Information Processing in Neural Networks

Understanding how a brain system processes information requires knowing what information is represented in the system, how the information is encoded and decoded, and what the computations are that transform the information throughout different stages of processing. Neural systems use a variety of coding schemes, including rate codes, temporal codes, and population codes. Any encoding scheme is constrained by the ability of downstream neurons to decode the information. The building blocks of these circuits include feedforward, feedback, and lateral connections; excitatory and inhibitory connections; and recurrent collaterals. Circuits show varying degrees of divergence, convergence, and parallel processing, depending on the information-processing task at hand. Because the systems need to adapt to countless behavioral and perceptual conditions, plasticity and neuromodulation are ubiquitous throughout most brain systems. These principles are exemplified in a number of well-studied neural circuits.