Moving Beyond Attitude Change in the Study of Dissonance-Related Processes - Mathematical Models, Neural Activations, and Affective Responses

Cognitive Dissonance: Reexamining a Pivotal Theory in Psychology - Eddie Harmon-Jones 2019

Moving Beyond Attitude Change in the Study of Dissonance-Related Processes
Mathematical Models, Neural Activations, and Affective Responses

An Update on the Role of Discomfort

Patricia G. Devine, John M. Tauer, Kenneth E. Barron, Andrew J. Elliot, Kristen M. Vance, and Eddie Harmon-Jones

In 1957, Leon Festinger outlined a straightforward set of assumptions concerning the motivational underpinnings of cognitive dissonance. Festinger posited that (a) an inconsistency between cognitions created an uncomfortable psychological tension state, and (b) people would be motivated to reduce the tension by implementing some change that would restore consonance among the inconsistent elements. Festinger’s theory is essentially a process model, as represented in Figure 12.1. In Festinger’s model, inconsistency between cognitions (A) leads to dissonance (B), an uncomfortable psychological tension and arousal state that the person will be motivated to reduce. According to the model, the motivational properties of dissonance will lead to a dissonance-reduction strategy (C), which, if effective, will alleviate dissonance (D).

FIGURE 12.1. Schematic of Festinger’s (1957) Process Model of Dissonance


Much of the imaginative work of dissonance researchers has been devoted to documenting the motivational properties of dissonance, and an enormous amount of evidence consistent with the predictions of dissonance theory has been obtained. For example, hundreds of studies have shown that, when people engage in counterattitudinal behaviors (A), ostensibly effective dissonance-reduction strategies, such as attitude change, are implemented (C) (see Cooper & Fazio, 1984, for a review).1 However, direct evidence regarding the arousal (B) and reduction (D) of dissonance has been more elusive. Given the extant evidence, the reasoning about dissonance arousal and its reduction has been necessarily indirect. The logic is as follows: Because dissonance-reduction strategies (e.g., attitude change) are implemented after a procedure designed to create inconsistencies between attitudes and behavior, it has been assumed that (a) dissonance was created, (b) dissonance motivated the use of the reduction strategy, and (c) the reduction strategy was successful in alleviating the dissonance. Although the well-established attitude-change findings are clearly consistent with theory-derived predictions, they do not provide direct tests of the process assumptions of dissonance theory. In short, the part of dissonance theory that has proven rather elusive has been the core assumption that the tension created is psychologically distressing, and that this distress is alleviated after the implementation of a dissonance-reduction strategy.

In this chapter, we argue that outcome measures such as attitude change or other ostensibly effective dissonance-reduction strategies (e.g., bolstering, self-affirmation, trivialization) cannot provide this type of evidence because such measures are silent regarding underlying processes. Thus, the methods most commonly used to test dissonance assumptions are limited in what they can reveal about the nature of the dissonance and whether dissonance-reduction strategies are effective in alleviating dissonance-related distress. Our position is that dissonance theorists have asked too much of attitude change (and other outcome measures). We suggest that more thorough and complete testing of dissonance theory may be possible if we expand our methodological tools in the assessment of dissonance. To this end, we offer one such tool and provide evidence supporting the efficacy of it as a measure that is sensitive to both dissonance induction (B) and reduction (D). Before introducing the measure, however, we review the historical approaches in which attitude change plays a central role.


Historically, the induced-compliance paradigm has been the most frequently used paradigm for studying dissonance hypotheses (see Cooper & Fazio, 1984). In this paradigm, participants are induced to freely choose to advocate a counterattitudinal position, which theoretically sets the stage for creating dissonance (commonly referred to as the high-choice condition). As a comparison condition, other participants are assigned to advocate a counterattitudinal position (commonly referred to as the low-choice condition); these participants theoretically do not experience dissonance, because they had no choice regarding the position they would advocate. After the counterattitudinal behavior, attitudes are assessed for participants in both choice conditions. In keeping with dissonance theory predictions, studies have repeatedly shown that high-choice participants show greater attitude change than their low-choice counterparts. Such findings have been interpreted to suggest that the dissonance experience is unpleasant and that it motivates attitude change that presumably alleviates dissonance-related distress. In addition, it has often been assumed that these attempts to reduce dissonance are successful. There are clearly some strengths to this empirical strategy.

The Good

Attitude change as the operational definition of dissonance reduction has served dissonance researchers well and yielded key insights regarding many aspects of cognitive dissonance. In the absence of any direct indicators of dissonance, testing outcomes that are consistent with the theory (e.g., attitude change among only high-choice participants after a counterattitudinal behavior) is a sensible and useful empirical strategy. Indeed, we are not suggesting that examining attitude change is in any way wrong or naive. Rather, we are suggesting that our overall understanding of the dynamics of dissonance processes may be improved by developing measures that can be used in conjunction with attitude change (and other outcome measures) to more fully reveal dissonance-related processes.

One of the primary goals early in the history of dissonance research was to demonstrate that the dominant theoretical paradigm in psychology at the time, reinforcement theory, could not account for the provocative findings generated in the tradition of dissonance experiments (see Aronson, 1992, for more detail). As such, the focus was on producing outcomes that were not readily interpretable from a reinforcement perspective. As an illustrative example, consider Festinger and Carlsmith’s (1959) classic study in which participants who chose to convince an unsuspecting participant (actually a confederate) that a boring task was actually interesting came to like the task more if they were offered low ($1) compared with high ($20) compensation for their efforts. Thus, early on, the goal of the research was to produce outcomes that were consistent with dissonance theory and not easily handled by alternative theoretical accounts. It was assumed, rather than tested, that the processes outlined by Festinger (1957) were responsible for the outcomes.

As dissonance research moved forward, however, issues arose concerning whether the motivational assumptions outlined by Festinger (1957) were responsible for the observed outcomes (Bem, 1967; Chapanis & Chapanis, 1964; Tedeschi, Schlenker, & Bonoma, 1971) that highlighted the need for evidence beyond attitude change to support dissonance interpretations for these outcomes. In the absence of direct measures of dissonance, these issues were, to say the least, challenging for dissonance theorists. Such theoretical and empirical challenges, however, ultimately served to showcase the cleverness and ingenuity of a generation of dissonance researchers, who developed compelling ways to circumvent the fact that there were no good direct ways to measure dissonance. This was perhaps the most positive by-product of the need to rely on indirect assessment of dissonance. The creativity of these dissonance theorists served to inspire, and continues to inspire, subsequent generations of social psychologists both within and beyond dissonance research.

Consider, as just one of many possible examples, the creative solution to the problem of having no direct measure of dissonance evidenced in Zanna and Cooper’s (1974) work on the arousal component of dissonance. Although they could not measure arousal directly, their theoretical resourcefulness led them to adopt a misattribution approach, which drew heavily on Schachter and Singer’s (1962) two-factor theory of emotion in characterizing dissonance as an arousal state open to various cognitive labels. Zanna and Cooper reasoned that participants who freely chose to advocate a counterattitudinal position would not change their attitude if given the opportunity to misattribute their arousal.2 In keeping with their predictions, high-choice participants given a placebo that would ostensibly make them feel tense changed their attitudes less than their high-choice counterparts, who had ingested a drug that supposedly would make them feel relaxed. Zanna and Cooper’s study, along with a variety of conceptual replications (see review by Fazio & Cooper, 1983), makes a strong case for the position that dissonance has arousal properties. The key point to be emphasized here is that in the absence of any direct way to assess the arousal component of dissonance, Zanna and Cooper used their theoretical resourcefulness and empirical imaginativeness to circumvent the problem. However, there are also some limitations to the emphasis on arousal that have created interpretational difficulties, if not theory-damning concerns, for dissonance theorists through the years.

