Effects of rest on V4 network coding in perceptual learning

EAGLEMAN, S.; MULAS, M.; FERNANDEZ LEON, J.A.; DRAGOI, V.

California

Conferencia; 2013 SfN Neuroscience; 2013

SfN

It is widely documented that behavioral performance in perceptual tasks is improved following a brief rest. However, the changes in neural network dynamics and coding underlying the behavioral improvement are unclear. In addition, the changes induced by rest in individual neuron and network responses are unknown. We devised a study that had two goals: 1. to determine whether the behavioral performance improvement observed following a period of rest is correlated with an improvement in neuronal discrimination and coding of stimuli, and 2. to employ a battery of analyses during the rest period to identify patterns of neuronal activity during rest (e.g., the presence of replay events and/or pronounced low frequency activity in local field potential responses) that correlate with the improvement in behavioral performance. To accomplish these goals we trained two monkeys to perform a visual same/different task in which successive natural images were judged to have been presented at the same or slightly different orientation. While animals performed the task, we simultaneously recorded the responses of multiple neurons in visual cortex area V4 using custom-made electrode arrays and laminar probes. Each session consisted of a block of trials in which monkeys performed the task followed by a 20-min rest in a dark, quiet room. During this time monkeys were observed to have their eyes closed and jaw-slack (indicative of sleep). Monkeys then performed a second block of trials after rest. We assessed behavioral performance by determining the percentage of correct trials at the monkeys? discrimination thresholds. Following rest, we found a significant increase in the percentage of correct trials. These improvements were correlated with the amount of time animals had their eyes closed during the nap period (Pearson Correlation, r =0.0365, P < 0.05). Subsequently, we found improvements in individual neural coding assessed using d? as well as a decrease in the trial by trial correlated variability in firing rate (noise correlation, 20%, Wilcoxon signed rank, P < 0.001) which has been previously associated with impoverished stimulus coding. Preliminary results indicate correlative relationships between activity observed during rest and subsequent neural and behavioral improvement. Collectively, our results provide evidence that neural populations exhibit changes in their coding properties that improve network performance following a brief nap.