Identity Pacts

What it is

Most self-control advice asks a person to win an argument with themselves, repeatedly, in the moment the temptation arrives. An identity pact changes what is being asked. Instead of committing to an action, you commit to a description of yourself, and the actions follow from the description without needing to be argued for each time. A vegetarian at a barbecue is not resisting the sausages. There is nothing to resist, because eating them is not a thing that person does.

The linguistic version of this is the difference between two refusals that mean the same thing on paper. "I can't eat that" positions the speaker as someone under a restriction, which implies an authority imposing it and, by extension, the possibility of an exemption. "I don't eat that" positions the speaker as a kind of person. The first invites negotiation, including internal negotiation. The second closes it.

The strongest evidence for the mechanism comes from Christopher Bryan's work at Stanford on voter turnout. Registered voters answered pre-election surveys phrased either with a verb, asking how important it was to them to vote, or with a noun, asking how important it was to them to be a voter. Their answers were then checked against public voting records rather than against what they said they had done. The noun framing produced what the authors described as among the largest experimental effects observed on measured turnout, and the result held in a second election. The work was published in PNAS. Bryan's own framing of why it works is that people may be more likely to vote when voting is represented as an expression of self, symbolic of a person's fundamental character, rather than simply as a behaviour.

A study in the Journal of Consumer Research found nearly twice as many people choosing the healthier snack after rehearsing "I don't eat that" rather than "I can't eat that." Nir Eyal, who named the technique in Indistractable, offers his own five years as a vegetarian as the personal case: the label made previously appealing food simply unpalatable, with no resistance involved. He also cites work by Eskreis-Winkler and Fishbach on teaching others as a form of self-reinforcement, and notes the finding that it works better when the teacher admits their own struggles rather than presenting as someone who has solved the problem.

One of the studies this chapter rests on has been retracted, and this page will not route around it. Eyal's argument about rituals, which sits alongside the identity material in the same chapter, depends on a Harvard Business School study in which an arbitrary three-step pre-meal ritual outperformed mindful eating on calories, fat and sugar. That study is Tian, Schroeder, Häubl, Risen, Norton and Gino (2018), "Enacting rituals to improve self-control," published in the Journal of Personality and Social Psychology. It was retracted in 2024, after the authors' own reanalysis found discrepancies against the underlying Qualtrics data which invalidated the self-control conclusion on the food-diary measure. It should not be cited, here or anywhere else. Those findings are contested in litigation, and the right thing to say is what was found and by whom, rather than anything about the person.

None of that touches the voter work or the "I don't" finding, which come from different authors and different laboratories. That is why this page carries a status of mixed rather than contested: the evidence is partly sound and partly withdrawn, and both halves are true of the same idea at the same time.