The Bad

Although the field got around the fact that there was not a direct indicator of dissonance (i.e., unpleasant tension and arousal) and progress was made, we argue that the field asked too much of attitude change for informing the conceptual analysis of dissonance. That is, attitude change was the indicator that dissonance was induced and reduced. Thus, the fact that high-choice participants changed their attitudes more than low-choice participants was taken as evidence that dissonance was experienced among the high-choice participants. Attitude change also served the function of telling researchers, by its absence, that dissonance was no longer present (although, as will become clear later, this may not always be a valid inference). For example, when people implement an ostensibly effective dissonance-reduction strategy, such as bolstering (Sherman & Gorkin, 1980), self-affirmation (Steele & Liu, 1983), or trivialization (Simon, Greenberg, & Brehm, 1995) and then attitudes do not change, it is assumed that dissonance motivation is no longer present. In short, attitude change has been the primary indicator that dissonance was created and that it was reduced.

We suggest that attitude change is at best an indirect indicator of whether dissonance has been induced or reduced. This observation is, of course, not novel. Overreliance on attitude change as an indicator of dissonance induction and reduction has created interpretational difficulties in many studies in which results did not conform with expected dissonance outcomes. For example, when attitude change is not observed in the induced-compliance paradigm, what can be concluded? Was dissonance not successfully created (e.g., perhaps one’s procedure was flawed)? Did participants find alternative ways to alleviate their discomfort before attitudes were assessed? Was attitude change not a viable strategy for all participants? Did attitude change reduce dissonance for those who used this strategy? These are crucial questions and ones that attitude change cannot answer.

To illustrate the conceptual ambiguities associated with the use of attitude change as the sole indicator of dissonance, consider a classic study by Cooper and Worchel (1970). The goal of the study was to demonstrate that advocating a counterattitudinal position does not lead to the arousal of cognitive dissonance unless that advocacy results in undesirable consequences. In this study, participants were asked to perform an extremely dull task (cf. Festinger & Carlsmith, 1959) and were offered varying incentives for telling a waiting participant (actually a confederate) that the task was interesting and enjoyable. Before reporting their attitudes toward the task, half of the participants learned that they had successfully convinced the waiting participant that the task was interesting. The other half learned that the participant still believed that the task would be dull.

Cooper and Worchel’s (1970) findings revealed that only those participants who believed that they had succeeded in convincing the waiting participant and complied for a small incentive came to believe the task was interesting (e.g., showed the classic dissonance-induced attitude shift). Cooper and Worchel suggested that those who were unsuccessful in convincing the participant (as well as those who received a sufficient incentive) did not change their attitudes because dissonance was not aroused in these participants. We would argue, however, that sole reliance on attitude change in this paradigm cannot reveal whether these participants experienced dissonance. It is at least conceptually possible that dissonance was created for the low-incentive participants who learned that their efforts to deceive the waiting participant failed but was reduced before participants’ attitudes were assessed. Simply agreeing to the counterattitudinal behavior (i.e., to deceive the waiting participant) would be sufficient to induce dissonance (e.g., Elliot & Devine, 1994; Rabbi, Brehm, & Cohen, 1959). Indeed, learning that they had been unsuccessful in their deception efforts may have been sufficient to alleviate any dissonance-related distress that the deception caused them.

Similar ambiguities arise in other paradigms (with other theoretical agendas) as well. For example, Steele (1988; Steele & Liu, 1983) suggested that classic dissonance manipulations, such as the induced-compliance paradigm, serve to threaten people’s global sense of self-integrity. He argued that to alleviate the discomfort, one need not make adjustments (e.g., attitude change) to the cognitions directly involved in the inconsistency but instead could engage in some activity that might restore (or reaffirm) one’s global sense of self-integrity. To support this logic, Steele and Liu had participants write a counterattitudinal essay. Then, before reporting their attitudes, participants filled out a value-affirming scale on a dimension that was either important or unimportant to their self-identities. In keeping with Steele’s theory, attitude change was observed only among participants for whom the self-affirmation opportunity was not important to their self-identities. However, Steele noted a couple of alternative explanations for the attitude data in Steele and Liu’s studies, which suggest that attitude change may not be an adequate measure to demonstrate that dissonance was alleviated. He suggested that the important self-affirmations may have bolstered or frozen participants’ initial attitudes, resulting in little attitude change. Thus, it appears that in the dissonance literature, the absence of a more direct measure of dissonance makes it difficult to make strong inferences about the presence or absence of dissonance.

Note that the studies reviewed in this section are not the only studies challenged by interpretational ambiguities that derive from having to reason indirectly about the arousal and reduction of dissonance. These studies were selected to illustrate the interpretational difficulties that arise under such circumstances. The ambiguities are clearly evident in other dissonance research as well. Consider, for example, the difficulties that could arise in a study if some participants’ responses fit the classic dissonance attitude-change pattern, but other participants’ responses did not. What is the researcher to make of such variability? Was dissonance successfully created for only some of the participants? Or is it possible that attitude change was not a dissonance-reduction option for some people (e.g., people who are highly committed to their attitudes; Hardyck & Kardush, 1968)? Similarly, inconsistent findings (e.g., failure to reliably obtain preference for attitude-consistent information in the selective-exposure literature) proved difficult to resolve in the absence of some indicator that provided independent evidence that dissonance was successfully induced (Cialdini, Petty, & Cacioppo, 1981).

In short, although attitude change can be informative and has been useful in providing support for dissonance theory, it simply cannot rule out alternative explanations (i.e., dissonance was never created; dissonance was created, but not reduced; dissonance was reduced through some other strategy besides attitude change). And, finally, virtually no evidence exists to suggest that if attitude change occurred, participants felt better—that any discomfort that had been created by the dissonance-induction procedure was alleviated. This is perhaps the issue that has received the least empirical attention in the dissonance literature and one that, as illustrated below, raises some questions about the effectiveness of attitude change in explaining the mechanisms underlying the dissonance process.

The Ugly

According to dissonance theory, attitude change was the outcome of a presumed motivational state. The field generally, and dissonance researchers more specifically, was ultimately dissatisfied with having to rely exclusively on indirect evidence for the motivational state. In response to this set of circumstances, an exciting line of research ensued, the goal of which was to provide evidence that dissonance was at least arousing, if not directly psychologically unpleasant. Several studies attempted to show the arousal properties of dissonance by using indirect research techniques, such as incidental retention, response competition, and misattribution of arousal (e.g., Kiesler & Pallak, 1976; Pallak & Pittman, 1972; Zanna & Cooper, 1974).

