Item

Abstract

Anatomical investigations of the primate prefrontal cortex revealed fundamental structural differences compared to early sensory areas, ranging from cell morphology to patterns of intra-areal connectivity. In order to make a bridge between anatomy and function in this area it is necessary to use measures that are functionally interpretable like noise correlation. In the present study, we characterized the spatial structure of pairwise noise correlations in the ventrolateral Prefrontal Cortex (vlPFC) to investigate potential differences in vlPFC functional connectivity compared to early sensory areas. We recorded the spiking activity of spatially distributed neural populations with a Utah array in the vlPFC of two anaesthetized monkeys during visual stimulation with short duration (10 seconds) movie clips. Our findings suggest that many of the correlation properties in the vlPFC are similar to those observed in early sensory areas (e.g., relationship between noise and signal correlations). However, in contrast to early sensory areas, we found that the vlPFC connectivity kernel is neither homogeneous nor monotonic. Specifically, we observed that following an initial monotonic decrease of correlations for intermediate distances (below 2 mm) correlations for remote neurons (inter-electrode distance above 2 mm) increase significantly, and are of equal strength to the magnitude of correlations for nearby neuronal pairs. To further examine the connectivity pattern, we built a functional connectivity graph of the array (based on pairwise noise correlations), and analyzed its topology using eigenvector centrality. This analysis revealed spatially segregated subnetworks with densely connected patches of neurons. The correlation structure within the patches contributes significantly to the overall structure of correlations. Our analysis suggests that the vlPFC circuits are organized in non-homogeneous subnetworks, compatible with anatomical studies of this region [1–3]. Such a connectivity pattern could constrain theoretical models of prefrontal function, as it might be instrumental to large-scale coordination of distributed information processing in prefrontal cortex.