Since the pioneering work of Alan Newell and Herb Simon in the late 1950s, researchers interested in human cognitive processes have used computer simulations to try to identify general principles of cognition. The strategy is to build computational models that embody putative principles and then examine how well the models capture human performance in cognitive tasks. Until the 1980s, this effort was undertaken largely within the context of the "computer metaphor" of mind: Researchers thought

more »

... the human mind as though it were a conventional digital computer and built computational models based on this conceptualization. Indeed, this approach has lead to considerable progress in modeling explicit reasoning, problem solving, and other high-level cognitive processes. By the late 70s and early 80s, however, many researchers began to think that an alternative framework was needed to capture the full range of cognitive behavior-one based more closely on the style of computation employed by the brain. The new approach, called connectionist or neural network modeling, or the parallel distributed processing approach (Elman et al., 1996; McLeod et al., 1998; , implements cognitive processes in terms of massively parallel cooperative and competitive interactions among large numbers of simple neuron-like computational units. Unit interactions are governed by modifiable excitatory and inhibitory weights on connections among the units. Although each unit exhibits nonlinear spatial and temporal summation, units and connections are not generally be taken as corresponding directly to individual neurons and synapses. Rather, the connectionist approach attempts capture the essential computational properties of the vast ensembles of real neuronal elements found in the brain using simulations of smaller networks of more abstract units. By linking neural computation to behavior, the framework enables developmental, cognitive and neurobiological issues to be addressed within a single, integrated formalism. REPRESENTATION. An issue of central relevance in understanding cognition is the nature of the representations used in cognitive processes. There are two basic approaches to representation within connectionist networks.