Mistakes Were Made (But Not by Me)

Tavris and Aronson apply cognitive dissonance theory to self-justification, arguing that people who hold two conflicting cognitions, typically I am a competent and decent person and I did something incompetent or harmful, resolve the conflict by revising their view of the action rather than of themselves. The revision happens automatically and below awareness, which is why self-justification is not the same as lying.

The central image is the pyramid, in which two people who start close together make slightly different initial choices and then justify them, which moves them apart, and each subsequent justification widens the distance until they occupy opposite positions and cannot understand how the other arrived there.

The applications are the strongest part. They cover prosecutors and police who refuse to accept DNA exonerations, because accepting them requires accepting that they imprisoned an innocent person; therapists who elicited false memories of abuse and never retracted; doctors who cannot acknowledge diagnostic errors; the escalation of commitment in politics and war; and the dynamics of marriages where each partner justifies their own behaviour and reinterprets the other's history. They argue that memory itself is reconstructed to support the current self-image, so people are not lying when they misremember their own past positions.

The practical prescription is limited: separate mistakes from identity so that admitting error does not threaten self-worth, and build institutions that make acknowledgment cheap. The book offers no evidence that this works. Some of the supporting social psychology has replicated poorly, and dissonance research has itself been re-examined. The pyramid model and the accounts of institutional refusal to admit error are the durable parts and apply directly to investment decisions held past the evidence.