The DOI matches its Crossref record; 44 of the 44 references it cites were found in a registry or an index; 1 of the 2 sources its claims cite were checked for support.
Common computations for metacognition and meta-metacognition
Open accessAbstract
Abstract Recent evidence shows that people have the meta-metacognitive ability to evaluate their metacognitive judgments of confidence. However, it is unclear whether meta-metacognitive judgments are made by a different system and rely on a separate set of computations compared to metacognitive judgments. To address this question, we asked participants (N = 36) to perform a perceptual decision-making task and provide (i) an object-level, Type-1 response about the identity of the stimulus; (ii) a metacognitive, Type-2 response (low/high) regarding their confidence in their Type-1 decision; and (iii) a meta-metacognitive, Type-3 response (low/high) regarding the quality of their Type-2 rating. We found strong evidence for the existence of Type-3, meta-metacognitive ability. In a separate condition, participants performed an identical task with only a Type-1 response followed by a Type-2 response given on a 4-point scale. We found that the two conditions produced equivalent results such that the combination of binary Type-2 and binary Type-3 responses acts similar to a 4-point Type-2 response. Critically, while Type-2 evaluations were subject to metacognitive noise, Type-3 judgments were made at no additional cost. These results suggest that it is unlikely that there is a distinction between Type-2 and Type-3 systems (metacognition and meta-metacognition) in perceptual decision-making and, instead, a single system can be flexibly adapted to produce both Type-2 and Type-3 evaluations recursively.
Overview
CoChat Summary
## TL;DR
This study tested whether meta-metacognitive judgments (Type-3) in perception are computed by a different system than metacognitive judgments (Type-2). Across a perceptual task in 36 participants, the authors found equivalent behavioral effects for a 4-point Type-2 rating and a binary Type-2 + Type-3 format, suggesting a common computation.
## Key findings
- In the Type-2/Type-3 condition, high Type-3 ratings were associated with higher metacognitive efficiency (Mratio) than low Type-3 ratings: mean = 0.96 vs 0.46, t(35) = 4.39, P < .001.
- In the Type-2-only condition, after converting the 4-point confidence scale into pseudo-Type-2/Type-3 ratings, high Type-3 ratings again had higher Mratio than low ratings: mean = 0.98 vs 0.45, t(33) = 4.90, P < .001.
- The Mratio difference between high and low Type-3 judgments did not differ between conditions: Type-2-only mean difference = 0.527, Type-2/Type-3 mean difference = 0.532, BF01 = 5.439; linear mixed models also found no task-condition effect or interaction.
- Type-1 accuracy and Type-1 decision RT both tracked confidence similarly across conditions: higher confidence meant higher accuracy and faster RT, with no main effect of task condition and no confidence × condition interaction. The authors also found no extra noise for Type-3 judgments: d' decreased with distance from the decision criterion in both conditions, with equivalent slopes (BF01 = 10.084).
## Methods
- 36 adult participants remained after exclusions from an initial 40 recruited on Prolific; each completed 400 trials in each of two conditions, with condition order counterbalanced across participants.
- The task was a perceptual dot-counting judgment: participants chose which of two black squares contained more white dots, after which they gave either a 4-point Type-2 confidence rating (“Type-2-only”) or binary Type-2 plus binary Type-3 ratings (“Type-2/Type-3”).
- Analyses used signal detection theory measures (d', meta-d', Mratio) and linear mixed-effects models with participant as a random effect; the authors also compared Type-1 accuracy, Type-1 RT, and d' as a function of confidence criteria.
## Limitations
- The paper states that its conclusions are specific to the perceptual domain and may not generalize to other domains such as memory or theory of mind.
- The authors note that the study cannot completely falsify the possibility of separate Type-2 and Type-3 systems performing the same computations in this task; they frame the unified-system account as the most parsimonious hypothesis.
- Evident from the text: the task used a single simple perceptual discrimination with only one stimulus type (100 vs 85 dots), so the evidence is narrow in scope.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Figures



Verification
Claims the paper makes
1 Valid · 0 Not valid · 1 Not checked · 40 Own results and interpretations
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“no interaction between Type-2 confidence and task condition (P = .368).”
§ Common computations for metacognition and meta-metacognition > Results > Equivalent accuracy and RT in Type-2/Type-3 and Type-2-only conditions
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“a Bayesian paired-sample t-test revealed that the Mratio difference between high and low pseudo-Type-3 ratings in the Type-2-only condition (mean = 0.527) did not differ from the same difference in the Type-2/Type-3 condition (mean = 0.532; Bayes Factor (BF)01 = 5.439).”
