Figure 1.

Illustration of the representation-communication framework for neuroscientific questions. (A) A pattern of activity within an area or population of neurons is captured as firing rates over a specified time window. In this rendering, each circle represents a neuron, and the color represents that neuron’s instantaneous activity, which continually changes over time. We measure representations not in an instant, but typically in a rate code, by integrating spiking activity over a time window ranging from tens to hundreds of milliseconds. (B) The rates of n neurons in Area X are collectively a multidimensional “representation” that varies over time. This representation may be as concrete as the joint firing rates across the population, or may be abstracted through dimensionality reduction. (C) An area X may be inferred to communicate with area Y if the representation within area X modulates the representation in area Y in a systematic way.