Séminaire - DIC-ISC-CRIA - 24 septembre 2026 par Reto Gubelmann

Value of the Information & Strength of the Argument

The talk provides a valuable perspective by reframing reasoning as a holistic action involving not just cognition but also affect and conation. Gubelmann builds a coherent argument using classical sources (Aristotle, Anselm) and contemporary psychology (system 1/2), illustrating with concrete examples like the HR evaluation scenario and the courage confrontation. The argument is logically structured, and the distinction between cognitive, conative, and affective failures is insightful. However, as a work-in-progress, the implications for AI reasoning are only briefly sketched, and the empirical basis for some claims (e.g., emotions affecting reasoning) is assumed rather than detailed.

Scientific Rigor, Source Quality, Title Accuracy

Gubelmann references his own papers (Gubelmann & Hongler 2026; Gubelmann 2026) which are relevant to the topic. He also draws on well-known philosophical sources without extensive citations, but the narrative is academically informed. The title ‘Grounding Reasoning’ accurately reflects the talk’s focus on making explicit the embodied and holistic nature of human reasoning. The video title ‘Séminaire - …’ correctly indicates an academic seminar format, and the content matches the announced topic. No public comments were analyzed.

The video title indicates a seminar talk and is accurate; the talk indeed presents a philosophical exploration of grounding reasoning.

7/10

The talk by an expert in philosophy and NLP presents original ideas grounded in classical philosophy and personal research. Work in progress but well-argued.

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

• 00:00Introduction by host, bilingual welcome.
• 05:00Discussion of grounding concept using School of Athens, contrast Plato and Aristotle.
• 10:00Anselm's ontological argument and mix of emotion and logic.
• 15:00System 1 vs System 2 thinking, requirement of willpower.
• 20:00Trilogy of mind: cognition, affect, conation, and courage example.
• 30:00Practical wisdom and implications for reasoning with LLMs.

• Too Fast, Too Shallow–LLMs, Including Reasoning LLMs, Are Unreliable Constitutional Reasoners — Referenced as a paper on legal AI, mentioned in the talk and description.
• Putting reasons back into reasoning: how genuine reasoning is inference-based and why neuro-symbolic NLI could achieve it — Referenced as a paper connecting philosophy and NLP, mentioned in talk and description.