-
logically inconsistent. Stated differently, two cognitions are dissonant if,
considering these two cognitions alone, the opposite of one follows
from the other. Since the occurrence of dissonance is presumed to be
unpleasant, individuals strive to reduce it by adding “consonant” cogni-
tions or by changing one or both cognitions to make them “fit togeth-

2An additional reason for the great number of experiments on dissonance theory is
completely nd horninern; Leon Festinger has an unmatched genius for translating interest-
ing hypotheses into workable experimental operations and for inspiring others to do so.
He has produced a great deal of research irrespective of any particular theoretical ap-
proach.
 THE THEORY OF COGNITIVE DISSONANCE 3

er” better: i.e.. so that they become more consonant with each other.”
To use Festinger’s time-worn (but still cogent) example, if a person be-
lieves that cigarette smoking causes cancer and simultaneously knows
that he himself smokes cigarettes. he experiences dissonance. Assum-
ing that the person would rather not have cancer, his cognition “I
smoke cigarettes” is psychologically inconsistent with his cognition
“Cigarette smoking produces cancer.” Perhaps the most efficient way
to reduce dissonance in such a situation is to stop smoking. But. as
many of us have discovered, this is by no means easy. Thus, a person
will usually work on the other cognition. There are several ways in
which a person can make cigarette smoking seem less absurd. He
might belittle the evidence linking cigarette smoking to cancer (“Most
of the data are clinical rather than experimental”); or he might asso-
ciate with other cigarette smokers (“If Sam, Jack. and Harry smoke,
then it can’t be very dangerous”); or he might smoke filter-tipped ciga-
rettes and delude himself that the filter traps the cancer-producing ma-
terials; or he might convince himself that smoking is an important and
highly pleasurable activity (“I’d rather have a shorter but more enjoy-
able life than a longer, unenjoyable one”); or he might actually make a
virtue out of smoking by developing a romantic, devil-may-care image
of himself, flaunting danger by smoking. All of these behaviors reduce
dissonance, in effect, by reducing the absurdity involved in going out of
one’s way to contract cancer. Thus, dissonance theory does not rest
upon the assumption that man is a rational animal; rather, it suggests
that man is a rationalizing animal-that he attempts to appear rational,
both to others and to himself. To clarify the theoretical statement and
to illustrate the kind of research generated by the theory a few experi-
ments will be briefly described.

B. DISSONANCE
FOLLOWING A DECISION

One of the earliest experiments testing derivations from dissonance

theory was performed by Brehm (1956). Brehm gave individuals their
choice between two appliances which they had previously evaluated.
He found that following the decision, when the subjects reevaluated the
alternatives, they enhanced their liking for the chosen appliance and
downgraded their evaluation of the unchosen one. The derivation is

3Although dissonance theory is an incredibly simple statement, it is not quite as

simple as a reading of this article will indicate. Many aspects of the theory (for example,
the propositions relevant to the magnitude of dissmance) will not be discussed here be-
cause they are peripheral to the major focus of this essay.
4 ELLIOT ARONSON

clear: Following a difficult choice, people experience dissonance. Cog-

nitions about any negative attributes of the preferred object are disso-
nant with having chosen it; cognitions about positive attributes of the
unchosen object are dissonant with not having chosen it. To reduce
dissonance, people emphasize the positive aspects and deemphasize the
negative aspects of the chosen objects while emphasizing the negative
and deemphasizing the positive aspects of the unchosen object (see
also Festinger, 1964).

c. RESULTINGFROM FFFORT
DISSONANCE
Aronson and Mills ( 1 959) reasoned that if people undergo a great
deal of trouble in order to gain admission to a group which turns out to
be dull and uninteresting they will experience dissonance. The cogni-
tion that they worked hard in order to become a member of the group
is dissonant with cognitions concerning the negative aspects of the
group. One does not work hard for nothing. To reduce dissonance,
they will distort their perception of the group in a positive direction. In
the Aronson-Mills experiment, college women underwent an initiation
in order to become a member of a group discussion on the psychology
of sex. For some of the girls the initiation was very embarrassing-it
consisted of reciting a list of obscene words in the presence of the male
experimenter. For others the initiation was a mild one. For still others
there was no initiation at all. All of the subjects then listened to the
same tape recording of a discussion being held by the group they had
just joined. As predicted, the girls in the Severe Initiation condition
rated the discussion much more favorably than did those in the other
two conditions [see also Aronson (1961); Zimbardo (1965); Lewis
(1964); Gerard and Mathewson (1966)l.

D. INSUFFICIENT JUSTIFICATION
Aronson and Carlsmith (1963) predicted that if threats are used to
prevent people from performing a desired activity, the smaller the
threat, the greater will be the tendency for people to derogate the activ-
ity. If an individual refrains from performing a desired activity, he ex-
periences dissonance: The cognition that he likes the activity is disso-
nant with the cognition that he is not performing it. One way to reduce
dissonance is by derogating the activity-in that way he can justify the
fact that he is not performing it. However, any threat provides cogni-
tions that are consonant with not performing the activity; and the more
severe the threat, the greater the consonance. In short, a severe threat
 THE THEORY OF COGNITIVE DISSONANCE 5

provides ample justification for not performing the activity; a mild

threat provides less justification. leading the individual to add justifica-
tions of his own in the form of convincing himself that he does not like
to perform the activity. In their experiment, Aronson and Carlsmith
found that children who were threatened with mild punishment for
playing with a desired toy decreased their liking for the toy to a greater
extent than did children who were severely threatened (see also Turner
and Wright, 1965; Freedman, 1965).

11. What Is Psychological Inconsistency?

The very simplicity of the core of the theory is at once its greatest
strength and its most serious weakness. We have already discussed the
heuristic value of its simplicity. I t should be emphasized that many of
the hypotheses which are obvious derivations from the theory are
unique to that theory; i.e., they could not be derived from any other
theory. This increases our confidence in dissonance theory as an ex-
planation of an important aspect of human behavior. The weakness
occurs primarily in the difficulty involved with defining the limits of the
theoretical statement. While at the “center” of the theory it is relatively
easy to generate hypotheses that are clear and direct, at its “fringes” it
is not always clear whether or not a prediction can be made from the
theory and, if so, exactly what that prediction will be.4 Although inves-
tigators who have had experience working with the theory seem to
have little difficulty intuiting its boundary conditions, they have had
considerable difficulty communicating this to other people; indeed, a
situation has evolved which can best be described by the statement: “If
you want to be sure, ask Leon.” This has proved to be both a source
of embarrassment for the proponents of the theory as well as a source
of annoyance and exasperation to its critics.
Why is it so difficult to make a more precise theoretical statement?
Perhaps the most basic reason has to do with the nature of the incon-
sistency involved in the core definition of dissonance theory. I t would
be easy to specify dissonant situations if the theory were limited to log-
ical inconsistencies. There exist relatively unequivocal rules of logic
which can be applied without ambiguity or fear of contradiction. But
recall that the inconsistency that produces dissonance, although it can
be logical inconsistency, is not necessarily logical. Rather, it is psycho-
logical inconsistency. While this aspect of the theory increases its

4Later in this article some attempt will be made to specify exactly what is meant
by “center“ and “fringes.”
6 ELLIOT ARONSON

power, range, and degree of interest, at the same time it also causes
some serious problems. Thus, returning to our friend, the cigarette
smoker, the cognition regarding smoking cigarettes is not logically in-
consistent with the cognition linking cigarette smoking to cancer; i.e.,
strictly speaking, having information that cigarette smoking causes can-
cer does not make it illogical to smoke cigarettes. But these cognitions
do produce dissonance because, taken together, they do not make
sense psychologically. Assuming that the smoker does not want cancer,
the knowledge that cigarettes cause cancer should lead to not smoking
cigarettes. Similarly, none of the research examples mentioned above
deals with logical inconsistency; e.g., it is not illogical to go through
hell and high water to gain admission to a dull discussion group; it is
not illogical to choose to own an appliance that one considers slightly
more attractive than the unchosen alternative: it is not illogical to re-
frain from playing with a toy at the request of an adult.
Festinger (1 957) lists four kinds of situations in which dissonance
can arise: (1) logical inconsistency; (2) inconsistency with cultural mores;
(3) inconsistency between one cognition and a more general, more encom-
passing cognition; and (4)past experience.
( 1 ) Logical inconsistency: Suppose a person believed that all men
are mortal but also held the belief that he, as a man, would live forever.
These two cognitions are dissonant because they are logically inconsis-
tent. The obverse of one follows from the other on strict logical grounds.
(2) Cultural mores: If a college professor loses his patience with one
of his students and shouts at him angrily, his knowledge of what he is
doing is dissonant with his idea about what is the proper, acceptable be-
havior of a professor toward his students -in our culture. In some other
cultures this might be appropriate behavior and, therefore, would not
arouse dissonance.
(3) Inconsistency between a cognition and a more encompassing
cognition: In a given election, if a person who has always considered him-
self to be a Democrat votes for the Republican candidate, he should expe-
rience dissonance. The concept “I am a Democrat” encompasses the
concept “I vote for Democratic candidates.”
(4)Past experience: If a person stepped on a tack while barefoot and
felt no pain, he would experience dissonance because he knows from ex-
perience that pain follows from stepping on tacks. If he had never had
experience with tacks or other sharp objects, he would not experience
dissonance.
The illustrations presented above are clear examples of dissonance.
Similarly, the situations investigated in the experiments described above
are clearly dissonant. But there are situations where for all practical pur-
 THE THEORY OF COGNITIVE DISSONANCE 7

