RuCCS - Dr. Ida Momennejad - In Person Talk - Tuesday, November 4, 2025

Value of the Information & Strength of the Argument

The talk provides a high-value synthesis of computational neuroscience and AI, clearly explaining complex concepts like successor representation and its neural correlates. The argumentation is solid, building from foundational studies to original experiments, and carefully distinguishes between model-free, model-based, and SR-based approaches. The speaker’s interdisciplinary background enriches the perspective, and she explicitly avoids overclaiming equivalence between brains and AI.

Scientific Rigor, Source Quality, Title Accuracy

The talk demonstrates scientific rigor by grounding claims in seminal papers (Tolman 1948, O’Keefe & Nadel, Mosers) and her own published work. She references a book chapter in ‘Space, Time and Memory’ (Oxford University Press). The title accurately reflects the content. No public comments were provided for analysis.

The title accurately reflects the content: a research talk by Dr. Ida Momennejad at RuCCS, covering her work on memory, planning, and AI.

8/10

The talk presents a coherent synthesis of established computational neuroscience concepts (Tolman, O'Keefe, Mosers) and original research, with clear methodological explanations. The speaker is a principal researcher at Microsoft Research with a strong interdisciplinary background. Claims are grounded in referenced literature, though the talk format limits detailed verification.

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

• 02:00Introduction of speaker and her interdisciplinary background.
• 10:00Overview of research question: how brains structure past experience for prediction and planning.
• 20:00Historical context: Tolman's 1948 experiment and cognitive maps.
• 30:00Introduction to successor representation and its mathematical formulation.
• 40:00Comparison of model-free, model-based, and successor representation agents.
• 50:00Experiments on reward changes and transition changes, showing SR predicts human behavior.
• 1:00:00Neural correlates: SR and hippocampal place fields, grid cells.
• 1:10:00Implications for AI and collective intelligence.
• 1:20:00Conclusion and Q&A.

• Tolman, E.C. (1948). Cognitive maps in rats and men. — Cited as foundational work on cognitive maps.
• O'Keefe, J., & Nadel, L. (1978). The Hippocampus as a Cognitive Map. — Cited for place cells and cognitive maps.
• Moser, E.I., et al. (2014). Grid cells and the brain's GPS. — Cited for grid cells in entorhinal cortex.
• Momennejad, I., et al. (2017). The successor representation in human reinforcement learning. — Cited as the speaker's own work on successor representation.
• Space, Time and Memory (Oxford University Press) — Book with a chapter by the speaker summarizing her research.

• Momennejad, I., et al. (2017). The successor representation in human reinforcement learning. — Directly supports the talk's claims about SR in human behavior.
• Russek, E.M., et al. (2017). Predictive representations can link model-based reinforcement learning to model-free mechanisms. — Supports the idea of SR as a bridge between model-free and model-based.