Please use this identifier to cite or link to this item:
http://hdl.handle.net/11375/6746
Full metadata record
DC Field | Value | Language |
---|---|---|
dc.contributor.advisor | Brooks, L. R. | en_US |
dc.contributor.author | Wood, Timothy J. | en_US |
dc.date.accessioned | 2014-06-18T16:36:45Z | - |
dc.date.available | 2014-06-18T16:36:45Z | - |
dc.date.created | 2010-06-02 | en_US |
dc.date.issued | 1998-03 | en_US |
dc.identifier.other | opendissertations/2055 | en_US |
dc.identifier.other | 2845 | en_US |
dc.identifier.other | 1339901 | en_US |
dc.identifier.uri | http://hdl.handle.net/11375/6746 | - |
dc.description.abstract | <p>When classifying a novel object, people often rely on similarity to previously learned instances to help them identify the category of the object. People also rely on more analytic knowledge, like classification rules, to identify the category of a novel object. The coexistence of two different procedures that could be used to classify the same novel object. The coexistence of two different procedures that could be used to classify the same novel object raises the issue of coordination; that is, what is the relation between rule-based and similarity-based classification. Learning a new category is much like acquiring any cognitive skill, therefore it is hypothesized that current theories of skill acquisition should provide a useful framework for studying the relation between rule-based and similarity-based classification. A popular view of skill acquisition, Logan's Instance Theory (1988), suggests that skilled performance can develop rapidly, possibly after only a few practice trials on specific instances. The underlying basis of this rapid skill development is attributed to a transition from reliance on slow analytic procedures to reliance on faster retrieval-based procedures. Given what is known about memory, this transition to a retrieval-based classification procedure makes a great deal of intuitive sense and should account for the faster, more efficient classification performance that is characteristic of highly practiced individuals. The experimental conditions were designed to test an extreme example of this transition from rule-based to retrieval-based categorization. If a transition to a fast, retrieval-based procedure is a reasonably automatic result given sufficient practice, then it should occur even when participants are given a classification rule that is simple, perfectly predictive, and easy to apply. Although there was some evidence of participants relying on similarity to prior instances for classification, this reliance on similarity never approached the levels that would be expected if participants had abandoned the rule. In fact, a reliance on similarity was limited to novel stimuli that were so similar to the training stimuli that they were actually falsely recognized as old. When people realized a novel stimulus as new, they relied on the classification rule. This "retrieve if believed old, rule if believed new" strategy held across a number of manipulations all designed to facilitate an increased reliance on similarity including extensive practice, training with multiple similar neighbors, speeded classification, and reducing the novelty cues present witthin the transfer stimuli. Continued reliance on rules to classify novel items believed to be new and false recognition of novel items believed to be old raises questions about the role of similarity and the assumptions that a transition from rules to retrieval is a relatively automatic occurrence of skilled classification.</p> | en_US |
dc.subject | Psychology | en_US |
dc.subject | Psychology | en_US |
dc.title | On the rules-to-episodes transition in classification: Generalization of similarity and rules with practice | en_US |
dc.type | thesis | en_US |
dc.contributor.department | Psychology | en_US |
dc.description.degree | Doctor of Philosophy (PhD) | en_US |
Appears in Collections: | Open Access Dissertations and Theses |
Files in This Item:
File | Size | Format | |
---|---|---|---|
fulltext.pdf | 6.6 MB | Adobe PDF | View/Open |
Items in MacSphere are protected by copyright, with all rights reserved, unless otherwise indicated.