University of Sussex
Browse
Compain, Clémence.pdf (11.27 MB)

Investigating the influence of biases on perceptual metacognition

Download (11.27 MB)
thesis
posted on 2023-10-27, 12:31 authored by Clémence Compain

Perception is generally biased by our expectations about the world. However, it is not known whether or how these biases influence people’s ability to evaluate their perceptual decisions. Are individuals’ evaluations of their perceptual decision accuracy impaired when they are biased? In this thesis, I investigated this question by asking how biases – intrinsic or induced by changing expectations – impact perceptual confidence and metacognition. This topic was explored over three studies, using behavioural psychophysics paradigms and methods. The first empirical chapter investigated whether confidence and metacognition are modulated by the precision of expectations by changing stimulus probabilities either frequently (low precision) or infrequently (high precision). Results showed that the effect of expectation on confidence and metacognition was not stronger when the precision of expectations was higher, contradicting some computational models of confidence. The second chapter asked whether stimuli with expected orientations are perceived differently (as indicated by contrast reports), and whether confidence and metacognition about appearance judgments changed accordingly. The main finding is that unexpected stimuli were perceived as less contrasted than they really were, supporting the view that expectations shape appearance. No difference in confidence between expected and unexpected stimuli was found and participants’ metacognitive sensitivity was at chance, suggesting that people were relatively unaware of their perceptual performance. The third chapter studied whether biases can be voluntarily controlled, and whether confidence and metacognition are impacted by this effort. The main finding is that people were able to voluntarily reduce their bias on instruction. No effects of the effort to control their bias on confidence and metacognition were found, suggesting that good performance monitoring is not necessary to control bias. Taken together, this thesis offers a characterisation of how metacognition is influenced by perceptual biases, and motivates future work on whether people are aware of their perceptual biases.

History

File Version

  • Published version

Pages

254

Department affiliated with

  • Informatics Theses

Qualification level

  • doctoral

Qualification name

  • phd

Language

  • eng

Institution

University of Sussex

Supervisor

Anil Seth and Maxine Sherman

Usage metrics

    University of Sussex (Theses)

    Categories

    No categories selected

    Exports

    RefWorks
    BibTeX
    Ref. manager
    Endnote
    DataCite
    NLM
    DC