Connectionist Models

Connectionist models, sometimes called artificial neural networks or parallel distributed processing models, refer to a family of mathematical models inspired by the way neurons in the brain work. The brain-like processing mechanisms that underlie connectionist models, as well as the ability of some models to learn from experience, to exhibit rule-like behavior without explicitly representing a rule, and to degrade gracefully when damaged (instead of exhibiting catastrophic failure) have made this family of computational models appealing to those who study various aspects of language processing. This entry describes the general features of connectionist models and highlights specific models from the language sciences to further explain these general features. The benefits of using connectionist models are also described.