My long-term research interests center around understanding the physiological mechanisms underlying the functional and computational properties brain neuronal networks. That is, I want to understand the brain as an organ of biological computation and by extension to understand brain dysfunction as examples of failed computation. I believe that the mechanisms underlying computation are best uncovered by detailed studies of the physiological properties of the synapses, cells and circuits involved in the performance of a given task. In particular, I am interested in how neuronal diversity, synchrony, circuitry, dendritic integration and synaptic plasticity may allow small groups of neurons to perform complex and interesting functions. Understanding such computational properties of brain networks often requires the simultaneous acquisition of data from several cells within a network and/or from multiple locations within a single cell. Thus, I also am interested in the application and development of physiological and optical techniques that facilitate this sort of parallel data acquisition in vitro and in vivo. One key aspect of this approach in the last few years has been gaining an understanding of the role of different sources of variability in neuronal computation. Students and postdocs in my lab have studied the ways in which apparently noisy signals can generate structured, synchronous activity across populations of neurons. This mechanism, which we have called stochastic synchrony, seems to provide a basis by which additional noise from synapses or even fro sensory stimuli can generate useful patterns of brain activity that may enhance sensitivity and selectivity in sensory systems. Another area of interest has been to examine the degree to which cell to cell variability in intrinsic biophysical properties can be considered to be a beneficial "feature" of brain computation rather than a "bug" of biological imprecision. In this vein we have combined computational and experimental approaches to characterize cell to cell variability and then to assess how this variability affects the ability of populations of neurons to represent information about stimuli. We have found that populations of neurons in the real "noisy" brain (in the sense of a brain in which cells are somewhat stochastic in their properties) can convey information about twice as efficiently as a "perfect" brtaion in which all neurons of a given type have identical properties. Finally we have recently been applying these approaches to study mouse models of autism. In this case we are looking at trial to trial variability as a kind of "noise". Human studies demonstrate that reliability of sensory evoked responses is impaired in autistic subjects. Our goal is to identify sources of variability of neuronal responses in our mouse models and determine whether this kind of noise differs between control mice and those with autism-related mutations.