Efficient coding predicts much about early sensory processing, particularly in the visual system. However, the impressive diversity of cell types and visual receptive field properties present in the retina have, thus far, not been predicted or explained by the theory. Perhaps this is one reason many visual neuroscientist continue to think about visual processing more in terms of labeled lines and cell types serving the needs of specific behavioral niches — e.g., ‘looming’ cells for detecting looming predators. I will discuss work from my lab as well as my collaborator, John Pearson, that shows the impressive predictive power of efficient coding theory and paints a possible way forward for understanding the origin and diversity of cell types in the early visual system.
Health Sciences G328 and Zoom