ThuPIS: A new affective image system for psychological analysis

Abstract

Traditional evaluations of human psychological status usually have a great dependence on the experience of the operator, and it is still a challenge to automate the evaluation process. In this study, a new image system ThuPIS (the Tsinghua Psychological Image System) is designed based on Minnesota Multiphasic Personality Inventory (MMPI), a classical personality inventory in clinical mental health diagnose. The goal of ThuPIS is to build an ontology of affective images, which can active difference of physiological reactions among people with different psychological status. Specific image contents were chosen according to the items of MMPI, and organized by a hierarchy structure. Images were rated according to an overall impression, and those with affective ambiguous were eliminated. An experiment was conducted based on ThuPIS, and the results provide evidence that images in this image system could lead to difference in response time of positive facial expression when participants hold different attitudes towards the background images. This new image system could also supplies materials for other studies on human cognition. It is a brand new work and more practical assessments on the new image system are expected along the way.