PURPOSE: It has been demonstrated that edge-based information is more important than surface-based information in object recognition. However, few studies have investigated the role of the two types of information in incidental category learning.To address this issue, the present study adopted the ERP technique and defined the categories either by four edge-based features or four surface-based features.
METHODS: The stimuli were cartoon animals that varied along 10 binary dimensions, with five in contour and five in surface. On each trial, a visual stimulus and an auditory sound was presented simultaneously and participants were asked to rate how likeable the animal and sound were in the training phase. Then, they were told that the animals can be divided into two categories according to the sounds and were instructed to classify the novel stimuli in the test phase.
RESULTS: The behavioral results showed that people could learn both categories simultaneously, and the accuracy was higher for the edge-based category than for the surface-based category. The ERP results revealed that the anterior N1 peak amplitude was more negative for the surface-based category than for the edge-based category, while the posterior N1 peak amplitude was larger for correct trials than for incorrect trials. Importantly, for correct trials, stimuli in the edge-based category elicited large anterior P2 and posterior P3 than that in surface-based category.
CONCLUSIONS: The results suggest that although surface-based information might attract more attention during the feature detection, edge-based information plays more important roles in the evaluation of the relevance of the information in making a decision in incidental category learning.