Action-Based Model of Dissonance: A Review, Integration, and Expansion of Conceptions of Cognitive Conflict

An action-based model of dissonance is presented. This model accepts the original theory’s proposal that a sufficient cognitive inconsistency causes the negative affective state of dissonance. It extends the original theory by proposing why cognitive inconsistency prompts dissonance and dissonance reduction. After reviewing past theoretical and empirical developments on cognitive dissonance theory, we describe the action-based model and present results from behavioral and physiological experiments that have tested predictions derived from this model. In particular, this evidence converges with recent neuroscience evidence in suggesting that the anterior cingulate cortex and left prefrontal cortical region are involved in conflict detection and resolution, respectively.We end by reviewing research on individual differences in dissonance arousal and reduction, and present a new measure designed to assess these individual differences. 1. Overview of the Chapter Cognitive dissonance theory, first proposed by Festinger (1957), has generated hundreds of experiments and is considered one of the most influential theories in psychology ( Jones, 1985). The theory and the research it has inspired have led to an increased understanding of attitude and behavior change processes, as well as an understanding of the relationships between cognition, perception, emotion, and motivation. In this article, we present the core ideas behind Festinger’s original theoretical statement and discuss some notable attempts by researchers to revise and extend the basic theory. We then describe a more recent theoretical conceptualization of dissonance, referred to as an action-based model, which provides an overarching framework for understanding dissonance processes, and for integrating a wide range of data and previous theoretical revisions to Festinger’s theory. Briefly stated, the action-based model begins with the assumption that many perceptions and cognitions serve to activate action tendencies with little or no conscious deliberation. This assumption is consistent with several perspectives in psychological science, such as William James’ (1890) ideomotor conception, Gibson’s (1966, 1979) ecological approach to perception, and subsequent elaborations of these basic ideas (Berkowitz, 1984; Dijksterhuis & Bargh, 2001; Fiske, 1992; McArthur & Baron, 1983; Smith & Semin, 2004). The action-based model goes further to suggest that when these ‘‘cognitions’’ with action implications come into conflict, a negative affective state is aroused, referred to as dissonance. Our model posits that dissonance affect is aroused because conflicting action-based cognitions have Action-Based Model of Dissonance 121 Author's personal copy the potential to interfere with effective action. The organism is motivated to reduce this negative affect and ultimately reduce the ‘‘cognitive inconsistency’’ in order to behave effectively. This way of conceptualizing dissonance processes addresses many problems with past theories concerned with dissonance. It also suggests a broad organizing framework for integrating and understanding a wide array of other nondissonance theories and research. 2. Overview of the Theory of Cognitive Dissonance The original theory of cognitive dissonance predicted that when an individual holds two or more elements of knowledge that are relevant to each other but inconsistent with one another, a state of discomfort is created. This unpleasant state is referred to as ‘‘dissonance.’’ According to the theory, the magnitude of dissonance in relation to a cognition can be formulated as equal to D/DþC, where D is the sum of cognitions dissonant with a particular cognition and C is the sum of cognitions consonant with that same particular cognition, with each cognition weighted for importance (see Sakai, 1999; Shultz & Lepper, 1999, for precise mathematical models). According to the original theory, the unpleasant state of dissonance motivates individuals to engage in psychological work in an effort to reduce the inconsistency between cognitions. Festinger (1957, p. 3) wrote, ‘‘The existence of dissonance, being psychologically uncomfortable, will motivate the person to try to reduce the dissonance and achieve consonance.’’ So, if a dieter consumed a fattening meal, he would likely be in a state of dissonance. Assuming that he stays committed to the diet, the theory would predict that he will reduce dissonance by adding consonant cognitions (e.g., ‘‘the diet will improve my appearance’’), subtracting dissonant cognitions (e.g., ‘‘fattening foods are not very tasty’’), increasing the importance of consonant cognitions (e.g., ‘‘my health is the most important thing in life’’), or decreasing the importance of dissonant cognitions (e.g., ‘‘sensory pleasures are not very important’’). Researchers have most often measured dissonance reduction with attitude change. Attitude change in response to a state of dissonance is expected to be in the direction of the cognition that is most resistant to change. In laboratory tests of the theory, knowledge about recent behavior is usually assumed to be the cognition most resistant to change. If one has recently performed a behavior, it is usually difficult to convince oneself that the behavior did not occur. Thus, attitudes often change to become more consistent with a recent behavioral commitment. 122 Eddie Harmon-Jones et al. Author's personal copy 2.1. Experimental paradigms used to test dissonance theory Three experimental paradigms constitute the majority of tests of dissonance theory. Each paradigm induces participants to experience an inconsistency between cognitions and then gives them an opportunity to express a change in attitudes. The change in attitudes is measured, and is presumed to reflect the degree of dissonance reduction. In this section, we describe the basic logic behind each of these paradigms to provide the readerwith a basis for evaluating much of the research conducted on dissonance over the past half century. 