Patterson & Kurtz (2016) Performance pressure and comparison in relational category learning

Image credit: Unsplash

Abstract

An important objective in higher-order cognition research is to understand how relational categories are acquired and applied. Much of the research on relational category learning has investigated the role of within-category comparison opportunities in category acquisition and transfer – guided by predictions from structure mapping theory that alignment leads to highlighting and abstraction of shared relational structure (Gentner, 1983). Recent research has yielded a within-category comparison advantage under the supervised observational learning mode (relative to twice as many single-item trials), but not under the supervised classification mode (Patterson & Kurtz, 2015). In the present study we investigate the role that pressure to succeed at the training task – a critical difference between the two learning modes – plays in the apparent ineffectiveness of learning by comparison within the classification mode. In a 2x2 between-subjects design we crossed two levels of performance pressure (elevated and standard) with two presentation formats (single-item and within-category pairs). The main findings are: (1) a significant interaction showing a negative impact of increased performance pressure for single-item learners, but not for comparison learners; and (2) a theoretically predicted, but empirically elusive effect of comparison over single-item in the classification mode. We conclude that: (1) performance pressure exerts a deleterious effect on relational category learning (in accord with findings in the attribute category literature) that opportunities to compare may compensate for; and (2) pressure to perform does not appear to underlie lackluster comparison + classification performance (relative to observational learning). Further, we offer new evidence on the role that within-category comparison plays in relational category learning.

Publication
In Proceedings of the 38th Annual Conference of the Cognitive Science Society (pp. 2333-2338)
Avatar
John D. Patterson
Postdoctoral Scholar

My research interests include how we learn and represent categories, how we can optimize learning in applied settings, and how we can capture learning through computational models.