Timothy D. Wilson

Confabulation

in Black box

Timothy D. Wilson (Timothy DeCamp Wilson), American social psychologist. Co-author with Richard Nisbett of "Telling More Than We Can Know" (Psychological Review, 1977); he spent his career at the University of Virginia. He later wrote Strangers to Ourselves (2002) on the adaptive unconscious and did influential work on affective forecasting.

Stake§

Scientific and reputational — the same challenge to the validity of verbal self-report, with no commercial or ideological interest.

Wilson's contribution to this topic is as co-author of the confabulation paper, and his later Strangers to Ourselves extends its claim: much of mental life runs in an adaptive unconscious the person cannot introspect, so the self they report is partly a construction. The position is the human counterpart to the machine question of whether a system's stated reasons match its computation, taken up for language models in Turpin.

Works in this corpus§

their concepts on the territory
ConfabulationConfabulation

1 concept in this scholar's webopen the full territory →

excerpts

Evidence is reviewed which suggests that there may be little or no direct introspective access to higher order cognitive processes.
Telling More Than We Can Know: Verbal Reports on Mental Processes (1977)

The sentence that founded the [[concept:confabulation|confabulation]] literature: people cannot watch their own higher cognition happen. Whatever they report about why they did something is produced by some other route than looking inward.

on Confabulation

It is proposed that when people attempt to report on their cognitive processes, that is, on the processes mediating the effects of a stimulus on a response, they do not do so on the basis of any true introspection. Instead, their reports are based on a priori, implicit causal theories, or judgments about the extent to which a particular stimulus is a plausible cause of a given response.
Telling More Than We Can Know: Verbal Reports on Mental Processes (1977)

The mechanism: a verbal report is a plausible story assembled from folk theories of cause, not a readout of the process it claims to describe. This is the exact structure [[source:turpin-2023-language-models-dont-always-say-what-they-think|Turpin]] finds in a language model's chain of thought — a plausible account that need not match the real cause.

on Post-hoc rationalisation, Confabulation