poses it is not perfectly clear whether two cognitions are dissonant o r

merely irrelevant. Because dissonance is nor limited to logical inconsis-
tencies, it is occasionally difficult to specify a priori whether o r not a cul-
tural more is being violated, whether o r not an event is markedly different
from past experience, or whether or not it is different from a more general
cognition. Recall the basic theoretical statement: Two cognitions are dis-
sonant if, considering these two cognitions alone, the obverse of one fol-
lows from the other. The major source of conceptual ambiguity rests upon
the fact that Festinger has not clarified the meaning of the words “follows
from.”
For example, if I learn that my favorite novelist beats his wife, does
this arouse dissonance? It is difficult to be certain. Strictly speaking, being
a wife-beater is not incompatible with being a great n ~ v e l i s t However,
.~
there may be a sense in which the term “great novelist” implies that such
a person is wise, sensitive, empathic, and compassionate- and wise, sen-
sitive, empathic, and compassionate people do not go around beating their
wives. This is not a logical inconsistency; nor is it a clear violation of a
cultural more; moreover, it may have nothing to do with past experience
-and it is not necessarily embedded in a more general cognition. Thus, a
knowledge of the kinds of situations in which dissonance can occur is not
always useful in determining whether dissonance does occur.
A rule of thumb which we have found useful is to state the situation
in terms of the violation of an expectancy. For example, one might issue
the following instructions: “Consider Mr. Roy Wilkins of the National
Association for the Advancement of Colored People. I’m going to tell you
something about his beliefs about the native 1Q of Negroes relative to
that of Caucasians. What do you expect these beliefs to be?” N o doubt
most people would have a firm expectancy that Mr. Wilkins would say
that there are no innate differences. Consequently, one could then con-
clude that if individuals were exposed to a statement by Mr. Wilkins to
the effect that Negroes were innately stupider than Caucasians, most
would experience dissonance. Let’s try our difficult example. Suppose we
confronted a large number of people with the following proposition:
“Consider the great novelist, X. I am about to tell you something about
whether o r not he beats his wife. What do you expect me to say?” Proba-
bly most people would shrug; i.e., they would not have a strong expect-
ancy (but, again, this is an empirical question; there is no certainty that

51f I had beaten my wife 1 might experience dissonance because of my violation of a

cultural more. But since I know that many people beat their wives, discovering that a partic-
ular person beats his wife is not necessarily inconsistent with my cognition about the world
and human nature, More will be said about this later.
8 ELLIOT ARONSON

it would come out this way). If this occurred, one could conclude that X’s
wife-beating behavior is irrelevant to his status as a novelist. An empirical
rule of thumb may be of practical utility but is, of course, no substitute for
a clearer, less ambiguous, more precise theoretical statement. Near the
end of the article this rule of thumb will be elaborated upon and it will be
indicated how it might be used conceptually.

111. Methodological Problems

Some critics have pointed to the ambiguities inherent in the theoreti-
cal statement and have concluded that they make the theory impossible to
disprove and, consequently, worthless. As stated above, some conceptual
ambiguities do exist and will be elaborated on shortly. But first, we should
make it clear that these conceptual ambiguities exist in a very small part
of the domain in which the theory has continued to make clear and precise
predictions; these predictions have been validated a number of times in a
number of different ways. Why, then, does the theory inspire what Mc-
Guire (1 966) referred to as “. . . more hostility than any other one ap-
proach”? We feel that a good deal of the hostility is misdirected - stem-
ming from a confusion between conceptual and methodological ambigui-
ties. Much of the difficulty in disproving dissonance theory arises from
weaknesses in the method of social psychological experimentation.
These weaknesses are hardly the fault of the theory. Moreover, these
methodological problems are not peculiar to research on dissonance
theory but are shared by research on all theories that predict social
psychological phenomena. They tend to have been associated with dis-
sonance theory precisely because of the great quantity of research gen-
erated (and, therefore, of methodological problems unearthed) by that
theory. The major methodological problems stem from the lack of tried
and true, standardized techniques for operationalizing conceptual vari-
ables in social psychology. Consequently, any single failure in a given
experiment can be attributed to a failure in the experimental operations
rather than an error of conceptualization. At the same time, repeated
failures across a wide variety of techniques would spell the end of dis-
sonance theory or any theory.

THEPROBLEM OF ALTERNATIVE
EXPLANATIONS
The lack of a standardized method in social psychology has con-
tributed to another major difficulty with research in this area: It is
frequently possible to come up with alternative explanations for em-
pirical results. Thus, like experiments testing other theories in social
psychology, many of the experiments testing dissonance theory are
 THE THEORY OF COGNITIVE DISSONANCE 9

subject to alternative explanations. If some of the data can be ex-

plained without recourse to dissonance theory, our confidence in
the theory is weakened. At the same time, dissonance theory does
provide the most parsimonious explanation for the data taken as
a whole-as McGuire has argued: “The whole set of dissonance stud-
ies would require accepting a tremendous variety of alternative explana-
tions, whereas dissonance theory alone explains a large subset of them”
(1966, p. 493). Although this is some recommendation, it is not wholly
sufficient. One still wants to be able to determine which explanation is
more nearly correct. The best way to distinguish among plausible alterna-
tive explanations is through a series of well-controlled systematic experi-
ments which are essentially conceptual replications using markedly differ-
ent sets of operations to test the same hypothesis. This technique has
been referred to as “purification”; the necessity for such procedures as
well as a fuller description is provided elsewhere (Aronson and Carlsmith,
1968).
Let us take, as an illustration, the initiation experiment by Aronson
and Mills (1959). Recall that the investigators predicted the results on the
basis of dissonance theory; specifically, the cognition that one has gone
through an unpleasant and embarrassing initiation in order to get into a
group was dissonant with the cognition that the discussion group was dull
and dreary. In order to reduce dissonance, subjects in the Severe Initia-
tion condition (but not in the Mild Initiation condition) convinced them-
selves that the “dull” group was really quite exciting. In order to maxi-
mize credibility and impact, the investigators constructed a rather novel
method for operationalizing “unpleasant effort”; they had the girls in the
Severe Initiation condition recite a list of obscene words and some lurid
passages from contemporary novels in the presence of a male experiment-
er. This procedure made sense in terms of the over-all “scenario” of the
experiment, thus effectively masking the true purpose of the experiment
and reducing the possibility of suspicion. It also seemed to be effective in
the sense that the girls appeared to be embarrassed- they tended to hesi-
tate, blush, cast their eyes downward, etc. Nevertheless, the use of sex-
ually related material opened the door for at least two plausible alterna-
tive explanations, both offered by Chapanis and Chapanis (1 964). One is
that while reciting the material the girls did not become embarrassed, but,
rather, became sexually aroused; this could have produced pleasure or
the expectation of pleasure which supposedly would increase the attrac-
tiveness of the discussion group. The second is quite the reverse: The
subjects in the Severe Initiation condition felt relief (from sexual anxi-
ety?) when they found the group discussion banal instead of embarrass-
ing. Supposedly, this could lead them to rate the discussion as not banal
at all.
10 ELLIOT ARONSON

Whether these explanations are more or less plausible than the disso-
nance explanation is not important. The important point is that they are at
least possible. In order to distinguish between the dissonance explanation
and these alternative explanations, the same hypothesis should be tested
using an operational definition of “unpleasant effort” which has nothing to
do with the pleasantness of sexual arousal or relief from sexual anxiety.
Such an experiment has been performed by Gerard and Mathewson
(1966), who replicated the Aronson-Mills (1959) experiment concep-
tually. In their experiment they advertised their group discussions as
being on the topic of college morals; the actual discussion was a rather
pallid one involving cheating on examinations. The initiation procedure
consisted of electric shocks instead of obscene words as used by Aronson
and Mills. The results paralleled those of Aronson and Mills and con-
firmed the prediction from dissonance theory: Those subjects who under-
went a series of severe electric shocks in order to gain admission to a dull
discussion group came to rate that group more favorably than those who
gained admission after having undergone mild electric shocks.
This single procedure, of course, does not eliminate all alternative
explanations. Let us return to the critique of the Aronson-Mills (1959)
experiment. To quote Chapanis and Chapanis:

It is interesting to speculate what would have happened if the girls had

been ‘initiated‘ into the group by the use of a more generally accepted painful
procedure, such as using electric shock. Somehow it seems doubtful that this
group would appreciate the group discussion more than the control group, unless
-and here is the crucial point- the conditions were so manipulated that Ss expe-
rienced a feeling of successful accomplishment in overcoming the painful obsta-
cle. It seems to us that if there is anything to the relationship between severity of
initiation and liking for the group, it lies in this feeling of successful accomplish-
ment. The more severe the test, the stronger is the pleasurable feeling of success
in overcoming the obstacle. There is no need to postulate a drive due to disso-
nance if a pleasure principle can account for the results quite successfully
(1964, p. 5).

