Natural statistics support a rational account of confidence biases
Abstract Previous work has sought to understand decision confidence as a prediction of the probability that a decision will be correct, leading to debate over whether these predictions are optimal, and whether they rely on the same decision variable as decisions themselves. This work has generally r...
Main Authors: | , , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
Nature Portfolio
2023-07-01
|
Series: | Nature Communications |
Online Access: | https://doi.org/10.1038/s41467-023-39737-2 |
_version_ | 1797784535114448896 |
---|---|
author | Taylor W. Webb Kiyofumi Miyoshi Tsz Yan So Sivananda Rajananda Hakwan Lau |
author_facet | Taylor W. Webb Kiyofumi Miyoshi Tsz Yan So Sivananda Rajananda Hakwan Lau |
author_sort | Taylor W. Webb |
collection | DOAJ |
description | Abstract Previous work has sought to understand decision confidence as a prediction of the probability that a decision will be correct, leading to debate over whether these predictions are optimal, and whether they rely on the same decision variable as decisions themselves. This work has generally relied on idealized, low-dimensional models, necessitating strong assumptions about the representations over which confidence is computed. To address this, we used deep neural networks to develop a model of decision confidence that operates directly over high-dimensional, naturalistic stimuli. The model accounts for a number of puzzling dissociations between decisions and confidence, reveals a rational explanation of these dissociations in terms of optimization for the statistics of sensory inputs, and makes the surprising prediction that, despite these dissociations, decisions and confidence depend on a common decision variable. |
first_indexed | 2024-03-13T00:41:17Z |
format | Article |
id | doaj.art-0d0ce14b85ec49c6b10e8f5781e6a146 |
institution | Directory Open Access Journal |
issn | 2041-1723 |
language | English |
last_indexed | 2024-03-13T00:41:17Z |
publishDate | 2023-07-01 |
publisher | Nature Portfolio |
record_format | Article |
series | Nature Communications |
spelling | doaj.art-0d0ce14b85ec49c6b10e8f5781e6a1462023-07-09T11:17:35ZengNature PortfolioNature Communications2041-17232023-07-0114111810.1038/s41467-023-39737-2Natural statistics support a rational account of confidence biasesTaylor W. Webb0Kiyofumi Miyoshi1Tsz Yan So2Sivananda Rajananda3Hakwan Lau4University of CaliforniaKyoto UniversityThe University of Hong KongHarvard UniversityLaboratory for Consciousness, RIKEN Center for Brain ScienceAbstract Previous work has sought to understand decision confidence as a prediction of the probability that a decision will be correct, leading to debate over whether these predictions are optimal, and whether they rely on the same decision variable as decisions themselves. This work has generally relied on idealized, low-dimensional models, necessitating strong assumptions about the representations over which confidence is computed. To address this, we used deep neural networks to develop a model of decision confidence that operates directly over high-dimensional, naturalistic stimuli. The model accounts for a number of puzzling dissociations between decisions and confidence, reveals a rational explanation of these dissociations in terms of optimization for the statistics of sensory inputs, and makes the surprising prediction that, despite these dissociations, decisions and confidence depend on a common decision variable.https://doi.org/10.1038/s41467-023-39737-2 |
spellingShingle | Taylor W. Webb Kiyofumi Miyoshi Tsz Yan So Sivananda Rajananda Hakwan Lau Natural statistics support a rational account of confidence biases Nature Communications |
title | Natural statistics support a rational account of confidence biases |
title_full | Natural statistics support a rational account of confidence biases |
title_fullStr | Natural statistics support a rational account of confidence biases |
title_full_unstemmed | Natural statistics support a rational account of confidence biases |
title_short | Natural statistics support a rational account of confidence biases |
title_sort | natural statistics support a rational account of confidence biases |
url | https://doi.org/10.1038/s41467-023-39737-2 |
work_keys_str_mv | AT taylorwwebb naturalstatisticssupportarationalaccountofconfidencebiases AT kiyofumimiyoshi naturalstatisticssupportarationalaccountofconfidencebiases AT tszyanso naturalstatisticssupportarationalaccountofconfidencebiases AT sivanandarajananda naturalstatisticssupportarationalaccountofconfidencebiases AT hakwanlau naturalstatisticssupportarationalaccountofconfidencebiases |