§ Common computations for metacognition and meta-metacognition > Results > Equivalent meta-metacognitive performance in Type-2/Type-3 and Type-2-only conditions
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Post hoc t-tests further confirmed that there was no difference in Type-1 accuracy between the two conditions for each of the four Type-2 confidence levels [confidence = 1: t(55.655) = 1.460, P = .150, BF01 = 1.586; confidence = 2: t(55.226) = 0.328, P = .744, BF01 = 3.903; confidence = 3: t(68.494) = −0.202, P = .841, BF01 = 4.043; confidence = 4: t(68.926) = 0.734, P = .465, BF01 = 3.246].”
§ Common computations for metacognition and meta-metacognition > Results > Equivalent accuracy and RT in Type-2/Type-3 and Type-2-only conditions
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“a linear LLM revealed that task order did not moderate the effect of Type-3 certainty on Mratio [F(1,34) = 0.183, P = .672].”
§ Common computations for metacognition and meta-metacognition > Results > Limited influence of task order on the observed behavioral similarities
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“another LLM revealed no effect of task order on the relationship between the distance of each confidence criterion to the decision criterion and the sensitivity (d') associated with that confidence criterion [F(1, 252) = 1.520, P = .210].”
§ Common computations for metacognition and meta-metacognition > Results > Limited influence of task order on the observed behavioral similarities
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
25 more claims
Create a free account to see all 42 claims and every source verdict.
- Probably not a substantial claim. 8
- It may state two separate claims as one. 5
Statistics that don’t add up
Each reported p-value recomputed from its test statistic and degrees of freedom.
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
“F(1, 252) = 1.520, P = .210”
§ Common computations for metacognition and meta-metacognition > Results > Limited influence of task order on the observed behavioral similarities
Reported p = .21 · recomputed p = .219
Rated minor: it would not touch a result — a value no result rests on, a p-value that stays on the same side of the significance level, or a conflict that is clerical, resolved elsewhere in the paper, or outside its own findings.
In the paper
“F(1, 247.56) = 4.361, P = .039”
§ Common computations for metacognition and meta-metacognition > Results > Limited influence of task order on the observed behavioral similarities
Reported p = .039 · recomputed p = .038
Statistics recomputed
Every reported test result, recomputed from its test statistic and degrees of freedom. A result shows as not adding up only once an independent check has confirmed it.
- Doesn’t add upF(1, 247.56) = 4.361reported p = .039 · recomputed p = .038
“F(1, 247.56) = 4.361, p = .039”
- Doesn’t add upF(1, 252) = 1.52reported p = .21 · recomputed p = .219
“F(1, 252) = 1.520, p = .210”
- Consistentt(35) = 4.39reported p < .001 · recomputed p < .001
“t(35) = 4.39, p < .001”
- Consistentt(33) = 4.9reported p < .001 · recomputed p < .001
“t(33) = 4.90, p < .001”
- Consistentt(55.655) = 1.46reported p = .15 · recomputed p = .150
“t(55.655) = 1.460, p = .150”
13 more recomputed tests
Create a free account to see every test the Stats check recomputed.
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
Self-check results appear here once an independent check has confirmed them.
Citations
Cites
Signal detection theory and psychophysics
Ingleby, J. D. (1967) · Journal of Sound and Vibration · doi:10.1016/0022-460x(67)90197-6 · Cited by 8,509
The Psychology of Learning and Motivation
Ross, B. H. (2012) · Cited by 2,633
Metamemory: A Theoretical Framework and New Findings
Nelson, T. O. (1990) · Psychology of Learning and Motivation · doi:10.1016/s0079-7421(08)60053-5 · Cited by 2,583
jsPsych: A JavaScript library for creating behavioral experiments in a Web browser
de Leeuw, J. R. (2015) · Behavior Research Methods · doi:10.3758/s13428-014-0458-y · Cited by 1,778 · cited 1×
“The experiment was programmed using jsPsych library (Version 5.0.3; de Leeuw 2015).”
