Cognitive Dissonance: Reexamining a Pivotal Theory in Psychology - Eddie Harmon-Jones 2019
Improving the 1957 Version of Dissonance Theory
Perspectives Employing the Original Version of the Theory
In this chapter, I am going to do something audacious. I am going to propose some improvements in Leon Festinger’s most important contribution to psychology, his 1957 theory of cognitive dissonance. Some of the proposed changes have to do with Festinger’s assumptions about the magnitude of avoidance of dissonance and about what occurs before a choice. A major proposed change is concerned with how dissonance is determined by desired consequences and importance of cognitions.
In Festinger’s last public reflections on dissonance theory made at the symposium, Reflections on Cognitive Dissonance: 30 Years Later, conducted at the 95th Annual Convention of the American Psychological Association (1987; see Appendix B, this volume), he described why he left social psychology: “I left and stopped doing research on the theory of dissonance because I was in a total rut. The only thing I could think about was how correct the original statement had been” (Appendix B, this volume, p. 290). If Festinger, the master theoretician, could not improve dissonance theory, what hope is there?
However, in his remarks, Festinger also said, “If a theory is at all testable it will not remain unchanged. It has to change. All theories are wrong. One doesn’t ask about theories, can I show they are wrong or can I show they are right, but rather one asks, how much of the empirical realm can it handle and how must it be modified and changed as it matures?” (Appendix B, this volume, p. 290). Referring to dissonance theory, Festinger said, “I am quite sure that there is enough validity to the theory, and as changes are made, emendations are made, there will be even more validity to the theory, that research on it will continue, and a lot will get clarified” (Appendix B, this volume, p. 291). So although Festinger felt unable to improve the theory and left dissonance research and social psychology in 1964, his retrospection gives us hope and encouragement to develop and improve the theory.
As is well known, dissonance theory has been extremely fruitful and has stimulated an enormous amount of research. Beyond that, as I noted when moderating the 1987 symposium on dissonance, Festinger’s theorizing about dissonance has had repercussions far outside the field of social psychology. It has changed the meaning of the word dissonance. No longer do educated people immediately think of music when the word dissonance is mentioned. Cognitive dissonance has become a part of the language. For example, the term has appeared in articles on the op-ed page of the Washington Post.
Why was dissonance theory, and specifically the 1957 version, so fruitful and important? One answer to that question emphasizes Festinger’s unique research style. In the obituary for Festinger in the American Psychologist, Zajonc (1990) likened Festinger to Dostoyevsky and Picasso. Zajonc wrote, “Like Dostoyevsky and like Picasso, Festinger set in motion a style of research and theory in the social sciences that is now the common property of all creative workers in the field,” (p. 661). Zajonc said about Festinger, “Leon is to social psychology what Freud is to clinical psychology and Piaget to developmental psychology,” (p. 661). Such a statement made by a Festinger advisee such as myself would be viewed as biased and possibly self-serving, but Zajonc, who is a very distinguished psychologist, was not a Festinger advisee.
I prefer to answer the question about why the theory became so important in terms of the content of the theory. I believe that it has been so important because it has uncovered a large number of new and interesting phenomena. These phenomena dealt primarily with the effect of actions on beliefs and attitudes. The theory went far beyond the simple idea that saying is believing. It specified the conditions under which saying is believing and doing is valuing. The concern with the effect of behavior on cognition that is the hallmark of dissonance theory can be seen in the earliest version of the theory, in an unpublished paper Festinger distributed in his seminar in 1954 (see Appendix A, this volume). In that first version of dissonance theory, Festinger hypothesized that there exists a tendency to make one’s cognition and one’s behavior consonant, to reduce dissonance between behavior and cognition.
