Purchase this article with an account.
Émilien Tlapale, Barbara Dosher, Zhong-Lin Lu; An anisotropic model of visual motion perception and perceptual learning. Journal of Vision 2016;16(12):667. doi: 10.1167/16.12.667.
Download citation file:
© 2017 Association for Research in Vision and Ophthalmology.
The performance on visual tasks involving stimuli oriented along the cardinal directions is significantly superior to the performance with the same stimuli oriented along oblique directions, a phenomenon classically designated as the "oblique effect" (Appelle 1972). Increased cell numbers or neural activity, and narrower tuning bandwidth for the cardinal directions have been suggested to account for the phenomenon (Li et al., 2003). Here we investigate the oblique effect in visual motion perception and perceptual learning by incorporating an anisotropy in the distribution of directional preferences of motion sensitive neurons into the integrated dynamical motion model (IDM) (Tlapale et al. 2014, 2015), a biologically plausible neural model including representations of several cortical areas (V1, MT, LIP) and their feedforward and lateral connections, with a novel neural mechanism for same-different decisions. Differential direction tuning bandwidths emerge as a direct consequence of the direction preference distributions. This biologically plausible model is more robust than previous static models of motion anisotropy (Rokem & Silver, 2009; Wong & Price, 2014). The anisotropic IDM correctly reproduces oblique effects in a range of representative experiments, including the classical visual motion perception experiments of Gros et al (1999), Matthews & Qian (1999) and Dakin et al (2005). Furthermore, the extended model accounts for the perceptual learning experiments of Ball & Sekuler (1982,1987), Watanabe et al (2001, 2002) and Liang et al (2015) through feedforward reweighting. It also generates predictions about the influence of aperture orientation on multi-stable motion displays. In summary, we developed a biologically plausible and robust dynamical neural model of visual motion perception and perceptual learning which includes neural dynamics of multiple stages of motion processing, precise descriptions of neuronal anisotropies and decision mechanisms, and is able to account for a large range of representative experiments from the literature on the oblique effect and perceptual learning.
Meeting abstract presented at VSS 2016
This PDF is available to Subscribers Only