Neural Information Processing Systems

Many neurons transform complex time-varying inputs into a string of output pulses. One can think of this transformation as the computation that the neuron performs upon its input. We seek methods to characterise this computation, working both from experimental data and from simple neural model dynamical systems. As biological systems must respond to signals which arise in widely different contexts, many systems employ adaptation to refine the neural computation taking into account the local context. We study this process and how it helps the system to efficiently process information in a variety of sensory systems, including the retina, somatosensory cortex and the visual motion detection of the fly.