The neuronal implementation of representational geometry in primate prefrontal cortex

Abstract

Modern neuroscience has seen the rise of a population-doctrine that represents cognitive variables using geometrical structures in activity space. Representational geometry does not, however, account for how individual neurons implement these representations. Leveraging the principle of sparse coding, we present a framework to dissect representational geometry into biologically interpretable components that retain links to single neurons. Applied to extracellular recordings from the primate prefrontal cortex in a working memory task with interference, the identified components revealed disentangled and sequential memory representations including the recovery of memory content after distraction, signals hidden to conventional analyses. Each component was contributed by small subpopulations of neurons with distinct spiking properties and response dynamics. Modeling showed that such sparse implementations are supported by recurrently connected circuits as in prefrontal cortex. The perspective of neuronal implementation links representational geometries to their cellular constituents, providing mechanistic insights into how neural systems encode and process information.

Sequential neuronal processing of number values, abstract decision, and action in the primate prefrontal cortex

“…We found neurons selective to numbers and decision in various combinations, as well as neurons exclusively selective to number, number order (only the first number or second number), or decision. The observation of mixed selective neurons together with more traditionally, selective single neurons in the dlPFC confirms previous findings [16,18,19,[21][22][23][24][25]. While strongly selective single neurons could be said to form important network nodes engaging in complex computations or processing, mixed selective neurons are thought to expand the range of possible computations [26,27].…”

Privileged representational axes in biological and artificial neural networks

“…We found that the axes of both biological networks and DCNNs facilitate sparser readouts of category information compared to arbitrary bases. This finding may relate to a recent study of neural responses in primate prefrontal cortex during a working memory task, revealing how neural implementations of representational geometries influence wiring costs [86]. Third, axes affect the capacity for generalization.…”