hide
Free keywords:
-
Abstract:
Repeated presentation of the same stimulus results in variability in the spiking of neurons, often being correlated across neuronal pairs. However, apart from common input, a number of factors like local circuitry or dominant global states are known to induce this correlated variability. For example, slow (0.5-2.5 Hz) state transitions arising during anesthesia drive correlations in the primary visual cortex (V1). To investigate if a mechanism operating on a similar temporal regime drives correlations in the final stages of the ventral visual stream we recorded from the ventro-lateral prefrontal cortex (vlPFC) of anesthetized macaques using 10*10 Utah Arrays during periods of repeated visual stimulation (10s movie clip) and silence (10s). A Gaussian Process based factorization algorithm, previously used to unravel the underlying temporal regime of correlations in V1. returned a timescale of 14-16ms as the width of the temporal kernel, which translated to a frequency between 13-17 Hz after using an appropriate attenuation value in the spectrum of the estimated latent variable. This result indicates that apart from slow, anesthesia-induced fluctuations, intrinsic activity on a faster scale could be the source of pairwise correlations. Indeed, previous analyses of the local field potentials (LFPs) in vlPFC revealed a dominant oscillation in the beta band (13-25Hz) in both anesthetized and alert animals. Moreover, spike{spike Coherence, spike{field Coherence and other measures of spike{spike/LFP coupling suggest a significant interaction between spikes and LFP in the beta band. Despite these findings the Gaussian Process algorithm did not capture a significant amount of variance in our data, and even after accounting for the latent variable, the residual covariances remain significantly different from zero. We are currently exploring other state-space models.
