Modelling developmental disorders

(from the chapter) In this chapter, we present examples of the use of computational modelling in the study of developmental disorders. In the next three sections, we illustrate three key ideas. The first section considers the use of models for testing the viability of theoretical proposals - in this case, to establish that certain kinds of deficits in the language system are sufficient to produce behavioural impairments across development, such as those found in Specific Language Impairment. This section focuses on the additional detail forced by implementation. The second section considers the role of the developmental process itself in producing the impairments, compared to building a static model and simulating deficits by breaking components of the model. This issue is considered in the context of models of reading development and dyslexia. The third section addresses the behaviour of complex cognitive architectures made up of many interacting components rather than individual systems, and explores the developmental consequences of initial limitations to individual components: Do deficits subsequently spread throughout the system, or can initially normal components serve to compensate for impaired ones, so alleviating deficits later in development? Once more, this example makes reference to dyslexia in reading. These three examples employ methods drawn from two approaches to computational modelling, artificial neural networks (sometimes called connectionism) and dynamical systems modelling. Following these sections, we consider some of the latest models in the field, including those considering the effects of anomalies in reward-based learning and those modelling deficits at the population level rather than just the individual. (PsycINFO Database Record (c) 2012 APA, all rights reserved).