Behaviorist Agent Architecture

2013 IEEE/WIC/ACM International Joint Conferences on Web Intelligence (WI) and Intelligent Agent Technologies (IAT) |

Publication | PDF

In this article we propose an agent architecture based on Behavior Analysis, a behaviorist psychology theory. The main characteristic of this theory is the description of complex behavior exclusively in terms of stimulation and behavioral responses. That is to say, in terms of observable and controllable environmental events, instead of the internal mental mechanisms typically found in other approaches to psychology. In this way, not only trivial reflexive actions, but also a number of sophisticated phenomena can be effectively modeled, such as learning and emotional states. Our work, which we call the Behaviorist Agent Architecture, provides a computational framework with which to build agents according to these principles. It accounts for five aspects of agent behavior in an integrated manner: (i) stimulus conditioning, (ii) reflexes, (iii) operant behavior, (iv) drives, and (v) emotions. The objective of this text and of the architecture itself is to show that a consistent and coherent behaviorist point of view may be useful, despite its current lack of popularity. Here we present our overall rationale and the main ideas underlying our agent architecture, including illustrative examples.