Two competing attentional mechanisms in category learning.

Abstract

We thank Julie Earles for helpful comments on a previous version of this article. Examples of the events used as stimuli in this research are available over the World Wide Web at http://www.psy.fau.edu/chez/ awk/home.html.

Event categorisation and language: A cross-linguistic study of motion

“…One reason to doubt the viability of a ranking comes from a recent eye-tracking study which measured English and Greek speakers' attention to manner and path in ongoing motion events : path (defined as the endpoint of the agent's trajectory) was found to attract more looks than the manner of motion (defined as the vehicle used for locomotion) from both populations. Relatedly, studies in the artificial category learning literature show that English-speaking adults have strong path biases in categorising novel event exemplars (Kersten, Goldstone, & Schaffert, 1998). Even within the word learning literature, studies comparing the potency of manner and path categories have produced inconsistent results: some studies have found that children prefer manner over result when interpreting novel action verbs, at least at the early stages of verb learning (Gentner, 1978), while other studies report that children overwhelmingly prefer to label the outcome rather than the manner of motion (Behrend, 1990;Forbes & Farrar, 1993;Gropen, Pinker, Hollander, & Goldberg, 1991; see also Behrend, Harris, & Cartwright, 1995).…”

A further investigation of category learning by inference

“…This experiment showed that the reason for the inference-learning condition's superiority on the single-feature test was due to the better learning of the queried features, the feature values having been predicted. Even so, the presented features were learned somewhat, at about the same level as in the classification-learning condition (see Kersten, Goldstone, & Schaffert, 1998, for a discussion of attentional mechanisms in category learning). These findings put strong constraints on theories of inference learning.…”

Reflections on Frequency Effects in Language Processing

“…Once you are trained to see the object in that way, that is the way you see it (or at least first see it), and those are the features whose strengths are incremented on each subsequent processing episode. Kersten, Goldstone, & Schaffert, 1998;Goldstone & Steyvers, 2001) presented a detailed analysis of the ways in which attentional persistence directs attention to attributes previously found to be predictive, elaborated a theory of conceptual and perceptual learning based on these mechanisms, and provided a connectionist model of the processes whereby category learning establishes detectors for stimulus parts that are diagnostic. These detectors, once established, bias the interpretation of subsequent objects to be segmented (Goldstone, 2000).…”