A gate formulation of emotional learning as a model of cognitive processes

Abstract

With establishment of computer age, interest towards cognitive processes has greatly increased, and comprehension of cognitive phenomena becomes extremely important and of dual value, for not only does it elucidate how human mind works, but also opens up new possibilities for efficient computer performance based on brain analogy. As such, this work suggests an alternative and fresh view to the paradigm of stimulus driven emotional learning, as part of a more integral field of machine learning, and provides a solid ground for further research in the area. Within the framework of this thesis, a comprehensive background on neurophysiology of emotional behaviour is presented, after which a computational model is suggested, where the actions of stimuli are represented as matrices acting on agent’s state vector. The model is then validated against several classical conditioning experiments. Finally, ways of further development are indicated and the conditioning phenomena not covered yet by the model are listed.