Figure 4: Spontaneous activity patterns recapitulate evoked network responses.
From: Spontaneous emergence of fast attractor dynamics in a model of developing primary visual cortex

(a) All spontaneous spikes over 500 ms, showing the sparseness of spontaneous activity. (b) Spontaneous spiking activity over 100 s is collected into 50 ms time bins, and the ∼30% most active time bins are selected (almost all time bins have either 0 or 1 spike for each neuron). (c) Same data as in b, with columns (time bins) sorted by similarity (same procedure as for Fig. 2b). Spiking activity is dominated by discrete patterns, which are the same multi-cell patterns evident in stimulus-evoked network responses, in a different order (Fig. 2b–see also e). (d) Correlation matrix for the columns in c. Clustering of response patterns is as evident as in Fig. 3c. (e) Cross-correlation matrix between the columns of c and of Fig. 2b. The high correlation between patterns in c and Fig. 2b reveals that the firing patterns in spontaneous and evoked activity are highly similar.