Thus, while Chapanis and Chapanis would appear to have been

wrong in their conviction that the effect demonstrated by Aronson and
Mills would not replicate if electric shock had been used, they have appar-
ently left themselves an escape hatch. Fortunately, however, there are
some data on this issue also. According to Chapanis and Chapanis (1964),
the more painful the situation one overcomes, the greater the feeling of
successful accomplishment. Although they do not explain how this feeling
of pleasure would make subjects like the discussion group better, one as-
sumes that they are using a rather simple contiguity model: If a person
feels good, contiguous stimuli (e.g., the discussion group) look and feel
 THE THEORY OF COGNITIVE DISSONANCE 11

good. Dissonance theory, of course, does not make use of such a contigu-
ity explanation; i.e., the group discussion looks good not because it is con-
tiguous with pain reduction (dissonance reduction) - rather, it comes to
look good as a means ofreducing dissonance. The crucial aspect of disso-
nance arousal in this situation is that getting into a group was contingent
upon going through a severe initiation; that is, it was an initiation, not
simply a stimulus that was contiguous with a pleasant feeling. Conse-
quently, if one simply hears a group discussion after having successfully
undergone a severe shock, dissonance theory would make no prediction
regarding the attractiveness of the group. It would make a prediction only
if the person had experienced dissonance; i.e., if the person had under-
gone a severe initiation in order ro get into a dull group.
Thus, a test between the Chapanis and Chapanis “successful accom-
plishment” explanation and the dissonance explanation can be arranged
simply by comparing an initiation (i.e., an “in order to” situation) with a
contiguous situation. Such a test was built into the Gerard-Mathewson
(1966) study. In this experiment some subjects underwent a severe shock
in order to get into a group (Initiation condition) while other subjects sim-
ply underwent severe shock (No Initiation condition). If a feeling of suc-
cess is aroused by getting through the shock situation, both groups had it.
All subjects were then exposed to a taped group discussion. Thus, for
subjects in both conditions the discussion was contiguous with feelings of
“successful accomplishment”; but only those in the Initiation condition
experienced dissonance. The results clearly support dissonance theory.
Those who went through severe electric shock in order to get into a dull
group rated the taped group discussion as more attractive than a “mild
shock” control condition. Those who went through a severe shock
(without dissonance) and then listened to the same tape rated the discus-
sion as less attractive than those in the Initiation condition-indeed, they
tended to rate the taped discussion as less attractive than subjects in the
parallel (No Initiation) condition who underwent mild electric shock. This
latter finding suggests that even in the absence of dissonance, “a feeling of
successful accomplishment” does not operate - but something else does;
more will be made of this later.6
To sum up this point, it should be made clear that neither the receiv-
ing of electric shock nor the recitation of obscene words is a perfect em-
pirical realization of the conception “unpleasant effort.” Neither, by it-

OOne additional piece of data is of relevance. One-half of the subjects in the Initiation
condition were told they passed the test and one-half were not told. The “told-not told”
manipulation did not interact with the severity of shock. This provides further evidence
against the “successful accomplishment”explanation.
12 ELLIOT ARONSON

self, is free of alternative explanations. The recitation of obscene words is

open to alternative explanations involving sexual matters -electric shock
is open to alternative explanations involving pain, fear, pain reduction,
and fear reduction. But taken together, they eliminate most possible alter-
native explanations. Accordingly, many of the results supporting disso-
nance theory have been and can continue to be strengthened by elimi-
nating alternative explanations through the purification of operations
afforded by conceptual replications. As this process continues, our
confidence in the validity and viability of the theory increases-in
spite of its simplicity and inelegance as a conceptual statement.‘
Of course, as indicated, not all the problems of dissonance theory
are methodological. Several additional conceptual problems will be dis-
cussed in a moment.

IV. The “Nothing But” Critique

Scientists tend to be conservative, parsimonious creatures. This is
generally a healthy attitude which most frequently manifests itself in a
reluctance to accept a new theory or a novel explanation for a phenome-
non if the phenomenon can be squeezed (even with great difficulty) into an
existing approach. In this regard, dissonance theory has been referred to
as “warmed-oversoup”; i.e., as nothing but a new name for an old explana-
tion. This has been most persistently stated in regard to that aspect of the
theory related to decision making. In this context dissonance theory has
been referred to as nothing but another name for conflict theory.

A. DISSONANCE
OR CONFLICT?

In fact, there are several differences. Conflict occurs before a deci-

sion is made, dissonance occurs after the decision. During conflict it is
assumed that an individual will devote his energies to a careful, dispas-
sionate, and sensible evaluation and judgment of the alternatives. He will

‘In struggling toward greater methodological sophistication, investigators working

with dissonance theory face the same problems as other experimental social psychologists.
Thus, the major critical review of dissonance theory to date (Chapanis and Chapanis, 1964)
is largely a methodologicalcritique. Although many of the points made in this review involve
reasonable methodological criticisms, the unfortunate illusion is created that, somehow,
“dissonance theorists” commit more methodological blunders than the rest of us. In articu-
lating this point, Chapanis and Chapanis attempt to cite examples of good (i.e.,
nondissonance) methodology in this area. Ironically, their principal example of good meth-
odology is an experiment where the subjects were allowed to assign themselves to experi-
mental conditions (p. 19), thus negating the major defining characteristic of an experiment.
 THE THEORY OF COGNITIVE DISSONANCE 13

gather all of the information, pro and con, about all of the alternatives in
order to make a reasonable decision. Following the decision, a person is
in a state of dissonance-all negative aspects of X are dissonant with hav-
ing chosen X;all positive aspects of Y are dissonant with not having cho-
sen Y.Far from evaluating the alternatives impartially (as in conflict), the
individual experiencing dissonance will seek biased information and eval-
uations designed to make his decision appear more reasonable. As in
Brehm’s (1956) experiment, he will seek to spread the alternatives apart.
The more difficulty a person had making a decision, the greater the tend-
ency toward this kind of behavior as a means of justifying his decision.
But how can we be certain that the “spreading apart” of the alterna-
tives in Brehm’s experiment occurred after the decision? Could it not
have occurred during the conflict stage? That is, it is conceivable that, in
order to make their decision easier, subjects in Brehm’s experiment began
to reevaluate the appliances in a biased manner before the decision. If this
were the case, then there is no essential difference between predecisional
and postdecisional processes; if so, this behavior can be considered part
of conflict-and there is, indeed, no need to complicate matters by bring-
ing in additional terminology.
Brehm’s experiment does not allow us to determine whether the eval-
uation of chosen and unchosen alternatives was spread apart before or
after the decision. Experiments by Davidson and Kiesler (1 964) and by
Jecker (1964) serve to clarify this issue. In Jecker’s experiment, subjects
were offered their choice between two phonograph records. In three con-
ditions there was low conflict; i.e., subjects were told that there was a very
good chance that they would receive both records no matter which they
chose. In three other conditions, high conflict was produced by telling
them that the probability was high that they would be given only the re-
cord that they chose. All of the subjects rated the records before the in-
structions; in each of the conflict conditions subjects rerated the records
either ( a ) after they discovered that they received both records, ( b )after
they discovered that they received only the one record they chose, or (c)
before they were certain whether they would get one or both. The results
are quite clear: N o spreading apart occurred when there was no disso-
nance; i.e., when the subject actually received both records or when he
was not certain whether he would receive one or both, he did not reeval-
uate the alternatives systematically. Where dissonance did occur there
was a systematic reevaluation; i.e., subjects spread their evaluation of the
alternatives when they received only one record - this occurred indepen-
dently of the degree of conflict. This experiment provides clear evidence
that conflict and dissonance are different processes; whatever else disso-
nance theory might be, it is not “nothing but conflict theory.”
14 ELLIOT ARONSON

B. DISSONANCE
OR SELF-JUDGMENT?

An intriguing variation on the “nothing but” theme is Bem’s (1965,

1967) analysis of the insufficient justification phenomenon. Speaking from
the point of view of “a radical behaviorist,” Bem suggested that the exper-
iments involving insufficient justification can be accounted for by a self-
judgment model. Accordingly, an aversive motivational state
(dissonance) is superfluous to a n understanding of these phenomena.
Bem’s model was based upon an individual’s ability to infer what his real
attitudes are by merely discriminating the circumstances which control
his behavior. According to Bern, each person is the observer of his own
behavior. The individual, then, in effect asks himself what the reinforce-
ments were which guided his actions. If the person observes that he per-
formed for a large reward, he is less apt to believe that the behavior was a
reflection of his real attitudes than if he performed it for a small reward.
To clarify the different approaches, let us examine the experiment by
Cohen (1 962) in which Yale students were induced to write an essay fa-
voring the repressive actions of the New Haven Police Department in
quelling a student riot. Cohen found that those students who were paid
SO# came to believe in the truth of their statements to a greater extent
than did those who were paid $1 .OO
According to dissonance theory, the cognition that one has written an
essay is dissonant with the cognition that one disagrees with the point of
view of the essay. The smaller the compensation, the greater the disso-
nance; the greater the dissonance, the greater the tendency to agree with
what one has written.
Bem suggested that what is called “dissonance” is really an instance
of self-judgment based upon the subject’s simple discrimination of the
reinforcement contingencies. According to Bem, the subject says, in ef-
fect, “If I wrote the essay for only SO$, then 1 must really believe it, where-
as if it required $1 .OO to get me to write it, then I probably don’t believe
it as much.” This reasoning, in and of itself, is not really different from the
way a dissonance theorist would conceptualize the process. But Bem car-
ried his reasoning one step further: He reasoned that an aversive motiva-
tional state is unnecessary. Consequently, an observer should be able to
arrive at the same inference as the subject himself-if the observer has
knowledge of the incentive offered to the subject to induce him to perform
a given behavior. Bem tested this prediction by describing to each of his
subjects one of the conditions in Cohen’s (1962) experiment. He found
that these observers could estimate the attitude of Cohen’s subjects-
even though these observers, of course, were not experiencing disso-
nance. In short, Bem’s observer-subjects estimated that Cohen’s subjects
 THE THEORY OF COGNITIVE DISSONANCE 15

