Back

Metacognitive Introspection Alters the Dynamics of Pre-Decisional Neural Evidence Accumulation

Dou, W.; Afrakhteh, S.; Samaha, J.

2024-11-26 neuroscience
10.1101/2024.11.26.625501 bioRxiv
Show abstract

Metacognitive introspection refers to the capacity to monitor and evaluate ones own performance, such as when expressing confidence in ones choice. In the context of perceptual decision making, current computational models of confidence typically assume that confidence is computed after the evidence collection process underlying the initial choice has unfolded. This account predicts that the act of introspection required in order to provide a confidence judgment about ones decision should have little impact on the dynamics of evidence accumulation leading up to that decision. We tested this by examining whether a neural proxy for perceptual evidence accumulation (the central-parietal positivity; CPP) exhibits different dynamics when participants do or do not rate confidence in their perceptual decisions. Behavioral results showed that confidence introspection increased decision accuracy and response time. Importantly, confidence introspection also resulted in a steeper CPP slope beginning shortly after ([~]250 ms) decision evidence was available. We further observed an effector-specific neural signal of confidence evidence which both built up prior to the perceptual choice and which was predictive of the amount of confidence subsequently expressed by the participants. Our results indicate that introspection can alter the dynamics of perceptual evidence accumulation, possibly by acting on the gain of evidence accumulation, and that motor-related confidence signals emerge in tandem with pre-decisional evidence accumulation signals rather than post-choice.

Matching journals

The top 5 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.