MethodInfluentialMacmillan, N. A., & Creelman, C. D. (2004) · doi:10.4324/9781410611147 · Cited by 1,269
Representation of Confidence Associated with a Decision by Neurons in the Parietal Cortex
Kiani, R., & Shadlen, M. N. (2009) · Science · doi:10.1126/science.1169405 · Cited by 1,027
Maniscalco, B., & Lau, H. (2012) · Consciousness and Cognition · doi:10.1016/j.concog.2011.09.021 · Cited by 883
Relating Introspective Accuracy to Individual Differences in Brain Structure
Fleming, S. M., Weil, R. S., Nagy, Z., Dolan, R. J., & Rees, G. (2010) · Science · doi:10.1126/science.1191883 · Cited by 809
Bahrami, B., Olsen, K., Latham, P. E., Roepstorff, A., Rees, G., & Frith, C. D. (2010) · Science · doi:10.1126/science.1185718 · Cited by 738 · cited 1×
“…(Fleming et al. 2010, Morales et al. 2018, Shekhar and Rahnev 2018, Zheng et al. 2021) and can guide learning (Guggenmos et al. 2016), cognitive offloading (Gilbert et al. 2020), information seeking (Desender et al. 2018 ), and social interactions (Bahrami et al. 2010, Pescetelli and Yeung 2021).”
BackgroundTwo-stage dynamic signal detection: A theory of choice, decision time, and confidence.
Pleskac, T. J., & Busemeyer, J. R. (2010) · Psychological Review · doi:10.1037/a0019737 · Cited by 722
Cited by
A comprehensive assessment of current methods for measuring metacognition
Rahnev, D. (2025) · Nature Communications · doi:10.1038/s41467-025-56117-0 · Cited by 98
Self-citationGoing ‘meta’: a systematic review of metacognition and functional neurological disorder
Sadnicka, A., Strudwick, A.-M., Grogan, J. P., Manohar, S., & Nielsen, G. (2024) · Brain Communications · doi:10.1093/braincomms/fcaf014 · Cited by 9
“35 Humans are considered to have the capacity to perform repeated hierarchical evaluations of their performance up to at least fourth-order judgements (i.e. meta-meta-metacognition) and such segregations may be relevant in FND. 36 There are also many potential reasons why the data do not yet…”
Automatic multisensory integration follows subjective confidence rather than objective performancePreprint
Gao, Y., Xue, K., Odegaard, B., et al. (2023) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2023.06.07.544029 · Cited by 7
Self-citationHaidemariam, T. (2026) · Frontiers in Artificial Intelligence · doi:10.3389/frai.2025.1728738 · Cited by 12
Sherman, M. T., & Seth, A. K. (2024) · Neuroscience of Consciousness · doi:10.1093/nc/niae020 · Cited by 12
“Taken together, we have shown, consistent with other recent work (Recht et al. 2022, Zheng et al. 2023), that human participants performing a simple visual task are able to discriminate the quality of their confidence reports, i.e. report when their confidence is a better versus worse predictor of their accuracy.”
Result
Metadata
- Authors
- Zheng, Y., Recht, S., & Rahnev, D.
- Affiliations
- Georgia Institute of Technology (US)
- University of Oxford (GB)
- Corresponding author
- Zheng, Yunxuan
- Venue
- Neuroscience of Consciousness
- Publisher
- Oxford University Press (OUP)
- Published
- 2023-01-01
- Volume
- 2023
- Issue
- 1
- Pages
- niad023-niad023
- ISSN
- 2057-2107 (ISSN-L)
- PMC
- PMC10693288
- OpenAlex
- W4389262286
- Cited by
- 15
- References
- 44
- Citations by year
- 2026: 4 · 2025: 5 · 2024: 6
- FWCI
- 2.24
- Type
- journal article
- Language
- en
- Open access
- gold
- doi.org
- Locations
- Neuroscience of Consciousness (publishedVersion, cc-by)
- PubMed Central (publishedVersion, cc-by)
- PubMed (publishedVersion)
- Funders
- Office of Naval Research (N00014-20-1-2622)
- National Institute of Health (R01MH119189)
- Topics
- Life Sciences › Neuroscience › Cognitive Neuroscience › Visual perception and processing mechanisms
- Life Sciences › Neuroscience › Cognitive Neuroscience › Memory Processes and Influences
- Life Sciences › Neuroscience › Cognitive Neuroscience › Neural and Behavioral Psychology Studies
- Keywords
- metacognition, confidence judgments, perceptual decision making, metacognitive skills, metacognitive confidence
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
A language model drew the claims from the paper on 2026-10-10, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-10-10 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
How to cite
Cite CoChat for the checks on this page, and the paper for its own findings.
- This page
- CoChat. (2026). Citation check: Common computations for metacognition and meta-metacognition. https://app.cochat.ai/papers/10.1093/nc/niad023
- The paper
- Zheng, Y., Recht, S., & Rahnev, D. (2023). Common computations for metacognition and meta-metacognition. Neuroscience of Consciousness. https://doi.org/10.1093/nc/niad023
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.