who wrote the essay for SO$ were more favorably disposed to the actions
of the New Haven police than those who wrote the essay for $ 1 .OO.
But the events experienced by a “real” subject and those experi-
enced by an observer are very different. Taking this position, Jones et al.
( 1968, in press) argue that Bern’s results are misleading. Picture the situa-
tion: Yale students are asked to write an essay favoring police suppres-
sion of Yale students. It seems reasonable to assume that Bern’s observ-
ers would infer that a typical Yale student would be unwilling to comply
with the experimenter’s request. But the subject Bem described to an
observer did, in fact, comply. Because of this, Bern’s observers in the SO$
condition are likely to infer that the behavior of that specific subject was
not typical; i.e., since he was quite willing to express an obviously un-
popular point of view for such a small sum of money, he must have been
more willing to comply than most Yale students. Consequently, it is pos-
sible that he favored the actions of the New Haven police in the first
place. In Cohen’s original experiment, of course, since the subject was
himself the complier, if he complied reluctantly and was initially opposed
to the actions of the New Haven police, it is more likely that he was aware
of it.
This is a subtle distinction, but it may be an important one. In a set of
factorial experiments, Jones er al. demonstrated that they can replicate
Bem’s results under Bern’s conditions; i.e., observers felt that subjects
who wrote the essay for SO$ were more favorably disposed to the New
Haven police than those who wrote the essay for $1.00. But under condi-
tions which effectively eliminated the possibility of observers attributing a
prioridifferences to the subjects, the results were opposite to Bern’s. Here
observers estimated that the original subjects in the $ 1 .OO condition were
more favorable to the actions of the New Haven police than were the
original subjects in the SO$ condition. These results, then, cast serious
doubt on the contention that the observer (who, of course, is not experi-
encing dissonance) can effectively infer the attitudes of a subject in a dis-
sonance experiment.
However, this experiment is not completely conclusive because it
involves a change in the conditions of Cohen’s original experiment. The
possibility remains that Cohen’s subjects did come to feel that they ini-
rially favored the actions of the New Haven police. Thus, Bern’s results
may, indeed, be an accurate translation of the Cohen experiment. What
must be established in future experiments is whether or not the subject’s
behavior (writing a counter-attitudinal essay) becomes so very salient that
it overwhelms his memory about his original position. This seems unlikely
when the issue is as personally involving for the Yale students as the ac-
tions of the New Haven police. Nevertheless, the question remains an
16 ELLIOT ARONSON

open one; at this time the most that can be said is that there is no compel-
ling evidence that dissonance-like phenomena can occur in the absence of
an aversive motivational state.

V. The Multiple Mode Problem

As indicated earlier, several problems are central to the theoretical
statement. One of the knottiest and most interesting conceptual problems
in dissonance theory involves the fact that in a given situation there is
usually more than one way for a person to reduce dissonance. For exam-
ple, the cigarette smoker has several techniques at his disposal. He may
use any one, or several simultaneously. Experimentally, this problem can
be eliminated by the simple device of blocking alternative techniques of
dissonance reduction. This is part of the definition of experimental con-
trol; a n y experimenter worth his salt will attempt to control the environ-
ment so that the behavior elicited by his independent variable will occur
in a manner which is measurable and at a time and place where the mea-
suring instruments have been set up. To illustrate: I n a typical communi-
-
cation persuasion experiment, if a highly credible communicator states a
position which is discrepant from the position of the recipient, the recipi-
ent experiences dissonance. H e can reduce dissonance in one of four
ways: ( 1 ) he can change his opinion to make it coincide with the commu-
nicator’s, (2) he can attempt to change the communicator’s opinion, (3) he
can seek social support from other members of the audience, or (4)he can
derogate the communicator. If one is interested in measuring opinion
change ( I ), one can eliminate (2) and (3) by making it impossible for the
subject to interact either with the communicator or his fellow subjects.
Furthermore, one can reduce the subject’s ability to derogate the commu-
nicator by assigning the latter high enough prestige so that he becomes
virtually nonderogatable. Thus, if these four techniques exhaust the uni-
verse, the only way that a subject can reduce dissonance is by changing
his attitude on the issue. The prudent experimenter will have built his
experiment to make it appear reasonable to measure the subject’s atti-
tudes after the communication, and he will use the most sensitive measur-
ing instrument he can construct.
Thus, if the question one asks is “Does dissonance occur in such a
situation and does it get reduced?” the answer can be easily determined
experimentally. But we may have a different question in mind: “ln a given
situation, how do people generally reduce dissonance?” And the answer
to this question may be strikingly different from the mode found in the
laboratory experiment. To illustrate, in the above example, most people
 THE THEORY OF COGNITIVE DISSONANCE 17

might prefer to argue with the communicator rather than change their
opinion.
The above argument suggests that the results from carefully con-
trolled laboratory experiments, on occasion, may be somewhat mislead-
ing. For example, suppose a young Ph.D. is being considered for a teach-
ing position in a major department at a prestigeous Ivy League university.
What happens if the members of that department decide not to hire him?
If he feels that he is a good and worthy scholar, he will experience cogni-
tive dissonance: His cognition that he is a good scholar is dissonant with
his cognition that he was rejected by members of a good department.
Thus, he can reduce dissonance in at least two ways: (1) he can convince
himself that his rejectors are, in reality, stupid, defensive, unprofessional,
and/or senile people who cannot or will not recognize a good man when
they see one; (2) he can convince himself that if they can reject him (as
good as he is), then their standards must be astronomically high and there-
fore they are a fine group of nonsenile professionals. Both of these tech-
niques succeed in reducing dissonance; moreover, they both protect the
individual’s ego-he leaves for his job at East Podunk State Teacher’s
College with the conviction that he is a good scholar. But note that the
results of his dissonance-reducing behavior can leave him with totally
opposite opinions about the members of the staff at the Ivy League uni-
versity. Thus, if one wanted to arouse dissonance in an individual for the
specific purpose of enhancing his impressions of the people at Ivy Univer-
sity, one had better be careful. The same dissonance-producing situation
can result in quite the opposite dissonance-reducing behavior.

A. CONSISTENCY WITH OTHER EVENTS

This is a serious conceptual problem. One way that it can be solved is
by coming up with a set of specific propositions that can lead one to state
the conditions under which one mode or the other is more likely to occur.
A possible solution was previously outlined in a specific situation
(Aronson, 1961). The situation was one involving alternative modes of
dissonance reduction following the unsuccessful expenditure of effort. If a
person struggles to reach a goal and fails, he experiences dissonance. His
cognition that he exerted effort to attain the goal is dissonant with his cog-
nition that he did not reach it. He could reduce dissonance by convincing
himself that the goal was not worth it anyway: recall that this was the way
that Aesop’s fox reduced dissonance in the fable of the sour grapes. There
is another reasonable way to reduce dissonance: by the person’s finding
something else in the situation to which he can attach value in order to
justify his expenditure of effort without achieving his avowed goal. Thus,
18 ELLIOT ARONSON

the fox might convince himself that he got some much-needed exercise
while leaping for the grapes, and that even though he failed to get those
luscious, sweet grapes, it was worth the effort because of the muscles he
developed while trying.
Under what conditions will an individual take one path rather than
the other? The first solution (Aronson, 196I ) is probably easier, but only
in a situation where the effort expended is of short duration. However, if
the situation consists of a long and repeated expenditure of effort, it be-
comes a less viable solution. To use the previous illustration, if the fox
made a few leaps at the grapes and failed, he could convince himself that
they were probably sour anway; but if he spent the entire afternoon strug-
gling to reach the grapes, it would not effectively reduce dissonance to
maintain that the grapes were sour-for if that were the case, why in the
world did he try to reach them over and over and over again? The data
from the above-mentioned experiment indicated that after the repeated
expenditure of effort people do attach value to an incidental stimulus;
however, the definitive factorial experiment remains to be done.
I t is encouraging to note that experimenters are beginning to focus
their efforts on this kind of problem. A good example of this trend is de-
scribed in a very recent article by Walster et al. (1967), who hypothesize
that individuals will choose that mode of dissonance reduction which is
least likely to be challenged by future events. In their experiment, chil-
dren were given their choice between two toys. In a situation like this,
individuals can reduce dissonance in two ways: by cognitively increasing
the attractiveness of the chosen alternative and/or by cognitively decreas-
ing the attractiveness of the unchosen alternative. One-half of the children
were led to expect that they would subsequently hear objective informa-
tion about the toy they chose: one-half of the children were led to expect
that they would hear objective information about the rejected toy. The
investigators found, as predicted, that individuals reduced dissonance by
distorting the attractiveness of that toy which they were not going to hear
information about; that is, they opted to reduce dissonance in a manner
which was less likely to run up against objective reality.

A N D VOLITION
B. COMMITMENT
In order to be of maximum use, such specific solutions should be re-
stated into more general propositions, where possible, and incorporated
into the theory. An important step in this direction was taken by Brehm
and Cohen ( 1962) in emphasizing the importance of commitment and vo-
lition in determining not only the strength of the dissonance involved, but
also, perhaps more important, in determining the nature of the dissonance
 THE THEORY OF COGNITIVE DISSONANCE 19

and, hence, the nature of the mechanisms needed to reduce dissonance.

