Modelling child memory

C. Philip Beaman & John Morton

This paper describes an implementation of a memory-updating theory of 3- and 4-year olds' performance at false belief tasks put forward by Morton (in press). The cognitive architecture used is adapted from one that has already been shown to be successful in accounting for the abilities of children to update the spatial locations of objects (Barreau, 1996). The model demonstrates that the failure of children under the age of 4 to succeed at a specific false belief task, the smarties task (Hogrefe, Wimmer & Perner, 1986), can be accounted for by a process of destructive updating. Further elaboration of the model also shows how, at a particular developmental stage, children can pass the smarties task if they can be induced to create a durable record of the smarties ``event'' prior to testing.

Integrating interactive activation into COGENT Workshop 2 Bayesian models of medical diagnosis