Skip to Main content Skip to Navigation

Recurrent network dynamics reconciles visual motion segmentation and integration

Abstract : In sensory systems, different computational rules are postulated to be implemented by different neuronal subpopulations, each one being characterised by a particular tuning function. For instance, in primate cortical area MT, different classes of direction-selective cells have been identified and related to either motion integration, segmentation or transparency. Still, how such different tuning properties are constructed is unclear. The dominant theoretical viewpoint postulates that differential weighting of MT inputs along the linear-nonlinear feedforward cascade is sufficient to build these different cell classes but it does not account for their complex temporal dynamics and their versatility when facing different input statistics. Here, we demonstrate that a recurrent network model of visual motion processing can reconcile these different properties. We show how excitatory and inhibitory recurrent connections, within a direction representation space, shape neuronal motion direction tuning and implement different computational rules such as vector averaging, winner-take-all or superposition. The model also captures ordered temporal transitions between these behaviours. In particular, depending on the inhibition regime the ring network can switch from motion integration to motion segmentation, thus being able to compute either a single pattern motion or to superpose multiple inputs as in motion transparency. We thus demonstrate that recurrent architectures can adaptively give rise to different cortical computational regimes depending upon the input statistics, thus reconciling the two facets of sensory processing: integration and segmentation.
Complete list of metadata

Cited literature [50 references]  Display  Hide  Download
Contributor : Pierre Kornprobst Connect in order to contact the contributor
Submitted on : Friday, March 3, 2017 - 2:30:57 PM
Last modification on : Tuesday, October 19, 2021 - 10:51:34 PM
Long-term archiving on: : Tuesday, June 6, 2017 - 12:54:54 PM


Files produced by the author(s)


  • HAL Id : hal-01482294, version 1


N. V. Kartheek Medathati, James Rankin, Andrew Meso, Pierre Kornprobst, Guillaume Masson. Recurrent network dynamics reconciles visual motion segmentation and integration. [Research Report] RR-9041, Inria Sophia Antipolis. 2017, pp.28. ⟨hal-01482294⟩



Record views


Files downloads