Bem, D. J., Self Perception Theory. In L. Berkowitz (ed). Advances in Experimental Social Psychology, Vol 6, 1972.

"Indi viduals come to know their own attitudes, emotions and internal states by inferring them from observations of their own behavior and circumstances in which they occur. When internal cues are weak, ambiguous, or uninterpretable, the individual is in the same position as the outside observer".

Skinner
Children learn by "point and name" with even internal stimuli. In fact, we have no knowledge at all until we have been explicitly trained.

Schachter & Singer's studies of drug-induced arousal show people can infer emotional sates from their environmental cues. Valins heatbeat studies show the same thing.

Personal behavior is another clue to inner state (in the context of the situation different attributions may be made).

A person might infer his own beliefs from his behavior if there are no external situational reasons for the behavior (ex Festingers $1/$20 study).

Support for Theory
In one study subjects are told to speak the truth when an amber light is on and lie when a green light is on. In subsequent ratings of prevriously judged neutral cartoons, the subject ranked those said as more funnier when the truth light was on.

Many of the pain perception experiments also support self-perception theory.

Reinterpretation of Cognitive Dissonance

Forced compliance Paradigm

In these experiments the person is induced to engage in counter-attitudinal behavior and their attitude is measured. According to self-perception theory, the observer sees his own behavior (telling lies to another subject for only $1) and concludes they must have liked the exercise more then they originally thought.

Pro-attitudinal advocacy
People who agree to give speeches supportive of their initial attitude (which shouldn't produce dissonance) does produce more attitude change. This makes sense from self-perception theory.

"Self-perception theory states that if external contingencies seem sufficient to account for the behavior, then the individual will not be led into using the behavior as a source of evidence for his self-attributions" "If any source implies irrelevance or low credibility, then self-attributions will not occur."

Forbidden toy experiment
Children to subsequently derogate a chosen toy when mildly threatened not to play with it (and they comply). Under severe threat they have "justification" for their not playing with the toy and so do not change their attitudes. While cognitive dissonance can explain this behavior, it's simpler to notice that if the subject just observed his own behavior he would arrive at the same conclusion.

Free-Choice Studies
After free choice between two similarly rated items, subsequent ratings are more different for the two items (higher for the chosen item, lower for the non-chosen item). Dissonance theory say the subject resolves the dissoance of rejecting the non-chosen item by devaluing it. Self-perception theory would indicated that an observer of the behavior would conclude by that action that they didn't like the non-chosen toy as much.

Interpersonal Simulations of Cognitive Dissonance studies
Rather than do the actual experiment, a subject is given a description or observes the conditions of a dissoance experiment and then estimate the attitude of the oberserved subject. The person thus "stands in" for the actual subject.

This method has been used to reproduce the free-choice experiments, Festinger/Carlsmith $1/$20, Brehm (0.50/$1.00) experiment.

Critique of Self-Perception
Many questioned the use of these interpersonal simulations because they didn't tell the subjects the pre-experiment attitudinal state of the observed subject. Yet it makes sense that any prior attidtudes are less salient after one observes their own behavior in the new situation. Results of many comparisons between actual experiment and interpersonal simulation show it's best to conduct both at the same time.

Bem agrees that the interpersonal simulations only provide weak support for the self-perception theory. It's too easy to manipulate the answers depending on what information is given the observer-subject, and too easy to get multiple interpretations. No simulation is as good as the actual experiment.

Yet, the Snyder-Ebbeson experiment designed to test dissonance and self-perception theory landed on the side of self-perception.

Mis-attribution Effects
Schachter's drug-induced work and Valins heartbeat experiments showed that attributionof internal states can come from external situational forces. Even false information can influence internal states (snake-phobia experiment). Nisbett and Schachter got less shock response when they misattributed their feelings to a pill. Storms and Nisbett got similar results with a placebo pill with insomniacs.

Freedman and Fraser's "foot-in-the-door" studies. Their conclusions are post-hoc very similar to self-perception theory (people who acceded to the first request began to see themselves as people who do things like that in general).

Lepper did a similar experiement (a variation on the forbidden toy experiment) where after the standard threat and response (didn't play with toy) Lepper had the children do another honesty-testing activity later. Those who complied to the orignial request under mild threat showed more resistance in the second experiment.

Lepper's follow up experiment where they used children who had been determined to intrinsically like a particular activity (with hidden camera) were taken to another room and asked to do the activity (either for a reward, no reward, or a suprised award). Those with the suprise award were more likely to engage in theactivity unprovoked later.

Differences Between Self-Perception and Interpersonal Perception
There is an insider-vs-outsider difference, intimate vs stranger difference (knowledge of past behavior) A stranger is more likely to infer attidtude from the outward behavior when the individual may compare behavor to past actions as well. A third difference may be self-vs-other (where the self tries to maintain self-esteem).

These differences "conspire to create a pervasive tendency for actors to attribute their actions to situations requirements, where as observers tend to attribute the same actions to stable personal dispositions of the actor.

Paridigm Shift in Social Psychology
In the Sixties all persons were characterized by drives toward consistency and uncertainty reduction. Now people are seen as less driven, constantly asking oneself "How do you feel?" based on available internal and external evidence.

In psychology its a shift from motivational/drive models to information processing/attribution models.

Kelly combined Jones and Davis thoughts on attribution with self-perception theory to create attribution theory.

Note that neither attribution or self-perception theory account for how cognitive control of physiological processes.

Schchter notes that once people figure out their emotional state they will 'behave accordingly".

Do Attributions Mediate Behavior?
So far they do sometimes, but not always.