Abstract

Many tasks in daily life involve a combination of perceptual decisions and navigation. Rodent parietal cortex has been implicated in both of these processes, with some studies focusing on its role in decisions and others on its role in navigation. Here we show that, when mice use vision to decide where to navigate, parietal cortex robustly encodes navigational, rather than perceptual or decision-related information. We trained mice in a two-alternative forced choice task, which required them to navigate in a virtual T-shaped corridor in which the correct choice was signaled by visual contrast on the corridor walls. 2-photon calcium imaging revealed that neurons in parietal cortex coded for combinations of the animal’s position and heading direction in the virtual room, and their responses were highly predictable based on these measures. Different neurons exhibited diverse heading-position tuning, so the population as a whole could be readily decoded to predict the mouse’s navigation paths in single trials. The neurons were also informative about the mouse’s choice, but the choice could be easily predicted from heading-position trajectories of the mouse through the room. Spatial coding in parietal cortex required active navigation, not just vision: during playback of previous navigation scenes to passive mice, activity in visual cortex matched that during active behavior, but activity in parietal cortex did not. We conclude that when mice use visual information to guide navigation, parietal cortex encodes spatial factors rather than visual information or abstract decisions. These spatial factors involve precise combinations of spatial position and direction of heading.

Matteo Carandini is the GlaxoSmithKline / Fight for Sight Professor of Visual Neuroscience at University College London, where together with Kenneth Harris he co-directs the Cortical Processing Laboratory (www.ucl.ac.uk/cortexlab). Before joining UCL in 2007, he ran laboratories at the Smith-Kettlewell Eye Research Institute in San Francisco (2002) and at the Swiss Federal Institute of Technology in Zurich (1998). He holds a Laurea in Mathematics from the University of Rome (1990) and a PhD in Neural Science from New York University (1996). The aim of his research is to understand the computations performed by neuronal populations in the visual system and beyond, the underlying circuits, and the way these computations lead to decisions and actions. He addresses these questions in mice that perform sensory and navigation tasks. Carandini is a McKnight Scholar (2005), a European Research Council Advanced Investigator (2007), a Wellcome Trust Senior Investigator (2012), and a Simons Investigator (2014).