Other investigators attempted to provide direct evidence regarding the arousal component of dissonance by measuring physiological changes that theoretically would accompany the dissonance. Studies provided evidence supporting the dissonance-as-physiological-arousal hypothesis (Elkin & Leippe, 1986; Harmon-Jones, Brehm, Greenberg, Simon, & Nelson, 1996; Losch & Cacioppo, 1990). For example, in a set of induced-compliance experiments, Elkin and Leippe’s (1986) participants displayed elevated galvanic skin responses (GSRs), as well as attitude change, after freely choosing to advocate a counterattitudinal position. Losch and Cacioppo (1990) obtained a similar pattern of results by means of a misattribution paradigm and frequency of nonspecific skin conductance responses (NS-SCRs) as the physiological indicator of dissonance arousal. Most recently, Harmon-Jones et al. (1996, Experiment 3) showed increased NS-SCRs in high- but not low-choice conditions. Taken together, these studies provide compelling evidence that there is a physiological component to the dissonance state. See also Chapters 4 and 11, this volume, for further evidence of physiological responses associated with cognitive dissonance processes.

With such strong evidence in hand for the arousal component of dissonance, Elkin and Leippe (1986; see also Harmon-Jones et al., 1996) also attempted to test the hypothesis that attitude change would lead to the reduction in the indicator of the presumed motivational state. Specifically, they tested the assumption that implementing a dissonance-reduction strategy, in this case attitude change, would lead to a reduction in dissonance. Elkin and Leippe argued that support for this assumption was critical to the contention that dissonance is a motivational state. Indeed, such studies provided the first opportunity to test the process assumptions suggested by Festinger’s (1957) model. However, these efforts to directly support the hypothesis that dissonance reduction occurs after attitude change were foiled. That is, although Elkin and Leippe were able to show that GSR increased reliably when participants freely chose to advocate a counterattitudinal position (A → B) and that these participants changed their attitudes in the direction of the position they advocated (B → C), they failed to show dissonance reduction in the form of a significant decrease in GSRs in the post-attitude-change period (C did not lead to D).

Such findings were ultimately troublesome for the theory and led Elkin and Leippe (1986) to call into question the veracity of Festinger’s (1957) assumptions regarding the motivational nature of dissonance arousal: “It is only though the arousal’s subsequent reduction that motivation can be implied, and we found no evidence that explicit attitude change reduced arousal. . . . Cognitive dissonance, then, may or may not be a motivational state” (p. 64). This type of statement was damning for the theory. Echoing these types of concerns, Wilder (1992) observed the following: “Questions of the reality of dissonance as a measurable tension or aversive state have always dogged the theory” (p. 352). In the wake of such disappointing findings regarding Festinger’s dissonance-reduction postulate as well as the need to rely on indirect reasoning regarding dissonance-related processing, empirical progress regarding dissonance-related hypotheses was largely stalled. However, more recent findings using an alternative measure of dissonance have paved the way to directly test Festinger’s assumptions regarding dissonance arousal and reduction. This measure serves as a “dissonance thermometer” of sorts and is sensitive to both the induction and reduction of dissonance-related distress.3 Used in conjunction with outcome measures such as attitude change, a measure of dissonance affect may yield insights concerning the dynamics of dissonance induction and reduction that would be impossible to obtain through outcome measures alone.


Both indirect (e.g., attitude change) and direct (e.g., arousal) indicators of dissonance have created difficulties for testing core assumptions of dissonance theory. A close reading of Festinger’s (1957) classic monograph, however, reveals that Festinger conceptualized dissonance in two distinguishable ways. He explicitly delineated psychological discomfort as a component of dissonance, and he alluded to dissonance as a bodily condition analogous to a tension or drive state like hunger (Croyle & Cooper, 1983). In Brehm and Cohen’s (1962) restatement of dissonance theory, they distinctly characterized dissonance as a state of arousal and focused extensively on its drivelike properties. As previously suggested, most research investigating the nature of dissonance, whether indirectly or directly, has primarily focused on Brehm and Cohen’s derived arousal component of dissonance rather than the psychological discomfort component identified by Festinger. Indeed, the discomfort component of dissonance has most often been assumed rather than measured directly.

We believe the field has been too narrowly focused on arousal as the motivational component of dissonance. Elliot and Devine (1994) argued that for a variety of reasons psychological discomfort may be the preferred component of dissonance to consider when exploring the dissonance-reduction process. First, physiological measures are imperfect measures of psychological processes, which may place limits on their use for tracking dissonance-reduction processes (Harmon-Jones & Beer, 2009). Second, Cooper and Fazio (1984) suggested that arousal plays only a distal role in dissonance reduction. That is, although arousal is posited to instigate attributional interpretation, Cooper and Fazio argued that it is the phenomenological experience of discomfort created by the attributional judgment that is the proximal motivational force encouraging the implementation of a dissonance-reduction strategy. Third, even if both arousal and discomfort serve proximal functions in the dissonance process, the time course of dissonance reduction may be different for arousal and psychological discomfort. For example, the dissonance-reduction experience may be marked by immediate psychological relief after the implementation of a dissonance-reduction strategy, followed by more gradual reduction of dissonance-based arousal. Under these circumstances, it may be more feasible to empirically demonstrate the alleviation of the psychological discomfort component of dissonance than a reduction of the arousal component, which may require a time sequence of unknown length. Empirical work (Elliot & Devine, 1994) designed specifically to assess the psychological component of dissonance suggests that efforts to assess the psychological component of dissonance may be revealing regarding both the nature of the dissonance experience and the motivational properties of the dissonance state.

Elliot and Devine (1994) argued that to the extent that dissonance is experienced as psychological discomfort (Festinger, 1957), it should be revealed as elevated feelings of discomfort (e.g., uncomfortable, uneasy, bothered) after a counterattitudinal advocacy (see also Devine, Monteith, Zuwerink, & Elliot, 1991). While acknowledging the potential limitations of self-report measures (Nisbett & Wilson, 1977), Elliot and Devine argued that feelings of discomfort could be sensitively assessed with a self-report measure of affect. To the extent that this self-report measure of discomfort was successful, it could serve as a dissonance thermometer sensitive to increases and decreases in psychological discomfort. Elliot and Devine argued further that in the induced-compliance paradigm, if attitude change was truly motivated by an effort to alleviate dissonance, discomfort feelings would be alleviated after the implementation of this reduction strategy. Thus, in an attempt to directly measure psychological discomfort after a counterattitudinal advocacy and its presumed alleviation following attitude change, Elliot and Devine varied the order of the placement of measures of affect and attitude in two induced-compliance studies. To the extent that such a pattern could be shown, it would provide the first direct evidence to support Festinger’s (1957) assertion that dissonance is fundamentally a motivational state. The minimum conditions needed to explore these issues are conditions that provide the opportunity to show that discomfort increases in the theoretically predicted circumstances (e.g., counterattitudinal advocacy), and that after an ostensibly effective dissonance-reduction strategy, the discomfort dissipates.

Initial Evidence for Dissonance as Psychological Discomfort

The utility of the affect measure for assessing dissonance is well illustrated in Elliot and Devine’s (1994) second study. In this study, all participants wrote a counterattitudinal essay arguing for a 10% tuition increase. The study had three conditions. The low-choice control provided baseline affect and attitude scores. The two high-choice conditions differed only in the order in which participants reported affect and attitude.

