ERP Correlates of Prototype Representation during Category Learning

Ragó, A. 1 , Czeto, M. 1 , Varga, M. 2 & Somos, E. 3

1 Cognitive Psychology Department, Eötvös Loránd University, Budapest, Hungary
2 Cognitive Science Department, Budapest University of Technology and Economics
3 Department of Psychology, University of Hull

Our aim was to test the emergence and stability of prototype abstraction in an implicit information-integration task using complex stimuli consist of individually identifiable exemplars following family resemblance structure.
In a supervised learning paradigm far-from-prototype (FP) exemplars were presented. In the immediate test phase prototypes, close-to-prototype (CP) exemplars, and new FPs were shown. We also presented few training FPs to test exemplar effect.
Hits, reaction times were recorded, and event related brain activity was registered from 32 electrode sites. Test 1 was repeated a week later.
Results show that subjects successfully learned to separate the two categories without an explicit knowledge of any rules. Prototype abstraction was demonstrated by the hit rates, and reaction times. Hits significantly decreased by the distance from the prototypes. Training FPs got as low hit rates as new FPs. Inversely, reaction times were lower with typicality. The rule based representation was also stable a week later. Familiarity has an effect to the early components as P1 was bigger to the training FPs, while the N1 was bigger to the prototype. This effect was also shown in the later components.
Our results confirm that specific information is overwritten by the generalized knowledge of category structure.