[
Back to list of talks
]
Extending a Computational Model of Transitive Inference to Novel Item Effects in Children and Birds
Will Shown and Thomas Shultz
Abstract:
A connectionist model using a cascade correlation algorithm which already displayed similar performance to that of humans on 6 common phenomena in transitive inference tasks was extended to cover past and recent findings on novel item sorting. The networks which the model produces show error effects which are analogous to data gathered from children and pinyon jays while also predicting two possible transitive inference phenomena in pinyon jays.