00100	THE MONKEY AND THE BANANAS
00200	
00300		If you put a monkey in a room with a bunch of bananas hanging
00400	from the ceiling too high to reach and a box in  the  corner  of  the
00500	room,  the monkey will often get around to dragging the box under the
00600	bananas and climbing on the box to get the bananas.   If  a  computer
00700	program  that  uses  logic  is to be as smart as a monkey, we must be
00800	able to express in logic what the monkey knows  about  the  situation
00900	and  the  laws  that  determine the effects of the actions the monkey
01000	might take.  Here is a somewhat simplified formulation of these facts
01100	and laws:
01200	
01300		1.     at(Box,under(Bananas),s)  and on(Monkey,Box,s) implies
01400	has(Monkey,Bananas,result(Monkey,reach-for(Bananas),s))
01500	
01600		2. at(Box,x,s) implies  at(Box,x,result(Monkey,climb(Box),s))
01700	and on(Monkey,Box,result(Monkey,climb(Box),s))
01800	
01900		3.    in(y,Room,s)  and  in(Monkey,Room,s) and in(z,Room) and
02000	isplace(z)            and             portable(y)             implies
02100	at(y,z,result(Monkey,move-to(y,z),s))
02200	
02300		4.in(Monkey,Room,Start)
02400	
02500		5.in(Box,Room,Start)
02600	
02700		6. portable(Box)
02800	
02900		Quite a few computer programs exist that can take these facts
03000	and find a sequence of actions for the monkey that will get  him  the
03100	bananas.  Of course, no-one imagines that monkeys solve such problems
03200	by  manipulating  logical  formulas,  but  progress   in   artificial
03300	intelligence  depends  on  our  ability  to  express  the  laws  that
03400	determine the effects  of  actions.    Even  toy  problems  like  the
03500	"monkey-and-bananas"  play  a role in the present beginning stages of
03600	this research.