Figure 3.

A simple feed‐forward model of replay fails to explain open field replay. (a) Top‐down view of a large open arena. Each dot represents a neural population with similar spatial representation. The place fields of two neuron populations are shown in red and blue. In this simple model, repeated traversal of the overlapping sections of these place fields serves to co‐activate these neuron populations sufficiently to strengthen the synaptic connections between them (represented with the bi‐directional arrow). (b) Left: Extended exploration of the environment would therefore serve to strengthen synaptic connections between populations of neurons representing adjacent locations. Right: This pattern of synaptic connectivity would be expected to exist across the entire cognitive map (assuming relatively equal exploration of all regions of the environment). (c) Using only simple feed‐forward network dynamics, the above model predicts that open field replay would encode a circular “wave” of representation that would start at the animal's current location (due to current sensory input driving those cells) and sweep in all directions simultaneously, following the established synaptic paths. Black dots indicate currently active neurons, arrows indicate currently active synapses. Critically, this is not what is observed in vivo (see Figure 1b), suggesting that more complicated network dynamics exist which regulate neuronal participation, likely involving inhibitory circuitry to force the representation to a single spatial location at a time