Figure 8.

PSD of neuron spike trains and spike-LFP coherence as a function of velocity. A, Individual power spectra for spike trains at low velocity (5–15 cm/s) and high running speed (35+ cm/s; color axes are to the same scale between plot) for interneurons and pyramidal cells. A spike threshold >0.25 Hz was applied to each running bin to ensure that only neurons with a considerable number of spikes were analyzed. The rows are sorted by the maximum frequency between 4 and 12 Hz. B, The average PSDs at low (blue) and high velocities (red); error bars are the SEM (different axes are used to allow the spectral shape to be compared). Note that there is an absence of a spike-frequency peak in the slow γ range although both the interneurons and pyramidal cells exhibit harmonic modulation. Moreover, at high velocity, there is a potential peak in the interneurons that coincides with the traditional γ range of 40–100 Hz (Bragin et al., 1995). Note that the axes are in arbitrary units as power in this circumstance is determined by the number of spikes and thus, each trace was normalized by total power. C, Individual spike-LFP coherence plots for interneurons and pyramidal cells at different running speeds. Once again, the color axis is equal across plots. D, Average spike-LFP coherence by cell type as a function of velocity. While there is a notable difference in overall coherence as a function of velocity, the only peak are at θ, the 14- to 18-Hz harmonic and the 21- to 27-Hz harmonic (prominent at high velocity in the interneurons). Broad modulation in the 25- to 50-Hz band (slow γ) is absent. We emphasize strong caution in interpreting small coherence values (for a similar opinion, also see Buzsáki and Schomburg, 2015). A coherence of 1 would indicate no variance in the phase difference across signals, while 0 would be indicative of a random phase assignment. As the majority of coherence values related to frequencies greater than the 16-Hz θ harmonic fall below 0.05, concluding that units are capable of organizing, or being organized into higher frequencies based on this data are dangerous.