Student or educator? Sign up with your school email and get 6 months of Pro free.
CoChat

The DOI matches its Crossref record; 44 of the 45 references it cites were found in a registry or an index; 10 of the 26 sources its claims cite were checked for support.

Folk psychological attributions of consciousness to large language models

Open access
Colombatto, Clara, Fleming, Stephen M · 2024 · Neuroscience of Consciousness

Abstract

Abstract Technological advances raise new puzzles and challenges for cognitive science and the study of how humans think about and interact with artificial intelligence (AI). For example, the advent of large language models and their human-like linguistic abilities has raised substantial debate regarding whether or not AI could be conscious. Here, we consider the question of whether AI could have subjective experiences such as feelings and sensations (‘phenomenal consciousness’). While experts from many fields have weighed in on this issue in academic and public discourse, it remains unknown whether and how the general population attributes phenomenal consciousness to AI. We surveyed a sample of US residents (n = 300) and found that a majority of participants were willing to attribute some possibility of phenomenal consciousness to large language models. These attributions were robust, as they predicted attributions of mental states typically associated with phenomenality—but also flexible, as they were sensitive to individual differences such as usage frequency. Overall, these results show how folk intuitions about AI consciousness can diverge from expert intuitions—with potential implications for the legal and ethical status of AI.

Overview

CoChat Summary

TL;DR

A survey of 300 US adults found that 67% attributed some possibility of phenomenal consciousness to ChatGPT, and these attributions increased with usage frequency. Folk attributions were driven by mental states related to “Experience” rather than “Intelligence,” and participants overestimated how conscious others would judge ChatGPT to be.

Key findings

  • 67% of participants attributed some possibility of phenomenal consciousness to ChatGPT, while 33% said it was definitely not an experiencer; consciousness ratings had M = 25.56, median = 16.00, SD = 27.36, range = 1–100.
  • Familiarity mattered: participants who had used ChatGPT attributed higher consciousness (M = 29.59) than those who never used it (M = 19.37; t(287) = 3.33, P < .001), with a linear increase across usage frequency (B = 4.94, SE = 0.99, t = 4.99, P < .001).
  • Ratings of 65 mental states reduced to two dimensions—Experience and Intelligence—explaining 58% of variance. ChatGPT was seen as more capable of Intelligence than Experience (r = 0.95 and r = −0.94, respectively), but only Experience-related states predicted consciousness attributions (r = 0.65, P < .001) and not Intelligence-related states (r = 0.03, P = .596).
  • Participants predicted others would attribute more consciousness to ChatGPT than they themselves did (predicted M = 41.11, median = 39.50, SD = 25.28 vs. own M = 25.56; t(299) = 10.90, P < .001), though own and predicted ratings correlated (r = 0.56, P < .001).

Methods

  • Nationally representative sample of 300 US residents recruited via Prolific Academic in July 2023 using proportional stratified random sampling with age and gender quotas based on US Census Bureau data (female = 152; male = 142; non-binary = 2; did not answer = 4; mean age = 46.13 years).
  • Participants read a description of ChatGPT and a description of phenomenal experience adapted from Peressini (2014), then rated ChatGPT’s capacity for conscious experience on a 1–100 scale, confidence in that judgment, predicted others’ judgments, and rated ChatGPT on 65 mental-state attributes from 1–100.
  • Analyses included principal component analysis of the 65 mental-state ratings, correlations, t-tests, and regression models; materials, anonymized raw data, and analysis code are available on OSF.

Limitations

  • The paper states that consciousness was measured via a single question, which may be sensitive to different measures and prompts; if the scale was interpreted as credence rather than degree, 23% of participants considered it more likely than not that ChatGPT has human-like experience.
  • The paper states that self-report may not fully capture intuitions about others’ mental states compared with indirect behavioral markers, and that it remains unclear how folk conceptions of phenomenality map onto philosophical constructs; the preceding explanation of phenomenality may have induced a concept participants would not otherwise spontaneously consider.
  • The paper states that the stratified US sample may not generalize across cultures, that online participants likely use computers daily, and that attitudes reflect a specific moment; the usage–consciousness link is correlational, so causal direction remains unclear.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.1093/nc/niae013

