FROM THE METAPHOR TO THE MODEL: A BRIEF HISTORY OF COMPUTATIONAL NEUROSCIENCE AND THE SEARCH FOR THE ENGRAM

M.R. Penner*; S.N. Burke

Neural Systems, Memory & Aging, Univ Arizona, Tucson, AZ, USA

Computational neuroscience is a field devoted to interpreting the information content of neuronal signals by modeling many levels of the nervous system. Although in the grand scheme of things, the field of computational neuroscience may seem new, the central question this field addresses is certainly not: how is information represented and stored in the brain? Following WWII, many physicists working on things such as the Manhattan Project, switched their attention to biophysics, thus providing some of the groundwork for modeling the brain. At one level, realistic brain models involve large-scale simulations that include as much cellular detail as possible such as the Hodgkin-Huxley (1952) model of the action potential in the squid giant axon. At the network level, simplifying brain models consider how to interpret the information encoded by the activity of a large neuronal population. In his now famous book, D.O. Hebb (1949) was one of the first to describe a mechanism whereby information can be represented in the brain in ensembles of nerve cells he called cell assemblies and phase sequences. Other notable contributions include that of Pitts and McCulloch who (1947) addressed the issue of pattern recognition; Steinbuch (1961) who proposed the learning matrix, providing a starting point for most other computational models; and David Marr who introduced the notion of feedforward inhibition to the learning matrix. Of course, there are many others who have also made substantial contributions to this field. In this presentation, the contributions of these pioneers are summarized along with the work of other less well-known, and contemporary scientists.