A motivational model of hunger for a cognitive architecture

Abstract

This paper presents a bio-inspired motivational model to generate, maintain and dismiss needs, specifically related with hunger. Because we want to obtain results nearer as possible to those of humans, the components of the model are developed based on suggested evidence from cognitive sciences and neurosciences. This model could be implemented in virtual creatures to induce behaviors, as we show in a simulation where the motivational part is implemented with the proposed model; it interacts with a more general agent model that abstracts higher-level cognitive functions. In the virtual scenario, characters are performing different physical activities, leading to two expected behaviors: a. Physical activities increment hunger depending on variable values of energy expenditure associated with each activity; and b. Hunger triggers a food-intake action. The results of the implementation show that the model responds as expected, and if more diverse activities are integrated (besides food-intake) we could have a more diverse behavior of looking for food. We believe that this model could be integrated with other models developed with the same approach, specifically for the part of higher-level cognitive functions, in order to obtain more specific behaviors, and therefore, a more realistic simulation.