That is, after freely choosing to write and prepare the counterattitudinal essay, half of the high-choice participants immediately reported their attitude and then their affect. In the other high-choice condition, the affect measure preceded reports of attitude.4

Replicating the standard induced-compliance effect, participants in the high-choice conditions reported more attitude change than their counterparts in the low-choice condition (see Table 12.1). However, in keeping with Festinger’s (1957) theorizing about dissonance induction and reduction, discomfort feelings were elevated only in the condition in which affect was reported before attitudes were assessed. These data suggest that preparing the counterattitudinal essay led to feelings of discomfort. Of critical importance for supporting Festinger’s dissonance-reduction postulate, the level of discomfort feelings reported after participants were provided with an attitude-change opportunity dropped to baseline levels and did not differ from the affect reported by low-choice participants, who reported affect before preparing their counterattitudinal essays. Thus, it appears that feelings of discomfort dissipate after a dissonance-reduction strategy is implemented. Moreover, the affect findings are unique to discomfort feelings. The affect measure in Elliot and Devine’s (1994) research included items that would tap other forms of psychological distress (e.g., guilt or depressed affect) as well as positive affect. None of these other affect measures were influenced by the experimental manipulations.

TABLE 12.1. Mean Attitude Change and Discomfort Ratings as a Function of Experimental Condition


Experimental condition

High-choice affect—attitude

High-choice attitude—affect

Low choice

Attitude change








Note. Attitude-change values greater than 1 represent change in the direction favoring the proposed tuition increase. Discomfort values had a possible range of 1 to 7, with 7 representing the highest level of dissonance affect. Within each dependent measure, means with different subscripts differ significantly at p < .01 by the Fisher least significant difference test. Data from Elliot and Devine (1994, Study 2).

These findings are important in a number of respects. By focusing on and measuring the psychological discomfort component of dissonance, these data both clarify the nature of the dissonance experience and directly demonstrate the alleviation of dissonance on the implementation of an ostensibly effective dissonance-reduction strategy. Moreover, dissonance appears to be a distinct aversive feeling and not an undifferentiated arousal state. Perhaps most important, by demonstrating that attitude change was in the service of reducing the discomfort created by the counterattitudinal advocacy, Elliot and Devine (1994) obtained the first direct support for both dissonance induction and reduction.

Using the dissonance thermometer, Matz and Wood (2005) extended this research to test whether the social group could be a source of cognitive dissonance, as Festinger (1957) had predicted. That is, in a group, disagreement from others causes dissonance, and cognitive changes within the group toward consensus may reduce this dissonance discomfort. They found that individuals in a group with others who ostensibly disagreed with them reported feeling more dissonance discomfort than individuals in a group with others who agreed with them. Follow-up studies revealed that standard moderators of dissonance—lack of choice and opportunity to self-affirm—decreased the dissonance discomfort caused by group disagreement. Finally, a third study revealed that the dissonance created by group disagreement was reduced via several routes to achieve group consensus; these included persuading others in the group, changing one’s own position, and joining a different group that possessed similar attitudes (see Matz, Hofstedt, & Wood, 2008, for a replication and extension).

Dowsett, Semmler, Bray, Ankeny, and Chur-Hansen (2018) used a version of the dissonance thermometer to examine dissonance discomfort after individuals were exposed to either information about the life of a meat lamb or information about the nutritional benefits of meat. They predicted that being exposed to the information about the meat lamb would evoke dissonance because of the “meat paradox” (Loughnan, Haslam, & Bastian, 2010). That is, individuals often experience dissonance over eating meat (animals) because they also love animals. Results revealed that individuals (who were meat eaters) exposed to the information about the life of a meat lamb reported significantly more dissonance discomfort than individuals exposed to the information about the nutritional benefits of meat. Illustrating the complexity of dissonance reduction strategies, men in this condition had more favorable attitudes toward eating meat, whereas women had less favorable toward meat eating, compared with men and women in the condition exposed to information about the nutritional benefits of meat.

Levy, Harmon-Jones, and Harmon-Jones (2018) found evidence of psychological discomfort in response to a very simple cognitive inconsistency, in line with Festinger’s (1957) idea that dissonance occurs in a wide range of situations. In this research, participants simply read sentences that ended with expected words or unexpected words (e.g., “She couldn’t start her car without the right teeth”). And implicit, self-report (single item valence rating), and psychophysiological measures of negative affect were collected after exposure to these words. Results revealed that the dissonance-arousing unexpected words, as compared with the expected words, evoked more dissonance discomfort across all three types of measures.

Attitude Importance, Resistance to Change, and the Dissonance Thermometer

We have argued that in dissonance research, there has been an overreliance on attitude change as the key dissonance-reduction strategy (see also Simon et al., 1995; Steele, 1988). We believe that this strategy is, in part, responsible for a relative lack of attention to the study of dissonance-related processes involving important attitudes, which by definition may be more resistant to change than relatively unimportant attitudes (Boninger, Krosnick, Berent, & Fabrigar, 1995; Zuwerink & Devine, 1996). In the induced-compliance literature, when participants freely choose to advocate a counterattitudinal position, the cognition about this behavior, partly because of its immediate salience, is highly resistant to change (hence attitudes are changed to restore consonance). When a counterattitudinal behavior conflicts with a personally important attitude, both the attitude and the cognition about one’s behavior are likely to be highly resistant to change. When attitude change is not an option, other strategies for reducing dissonance-related distress would be required. A by-product of studying relatively unimportant attitudes is that there has been comparatively little focus on alternative dissonance-reduction strategies.

The issue of attitude importance did not escape the attention and theorizing of Festinger (1957) or the early dissonance researchers. Although it has long been assumed that attitude change may not be a viable dissonance-reduction strategy when the dissonance is associated with important attitudes, empirical attempts to explicate the role of attitude importance are scant. Theorists have articulated some of the challenges associated with studying important attitudes within the dissonance framework (e.g., Cooper & Mackie, 1983; Hardyck & Kardush, 1968; Pilisuk, 1968; Sherman & Gorkin, 1980). As noted by Sherman and Gorkin (1980), “when the original attitude is an especially strong and central one, involving a large degree of relevance and prior commitment, attitude change is unlikely” (p. 389). In response, Sherman and Gorkin examined a form of attitude bolstering, arguing that neither attitude change nor denying the behavior was possible.

In our work, we have sought to document the role of attitude importance in the dissonance process. Specifically, we have begun to investigate how attitude importance influences the nature of the dissonance experience and the types of dissonance-reduction strategies that may prove effective (or ineffective) in alleviating distress. In one study, using a counterattitudinal advocacy paradigm with recycling as the issue, we replicated the essential design of Elliot and Devine (1994) but included participants who varied in the self-reported importance of their recycling attitudes (Devine, Froning, & Elliot, 1995). That is, some of the participants reported that their recycling attitudes were highly personally important, whereas others, although equally in favor of recycling, indicated that their attitudes were less personally important. Low- and high-importance participants were then randomly assigned to one of three conditions. Following Elliot and Devine (1994), this study included a low-choice condition, to establish baseline affect and attitude scores, and two high-choice conditions that differed only in the order in which participants reported their affect and their attitude.

Under low choice, as can be seen in Table 12.2, both low- and high-importance participants reported little discomfort or attitude change.5 Under high choice, the effects were moderated by attitude importance. For low-importance participants, both the attitude change and the discomfort data replicated the findings of Elliot and Devine (1994). Specifically, attitude change for low-importance participants was elevated under both high-choice conditions; however, discomfort feelings were elevated only for low-importance, high-choice participants when affect was reported before attitudes were assessed. For high-importance participants, however, attitude change was not a viable dissonance-reduction strategy. In neither high-choice condition did high-importance participants change their attitudes against campus recycling. Moreover, their discomfort was elevated whether they reported attitude or affect first.

