AI ClaimsRapid review

Carol Tavris and Elliot Aronson · 2007

Mistakes Were Made (But Not by Me)

Why We Justify Foolish Beliefs, Bad Decisions, and Hurtful Acts

Cover via Open Library

Rough AI truth score

67/100

The book's central warning is substantially right: people reinterpret choices and actions to protect coherent, favorable self-views, and that can impede correction. Yet cognitive dissonance is a family of mechanisms and paradigms rather than one settled universal process. Robust choice-induced preference change coexists with failed induced-compliance replications and important alternative explanations.

Based on three central claims · medium confidence

The three claims

01Mostly supported

Choosing between close alternatives can shift later preferences toward the chosen option and away from the rejected one.

Artifact-aware experiments and meta-analysis find choice-induced preference change after procedures designed not to confuse choice with previously unmeasured preference. Effects vary by design and do not establish that every rationalization is unconscious or caused by one dissonance mechanism.

02Mostly supported

Self-justification can escalate commitment and impede apology, error correction, or belief revision.

Dissonance research documents preference change, effort justification, selective exposure, and behavior-consistent attitude shifts. Institutional escalation also depends on incentives, reputation, sunk costs, group norms, and power, so psychological consistency is a contributor rather than a sufficient explanation.

03Mixed

A single cognitive-dissonance mechanism explains most persistent false beliefs and institutional mistakes.

Consistency and self-image motives explain important cases, but the induced-compliance paradigm recently failed in a large multilab replication and researchers dispute the correct manipulation and inference. Motivated reasoning, identity, incentives, information environments, memory, and coordination failures provide additional mechanisms.

Other claims worth checking
  • Public commitments can make reversal reputationally costly.
  • Acknowledging error is easier when identity and belonging are not simultaneously threatened.

What this number means. It is an AI-generated first-pass judgment of three central factual or causal claims—not a rating, exhaustive fact-check, or human peer review. Claim credits are 100% for supported, 75% for mostly supported, 50% for mixed, and 25% for weak, then averaged and rounded. Lower confidence means the score should move more as better evidence arrives.

Method three-central-claims/0.1.0 · checked 2026-09-01 · 3/3 selected claims assessed · method and source audit