Carnegie Mellon neuroscientist proposes new theory of brain f...( PITTSBURGH Carnegie Mellon Universit...),Health

PITTSBURGH Carnegie Mellon University neuroscientist Marcel Just and Stanford postdoctoral fellow Sashank Varma have put forward a new computational theory of brain function that provides answers to one of the central questions of modern science: How does the human brain organize itself to give rise to complex cognitive tasks such as reading, problem solving and spatial reasoning? Just and Varmas theory, called 4CAPS, is described in the fall issue of the journal Cognitive, Affective, and Behavioral Neuroscience.

More than a decade of research involving functional Magnetic Resonance Imaging brain scans in hundreds of laboratories has yielded a tremendous amount of information about what parts of the brain are activated when a person performs various tasks. Some researchers have been tempted to conclude that a simple one-to-one relationship exists between high-level mental tasks and brain areas. For example, some believe that a specific brain area is responsible for a specific cognitive task, such as identifying a face.

Just and Varma, however, propose that the evidence reveals a more complex picture in which thinking is a network function a collaboration of several brain areas that is constantly adapting itself, based on the task at hand and the brains own resources and biological limitations. The collaborating parts of the brain, according to Just, are like members of a sports team whose players substitute in and out of the action. 4CAPS (an acronym for Capacity Constrained Concurrent Cortical Activation-based Production System), proposes a decentralized process by which members of the cortical team volunteer themselves when their strengths are called for, but also permits less efficient but capable members to step forward when the primary player is injured or disabled, as might occur as a result of a stroke. Just and Varma have constructed a number of computational models to demonstrate this process, such as a model that understands Engli
'/>"/>