30. Role of feedback connections: A Bayesian inference model

¹Centre for Theoretical and Computational Neuroscience, University of Plymouth, Plymouth, UK

There exists substantial evidence, including recently published papers, showing V1/V2 cortical activity responding to illusory contours (eg. Kanizsa019s triangle). This means V1 is representing information which correlates with the visual perception but which does not arise from the ordinary feedforward paths (retina and LGN). The same evidence strongly suggests this response is driven by feedback connections from higher-level areas. On the other hand, several authors have emphasized the important role played by feedback connections in visual perception and have suggested hierarchical Bayesian inference as a plausible underlying interpretation (Rao & Ballard, 1997; Friston, 2003; Lee & Mumford, 2003; Murray et al., 2004; Olshausen & Field, 2005; Sillito et al., 2006). The proposed model aims at providing a functional interpretation of the illusory contour phenomenon, and more generally, of the role of feedback connections in visual perception, under the perspective of a Bayesian inference framework. More precisely, we attempt to implement a feedback model based on the Bayesian Belief propagation algorithm, taking as a startpoint a well-known feedforward cortex-based object recognition model (HMAX model - see Riesenhuber & Poggio, 1999; Cadieu et al., 2007).