A Bottom-Up Mechanism for Behavior Selection in an Artificial Creature

In this paper we propose a mechanism for motivational competition and selection of behavior. One important characteristic of this mechanism is that the selection of behavior is modelled as an emergent property of a parallel process. This in contrast with mechanisms for behavior selection and motivational competition proposed earlier, which are based on a hierarchical. preprogrammed control structure. We show that selection of behavior can be modeled in a bottom-up way using an activation/inhibition dynamics among the different behaviors that can be selected. There is no weighing up of behaviors in a cognitive manner and neither are hierarchical or bureaucratic structures imposed. The paper elaborates upon the results we obtained with simulated creatures based on this mechanism. It draws parallels between characteristics observed in animal behavior and characteristics demonstrated by our artificial creatures. Examples are: displacement behavior, opportunistic behavior, fatigue, selective attention, and so on.