Towards a theory of affective mind: Computationally modeling the generativity of goal appraisal.

摘要

A cognitive theory of how humans evaluate the goodness or badness of situations is formulated and tested using data from human participants as well as computer modeling. A specific problem of cognitive appraisal theory, goal appraisal, is formulated: how do people cognitively evaluate a situation with respect to a focal goal in the context of assumed background goals? It is shown that cognitive appraisal theory may provide useful details of how humans form beliefs about others' emotions. In other words, cognitive appraisal theory is viewed as an affective theory of mind. Three studies are described. In Study 1, participants generated inferences about the emotions of others. Using these data, a theory of how these appraisals were formed was created. In Study 2, some of the postulates of the theory were empirically verified by asking a second group of participants to rate the believability of items created and manipulated on the basis of the theory. Theory predictions about believability were upheld. In Study 3, a computer model of the theory was created and tested using believability judgments of a third group of human participants. Results indicated that the computer model created affective evaluations that were, in most cases, as believable as human generated evaluations. Desired improvements to the model and the theory were inferred from data showing cases in which computer generated appraisals were not as believable as human generated appraisals. Implications for future research and practice of special relevance to individuals with autism are identified.