Executive control of cognitive agents using a biologically inspired model architecture of the prefrontal cortex

Prefrontal cortex (PFC) is implicated in executive control, herein defined as the ability to map external inputs and internal goals toward actions that solve problems or elicit rewards. In this work, we present a computational model of PFC and accompanying brain systems (e.g., limbic, motor control, and sensory areas) based on a review of structure and function. The current design leverages previous models, but organizes them in novel ways to provide transparent and efficient computation. We propose this model provides a biologically plausible architecture that learns from and uses multimodal spatio-temporal working memories to develop and refine reward-eliciting behaviors. It addresses several anatomical and physiological constraints consistent with neurophysiology. The functional aspects of the model are illustrated using two example studies: the “Egg Hunt” and “Widget Factory” scenario. These examples illustrate the flexibility of the proposed model to assist cognitive agents to become adaptive during their interaction with their non-stationary environments.