The DOI matches its Crossref record; 33 of the 33 references it cites were found in a registry or an index.
Metacognition and Uncertainty Communication in Humans and Large Language Models
Open accessAbstract
Metacognition—the capacity to monitor and evaluate one’s own knowledge and performance—is foundational to human decision-making, learning, and communication. As large language models (LLMs) become increasingly embedded in both high-stakes and widespread low-stakes contexts, it is important to assess whether, how, and to what extent they exhibit metacognitive abilities. Here, we provide an overview of the current knowledge of LLMs’ metacognitive capacities, how they might be studied, and how they relate to our knowledge of metacognition in humans. We show that although humans and LLMs can sometimes appear quite aligned in their metacognitive capacities and behaviors, it is clear many differences remain; attending to these differences is important for enhancing the collaboration between humans and artificial intelligence. Last, we discuss how endowing future LLMs with more sensitive and more calibrated metacognition may also help them develop new capacities such as more efficient learning, self-direction, and curiosity.
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
The cited source doesn’t support the claim as written.
In the citing paper
“Steyvers and Peters [58] systematically demonstrated that while LLMs produce well-calibrated confidence judgments”
From Self-Referential Introspection in Large Language Models: The Critical Threshold for Recursive Self-Improvement (2026) · cites this paper as [58]
Checked by a reasoning model — may be wrong.
Experimental — accuracy not yet measuredThe source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“this capability relies on surface-level statistical regularities rather than genuine metacognitive monitoring”
From Self-Referential Introspection in Large Language Models: The Critical Threshold for Recursive Self-Improvement (2026) · cites this paper as [58]
Experimental — accuracy not yet measuredThe cited source says what the claim attributes to it.
In the citing paper
“Steyvers and Peters emphasize the need for careful study of metacognition and communication of uncertainty in humans and LLMs, while highlighting important differences between the two [20].”
From Machine Psychometrics: A Mathematical Psychology of Artificial Intelligence (2026) · cites this paper as [20]
The cited source says what the claim attributes to it.
In the citing paper
“masking a fundamental mechanistic divergence from human cognition.”
From Self-Referential Introspection in Large Language Models: The Critical Threshold for Recursive Self-Improvement (2026) · cites this paper as [58]
In the citing paper
“and that people’s confidence judgments about AI can be poorly aligned with its actual accuracy (Steyvers and Peters, 2026).”
From Cognitive transparency and trust calibration in explainable AI systems: a psychological model of human-AI decision reliance (2026) · cites this paper as Steyvers and Peters, 2026
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
Fleming, S. M., & Lau, H. C. (2014) · Frontiers in Human Neuroscience · doi:10.3389/fnhum.2014.00443 · Cited by 1,408 · cited 2×
“…probability that a randomly sampled correctly answered question receives a higher confidence score than a randomly sampled incorrectly answered question—and a signal detection theoretic metric known as “meta-d ′ ” (analogous to d ′ from signal detection theory), among others (Fleming & Lau, 2014).”
Can LLMs Express Their Uncertainty? An Empirical Evaluation of Confidence Elicitation in LLMs
Xiong, M., Hu, Z., Lu, X., Li, Y., Fu, J., He, J., & Hooi, B. (2023) · arXiv · doi:10.48550/arxiv.2306.13063 · Cited by 618 · cited 2×
“…nature: the model can be prompted multiple times with the same question, and confidence can be inferred from the consistency of responses (Xiong et al., 2024) — similar to the other implicit measures of metacognition discussed above — or their semantic coherence (Farquhar et al., 2024).”
Confidence and certainty: Distinct probabilistic quantities for different goals
Pouget, A., Drugowitsch, J., & Kepecs, A. (2016) · Nature Neuroscience · doi:10.1038/nn.4240 · Cited by 528 · cited 2×
“We provide insights into the parallels and divergences between human and LLM metacognition throughout and discuss potential pathways for enhancing metacognitive interactions between humans and LLMs.”
