Psychological scaling with method of fuzzy categories and its experimental verifications

Abstract

This paper aims to discuss a method for assigning numbers to subjective degrees that are measured with verbal categories such as "very tall". We call this method "method of fuzzy categories" and derive it from the model for subjective rating process published by Yoshikawa et al. (1991). This method makes the psychological scale value for each category from the membership function of linguistic truth value that has the same verbal hedge as the category has. This method is appropriate to handle vagueness of categories, difference of meaning between persons and effects of context. We carry out psychological experiment to verify this method. Nineteen random dots patterns are used as stimuli. Eight male subjects rate subjective numerousness of those patterns with method of fuzzy categories and fuzzy graphic rating scale simultaneously. The results obtained are : First, the data obtained by two methods are equal in degree of matching between two fuzzy set and degree of linearity. Next, the data by method of fuzzy categories satisfies two axioms of interval scale. Therefore those results show that this method works as phychological scaling.