Measurement of propagating waves from local field potentials and unit activity in the cortex of human and monkey

Abstract

The existence of propagating waves, either spontaneous or stimulus-evoked, in neocortex during the awake state has been a subject of recent interest [1,2]. Here, following work done previously in voltage-sensitive dye imaging of the primary visual cortex in the awake monkey [3], we apply an analysis method for non-parametric, automated detection of propagating waves in noisy multichannel data to multielectrode array (MEA) recordings taken from the neocortex of the human (middle temporal gyrus, as studied in [4]) and monkey (primary visual cortex). In the spontaneous awake state, we detect transient, intermittent propagations occurring in the local field potential (LFP) across a wide frequency range. After selecting the propagation epochs within the highest confidence interval, we observe that the speed distribution falls naturally into the range of the horizontal fibers (0.1 - 0.5 m/s), suggesting a similar propagation substrate as that observed previously in voltage-sensitive dye imaging data [3]. With this in mind, we go on to compare the spatiotemporal dynamics on the array across various states of arousal. One question raised by our earlier VSD analysis is the extent to which the wave evoked by a small visual stimulus is supra- or sub-threshold. Specifically, because neurons during awake, activated cortical states sit a few millivolts below threshold [5,6], operating in a fluctuation-driven regime [7], the transient depolarization detected in the VSD signal as the wave passes may change the background spiking probability in the local circuit. To address this, we analyzed the relationship between single- and multi-unit activity (SUA/MUA) and LFP. We study the spatiotemporal dynamics of this relationship to determine the coupling of spiking activity with transient propagating waves, and compare the results with computational models of spiking neurons. In conclusion, we show here that a combination of multi-electrode recordings, time-series analysis, and computational modeling provides a powerful set of tools for quantifying spatiotemporal dynamics in awake states.