TABLE 12.2. Mean Attitude Change and Discomfort Ratings as a Function of Experimental Condition and Importance Level

Experimental condition

High-choice affect—attitude

High-choice attitude—affect

Low choice


Low imp

High imp

Low imp

High imp

Low imp

High imp

Attitude change














Note. Imp = importance. Attitude-change values greater than 1 represent change in the direction favoring reduction of recycling efforts. Discomfort values had a possible range of 1 to 7, with 7 representing the highest level of dissonance affect. Within each dependent measure, means with different subscripts differ significantly at p < .01 by the Fisher least significant difference test.

When attitude change is not viable, are there strategies that can be effective for alleviating dissonance-related distress? Two recent lines of research suggest an affirmative answer to this question. One line of work follows directly from the Devine et al. (1995) study examining dissonance and attitude change in people of varying levels of attitude importance. Because high-importance people continue to experience discomfort after the attitude-change opportunity, they will still be motivated to reduce their distress. Thus, in a follow-up study, Tauer and Devine (1998) replicated Devine et al.’s (1995) basic design, but they provided all participants with a subsequent alternative dissonance-reduction opportunity, in this case, altering the perception of the strength of their essay (cf. Scheier & Carver, 1980; Simon et al., 1995). Theoretically, individuals with attitudes of low importance, who are likely to change their attitudes and therefore be free from dissonance motivation, would not take advantage of this subsequent opportunity. However, individuals with attitudes of high importance would alter the perception of the strength of their essay because they continued to experience high levels of discomfort and thus dissonance motivation. To the extent that altering the perception of essay strength was an effective dissonance-reduction strategy for these individuals, they would experience a reduction of discomfort. This line of work is important because it will help to shed light on the effectiveness of different strategies and the conditions under which alternative strategies are most optimal.

In another study, Tauer, Devine, and Elliot (1998) garnered evidence that self-affirmations are effective at reducing dissonance for people low and high in attitude importance. Once again using the recycling issue, half of the low- and high-importance participants completed a self-affirmation task before reporting their affect; for the other half of the participants, the order of these tasks was reversed. The self-affirmation task involved generating four examples of times when they had demonstrated their most cherished characteristics (see Vance, Devine, & Barron, 1997). When the affect measure preceded the self-affirmation task, feelings of discomfort were elevated for both high- (M = 3.83) and low- (M = 2.94) importance participants. When the affect measure followed the self-affirmation task, feelings of discomfort for both high- (M = 3.09) and low- (M = 2.22) importance participants were reduced, although discomfort was still somewhat elevated for high-importance participants. It is also of interest to note that there was no evidence of attitude change for either high- or low-importance participants in this study. In all conditions, attitudes were assessed after the affect and self-affirmation opportunities. High-importance participants, of course, were not expected to show evidence of attitude change (cf. Devine et al., 1995). Of particular interest, however, was that low-importance participants, after having completed a self-affirmation task, did not change their attitudes.

In other research, Galinsky, Stone, and Cooper (2000) replicated the results of Elliot and Devine (1994) and also found evidence that suggested that self-affirmations decrease attitude change because they also decrease the negative affect associated with dissonance. Additionally, Holland, Meertens, and Van Vugt (2002, Study 2) induced dissonance by providing participants information that indicated they were prejudiced toward groups with whom that they do not want to be prejudiced (i.e., Dutch participants and Turkish and Moroccan groups). This information increased dissonance discomfort among participants low in self-esteem but not among participants high in self-esteem. These results are consistent with the self-affirmation theory prediction that high self-esteem serves as a resource or buffer against threatening, dissonance-arousing information. Taken together, these studies suggest that self-affirmation alleviates dissonance discomfort and, thus, reduces the motivational force that produces attitude change (Elliot & Devine, 1994; Festinger, 1957).


Several methodological and theoretical benefits accrue from having a measure that is both easy to implement and sensitive to dissonance induction as well as reduction. First, and most obviously, the measure can serve as a manipulation check for both dissonance induction and reduction. As noted previously, the absence of a manipulation check created interpretational difficulties for dissonance researchers when results did not conform to theoretical predictions. It was not possible to determine whether the experimental procedures had failed to evoke dissonance or whether the dissonance that had been created was reduced through the implementation of an alternative reduction strategy. Cialdini et al. (1981) suggested that research on selective exposure and dissonance was (temporarily) abandoned against a backdrop of seemingly inconsistent results. These inconsistencies proved difficult to interpret, in part due to the absence of an effective dissonance manipulation check. Use of the dissonance thermometer may facilitate progress in such areas.

Having an effective dissonance manipulation check may also provide efficient ways for validating new procedures for instilling dissonance, such as the hypocrisy procedure developed by Aronson and colleagues (see Aronson, 1992; Stone, Aronson, Crain, Winslow, & Fried, 1994). The goal of the hypocrisy manipulation is to show that dissonance can be created and can lead to dissonance-related cognitive and behavioral changes even when participants advocate a proattitudinal position (e.g., in favor of safe sex). In developing the technique, Stone et al. (1994), in the long tradition of dissonance research, relied on producing theory-consistent changes in behavior or attitudes to validate the hypocrisy procedure. We suggest that more efficient and potentially more informative progress can be made by using a measure like the dissonance thermometer during the validation process. For example, it would be immediately apparent if new procedures were effective if they led to increases in psychological discomfort.

In a similar fashion, the dissonance thermometer can be used to assess the efficacy of alternative dissonance-reduction strategies without relying exclusively on attitude change (see Tauer & Devine, 1998). The most often used strategy for validating the efficacy of alternative dissonance-reduction strategies (e.g., trivialization, self-affirmation) is to show that after implementing an ostensibly effective strategy, attitudes do not change (e.g., Simon et al., 1995; Steele & Liu, 1983, respectively). The dissonance thermometer can provide information on the efficacy of dissonance-reduction strategies much more directly (i.e., discomfort decreases) and may, as a result, enable new progress to be made in exploring alternative strategies for alleviating dissonance-related discomfort. Along these lines, Gosling, Denizeau, and Oberlé (2006) investigated denial of responsibility as a mode of dissonance reduction, and found that following counterattitudinal behavior, participants used the mode of discrepancy reduction that was first presented to them, regardless of whether it was attitude change, trivialization, or denial of responsibility. In a subsequent experiment, they found that the denial of responsibility reduced dissonance discomfort.

Finally, use of a dissonance manipulation check may ultimately permit the use of more efficient experimental designs. Historically, the dissonance literature has included low-choice conditions primarily to validate that little attitude change occurs when theoretically it should not. A direct measure of dissonance may obviate the need for such control conditions, at least in some circumstances.

Although the benefits of a dissonance manipulation check are considerable, some of the potentially more powerful benefits of a more direct dissonance measure may come in the form of improved theory testing and elaboration. As stated in the beginning of this chapter, Festinger (1957) proposed essentially a process model: A (inconsistency between cognitions) → B (dissonance) → C (some dissonance-reduction strategy) → D (alleviation of dissonance). However, empirical strategies to date have been largely limited to investigations in which A is created and C is observed. Although research investigating the arousal component of dissonance has attempted to explore the mediational processes suggested by the model, complications arise when arousal does not appear to dissipate after an ostensibly effective dissonance-reduction strategy is implemented (e.g., Elkin & Leippe, 1986). A measure of the psychological component of dissonance, which has been shown to be sensitive to dissonance induction and reduction, provides an additional tool for directly testing the process assumptions as they unfold over time. Specifically, a complete understanding of dissonance theory requires testing whether dissonance is created, whether a reduction opportunity is used, and whether dissonance reduction follows the implementation of an ostensibly effective dissonance-reduction strategy. Exploring these ideas calls for a mediational analysis that will allow researchers to directly test the process of dissonance induction and reduction.

