By Frederic Chavane, CNRS and Aix-Marseille University

Visual motion integration is traditionally viewed as a cascade of hierarchical processing steps identified using local stationary inputs. However naturalistic inputs are most often non-stationary, such as an object moving along smooth trajectories. This later will generate a sequence of spatio-temporally coherent feed-forward inputs that shall interact with spreads of activity within and between cortical retinotopic maps. In awake monkeys, we investigated how these nested propagations shape the cortical mapping of a motion trajectory. Recording the population response dynamics of apparent motion stimuli using voltage-sensitive dye imaging, we show that non-linear interactions between feedforward input and lateral interactions are essential to shape the spatio-temporal representation of the stimulus velocity (speed and direction). In response to continuous motion along a trajectory, such interactions are responsible for the emergence of anticipatory spiking activity. Predictive and accurate representation of non-stationary motion signals along trajectories thus results from the convergent non-linear interplay of intra- and inter-cortical inputs propagating information faster than the feed-forward sequence.