The effect of attitude towards computer generated faces on face perception

Abstract

Attitudes towards computers have been becoming more negative over time. Earlier research has shown an unwillingness to cooperate with computers once their nature is revealed. Correspondingly, computer generated faces are perceived as less trustworthy than natural faces. This effect can be explained by either the synthetic appearance of the computer generated faces, or by a general bias towards artificial faces. Merely knowing that a face is computer generated could elicit a lower trustworthiness rating due to social categorization. Therefore, the aim of the current study was to examine whether the belief that a face is computer generated can impact how trustworthy and attractive it appears, using two experiment. In the first experiment, participants were presented two sets of natural faces and asked to rate these on trustworthiness and attractiveness. One of the sets was labeled computer generated and the other natural. In the second experiment, attitude towards computer generated faces was manipulated beforehand by presenting either a positive or a negative text before data collection. The results of both experiments show that merely labeling a face as computer generated is enough to make it appear less trustworthy but not less attractive. However, providing a positive or negative text before the task did not influence the ratings. It is implied that social categorization is involved in the discrimination between computer generated and natural faces.