Figure 3.

A model of cortical state. a, We model the activity state at any instant by (v, w) (asterisk) and model the dynamic state as a vector field governing the dynamics of the activity state [gray arrows represent flow lines for various values of (v, w)]. b, The FHN equations provide a parametric family of vector fields that are easily and robustly fit from short data segments. c, A possible physiological interpretation of different terms in the model; v represents the firing rate of both pyramidal cells and interneurons, whereas w represents the combined effect of multiple adaptive phenomena. d, Sample v (red) and w (green) traces obtained directly from the data. Three seconds of spontaneous activity preceding the stimulus (time 0) are used to fit the model parameters. The resulting phase diagram (far right) can be used as a means of characterizing the dynamic state.