2.1.1. Free choice After a decision between alternatives, all of the cognitions that favor the chosen alternative are consonant with the decision, whereas all the cognitions that favor the rejected alternative are dissonant. An individual’s experience of dissonance is greater when the number and importance of dissonant cognitions is higher, and/or when the number and importance of consonant cognitions is lower. The dissonance an individual experiences is typically greater after choosing between alternatives that are closer in attractiveness (as long as each alternative has several distinguishing characteristics). Dissonance caused by a decision can be reduced by viewing the chosen alternative as more attractive and/or viewing the rejected alternative as less attractive. Brehm (1956) conducted the first free choice experiment. In it, participants made either an easy or a difficult decision between two alternatives (i.e., household objects such as an automatic toaster and a fluorescent desk lamp). The difficult decision was one in which the alternatives were close in attractiveness, whereas the easy decision was one in which the two alternatives were very different in their attractiveness (i.e., one alternative was much more attractive than the other). Participants were asked to evaluate each of the alternatives before and after their decision to choose one of the alternatives. After an easy decision, attitudes toward the alternatives did not change. In contrast, after a difficult decision, attitudes toward the alternatives changed, such that they became more negative toward the rejected alternative (and slightly more positive toward the chosen alternative). This method of reducing dissonance by changing one’s attitudes toward the two choice options to be more consistent with a decision has been referred to as ‘‘spreading of alternatives.’’ 2.1.2. Induced compliance Dissonance should also be aroused when a person acts in a way that is contrary to his or her attitudes, because the recent behavior is inconsistent with one’s preexisting attitude. But how can an experimenter unobtrusively induce a research participant to perform such an act? In the first test of this prediction, Festinger and Carlsmith (1959) had participants perform a boring task that involved turning a series of wooden pegs. After completing Action-Based Model of Dissonance 123 Author's personal copy this very tedious task, participants were paid either $1 or $20 to tell ‘‘another participant’’ that the task was interesting. Festinger and Carlsmith reasoned that lying for a payment of $20 should not arouse much dissonance, because $20 provides sufficient justification for the counterattitudinal behavior (i.e., it adds cognitions consonant with the behavior). By comparison, being paid $1 for performing the same behavior should arouse much dissonance, because $1 was just enough justification for the behavior (i.e., it adds fewer consonant cognitions than $20). As expected, participants in the $1 (low-justification) condition changed their attitudes to be more positive toward the task, whereas participants in the $20 (high-justification) condition did not change their attitudes. Thus, this paradigm was successful in arousing dissonance and motivating dissonance-reducing attitude change. 2.1.3. Effort justification Dissonance is aroused whenever a person engages in an unpleasant activity to obtain some desirable outcome. From the cognition that the activity is unpleasant, it follows that one would not engage in the activity. In other words, the cognition that the activity is unpleasant is dissonant with engaging in the activity. As an individual puts increasing effort into an unpleasant activity, the dissonance he or she feels as a result of the activity should increase. Dissonance can be reduced by changing one’s view of the outcome to be even more desirable (a means for adding consonant cognitions). In the first experiment designed to test these theoretical ideas, Aronson and Mills (1959) had women undergo a severe or mild ‘‘initiation’’ to become a member of a group. In the severe initiation condition, the women engaged in an embarrassing activity to join the group, whereas in the mild initiation condition, the women engaged in an activity that was not very embarrassing to join the group. The group turned out to be rather dull and boring. The women in the severe initiation condition evaluated the group more favorably than the women in the mild initiation condition. The above paradigms continue to be used fruitfully in research (e.g., Beauvois & Joule, 1996; Cooper, 2007; Harmon-Jones & Mills, 1999; Olson & Stone, 2005). Other experimental paradigms have been used to test the theory but they are used less frequently and because of space limitations are not described here (Wicklund & Brehm, 1976). 2.2. Alternative theoretical explanations After these and other dissonance results appeared, some theorists began to question whether the results were due to motivational processes. These theorists suggested that attitude change was due to cold, purely cognitive processes such as self-perception (Bem, 1967) or to managing one’s impression to others (Tedeschi et al., 1971). However, subsequent research confirmed that dissonance is best characterized as a motivated process (for 124 Eddie Harmon-Jones et al. Author's personal copy reviews, Harmon-Jones, 2000a,b). For example, individuals experiencing the state of dissonance have been found to exhibit heightened electrodermal activity (which is associated with activation of the sympathetic nervous system; Elkin & Leippe, 1986; Harmon-Jones et al., Simon & Nelson, 1996) and report increased negative affect (e.g., Elliot & Devine, 1994; Harmon-Jones, 2000c; Zanna & Cooper, 1974). After cognitive discrepancy is reduced (i.e., attitude change occurs), self-reported negative affect is reduced (Elliot & Devine, 1994; Harmon-Jones, 2000c). Moreover, research using a misattribution paradigm reveals that discrepancy reduction is motivated by the need to reduce negative affect (Zanna & Cooper, 1974). Thus, this research showing that negative affect occurs as a result of cognitive dissonance and that it creates a motivation to engage in dissonance-reducing activities strongly suggests that the dissonance process is a motivated one. Beginning in the late 1960s, researchers began to propose motivational explanations for dissonance effects that differed from Festinger’s originally proposed theory. Whereas the original theory focused on a very basic incompatibility between cognitions, these newer theories invoked higher-order, more complex processes. They changed the focus from inconsistency to the individual’s self-concept and the individual’s concern with harming others. 2.2.1. Self-consistency In self-consistency theory, Aronson (1969, 1999) proposed that dissonance only occurs when a person acts in a way that violates his or her self-concept, that is, when a person performs a behavior inconsistent with his or her view of the self. Because most persons view themselves in a positive light, such that they are competent, rational, and moral, dissonance is experienced when a person behaves in an incompetent, irrational, or immoral way. One of the primary predictions derived from this revision is that high selfesteem individuals should respond with more dissonance reduction than low self-esteem individuals, because dissonance experiments induce individuals to act in ways discrepant from a positive self-view. Studies testing this prediction have produced mixed results: some showed that high self-esteem individuals showed greater attitude change, some showed that low selfesteem individuals showed greater attitude change, and some found no differences between self-esteem groups (see Stone, 2003, for review). Also, Beauvois and Joule (1996, 1999) obtained results that appear incompatible with this self-consistency revision. Therefore, the experience of dissonance and the engagement in dissonance-reducing activities does not appear to be limited to discrepancies involving the self-concept. 