École d'été | 2 juin 2026 : Reasoning with causes and reasons par Serge Robert
Keywords
Summary
163 words
Critical Evaluation
Value of the Information & Strength of the Argument
The talk offers a valuable theoretical contribution by proposing a unified explanation for divergent results in reasoning experiments. It integrates logical analysis with psychological findings, providing a computational-level account. The argumentation is systematic, building from truth tables to experimental evidence and evolutionary considerations. However, some claims are speculative, such as the existence of a ‘behavior appraisal device’ and the evolutionary necessity of altruism detection, which are not directly supported by empirical data. The speaker acknowledges the speculative nature and invites further research.
Scientific Rigor, Source Quality, Title Accuracy
The talk references key experiments (Wason, Griggs & Cox, Cosmides, Tooby & Cosmides, Rume, etc.) and theories (mental models, probabilistic, active inference) without providing formal citations. The title accurately reflects the content. The presentation is rigorous in its logical analysis, but the lack of explicit references and the speculative extensions reduce its scientific robustness. The speaker does not provide a list of sources, but the talk is based on established literature.
168 words
Title / Content Match
The title accurately reflects the content, which contrasts reasoning with causes (abductive) and reasons (deductive) in different contexts.
Quality & Reliability
7/10
The talk is a theoretical presentation by a researcher in cognitive science, grounded in established experimental psychology of reasoning and logical analysis. It references well-known experiments and theories, but does not provide original empirical data. The argumentation is coherent and plausible, but some claims are speculative and not fully verified.
Key Moments
Markers derived by PSI from the transcript: the creator did not define chapters.
- Introduction: goal to bridge logic and experimental psychology of reasoning.
- Overview of four parts: causal cognition, conditionals, normative context, dynamics of correction, dual processes.
- Introduction of the abductive conditional and its truth table.
- Application to Wason selection task: explanation of typical behavior via abduction.
- Four types of abduction: explanation/prediction, affirmative/negative.
- Connection to active inference and long-term memory.
- Discussion of normative context: barman task and cheater detection.
- Extension to behavior appraisal device detecting altruists.
- Contrast between causal and normative reasoning: discovery vs justification.
- Dynamics of correction: alternative causes and disablers.
Cited Sources
- Wason selection task — Classic experiment in psychology of reasoning, mentioned as the causal context.
- Griggs & Cox barman task — Normative context experiment, mentioned as the normative context.
- Cosmides & Tooby cheater detection theory — Evolutionary explanation for normative reasoning.
- Rume's experiments on disablers — Experiments on conditionals with disablers, mentioned in the talk.
- Bonnefon & Maravitz on alternative causes — Research on alternative causes, mentioned in the talk.
Concurring Sources
- Cosmides, L. (1989). The logic of social exchange: Has natural selection shaped how humans reason? Studies with the Wason selection task. — Supports the cheater detection theory and normative reasoning.
- Johnson-Laird, P. N. (2006). How we reason. — Mental models theory, which the speaker positions as complementary.
- Oaksford, M., & Chater, N. (2007). Bayesian rationality: The probabilistic approach to human reasoning. — Probabilistic theory, also complementary.
Dissenting Sources
- Stenning, K., & van Lambalgen, M. (2008). Human reasoning and cognitive science. — May offer alternative formal accounts of reasoning that differ from the speaker's abductive conditional.
Contribution & Novelties
The talk provides a novel computational-level account of conditional reasoning, distinguishing between abductive (causal) and deductive (normative) uses. It extends existing theories by proposing a unified framework and introduces the concept of a ‘behavior appraisal device’ that goes beyond cheater detection. The discussion of learning through counterexamples offers a dynamic perspective on reasoning.
Pour aller plus loin :
- Abductive reasoning — Overview of abduction in philosophy and science.
- Wason selection task — Description of the classic experiment.
- Active inference — Theory by Karl Friston, relevant to the talk’s connection.
- Mental models theory — Theory by Johnson-Laird, mentioned as complementary.
- Probabilistic reasoning — Overview of probabilistic approaches to reasoning.
108 words
Radar Profile
The radar profile shows high scores in quantity of information and technical level, indicating a dense and specialized presentation. Quality and reliability are moderate, reflecting the theoretical and speculative nature. The overall balance suggests a valuable but not fully empirically grounded contribution.