Arquitecturas emocionales en inteligencia artificial: una propuesta unificadora

Abstract

Emotions can be explained in different ways depending on the issues we consider (psychology, neuroscience, philosophy, etc). In particular, from an artificial intelligence viewpoint, studying emotions is searching which structural conditions corresponds to emotional states and which functional processes underlie emotional behavior in living beings, especially in humans, with the aim of reproducing them in artificial architectures. This paper sets out a review of the main architectures in Artificial Intelligence and explains where are the errors and constraints in classical models: basically, a dualistic approach to understand cognition that has considered emotions and reasons as antagonists. In the second part of the paper, is introduced the recent interest that researchers show on works about emotions that neurophysiology are involved in. In the next section it is proposed a complex architecture that allows implementing some neurophysiological features with results that generate important expectations. Finally, the paper concludes with a discussion about emotions and Artificial Intelligence, beyond issues of engineering, trying to understand what it means for humans to have emotional behaviours from an adaptive approach.