I am going to concentrate on the 1957 version of dissonance theory, which has been, and I believe continues to be, a very useful theoretical statement. The 1957 version started with a (seemingly) simple definition of dissonance. “Two elements are in dissonant relationship if, considering these two alone, the obverse of one element would follow from the other. To state it a bit more formally, x and y are dissonant if not-x follows from y” (Festinger, 1957, p. 13). I prefer to focus on the second sentence of the definition because it avoids the obscure term obverse.1
The key element of dissonance theory that made the theory so useful was, I believe, the assumption that dissonance varies in magnitude, that the total amount of dissonance depends on the proportion of relevant elements that are dissonant with the one in question. The focal element was typically a behavior or what Festinger called a behavioral element. The assumption concerning the magnitude of the dissonance, with the assumptions that dissonance was uncomfortable and that the pressure to reduce dissonance was a function of the magnitude of the dissonance, allowed the derivation of many interesting predictions, for example, the prediction that the less money one receives for convincing someone that a boring task is enjoyable, the more positive one’s attitude toward the boring task will be (Festinger & Carlsmith, 1959).
Festinger’s assumption about the magnitude of dissonance was very different than what was assumed in the other theories of cognitive consistency common in the 1950s and 1960s. It allowed not only derivations about when dissonance would be more likely to be created and thus when tendencies to reduce dissonance would be stronger but predictions about how dissonance would be reduced. The assumption that dissonance was a function of the proportion of dissonant elements enabled predictions concerning the reduction of dissonance by adding consonant cognitions as well as by removing dissonant cognitions. Those predictions were not made by the other cognitive consistency theories, which did not make assumptions about degrees of inconsistency or alleviating an inconsistency by adding a different consistency.
THE MAGNITUDE OF AVOIDANCE OF DISSONANCE
There are some assumptions in the 1957 version of dissonance theory that have received insufficient attention and are in need of revision. One is Festinger’s assumption that the greater the magnitude of existing dissonance, the greater will be the magnitude of avoidance of new information expected to increase dissonance (up to an extreme point; p. 130).
Some of the earliest research on dissonance theory examined the implications of the theory for selectivity in exposure to information. It was known in the 1950s that the mass media were generally ineffective in changing socially significant attitudes and that one prominent reason for this was self-selection of mass media audiences (Hovland, 1959). Dissonance theory was particularly useful in understanding bias in voluntary exposure to information because it specified conditions under which persons will seek out or avoid information. It predicted that persons will seek out information expected to increase consonance (consonant information) and avoid information expected to increase dissonance (dissonant information). It also predicted that the greater the existing dissonance, the stronger will be the tendencies to seek out consonant information and to avoid dissonant information.
The early dissonance studies of selective exposure to information showed that there was a preference for consonant information (Ehrlich, Guttman, Schonbach, & Mills, 1957; Mills, Aronson, & Robinson, 1959). However, the fact that people prefer consonant information does not provide evidence of avoidance of dissonant information, because such a preference can be due solely to the seeking of consonant information. To show actual avoidance of dissonant information, it is necessary to make comparisons against a neutral baseline.
Later experiments on interest in consonant and dissonant information had a neutral baseline, and evidence of avoidance of dissonant information was found (Mills, 1965a). The effect of the amount of existing dissonance on interest in consonant and dissonant information was tested (Mills, 1965b). Evidence was found that the greater the postdecision dissonance, the greater the interest in consonant information. Interest in information favoring the chosen alternative was stronger when the chosen and rejected alternatives were closer in attractiveness. However, it was not found that avoidance of dissonant information was greater, the greater the postdecision dissonance. Avoidance of information favoring the rejected alternative was not stronger when the chosen and rejected alternatives were closer in attractiveness.
The failure to find that the magnitude of avoidance of dissonance was influenced by the amount of existing dissonance in the experiment by Mills (1965b) cannot be explained away on the basis of inadequate procedures. It cannot be attributed to inadequate manipulation of amount of dissonance or inadequate measurement of interest in information. The reason is that, in the same experimental situation using the same manipulation of amount of dissonance and the same measure of interest in information, it was found that interest in consonant information was greater, the greater the amount of existing dissonance. The positive result for interest in consonant information would not have occurred unless the manipulation of amount of dissonance and the measurement of interest in information had been adequate in that experiment. That positive result provides evidence that the negative result for the effect of amount of existing dissonance on the magnitude of avoidance of dissonance was not due to inadequate procedures.
