基于规则和基于相似性的分类学习在复杂视觉分类中的比较

A Comparison Between Rule-Based and Similarity-Based Processes in Complex Visual Categorization

采用同一复杂视觉分类任务,通过比较对基于规则的和基于相似性的分类学习异同,探讨了2种不同分类过程之间的相互关系,结果发现:在不同感知类型的刺激项目上,基于相似性的分类成绩都明显高于基于规则的分类成绩.在基于相似性的分类学习模式下,不同感知难度的刺激正确率之间都存在显著差异;在基于规则的分类学习模式下,部分类型的刺激正确率之间不存在明显差异.实验结果表明基于规则的和基于相似性的2种不同加工过程同时存在于同一分类过程中并且互相作用,研究支持了多系统理论模型.

In this paper,a complex visual categorization task is designed to investigate the relationship between rule-based and similarity-based processes.The results revealed that the similarity-based group exhibited a better performance than the rule-based group in each type of image matrix and that significant differences existed in accuracy between different types of image matrix under similarity-based condition whereas no significant differences were detected between some types of image matrix in their accuracy under rule-based condition.These results indicated that rule-based and similarity-based processes are two different processes which can operate simultaneously in one task,thus providing a new support to the theory of the multiple system model.