Week 7: Modelling the modeller
Why a self-model must be a summary, and what that predicts about introspective reports.
A complete model of a process cannot run inside that process, so self-models compress — and they do not mark what they left out.
Johansson and colleagues (2005) handed people the face they had rejected and asked why they chose it. Under a fifth of manipulations were detected; 84% had said in advance they would notice. Petitmengin and colleagues (2013) then showed the effect largely dissolves when people are given room to examine their experience rather than pressure to explain themselves, which weakens the strong reading and sharpens the interesting one.
The mirror test dispute closes the lecture, as an example of an assay asked to carry more than it can.
Teaching team
- Tuyet Nabarro — lecturer