Activities
Eero Simoncelli
Characterizing Neural Encoding with Single Spiking Models
A fundamental goal of sensory systems neuroscience is the
characterization of the functional relationship between
environmental stimuli and neural response. In seeking
computational models that can be used for this purpose, there
is a natural tradeoff between biological realism and
computational simplicity/tractability. I'll describe a family
solutions we've been investigating that inherit much of the
simplicity and computational advantage of linear systems, and
yet are capable of faithfully reproducing many of the complexities
of spiking neural responses. I'll discuss application of these
methods to visual neurons in retina and cortex.