Abstract
What are the neural mechanisms underlying motion integration of translating objects? Visual motion integration is generally conceived of as a feedforward, hierarchical, information processing. However, feedforward models fail to account for many contextual effects revealed using natural moving stimuli. In particular, a translating object evokes a sequence of transient feedforward responses in the primary visual cortex but also propagations of activity through horizontal and feedback pathways. We investigated how these pathways shape the representation of a translating bar in monkey V1. We show that, for long trajectories, spiking activity builds-up hundreds of milliseconds before the bar enters the neurons receptive fields. Using VSDI and LFP recordings guided by a phenomenological model of propagation dynamics, we demonstrate that this anticipatory response arises from the interplay between horizontal and feedback networks driving V1 neurons well ahead of their feedforward inputs. This mechanism could subtend several perceptual contextual effects observed with translating objects.
Laurent U Perrinet
Researcher in Computational Neuroscience
My research interests include Machine Learning and computational neuroscience applied to Vision.