Contributed Talk Sessions | Poster Sessions | All Posters | Search Papers

Poster Session A: Tuesday, August 12, 1:30 – 4:30 pm, de Brug & E‑Hall

How neurocomputational mechanisms of number perception adapt to prior expectations

Gilles de Hollander¹, Arthur Prat-Carrabin², Saurabh Bedi¹, Samuel J. Gershman³, Christian C. Ruff¹; ¹University of Zurich, ²Harvard University, Harvard University, ³Harvard University

Presenter: Gilles de Hollander

Efficient coding offers a normative theory for how the brain should allocate resources to represent the world, and growing evidence demonstrates that perceptual systems of humans and animals adhere to its principles. However, most existing studies have focused on simple stimuli like Gabor patches and have assumed relatively fixed encoding functions. Here we demonstrate that cognitive and neural representations of numerosity - a higher-level cognitive function - can rapidly adapt to context in ways consistent with efficient coding. Using fMRI (n=39), we show that neural populations tuned to specific numerosities shift their tuning with context, aligning with with a Thurstonian perceptual model in which part of an unconstrained objective stimulus space is linearly mapped to a constrained representational space. Our findings demonstrate how the brain adapts to changing conditions and how neurocomputational modeling of fMRI data can deepen our understanding of the neural representations driving behavior.

Topic Area: Memory, Spatial Cognition & Skill Learning

Extended Abstract: Full Text PDF