The experiment on the effect of amount of existing dissonance on interest in consonant and dissonant information (Mills, 1965b) is the only one I am aware of that has tested Festinger’s assumption about the effect of existing dissonance on the magnitude of avoidance of new dissonance. There is no evidence that I know of that supports the assumption that avoidance of new dissonance is greater, the greater the amount of existing dissonance. On the basis of what little research there is on the topic, I conclude that the assumption in the 1957 version that the magnitude of avoidance of dissonance is influenced by the amount of existing dissonance is in need of revision.
WHAT HAPPENS BEFORE A CHOICE
Another aspect of the 1957 version that is in need of revision is Festinger’s assumption about what happens before a choice. Festinger assumed that there is no cognitive bias in the prechoice situation. Festinger took the position, which, strictly speaking, was not integral to the theory, that “the preaction or predecision situation will be characterized by nonselective seeking of relevant information” and “there will be a lack of resistance to accepting and cognizing any relevant information” (p. 126).
Contrary to what Festinger assumed, research on the topic of exposure to information before a choice has found evidence of selective exposure to information. When people are not committed to a position, the more certain they are that their position is the best one, the more they prefer information supporting their position (Mills & Ross, 1964), whereas the opposite occurs when commitment is high. Before a choice, people who are certain that an alternative is not the best are less interested in information favoring that alternative than people uncertain about which alternative is best (Mills, 1965c). People certain that an alternative is not the best before a choice show evidence of avoidance of information favoring that alternative by displaying less interest in it than people who can not even choose that alternative (Mills & Jellison, 1968).
The research on selective exposure before commitment has been based on the assumption that people want to be certain when they take an action that it is better than the alternatives. That conception was termed choice certainty theory (Mills, 1968). Another line of research based on choice certainty theory found evidence of cognitive bias before a choice. The anticipation of making a choice about other people increases the halo effect in the impressions of those other people on positive traits, reflecting a greater difference in evaluations of those other people (Mills & O’Neal, 1971; O’Neal, 1971; O’Neal & Mills, 1969). When a prospective choice is nearer in time, there is greater difference in private evaluations of the alternatives (Brounstein, Ostrove, & Mills, 1979) but smaller difference in public evaluations of the alternatives.
The effects of importance of a prospective choice on private and public evaluations of the alternatives were tested in a recent experiment (Mills & Ford, 1995). It found that the more important the prospective choice, the greater the difference in private evaluations of the alternatives and the less the difference in public evaluations of the alternatives. Those results can be interpreted in terms of dissonance, specifically, avoidance of dissonance expected to occur after the choice.2
The greater difference in private evaluations of the alternatives when the prospective choice was more important can be interpreted in terms of greater motivation to avoid the dissonance that would otherwise be expected to occur after the choice. Spreading the attractiveness of the alternatives in private before a choice will reduce dissonance that might otherwise be expected after the choice. The dissonance expected after the choice should be greater, the more important the choice, and so spreading the attractiveness of the alternatives in private to avoid dissonance after the choice should be greater, the more important the choice.3 The finding that the more important the choice, the less the difference in public evaluations of the alternatives provides evidence that the private and public evaluations were made in a prechoice situation. If the evaluations were made in a postchoice situation, opposite effects for private and public evaluations should not have occurred.
The smaller difference in public evaluations of the alternatives when the prospective choice is more important can be explained in terms of the assumption Festinger (1957) made that the fear of dissonance may lead to a reluctance to commit oneself (p. 30). The explanation involves the assumption that the public expression of a preference for one alternative before a choice is regarded as involving commitment to choose that alternative. Commitment can be avoided by minimizing the difference in public evaluations of the alternatives. The greater the importance of the prospective choice, the greater the fear of dissonance and of the minimizing of the difference in public evaluations.
