Scott Alexander, curated
← Back to curation

The limits of introspection

Quality
74
Strong
Claude Shift
45
Moderate
RWI
2
of 10

Summary

Marshals Nisbett & Wilson's classic 'Verbal Reports on Mental Processes' to argue we have no privileged introspective access to our own cognition — we infer our reasons the same way outside observers do, and often guess wrong. A parade of studies: the shock-tolerance placebo pill (subjects credit made-up reasons like 'I played with radios as a child'), bug-eating cognitive dissonance, Maier's pendulum cue (unrecognized), the bystander effect (others' presence unmentioned), Tide word-priming, the clothing position effect (subjects deny it 'with a worried glance...as if dealing with a madman'), and the teacher halo effect. The clincher: outsiders predicting the experiments guess the same wrong reasons subjects give, suggesting both are running the same blind inference. Supports Scott's thread that our introspected goals/preferences are produced by the machinery that infers others' goals.

Why this score

Quality 74 · Strong. Strong. A clear, well-marshaled review with an important, load-bearing conclusion (no privileged introspection) that anchors his eliminativist-about-preferences thread. Held at 74 as primarily a synthesis of established (Nisbett & Wilson) work. 74.

Claude’s paradigm shift 45 · Moderate. Moderate. Conveys established 1977-era cognitive science; the fresh element is fitting it to the 'we infer our own desires as we infer others' framing. 45.

Real-world impact 2 · Minor. A clear, well-marshaled review (via Nisbett & Wilson) of the load-bearing conclusion that we have no privileged introspective access to our own cognition — anchoring his eliminativist-about-preferences thread. Conceptual influence within rationalist discourse, primarily synthesis, no material change — low RWI.