Having a more direct measure of dissonance may ultimately help in addressing the long-standing debate concerning the necessary and sufficient conditions for dissonance arousal. Cooper and Fazio (1984) maintained that taking responsibility for the production of aversive consequences is necessary for the arousal of dissonance. However, in using a discomfort measure of dissonance, Harmon-Jones (2000) showed that the production of aversive consequences is not necessary to create dissonance-related discomfort. That is, when participants advocated a counterattitudinal position for which there could be no aversive consequences (i.e., the evidence of their counterattitudinal behavior is literally thrown away), discomfort feelings were elevated. Thus, although the production of aversive consequences may heighten the dissonance experience, Harmon-Jones’s research suggests that aversive consequences are not necessary. As these findings suggest, when used in conjunction with other measures, the dissonance thermometer may enable more precise evaluation of the plausibility of the core assumptions of the various revisions of dissonance theory proposed over the years.

A final set of issues that may be fruitful to explore with the dissonance thermometer concerns the qualitative nature of the affect associated with different types of cognitive inconsistencies and the possibility that not all dissonances are created equal. Because the affect measure was designed to be sensitive to both global and more specific forms of affective distress, it may play a role in elucidating theory and enable the field to address issues that have long proved vexing for dissonance theorists. These include assessing the magnitude of dissonance experienced and how people resolve dissonance when inconsistencies involve important, central, or self-defining attitudes.

For example, following in the tradition of Aronson’s (1968) reconceptualization of dissonance, Elliot and Devine (1994) suggested that the self may be implicated to varying degrees in the dissonance process. Specifically, the self-relevance of the threatened cognition may be critical in determining the qualitative nature of the affect experienced as a result of dissonance. Similarly, Aronson (1992) argued that dissonance-induction procedures, such as the hypocrisy manipulation, were likely to lead to feelings of guilt and not just global discomfort. The typical strategies used to validate hypocrisy-induced dissonance reduction (e.g., cognitive or behavioral change) are silent on the qualitative nature of the affect created by the manipulation. To the extent that the quality of affect experienced is important in determining whether particular strategies are likely to be effective in reducing dissonance-related distress, it will be important to establish precisely what type of affect is evoked by alternative procedures (see Vance et al., 1997). Our measure, which was developed with the goal of assessing various qualities of affect elicited in response to cognitive inconsistencies, is well suited to these tasks (see Devine et al., 1991).

For example, our work on the affective consequences of prejudice-related discrepancies (i.e., actual responses revealing more prejudice than is permitted by one’s nonprejudiced standards) has shown that violations of well-internalized, self-defining standards generate general negative affect (i.e., discomfort) and a more specific negative self-directed affect (i.e., guilt; see Devine et al., 1991). Violations of less internalized standards simply elicit more global negative affect. The dissonance thermometer was essential for detecting this difference in the experience of dissonance for attitudes that are more, compared with less, self-defining. Moreover, Vance and Devine (1997) used the dissonance thermometer after a hypocrisy manipulation relevant to nonprejudiced standards and behavior. Participants advocated the importance of treating Blacks in a nonprejudiced manner, then recalled times they had responded with prejudice. The combination of these experiences resulted in increased global discomfort, as well as elevated feelings of guilt and self-criticism. Similarly, Son Hing, Li, and Zanna (2002) found that when individuals who scored high in aversive racism were induced to experience hypocrisy about their prejudice, they felt more guilt and discomfort than a comparison condition. Moreover, after these individuals acted in a nonprejudiced manner, they felt less guilt and discomfort. These findings support Aronson’s (1992) general proposal about the consequences of hypocrisy but, perhaps more important, when combined with our previous findings, suggest that it may be prudent to think more completely about the specific qualities of affect associated with alternative dissonance manipulations.

Along these lines, research has found that individuals experiencing dissonance report feeling tense (induced compliance; Zanna & Cooper, 1974), mental discomfort and frustration (effort justification; Shaffer & Hendrick, 1974; induced compliance; Shaffer, 1975), bashful (induced compliance; Kidd & Berkowitz, 1976), and distressed, threatened, angry, and frustrated (belief disconfirmation; Burris, Harmon-Jones, & Tarpley, 1997). Thus, the words associated with the feeling state of dissonance may depend on the type of cognitive discrepancy aroused, as well as the emotion word vocabularies of the individuals investigated. For instance, individuals may report feeling more distressed when exposed to belief disconfirming information, but more regret after making difficult decisions. In addition, the topic of the counterattitudinal essay or belief disconfirmation may also influence the emotion words individuals feel most strongly. Consequently, it is not surprising that factor analyses of the negative affect reported in dissonance experiments (e.g., Gosling et al., 2006) do not always converge with the ones reported by Elliot and Devine (1994). As this evidence suggests, it appears that not all dissonances are created or resolved equally. The use of a direct measure of affect makes it possible to more fully explore these issues.


Are individuals always aware of their dissonance discomfort? Theory and research have suggested that individuals are not always consciously aware of their affective states (e.g., Winkielman & Berridge, 2004). Moreover, research using the misattribution paradigm (e.g., Zanna & Cooper, 1974) suggests that the source of the dissonance discomfort is not very well known to participants, because if the source were well known, they would not be able to misattribute their discomfort to another source. The misattribution paradigm relies on the individuals being less than perfectly aware of the source of their emotion. If individuals are not very aware of the source of their dissonance discomfort (e.g., their counterattitudinal essay), they may not be very aware of their discomfort. In other words, the dissonance discomfort (evoked in the lab) may be a subtle or vague feeling, and consequently at least some individuals may be unable to accurately report feeling it. If so, some situations or individuals may not provide evidence of dissonance discomfort on self-report measures.

Might the measurement of self-reported discomfort or negative affect influence the process of dissonance reduction? Yes, according to some research. For instance, Pyszczynski, Greenberg, Solomon, Sideris, and Stubing (1993) posited that the expression of negative affect may reduce the motivation to engage in discrepancy reduction (e.g., attitude change). They based this prediction on the idea that discrepancy reduction functions to protect individuals from the negative affective state of dissonance. If, however, individuals acknowledge and express their dissonance discomfort, they would be less motivated to engage in discrepancy reduction. In one high-choice condition of the Pyszczynski et al. (1993) experiment, participants were instructed, prior to writing the counterattitudinal essay, to express any subtle feelings of anxiety or tension they experienced. Results revealed that this condition had less attitude change than a standard high-choice condition and a high-choice condition in which participants were instructed to suppress negative feelings. In a similar manner, Stice (1992) found that individuals induced to express their feelings about their counterattitudinal behavior to the experimenter had less attitude change, compared to individuals who engaged in the same counterattitudinal behavior but did not express their feelings. These experiments suggest that expressing the negative affect associated with dissonance following the evocation of cognitive discrepancy may reduce the motivation to engage in discrepancy reductions such as attitude change. Although these experiments used fairly strong inductions to have participants express their dissonance discomfort, it is possible that having individuals express their dissonance discomfort by completing a self-report measure of negative affect could have a similar effect of reducing the motivation to engage in discrepancy reduction. Indeed, Galinsky et al. (2000) conducted an induced compliance experiment in which dissonance discomfort was measured. Their results revealed that participants who wrote a counterattitudinal essay (under high choice) reported more dissonance discomfort, but did not show evidence of attitude change. Along the lines suggested above, Galinsky et al. (2000) posited that “the expression of dissonance-produced negative affect may have reduced participants’ dissonance motivation and their need to engage in attitude change” (p.137). Thus, use of measures of self-reported negative affect in dissonance experiments may have the unfortunate consequence of eliminating effects on measures of cognitive discrepancy. However, if the self-report measures are presented to participants in a manner that avoids a full expression of negative affect, this unfortunate consequence may be eliminated.


