Computational model for affective processing based on Cognitive Sciences: An approach using deterministic finite automata’s and temporal heterogeneity

Abstract

Cognitive architectures represent an alternative in the quest to develop general purpose artificial intelligence, for which cognitive sciences are studied. In this work we will focus on modeling affective processing, an important component to enable basic emotional capabilities. This component was developed with the aim of generating affective responses in the presence of stimuli, necessary to feed a basic emotion model already proposed within our research group. For the proposal we used a layer-based model with Deterministic Finite Automata’s (DFA) to process stimuli along the time, which works as structures to store and represent stimulus–response associations. This approach provides an independent component, contrary to the proposals commonly seen in the state of the art, where this process is often embedded in the feelings and emotions calculations. This model was tested to respond to the sounds consonance, showing that is capable to provide and reinforce responses for specific stimuli features. The results obtained show that the model is capable of making associations between the encoded stimuli and the expected responses, taking advantage of the fact that it is not necessary to be trained to identify stimulus patterns but only to learn to respond to them.