Whether or not a high degree of volition is present can change the nature
of the prediction even though both situations may involve cognitive disso-
nance. For example, in a minor part of their experiment, Aronson et af.
( 1963)reasoned that disagreement with a highly credible source produces
more dissonance then disagreement with a source having low credibility.
The cognition that a highly sentient person believes X is dissonant with
the cognition that I believe not A’. The higher the credibility of the source,
the greater the dissonance-because the less sense it makes to be in dis-
agreement with him. This should lead to greater attitude change in the
Highly Credible condition- to reduce dissonance. The results of their
experiment were consistent with this reasoning. On the other hand, Zim-
bardo ( 1 960) and Brehm and Cohen ( 1 962) reasoned that under certain
conditions a source having low credibility would produce greater attitude
change than one having high credibility. Specifically, if a person had cho-
sen of his own volition to go to hear a speech by a low credibility source,
he would experience a great deal of dissonance. The cognition involving
volition and commitment is dissonant with the cognition that the credibil-
ity of the communicator is low; after all, it is absurd to choose to go out of
one’s way to hear a low prestige source make a speech which is discrepant
with one’s own opinion. I n order to reduce dissonance, one might con-
vince oneself that there was no essential discrepancy- that one always
held the position espoused by the low credibility source. Thus, Zimbardo
and Brehm and Cohen suggested that under conditions of high commit-
ment one might get greater agreement with a low credibility source than
with a high credibility source. This prediction made by Zimbardo and by
Brehm and Cohen is consistent with other data involving choice and
commitment. For example, Smith (1961) found that soldiers who volun-
teered to eat grasshoppers when induced by an unpleasant leader, came to
like the grasshoppers better than did those who volunteered to eat them
when induced by an affable leader. Similar results are reported by Zim-
bardo ( 1 964a,b).
I t should be clear that the prediction made by Aronson et af.and that
made by Zirnbardo and by Brehm and Cohen are not mutually exclusive;
rather, they apply to a crucially different set of circumstances. Although
both predictions are derived from dissonance theory, they involve
different aspects of the theory; the crucial distinction is whether or not a
high degree of volition is present. Nonetheless, to avoid confusion, these
distinctions should be articulated with even greater clarity.
To sum up this section, dissonance theory, as originally stated, does
have some areas of conceptual fuzziness. Much of this fuzziness can be
eliminated by empirical research. Again, this research should be focused
20 ELLIOT ARONSON

on the conditions and variables which maximize and minimize the occur-
rence of dissonance and dissonance reduction as well as the conditions
which lead to one or another mode of dissonance reduction. This position
will be elaborated upon in a moment.

VI. Dissonance Theory and Reward-Incentive Theory

A. NOT WHICHBUT WHEN

One of the intriguing aspects of dissonance theory is that it frequently
leads to predictions which stand in apparent contradiction to those made
by other theoretical approaches, most notably, to a general reward-incen-
tive theory. The words “stand in apparent contradiction” were carefully
chosen, for as we shall see, these theories are not mutually exclusive on a
conceptual level. N o advocate of dissonance theory would take issue with
the fact that people frequently perform behaviors in order to obtain re-
wards or that activities associated with rewards tend to be repeated. What
they would suggest is that under certain carefully prescribed conditions,
cognitive events are set in motion which result in behaviors quite different
from what one would expect from reward-incentive theories. Moreover,
they might also suggest that such situations are not rare and, therefore,
such behaviors are not flukey. Rather, they are quite common; one reason
that they seem strange or “uncommonsensical” to us is that total reliance
on other theoretical approaches (explicitly or implicitly) have blinded us
to alternative possiblities or have made us disinclined to look beyond the
obvious events generated by reward-reinforcement theories. The much
discussed “nonobvious” predictions generated by dissonance theory are
nonobvious only in an apparent sense; they become obvious and make
sense once we gain an understanding of the dissonance-reducing process.
In the previous section, when discussing alternative ways of reducing
dissonance, the author tried to make the point that it is not very fruitful to
ask what the mode of dissonance reduction is; rather, it is far more mean-
ingful and instructive to isolate the various modes of reducing dissonance
and to ask what the optimum conditions are for each. Similarly, rather
than ask whether dissonance theory or reward-incentive theory is the
more valid, one should attempt to determine the optimal conditions for
the occurrence of processes and behaviors predicted by each theory.
One example of this approach has already been discussed. Recall
that in the Gerard and Mathewson (1 966) conceptual replication of the
Aronson-Mills ( 1959) experiment, they found that when dissonance was
eliminated from the experimental situation (in the No Initiation condition)
 THE THEORY OF COGNITIVE DISSONANCE 21

subjects tended to rate the group discussion as being less attractive if it

followed severe electric shock. Recall also that this is opposite to the feel-
ings of “successful accomplishment” interpretation proposed by Cha-
panis and Chapanis (1 964); rather, it can be considered as consistent with
a general reward theory: i.e., stimuli contiguous with severe shock are
considered to be unattractive. Similar findings relevant to reward theory
are reported by Aronson ( I96 1).
Another example of this approach can be found in an experiment by
Freedman ( I963), who had subjects perform a dull task after first inform-
ing them that either (a) the data would definitely be of no value to the ex-
perimenter since his experiment was already complete, or (b) the data
would be of great value to the experimenter. According to dissonance
theory, performing a dull task is dissonant with the fact that it is not very
valuable; in order to reduce dissonance, subjects should attempt to con-
vince themselves that they actually enjoyed performing the task for its
own sake. However, if the data are valuable, there is little dissonance,
hence, little need to convince one’s self that the task was enjoyable.
Freedman’s results confirmed his prediction: Subjects in the No-Value
condition enjoyed the task to a greater extent than did subjects in the
High-Value condition. In addition, he ran a parallel set of conditions
except that he withheld information about how valuable the task perfor-
mance was for the experimenter until after the subjects had completed the
task. With this modification he found the opposite effect: Those who were
told the task was valuable enjoyed it more than those who were told it was
useless.
A moment’s reflection should indicate that there is little or no disso-
nance in the above situation. No subject can have any reason to suspect
that an experimenter is observing him for no reason at all. If the subject
performed the task in good faith, he had no way of knowing his data would
not be used by the experimenter; that is, experimenters do not generally
collect data that they have no intention of using. Accordingly, the subject
does not need to seek justification for performing the task-the fact that
his performance turned out to be futile was nothing that he could have
possibly foreseen. On the other hand, if, in advance, he had some reason
for believing that his efforts might be futile (as in the previous condition),
he does need additional justification- he must convince himself that he
chose to do it for its own sake. The point stressed here is that where little
or no dissonance exists, an incentive effect emerges: The more valuable
the task, the “better” it is; the “better” it is, the more the subjects enjoyed
doing it. This experiment clearly demonstrates that dissonance effects and
incentive effects can exist side by side. Moreover, it helps define some of
the limiting conditions of each.
22 ELLIOT ARONSON

In a similar vein, a recent experiment by Carlsmith ef a f .( 1 966) has

taken us a long way toward an understanding of the conditions optimal for
the emergence of incentive and dissonance phenomena following counter-
attitudinal advocacy. According to dissonance theory, if a person says
something he feels is untrue, he experiences dissonance: The cognition “I
said X” is dissonant with the cognition “I believe not X.” In order to re-
duce dissonance, he might attempt to convince himself that what he said
was not so very untrue. Thus, dissonance theory suggests that advocating
an opposite position increases one’s tendency to believe in that position.
However, if one is provided with a great deal of justification for advocat-
ing an opposite position (for example, if one is paid a great deal of money
for telling a lie), one experiences less dissonance; that is, if I told a small
lie for $53,000, I would have ample justification for having lied: The cog-
nition that I received $53,000 is consonant with having lied. Consequent-
ly, 1would have less need to justify my action by convincing myself that I
really believed what I said than if 1 had been paid a mere 53$ for lying.
This general prediction has been confirmed by several experiments (e.g.,
Festinger and Carlsmith, 1959; Cohen, 1962; Nuttin, 1964; Lependorf,
1964), These experiments have shown greater attitude change for less
reward across a wide range of topics; moreover, it has been confirmed
across a wide range of rewards, from $20.00 (high) and $1 .OO (low) in the
Festinger-Carlsmith experiment, to SO$ (high)s and 5$ (low) in the Lepen-
dorf experiment. Thus, it would appear that this is a sturdy finding. On the
other hand, there is some evidence that under certain conditions the op-
posite effect might emerge (Janis and Gilmore, 1965; Elms and Janis,
1965; Rosenberg, 1965).9Briefly, under certain conditions, offering a high
incentive for advocating a given position may lead to a better perfor-
mance, i.e., thinking up more and better arguments. This could lead to
greater attitude change; i.e., a person changes his attitude because he has
exposed himself to more arguments because he has looked harder be-
cause he was paid more money.

B. COMMITMENT
AND COMPLEXITY

But what are these conditions? Or, better still, what conditions are
optimum for the dissonance effect and what conditions are optimum for
the incentive effect? The experiment by Carlsmith et al. (1966) provides

B“High‘’and “low” means, of course, relative to the other conditions; thus, 5 9 is

high because it is higher than 5$.

sFor a more detailed critical analysis of all of these experiments,see Aronson (1966).
 THE THEORY OF COGNITIVE DISSONANCE 23

us with a solid clue. In their experiment subjects were put through a dull
task and were then asked to describe the task as interesting. The depen-
dent variable was the extent to which the subjects convinced themselves
that the task really was interesting. The results showed a dissonance ef-
fect (the smaller the reward, the greater the opinion change) only under
conditions where subjects lied to another person in a highly committing
face-to-face situation. In other conditions, subjects wrote an essay, were
assured complete anonymity, and were told that only bits and pieces of
their argument would be used. Here an incentive effect emerged: The
greater the reward, the greater the opinion change. In the early experi-
ments (e.g., Festinger and Carlsmith, 1959) the importance of the face-to-
face situation was not fully appreciated by the investigators because this
variable was not systematically manipulated. In a recent analysis of this
area (Aronson, 1966) it was suggested that the important distinction be-
tween the above conditions is “degree of commitment”; i.e., in the face-
to-face situation the subject was saying things to a person which he him-
self believed were untrue. In our opinion, this situation involves much
more commitment and, hence, arouses much more dissonance than the
writing of an anonymous essay which the subject has been told would
not be used in its original form.
At the same time, it should be noted that the complexity of the exper-
imental operations employed by Carlsmith er af. ( 1966) allow for alterna-
tive explanations. One of the most serious of these alternative explana-
tions is in terms of the complexity of the counter-attitudinal task involved.
Rosenberg (1 966) has argued that dissonance theory may be limited to
situations where not much cognitive elaboration is required; he contended
that where the task is more complex, incentive effects might occur. In
analyzing the study by Carlsmith et al., Rosenberg made the reasonable
point that writing an essay and telling a lie not only differ in degree of
commitment but also may differ in the degree of cognitive complexity
required. Consequently, this experiment cannot be taken as offering un-
ambiguous support for our suggestion that degree of commitment is the
decisive factor.
Two very recent experiments shed some additional light on this prob-
lem. In one, Linder er af.(1967) were careful to hold the complexity of the
task constant. The task was a complex one in all conditions: College stu-
dents were asked to write an essay favoring more stringent paternalistic
supervision of students by the college administration. The experimenters
varied (a) the degree of commitment (in terms of whether or not the sub-
jects were allowed to feel that they had a clear choice as to whether or not
to write the essay) and (6)the magnitude of monetary incentive for writing
the essay. The results are quite clear: When commitment was high there
24 ELLIOT ARONSON

