MIT CogNet, The Brain Sciences ConnectionFrom the MIT Press, Link to Online Catalog
SPARC Communities
Subscriber : Stanford University Libraries » LOG IN

space

Powered By Google 
Advanced Search

 

An fMRI Study of Expert and Novice Categorization

 Corinna Cincotta and Carol A. Seger
  
 

Abstract:
Abstract: We examined differences in brain activation between expert and novice concept learning by comparing a well-learned and a novel classification task within-subjects. The stimuli used in each task were defined according to Ashby & Maddox (1993), and were simple visual figures (rectangles; intersecting lines) that could vary along two dimensions (height and width; line length and angle between lines). For each task, prototypes were formed for categories A and B. Exemplars were formed by varying the dimensions according to a normal distribution with a fixed standard deviation. Prototype mean feature values and standard deviations were chosen so that participants learning an ideal decision bound between categories would be able to correctly classify stimuli with an accuracy of 80%. For the Well-learned concept, participants were trained to criterion (2 consecutive blocks of >75% accuracy) prior to scanning. During scanning, participants classified exemplars from the Well-learned and Novel tasks in separate blocks alternating with a simple baseline task. With respect to baseline, both Well-learned and Novel classifications activated bilateral inferior parietal lobes and bilateral fusiform gyri. The Well-learned classification activated left prefrontal areas, whereas Novel classification activated bilateral prefrontal areas. These results are consistent with those of Seger et al. (in press), who associated right frontal activation with processing of novel exemplars early in classification training, and associated left frontal activation with successful learning.

 
 


© 2010 The MIT Press
MIT Logo