Inferring eye position from populations of lateral intraparietal neurons.

Arnulf Ba Graf,Richard A Andersen

doi:10.7554/elife.02813

Abstract

Understanding how the brain computes eye position is essential to unraveling high-level visual functions such as eye movement planning, coordinate transformations and stability of spatial awareness. The lateral intraparietal area (LIP) is essential for this process. However, despite decades of research, its contribution to the eye position signal remains controversial. LIP neurons have recently been reported to inaccurately represent eye position during a saccadic eye movement, and to be too slow to support a role in high-level visual functions. We addressed this issue by predicting eye position and saccade direction from the responses of populations of LIP neurons. We found that both signals were accurately predicted before, during and after a saccade. Also, the dynamics of these signals support their contribution to visual functions. These findings provide a principled understanding of the coding of information in populations of neurons within an important node of the cortical network for visual-motor behaviors.DOI: http://dx.doi.org/10.7554/eLife.02813.001.

Highlights

Understanding how the brain computes eye position is essential to unraveling highlevel visual functions such as eye movement planning, coordinate transformations and stability of spatial awareness
To address population coding in lateral intraparietal area (LIP), we used Bayesian inference (Dayan and Abbott, 2001; Jaynes, 2003; Pouget et al, 2003; Graf et al, 2011) to compute the neuronal population code (NPC): the accuracy with which pre- and postsaccade eye positions and the saccade direction could be predicted from the population response
We have shown that the eye position signal (EPS) is accurately carried by populations of LIP neurons before, during and after an eye movement

Summary

Introduction

Understanding how the brain computes eye position is essential to unraveling highlevel visual functions such as eye movement planning, coordinate transformations and stability of spatial awareness. To illustrate the mechanisms of predicting oculomotor behaviors from neuronal populations, we computed the population activity in the memory epoch for two planned saccade directions (135 and 270 deg) starting at the preferred eye position (red and blue curves, bottom panel, Figure 1B).

Results

Conclusion