Dynamics of Cognitive Processing and Learning: Linking Connectionist, Neuroscience, and Dynamical Systems Approaches

Abstract:
Connectionist or parallel-distributed processing models provide mechanistic accounts of cognitive processes, demonstrating how emergent dynamics of higher-order cognitive states such as percepts, decision states, and actions can arise from the micro-dynamics of the interactions of simple neuron-like processing units. These models can also address changes in representation and processing that occur as a result of experience, via adjustments in the strengths of the connections among the units that participate in processing. As such these models make several points of contact with the literature on dynamical systems in development. The models also link to many other recent models on the neural population dynamics underlying decision making in the brain. This talk will discuss recent developments related to these issues, building on a model of the dynamics of simple perceptual decision making (Usher and McClelland, 2001). http://tinyurl.com/cb2wuv