A computational model of human emotion

TR Number



Journal Title

Journal ISSN

Volume Title


Virginia Tech


To date, few computer programs have been constructed to express or understand human emotional states. None of these programs can detect a large spectrum of emotions from emotional situations, nor do any of them express emotions over more than one temporal dimension. A number of these programs used representations with domain restrictive structures. We have constructed a computational model of emotion that detects twenty-eight emotions and is non-domain specific. It reports emotional episodes, moods, and dispositions. The psychological theory behind the model draws extensively on the ideas developed in the book “The Cognitive Structure of Emotions,” by Ortony, Clore and Collins [Ortony et al. 1988]. To test the model, we implemented a contextual front end to the system to provide input data. The domain chosen was doctor/patient interaction scenarios. Because our model is domain independent, any context could have been chosen, and other such contexts are explored in this paper. Our model demonstrated its ability by detecting a considerable range of appropriate emotions from the test case.