Quantifying confidence in perception and knowledge
Matteo Lisi

back to Overview

Date: Wednesday, 21.12.2022 15:20 CET

Signup: If you would like to attend the talks please register here to get a Zoom link.

Abstract:

Every choice we make in life is accompanied by a sense of confidence – a subjective feeling about how likely it is that we are choosing the best possible course of action. Importantly, confidence is not only used to critically re-evaluate past decisions but serves a pivotal role in successfully guiding future behaviour in many situations, in particular when outcomes are not immediately available. In the first part of this talk, I will present a psychophysical method based on a ‘dual-decision’ task that allows inferring miscalibration of confidence (such as systematic biases of over/under-confidence) in a principled manner by modelling how confidence is used to guide sequences of decisions. I will show that even in simple perceptual decision tasks humans display systematic biases of confidence that make them fall short of idealised Bayesian models. In the second part of the talk, I will show how ratings of confidence can be used to quantify the effect of misinformation on public belief. Using data from a large sample representative of GB population I will show that the sensitivity of confidence (the extent to which confidence ratings are diagnostic of accuracy) around COVID-19 information predicted health protective behaviours and compliance to public health measures during the lockdown, suggesting that it reflects the assimilation of misinformation. This work opens up new pathways to understanding the role of confidence in a wide range of human behaviours.