Ethologists have long believed that the control of action selection in animals is hierarchical in structure [Lorenz1981, Tinbergen1951, Dawkins1976]. This classical view is supported by extensive observation of and numerous experiments on animal behavior. Tinbergen Tinbergen51 described a ``working hypothesis'' of an action selection mechanism composed of a hierarchy of nodes. The top level nodes represent high-level motivational behaviors, such as ``reproduction''. The lowest level nodes correspond to simple sub-behaviors such as ``lay eggs'' and ``comb feathers''. There is a feeding of activation down through the hierarchy. Nodes in the same level inhibit one another so that only one can be active at any time. Lorenz Lorenz81 described a similar idea which emphasizes the effect of both internal and external stimuli in activating a specific behavior.
|Xiaoyuan Tu||January 1996|