Home

About

Advanced Search

Browse by Discipline

Scientific Societies

E-print Alerts

Add E-prints

E-print Network
FAQHELPSITE MAPCONTACT US


  Advanced Search  

 
Copyright 2004 Psychonomic Society, Inc. 1318 Perception & Psychophysics
 

Summary: Copyright 2004 Psychonomic Society, Inc. 1318
Perception & Psychophysics
2004, 66 (8), 1318-1340
Traditionally, category learning has been investigated
by examining how response accuracy changes with ex-
perience. Often, such data are presented in the form of a
learning curve, which plots proportion correct against
trial or block number. Learning curves are a good non-
parametric method for investigating category learning,
because no model needs to be specified during their con-
struction. Learning curves are also relatively simple to
compute and often provide an effective method for com-
paring task difficulty across different conditions of an
experiment (e.g., Shepard, Hovland, & Jenkins, 1961).
On the other hand, the use of learning curves to test
among competing models is severely limited because, in
most cases, a variety of different models will be capable
of predicting the same learning curves. For example, a
popular assumption of many category-learning models is
that the trial-by-trial learning of categories is accom-

  

Source: Ashby, F. Gregory - Department of Psychology, University of California at Santa Barbara

 

Collections: Biology and Medicine; Computer Technologies and Information Sciences