Contributed Talks: “Metacontrol & flexibility in inference and decision-making” - CCN 2025

Value of the Information & Strength of the Argument

The talks provide valuable insights into cognitive mechanisms, each presenting original research with clear hypotheses and computational models. The first talk offers a novel unifying account of probability distortions, supported by pre-registered experiments and simulations. The second talk provides a detailed analysis of fatigue dynamics and its impact on confidence, with a computational model that fits the data well. The fourth talk demonstrates the power of recurrent neural networks in capturing flexible behavior across tasks, and the reduced-complexity model offers interpretable insights. The argumentation is solid, with each speaker explaining their methods and results clearly, and they address questions from the audience. However, the talks are brief, and some details are omitted, but overall the scientific value is high.

Scientific Rigor, Source Quality, Title Accuracy

The scientific rigor is high, as the talks present pre-registered experiments, computational modeling, and clear statistical analyses. The sources cited are primarily the speakers’ own work and relevant literature, but specific references are not explicitly mentioned in the transcription. The title accurately reflects the content, and the video is a professional recording of a conference session. The quality of the video is adequate, with some technical issues, but the content is clear. The adequacy between title and content is excellent, as it directly describes the session.

The title accurately describes the content: a session of contributed talks on metacontrol and flexibility in inference and decision-making.

8/10

The video presents four original research talks from a scientific conference, with clear methodology, pre-registered experiments, and computational modeling. The content is peer-reviewed in the sense of conference presentation, and the speakers are from reputable institutions. However, the video is a recording of a live session with limited production quality, and the talks are concise, lacking full methodological details.

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

• 00:00Introduction and first talk by Saurabh Bedi on probability distortions.
• 10:00Discussion of boundary repulsions and pre-registered experiments.
• 20:00Second talk by Katarzyna Dudzikowska on fatigue and confidence.
• 30:00Third talk by Peiyue Liu on proactive counterfactual inference (brief).
• 40:00Fourth talk by Jessica Passlack on general strategy for flexible behavior.

• Prospect Theory: An Analysis of Decision under Risk — Mentioned in the first talk as the basis for probability distortions.
• Efficient coding and Bayesian decoding in neural systems — Referenced in the first talk as theoretical foundations.

• Prospect Theory: An Analysis of Decision under Risk — Supports the existence of probability distortions.
• A tutorial on Bayesian models of perception — Relevant to Bayesian decoding in the first talk.

• Alternative accounts of probability distortions — Some existing models assume specific encoding/decoding distortions, which the first talk challenges.