In Conflict, Decision, and Dissonance, Festinger (1964) emphasized that dissonance does not occur until after a choice. However, in discussing the consequences of the anticipation of postdecision dissonance for predecision behavior, Festinger (1964) stated, “If a person anticipates dissonance as a consequence of making a decision, he would be expected to react by attempting to minimize, or to avoid completely, the anticipated dissonance” (1964, pp. 144—145). If the need to be certain about the correctness of a prospective choice is construed in terms of the need to avoid dissonance that would otherwise be expected to occur after the choice, the research based on choice-certainty theory (which found bias in information seeking and evaluations before a choice) can be interpreted in terms of dissonance. As mentioned in the article by Mills and Ford (1995), choice-certainty theory can be integrated with dissonance theory.4
On the basis of a fair amount of research dealing with the issue, I conclude that when faced with a prospective choice, people will be motivated to avoid dissonance anticipated as a consequence of making a decision. They will avoid information expected to increase dissonance after the decision, and they will spread the attractiveness of the alternatives, in private, to avoid the dissonance that would otherwise be expected to occur after the choice. That is contrary to what Festinger assumed. The avoidance of dissonance is an important aspect of the theory, which has been neglected in most recent work on dissonance. It needs to be addressed in any revision of the theory.
HOW DISSONANCE IS DETERMINED BY DESIRED CONSEQUENCES AND IMPORTANCE OF COGNITIONS
There are some other important aspects of the 1957 version that I feel have been ignored or have not received sufficient attention. One is that Festinger said that “motivations and desired consequences may also be factors in determining whether or not two elements are dissonant” (p. 13). Another is Festinger’s assumption that if two elements are dissonant, the magnitude of dissonance will be a function of the importance of the elements (p. 16). Those neglected aspects of the theory are involved in a proposal I make for a change in the definition of dissonance.5 The change I will propose also stems from some aspects of the theory that have concerned me for a long time, ever since the time I was working as Festinger’s research assistant in the period 1954—1957, when he was developing the theory.
One thing about the 1957 version that has always troubled me is that the definition of dissonance disregards the existence of all the other cognitive elements that are relevant to either or both of the two under consideration and simply deals with those two alone (p. 13). For a long time, I have thought that if motivations and desired consequences determine whether there is dissonance, then one would not simply consider the two cognitive elements alone. Desired consequences would seem to constitute a third cognition that must be taken into account.
Immediately after Festinger (1957) made the statement about the role of desired consequences in determining whether cognitions are dissonant, he gave the following example:
A person in a card game might continue playing and losing money while knowing that the others in the game are professional gamblers. This latter knowledge would be dissonant with his cognition about his behavior, namely, continuing to play. But it should be clear that to specify the relation as dissonant is to assume (plausibly enough) that the person involved wants to win. If for some strange reason this person wants to lose, this relation would be consonant. (p. 13)
In the situation in Festinger’s example, it would seem reasonable to assume that if the person wants to win, there would be a cognition that the person wants to win. If the person wants to win but for some strange reason does not have the cognition that he wants to win, it would seem that the playing the game would not be dissonant with the knowledge that the others in the game are professional gamblers. Taking desired consequences into account when determining the presence of dissonance seems to require considering more than just two cognitive elements alone.
Another aspect of the theory that has long bothered me has to do with the assumption that elements either follow or do not follow from one another but that there is no variation in the degree to which one element follows from another. I recall an argument with Festinger about that issue. It occurred when I was working on the study that was to be my doctoral dissertation. That study examined the effect of resistance to temptation on changes in attitudes toward cheating (Mills, 1958). One aspect of that study varied restraints against cheating by making it seem highly likely that cheating would be detected or very unlikely that cheating would be detected.
I suggested to Festinger that the cognition that there was a high likelihood of being caught was more dissonant with cheating than the cognition that there was a low likelihood of being caught, and I went further than that and suggested that a 95% chance of being caught was more dissonant with cheating than a 55% chance of being caught. Festinger maintained that dissonance was either/or, that there was no degree to which a behavior followed from a cognition. He took the position that there were a larger number of cognitions dissonant with cheating if there was a 95% chance of being caught than if there was a 55% chance. As with every argument I ever had with Festinger, I did not win that argument.
Festinger did make an assumption in the 1957 version about degrees of dissonance between two cognitions, when he assumed that the magnitude of the dissonance between two cognitions depends on the importance of the cognitions. Unfortunately, the variable of importance of cognitions has not been given much emphasis in research and theorizing about dissonance. One possible reason is that Festinger (1957) did not say much about what he meant by importance beyond saying that “the more these elements are important to, or valued by the person, the greater will be the magnitude of a dissonant relation between them” (p. 16). Another possible reason, perhaps more consequential, is that Festinger did not do any studies varying importance.
