A researcher reports a chimpanzee in captivity filling his mouth with water in order to penalize a not-so-friendly keeper. The chimpanzee coaxes the keeper and tries to lure him to proximity to spit water in his face. Sometimes the plan fails, and we would expect the chimpanzee not to spit water. (He might spit water, but not for that purpose. Spit, therefore, means something more as part of a plan.) His actions depend on how much the keeper conforms to his role as part of the chimpanzee’s plan. In this sense, it is instrumental planning. It is worth formalizing some aspects of this planned action to see that at this level of instrumentalizing we are dealing with context-free one-agent-centered dependencies.

Spitting depends on achieving LureKeeper, which might enter the chimpanzee’s plan by perceiving and interpreting the actions of the keeper. If coaxing him fails, we might still have some acts of stalking the keeper by the chimpanzee, but they would presumably not amount to a plan of spitting at him. This is a dependency that—let’s say—he himself established as part of the plan. Below is context-free grammar on behalf of the chimpanzee for some potential ways to get what he wants.

 

THE QUESTION: 

Show a tree of chimpanzee actions where he coaxes the keeper by stalking him for a few actions, then asks for banana, but does not spit because he does not have water in his mouth. Can we tell externally the causes of him not spitting?

S Fillwater LureKeeper Spit LureKeeper -> Coax | Hoax Coax -> Stalk Coax | AskBanana