Mechanistic Basis of Reasoning (in Brains and AI)

This workshop will focus on the mechanistic side of reasoning, including both neuroscience and mechanistic interpretation research in AI. The goal will be to explore different mechanisms by which complex cognition and reasoning capabilities are implemented, and discuss the advantages and limitations of biological and artificial neural networks in this role. This workshop is a natural continuation of the previous event and has the objective to ground discussions in implementation strategies.

Stephanie C.Y. Chan (Google DeepMind), Ila Fiete (MIT McGovern Institute), Uri Hasson (Princeton University), Alexia Jolicoeur Martineau, David Klindt (Cold Spring Harbor Lab), Benno Krojer (McGill University, Mila), Sarthak Mittal (Université de Montréal, Mila), Ida Momennejad (Microsoft Research), Siva Reddy (McGill University, Mila), Irina Rish (Université de Montréal, Mila), Naomi Saphra (Harvard University), Nino Scherrer (Google), Dhanya Sridhar (Université de Montréal, Mila), Claire Stevenson (University of Amsterdam), Hidenori Tanaka (Harvard University), Siddarth Venkatraman (Université de Montréal, Mila).

TUESDAY, FEBRUARY 24th, 2026

9:30 – 9.50 a.m.: Welcome and Registration

9:50 – 10 a.m.: Welcome Address

10 – 10:45 a.m.: The Effects of Data Pruning, Representation Capacity, and Latent Trajectories
Irina Rish (Université de Montréal, Mila)

10:45 – 11:30 a.m.: The Next Frontier: Generative and Decodable Embeddings from Frozen LLMs
Siva Reddy (McGill University, Mila)

11:30 a.m. – 12 p.m.: Recap. Discussion Audience/ G. Speakers

12 – 1.45 p.m.: Lunch on your own

1:45 – 2.30 p.m.: Fluid Reasoning Development: Comparing Mechanisms in Minds and Machines
Claire Stevenson (University of Amsterdam)

2.30 – 3.15 p.m.: Deep Test-Time Thinking with Recursive Self-Aggregation
Siddarth Venkatraman (Université de Montréal, Mila) & Sarthak Mittal (Université de Montréal, Mila)

3.15 – 3.45 p.m. : Recap. Discussion Audience/ G. Speakers

WEDNESDAY, FEBRUARY 25th, 2026

9:30 – 10 a.m.: Welcome and Coffee

10 – 10:45 a.m.: Causal Representation Learning: A Natural Fit for Mechanistic Interpretability
Dhanya Sridhar (McGill University, Mila)

10:45 – 11:30 a.m.: Interpreting the Relation Between Visual and Linguistic Representations in a VLM
Benno Krojer (McGill University, Mila)

11:30 a.m. – 12 p.m.: Recap. Discussion Audience/ G. Speakers

12 – 1.45 p.m.: Lunch on your own + Move to Mila

Afternoon at Mila

1:45 – 2.30 p.m.: Identifying the Atoms of Meaning in Biological and Artificial Visual Processing
David Klindt (Cold Spring Harbor Lab)

2.30 – 3.15 p.m.: The Compositional Geometry of Open-Ended Reasoning
Ida Momennejad (Microsoft Research) – Remote

3.15 – 3.45 p.m. : Coffee Break

3.45 – 5 p.m.: Expert Panel: “Mechanisms of Reasoning in Brains and AI – Are We on the Right Track?” with Benno Krojer (McGill University, Mila), David Klindt (Cold Spring Harbor Lab) and Nino Scherrer (Google) + Discussion Audience/ G. Speakers

5 – 6 p.m.: Social Hour/Mixer

THURSDAY, FEBRUARY 26th, 2026

9:30 – 10 a.m.: Welcome and Coffee

10 – 10:45 a.m.: Associative and Episodic Memory from Pre-Structured Spatial Scaffolds in the Hippocampus
Ila Fiete (MIT McGovern Institute) – Remote

10:45 – 11:30 a.m.: Deep Language as a Cognitive Model for the Development and Processing of Natural Language
Uri Hasson (Princeton University) – Remote

11:30 a.m. – 12 p.m.:  Recap. Discussion Audience/ G. Speakers

12 – 1.45 p.m.: Lunch on your own

1:45 – 2.30 p.m.: Connecting Perspectives on Dynamical Phenomena in Transformers
Stephanie C.Y. Chan (Google DeepMind) – Remote

2.30 – 3.15 p.m.: Uncovering Mesa-Optimization Algorithms in Transformers & Building a Layer for Optimal Test-Time-Training
Nino Scherrer (Google)

3.15 – 3.45 p.m. : Coffee Break

3.45 – 4.30 p.m.: Title TBA
Alexia Jolicoeur Martineau

4.30 – 5 p.m.: Recap. Discussion Audience/ G. Speakers

FRIDAY, FEBRUARY 27th, 2026

9:30 – 10 a.m.: Welcome and Coffee

10 – 10:45 a.m.: When Can We Predict Model Behavior?
Naomi Saphra (Harvard University)

10:45 – 11:30 a.m.: Mechanistic Principles of Concept Learning and Compositional Reasoning
Hidenori Tanaka (Harvard University)

11:30 a.m. – 12 p.m.:  Recap. Discussion Audience/ G. Speakers

12 – 12.05 p.m.: Closing Remarks