Festinger often made his points by giving examples, which he was a master at constructing. He gave an example of reducing dissonance by reducing importance, when a habitual cigarette smoker has the belief that smoking is bad for health:
Our smoker, for example, could find out all about accidents and death rates in automobiles. Having then added the cognition that the danger of smoking is negligible compared to the danger he runs driving a car, his dissonance would also have been somewhat reduced. Here the total dissonance is reduced by reducing the importance of the existing dissonance. (Festinger, 1957, p. 22)
Like most of Festinger’s examples, this one seems clear, but it does not provide a general conceptualization of the variable of importance.
The variable of importance seems to be involved with the consequences of the behavior for things that are valued or desired or with the consequences of the behavior for things regarded as undesirable. Now that appears similar to the idea that motivation and desired consequences determine dissonance, except that the variable of importance varies the magnitude of the dissonance. It seems reasonable to assume that a behavior follows from a cognition about a consequence of the behavior, if there is a desire for that consequence, and that the obverse (or opposite) of the behavior follows from a cognition about a consequence of the behavior, if there is a desire to avoid that consequence. Such a formulation would incorporate desired (and undesired) consequences within the definition of dissonance.
One result of incorporating the desirability (or undesirability) of consequences of the behavior within the definition of dissonance is that the determination of dissonance would involve three cognitions: (a) a cognition about the behavior, (b) a cognition about a consequence of the behavior, and (c) a cognition about the desirability (or undesirability) of the consequence. Using three cognitions when specifying dissonance would make dissonance theory more like balance theory. Some years ago, Insko and collaborators (Insko, Worchel, Folger, & Kutkus, 1975) proposed a balance theory interpretation of dissonance, although in somewhat different terms. In their recent constraint-satisfaction neural network model of dissonance, Shultz and Lepper (1996) made assumptions that they noted were reminiscent of cognitive balance theory.
Sticking as closely as possible to the language of the 1957 version, my suggestion for specifying dissonance is that a behavior follows from (is consonant with) a cognition about a consequence of the behavior, if the consequence is desirable, and a behavior does not follow from (is dissonant with) a cognition about a consequence of the behavior, if the consequence is undesirable. Thinking about dissonance in terms of three cognitions clarifies why dissonance is sometimes reduced by changing an attitude and sometimes by changing a belief. A smoker can reduce dissonance by believing there is no evidence smoking causes cancer or by downplaying the undesirability of having cancer, although the latter is less likely because negative attitudes toward cancer are very strong and resistant to change. Most dissonance research has focused on reducing dissonance by changing an attitude (e.g., disliking someone to whom one has made disparaging comments; Davis & Jones, 1960) and has used situations in which it was very difficult to reduce dissonance by changing a belief (e.g., by denying that one’s comments were disparaging to the other).
The idea that a behavior is dissonant with a cognition about a consequence of the behavior if there is a desire to avoid the consequence sounds very similar to the assumption of Cooper and Fazio (1984) about the necessity of aversive consequences for the arousal of dissonance. However, the similarity is not quite so clear when one looks at their definition of an aversive event as “an event that blocks one’s self-interest or an event that one would rather not have occur” (Cooper & Fazio, 1984, p. 232). Their definition of an aversive event has two distinct aspects. An event may be one that one would rather not have occur, but it may not be an event that blocks one’s self-interest.
It seems clear that the consequence of the behavior has to be undesirable, something the person is motivated to avoid, for dissonance to be aroused. Using ideas similar to the general version of balance proposed by Rosenberg and Abelson (1960), it is possible to state in general terms what constitutes a desirable or undesirable consequence. A consequence is desirable if it has a positive effect on or promotes something that is positively evaluated or if it has a negative effect on or negates something that is negatively evaluated. A consequence is undesirable if it promotes something that is negatively evaluated or if it negates something that is positively evaluated.
As Brehm and Cohen (1962) emphasized, dissonance trades on the frustration of other motives. But whether the self has to be involved is another matter. The recent research by Harmon-Jones, Brehm, Greenberg, Simon, and Nelson (1996), showing that a public statement of one’s position is not necessary for the arousal of dissonance, indicates that blocking one’s self-interest or harming another person is not required for dissonance arousal. The way in which the self is involved in dissonance is an important matter that I return to later.
