Mark Goldman

Animal behavior typically involves interactions among networks of large numbers of interconnected neurons, but experimental techniques in most systems are limited to the direct measurement of single or small numbers of neurons. My laboratory uses computational modeling to bridge the gap between single-neuron measurements and hypothesized network function. We study a wide variety of systems and seek to address questions ranging from cellular and network dynamics to sensory coding to memory and plasticity. 

One major project involves constructing models of networks involved in motor control and working memory in which a briefly presented stimulus leads to long-lasting, or “persistent”, neural activity whose level encodes the triggering stimulus. Our work focuses primarily on a model system for studying such activity, the oculomotor neural integrator, in which transient eye movement commands are accumulated into persistent neural signals that control the horizontal position of the eyes. By modeling electrophysiological data, we seek to dissect the contributions of synaptic inhibition, synaptic excitation, and intrinsic cellular properties to the generation of the observed neural activity. More recently, we have also been studying the role of the cerebellum in controlling the plasticity of oculomotor responses.

In a second set of projects, we have been modeling how neurons in sensory areas encode information about stimuli in the environment. In one project, we used methods from information theory to address fundamental questions about the manner in which noise in the nervous system affects the neural code. In a second project, we are analyzing how the response properties of color selective cells in the primary visual cortex may explain “color constancy” – the observation that the color of objects looks the same to us even when the light reflecting from them can differ dramatically depending on whether the illuminant is natural or artificial light.