Possibly Optimal Decision-Making under Self-sufficiency and Autonomy

Any self sufficient autonomous agent faces multiple tasks to which it must sequence its behaviour if it is to manage its time appropriately. This paper considers the minimal multiple task scenario — the two resouce problem. Some possible mechanisms, inspired from ideas in ethology and having claim to perform optimally when analysed mathematically, are considered: one based on cost function and another based on motivational tendancies. Several variants of the mechanisms are implemented and tested in a continuous two dimensional simulation environment under a variety of conditions. Those mechanisms closely linked to cost functions do not perform well, are not adaptive, and dither. The mechansisms based on motivational tendancies perform better, are more adaptive, and demonstrate opportunism. The difficulty of finding effective mechanisms by using a functional perspective is noted.