What Are Emotions For? Commitments Management and Regulation Within Animals/Animats Encounters

The hypothesis that emotional mechanisms are wired in the repertoire of higher species is not as new as is its impact on the nature of autonomous agents underlying animats' design. In an evolutionary perspective, it looks as if, the more rational a species gets to be, the more emotional it happens to be as well. Thus, the study of emotional behavior raises the question of adaptiveness: what are emotions for, in terms of benefits for the individual? This paper attempts to answer such questions by proposing a model that links autonomy to motivation, to resource management and to emotions. The model distinguishes two kinds of resources: first-order ones, directly consumable, and second-order ones whose access is only guaranteed by a mediating agent, committed to give them. The model defines needs (such as those associated with states of hunger or thirst) as motivational processes insuring management and regulation of directly consumable first-order resources, while emotions (such as anger or fear) appear as motivational processes insuring management and regulation of second-order resources - namely commitments - within complex species for which they become the most important of all resources, and the key for achieving multiple goals. The paper concludes that emotions, like needs, are powerful control structures that have been designed in to protect agents and societies of agents from running out of resources. Such mechanisms meet some essential requirements that autonomous agents are faced with.