Attitude generalization: Similarity, valence, and extremity

Attitude generalization was explored as a function of object similarity and attitude valence and extremity. Participants in a computer game formed attitudes toward positive and negative, mild or extreme stimuli. How well these attitudes generalized to similar, novel stimuli was then examined. Visual similarity to game targets affected categorization of novel stimuli, such that greater resemblance resulted in more similar classification. However, generalization varied by valence and extremity. Negative attitudes generalized more than positive attitudes, requiring less resemblance for a novel target to be classified as negative. This pattern was more obvious with extreme attitudes than mild attitudes. That is, extreme attitudes were more influential and given more weight than mild attitudes. Also, specific conditions were identified under which positive attitudes proved more influential than negative attitudes.