Combining cognition and emotion in virtual agents

Purpose The paper aims to explain the limitations of existing cognitive architectures and affective models, and propose a new cognitive-affective architecture that can be integrated in real intelligent agents to make them more realistic and believable. Design/methodology/approach The paper evaluates the state of the art, and describes the design and implementation of the cognitive-affective architecture in an agent. A brief evaluation of the agent is provided. Findings The paper clearly states that it is possible to use cognitive architectures to help, but there is a lack of architectures that address the problem of combining cognition and emotion in agents in a unified, simplified way. A cognitive-affective architecture is useful to make believable intelligent agents in an easier way. Research limitations/implications The paper does not explore a lot of possible future work that can be done to extend the emotional expressions of the agent, as well as including direct emotional-sensing capabilities in real time. Practical implications The paper argues about the need to include cognitive-affective architectures in modern intelligent agents. The architecture allows to influence and modify the behavior of the agent in real time, to achieve a more realistic and believable interaction with the user. Social implications The paper remarks the importance of a cognitive-affective architecture that makes intelligent agents able to help the users in different tasks and environments. Originality/value The paper describes a new cognitive-affective architecture and its utility for modern intelligent agents. This is proven by including it in a previous agent, which boosts its behavior and emotional expression possibilities and thus improves user experience.