2.2.2. Self-affirmation In his alternative to Festinger’s dissonance theory, Steele (1988) proposed that individuals possess a motive to maintain an overall self-image of moral and adaptive adequacy. He stated that dissonance-induced attitude change Action-Based Model of Dissonance 125 Author's personal copy occurs because dissonance threatens this positive self-image. Whereas Festinger’s dissonance theory posited that individuals are motivated to reconcile inconsistent cognitions, Steele proposed that, instead, individuals are merely motivated to affirm the integrity of the self or maintain a ‘‘perception of global integrity, that is, of overall moral and adaptive adequacy’’ (Steele et al., 1993, p. 885; see Sherman & Cohen, 2006, for a recent review). In support of this idea, Steele presented experiments where, following a dissonance induction, participants either were or were not presented with an opportunity to affirm an important value. When participants were allowed to affirm an important value, dissonance-related attitude change did not occur. However, Simon et al. (1995) presented evidence supporting an alternative explanation for Steele’s findings that was in line with the original theory of dissonance. Festinger’s original theory proposed that the degree of dissonance experienced depended upon the importance of the dissonant and consonant cognitions. Simon et al. proposed that the mechanism by which self-affirmation reduced dissonance was by reducing the importance of the cognitions involved in the dissonance. They hypothesized that making an important value salient could reduce dissonance by reducing the individual’s perception of the importance of the dissonant act, even if the value was unrelated to the self-concept. They conducted an experiment in which, following the induction of dissonance, participants were either given an opportunity to affirm an important value (i.e., ‘‘a self-affirmation condition, rank issues such as politics in term of their personal importance’’), asked to consider a value that was not important to them personally but was of general importance (i.e., ‘‘an issuesalient condition, rank the same issues as above but in terms of their importance in general’’), or were given no special instructions (control condition). Participants in the control condition changed their attitudes to be more consistent with the induced compliance behavior, as expected. Participants in both the self-affirmation and issue salient conditions did not change their attitudes. Writing about an important value caused participants to reduce the importance of the behavior and attitude to the point that attitude change did not occur. This occurred even when the values were not personally important and thus not self-affirming. Other evidence has been presented that is difficult to interpret in self-affirmation theory terms, such as evidence suggesting that self-affirmations relevant to the recent dissonant act increase rather than decrease dissonance-related attitude change (Aronson et al., 1999). The self models of dissonance also have difficulty explaining the dissonance effects produced in rats (Lawrence & Festinger, 1962), as rats are believed to lack self conceptions of morality, rationality, and competence. Recent research has revealed that four-year-old humans and capuchin monkeys, who also lack the complex self-concepts which would seem to be required by self models of dissonance, show evidence of dissonance reduction (Egan et al., 2007). Hence, although self aspects appear to 126 Eddie Harmon-Jones et al. Author's personal copy moderate dissonance processes, they are not necessary to cause dissonance (Harmon-Jones, 2000d; Stone & Cooper, 2003). In terms of the original theory, self-related cognitions would be expected to affect the magnitude of dissonance, as cognitions related to the self are often important to an adult human. In other words, the experimental results derived from the self models are compatible with the original theory. Furthermore, the self models are unable to explain basic dissonance motivation effects concerning discrepancies that do not involve the self. 2.2.3. Aversive consequences Cooper and Fazio (1984) proposed that the discomfort experienced in dissonance experiments was not due to an inconsistency between the individual’s cognitions, but rather to feeling personally responsible for producing an aversive consequence. In support of this idea, Cooper and Worchel (1970) replicated and extended Festinger andCarlsmith’s (1959) classic experiment in which participants were given low or high justification to claim that a boring task was interesting. In addition to the conditions of the original experiment, Cooper and Worchel added a condition in which, when the participant told the confederate that the boring task was interesting, the confederate was not convinced. Attitude change occurred only in the low-justification condition where the confederate believed the participant. This result and others (for review, see Cooper & Fazio, 1984) have been interpreted as indicating that dissonance-related attitude change only occurs when individuals feel personally responsible for producing an aversive consequence. According to the original theory of cognitive dissonance, the production of aversive consequences would be expected to increase the amount of dissonance produced because an aversive consequence in itself may be an important dissonant cognition, or it may further strengthen one’s behavioral commitment (see Harmon-Jones, 1999). However, the original theory would deny that an aversive consequence is necessary to produce dissonance. In the induced-compliance experiments testing the necessity of aversive consequences, there are a number of reasons why attitude change may have occurred only when participants’ behavior led to aversive consequences. The null finding that attitudes were unchanged in the no-aversive-consequences conditions, like all null effects, is difficult to explain and subject to multiple alternative explanations. One possibility is that attitude change was produced, but the small sample sizes in these experiments may have had insufficient power to detect the change. Another possibility is that not enough dissonance was aroused in these experiments to produce attitude change without the additional important cognition of an aversive consequence. Finally, the dissonance in the no-aversive consequences conditions may have been reduced by some other route