The Illusion of Conscious Will (30 page)

Everyone did write an essay, probably because of the subtle social pressure to finish something one has started for an experiment. But, as compared to participants who were not given the sense of choice (an illusory sense, because everyone did write the essay), those who perceived choice showed a classic cognitive dissonance effect. The ones who wrote the essay against their own attitudes for 50 cents believed it more afterwards than did those who wrote for the larger sum of $2.50. Performing a dissonant act under conditions designed to arouse a feeling of choice made people become more positive toward the topic they were paid less to espouse. These results suggest that people will become more positive toward their action, the more negative the consequences of it that they knew in advance of choosing to do it. This curious turn of events—in which people come to like something more the
less
rewarding it is—is the hall-mark of dissonance phenomena and is surely the reason that many psychologists have studied dissonance over the years. It is always interesting to learn about seemingly irrational behavior, and liking things precisely because they are unrewarding seems irrational indeed.

Now, although cognitive dissonance theory is usually understood to be about attitudes and attitude change, it can also be understood to characterize a process of the revision of intention.
¹²
This is because dissonance always arises in the conindent of action or the choice to act. The change that occurs after the person has written a counterattitudinal essay, for instance, can be understood as a change in conceptualizing what was intended. From an initial state in which the intention to write the essay is weak or contrary (as marked by the negative attitude on the topic), the person moves to perform the action in response to complex pressures that are not fully represented in terms of any intentions. But having now done the act, and recalling having been told both that it was “your choice” and that a paltry 50 cents would be the pay, the person develops a positive attitude toward the act. This new attitude, in turn, suggests that there must have been a prior conscious intention that would predict writing that essay for only 50 cents. Following the action of essay writing, in other words, the person may develop the idea that there was a prior intention that would predict the action (“I wanted to write an essay on what I believe”).

12.
Researchers studying dissonance have not made a practice of asking people their intentions, but when they do, it typically turns out that attitudes do predict intentions in this way (e.g., Aronson, Fried, and Stone 1991). The relation between attitudes and behavioral intentions outside cognitive dissonance theory has been well-established (e.g., Ajzen 1985; 1991).

The key process underlying this transformation, at least for dissonance theory, is the resolution of conflict between the person’s old intention (which wouldn’t be sufficient to motivate the act or might even cause its opposite) and the action itself. In this sense, the theory suggests that actions can “sneak by” without sufficient intention, but that once having sneaked, they can become unpleasantly inconsistent with those contrary prior intentions and prompt the revision that creates a new intention. At some level of mind, then, the person knows that there is an insufficient old intention. The theory suggests that the conflict of the action with the old intention is what motivates the creation of the new intention.

The circumstances that create dissonance effects say some important things about how the confabulation of intention comes about. Apparently, the circumstances in which the action occurs must suggest to the person that the action was freely chosen and must also allow the person to believe that the consequences of the action were known or at least foreseeable (Wicklund and Brehm 1976). People do not commonly resolve dissonance for
fait accompli
consequences of action. So, for instance, dissonance forces might lead one to become particularly happy with an automobile one had purchased while knowing that it needed an engine overhaul (“It must be a marvelous car indeed if I bought it even with that engine problem”). But learning about the overhaul after the choice would do nothing to enhance one’s love of the car and instead would probably undermine it.
Fait accompli
consequences of action do not entail conscious will, and so do not lead through dissonance to the confabulation of intention.

Self-Perception

A different way to explain the confabulation of intention is to say that we have no attitudes at all before action and instead often compute our attitudes and the associated intentions
only
after we have acted. If this is true, then intentions are often matters of self-perception following action, not of self-knowledge prior to action or of any conflict between action and prior intents or attitudes. In fact, this was suggested before cognitive dissonance theory was born. Bertrand Russell (1921, 31) said it this way: “I believe that the discovery of our own motives can only be made by the same process by which we discover other people’s, namely, the process of observing our actions and inferring the desire which could prompt them.” Subsequently, this idea was expressed yet more fully by Gilbert Ryle in
The Concept of Mind
(1949), a volume that served as a guide for the behaviorist philosophy of mind. Ryle proposed that the various reasons we invent to account for our behavior might all be an exercise in creative interpretation, no matter whether we invent them before or after we act.

Ryle’s insight was to compare the perspective of a person performing an action with that of an observer of the action—for example, a speaker and a listener. Although we normally think that the speaker knows everything about what is being said while the listener knows nothing, the difference between them may not be that great. Ryle granted that “the listener may be frequently surprised to find the speaker saying something, while the speaker is only seldom surprised,” and he also noted that “while the speaker intends to say certain fairly specific things, his hearer can anticipate only roughly what sorts of topics are going to be discussed.” He went on to argue, however, that “the differences are differences of degree, not of kind. The superiority of the speaker’s knowledge over that of the listener does not indicate that he has Privileged Access to facts of a type inevitably inaccessible to the listener, but only that he is in a very good position to know what the listener is in a very poor position to know” (179). Ryle reasoned that the internal states such as attitudes or intentions that we often attribute to a person’s “Privileged Access” to some repository of his or her own mind could well be no more than the result of the person’s self-perceptions, derived from processes like those that might be applied by outside observers. Perhaps people don’t just
know
their own attitudes or intentions but instead
discern
them because they are in a particularly good position to figure them out.

