Experimental Neuroscience has successfully identified the molecules and many of the neural pathways of the mind. In a number of important cases, e.g., spatial memory in rats, it has directly linked these players and pathways to behavior. In many cases it remains, however, unclear to what extent these finding actually “explain” behavior. For although our neurons share a common chemical composition, there are over 100 billion neurons per brain–each talking with approximately 10000 of its neighbors across synapses that are rapidly strengthened or weakened as a function of activity.
In order to bridge mind and molecule we must tame this neural net. The complexity of the net, together with its ability to change under our eyes, argues against relying solely on intuition and for the construction of a theoretical framework that yields computationally tractable predictions and helps guide further experiment.
Traditional Neuroscience uses reductionism to formulate hypotheses and tests them experimentally, while Theoretical and Computational Neuroscience builds on Information Theory, Dynamical Systems Theory, and Computer Science to create theoretical models to be tested numerically. Collaborations of neuroscientists, individually trained in experimental and computational approaches, are not unusual on the basis of experimental data. In extension of this, we advocate a synergistic use of both approaches to control the experiment itself. Commensurate with our escalating knowledge of neural function, the complexity of experiments to analyze both healthy and diseased brain function is ever-increasing. In this situation, it is necessary to utilize the analytic and predictive nature of Theoretical and Computational Neuroscience not only between but rather during experiments.