The role of metacognition in human social interactions
Frith, C. D. (2012) · Philosophical Transactions of the Royal Society B: Biological Sciences · doi:10.1098/rstb.2012.0123 · Cited by 407 · cited 1×
“We provide insights into the parallels and divergences between human and LLM metacognition throughout and discuss potential pathways for enhancing metacognitive interactions between humans and LLMs.”
The self-consistency model of subjective confidence.
Koriat, A. (2012) · Psychological Review · doi:10.1037/a0025648 · Cited by 320 · cited 2×
“Confidence may reflect internal consistency or access to task-relevant information (Koriat, 2012) or the formation of second-order beliefs (Peters, 2022) Token likelihoods and response consistency are used to estimate uncertainty (Kadavath et al., 2022; Liu et al., 2025) Metacognitive training Some…”
Self-Directed Learning: A Cognitive and Computational Perspective
Gureckis, T. M., & Markant, D. B. (2012) · Perspectives on Psychological Science · doi:10.1177/1745691612454304 · Cited by 292
Morales, J., Lau, H., & Fleming, S. M. (2018) · The Journal of Neuroscience · doi:10.1523/jneurosci.2360-17.2018 · Cited by 274 · cited 2×
“…some shared processes underlie metacognition about perception, memory, and cognition may exist and rely on common neural structures, while others may be domain specific—i.e., separable computational or neural modules for perceptual versus cognitive or memory metacognition (Morales et al., 2018).”
Metacognition and Confidence: A Review and Synthesis
Fleming, S. M. (2024) · Annual Review of Psychology · doi:10.1146/annurev-psych-022423-032425 · Cited by 256 · cited 3×
“As discussed above, metacognitive sensitivity is the degree to which confidence judgments can discriminate between right and wrong answers, which is critical to effective decision-making in humans (Fleming, 2024).”
Three Challenges for AI-Assisted Decision-Making
Steyvers, M., & Kumar, A. (2024) · Perspectives on Psychological Science · doi:10.1177/17456916231181102 · Cited by 217 · cited 1×
“In closing, we consider how advances in LLM metacognition might contribute to the emergence of other cognitive functions relevant to intelligence.”
Self-citationWhat large language models know and what people think they know
Steyvers, M., Tejeda, H., Kumar, A., Belem, C., Karny, S., Hu, X., Mayer, L. W., & Smyth, P. (2025) · Nature Machine Intelligence · doi:10.1038/s42256-024-00976-7 · Cited by 209 · cited 14×
“Recent findings show that users to interpret longer LLM responses as more confident, even when the model’s internal confidence remains unchanged (Steyvers, Tejeda, et al., 2025).”
InfluentialSelf-citation
Research works printed but not linked: 6
- Betley J. Bao X. Soto M. Sztyber-Betley A. Chua J. Evans O. (2025). Tell me about yourself: LLMs are aware of their learned behaviors. arXiv. https://doi.org/10.48550/arXiv.2501.11120 · DOI
- Didolkar A. Goyal A. Ke N. R. Guo S. Valko M. Lillicrap T. Rezende D. Bengio Y. Mozer M. Arora S. (2024). Metacognitive capabilities of LLMs: An exploration in mathematical problem solving. arXiv. https://doi.org/10.48550/arXiv.2405.12205 · DOI
- Kadavath S. Conerly T. Askell A. Henighan T. Drain D. Perez E. Schiefer N. Hatfield-Dodds Z. DasSarma N. Tran-Johnson E. Johnston S. El-Showk S. Jones A. Elhage N. Hume T. Chen A. Bai Y. Bowman S. Fort S. . . . Kaplan J. (2022). Language models (mostly) know what they know. arXiV. https://doi.org/10.48550/arXiv.2207.05221 · DOI · Paper page
- Li Z., Steyvers M. (2025). The importance of metacognitive sensitivity in human-AI decision-making. Proceedings of the Annual Meeting of the Cognitive Science Society, 47, 1773–1779.
