A sparse coding model of V1 produces surround suppression effects in response to natural scenes

Recent electrophysiology research has made significant advancements toward revealing the neural basis of early visual processing. The brain is optimized to draw conclusions from natural scenes, and models of the human visual system may uncover principles by which to develop better automated vision systems. In turn, the neuroscience community would benefit from deeper understanding of human vision through the implementation and testing of models of this neural system. While many neural coding models have been proposed for the primary visual cortex (V1), it remains an open question as to which model best describes the diversity of observed response properties. For instance, the canonical linear-nonlinear model (LN) partially explains some fundamental mechanistic and phenomenological properties of V1, but is unable to explain many nonlinear response properties that are likely associated with the keys to efficient and robust human vision. Surround suppression is one such nonlinear response property in which visual stimuli extending beyond the classical receptive field (CRF) selectively diminish neural responses. This property has been studied through electrophysiology experiments with synthetic stimuli (e.g., gratings). Surprisingly, high level sparse coding models implemented in a biologically plausible dynamical system have been shown to produce surround suppression effects that match individual and population observed responses. More recently, surround suppression has been investigated experimentally using natural stimuli, and these experiments have shown an increase in the sparsity of measured responses. Despite these findings, it remains unclear whether a functional sparse coding model is sufficient to produce the types of surround suppression observed with natural stimuli. This thesis demonstrates that the surround suppression effects recently observed with natural stimuli are also emergent properties of a sparse coding model. First, relevant literature in human vision and signal processing will be reviewed. The methods for implementing the model and the results from simulations will then be presented followed by discussion of implications of these results and future work.Butera, Robert - Faculty Mentor; Rozell, Christopher - Faculty Mento