Explicitly including desired (and undesired) consequences in the definition of dissonance has other advantages. It includes importance or value directly in the specification of dissonance. In addition, it makes it easy to take the next step of allowing for degrees of dissonance between cognitions, depending on the strength of the motivation to avoid the particular consequence. What determines importance is not just that a behavior has consequences for something that is desirable or undesirable but also the degree of likelihood that a behavior has the consequence and the degree to which the consequence is desirable or undesirable.
I propose that the degree to which the obverse (or opposite) of a behavior follows from a cognition about a consequence depends on the amount of the desire to avoid the consequence. Dissonance should depend on the degree of certainty that a behavior will lead to a consequence and on the degree to which the consequence is desirable or undesirable. To give some examples, there should be more dissonance if a smoker thinks there is a 100% probability that smoking causes cancer than if she or he thinks the chance that smoking causes cancer is only 1%. There should be more dissonance if a smoker believes that smoking will cause cancer within 5 years than if she or he believes smoking will cause cancer within 50 years. There should be more dissonance if a smoker believes smoking will cause large inoperable tumors than if she or he thinks it will cause only very small tumors that are easily removed. And, of course, there should be more dissonance for a smoker to believe that smoking causes cancer than to believe that smoking causes something less undesirable, such as discolored teeth.
This formulation of dissonance in terms of the degree to which a behavior follows from a consequence and the desirability of the consequence has another advantage. It avoids the awkwardness of the original version that requires that degrees of dissonance always be described in terms of the number of consonant and dissonant cognitions. In some cases, such as the amount of money received for convincing someone a boring task is enjoyable (Festinger & Carlsmith, 1959), it is not unreasonable to talk about a person receiving $20 as having a larger number of consonant cognitions than a person receiving $1. But in other cases, such as the amount of effort involved in engaging in an action, it is rather awkward to talk in terms of number of cognitions. It is strained to say that a highly effortful action involves more cognitions than one involving moderate effort. That kind of awkward usage can be avoided if one thinks of dissonance in terms of the strength of the relationship between the cognitive elements as opposed to simply counting the number of consonant versus dissonant elements.
The formulation which I propose which explicitly takes into account the desirability of consequences and thus, in effect, the importance of the elements, has another potential advantage. It could help to reconcile the current theories that place emphasis on the role of the self in dissonance processes (Thibodeau & Aronson, 1992) with the formulation of dissonance in the original 1957 version. When considering the role of the self, I am going to assume that there is a distinction between the motivation of the person and the role of the self. If lower animals such as rats have dissonance and reduce dissonance as Festinger believed (Lawrence & Festinger, 1962), then the self does not have to be involved for dissonance to occur. We do not assume that rats have self-concepts. But, of course, rats do have motivations. For rats, there are desired and undesired consequences, such as the presence of food or electric shock.
In the case of humans, in addition to motives that do not require invoking the concept of the self, there are, of course, self-concepts, self-esteem, self-beliefs, and self-attitudes. All of these self-aspects have important implications for motivation. In my interpretation of dissonance, the engagement of the self may increase desired consequences or undesired consequences. If we use the definition of dissonance that allows for degrees of dissonance depending on the degree of the consequences and the degree of the desirability or undesirability of the consequences, we can unify the current thinking about the role of the self in dissonance with the general framework of the original version of the theory.
In some cases, self-relevance may increase dissonance by increasing the motivation to avoid certain consequences and, in other cases, a focus on the self may reduce dissonance by reducing the motivation to avoid the consequences. Some important work recently by Simon, Greenberg, and Brehm (1995) has provided evidence that the self-affirmation manipulation of Steele and Liu (1983), which has been shown to reduce dissonance, can be explained in terms of the lowering of the importance of the dissonant action. The present formulation, which explicitly includes importance in the definition of dissonance, can explain why sometimes dissonance is greater when the self is involved and also why it is less under other circumstances when the self is involved.
