Population Coding

Population coding is the view that groups of neurons, not only individual neurons, jointly encode perception, location, action, or internal state. In Claire Isabel Webb & Nina Miolane: The Geometry of Consciousness, Nina Miolane presents it as a response to the limits of the single-neuron doctrine: individual neurons can be informative, but brains contain many neurons that often participate in multiple codes.

Key Claims

• Single-neuron firing rates can encode continuous variables, but isolated neurons rarely explain the full representation.
• Treating a population’s firing rates as coordinates creates the setting for Neural Geometry.
• Population coding can turn large neural recordings into trajectories and shapes that can be compared across animals and artificial networks.
• The Spatial Navigation Torus depends on periodic activity across many neurons rather than on one neuron standing for one location.

Connections

• Nina Miolane - source speaker using population codes as the bridge from recordings to geometry.
• Neural Geometry - geometric form population codes can take.
• Mathematical Theory Of Intelligence - larger project that population coding supports.
• Spatial Navigation Torus and Fourier Spatial Encoding - concrete spatial-navigation case.
• Representation Learning - AI-adjacent concept for learned internal codes.