Ordered Information and Word Learning: An Associative Learning Perspective

Abstract

A computational associative learning simulation is implemented to gauge the role of temporally ordered information in regards to establishing differently weighted representations. Simulation comparisons are made between two approaches. One approach simultaneously learns multiple sets of cues, which are trained to be associated with their respective outcomes. The other approach separates training into temporally distinct early and late phases. Qualitative differences in weighted representations are observed between the two types of training. Asymmetrical learning of features occurs in the simulation with temporally distinct learning instances, and not in the simultaneous learning model. Negative associations are learned for specific cues when competing information is introduced at a later time point. These findings provide discussion about the influence of order effects in terms of word acquisition by attending to relevant features over time, and in general, the role cue competition in relation to temporal learning during early development.


Back to Table of Contents