>* It is not, actually, a realistic simulation of how the real world
>works. IIRC from the little cognitive psychology I've studied,
>ability CAN be rated by the number of times a skill has been used. I
>expect Thom could put us straight here.
As I recall this is only true for very simple tasks. So a skill of "bounce ball off wall and catch" might* work along these lines, but "play tennis" doesn't, because it is a complex or meta-task. As soon as you get a task which involves co-ordinating, planning, decision making, weighing options, breaking the task into sub-tasks, etc. then the simple model breaks, and you get the more expected diminishing returns model.
Not an expert on cognitive psychology per se, but I am interested in AI, and therefore know something of the field.
Cheers,
Graham
-- Graham Robinson graham_at_... Albion Software Engineering Ltd.
Powered by hypermail