A theoretical framework in which networks of interconnected nodes (analogous to neural networks) process information through an activation dynamics while their connectivity evolves according to learning dynamics. Connectionist networks receive inputs that ultimately arise from sensor systems. Particular output layers are assumed to ultimately determine behavior. By generating activation at the output nodes in response to sensory inputs, connectionist networks process information. Through the learning dynamics, the connectivity of the network over the long run comes to reflect the sensory environment of the system.
See Activation (in a connectionist model), Artificial intelligence, Auto-encoder networks, Backpropogation, Cognitive neuroscience, Computational models, Connectionist models, Neural net