Obviously, my proposal (that the degree of dissonance depends on the degree to which the opposite of the behavior follows from the cognitions about the degree of the consequences of the behavior and the degree of desirability or undesirability of the consequences) is something that goes beyond Festinger’s original definition of dissonance. However, I feel it is within the spirit of Festinger’s original 1957 version and fits with many of the examples that he gave. There are many complexities in this formulation that need to be addressed, which could pose formidable problems. How does one measure the degree to which a behavior follows from a consequence? How does one measure the degree of desirability or undesirability of a consequence? A number of important issues need to be considered.
CONCLUSION
I am frequently reminded of the statement attributed to Einstein that everything should be made as simple as possible, but no simpler. This formulation of dissonance is not as simple as the 1957 version, which I believe was deceptively simple. However, I believe it provides a more precise and accurate account of the domain of dissonance research. If that is true, it should prove useful in providing new insights. Hopefully, it will promote the continued development of dissonance theory. I believe we honor Festinger’s contribution by continuing to develop the theory.
REFERENCES
Brehm, J. W., & Cohen, A. R. (1962). Explorations in cognitive dissonance. New York, NY: Wiley. http://dx.doi.org/10.1037/11622-000
Brounstein, R. J., Ostrove, N., & Mills, J. (1979). Divergence of private evaluations of alternatives prior to a choice. Journal of Personality and Social Psychology, 37, 1957—1965. http://dx.doi.org/10.1037/0022-3514.37.11.1957
Cooper, J., & Fazio, R. H. (1984). A new look at dissonance theory. In L. Berkowitz (Ed.), Advances in experimental social psychology (Vol. 17, pp. 220—262). New York, NY: Academic Press.
Davis, K. E., & Jones, E. E. (1960). Changes in interpersonal perception as a means of reducing cognitive dissonance. Journal of Abnormal and Social Psychology, 61, 402—410. http://dx.doi.org/10.1037/h0044214
Ehrlich, D., Guttman, I., Schonbach, P., & Mills, J. (1957). Postdecision exposure to relevant information. Journal of Abnormal and Social Psychology, 54, 98—102. http://dx.doi.org/10.1037/h0042740
Festinger, L. (1957). A theory of cognitive dissonance. Evanston, IL: Row, Peterson.
Festinger, L. (1964). Conflict, decision, and dissonance. Palo Alto, CA: Stanford University Press.
Festinger, L., & Carlsmith, J. M. (1959). Cognitive consequences of forced compliance. Journal of Abnormal and Social Psychology, 58, 203—210. http://dx.doi.org/10.1037/h0041593
Harmon-Jones, E., Brehm, J., Greenberg, J., Simon, L., & Nelson, D. E. (1996). Evidence that the production of aversive consequences is not necessary to create cognitive dissonance. Journal of Personality and Social Psychology, 70, 5—16. http://dx.doi.org/10.1037/0022-3514.70.1.5
Hovland, C. I. (1959). Reconciling conflicting results derived from experimental and survey studies of attitude change. American Psychologist, 14, 8—17. http://dx.doi.org/10.1037/h0042210
Insko, C. A., Worchel, S., Folger, R., & Kutkus, A. (1975). A balance theory interpretation of dissonance. Psychological Review, 82, 169—183. http://dx.doi.org/10.1037/0033-295X.82.3.169
Lawrence, D. F. I., & Festinger, L. (1962). Deterrents and reinforcement. Palo Alto, CA: Stanford University Press.
Mills, J. (1958). Changes in moral attitudes following temptation. Journal of Personality, 26, 517—531. http://dx.doi.org/10.1111/j.1467-6494.1958.tb02349.x
Mills, J. (1965a). Avoidance of dissonant information. Journal of Personality and Social Psychology, 2, 589—593. http://dx.doi.org/10.1037/h0022523
Mills, J. (1965b). Effect of certainty about a decision upon postdecision exposure to consonant and dissonant information. Journal of Personality and Social Psychology, 2, 749—752. http://dx.doi.org/10.1037/h0022676
Mills, J. (1965c). The effect of certainty on exposure to information prior to commitment. Journal of Experimental Social Psychology, 1, 348—355. http://dx.doi.org/10.1016/0022-1031(65)90014-4
Mills, J. (1968). Interest in supporting and discrepant information. In R. P. Abelson, E. Aronson, W. J. McGuire, T. M. Newcomb, M. J. Rosenberg, & P. H. Tannenbaum (Eds.), Theories of cognitive consistency: A source book. Skokie, IL: Rand McNally.
