Current Theoretical Conceptualization of Brain Organization

Current Theoretical Conceptualization of Brain Organization

Current thinking in the cognitive neurosciences strikes a balance between local and holistic organization of functions. The current emphasis is on “connectivity”, as both a theoretical heuristic and a set of models of neural behavior. The localizationist perspective of Wernicke introduced the rudimentary idea that interconnected brain regions could work in a serial fashion to perform a set of oper-ations such as language. The Russian neuropsychologist Alexander Luria, who influenced much contemporary thinking in cognitive neuroscience, was himself influenced by the idea of connectivity.

By borrowing from the then emerging field of machine intel-ligence, the model of the brain as a digital computer operating in a serial processing framework was introduced. In this model, brain functions were seen as the constellation of interconnected neural net-works, each operating in serial fashion, to produce macrocognitive processes from series of local microcognitive operations. Although the early artificial intelligence and information-processing traditions of cognitive psychology offered elegant models, they were limited by their underlying assumptions. First, the cognitive models that pos-tulated the premise of connectivity between local networks did not explicitly link the cognitive realm with the underlying anatomical substrate, lending these theories a lack of neural reality. For example, this yielded models of attention that “worked” from a cognitive per-spective but did not suggest an underlying neural mechanism.

Connectionist theories of the brain within the cognitive sciences have begun to offer models with greater neural and com-putational realism. In the computer realm, the development of massively parallel computing has suggested that extremely com-plex functions can be performed simultaneously, not in a serial fashion. The obvious extrapolation to neural science is to think of brain functions as having a parallel architecture, which would perform several operations simultaneously. In contrast to serial processing, parallel processing allows the relatively slow indi-vidual neurons to accomplish tasks in rapid real time. The other current concept applied to neural processing is that of distributed processing. Distributed processing refers to the coordination of functions that are distributed within and across brain regions. A particular function is therefore emergent from neural process-ing that is both parallel and distributed. This concept of neural function represents a compromise between local and holistic perspectives. The model of the brain in terms of parallel distrib-uted processing (PDP) is both a heuristic framework from which to view neural organization and a formal set of models within cognitive science. The formal elements of PDP involve the rela-tionship of neural science and computer science and have been applied to the development of models of language, vision, motor learning and memory. As a heuristic framework, the notion that higher cortical functions are best described as parallel and dis-tributed is quite influential to most current thinking in the cogni-tive neurosciences.