- 10.52202/079017-1364 · doi:10.52202/079017-1364
- Steyvers M. Belem C. Smyth P. (2025). Improving metacognition and uncertainty communication in language models. arXiv. https://doi.org/10.48550/arXiv.2510.05126 · DOI
Cited by
Nash, B. L., Zengilowski, A., Schuetze, B. A., & Schallert, D. L. (2025) · Educational Psychologist · doi:10.1080/00461520.2025.2592823 · Cited by 8
“Although some users may be able to interrogate the results of LLMs before using the output, it seems unlikely that most users have this ability (Steyvers & Peters, 2025), suggesting the need for MTC researchers to draw on algorithmic literacy studies to design interventions to support LLM-aided…”
Persuading large language models to comply with objectionable requests
Meincke, L., Shapiro, D., Duckworth, A. L., Mollick, E., Mollick, L., den Bulte, C. V., & Cialdini, R. B. (2026) · Proceedings of the National Academy of Sciences · doi:10.1073/pnas.2535868123 · Cited by 9
Zhang, J., Yuan, B., & Zhang, Q. (2026) · Entropy · doi:10.3390/e28090951 · Cited by 2
“Steyvers and Peters [58] systematically demonstrated that while LLMs produce well-calibrated confidence judgments, this capability relies on surface-level statistical regularities rather than genuine metacognitive monitoring, masking a fundamental mechanistic divergence from human cognition.”
Cybersophy: A framework for human cognition, identity, and values in the age of artificial agents
Ning, H., & Ding, J. (2026) · Computers in Human Behavior Artificial Humans · doi:10.1016/j.chbah.2026.100385 · Cited by 1
Towards an integrative neuroscience of metacognition
Fleming, S. M. (2026) · Nature reviews. Neuroscience · doi:10.1038/s41583-026-01081-x · Cited by 5
“In either case, the development of such metacognitive capacities in AI systems will likely be important in managing both self and social regulation in human–AI interactions 187 .”
Metadata
- Authors
- Steyvers, M., & Peters, M. A. K.
- Affiliations
- University of California, Irvine (US)
- Canadian Institute for Advanced Research (CA)
- Corresponding author
- Steyvers, Mark
- Venue
- Current Directions in Psychological Science
- Publisher
- SAGE Publications
- Published
- 2025-11-18
- Volume
- 35
- Issue
- 3
- Pages
- 131-139
- ISSN
- 0963-7214 (ISSN-L)
- 1467-8721
- OpenAlex
- W4416402838
- Cited by
- 18
- References
- 29
- Citations by year
- 2026: 16 · 2025: 2
- FWCI
- 10.26
- Type
- journal article
- Language
- en
- Open access
- hybrid
- doi.org
- Licence
- cc-by-nc
- creativecommons.org/licenses/by-nc/4.0/ (vor)
- journals.sagepub.com/page/policies/text-and-data-mining-license (tdm)
- Locations
- Current Directions in Psychological Science (publishedVersion, cc-by-nc)
- arXiv (Cornell University) (submittedVersion)
- Funders
- Canadian Institute for Advanced Research ((will be provided later))
- Topics
- Life Sciences › Neuroscience › Cognitive Neuroscience › Memory Processes and Influences
- Physical Sciences › Computer Science › Artificial Intelligence › Topic Modeling
- Physical Sciences › Computer Science › Artificial Intelligence › Psychiatry, Mental Health, Neuroscience
- Keywords
- metacognition, large language models, uncertainty communication, metacognitive calibration, human-AI collaboration, curiosity
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.
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: Metacognition and Uncertainty Communication in Humans and Large Language Models. https://app.cochat.ai/papers/10.1177/09637214251391158
- The paper
- Steyvers, M., & Peters, M. A. K. (2026). Metacognition and Uncertainty Communication in Humans and Large Language Models. Current Directions in Psychological Science. https://doi.org/10.1177/09637214251391158
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.