Bio-inspired task-rule retrieval model with auditory sorting test

Abstract

Cognitive architectures (CAs) are currently used to bring the behavior of computer systems closer to human behavior. One of the main capacities of humans is the ability to plan and make decisions. Thus, part of the human behavior is based on rules associated with the relevant environmental stimuli. Rule management can be divided into six needed processes: rule learning, rule retrieving, rule coding, rule updating, rule reinforcement, and rule changing. A key aspect of rule processing is retrieval, which involves the use of information extracted from memory. This work deals with how rules are stored in coded form in the brain, retrieved and used as the need arises. The proposed model takes inspiration mainly from the processes in the VLPFC and MTL brain areas to extract rules from memory and create a rule-set that is sent to the DLPFC. The DLPFC and VLPFC prefrontal area operations are also described, and a process is proposed to select the appropriate rule and give a response. The experimentation of the implementation of our proposal with different configuration parameters gives rise to different approaches to human behavior in rule retrieval. Our conclusion is that after a virtual entity is endowed with this proposal, it computes in a way similar to human behavior.