Deviations in the emergence of representations: a neuroconstructivist framework for analysing developmental disorders

A common way of studying developmental disorders is to adopt a static neuropsychological deficit approach, in which the brain is characterized in terms of a normal brain with some parts or 'modules' impaired. In this paper we outline a neuroconstructivist approach in which developmental disorders are viewed as alternative developmental trajectories in the emergence of representations within neural networks. As a concrete instantiation of the assumptions underlying this general approach, we present a number of simulations in an artificial neural network model. The representations that emerge under different architectural, input and developmental timing conditions are then analysed within a multi-dimensional state space. We explore alternative developmental trajectories in these simulations, demonstrating how initial differences in the same parameter can lead to very different outcomes, and conversely how different starting states can sometimes result in similar end states (phenotypes). We conclude that the assumptions of the neuroconstructivist approach are likely to be more appropriate for analysing developmental deviations in complex dynamic neural networks, such as the human brain.