Abstract
Stereotype change is simulated in a distributed recurrent network. Although it operates using iterative application of simple learning rules as each new group member is presented, the network can nonetheless mimic both bookkeeping (Rothbart, 1981) and subtyping (Brewer, 1981; Taylor, 1981) patterns of results. It produces these effects through learning of reliable covariations between counterstereotypic units. Advantages and disadvantages of using a distributed recurrent network to model the representation of stereotypes are discussed. Key among the advantages are those relevant to the dynamic nature of these models.
Get full access to this article
View all access options for this article.