In a clever theoretical advance, Daryl Bem (1967; 1972) applied this analysis to the cognitive dissonance studies. He reasoned that one might not need to propose an internal tension resulting from dissonant behavior if, when people behave, they simply observe themselves and draw inferences about their attitudes and intentions from what they find they have done and the circumstances in which they did it. So, when a person chooses Thai food over German food, for example, and does this despite the fact that the German meal is nearby and the Thai restaurant is across town through traffic, any observer would probably infer that the person prefers noodle to strudel. In fact, the person might make this same inference. It could be that the person simply infers a positive attitude toward Thai food after making this choice and so actually embraces that attitude. The “changed” attitude does not come from dissonance between the chosen behavior and some prior preference for German food. Rather, the person just self-perceived. In this analysis, the behavior itself comes from subtle situational cues of which the person is unaware (perhaps the person was unconsciously primed to recall a pleasant flirtation that once happened over satay at a Thai restaurant?). Prior conscious intention is simply not a factor. Instead, intention is invented after the fact in the process of self-perception.

This logic suggests that people who merely
observe
a participant in a dissonance experiment should estimate the participant’s attitudes quite correctly. And indeed, this is just what happens. Bem (1967) tested this by having two groups of observers each learn of a participant in a dissonance study. One group heard of a participant who wrote an essay for $5, whereas another group learned that a participant wrote an essay for 50 cents. In an original dissonance study exactly like this (Cohen 1962), of course, the essay topic had been assumed to be something that countered participants’ initial attitudes. And in the original dissonance study, it was found that participants in the 50 cent condition changed more in the direction of the essay than did those in the $5 condition, presumably because of the dissonance between their initial attitudes and the attitudes they espoused in the essay. In Bem’s observer study, no one made any mention of initial attitudes. Still, the groups of observers of people in each of these conditions were found to make estimates of the person’s attitude that corresponded precisely to the dissonance prediction: they thought that the 50 cent person believed the essay more than did the $5 person. It seems the observers asked themselves, “What sort of person would write an essay in favor of this position for this amount of money?”

When the money was less, they inferred that the person had a stronger intention to write the essay because he must have believed it.

This finding suggests that conflict with an initial attitude may not be critical for the occurrence of dissonance-type effects. Observers didn’t know anything about such initial attitudes and still were able to guess postbehavior attitudes quite nicely. Perhaps, then, this is just the same position that the original participants were in when they wrote essays for money. Their attitudes or intentions may have been quite obscure to begin with and may only have become clear after they perceived what they had done and the circumstances in which they had done it. Seeing that they had written an essay in circumstances that would not justify this—there was very little money—they perceived themselves as believers in the essay. Just as Ryle had suggested, people seemed to be observers of themselves. Even without “Privileged Access” to their own attitudes, they nonetheless figured out what their attitudes ought to be by using the same information available to any observer of their behavior.

Self-perception was highlighted in a different way in a further study by Bem and McConnell (1970). They asked people who had written counterattitudinal essays to report not their final (postessay) opinion on the issue but their
prior
opinion. People couldn’t do this. Instead, these reports of prior attitudes mirrored faithfully the standard dissonance effect: Participants led to believe that they had a high degree of choice in whether to write the essay reported that they had agreed with the essay
all along,
whereas those led to believe they had low choice reported no such agreement. What happened, apparently, was that people looking back after they had written the essay had no conscious memory of their pre-essay attitudes. It is as though the confabulation of intention erases its tracks, leaving people with no memory of ever having wanted other than what they currently see as their intention.

One way to interpret this finding is to say that it undermines the idea that people in the high-choice condition experienced dissonance. Bem and McConnell said just this, arguing that the participants’ failure to remember their previously conflicting thoughts indicated that there really was no mental conflict of the kind that dissonance theory suggests. At the same time, one might also argue that such failure is in fact evidence of a successful conflict resolution process and so supports a dissonance theory interpretation. A number of experiments subsequently focused on this problem, and it is probably safe to say that neither theory is clear enough about what happens to prior attitudes following the action to allow us to decide between them on this point (Greenwald 1975; Ross and Schulman 1973; Snyder and Ebbesen 1972).

The question of whether dissonance or self-perception theory offers a better account of these effects turns out to hinge on a different issue— the occurrence of the actual bodily tension created by dissonance. Self-perception theory holds that there is no conflict and that there should be no experience or sign of an emotional tension state when people behave in a way that counters their attitudes. Dissonance theory, on the other hand, portrays the process as driven by the unpleasant emotional state created when one behaves in a way that clashes with one’s convictions.

Lots of clever studies have been aimed at this distinction, and the conclusion they reach is simple: When people behave in a way that doesn’t conflict very strongly with their prior attitudes or intentions, no measurable tension state is created, so self-perception theory is all that is needed to understand their self-justification behavior. When people are led to behave in a way that does conflict strongly with their prior attitudes, however, measurable tension is created, and thus, dissonance theory is more clearly applicable (Cooper and Fazio 1984; Fazio 1987; Fazio, Zanna, and Cooper 1977). In the end, then, both theories meld into a kind of supertheory that allows us to understand how people justify their behavior in general.

The picture we are left with emphasizes the impromptu nature of reports of intentions. Intentions do not seem to persist in some pristine, timeless format, etched in mind permanently as they guide behavior. Rather, some fair amount of human behavior seems to occur without much influence by intentions, especially when the behavior is not particularly discrepant from prior beliefs and so can slip by without producing dissonance. But regardless of whether behaviors merely produce self-perception or actually cause dissonance, they come to be understood afterward as having accrued from certain intentions that make them seem sensible. These intentions, however, may not have existed in advance of the behavior, and in many of these studies, we know that the intentions surely did not predate the behavior. They are created post hoc as a way of protecting the illusion that we are conscious agents.