Overreliance on attitude change has limited our ability to test core assumptions of dissonance theory. In response, we have offered an additional methodological tool and have reviewed evidence supporting its use. We have reviewed evidence from traditional dissonance paradigms that supports the efficacy of the dissonance thermometer and encourages dissonance researchers to revisit conceptual and theoretical issues that have been difficult to explore in the absence of a direct measure of dissonance. The use of an affect measure is not a panacea for the dissonance literature. Our purpose has been to illustrate how an affect measure can be used and may facilitate progress on central questions in the dissonance literature. Through its use, this research provides direct evidence to support the theoretical process outlined by Festinger (1957). Indeed, we expect that the most important developments afforded by the dissonance thermometer lie ahead and look forward to the next 60 years of dissonance-related theory and research.


Aronson, E. (1968). Dissonance theory: Progress and problems. In R. Abelson, E. Aronson, W. McGuire, T. Newcomb, M. Rosenberg, & P. Tannenbaum (Eds.), The cognitive consistency theories: A sourcebook (pp. 5—27). Chicago, IL: Rand McNally.

Aronson, E. (1992). The return of the repressed: Dissonance theory makes a comeback. Psychological Inquiry, 3, 303—311.

Bem, D. J. (1967). Self-perception: An alternative interpretation of cognitive dissonance phenomena. Psychological Review, 74, 183—200.

Boninger, D. S., Krosnick, J. A., Berent, M. K., & Fabrigar, L. R. (1995). The causes and consequences of attitude importance. In R. E. Petty & J. A. Krosnick (Eds.), Attitude strength: Antecedents and consequences (pp. 159—189). Hillsdale, NJ: Erlbaum.

Brehm, J., & Cohen, A. (1962). Explorations in cognitive dissonance. New York, NY: Wiley.

Burris, C. T., Harmon-Jones, E., & Tarpley, W. R. (1997). “By faith alone”: Religious agitation and cognitive dissonance. Basic and Applied Social Psychology, 19, 17—31.

Chapanis, N. P., & Chapanis, A. (1964). Cognitive dissonance: Five years later. Psychological Bulletin, 61, 1—22.

Cialdini, R., Petty, R., & Cacioppo, J. (1981). Attitude and attitude change. Annual Review of Psychology, 32, 357—404.

Cooper, J., & Fazio, R. (1984). A new look at dissonance theory. In L. Berkowitz (Ed.), Advances in experimental social psychology (Vol. 17, pp. 229—266). San Diego, CA: Academic Press.

Cooper, J., & Mackie, D. (1983). Cognitive dissonance in an intergroup context. Journal of Personality and Social Psychology, 44, 536—544.

Cooper, J., & Worchel, S. (1970). Role of undesired consequences in arousing cognitive dissonance. Journal of Personality and Social Psychology, 16, 199—206.

Croyle, R. T., & Cooper, J. (1983). Dissonance arousal: Physiological evidence. Journal of Personality and Social Psychology, 45, 782—791.

Devine, P., Froning, D., & Elliot, A. (1995). [Attitude importance, dissonance, and resistance to attitude change]. Unpublished raw data.

Devine, P. G., Monteith, M. J., Zuwerink, J. R., & Elliot, A. J. (1991). Prejudice with and without compunction. Journal of Personality and Social Psychology, 60, 817—830.

Dowsett, E., Semmler, C., Bray, H., Ankeny, R. A., & Chur-Hansen, A. (2018). Neutralising the meat paradox: Cognitive dissonance, gender, and eating animals. Appetite, 123, 280—288.

Elkin, R. A., & Leippe, M. R. (1986). Physiological arousal, dissonance, and attitude change: Evidence for a dissonance-arousal link and a “don’t remind me” effect. Journal of Personality and Social Psychology, 51, 55—65.

Elliot, A. J., & Devine, P. G. (1994). On the motivational nature of cognitive dissonance: Dissonance as psychological discomfort. Journal of Personality and Social Psychology, 67, 382—394.

Fazio, R. F. I., & Cooper, J. (1983). Arousal in the dissonance process. In J. Cacioppo & R. Petty (Eds.), Social psychophysiology (pp. 122—152). New York, NY: Guilford Press.

Festinger, L. (1957). A theory of cognitive 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.

Galinsky, A. D., Stone, J., & Cooper, J. (2000). The reinstatement of dissonance and psychological discomfort following failed affirmations. European Journal of Social Psychology, 30, 123—147.<123::AID-EJSP981>3.0.CO;2-T

Gosling, P., Denizeau, M., & Oberlé, D. (2006). Denial of responsibility: A new mode of dissonance reduction. Journal of Personality and Social Psychology, 90, 722—733.

Hardyck, J., & Kardush, M. (1968). A modest modish model for dissonance reduction. In R. Abelson, E. Aronson, W. McGuire, T. Newcomb, M. Rosenberg, & P. Tannenbaum (Eds.), Theories of cognitive consistency: A sourcebook (pp. 684—692). Chicago, IL: Rand McNally.

Harmon-Jones, E. (2000). Cognitive dissonance and experienced negative affect: Evidence that dissonance increases experienced negative affect even in the absence of aversive consequences. Personality and Social Psychology Bulletin, 26, 1490—1501.

Harmon-Jones, E., & Beer, J. S. (2009). Methods in social neuroscience. New York, NY: Guilford Press.

Harmon-Jones, E., Brehm, J., Greenberg, J., Simon, L., & Nelson, D. (1996). Evidence that the production of aversive consequences is not necessary to create cognitive dissonance. Journal of Personality and Social Psychology, 70, 5—16.

Holland, R. W., Meertens, R. M., & Van Vugt, M. (2002). Dissonance on the road: Self-esteem as a moderator of internal and external self-justification strategies. Personality and Social Psychology Bulletin, 28, 1713—1724.

Kidd, R. F., & Berkowitz, L. (1976). Effect of dissonance arousal on helpfulness. Journal of Personality and Social Psychology, 33, 613—622.

Kiesler, C. A., & Pallak, M. S. (1976). Arousal properties of dissonance manipulations. Psychological Bulletin, 83, 1014—1025.

Levy, N., Harmon-Jones, C., & Harmon-Jones, E. (2018). Dissonance and discomfort: Does a simple cognitive inconsistency evoke a negative affective state? Motivation Science, 4, 95—108.

