Neural Network Models and Mechanisms of Strategy Development

Agent computing themes in biologically inspired models of learning and development

“…Study of external memory strategy (EMS) learning and use combines experimental and computing elements. Bray et al (1997Bray et al ( , 1998a and Grupe (2001) cover experiments and some models for object placement tasks, our main emphasis, and arithmetic (addition), mentioned occasionally. A considerable amount of modeling, not in these papers, is covered here and in other references (cited in the following sections).…”

“…It is believed to be more effective to explore accuracy model parameters, such as neuron-to-neuron activity transfer and neuron decay rates (α and δ, respectively). Table 1 provides a (small) sample of the results, reporting these two parameters for data like some found in Bray et al (1997); these data are based on 36 subjects, each performing over 32 trials. The probabilities attached to the Chi-squared values are based on ndf = 4, the fraction of correct responses in each class being independent.…”

“…As inhibition enters a pool it produces a signal enhancement effect, appropriate to the lateral inhibition it introduces into the model. The enhancement effect (Bray et al, 1997;Reilly and Bray, 2000a,b) provides higher differences in activation for an instruction's assigned items relative to other items in the pools, which compete with the correct items and potentially produce errors. These activation differences, which grow with improved (modeled) EMS levels, can be exploited to achieve a model for increments in accuracy due to strategy (choice) improvements, with a base accuracy treated as a displacement from model activation levels.…”

Models of atypical development must also be models of normal development

“…Connectionist modeling plays a crucial role in the arguments put forward by T&K-S. As highlighted in the target article, connectionists have recently begun to model various developmental disorders, including dyslexia Brown & Loosemore 1995;Harm & Seidenberg 1999), autism O'Laughlin & Thagard 2000), selective language impairment (Hoeffner & McClelland 1993), Williams syndrome (WS) (Thomas & Karmiloff-Smith, in press), mental retardation (Bray et al 1997), and schizophrenia (Hoffman & McGlashan 1997). However, in a recent review of connectionist modeling of cognitive developmental disorders, we found that most models suffer from a variety of shortcomings (Conway et al, in preparation).…”

Can Connectionism Model Developmental Change?

“…One area where the interplay between models derived from empirical research and c-nets has been particularly fruitful is in the area of memory which receives little attention in Rethinking Innateness. Bray, Reilly, Villa and Grupe (1997) describe a research programme which has attempted to model the development of memory strategies in typically and atypically developing children, using cnets. They highlight some of the special challenges that arise in connectionist modelling of developmental phenomena.…”