Modeling the Shepard, Hovland, and Jenkins (1961) categorization task using configural memory with network reinforcement learning

Loading...
Thumbnail Image
Authors
Veksler, Vladislav D.
Issue Date
2006-05
Type
Electronic thesis
Thesis
Language
ENG
Keywords
Cognitive science
Research Projects
Organizational Units
Journal Issue
Alternative Title
Abstract
Categorization is central to cognition. Current work proposes a categorization-based cognitive architecture with distinct declarative and procedural categorization components. The declarative component is a configural memory network (Gluck & Bower, 1988; Heydemann, 1995). The procedural component is a specialized reinforcement learning mechanism - Network Reinforcement Learning (NRL). The Shepard, Hovland, and Jenkins (1961) benchmark categorization experiment is used to lay the groundwork for this architecture. NRL is the only active learning mechanism (no configural learning) used to explain data from this experiment. Two free parameters (the values of positive and negative reinforcement) were varied to find the best fits to psychological data from six versions of the Shepard et al. task, including human supervised and unsupervised category learning (Love, 2002), and category learning in rhesus monkeys (Smith, Minda, & Washburn, 2004).
Description
May 2006
School of Humanities, Arts, and Social Sciences
Full Citation
Publisher
Rensselaer Polytechnic Institute, Troy, NY
Terms of Use
Journal
Volume
Issue
PubMed ID
DOI
ISSN
EISSN