Attenuation of noise correlations in the transformation from the frontal eye field to movement.

Abstract

Correlated activity between neurons can cause variability in behavior across trials, as trial-by-trial cofluctuations can propagate downstream through the motor system. The extent to which correlated activity affects behavior depends on the properties of the translation of the population activity into movement. A major hurdle in studying the effects of noise correlations on behavior is that in many cases this translation is unknown. Previous research has overcome this by using models that make strong assumptions about the coding of motor variables. We developed a novel method that estimates the contribution of correlations to behavior with minimal assumptions. Our method partitions noise correlations into correlations that are expressed in a specific behavior, termed behavior-related correlations, and correlations that are not. We applied this method to study the relationship between noise correlations in the frontal eye field (FEF) and pursuit eye movements. We defined a distance metric between the pursuit behavior on different trials. Based on this metric, we used a shuffling approach to estimate pursuit-related correlations. Although the correlations were partially linked to variability in the eye movements, even the most constrained shuffle strongly attenuated the correlations. Thus, only a small fraction of FEF correlations is expressed in behavior. We used simulations to validate our approach, show that it captures behavior-related correlations, and demonstrate its generalizability in different models. We show that the attenuation of correlated activity through the motor pathway could stem from the interplay between the structure of the correlations and the decoder of FEF activity.NEW & NOTEWORTHY The effect of noise correlations on neural computations has been studied extensively. However, the degree to which correlations affect downstream areas remains unknown. Here, we take advantage of precise measurement of eye movement behavior to estimate the degree to which correlated variability between neurons in the frontal eye field (FEF) affects subsequent behavior. To achieve this, we developed a novel shuffling-based method and verified it using different models of the FEF.