Spike count correlations in visual, visuomotor, and motor neurons of macaque prefrontal area 8A during working memory maintenance

Abstract

We examined spike count correlations between neurons in prefrontal area 8A of two macaca fascicularis during a spatial working memory task by recording the responses of multiple units using microelectrode arrays (Blackrock Inc., UT). The task consisted of fixation on a central spot for 494-800ms, then presentation of a circular sine wave grating at one of 16 randomly selected locations for 507ms. After extinguishing the grating, there was a delay period of 494-2000ms, followed by extinguishing of the central fixation point, which instructed the animals to make a saccade to the remembered stimulus location. We recorded the activity of neurons in blocks of 32 channels and sorted spikes using Plexon software (Plexon Inc, TX). We isolated responses of 191 single units for a total of 1170 neuronal pairs. Neurons were classified as being selective (one-way ANOVA, p<.05) for visual stimuli (visual, n = 29, or 15%), saccades (motor, n = 22, or 12%), or both (visuomotor, n = 78, or 41%). We then quantified the proportion of significant positive and negative spike count correlations as well as the proportion predicted by chance (shuffled correlations) in each group. In visual units the proportion of significant positive and negative correlations during the memory period were both significantly greater than expected by chance (positive = .22 and negative = .13, p<0.05). In motor and visuomotor units, only the proportion of significant positive correlations was significantly greater than chance (motor = .18, visuomotor = .17, p<0.05). Our results show that during working memory maintenance visual, motor, and visuomotor units in area 8A exhibit different degrees of correlated firing. This suggests that these different cell types may play different roles in the network dynamics underlying working memory maintenance by prefrontal neurons. Meeting abstract presented at VSS 2012