Confiance, métacognition et le « problème difficile » de la conscience

Confiance, métacognition et le « problème difficile » de la conscience

by Megan Peters

Life & Natural Systems Neuroscience PSANNeurosciencesPSAN5Cognitive and behavioural neuroscience
🎙 Megan Peters 👥 305 📅 September 11, 2025 ⏱ 77 min 👁 87 📄 expert opinion 🧭 2026-08-16
Available in: English (current) Français

Keywords

metacognitionconfidenceconsciousnesshard problempsychophysicsBayesian modelsintrospectionvisual perceptionuncertaintyneuroimaging

Summary

In this seminar, Megan Peters, an associate professor at UC Irvine, explores whether computational models of metacognitive confidence and uncertainty can shed light on the hard problem of consciousness. She begins by illustrating the dissociation between objective visual information and subjective experience using a real-world example. She then outlines her research approach, which combines psychophysics, computational modeling, and neuroimaging to study introspective psychometric functions. Peters presents a series of experiments showing that introspection is not constant across conditions: visual masking did not induce blindsight-like behavior, TMS over early visual cortex paradoxically increased confidence, and higher dot density or invalid attention cues increased confidence without improving performance. These findings suggest that subjective confidence is influenced by the amount of evidence supporting a decision, rather than being a direct readout of objective performance. She also discusses how the observer’s policy for reporting confidence affects the relationship. Peters then moves to the neural basis, mentioning an ECoG study that decoded evidence for decisions and confidence. Finally, she touches on theoretical implications, suggesting that understanding the computations underlying metacognition might offer a way to reframe or potentially sidestep the hard problem. The talk concludes with a discussion of future directions and the potential for artificial systems to have subjective experiences.

206 words

Critical Evaluation

Value of the Information & Strength of the Argument

The talk provides valuable insights into the empirical study of metacognition and its relevance to consciousness. Peters presents a coherent argument that introspection is not a monolithic phenomenon but varies systematically with stimulus and task conditions. The use of introspective psychometric functions is a novel methodological contribution that allows for the isolation of subjective experience from objective performance. The argumentation is solid, grounded in multiple experiments and computational models, though some conclusions are speculative. The speaker acknowledges the limitations and open questions, which enhances the credibility of the presentation.

Scientific Rigor, Source Quality, Title Accuracy

The talk demonstrates scientific rigor through the use of established psychophysical methods and computational modeling. Peters cites several of her own peer-reviewed publications, including a 2015 eLife paper and a 2025 Cerebral Cortex paper, which are relevant to the content. The title accurately reflects the content, focusing on confidence, metacognition, and the hard problem. The presentation is well-structured and the claims are supported by data. However, as a seminar, it does not provide the full methodological details, and some references are mentioned but not fully elaborated.

190 words

Title / Content Match

The title accurately reflects the content, which explores the relationship between confidence, metacognition, and the hard problem of consciousness.

Quality & Reliability

8/10

The talk is given by a recognized expert in cognitive neuroscience, presenting a coherent overview of her research program with references to peer-reviewed publications. The claims are grounded in empirical studies and computational modeling, though the presentation is a seminar rather than a peer-reviewed article, and some speculative elements are acknowledged.

Key Moments

Markers derived by PSI from the transcript: the creator did not define chapters.

Cited Sources

Concurring Sources

Dissenting Sources

Contribution & Novelties

The talk contributes to the field by systematically demonstrating that introspection is not a fixed function of objective performance, but varies with stimulus properties, attention, and decision policies. It introduces the concept of introspective psychometric functions as a tool to isolate subjective experience. The findings challenge previous assumptions about visual masking and blindsight, and suggest that confidence may be biased by evidence supporting the decision. The talk also bridges empirical work with philosophical questions about consciousness, proposing that understanding metacognitive computations might offer a new angle on the hard problem.

Pour aller plus loin :

  • Higher-order theories of consciousness — Stanford Encyclopedia of Philosophy entry on higher-order theories, relevant to the discussion of metacognition and consciousness.
  • Signal detection theory — Wikipedia article on SDT, foundational to the psychophysical methods used.
  • Blindsight — Wikipedia article on blindsight, providing background on the condition discussed.

142 words

Radar Profile

The radar profile shows high scores in information quantity, quality, and technical level, indicating a dense and well-supported presentation. The fiabilite_globale is also high, reflecting the speaker's expertise and the use of peer-reviewed references. The overall profile suggests a scientifically rigorous talk that is accessible to a specialized audience.

Reliability 8/10

💬 No comments were provided for analysis.