D. Mareschal et Tr. Shultz, GENERATIVE CONNECTIONIST NETWORKS AND CONSTRUCTIVIST COGNITIVE-DEVELOPMENT, Cognitive development, 11(4), 1996, pp. 571-603
This article presents a novel computational framework for modeling cog
nitive development. The new modeling paradigm provides a language with
which to compare and contrast radically different facets of children'
s knowledge. Concepts from the study of machine learning are used to e
xplore the power of connectionist networks that construct their own ar
chitectures during learning. These so-called generative algorithms are
shown to escape from Fodor's (1980) critique of constructivist develo
pment We describe one generative connectionist algorithm (cascade-corr
elation) in detail. We report on the successful use of the algorithm t
o model cognitive development on balance scale phenomena; seriation; t
he integration of velocity, time, and distance cues; prediction of eff
ect sizes from magnitudes of causal potencies and effect resistances;
and the acquisition of English personal pronouns. The article demonstr
ates that computer models are invaluable for illuminating otherwise ob
scure discussions.