Cites

44 of 45 references verified1 citation errors · 0 not checkable

Statistics

Figures

Figure 1. Attributions of consciousness to an LLM. Participants attributed varying levels of consciousness to ChatGPT (panel a) and these attributions increased with usage frequency (b). When asked to predict the extent to which other people on average would think ChatGPT is conscious (c), participants consistently overestimated public opinion (d). Error bars and bands represent 95% CIs.
Figure 1. Attributions of consciousness to an LLM. Participants attributed varying levels of consciousness to ChatGPT (panel a) and these attributions increased with usage frequency (b). When asked to predict the extent to which other people on average would think ChatGPT is conscious (c), participants consistently overestimated public opinion (d). Error bars and bands represent 95% CIs.CC BY-NC · PubMed Central
Figure 2. Structure of mental state attributions to ChatGPT. Participants’ ratings of ChatGPT’s mental capacities mapped onto two dimensions—’Experience’ and ‘Intelligence’. While ChatGPT was seen as more capable of mental states related to Intelligence than Experience (a), only those related to Experience were predictive of phenomenal consciousness attributions (b).
Figure 2. Structure of mental state attributions to ChatGPT. Participants’ ratings of ChatGPT’s mental capacities mapped onto two dimensions—’Experience’ and ‘Intelligence’. While ChatGPT was seen as more capable of mental states related to Intelligence than Experience (a), only those related to Experience were predictive of phenomenal consciousness attributions (b).CC BY-NC · PubMed Central

Verification

Claims the paper makes

7 Valid · 0 Not valid · 8 Not checked · 24 Own results and interpretations

9 more claims

Create a free account to see all 39 claims and to verify the 2 references still open.

See all claims — free account
  • It may state two separate claims as one. 16
  • Probably not a substantial claim. 16
  • The quote only partly states this claim. 1

Create a free account to see these claims.

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?

7 more claims about this paper

Create a free account to see every claim other papers make about this one.

See all claims — free account

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.

  • ConsistentF(1, 297) = 53.24reported p < .001 · recomputed p < .001

    “F(1, 297) = 53.24, p < .001”

  • Consistentt(287) = 3.33reported p < .001 · recomputed p < .001

    “t(287) = 3.33, p < .001”

  • Consistentt(299) = 10.9reported p < .001 · recomputed p < .001

    “t(299) = 10.90, p < .001”

3 of 3

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

Show all 31 references

Research works printed but not linked: 1

  1. Wiese, 2023, Could large language models be conscious? A perspective from the free energy principle

Cited by

See all 96

Metadata

Authors
Colombatto, C., & Fleming, S. M.
Affiliations
  • University of Waterloo (CA)
  • Wellcome Centre for Human Neuroimaging (GB)
Corresponding author
Colombatto, Clara
Venue
Neuroscience of Consciousness
Published
2024-01-01
Volume
2024
Issue
1
Pages
niae013-niae013
ISSN
2057-2107 (ISSN-L)
OpenAlex
W4394716098
Cited by
96
References
31
Citations by year
2026: 47 · 2025: 44 · 2024: 5
FWCI
39.06
Type
journal article
Language
en
Open access
Locations
  • Neuroscience of Consciousness (publishedVersion, cc-by-nc)
  • PubMed Central (publishedVersion, cc-by-nc)
  • PubMed (publishedVersion)
  • UCL Discovery (University College London) (submittedVersion)
  • MPG.PuRe (Max Planck Society) (submittedVersion)
  • PsyArXiv (OSF Preprints) (submittedVersion, public-domain)
Funders
  • Engineering and Physical Sciences Research Council (EP/V000748/1)
  • Erc (Consolidator Award 'ConsciousComputation')
Topics
  • Social Sciences › Arts and Humanities › Philosophy › Philosophy and Theoretical Science
  • Social Sciences › Arts and Humanities › History and Philosophy of Science › Cybernetics and Technology in Society
  • Life Sciences › Neuroscience › Cognitive Neuroscience › Embodied and Extended Cognition
Keywords
phenomenal consciousness, large language models, folk psychology, artificial intelligence, subjective experience, public attitudes toward AI, individual differences, usage frequency

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.

The claim verdicts this report counts are experimental: their accuracy is not yet measured.

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).

Cited sources are checked with a language model’s help (gpt-6-sol).

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: Folk psychological attributions of consciousness to large language models. https://app.cochat.ai/papers/10.1093/nc/niae013
The paper
Colombatto, C., & Fleming, S. M. (2024). Folk psychological attributions of consciousness to large language models. Neuroscience of Consciousness. https://doi.org/10.1093/nc/niae013

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.