Keywords
spatial distortions, short term synaptic plasticity, natural vision, motion perception, motion adaptation aftereffect
Abstract
Motion information is processed in a neural circuit formed by synaptic organization of feedforward (FF) and feedback (FB) connections between different cortical areas. However, the contribution of a recurrent FB information to adaptation process is not well explored. Here, a biologically plausible neural model that predicts motion adaptation aftereffect (MAE) induced by exposure to geometrically skewed natural image sequences is suggested. The model constitutes two stage recurrent motion processing within cortical areas V1 and MT [1]. It comprises FF excitatory, FB modulatory and lateral inhibitory connections, and a fast and a slow adaptive synapse in the FF and FB streams, respectively, to introduce plasticity. Simulation results of the model show the following main contributions of FB in distortion induced motion adaptation:
- FB disambiguates the main signal from a noisy natural stimulus input: results in adaptation to globally consistent salient information.
- A model with distinct adaptive mechanisms in FF and FB streams predicts MAE at different time scales of exposure to skewed natural stimuli more accurately than other model variants constituting single adaptive mechanism: Multiple adaptive mechanisms might be implemented via FB pathways.
- FB allows similar response tuning in model area V1 and MT during adaptation in line with physiological findings [2].
[1] Bayerl, P. and H. Neumann, Disambiguating visual motion through contextual feedback modulation. Neural computation, 2004. 16(10): p. 2041-2066.
[2] Patterson, C.A., et al., Similar adaptation effects in primary visual cortex and area MT of the macaque monkey under matched stimulus conditions. Journal of neurophysiology, 2013. 111(6): p. 1203-1213.
Start Date
16-5-2018 12:00 PM
End Date
16-5-2018 12:25 PM
Included in
Model investigation on contribution of feedback in distortion induced motion adaptation
Motion information is processed in a neural circuit formed by synaptic organization of feedforward (FF) and feedback (FB) connections between different cortical areas. However, the contribution of a recurrent FB information to adaptation process is not well explored. Here, a biologically plausible neural model that predicts motion adaptation aftereffect (MAE) induced by exposure to geometrically skewed natural image sequences is suggested. The model constitutes two stage recurrent motion processing within cortical areas V1 and MT [1]. It comprises FF excitatory, FB modulatory and lateral inhibitory connections, and a fast and a slow adaptive synapse in the FF and FB streams, respectively, to introduce plasticity. Simulation results of the model show the following main contributions of FB in distortion induced motion adaptation:
- FB disambiguates the main signal from a noisy natural stimulus input: results in adaptation to globally consistent salient information.
- A model with distinct adaptive mechanisms in FF and FB streams predicts MAE at different time scales of exposure to skewed natural stimuli more accurately than other model variants constituting single adaptive mechanism: Multiple adaptive mechanisms might be implemented via FB pathways.
- FB allows similar response tuning in model area V1 and MT during adaptation in line with physiological findings [2].
[1] Bayerl, P. and H. Neumann, Disambiguating visual motion through contextual feedback modulation. Neural computation, 2004. 16(10): p. 2041-2066.
[2] Patterson, C.A., et al., Similar adaptation effects in primary visual cortex and area MT of the macaque monkey under matched stimulus conditions. Journal of neurophysiology, 2013. 111(6): p. 1203-1213.