Adaptive Modelling of Social Decision Making by Affective Agents Integrating Simulated Behaviour and Perception Chains Alexei Sharpanskykh and Jan Treur Vrije Universiteit Amsterdam, Department of Artificial Intelligence De Boelelaan 1081, 1081 HV Amsterdam, The Netherlands http://www.few.vu.nl/~{sharp,treur} {sharp, treur}@few.vu.nl Abstract. It is widely recognized that both cognitive and affective aspects play an important role in human decision making. In most recent approaches for computational modelling of affective agents emotions have a cognitive origin. In contrast to these approaches, the computational social decision making model proposed in this paper is grounded on neurological principles and theories, thus providing a deeper level of understanding of decision making. The proposed approach integrates existing neurological and cognitive theories of emotion in a decision model based on evaluation of simulated behaviour chains. The application of the proposed model is demonstrated in the context of an emergency scenario. Keywords: Social decision making, affective, cognitive, simulated behavioural chains, neurological modelling. 1 Introduction Traditionally, human decision making has been modelled as the problem of rational choice from a number of options using economic utility-based theories [17, 18]. In the last decades such approaches were criticized by many authors [18] for the lack of realism and limited applicability. In particular, it is imputed to the traditional decision making modelling methods that the role of human cognitive heuristics and biases, and affective states is totally neglected. Much evidence exist [1, 4, 5, 10] that affective states have a significant impact on a human’s decision making. However, computational models to explain this evidence are rare. In this paper the focus is on the integration of cognitive and affective aspects in a computational social decision making model which is grounded in neurological theories. In the areas of Artificial Intelligence and Cognitive Science a number of computational models of decision making with emotions have been developed [9, 21, 22], which use variants of the OCC model developed by Ortony, Clore and Collins [19] as a basis. The OCC model postulates that emotions are valenced reactions to events, agents, and objects, where valuations are based on similarities between achieved states and goal states; thus emotions in this model have a cognitive origin.