besides attitude change. To examine whether attitude change could occur in an induced compliance setting in which aversive consequences were not produced, we Action-Based Model of Dissonance 127 Author's personal copy conducted several experiments (Harmon-Jones, 2000c; Harmon-Jones et al., 1996). Under the guise of an experiment on memory, participants were exposed to an attitudinal object (e.g., a boring passage they read). Participants were assured of privacy and anonymity, and then given high or low choice to write a counter-attitudinal statement (to manipulate justification) about the object. They were asked to discard the statement in the trash after writing it, so that there was no chance of the statement causing an aversive consequence. This manipulation was based on Cooper and Fazio’s (1984) statement that, ‘‘making a statement contrary to one’s attitude while in solitude does not have the potential for bringing about an aversive event’’ (p. 232). In other words, the experiments were designed so that aversive consequences were clearly absent to demonstrate that cognitive dissonance processes could occur in such situations. In one experiment (Harmon-Jones et al., 1996), participants were asked to read a boring passage. They were then given high or low choice to write that they found the boring passage interesting. Although no aversive consequences were produced, persons in the high-choice condition changed their attitudes to be more favorable toward the passage. In addition, participants in high-choice condition evidenced more sympathetic nervous system arousal, as measured by nonspecific skin conductance responses, than those in the low-choice condition. In another experiment, chocolate-loving participants wrote a statement that they disliked a piece of chocolate they had just eaten under conditions of low or high choice (Harmon-Jones, 2000c). Participants in the highchoice condition changed their attitudes to report a decrease in their enjoyment of chocolate. In addition, self-reported negative affect was increased following dissonance-producing behavior and was reduced following the attitude change. These experiments also demonstrate that the experience of cognitive dissonance evokes an unpleasant state that motivates discrepancy reduction. The results obtained in these experiments indicate that dissonance affect and dissonance-related attitude change can occur in situations in which a cognitive inconsistency is present but does not involve the possibility of aversive consequences. Because participants’ counterattitudinal statements were produced in private and with anonymity and were discarded after they were written, the participants did not cause an aversive consequence. In these experiments, participants did not lose a reward, gain a punishment, tell a lie to another person, or inflict any other kind of injury on other persons. There was simply an abstract benefit of helping in research. The discrepancy between the participants’ perception of a stimulus and the participants’ knowledge of what they had been induced to state about that stimulus was sufficient to create dissonance. These experiments supported the original conception of dissonance theory over this revision. McGregor et al. (1999) have also discussed and 128 Eddie Harmon-Jones et al. Author's personal copy demonstrated that attitudinal ambivalence research has provided evidence of dissonance-related negative affect in the absence of feeling personally responsible for producing negative consequences. More specifically, they found that the simultaneous accessibility of participants’ conflicting attitudes (i.e., how quickly and equally quickly conflicting evaluations came to mind) predicted the ambivalence participants felt (Newby-Clark et al., 2002). Nevertheless, some important questions regarding the basic mechanism underlying dissonance effects remained: Why does dissonance evoke this negative motivational state? Why does this state motivate attitude change? 3. Action-Based Model of Dissonance: Why do Dissonance Processes Occur? Festinger (1957) posited no answer to the question of why dissonance processes occur other than to state that inconsistency is motivating. Brehm and Cohen (1962) and Beauvois and Joule (1996, 1999) pointed out that a behavioral commitment is an important component of the dissonance process. However, in these previous statements, these theorists did not indicate why cognitions with implications for action motivate persons to engage in discrepancy reductions. The action-based model of cognitive dissonance was proposed to answer this ‘‘Why?’’ question (Harmon-Jones, 1999). The action-basedmodel concurs with theorizing in other areas of psychology in proposing that perceptions and cognitions can serve as action tendencies (Berkowitz, 1984; Dijksterhuis & Bargh, 2001; Fiske, 1992; Gibson, 1979; James, 1890; McArthur & Baron, 1983; Smith & Semin, 2004). Indeed, this perspective on perception/cognition is quite consistent with the situated cognition approach of Smith and Semin (2004), which proposes, among other things, (1) that mental representations are action oriented; (2) that cognition is embodied in that it draws on our sensorimotor abilities, environments, brains, and bodies; and (3) that cognition and action are the result of dynamic processes of interactions between an agent and environment. The action-based model further proposes that dissonance between cognitions evokes a negative affective state because it has the potential to interfere with effective and unconflicted action. In essence, discrepant cognitions create problems for the individual when those cognitions have conflicting action tendencies. Dissonance reduction, by bringing cognitions into line with behavioral commitments, serves the function of facilitating the execution of effective and unconflicted action (see also, Jones & Gerard, 1967). The action-based model proposes both a proximal and a distal motivation for the existence of dissonance processes. The proximal motive for reducing dissonance is to reduce or eliminate the negative emotion of dissonance. The distal motivation is the need for effective and unconflicted action. Thus, Action-Based Model of Dissonance 129 Author's personal copy consistent with the socially situated cognition approach (Smith & Semin, 2004), the action-based model assumes that emotion, cognition, and action constitute adaptive regulatory processes that ultimately serve survival needs. Past discussions of the theory of cognitive dissonance have referred to two different constructs as ‘‘cognitive dissonance.’’ One is the inconsistency between cognitions. The second is the unpleasant emotional/motivational state that occurs when a person holds two contradictory cognitions. In order to better understand the processes of dissonance, the action-based model distinguishes between the two. We refer to inconsistency between cognitions as ‘‘cognitive discrepancy,’’ whereas we call the unpleasant emotive state ‘‘dissonance.’’ The unpleasant emotive state of dissonance providesmotivation to change one’s attitudes or engage in other discrepancy-reduction processes. After an individual makes a difficult decision, psychological processing should assist with the execution of the decision. The tendency of participants in dissonance research to view the chosen alternative more favorably and the rejected alternative more negatively after a decision may help the individual to follow through, to effectively carry out the actions that follow from the decision. As an example, consider an important, effortful behavioral decision, such as beginning an exercise program. In this situation, the ‘‘actions’’ implied by the decision are the exercise behaviors. The benefits of exercise, from better-fitting clothes to improved long-term health, constitute consonant cognitions. The drawbacks of exercise, including the time commitment and muscle soreness, constitute dissonant cognitions. Dissonance affect comes from the conflict aroused by the dissonant cognitions, and this unpleasant affect motivates the individual to decrease the discrepancy by bringing the cognitions in line with the behavioral commitment. The better an individual is able to reduce the number and importance of dissonant cognitions and increase the number and importance of consonant cognitions, the more likely it is that he or she will faithfully perform the actions required by the exercise program over the long-term and reap its benefits. In contrast to models of cognitive dissonance that view dissonance processes as irrational and maladaptive (Aronson, 1969), the action-based model views dissonance processes as adaptive. Of course, adaptive, functional psychological processes that are useful and beneficial in most circumstances may not be beneficial in all circumstances. Occasionally, dissonance reduction may cause persons to maintain a prolonged commitment to a harmful chosen course of action, when it would be better to disengage. However, when we state that dissonance processes are adaptive, we mean that they benefit the organism in the majority of cases. In addition, we must distinguish between dissonance motivation and dissonance reduction. The action-based model, like the original theory, proposes that cognitive discrepancy produces negative affect, and that the negative affect motivates the individual to change his or her attitudes. 130 Eddie Harmon-Jones et al. Author's personal copy However, it is possible for a person to continue to maintain conflicting attitudes (although negative affect may persist). Furthermore, there are some situations in which individuals do disengage from harmful chosen courses of action, even though they may experience high levels of negative affect in the process. 4. Tests of the Action-Based Model 4.1. Action-orientation and spreading of alternatives According to the action-based model of dissonance, the post-decisional state is similar to an action-oriented state (Beckmann & Irle, 1985; Gollwitzer, 1990; Kuhl, 1984), where the individual is in a mode of ‘‘getting things done.’’ Once a decision is made, an organism should be motivationally tuned toward enacting the decision and behaving effectively with regard to it. An implemental or action-oriented mindset is one in which plans are made to effectively execute behaviors associated with the decision (Gollwitzer & Bayer, 1999). We suggest that this implemental or action-oriented state is similar to an approach motivational state. When a person is in an action-oriented state, implementation of decisions is enhanced (Gollwitzer & Sheeran, 2006). We suggest that these actionoriented states and implemental states are similar to Jones and Gerard’s (1967) concept of an unequivocal behavior orientation. We proposed that the action-oriented state that follows decision-making is equivalent to the state in which dissonance motivation operates and discrepancy reduction occurs (Harmon-Jones & Harmon-Jones, 2002). Thus, experimentally manipulating the degree of action-orientation experienced following a decision should affect the degree of discrepancy reduction. In one experiment, participants were asked to make either an easy decision or a difficult decision. Participants then completed a mindset questionnaire. The neutral mindset asked participants to list seven things they did in a typical day, whereas the action-oriented mindset questionnaire asked participants to list seven things they could do to perform well on the physical exercise they had chosen. Participants then reevaluated the exercises. Participants who made a difficult-decision in the action-oriented condition demonstrated a greater increase in preference for the chosen over the rejected exercise (i.e., spreading of alternatives) than participants in the other three conditions. In a second experiment, we replicated the results of the first experiment using a different manipulation of action-orientation (Harmon-Jones & Harmon-Jones, 2002). In this experiment, action-orientation was induced by asking participants to think about a project or goal that they intended to accomplish, and to list the steps they intended to use to successfully follow Action-Based Model of Dissonance 131 Author's personal copy through with their decision (Gollwitzer, 1990). Two comparison conditions were also included, one in which participants wrote about a neutral, ordinary day and one in which participants wrote about an unresolved problem, which was defined as a problem characterized by the fact that they were not yet sure whether to take action to change things. Thus, as in the previous experiment, participants first made a difficult decision, but this time the decision was between two equally attractive research studies in which they could participate. Following the decision, participants completed the action-orientation manipulation described above, and then rerated their attitudes toward the research studies. Results indicated that the participants in the action-orientation condition engaged in more spreading of alternatives following a difficult decision than did participants in the comparison conditions. This study provided stronger support for the action-based model because, in this case, the action-orientation induction was unrelated to the decision in the experiment. Correlational evidence also suggests that action-oriented processing facilitates discrepancy reduction (Beckmann & Kuhl, 1984). In this study, dispositional action orientation was measured by Kuhl’s (1980, 1984) action versus state orientation questionnaire. A sample item from the scale says, ‘‘When I have decided to buy one item of clothing and I find several things I like: (1) ‘‘I often waver back and forth, trying to decide which I should buy’’ (state-oriented answer); and (2) ‘‘I usually don’t think much about it and make a quick decision’’ (action-oriented answer). Participants were individuals searching for an apartment and they were shown information about 16 apartments. Participants rated the attractiveness of the apartments before and after choosing the apartment they preferred (i.e., before and after a tentative decision). After the decision, individuals who were dispositionally high in action-orientation increased the attractiveness rating of the chosen apartment more than individuals who were dispositionally low in action-orientation. Thus, both state and trait evidence support our contention that dissonance reduction occurs in an action-oriented state—a state that assists in the implementation of decisions and in effective action. 