A Web of Emotions

The fast growing technology of agents and the rise of motional intelligence opened the doors to many new and inter es ing ideas. Among these ideas, one that has captured the intere st of many researchers within the AI community is the developm ent of emotional agents. While research on emotional agent s is in a very early stage, many models have been developed to sim ulate emotions in agents. Some of the most recent computa tional models of emotions attempt to account for the inevi table links between emotions and many other processes residing in the mind, such as planning, cognition, personality, and motiv ation. Most of these models, however, simulate an incomplete pictu re of the links involved. Furthermore, most of these models t end to simulate emotional triggered expressions or behavio rs according to one emotion, which is an invalid assumption, sin ce emotions, realistically, occur in a mixture. We have taken a relatively new approach in emotional agents’ development. Rather t han defining the emotional process as a single independent proce ss, we define the emotional process as a web whose different emot ional states act as strings that pass through many other process es or states affecting or becoming affected by them. Thus, inste ad of triggering emotions and then defining a behavior ac cording to one emotion, we use other processes, such as learning a d experience, to trigger mixtures of emotions. These mixtures the n propagate through the web affecting or becoming affected by o ther processes and states, including motivations and behaviors. In this paper, we describe the design and implementation of the emoti onal web. Additionally, we demonstrate empirically through a computer simulation of a pet that simulating the link betwee n emotions and learning is crucial to users’ assessments of the be li vability of an agent’s behaviors.