A method for analysing neural computation using receptive fields in state space.
The behaviour of spiking neurons which are involved in a control task can be quantified by mapping receptive fields in the state space of the control problem. These receptive fields link spikes, the operands of neural computation, to state variables, the operands of conventional control theory. They allow neural computation underlying control tasks to be quantitatively analysed, and meaningfully discussed in ordinary language, by providing a rigorous way to interpret single spikes as assertions about dynamical state.
Journal Article↗