Similarity-Based Representation And Rule-Based Representation In The Category Learning

Category learning is the basic ability for human adaptation to the environment, and has an important influence on people’s production and life.Category learning is the basis for human cognition and also as one of the hot issues in the research of cognitive psychology.This paper based on behavioral experiments, revealed the processing mechanism of category learning.The results of this study not only make up the blank of the related theory,but also provide a theoretical basis for promoting the effection of human category learning.Here we review the following important issues related to the category theory, category representation theory and the related research of category representation.we found that the study on the relationship between category structure and category representation is relatively few, simple geometric shapes were always used. Aiming at this problem, this study uses complex materials to explore the effect of category structure to the category representation. Previous experimental evidence show that learning strategy plays a critical role to the category representation we propose that the effects was not same within different category structure.We also claim that mutual effect is exist between different category representation systems in category learning.Here we use complex artificial animal as category learning materials, and according to the type of sample features, divided into rule-based category structure, similar-based category structure and ambiguous category structure. Experiment1, the main purpose was to explore the category representation within the ambiguous category structure. The result was that part of the objects achieve standard accuracy of similarity representation, and other part achieve standard accuracy of rule representation. These findings suggest that both of the two category representation can represent the category. Experiment2, the main purpose was to explore the effects of category learning strategy for category representation within similarity-based and rule-based category structure.The result was that within rule learning strategy, people achieve the accuracy level of similarity representation in the similarity-based category structure, people achieve the accuracy level of rule representation in the simple rule-based category structure, but not the accuracy level of rule representation in the complex rule-based category structure. Within similarity learning strategy, people did not achieve the accuracy level of similarity representation in the similarity-based category structure, people achieve the accuracy level of rule representation in both of the rule-based category structure. These findings suggested that within the proper category structure, learning strategy played an important role to the category representation. Experiment3,the main purpose was to examine how the existing category representation impact on the subsequent category representation. The result was that people acquired similarity representation, then they accepted the rules strategy guide within complex rule category structure, they learning performance was significantly lower than accuracy level of rule representation, but this result was not suit for simple rule category structure. People acquired rule representation, then they accepted the similarity strategy guide within similarity category structure, they learning performance was significantly lower than accuracy level of similarity representation. This suggested that existed-rule categories representation hindered the formation of subsequent similarity category representation. existed-similarity category representation also hindered the formation of subsequent rule category representation.Three experimental results show that the best category representation of different category structures in a different way. Rule category representation is the best representation mode of rule category structure. Similarity category representation is the best representation of similarity category structure. The existed-representation may play a role in the development of subsequent category representation. Rules-based representation and similarity-based representation interacted with each other in the category learning.