John Hopfield's famous 1982 paper unleashed a flood of publications by theoretical physicists on neural networks. By the 1990s this flood had subsided, but in its wake was a group of physicists who continued to pursue their goal of understanding the brain. This group has since fostered a new generation of colleagues, and pioneered new applications of physical thinking to problems in brain science. We propose to organize a full semester program at the Institute for Theoretical Physics (ITP). The objectives of this program are to heighten and focus the interactions among members of this group, to recruit new physicists to the study of neuroscience, and to insure that experimentalists in neuroscience can interact with the experts in their specialty within computational neuroscience. The initial ideas of physicists about the dynamics of neural networks were metaphor more than science. This situation has changed during the past decade, as these ideas have begun to be more firmly grounded in cellular biophysics and the neural correlates of behavior. As practiced by theoretical physicists, neural network modeling was initially separate from the biophysical tradition of Hodgkin and Huxley, which focused on the properties of single neurons. One of the most exciting and encouraging developments of the 1990s is the convergence of these two strands of research. Another exciting development is the use of neural network models to understand the data obtained from single unit recordings by neurophysiologists. As mathematical models of neural networks have moved closer to experimental realities, three fundamental theoretical issues have emerged: (i) The dichotomy between rate-based and spike-based descriptions of neural activity: (ii) The role of recurrent synaptic feedback; and (iii) The effects of short- and long-term synaptic plasticity. These three issues in the dynamics of neural networks form the unifying themes of the proposed ITP program. As a complementary venue to the research aspect of the full program, we propose to hold a two-week Pedagogical Workshop on topics that exemplify the confluence of experimental and theoretical neuroscience. This workshop will serve as an introduction to Neuroscience and Neural Network Modeling for individuals, particularly junior scientists, with a physical or mathematical background. It will act as a catalyst to attract individuals that are appropriate for the NIH (K25)Mentored Quantitative Research Career Development Awards. It may also act to increase the impact of the larger physics community in neuroscience, both through theoretical studies and advanced biophysical instrumentation.
