Area of Research
Cognitive/Systems Neuroscience, Theoretical Neuroscience, Synapses and Circuits, Sensory Physiology
Models of Visual Cortical Circuitry and Development, Analysis of Visual Coding
My lab's interests focus on understanding the cerebral cortex. We use theoretical and computational methods to unravel the circuitry of the cerebral cortex, the rules by which this circuitry develops or "self-organizes", and the computational functions of this circuitry. Our guiding hypothesis -- motivated by the stereotypical nature of cortical circuitry across sensory modalities -- is that there are fundamental computations done by the circuits of sensory cortex that are invariant across highly varying input signals. This commonality is likely to extend in important ways to motor and "higher-order" cortex as well, although these structures show more prominent circuit differences with sensory cortex, consistent with their role in producing internally generated activity as well as in integrating their inputs. In some way that does not strongly depend on the specific content of the input, cortical circuits extract invariant structures from their input and learn to represent these structures in an associative, relational manner. We (and many others) believe the atomic element underlying these computations is likely to be found in the computations done by a roughly 1mm-square chunk of the cortical circuit. To understand this element, we have focused on one of the best-studied cortical systems, primary visual cortex, and also have interest in any cortical system in which the data gives us a foothold (such as rodent whisker barrel cortex, studied here at Columbia by Randy Bruno, and monkey area LIP, studied here by Mickey Goldberg and Jackie Gottlieb).
The function of this element depends both on its mature pattern of circuitry and on the developmental and learning rules by which this circuitry is shaped by the very inputs that it processes. Thus we focus both on understanding how the mature circuitry creates cortical response properties and on how this circuitry is shaped by input activity during development and learning. We also use theoretical methods to analyze sensory cortical data to infer functional properties.
With respect to the mature circuitry: In work done in 1998-2003, we showed how most "classical receptive field" properties -- the basic tuning of a primary visual cortical neuron for patterns of visual input -- could be accounted for by "feedforward" circuitry, including feedforward excitatory input relayed from the eyes and feedforward inhibitory input activated by this feedforward excitation. This left unexplained the role of the strong recurrent connections among cortical cells. In more recent work, we have shown how a balance of strong excitatory and inhibitory recurrence can serve as an adjustable gain mechanism, which we hypothesize adjusts gain without significantly altering tuning. In one study, we examined "surround suppression", defined as follows: the region of visual space in which appropriate visual stimuli can drive a neuron's spiking responses is called the neuron's classical receptive field or CRF; stimuli outside of this region (in the neuron's "surround)l can suppress the neuron's response to CRF stimuli. We showed, in collaboration with the experimentalist lab of David Ferster at Northwestern, that surround suppression operates as such a turning down of gain -- a de-amplification, rather than an addition of inhibition. In another study, we showed that this mechanism allows signal amplification without slowing the dynamics of the signal, unlike previously proposed mechanisms of amplification that amplify by dynamical slowing. Separately, we showed, in collaboration with the experimental labs of Mickey Goldberg at Columbia and Mike Shadlen at U Wash, that the more classic form of amplification (by slowing) can explain a set of puzzling observations in monkey area LIP, provided that only a single pattern of neural activity shows such slowing, and we provided evidence that this is indeed the case. We are currently building on this work to address a wide range of spatial, temporal, and nonlinear behaviors of sensory circuits.
In older work on the development of circuitry, we proposed and analyzed models showing how fundamental response properties of primary visual cortical neurons -- orientation selectivity and ocular dominance -- and their arrangement into periodic maps across the cortical surface (orientation columns and ocular dominance columns) could be explained by simple activity-dependent mechanisms of "correlation-based" synaptic learning from spontaneous activity, without structured visual input (since orientation columns and some aspects of ocular dominance columns develop without structured input). In our recent and current developmental work, we are addressing a number of issues related to the onset of the critical period for ocular dominance plasticity, a period in the life of a young animal when closing of an eye leads over a few days to a strong loss of cortical input from that eye and over a few more days to a strengthening of the open eye's inputs. Before the critical period, closing of an eye has effects on cortical development -- that is, the cortex "knows" that one eye is closed -- yet it does not change the balance of the two eyes' inputs. The onset of the critical period has been shown to be determined by the maturation of cortical inhibition, which depends on visual experience. We have shown how this maturation of inhibition can also account for the equalization of the two eyes' inputs to cortex, which also depends on visual experience and occurs coincident with the onset of the critical period. In ongoing work, we are addressing why the maturation of inhibition initiates the critical period and how existing synaptic learning rules must be modified to account for the observed dynamics of ocular dominance plasticity.
Ozeki H., I.M. Finn, E.S. Schaffer, K.D. Miller and D. Ferster (2009). Inhibitory stabilization of the cortical network underlies visual surround suppression. Neuron 62:578-592.
Toyoizumi T. and Miller K.D. (2009). Equalization of ocular dominance columns induced by an activity-dependent learning rule and the maturation of inhibition. Journal of Neuroscience 29:6514-25.
Murphy, B.K. and K.D. Miller (2009). Balanced amplification: A new mechanism of selective amplification of neural activity patterns. Neuron 61:635-648.
Ganguli, S., J.W. Bisley, J.D. Roitman, M.N. Shadlen, M.E. Goldberg, and K.D. Miller (2008). One-dimensional dynamics of attention and decision making in LIP. Neuron 58:15-25.
Sharpee, T.O., H. Sugihara, A.V. Kurgansky, S.P. Rebrik, M.P. Stryker and K.D. Miller (2006). Adaptive Filtering Enhances Information Transmission in Visual Cortex. Nature 439, 936-942.