4.2. Neural activity underlying dissonance and dissonance reduction The action-based model of cognitive dissonance corresponds closely to recent models of self-regulation developed in the field of cognitive neuroscience, and it provides an important theoretical framework for placing neural processes in the context of motivated cognition. In this section, we describe findings from research on the neural processes associated with the monitoring of response conflicts and the implementation of intended behavior that are consistent with the action-based model of dissonance. 132 Eddie Harmon-Jones et al. Author's personal copy 4.2.1. Dissonance arousal, conflict monitoring, and the anterior cingulate cortex According to the action-based model, dissonance is aroused by the activation of cognitions that interfere with goal-driven behavior. Although few studies have directly examined the process of dissonance arousal in the brain, much attention has been given to questions of how the brain processes response conflicts on task such as the color-naming Stroop (1935) task. For example, when completing the color-naming Stroop task, one’s goal is to identify the ink color of a word stimulus, regardless of the word’s meaning. However, the processing of word meaning is typically automatic, and when a word’s meaning is incongruent with one’s goal to judge the word’s color, such as when the word ‘‘red’’ is presented in blue ink, there is conflict between the intended and the automatic response tendencies. In studies examining neural activity during the Stroop task, anterior cingulate cortex activity is greater during incongruent trials than congruent trials (Carter et al., 1998). Similar findings have been observed using other response-conflict tasks, such as the Eriksen flanker’s task (Gerhing et al., 1993; Eriksen & Eriksen, 1974), and the Go/No-Go task (Botvinick et al., 1999; Kiehl et al., 2001). Researchers have interpreted these findings as evidence that the anterior cingulate cortex plays an important role in monitoring the moment-to-moment representations of action tendencies for potential conflicts, presumably so that other neurocognitive mechanisms may be engaged to override the unwanted tendency and to promote an effective goal-directed response (Botvinick et al., 2001). Thus, conflict monitoring represents the first component of a dual-process model of cognitive control, whereby the need for control is initially detected. Recently, we have suggested that the anterior cingulate cortex, and its associated role in conflict monitoring, corresponds well to the process of dissonance arousal (Harmon-Jones, 2004). The conflict-monitoring account is consistent with the action-based model of dissonance, because it too focuses on conflicts between action tendencies. Amodio et al. (2004) integrated the conflict-monitoring framework with social psychological theories of selfregulation by examining conflict between automatic stereotyping tendencies and participants’ goals to respond without prejudice. In this study, anterior cingulate cortex activity was monitored using an event-related potential measure referred to as the ‘‘error-related negativity’’ component (Gerhing et al., 1993; van Veen & Carter, 2006). When participants—who reported low-prejudice attitudes—accidentally made responses that reflected the application of racial stereotypes, thus constituting a clear response conflict, the anterior cingulate cortex was strongly activated. By comparison, anterior cingulate cortex activity was lower on other trial types that did not elicit conflicting actions. In subsequent research, Amodio et al. (2008) demonstrated that heightened anterior cingulate cortex activity associated with racially-biased responses was only observed for participants with strong personal Action-Based Model of Dissonance 133 Author's personal copy motivations to respond without prejudice. Participants without personal motivations (i.e., high-prejudice participants) did not show enhanced anterior cingulate cortex activity when their responses reflected the application of stereotypes. Thus, when participants made responses that were dissonant with their attitude-based intentions, anterior cingulate cortex activity was high. Furthermore, across studies, participants with stronger anterior cingulate cortex activity to dissonant responses were more likely to engage in controlled behavior (slower, more careful responding). These studies provided initial evidence for the role of the anterior cingulate cortex, and its associated conflict monitoring function, as a critical process underlying dissonance arousal. Importantly, this line of research demonstrated that high-level conflicts, the type with which dissonance theory has been most concerned, also activate the anterior cingulate cortex, in line with lowerlevel forms of conflict typically studied in the cognitive neuroscience literature (e.g., in studies using the Stroop task). More recently, van Veen et al. (2007) examined dissonance-related brain activity more directly in a study that used the induced compliance paradigm. The authors observed heightened anterior cingulate cortex activity during the manipulation of dissonance, and participants’ degree of anterior cingulate cortex activationwas significantly associatedwith attitude change. The finding that dissonance reduction was associated with increased anterior cingulate cortex activation is consistent with the action-based model, which suggests that discrepancy reduction results from the need for effective and unconflicted action (distal motive). Although this interpretation of anterior cingulate cortex activity is compatible with the original theory of dissonance, it is not clearly compatible with other versions of dissonance, because these versions focus primarily on high-level self-consistencies (or other nonconsistency-oriented motivations such as aversive consequences or self-affirmation), and thus are not directly associated with coordinating action. Response conflict tasks used in studies of the anterior cingulate cortex have also been found to cause increases in skin conductance, which indexes sympathetic nervous system arousal (Hajcak et al., 2003, 2004), and measures of negative affect such as the startle eyeblink response (Hajcak & Foti, 2008). Situations that typically evoke cognitive dissonance also cause increased skin conductance (Elkin & Leippe, 1986; Harmon-Jones et al., 1996; Losch & Cacioppo, 1990) and negative affect (Elliot & Devine, 1994; Harmon-Jones, 2000c; Zanna & Cooper, 1974). Taken together, these studies suggest that the anterior cingulate cortex is involved in generating the negative affective state of dissonance. 