2026 GAC 3

Representations or Transformations: What Should Models of Neural Population Activity Explain?

Julian Kędys, Poznan Supercomputing and Networking Center

Cezary Mazurek, Poznan Supercomputing and Networking Center

SueYeon Chung, Harvard University

Sarah Harvey, Flatiron Institute

Gustavo Deco, Universitat Pompeu Fabra

Il Memming Park, Champalimaud Centre for the Unknown

Maneesh Sahani, University College London

Thomas Anderson Keller, Harvard University

Short description

Modern computational neuroscience analyses neural population activity through two broad lenses. The representation-centric program asks how stimuli are encoded in the geometry of population responses — their dimensionality, separability, and organisation — and treats this geometry as the primary object of explanation. The transformation-centric program asks what functional map the system computes — the canonical operations it performs, whether as dynamical motifs and attractor structures unfolding over time or as hierarchical feedforward transformations over processing depth — and treats this map as primary instead. The two can be applied to the same recordings, yet they make different commitments about what counts as a sufficient explanation of neural computation, and recent results show these commitments can yield contradictory verdicts on identical data — including when comparing artificial models to brains. This GAC brings together leading proponents of both views, plus a bridging perspective grounded in formal structural constraints, to identify concrete pre-registered predictions on which the frameworks diverge, to adjudicate them on a shared benchmark, and to determine for which neural systems and cognitive functions each account suffices — and where they can be integrated.

GAC kickoff workshop, August 4th at CCN

Schedule of events

TBA

Team

Comment on this GAC proposal!

Instructions for CCN community members: 

1. Download the proposal PDF using the button above. 

2. Use the Disqus box below to leave comments and reviews, and discuss the submission. You'll need to create a Disqus account or log into Disqus using a google account or similar in order to leave a comment/review.