Soutenance de thèse Inès Leproult - 01 décembre 2025

Value of the Information & Strength of the Argument

The value of the information is high, as it presents original experimental data and computational modeling that addresses a paradox in the literature on semantic strategies in working memory. The argumentation is solid, systematically testing hypotheses and using computational modeling to validate theoretical claims. The progression from behavioral experiments to model implementation is logical and strengthens the conclusions. However, the presentation is a thesis defense, so some details on statistical methods and potential limitations are not fully elaborated.

Scientific Rigor, Source Quality, Title Accuracy

The scientific rigor is evident in the careful experimental design and the use of computational modeling to test theoretical predictions. The sources cited are relevant and include key studies in the field, though specific references are not explicitly listed in the video. The title accurately reflects the content, and the presentation adheres to academic standards. The thesis defense format includes a jury, which adds credibility, but the lack of explicit citations in the video limits the ability to verify all sources.

The title accurately reflects the content: a thesis defense on the role of semantic representations in working memory maintenance.

8/10

The thesis defense presents original research with a clear methodology, computational modeling, and peer-reviewed context. The presentation is rigorous, but as a defense, it may lack full detail on statistical analyses and limitations.

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

• 00:00Introduction by jury president and presentation of the thesis topic.
• 02:00Definition of working memory and examples of everyday situations.
• 05:00Explanation of maintenance strategies: rehearsal, refreshing, and semantic strategies.
• 10:00Presentation of the paradox in the literature: semantic strategies beneficial but not when instructed.
• 15:00Description of experiments 1 and 4: manipulation of free time distribution and results.
• 20:00Experiments 3 and 5: testing semantic vs. phonological strategies with pseudowords and abstract words.
• 25:00Introduction of the computational model TBRS* and its failure to reproduce results.
• 30:00Development of TBRS*S+ with semantic boost mechanism and successful simulation.
• 35:00Aging studies: investigating the role of semantic strategies in compensating WM decline.
• 40:00Conclusion and discussion of theoretical implications.

• TBRS model (Time-Based Resource Sharing) — Mentioned as the basis for the computational model used in the thesis.
• Barrouillet, P., & Camos, V. (2015). Working memory: Loss and reconstruction. — Likely cited in the thesis as a foundational model of working memory.

• Camos, V., & Barrouillet, P. (2018). Working memory in development. — Supports the TBRS model and its application to development and aging.

• Studies showing no benefit of semantic strategies in WM (e.g., Bailey & Dunlosky, 2017) — The thesis addresses contradictory findings in the literature, proposing a resolution via the direct semantic maintenance hypothesis.