Losch, M., & Cacioppo, J. (1990). Cognitive dissonance may enhance sympathetic tonus, but attitudes are changed to reduce negative affect rather than arousal. Journal of Experimental Social Psychology, 26, 289—304.

Loughnan, S., Haslam, N., & Bastian, B. (2010). The role of meat consumption in the denial of moral status and mind to meat animals. Appetite, 55, 156—159.

Matz, D. C., Hofstedt, P. M., & Wood, W. (2008). Extraversion as a moderator of the cognitive dissonance associated with disagreement. Personality and Individual Differences, 45, 401—405.

Matz, D. C., & Wood, W. (2005). Cognitive dissonance in groups: The consequences of disagreement. Journal of Personality and Social Psychology, 88, 22—37.

Nisbett, R., & Wilson, T. (1977). Telling more than we can know: Verbal reports on mental processes. Psychological Review, 84, 231—259.

Pallak, M. S., & Pittman, T. S. (1972). General motivational effects of dissonance arousal. Journal of Personality and Social Psychology, 21, 349—358.

Pilisuk, M. (1968). Depth, centrality, and tolerance in cognitive consistency. In R. Abelson, E. Aronson, W. McGuire, T. Newcomb, M. Rosenberg, & P. Tannenbaum (Eds.), The cognitive consistency theories: A sourcebook (pp. 693—699). Chicago, IL: Rand McNally.

Pyszczynski, T., Greenberg, J., Solomon, S., Sideris, J., & Stubing, M. J. (1993). Emotional expression and the reduction of motivated cognitive bias: Evidence from cognitive dissonance and distancing from victims’ paradigms. Journal of Personality and Social Psychology, 64, 177—186.

Rabbi, J. M., Brehm, J. W., & Cohen, A. R. (1959). Verbalization and reactions to cognitive dissonance. Journal of Personality, 27, 407—417.

Schachter, S., & Singer, J. E. (1962). Cognitive, social, and physiological determinants of emotional state. Psychological Review, 69, 379—399.

Scheier, M., & Carver, C. (1980). Private and public self-attention, resistance to change, and dissonance reduction. Journal of Personality and Social Psychology, 39, 390—405.

Shaffer, D. R. (1975). Some effects of consonant and dissonant attitudinal advocacy on initial attitude salience and attitude change. Journal of Personality and Social Psychology, 32, 160—168.

Shaffer, D. R., & Hendrick, C. (1974). Dogmatism and tolerance for ambiguity as determinants of differential reactions to cognitive inconsistency. Journal of Personality and Social Psychology, 29, 601—608.

Sherman, S., & Gorkin, L. (1980). Attitude bolstering when behavior is inconsistent with central attitudes. Journal of Experimental Social Psychology, 16, 388—403.

Simon, L., Greenberg, J., & Brehm, J. (1995). Trivialization: The forgotten mode of dissonance reduction. Journal of Personality and Social Psychology, 68, 247—260.

Son Hing, L. S., Li, W., & Zanna, M. P. (2002). Inducing hypocrisy to reduce prejudicial responses among aversive racists. Journal of Experimental Social Psychology, 38, 71—78.

Steele, C. M. (1988). The psychology of self-affirmation: Sustaining the integrity of the self. In L. Berkowitz (Ed.), Advances in experimental social psychology (Vol. 21, pp. 261—302). New York, NY: Academic Press.

Steele, C. M., & Liu, T. J. (1983). Dissonance processes as self-affirmation. Journal of Personality and Social Psychology, 45, 5—19.

Stice, E. (1992). The similarities between cognitive dissonance and guilt: Confessions as a relief of dissonance. Current Psychological Research & Reviews, 11, 69—77.

Stone, J., Aronson, E., Crain, A., Winslow, M., & Fried, C. (1994). Inducing hypocrisy as a means of encouraging young adults to use condoms. Personality and Social Psychology Bulletin, 20, 116—128.

Tauer, J., & Devine, P. (1998). [Attitude importance and the efficacy of alternative dissonance reduction strategies]. Unpublished raw data.

Tauer, J., Devine, P., & Elliot, A. (1998, May). Attitude importance, dissonance, and resistance to attitude change reduction. Paper presented at the 7th annual meeting of the Midwestern Psychological Association, Chicago, IL.

Tedeschi, J. T., Schlenker, B. R., & Bonoma, T. V. (1971). Cognitive dissonance: Private ratiocination or public spectacle? American Psychologist, 26, 685—695.

Vance, K., & Devine, P. (1997). [How self-affirmations influence the motivation and behaviors associated with prejudice reduction]. Unpublished raw data.

Vance, K., Devine, P., & Barron, K. (1997, May). The effects of self-affirmation on the prejudice reduction process. Paper presented at the Midwestern Psychological Association Meeting, Chicago, IL.

Wilder, D. (1992). Yes, Elliot, there is dissonance. Psychological Inquiry, 3, 351—352.

Winkielman, P., & Berridge, K. C. (2004). Unconscious emotion. Current Directions in Psychological Science, 13, 120—123.

Zanna, M. P., & Cooper, J. (1974). Dissonance and the pill: An attribution approach to studying the arousal properties of dissonance. Journal of Personality and Social Psychology, 29, 703—709.

Zuwerink, J., & Devine, P. (1996). Attitude importance and resistance to persuasion: It’s not just the thought that counts. Journal of Personality and Social Psychology, 70, 931—944.


1Throughout the chapter, we focus on induced-compliance paradigms that have traditionally used attitude change as the indicator of dissonance. We should be clear, however, that the induced-compliance paradigm is not the only one that suffers from an overreliance on the use of attitude measures as the indicator of dissonance. For example, the free-choice paradigm relies on the spreading of alternatives, and the selective-exposure paradigm relies on preference for attitude-consistent information. We suggest that these paradigms, with their emphasis on outcome measures to indicate dissonance arousal and reduction, are also limited in what they can reveal about the nature of the dissonance experience or the process of dissonance induction and reduction. Over the years, the induced-compliance paradigm, however, has been the most frequently used paradigm in dissonance research. Therefore, we use this paradigm to illustrate our concerns regarding the overreliance on such outcome measures as indicators of both dissonance arousal and reduction.

2It is of interest to note that Zanna and Cooper (1974) collected a measure of felt tension with a single item, ranging from calm (1) to tense (31). In general, participants who were expected to feel tension (e.g., standard high-choice condition and participants in the “drug”-creating arousal conditions) reported higher levels of felt tension. Although these findings were reported, Zanna and Cooper’s primary focus was to provide evidence that dissonance had arousal properties. Very few others have attempted to measure the felt discomfort created by dissonance tasks, and studies that did typically suffered from shortcomings that limited the informativeness of the measures (see Elliot & Devine, 1994, for a discussion).

3We thank Mark Zanna for the suggestion of the label dissonance thermometer. We think this label is intuitive and captures the essence of the measure.

4In all of our studies reviewed, choice manipulation checks indicated the effectiveness of the choice manipulation.

5In this study, all measures were taken after the essay was prepared. In Elliot and Devine (1994), low-choice participants reported affect before the essay task was introduced. Although this provided a nice baseline affect measure, it did not directly address whether the low-choice instructions were dissonance inducing. The data from the Devine et al. (1995) study suggest that the low-choice instructions did not lead to elevated levels of discomfort for high- or low-importance participants. Assessing affect after rather than before the essay task is, we think, generally a preferred strategy.