was a dissonance effect; i.e., the smaller the incentive, the greater the
opinion change. When commitment was relatively low there was an incen-
tive effect. A different experiment (Helmreich and Collins, 1968) pro-
duced similar results. Here the task was also held constant, but instead of
being complex (as in the study by Linder et a / . ) it was a simple one.
Subjects were asked to record a statement which would be played
to a large classroom of other students. In two relatively high com-
mitment conditions the subject’s simple statement was put on video
tape along with his name, class, major, and hometown. In a low-
commitment condition the subjects made statements anonymously on
audio tape. The results paralleled those obtained by Linder et af. In the
high-commitment conditions the smaller the incentive, the greater the
opinion change (dissonance effect): in the low-commitment condition the
greater the incentive, the greater the opinion change (incentive effect).

VII. The “Underlying Cognition” Problem

The importance of commitment emerges most clearly when we scm-
tinize the phenomenon of the white lie more thoroughly. Clearly, every
time we say something that we do not believe, we do not experience dis-
sonance. Under certain conditions there are some underlying cognitions
which serve to prevent the occurrence of dissonance. For example, if we
stated a counter-attitudinal position in the context of a formal debate, we
would not experience dissonance (see Scott, 1957,1959; Aronson, 1966).
I t is clearly understood both by the speaker and the audience that a debat-
er’s own personal views have nothing to do with the opinions he ex-
presses. The rules of the game of debating are an underlying cognition
which prevents the occurrence of dissonance. Similarly, as teachers we
frequently get exposed to a great many stupid ideas from our students.
Unless we know the student well - know that he is capable of better ideas
and know that he is capable of “taking it”-most teachers refrain from
tearing the idea to pieces. Instead, we tend to give the student our atten-
tion, nod and smile, and suggest that it is not such a bad idea. We do this
because we have a general underlying cognition that we should not dis-
courage students early in their careers and that it is wrong to be unkind to
people who are relatively powerless to fight back. It would be ludicrous to
suggest that teachers begin to believe that a student’s poor idea is really a
pretty good one simply because the teacher had said “pretty good idea” to
the student. The underlying cognition prevents the occurrence of disso-
nance. But observe how commitment can make it a dissonant situation: If,
on the basis of the teacher’s statement, the student had decided to read his
paper at an APA convention,the teacher might begin to convince himself
 THE THEORY OF COGNITIVE DISSONANCE 25

that it was not such a bad idea-because the teacher has now been com-
-
mitted he has misled the student into taking some action. This increases
the teacher’s commitment to the situation and is probably more powerful
than the underlying consonant cognition “this is how we treat students.”
The teacher now seeks additional justification for having misled the stu-
dent, perhaps by convincing himself that it was not such a bad idea
after all.
The general point to be made here is an important one. Inconsistency
is said to arise between two cognitive elements if “considering these two
alone, the obverse of one element follows from the other” (Festinger,
1957, pp. 260-261). But we know that in most situations twocognitions
are almost never taken by themselves. Occasionally, two cognitions,
which in the abstract would appear to be dissonant, fail to arouse disso-
nance because of the existence of a “neutralizing” underlying cognition.
For example, suppose 1 know a brilliant fellow who is married to an in-
credibly stupid woman. These cognitions are inconsistent but I would con-
tend that they do not necessarily produce dissonance; i.e., I can tolerate
this inconsistency-it does not cause me pain, it does not necessarily lead
me to change my opinion about the brilliant fellow or his wife. I do not
conclude that he is dumber than I thought or that she is smarter. Why?
Because 1 have a general, underlying, pervasive cognition that there are a
multitude of factors which determine mate selection- similarities of intel-
ligence being only one of them. Moreover, 1 know that it is extremely rare
for all of these to be matched in a marital relationship. Therefore, al-
though taken by themselves, the above two cognitions are incompatible, l
simply do not ever take them by themselves.
Festinger suggested that one way to reduce dissonance is to martial
consonant cognitions; thus, he might say that the above reasoning is one
way of reducing dissonance. But it is a moot yet important point whether
I martialed the above cognitions as a result of the inconsistency. or
whether I walked around with these cognitions about mate selection be-
fore the fact. If the latter is the case, then it can hardly be said that I
dredged up this overriding cognition as a means of reducing dissonance.
For example, let us look at the finding (Aronson and Carlsmith, 1963;
Turner and Wright, 1965: Freedman, 1965) that children threatened with
mild punishment for playing with a toy tend to derogate that toy after re-
fraining from playing with it. Suppose that many children entered the situ-
ation with the strong feeling that adults must be obeyed always, even
when commands are arbitrary and threats are nonexistent (“My mother,
right or wrong!”). Put another way (which will become important in a
moment), suppose that part of the self-concept of these children involved
“obedience to adult authority.” If this were the case there would have
26 ELLIOT ARONSON

been no dissonance even though, taken by itself, the cognition “ I like that
toy” is dissonant with the cognition “I’m not playing with it.” If this were
not already a part of the person’s self-concept, it might have become one
as a function of the experiment; i.e., developing a belief in the importance
of obedience is one way of reducing dissonance in the above situation.
But if it were already there, there would have been no dissonance to begin
with.
This added complexity should not lead us to throw up our hands in
despair. Rather, it should lead us to a more careful analysis of the situa-
tions we are dealing with and perhaps even to a greater concern with indi-
vidual differences.

VIII. The Importance of the Self-concept and Other Expectancies

In discussing the difficulties involved in making precise predictions
from dissonance theory in some situations, we have purposely tiptoed
around the problem of individual differences. The fact that all people are
not the same presents intriguing problems for dissonance theory as it does
for all general motivational theories. Of course, one man’s problem is an-
other man’s primary datum; i.e., psychologists who are interested in per-
sonality regard individual differences as being of great interest. For those
who are primarily interested in establishing nomothetic laws, individual
differences usually constitute nothing more than an annoying source of
error variance. Nevertheless, whether or not we are interested in indi-
vidual differences per se, an understanding of the way people differ in dis-
sonant situations can be an important means of clarifying and strengthen-
ing the theory. Basically, there are three ways that individuals differ
which should be of concern to people investigating dissonance theory:

( I ) People differ in their ability to tolerate dissonance. It seems rea-

sonable to assume that some people are simply better than others at
shrugging off dissonance; i.e., it may take a greater amount of dissonance
to bring about dissonance-reducing behavior in some people than in
others.
(2) People probably differ in their preferred mode of dissonance re-
duction; e.g., some people may find it easier to derogate the source of a
communication than to change their own opinion. Others may find the
reverse resolution easier.
(3) What is dissonant for one person may be consonant for someone
else; i.e., people may be so different that certain events are regarded as
dissonant for some but not for others.
 THE THEORY OF COGNITIVE DISSONANCE 27

The first two possibilities are covered in depth elsewhere (see Abel-
son et al., 1968). We shall not dwell on them here except to say that
earlier in this article we underscored the difficulty of ascertaining the
proper conditions for establishing whether or not dissonance exists for
most people and the conditions for determining which mode of dissonance
reduction most people will use; the existence of individual differences
complicates matters further by adding another important dimension
which should eventually be specified. The third case will be discussed
here because it is of great relevance for the general theory. Furthermore,
it is prior to the other two, for before one can determine whether ( a )an
individual is experiencing enough dissonance to reduce it or ( 6 )how he
will reduce it, we must first determine whether the events are indeed dis-
sonant, consonant, or irrelevant to him.
Dissonant or consonant with what? Recall the earlier discussion
wherein a rule of thumb based upon an expectancy was described (e.g.,
the Mr. Roy Wilkins of the NAACP and wife-beating novelist
illustrations). Dissonance theory makes a clear prediction when a firm
expectancy is involved as one of the cognitions in question. Thus, our
cognition about Roy Wilkin’s behavior can be dissonant with our expect-
ancy about how he will behave. Dissonance theory is clearer still when
that firm expectancy involves the individual’s self-concept, for -almost
by definition-our expectancies about our own behavior are firmer than
our expectancies about the behavior of another person. Thus, at the very
heart of dissonance theory, where it makes its clearest and neatest predic-
tion, we are not dealing with any two cognitions; rather, we are usually
dealing with the self-concept and cognitions about some behavior. If dis-
sonance exists it is because the individual’s behavior is inconsistent with
his self-concept.
As we suggested several years ago (Aronson, 1960), this point has
been elusive because almost all of the experiments testing dissonance
theory have made predictions based upon the tacit assumption that people
have a high self-concept. Why do people who buy new cars selectively ex-
pose themselves to advertisements about their own make of car (Ehrlich
et al., 1957) and try to convince themselves that they made the right
choice? Because the knowledge that one has bought a junky car is disso-
nant with a high self-concept. But suppose a person had a low self-con-
cept? Then the cognition that he bought a junky car would not be disso-
nant. Indeed, if the theory holds, such a person should engage in all kinds
of “bizarre” behavior such as exposing himself to advertisements about
other cars, hearing squeaks and rattles that are not even there, and saying,
in effect, “Just my luck, I bought a lemon-these things are always hap-
pening to me.” In short, if a person conceives of himself as a “schnook,”
28 ELLIOT ARONSON

