Definition
Self-perception theory, developed by Daryl Bem in 1967, holds that we determine our own attitudes, preferences, and identity largely by observing our own behavior — much as we infer the attitudes of others by watching theirs. Internal states are often ambiguous or weak, and the brain treats behavior as the most reliable evidence about who we are. If I gave to charity last month, I learn that I am the kind of person who gives to charity, and that inferred identity then writes the default for next month's charity decision.
Bem proposed self-perception as a simpler alternative to cognitive dissonance theory. The two theories make similar predictions about how attitudes follow behavior, but they differ in mechanism — dissonance posits an aversive tension that motivates attitude change, while self-perception posits a more dispassionate inferential process.
Why it matters
How it works
When asked to introspect on our attitudes, we rarely have direct access to a clear internal answer. We look instead at what we have done lately. If I sorted recycling carefully this morning, I infer that I care about the environment. If I gave to a charity solicitation, I infer that I support that cause. These inferences then become the cached "attitude" we report when asked, and they shape subsequent decisions.
The implication for influence is profound. A compliance professional who can produce any behavior in the target — sign a petition, accept a small gift, attend a meeting — has produced evidence the target's brain will use to update its self-model. The next, larger request is no longer evaluated by the original self; it is evaluated by a slightly modified self that already supports the cause.
The defense is to be intentional about which behaviors you allow to become evidence. Performative behaviors under social pressure should be discounted in self-evaluation. Decisions made for engineered reasons (a deadline, a free gift) should be flagged as not load-bearing on your identity.