POSTER DETAILS
Distinct neuronal states encode task identity in frontal eye field and interact with its core spatial properties
Axel Mouille, Corentin Gaillard, Elaine Astrand, Claire Wardak, Julian Amengual, Suliann Ben Hamed
Date / Location: Monday, 11 July 2022 / S04-093
Add this poster event to your π
Add this poster event to your π
Human cognition relies on the coordinated activity of multiple neuronal populations dynamically interacting together and shaping our behavior. Interestingly, single-neuron activity in the prefrontal cortex displays mixed selectivity, showing simultaneous tuning to different task-related processes. For example, prefrontal populations can jointly represent the position of a visual spatial information, spatial attention and working memory content. This feature is a hallmark of high-dimensionality, which provides the ability to encode different types of information simultaneously. In this context, the question we tackle is whether the prefrontal cortex population also represents task identity and how this impacts on its core specific functional computations. We trained two monkeys to perform three different task: a memory guided saccade task and two cued target detection tasks: peripherally cued (cue at expected target location, exogenous) and centrally cued (central cue, endogenous). During their performance, multi-unit activity (MUA) was recorded in both Frontal eye fields (FEF). Using demixed Principal Component Analysis, we found a two-dimensional neural states that fully characterized each of these tasks. This result indicates a task-related neural state in the recorded population. Furthermore, we observed that encoding of the spatial information was task-dependent. Such interaction between task and position coding indicates that task and spatial information are non-linearly mixed which is considered as a signature of a high-dimensional neuronal representation. Overall, this indicates that the FEF encodes ongoing task-identity in an identifiable neuronal dimension that interacts with its core spatial computations.