The Influence Of Stimulate Description And Distribution On The Category Representation

In this paper, on the base of the former studies about category learning theory and representation of theory, discusses the previous category learning research. Based on multiple category learning system theory, analysis the way of representation of rule-based category learning and information-integration category learning which under the different descriptions and different distributions.The hypothesis of experiment 1 is that rule-based category learning, no matter what kind of description, process by rules representation. Well under the information-integration category structure, words describe present group use the rule representation, and pictures describe present group use the similarity representation. The results is that rule-based category structure there is no significant difference in accuracy and reacting time between words describe present group and pictures describe present group. Information-integration category structure there is significantly different in reacting time between words describe present group and pictures describe present group. The results indicate that under the rule-based category structure the classification rules can be learned through the hypothesis-test,both words describe present group and pictures describe present group can master rules and do the classification by rule representation. Information-integration category structure subjects in words describe present group were out of similarity information and do the classification by rules and pictures describe present group do the same way as words describe present group, cause the uniform distribution lessen the similarity between the stimulus which belong the same category.Subjects can not find the best rule judgment or compare the perception similarity degree, so they can only choose rules representation.Experiment 2 focus on discussing the representation of information-integration category structure with the distribution of the stimulus changed. The hypothesis of experiment 2 is that increased similarity between the stimulus which belong the same category, subjects in words describe present group can not do the classification through similarity comparison; will try to through the hypothesis test form classification rule. Because the rule which only achieve single dimension can only reach the accuracy by 50% and the best rule can not be describe by words, subjects can master the second-best rule of classification which will need to take more reaction time. Subjects in pictures describe present group will also try to achieve classification rule by hypotheses-test but not because of high accuracy to classify through comparison similarity judgment. Judging by similarity requires less reaction time. The results that there is no significant difference in accuracy and reacting time between words describe present group and pictures describe present group and the different in response time is significantly. The results show that pictures describe present subjects compare the similarity between the stimulus, use similarity representation direct the classification, words describe present subjects through hypothesis test form the rules, use rule representation direct the classification.Different categories structure in the category learning through different ways of representation, stimulate distribution and stimulate description both affect the category learning. The rule-based category learning is rule representation both in words describe group and picture describe group. Well the information integration category learning, if the stimulate evenly distributed in a larger area range within the categories, the low degree of similarity stimulus, learners tend to through hypotheses-test and form the rules of classification. If the stimulate linear distributed in smaller area within the categories, picture describes present tend to use similarity representation and words description present tend to use rules representation.