The Mood and Memory of Believable Adaptable Socially Intelligent Characters Mark Burkitt and Daniela M. Romano Department of Computer Science, University of Sheffield, Regent Court, 211 Portobello Rd, Sheffield, S1 4DP, United Kingdom. {acp07mab,D.Romano}@sheffield.ac.uk Abstract. In this paper a computational model for believable adapt- able characters is presented, which takes into account several psycho- logical theories: five factors model of personality[1], the pleasure arousal dominance[2] and the social cognitive factors[3] to create a computation model able to process emotionally coded events in input, alter the char- acter’s mood, the memory associate with the set of entities connected with the event, and in the long run the personality; and produce an im- mediate emotional reaction in the character, which might or might not be displayed according to the social cognitive factors, the goal of the character and the environment in which the event is taking place. 1 Introduction Virtual characters are increasingly being used in different applications to enhance a user’s experience and to increase the perception of realism. One way this can be achieved is through elicitation of social presence. Biocca[4] defines social presence as the feeling that you are interacting, through mediated communication, with an intelligent being. In this context, the measure of social presence is the degree to which the user is able to perceive the intelligence of another[4]. Various psychological theories have been combined together to obtain a com- putational model able to process emotionally coded events in input and gener- ate a reaction, which might or might not be displayed, according to the mood, memories, personality and social cognitive factors of the character. Roseman & Smith[5] suggested that the judgment (appraisal) we make about ourselves or the world are what causes an emotion to arise. They propose an adaptive di- mension, where for example if we appraise a situation as being negative, but there is not relevant action that can be taken, we experience sadness (pas- sive reaction); instead if there is a relevant action that might be of benefit, we experience anger (active reaction). Computational models of emotion like EMA[6] are concerned with modelling such appraisal of the world. Scherer at al.[7] report that the more recent literature conceptualises emotional states as two-dimensional models where a pleasantness (or “valence”) dimension is added to the arousal dimension, or positive affect and negative affect are conceptualised as two fundamental dimensions of emotional experience, with the PAD[2] being