4.2.2. Dissonance reduction and the prefrontal cortex The arousal of negative affect by cognitive discrepancy drives efforts to reduce the dissonant state, either through actions or cognitive restructuring (e.g., attitude change). The process of cognitive discrepancy reduction can 134 Eddie Harmon-Jones et al. Author's personal copy occur rapidly. Indeed, research has revealed that dissonance-related attitude change can occur immediately after individuals commit to engage in behavior and before they actually engage in the behavior (e.g., essay writing; Rabbie et al., 1959). According to the action-based model, the process of discrepancy-reduction engages approach-oriented motivational processes, as the individual works to successfully implement the new commitment. To our knowledge, only the action-based model makes the prediction that discrepancy reduction following commitment to action involves approach motivational processes, which the model views as part of the distal motive of effecting unconflicted behavior. Recent neurocognitive models of control posit that the prefrontal cortex governs the implementation of a controlled response following the detection of conflict by the anterior cingulate cortex (Botvinick et al., 2001; Miller & Cohen, 2001). That is, as discrepancy-related activity in the anterior cingulate cortex rises, anterior cingulate cortex-to-prefrontal cortex communication or signaling increases. The prefrontal cortex is then believed to play a critical role in responding to the discrepancy by amplifying an intended response tendency to override the unintended tendency (Kerns et al., 2004). In relating the neurocognitive model of control to cognitive dissonance, the action-based model suggests that whereas the anterior cingulate cortex is associated with dissonance arousal, regions of the prefrontal cortex are critical for dissonance reduction. The dissociation between the neural processes related to dissonance arousal and discrepancy reduction supports the idea that these two processes reflect the operation of independent underlying mechanisms. However, the neurocognitive model of control does not clearly specify which regions of the prefrontal cortex contribute to different aspects of discrepancy reduction and action control, and it is silent on the role of motivation in the process of control. Converging evidence from studies using a range of methods suggest that prefrontal cortex activity is lateralized on the basis of motivational direction, with the left frontal region being involved in approach motivational processes (‘‘going toward’’), and the right frontal region being involved in inhibitory or withdrawal motivational processes (‘‘going away’’). For instance, Robinson and colleagues (e.g., Robinson & Downhill, 1995) have observed that damage to the left frontal lobe causes depressive symptoms, with stronger depressive symptoms among patients with damage closer to the frontal pole. Given that depression relates to impaired approachrelated processes (and associated approach-related emotion), damage to brain regions involved in approach motivation would lead to depression. A growing body of research assessing electroencephalographic (EEG) activity has similarly found that increased left-frontal cortical activation relates to state and trait approach motivation (Amodio et al., 2007, 2008; Harmon-Jones, 2003, 2004; Harmon-Jones & Allen, 1997, 1998). Source localization of frontal asymmetry in alpha power, which comprises the Action-Based Model of Dissonance 135 Author's personal copy index of frontal asymmetry in EEG studies, has demonstrated that it reflects activity in the dorsal prefrontal cortex (Pizzagalli et al., 2005). Initial studies of the prefrontal cortex’s role in motivation examined the association between greater left-sided frontal activity and questionnaire measures of behavioral approach sensitivity (Harmon-Jones & Allen, 1997) and the approach-related emotion of anger (Harmon-Jones & Allen, 1998). Subsequent research has related greater left-sided frontal activity to the state engagement in approach-related responses (Amodio et al., 2007; Harmon-Jones & Sigelman, 2001) and to the accessibility of approachrelated goals (Amodio et al., 2004). In addition, several fMRI studies have observed greater left-sided prefrontal cortex activity during the retrieval of approach-related action words (Bunge, 2004; Petersen et al., 1988). These findings are consistent with the idea that the left prefrontal cortex is particularly involved in the implementation of intended action and the formation (and restructuring) of goals to guide future action. This body of findings is in line with the action-based model’s position that the discrepancy reduction process serves to promote goal-directed behavior through the restructuring of goal-relevant attitudes and beliefs. It is notable that the relation between right-sided prefrontal cortex activity and withdrawal motivation is less clear, with few EEG studies reporting an association between right-sided frontal activity and either state or trait assessments of withdrawal motivation. By comparison, several recent studies suggest that the right prefrontal cortex plays a special role in the inhibition of action (Aron et al., 2004). This evidence represents data from fMRI studies of normal participants as well as brain lesion patients. Given the remaining ambiguities concerning the frontal asymmetry and withdrawal motivation, more research is currently needed to clarify the relation between withdrawal motivation and response inhibition, both at the conceptual and neurocognitive levels of analysis (Amodio et al., 2008). Considered as a whole, research on left prefrontal cortex function suggests that it is involved in approach motivational processes aimed at resolving inconsistency (MacDonald et al., 2000; van Veen & Carter, 2006). In what follows, we describe a set of studies that have examined the role of left prefrontal cortex activity and approach motivation as they relate directly to the resolution of dissonance-arousing discrepancies. The overarching prediction of the action-based model is that commitment to a chosen course of action should lead to an enhancement in relative left frontal cortical activity, which in turn should be associated with attitude change in support of the