TLDR: We link the distribution of tuning properties to representational geometry and readout for perceptual tasks.
Different types of tuning diversity reshape population codes in distinct — and beneficial — ways.

Tuning diversity is a mechanism for enhancing population representational efficiency while respecting these constraints.

Of course, the geometry could change in this way for other reasons — such as increasing the number of neurons or their firing rates. But metabolic constraints prevent arbitrarily high amplitudes, bandwidths, and numbers of neurons.

BUT, amplitude diversity helps discrimination more, while bandwidth diversity helps identification more. Because of their distinct impacts on geometry, the two types of diversity affect different perceptual tasks more.

These geometric changes increase the population’s coding efficiency for two types of perceptual tasks: Discrimination and identification

Intuitively:
– Amplitude diversity helps the population use more of its firing rate range
– Bandwidth diversity lets it exploit more dimensions in firing rate space

We focus on two types of diversity: amplitude (height) and bandwidth (width) of tuning curves. Each shapes geometry differently.
Amp Div: expands distances between the centers of representations for different stimuli
BW Div: decorrelates the centers

TLDR: We link the distribution of tuning properties to representational geometry and readout for perceptual tasks.
Different types of tuning diversity reshape population codes in distinct — and beneficial — ways.

Tuning diversity is a mechanism for enhancing population representational efficiency while respecting these constraints.

Of course, the geometry could change in this way for other reasons — such as increasing the number of neurons or their firing rates. But metabolic constraints prevent arbitrarily high amplitudes, bandwidths, and numbers of neurons.

BUT, amplitude diversity helps discrimination more, while bandwidth diversity helps identification more. Because of their distinct impacts on geometry, the two types of diversity affect different perceptual tasks more.

These geometric changes increase the population’s coding efficiency for two types of perceptual tasks: Discrimination and identification

In many brain areas, neuronal tuning is heterogeneous. But how does this diversity help behavior? We show how tuning diversity shapes representational geometry and boosts coding efficiency for perception in our new preprint: www.biorxiv.org/content/10.1...
(w/ @sueyeonchung.bsky.social&Tony Movshon)

Variations in neuronal selectivity create efficient representational geometries for perception https://www.biorxiv.org/content/10.1101/2025.06.26.661754v1

Already feeling #cosyne2025 withdrawal? Apply to the Flatiron Institute Junior Theoretical Neuroscience Workshop! Applications due April 14th

jtnworkshop2025.flatironinstitute.org