Mills, J., Aronson, E., & Robinson, H. (1959). Selectivity in exposure to information. Journal of Abnormal and Social Psychology, 59, 250—253. http://dx.doi.org/10.1037/h0042162
Mills, J., & Ford, T. E. (1995). Effects of importance of a prospective choice on private and public evaluations of the alternatives. Personality and Social Psychology Bulletin, 21, 256—266. http://dx.doi.org/10.1177/0146167295213007
Mills, J., & Jellison, J. M. (1968). Avoidance of discrepant information prior to commitment. Journal of Personality and Social Psychology, 8, 59—62. http://dx.doi.org/10.1037/h0025319
Mills, J., & O’Neal, E. (1971). Anticipated choice, attention, and halo effect. Psychonomic Science, 22, 231—233. http://dx.doi.org/10.3758/BF03332586
Mills, J., & Ross, A. (1964). Effects of commitment and certainty upon interest in supporting information. Journal of Abnormal and Social Psychology, 68, 552—555. http://dx.doi.org/10.1037/h0043278
O’Neal, E. (1971). Influence of future choice importance and arousal upon the halo effect. Journal of Personality and Social Psychology, 19, 334—340. http://dx.doi.org/10.1037/h0031466
O’Neal, E., & Mills, J. (1969). The influence of anticipated choice on the halo effect. Journal of Experimental Social Psychology, 5, 347—351. http://dx.doi.org/10.1016/0022-1031(69)90059-6
Rosenberg, M. J., & Abelson, R. R. (1960). An analysis of cognitive balancing. In C. I. Hovland & M. J. Rosenberg (Eds.), Attitude organization and change (pp. 112—163). New Haven, CT: Yale University Press.
Shultz, T. R., & Lepper, M. R. (1996). Cognitive dissonance reduction as constraint satisfaction. Psychological Review, 103, 219—240. http://dx.doi.org/10.1037/0033-295X.103.2.219
Simon, L., Greenberg, J., & Brehm, J. (1995). Trivialization: The forgotten mode of dissonance reduction. Journal of Personality and Social Psychology, 68, 247—260. http://dx.doi.org/10.1037/0022-3514.68.2.247
Steele, C. M., & Liu, T. J. (1983). Dissonance processes as self-affirmation. Journal of Personality and Social Psychology, 45, 5—19. http://dx.doi.org/10.1037/0022-3514.45.1.5
Thibodeau, R., & Aronson, E. (1992). Taking a closer look: Reasserting the role of the self-concept in dissonance theory. Personality and Social Psychology Bulletin, 18, 591—602. http://dx.doi.org/10.1177/0146167292185010
Zajonc, R. B. (1990). Obituary: Leon Festinger (1919—1989). American Psychologist, 45, 661—662. http://dx.doi.org/10.1037/h0091620
1The term obverse does not seem to have been used in accord with the dictionary definition of obverse as a proposition inferred immediately from another by denying the opposite of that which the given proposition affirms, for example, the obverse of “all a is b” is “no a is not b.”
2The interpretation in terms of dissonance was relegated to a footnote in the article by Mills and Ford, in accordance with the wishes of the article’s editor.
3The assumption that the magnitude of avoidance of dissonance is determined by the amount of dissonance otherwise expected to occur is separate and distinct from the assumption questioned earlier that the magnitude of avoidance of dissonance is determined by the amount of existing dissonance.
4The mention occurred in a footnote about the dissonance interpretation.
5I was stimulated to consider the role of importance in dissonance by conversations with Haruki Sakai when he visited Maryland during the summer of 1995. Readers can find an essay by Sakai in the first edition of this volume (Chapter 11, “A Multiplicative Power-Function Model of Cognitive Dissonance: Toward an Integrated Theory of Cognition, Emotion, and Behavior After Leon Festinger”).