Synchronous rhythmic activity in area V4 can impair shape detection and neuronal reliability.

Abstract

Rhythms at a population level are a defining characteristic of both normal and pathological cortical activity, but it is unclear howsuch rhythms interactwith activity of specific neurons to impact task performance on a trial-by-trial basis. We address this by employing a challenging visual detection task in which male rhesus macaques must signal the presentation of a shape embedded in a noisy background. We analyzed the rhythmic activity in the local field potential (LFP) and single neuron activity in area V4, a brain area strongly implicated in shape perception, prior to such presentations and focused on two different frequency ranges: alpha/beta (10-30 Hz), in which coherence was particularly strong and spatially extensive, and gamma (50-70 Hz), which has traditionally been strongly associated with single unit activity. We find that within sessions there were periods of time during which successful detection was associated with the absence of rhythmic activity prior to shape presentation in either frequency range. During these periods, rhythmic activity in both frequency bands could predict whether the shape would be detected by the animal at the time of, as well as before, shape presentation on a trial-to-trial basis with high accuracy. Importantly, for both frequency ranges, the individual neurons carrying the most relevant information with regard to the task had the weakest coupling to the LFP rhythms. These results are consistent with spatially-distributed rhythmic activity acting as a source of decision noise in the context of rapid visual detection by reducing the moment-to-moment reliability of task-relevant information carried by individual neurons.Significance Statement Although rhythmic activity in the brain has been studied for over 100 years, its relevance to information processing remains unresolved. In this study we show for the first time that, in the context of a challenging visual detection task, rhythmic activity in local populations of neurons prior to appearance of the visual stimulus can predict mistakes on a trial-by-trial basis. Furthermore, this activity is linked to task-relevant signals at a neuronal level because the individual neurons with the weakest coupling to these rhythms are the most reliable.