Figure 6.

Relationship between the receptive field and spike train crosscovariance of neighboring and distant neurons. The STRFs of four example neuron pairs are shown (A–D, columns 1 and 2). The examples were chosen from the experimental (A, ST; B, AT) and control groups (C, DS; D, DSBF). For each pair, we computed the spike train crosscovariance function (SCC, normalized between -1 and 1, see Materials and Methods) to determine the strength and timing of the temporal correlations between the spike trains of neighboring neurons (column 4, black). We used the STRFs for each pair to predict the SCC. For each pair of neurons, this is done by performing a two-dimensional cross-correlation between the STRF of each neuron (column 3) and subsequently selecting the temporal cross-section about zero frequency shift (red horizontal section, in column 3; see Materials and Methods). For instance, the example neuron pairs of A and B have STRFs with similar structure and thus the RFCC shows a strong positive peak about zero frequency shift. The predicted spike train covariance mirrors the SCC for each pair (column 4, red). In C, the neurons are highly overlapped in frequency, however, STRFs fail to predict the lack of correlation observed between the neural spike trains. The pair shown in D has receptive fields that are temporally delayed (neuron 1 leads neuron 2) and which differ in frequency by 1 octave. Because of the lack of frequency overlap the predicted (red) and measured SCC (black) are both zero.