he will expect to behave like a “schnook”; consequently, wise, reason-

able, successful, “un-schnooky” behavior on his part should arouse dis-
sonance. One of the advantages of this kind of statement is that it allows
us to separate the effects of dissonance from other hedonic effects; that is,
people with high self-concepts who fail doexperience dissonance,but they
experience many other negative feelings as well simply because failure is
unpleasant. No one can deny that success brings pleasant consequences
for people with high and low self-concepts alike; that is, regardless of a
person’s self-concept, successful achievement is often accompanied by
silch pleasant things as acclaim, money, fame, admiration, and popularity.
But dissonance theory allows us to predict that for people with low self-
concepts the “good feelings” aroused by the products of success will be
tempered by the discomfort caused by dissonance-the dissonance be-
tween a low self-concept and cognitions about high performance. Several
experiments have demonstrated that people who expect failure are some-
what discomforted by success (Aronson and Carlsmith, 1962; Cottrell,
1965; Brock et al., 196% but the data are by no means unequivocal (see
Abelson et al., 1968).
Thus, although we were not fully aware of it at the time, in the clear-
est experiments performed to test dissonance theory, the dissonance in-
volved was between a self-concept and cognitions about a behavior that
violated this self-concept. In the experiments on counter-attitudinaladvo-
cacy, for example, we maintain that it is incorrect to say that dissonance
existed between the cognition “I believe the task is dull” and “I told some-
one that the task was interesting.”This is not dissonant for a psychopathic
liar-indeed, it is perfectly consonant. What is dissonant is the cognition
“I am a decent, truthful human being” and the cognition “I have misled a
person: I have conned him into believing something which just isn’t true;
he thinks that I really believe it and I cannot set him straight because I
probably won’t see him again.” In the initiation experiment, in our opinion
dissonance does not exist between the cognition “I worked hard to get
into a group” and the cognition “The group is dull and stupid.” Recall that
for a “schnook” these cognitions are not at all dissonant. What is dis-
sonant in this situation is the cognition “I am a reasonable and intelligent
person” and the cognition “I have worked hard for nothing.” Reasonable,
intelligent people usually get a fair return for their investment - they
usually do not buy a pig in a poke (unless there is some reasonably im-
plicit guarantee, as in Freedman’s [ 19631 experiment discussed above).
As an empirical refinement this self-concept notion is probably triv-
ial. Experimenters have made the tacit assumption that people have high
self-concepts- and these experimenters achieved positive results; this
implies that the assumption is valid for most people in these situations.
 T H E THEORY OF COGNITIVE DISSONANCE 29

But the self-concept notion may constitute a valuable and interesting

rheoretical refinement. A theory becomes infinitely more meaningful
when its domain is clearly demarcated; i.e., when it states clearly where it
does not apply. If it is the case that dissonance theory makes unequivocal
predictions only when the self-concept or another strong expectancy is
involved, then an important set of boundary conditions has been drawn.
What we have described earlier as a rule of thumb may actually be a
conceptual clarification.
It was stated early in this article that “at the center of the theory”
predictions are unequivocal, but at the “fringes” they are somewhat
fuzzy. At this point, we can assert that “at the center” means situations in
which the self-concept or other firm expectancies are involved-and in
which most people share the same self-concepts or other firm expectan-
cies. Thus, most people have self-concepts about being truthful and hon-
est so that we can make clear predictions intuitively, as in the Carlsmith
et 01. ( 1966) experiment. Most people have self-concepts involving mak-
ing reasonable and wise decisions so that we can intuit clear predictions,
as in the Brehm ( 1956) or Jecker ( 1964) experiments. Also, most people
have firm expectancies about what Mr. Wilkins might say about Negro
intelligence, so that a dissonance theory prediction makes sense and can
be made clearly, even though a self-concept is not involved. The predic-
tion about the great novelist who beats his wife gives the theory trouble
precisely because people differ tremendously with regard to whether or
not they expect a particular novelist to be a gentle and considerate man.
In a specific instance, the knowledge of whether or not individual X has
this expectancy would increase the accuracy of the prediction. In our
opinion, this is of no great importance. What we consider important is the
recognition of the fact that dissonance theory may be best suited for mak-
ing general predictions in situations where expectancies are firm and
nearly universal.
Several years ago, Zajonc ( 1960) raised a very interesting and rea-
sonable question: If dissonance is bothersome, why do we enjoy magi-
cians? That is, magicians can be thought of as people who arouse disso-
nance. Should we not experience pain and discomfort when we see rab-
bits pulled from hats, women sawed in half, or dimes turned into quarters?
Perhaps the reason that we are not upset by magicians is because the be-
havior of a magician is consonant with our expectancy regarding magi-
cians. That is, since we know in advance that a magician uses tricks and
sleight-of-handtechniques to produce interesting illusions, why should we
experience dissonance when we see him do these things? Is this not akin
to the “schnook” who expects to purchase an inferior car?
Before the reader dismisses this as mere sophistry, it should be re-
30 ELLIOT ARONSON

marked that this is an empirical question. What is suggested is that we

enjoy magicians only when they are billed as magicians. If they were not
billed as magicians, they would cause quite a bit of discomfort. If the fel-
low sitting next to us at the bar suddenly “became” a fat woman, this
would be very upsetting-unless the bartender had forewarned us that we
were sitting next to a professional quick-change artist known as “Slippery
Sam, the man of a thousand faces.” If he then “became” a fat woman, we
would be thrilled and delighted. It is interesting to note that the bartender
could have produced a similar result if he had forewarned us that he had
placed some LSD in our drink. In short, either being told a man is a magi-
cian or being told we were fed a halucinogen is consistent with seeing a
man “become” a fat woman.
Empirically, this can be tested by finding some young children o r
some people from a different culture who have never seen or heard of
magicians. Without the expectancy regarding magicians that Zajonc and
the author share, these subjects might be quite upset by the goings on.

IX. Man Cannot Live by Consonance Alone

The implication of this article is that dissonant situations are ubiqui-
tous and that man expends a great deal of time and energy attempting to
reduce dissonance. It should be obvious that man does many other things
as well. Festinger never intended dissonance theory to be imperial or
monolithic. In 1957, he emphasized the fact that dissonance reduction is
only one of many motives and can be counteracted by more powerful
drives. We have already discussed how dissonance effects and reward-
incentive effects can both occur in the same experimental design. Even
more basic is the confrontation that occurs when consonance needs meet
utility needs head-on. An extremely high drive to reduce dissonance
would lead man to weave a cocoon about himself; he would never admit
his mistakes and would distort reality to make it compatible with his be-
havior. But if a person is ever going to grow, improve, and avoid repeating
the same errors, he must sooner or later learn to profit from past mistakes.
One cannot profit from one’s mistakes without first admitting that one has
made a mistake. And yet, the admission of error almost always arouses
some dissonance. The fact is, people frequently do profit from their mis-
takes; thus, people occasionally do not avoid or reduce dissonance.
To illustrate, if a man spends $50,000 for a home, dissonance theory
would suggest that he may be the last to notice that during the rainy sea-
son there is water in the basement. Noticing water would arouse disso-
nance by making his purchase appear to have been a mistake. But to no-
tice the water has great utility-for he must notice it in order to repair it,
 THE THEORY OF COGNITIVE DISSONANCE 31

or at least to prepare for the flood. Moreover, if he does not take cogniz-
ance of his leaky basement he may walk into the same problem the next
time he purchases a house. Thus, dissonance and utility are in constant
tension by virtue of the fact that under certain conditions dissonant infor-
mation may be extremely useful and, conversely, useful information can
arouse dissonance. This phenomenon was discussed by Mills et al.
(1959). who suggested that one reason why people frequently do not
avoid dissonant information is that it often has great utility. In their exper-
iment, they found that many subjects who had recently committed them-
selves to taking essay examinations as opposed to multiple-choiceexami-
nations opted to read articles explaining why essay examinations were
more difficult, anxiety-provoking, etc. In this situation, apparently, the
utility of the information was considered worth the price to be paid in dis-
sonance. More recent experiments by Canon ( 1964) and Aronson and
Ross (in preparation) have begun to indicate the requisite conditions for
these effects. Precise predictions can be made by manipulating the
strength of the opposing drive. As utility increases and dissonance be-
comes weaker, individuals begin to show a preference for dissonance-
arousing but useful information. But as dissonance increases (i.e., imme-
diately after a decision or when commitment is high, etc.), individuals
tend to manifest dissonance-reducing behavior in spite of the fact that the
future consequences of such behavior tend to be unpleasant.

