Self-Signalling

What it is

The older name is self-perception theory. Bem set it out in a 1967 paper in Psychological Review and a long 1972 chapter in Advances in Experimental Social Psychology. People, he argued, have no direct readout of what they believe and prefer, so they infer it from what they do and from the circumstances in which they did it. The inference is strongest when internal cues are weak or ambiguous.

The later name is self-signalling. It comes from behavioural economics, and it changes the direction of travel. Bem's observer reads behaviour that has already happened. The self-signaller chooses the behaviour partly because of the reading it will produce. Drazen Prelec and Ronit Bodner gave this a formal version in a 2003 chapter. A choice, on their account, has two kinds of reward. Outcome utility comes from what the action directly produces. Diagnostic utility is the lift or the dent to self-image from what the action reveals about the person who took it.

Their everyday examples include giving to charity, drinking before noon, jogging in a heat wave, voting, and buying an expensive item on credit. The last of these is why the concept matters to a site about consumer behaviour. It means that a purchase can be partly a message to the buyer.

The popular version runs the same logic in reverse, as a method. Act like the person you want to be and you hand yourself the evidence. Benjamin Hardy uses it that way in Personality Isn't Permanent, quoting Adam Grant's line from a commencement speech that he "wasn't being true to myself, I was being true to the self I wanted to become" (p. 61). Hardy never names Bem or the literature, so the book is a source for the practice and not for the evidence.

In effect

The standard laboratory demonstration is the cold-water experiment in a 1984 paper by George Quattrone and Amos Tversky. Students held an arm in cold water until it was intolerable, exercised, and repeated the test. They had been told that the way exercise changed their tolerance pointed to a long, healthy life, or to the reverse, depending on their group. On the second trial their tolerance moved toward the good-news pattern. A second experiment in the same paper, with undergraduates, found that more people said they would vote when a theory let them believe their own vote showed what like-minded people would do, although a single vote caused the same thing under either theory.

The cost of the act appears to matter. Ayelet Gneezy and four colleagues reported in Management Science in 2012 that a costly prosocial act signals a prosocial identity, and that people then keep behaving in line with it. A costless act signals little, so later behaviour is less consistent and may decline, which the authors tie to moral licensing. A laboratory experiment and a large field experiment supported the account. The tested behaviour was prosocial, and the published abstract does not separate observed choices from unobserved ones.

The cleanest consumer-adjacent test of the underlying mechanism is the foot-in-the-door effect, where agreeing to a small request makes a larger one easier to grant. A 1975 field experiment tested the self-perception explanation directly: about half of the people agreed to a moderate request after a small yes, against about a third of a control group and about a fifth after a large request they were almost sure to refuse. Later reviews were less kind to the explanation. A 1983 meta-analysis judged it too imprecise to make clear predictions, and a 1984 meta-analysis called it seriously flawed as an account of the effect.

What it does not say

It does not say that purchases are mainly self-directed. The experiments that support self-signalling used a cold-water task, a voting question and charitable acts, not an ordinary retail purchase set against an unobserved control.

It does not settle whether the signal needs a witness. Simler and Hanson, in The Elephant in the Brain, argue that incentives that begin and end inside one's head "ultimately lead nowhere" (p. 295) and that "ignorance is at its most useful when it is most public" (p. 79). Taken straight, that turns self-signalling into an audience effect in disguise. The rule is asserted in the book and not tested, and a person who enjoys an act and a person who wants credit for it both confirm the model, which is the pattern this site calls the unfalsifiable frame. Bem's mechanism, in its own formulation, does not require an audience.

It does not say that Prelec and Bodner's model is an experimental result. Their chapter is theoretical and reports no new experiments. They list limits themselves: the model needs people to discount their own motives when they read their actions, which may not be psychologically accurate, and it assumes people do not learn across repeated choices.

And it does not make the cold-water result a claim about shopping. What survives is narrower and still useful. People read their own actions as evidence about themselves, they shade those actions when the reading matters to them, and the size of the signal depends on what the act cost.

---