chosen course of action. Induced Compliance and Relative Left Frontal Cortical Activation In an experiment by Harmon-Jones et al. (2008), participants were randomly assigned to a low versus high choice condition in an induced compliance paradigm. Immediately after starting to write the counterattitudinal essay 136 Eddie Harmon-Jones et al. Author's personal copy (regarding a tuition increase at their university), participants’ EEG activity was recorded. After essay completion, attitudes were assessed. Participants in the high choice condition evidenced greater relative left frontal activation than individuals in the low choice condition (Harmon-Jones et al., 2008). Moreover, commitment to write the counter-attitudinal essay (high-choice) caused attitudes to be more consistent with the behavior, as compared to a lowcommitment (low-choice) condition. However, in this experiment, relative left frontal activation did not relate to attitudes, perhaps because the attitude measure lacked the needed sensitivity (e.g., it did not tap attitude change from precommitment, but only tapped attitudes following the commitment). Neurofeedback of Relative Left Frontal Cortical Activity and Free Choice In the previous experiment, when the psychological process (commitment to a chosen course of action) was manipulated and the proposed physiological substrate was measured (left frontal cortical activation), commitment to a chosen course of action increased relative left frontal cortical activation (Harmon-Jones et al., 2008). To provide stronger causal inferences regarding the role of the left frontal cortical region in following through with the commitment (discrepancy reduction), it is important to manipulate the physiology and measure the psychological outcome. Manipulation of the mediator also provides stronger causal evidence than simply correlating the proposed mediator with the outcome (Sigall &Mills, 1998; Spencer et al., 2005). Therefore, we conducted another experiment in which relative left frontal cortical activation was manipulated after dissonance was aroused to test whether a manipulated increase in relative left frontal cortical activation would increase dissonance reduction (attitude change). Tomanipulate relative left frontal cortical activity, we used neurofeedback training of EEG. Neurofeedback presents the participant with real-time feedback on brainwave activity. If brainwave activity over a particular cortical region changes in the direction desired by the experiment, then the participant is given ‘‘reward’’ feedback; if brainwave activity does not change in the desired direction, either negative feedback or no feedback is given. Rewards can be as simple as the presentation of a tone that informs the participant that brain activity has changed in the desired way. Neurofeedback-induced changes result from operant conditioning, and these changes in EEG can occur without awareness of how the brain activity changes occurred (Kamiya, 1979; Siniatchkin et al., 2000). Participants typically are not aware of how they brought about changes in brain activity; in fact, extensive practice is required to gain awareness of how one may intentionally cause changes in brain activity (e.g., 8 weeks of practice, Kotchoubey et al., 2002). In past research, neurofeedback was effective at decreasing but not increasing relative left frontal activity after only 3 days of training. The decrease in relative left frontal activity brought about with this brief neurofeedback training caused less approach-related emotional responses (Allen Action-Based Model of Dissonance 137 Author's personal copy et al., 2001). Based on these past results, we predicted that a decrease left frontal condition would be more successful at changing brain activity than an increase left frontal condition. Most importantly, we predicted that a decrease in relative left frontal activity would lead to a decrease in discrepancy reduction as measured by spreading of alternatives. To test these predictions, we used the decision paradigm developed by Brehm (1956). First, participants were randomly assigned to increase or decrease relative left frontal activation during 2 days of neurofeedback training. Then, on the third day, immediately following a difficult decision, participants received neurofeedback training in the same direction as the previous 2 days. Finally, attitudinal spreading of alternatives was assessed. In support of predictions, neurofeedback training caused a reduction in relative left frontal cortical activity, which caused an elimination of the familiar spreading of alternatives effect (Harmon-Jones et al., 2008). Together with past research showing that commitment to a chosen course of action increases activity in the left frontal cortex (Harmon-Jones et al., 2008), this experiment’s manipulation of relative left frontal cortical activity, a presumed mediator of the effect of commitment on discrepancy reduction, provides strong support for the role of relative left frontal activity in discrepancy reduction processes. Action-Oriented Mindset and Relative Left Frontal Cortical Activation A follow-up experiment (Harmon-Jones et al., 2008, Experiment 2) was designed to conceptually replicate the experiment described in the previous section. In this experiment, we manipulated action-oriented mental processing following a difficult decision. We expected to replicate past research in which the action-oriented mindset increased discrepancy reduction following a decision (Harmon-Jones & Harmon-Jones, 2002). Secondly, we expected the action-oriented mindset would increase relative left frontal cortical activity. Finally, we expected this increase in left frontal cortical activity would relate to discrepancy reduction, as assessed by spreading of alternatives. To further extend past research, we included a condition to manipulate positive affect that was low in approach motivation (i.e., participants wrote about a time when something happened that caused them to feel very good about themselves but was not the result of their own actions). This was done to distinguish between the effects of positive affect and of approach motivation on spreading of alternatives. Past research suggested that actionoriented mindsets increase positive affect (Taylor & Gollwitzer, 1995), but we do not predict that positive affect, itself, causes increased left frontal cortical activity or an increase in spreading of alternatives. Results from the experimentwere consistentwith predictions and revealed that the action-oriented mindset increased relative left frontal cortical activity and spreading of alternatives, as compared to a neutral condition and a positive affect/low-approachmotivation condition. See Figs. 3.1 and 3.2. These results Neutral 0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 Positive/no-action R el at iv e le ft m id -f ro nt al a ct iv at io n