Computational issues in natural auditory scene analysis

Scene analysis is a complex process involving a hierarchy of computational problems ranging from sensory representation to feature extraction to active perception. It is studied in a wide range of fields using different approaches, but we still have only limited insight into the computations used by biological systems. Experimental approaches often implicitly assume a feature detection paradigm without addressing the true complexity of the problem. Computational approaches are now capable of solving complex scene analysis problems, but these often defined in a way that is of limited relevance to biology. The challenge is to develop approaches for deducing computational principles relevant to biological systems. I will present the view that scene analysis is a universal problem solved by all animals, and that we can gain new insights by studying the problems that animals face in complex natural environments. From this, I will present framework for studying scene analysis comprising four essential propertie...