Comparing apples to oranges from photoreceptors to inferior temporal cortex through neural manifolds

IT population responses enable linear separation of many more natural fruit and vegetable pairs than photoreceptor or retinal ganglion cell responses, using only color distributions.

• Why it matters: Color processing transforms signals across retinal and cortical stages, but the computational benefit of these transformations for distinguishing natural objects is unclear.
• What they did: The study simulated Poisson-variable responses for photoreceptors, retinal ganglion cells, and IT populations, then tested whether linear classifiers could separate fruit–vegetable pairs from hyperspectral-derived color distributions.
• The result: Low-dimensional manifold shapes extracted from population responses explain why IT supports more pairwise linear separations, suggesting manifold-based transformations improve discriminability.

AI summary of the abstract · may contain errors · not medical advice · read the paper.