2021
DOI: 10.1037/amp0000937
|View full text |Cite
|
Sign up to set email alerts
|

Visual metacognition: Measures, models, and neural correlates.

Abstract: Visual metacognition is the ability to evaluate one's performance on visual perceptual tasks. The field of visual metacognition unites the long tradition of visual psychophysics with the younger field of metacognition research. This article traces the historical roots of the field and reviews progress in the areas of (a) constructing appropriate measures of metacognitive ability, (b) developing computational models, and (c) revealing the neural correlates of visual metacognition. First, I review the most popul… Show more

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
2
1
1
1

Citation Types

1
21
0

Year Published

2022
2022
2025
2025

Publication Types

Select...
6
3

Relationship

2
7

Authors

Journals

citations
Cited by 28 publications
(22 citation statements)
references
References 93 publications
1
21
0
Order By: Relevance
“…We examined six foundational signatures of human perceptual decision making that have already been established in studies of 2-choice tasks: 1) Human decisions are stochastic, meaning that the same stimulus can elicit different responses on different trials 17,18 , 2) increasing speed stress shortens RT but decreases accuracy (speed-accuracy trade-off) 10,19,20 , 3) more difficult decisions lead to reduced accuracy and longer RT 10,21,22 , 4) RT distributions are right-skewed, and this skew increases with task difficulty 10 , 5) RT is lower for correct than for error trials [22][23][24][25][26] , and 6) confidence is higher for correct than for error trials 27 . For each of these signatures, we confirmed that the signature also occurs for our 8-choice task with naturalistic images, and then tested whether RTNet and Anytime Prediction exhibit the same signature.…”
Section: Signatures Of Human Decision-makingmentioning
confidence: 99%
“…We examined six foundational signatures of human perceptual decision making that have already been established in studies of 2-choice tasks: 1) Human decisions are stochastic, meaning that the same stimulus can elicit different responses on different trials 17,18 , 2) increasing speed stress shortens RT but decreases accuracy (speed-accuracy trade-off) 10,19,20 , 3) more difficult decisions lead to reduced accuracy and longer RT 10,21,22 , 4) RT distributions are right-skewed, and this skew increases with task difficulty 10 , 5) RT is lower for correct than for error trials [22][23][24][25][26] , and 6) confidence is higher for correct than for error trials 27 . For each of these signatures, we confirmed that the signature also occurs for our 8-choice task with naturalistic images, and then tested whether RTNet and Anytime Prediction exhibit the same signature.…”
Section: Signatures Of Human Decision-makingmentioning
confidence: 99%
“…We examined six foundational signatures of human perceptual decision making that have already been established in studies of 2-choice tasks: 1) Human decisions are stochastic, meaning that the same stimulus can elicit different responses on different trials (Beck, Ma, Pitkow, Latham, & Pouget, 2012; Renart & Machens, 2014), 2) increasing speed stress shortens RT but decreases accuracy (speed-accuracy trade-off) (Forstmann et al, 2016; Heitz, 2014; Heitz & Schall, 2012), 3) more difficult decisions lead to reduced accuracy and longer RT (Forstmann et al, 2016; Ratcliff & Rouder, 1998; Wagenmakers & Brown, 2007), 4) RT distributions are right-skewed, and this skew increases with task difficulty (Forstmann et al, 2016), 5) RT is lower for correct than for error trials (Brown & Heathcote, 2008; Forstmann et al, 2008; Luce, 1986; Ratcliff, 2002; Wagenmakers & Brown, 2007), and 6) confidence is higher for correct than for error trials (Rahnev, 2021). For each of these signatures, we confirmed that the signature also occurs for our 8-choice task with naturalistic images, and then tested whether RTNet and Anytime Prediction exhibit the same signature.…”
Section: Resultsmentioning
confidence: 99%
“…Finally, a ubiquitous feature of confidence ratings is that they are higher for correct compared to incorrect decisions (Rahnev, 2021; Yeung & Summerfield, 2012). Our human data replicated this effect (F(1,59) = 472.17, p < 0.0001; Figure 4F ).…”
Section: Resultsmentioning
confidence: 99%
“…However, people are typically not fully aware of the accuracy of a decision; observers often display metacognitive inefficiency (i.e., M-ratio < 1 and log M-ratio < 0) (62). In contrast, observers occasionally exhibit superefficiency (i.e., M-ratio > 1 and log M-ratio > 0) in that they seemingly use more information than the theoretical maximum (63, 64). Although superefficiency is not well understood, the nonoptimal metacognition (i.e., either inefficiency or superefficiency) implies (at least partially) distinct mechanisms for first-order decisions and confidence ratings.…”
Section: Methodsmentioning
confidence: 99%