X. Epilogue
The theory of cognitive dissonance is much more complicated than
we thought it was some 10 years ago. A good deal of research has been
done since then. Many of the problems which were specified earlier have
been solved; many new problems have been unearthed, some of which re-
main to be solved. Hopefully, future research will lead to the emergence
of still new problems, which will lead to still more research, which will
continue to yield an increased understanding of human behavior. Per-
haps this is what the scientific enterprise is all about.
I n their critique of five years of dissonance theory, Chapanis and
Chapanis ( 1964) concluded with the pronouncement “Not proven.”
Happily, after more than 10 years, it is still not proven; all the theory ever
does is generate research.
32 ELLIOT ARONSON

REFERENCES

Abelson, R.,Aronson, E., McGuire, W., Newcomb, T., Rosenberg, M., and Tannenbaum,
P. (Eds.), The cognitive consistency theories: a source book. Chicago: Rand McNally,
1968.
Aronson, E. The cognitive and behavioral consequences of the confirmation and disconfir-
mation of expectancies. Proposal to the National Science Foundation, 48 pp.. August,
1960.
Aronson, E. The effect of effort on the attractiveness of rewarded and unrewarded stimuli.
Journal ofAbnorma1 and Social Psychology, 1961,63,375-380.
Aronson, E. The psychology of insufficient justification: An analysis of some conflicting
data. In S. Feldman (Ed.), Cognitive consistency. New York: Academic Press, 1966. Pp.
115-133.
Aronson, E., and Carlsmith, J. M. Performance expectancy as a determinant of actual per-
formance. Journal of Personality and Social Psychology, 1962.65, 178-182.
Aronson, E., and Carlsmith, J. M. Effect of the severity of threat on the devaluation of for-
bidden behavior. Journal of Abnormal and Social Psychology, 1963,66,584-588.
Aronson, E., and Carlsmith, J. M. Experimentation in social psychology. In G. Lindzey and
E. Aronson (Eds.), Handbook of social psychology. (Rev. ed.), Vol. 11. Reading, Mass.:
Addison-Wesley, 1968.
Aronson, E., and Mills, J. The effect of severity of initiation on liking for a group. Journal of
Abnormal and Social Psychology, 1959.59, 177- I8 I .
Aronson, E., and Ross, A. The effect of support and criticism on interpersonal attractive-
ness. In preparation.
Aronson, E., Turner, Judith, and Carlsmith, J. M. Communicator credibility and communi-
cation discrepancy as determinants of opinion change. Journal of Abnormal and Social
Psychology, 1963,67,3 1-36.
Bern, D. J. An experimental analysis of self-persuasion. Journal of Experimental Social
Psychology, 1965,1, 199-218.
Bern, D. J. Self-perception: An alternative interpretation of cognitive dissonance phenom-
ena. Psychological Review, 1967,74, 183-200.
Brehm, J. W. Postdecision changes in the desirability of alternatives. Journal of Abnormal
and Social Psychology, 1956,52,384-389.
Brehm, Mary L., Back, K. W., and Bogdonoff, M. D. A physiological effect of cognitive dis-
sonance under stress and deprivation. Journal of Abnormal and Social Psychology, 1964,
69,303-3 10.
Brehm, J. W., and Cohen, A. R. Explorations in cognitive dissonance. New York: Wiley,
1962.
Brock, T. C., Adelman, S. K., Edwards, D. C., and Schuck, J. R. Seven studies of perfor-
mance expectancy as a determinant of actual performance. Journal of Experimental So-
cial Psychology, 1965, l, 295-310.
Canon, L. K. Self-confidence and selective exposure to information. ln L. Festinger (Ed.)
ConjZict, decision and dissonance. Stanford University Press, 1964. Pp. 83-96.
Carlsmith, J. M., Collins, B. E., and Helmreich, R. L. Studies in forced compliance: 1. The
effect of pressure for compliance on attitude change produced by face-to-face role playing
and anonymous essay writing. Journal of Personality and Social Psychology, 1966,4,1-3.
Chapanis, Natalia P., and Chapanis, A. Cognitive dissonance: Five years later. Psychologi-
cal Bulletin, 1964,61, 1-22.
Cohen, A. R. An experiment on small rewards for discrepant compliance and attitude
change. In J. W. Brehm and A. R. Cohen, Explorations in cognitive dissonance. New
York: Wiley, 1962. Pp. 73-78.
 THE THEORY OF COGNITIVE DISSONANCE 33

Collins, B. E. An experimental study of satisfaction, productivity, turnover, and comparison

levels. Unpublished paper, Northwestern University, 1963.
Cottrell, N. B. Performance expectancy as a determinant of actual performance: A replica-
tion with a new design. Journal of Personality and Social Psychology, 1965,2,685-691 .
Davidson, J. R., and Kiesler, Sara B. Cognitive behavior before and after decision. In L.
Festinger ,Ed.., Co@ict, decision and dissonance. Stanford University Press, 1964. Pp.
10-21.
Ehrlich, D., Guttman, I., Schonbach, P., and Mills, J. Post-decision exposure to relevant
information.Journal ofAbnormal and Social Psychology, 1957,54,98-102.
Elms, A. C., and Janis, 1. L.Counter-norm attitudes induced by consonant versus dissonant
conditions of role-playing. Journal of Experimental Research in Personality, 1965, 1,
50-60.
Festinger, L. A theory of cognitive dissonance. Evanston, Ill.: Row, Peterson, 1957.
Festinger, L. (Ed.), Co@ict, decision and dissonance. Stanford University Press, 1964.
Festinger, L., and Aronson, E. The arousal and reduction of dissonance in social contexts.
In D. Cartwright and A. Zander (Eds.), Group dynamics: research and theory. Evanston,
Ill.: Row, Peterson, 1960. Pp. 2 14-231.
Festinger, L. and Bramel, D. The reactions of humans to cognitive dissonance. In Bachrach,
A. J. (Ed.), Experimental foundations of clinical psychology. New York: Basic Books,
1962.
Festinger, L., and Carlsmith, J. M. Cognitive consequences of forced compliance. Journal
ofAbnorma1 and Social Psychology, 1959,58,203-210.
Festinger, L., and Freedman, J. L. Dissonance reduction and moral values. In P. Worchel
and D. Byrne (Eds.), Personality change. New York: Wiley, 1964, Pp. 220-243.
Festinger, L., Riecken, H. W., and Schachter, S . When prophecy fails. Minneapolis: Uni-
versity of Minnesota Press, 1956.
Freedman, J. L. Attitudinal effects of inadequate justification. Journal of Personality, 1963,
31,371-385.
Freedman, J. L. Long-term behavioral effects of cognitive dissonance. Journal of Experi-
mentalSocial Psychology, 1965,1, 145-155.
Gerard, H. B., and Mathewson, G. C. The effects of severity of initiation on liking for a
group: A replication. Journal of Experimental Social Psychology, 1966,2,278-287.
Helmreich, R., and Collins, B. E. Studies in forced compliance: IV.Commitment and incen-
tive magnitude as determinants of opinion change. Journal of Personality and Social
Psychology, in press.
Janis, 1. L., and Gilmore, J. B. The influence of incentive conditions on the success of role
playing in modifying attitudes. Journal of Personality and Social Psychology, 1965, 1,
17-27.
Jecker, J. D. The cognitive effects of conflict and dissonance. In L. Festinger ,Ed.., Con-
Jict, decision and dissonance. Stanford University Press, 1964.
Jones, R. A., Linder, D. E., Kiesler, C. A., Zanna, M.,and Brehm, J. W. Internal states or
external stimuli: Observers’ attitude judgments and the dissonance theory-self-persuasion
controversy. Journal of Experimental Social Psychology, 1968, in press.
Lawrence, D. H., and Festinger, L. Deterrents and reinforcement. Stanford University
Press, 1962.
Lependorf, S. The effects of incentive value and expectancy on dissonance resulting from
attitude-discrepant behavior and disconfirmation of expectancy. Unpublished doctoral
dissertation, State University of New York at Buffalo, 1964.
Lewis, M. Some nondecremental effects of effort. Journal of Comparative and Physiologi-
cal Psychology, 1964,57,367-312.
Linder, D. E., Cooper, J., and Jones, E. E. Decision freedom as a determinant of the role of
34 ELLIOT ARONSON

incentive magnitude in attitude change. Journal of Personality and Social Psychology,

1967,6,245-254.
McGuire, W. J. Attitudes and opinions. In P. R. Farnsworth, Olga McNemar, and Q. Mc-
Nemar (Eds.), Annual Review of Psychology, 1966,17,475-5 14.
Mills, J., Aronson, E., and Robinson, H.Selectivity in exposure to information. Journal of
Abnormal and Social Psychology, 1959,59,250-253.
Nuttin, J. M., Jr. Dissonant evidence about dissonance theory. For Second Conference of
Experimental Social Psychologists in Europe. Frascati, 1964.
Rosenberg, M. J. When dissonance fails: On eliminating evaluation apprehension from atti-
tude measurement. Journal of Personality and Social Psychology, 1965,1,28-42.
Rosenberg, M. J. Some limits of dissonance: toward a differentialview of counter-attitudinal
performance. In S. Feldman ,Ed.., Cognitive consistency. New York: Academic Press,
1966. Pp. 135-170.
Scott, W. A. Attitude change through reward of verbal behavior. Journal ofAbnormal and
Social Psychology, 1957.55,72-75.
Scott, W. A. Attitude change by response reinforcement: Replication and extension.
Sociometry, 1959,22,328-335.
Smith, E. E. The power of dissonance techniques to change attitudes. Public Opinion Quart-
erly, 1961, %, 626-639.
Turner, Elizabeth A,, and Wright, J. C.Effects of severity of threat and perceived availabil-
ity on the attractiveness of objects. Journal of Personality and Social Psychology, 1965,
2, 128-132.
Walster, Elaine, Berscheid, EUen, and Barclay. A. M.A determinant of preference among
modes of dissonance reduction. Journal of Personality and Social Psychology, 1967, 7 ,
211-216.
Zajonc, R. B. The concepts of balance, congruity, and dissonance. Public Opinion Quar-
terly, 1960.24,280-286.
Zimbardo, P.G . Involvement and communication discrepancy as determinants of opinion
conformity. Journal of Abnormal and Social Psychology, 1960,60,86-94.
Zimbardo, P. G. A critical analysis of Smith's "grasshopper" experiment. Dept. Psychol.,
New York University. I3 pp., 1964. mimeo (a).
Zimbardo, P.G. A reply to Jordan's attack on dissonance theory. Contemporary Psychol-
ogy, 1964,9,332-333.(b).
Zimbardo, P. G . The effect of effort and improvisation on self-persuasion produced by role
playing. Journal of Experimental Social Psychology, 1965,1,103-120.