Connectionism refers to a theoretical approach in cognitive science and artificial intelligence that models cognitive processes through networks of simple, interconnected units.
These networks, often called artificial neural networks, are connected by weighted edges whose values are adjusted by learning algorithms such as backpropagation.
Connectionism is based on the assumption that knowledge is not symbolically encoded but is represented in the distribution of connection strengths.
It enables the modeling of parallel processing, pattern recognition, and robust learning despite noisy data.
Definition:
Connectionism is an approach that explains cognitive functions through distributed representations in networks of simple, interconnected units.
Source:
Rumelhart, D. E., McClelland, J. L. (1986). Parallel Distributed Processing: Explorations in the Microstructure of Cognition. MIT Press.