When people (or other animals) run to catch a flying ball, how do they ensure that they get to the right place at just the right time?
The computational theory of the mind says that we have a "mental model" of how things fly through the air. We use this to work out the likely trajectory of the ball, and run to where our model says the ball will come to earth. The embodied cognition people point out that this is wrong. We use a heuristic: we run so as to keep the line of sight to the ball at a constant angle to the horizontal and this simple rule ensures that we get to the right spot just at the right time.
Gerd Gigerenzer touches on this topic during this excellent talk I came across via a tweet by Yves Smith. He gives the clearest explanation I've seen of how the two approaches differ in their predictions, and how the computational theory is simply wrong even if our goal is not a realistic description of how people actually manage to catch balls but is, as Friedman might have argued, merely to be able to come up with good predictions of their observable behavior.
Of course, his topic is not embodied cognition, but rationality in general, and how decision making under uncertainty differs from decision making under risk, how heuristics are not "second best", and how complex problems don't need complex solutions. Robustness is more importance than optimization, when you cannot know the optimum. His argument in favor of a simple 1/N heuristic, as compared to Markowitz's portfolio allocation theory, was quite astonishing.
Of course, his topic is not embodied cognition, but rationality in general, and how decision making under uncertainty differs from decision making under risk, how heuristics are not "second best", and how complex problems don't need complex solutions. Robustness is more importance than optimization, when you cannot know the optimum. His argument in favor of a simple 1/N heuristic, as compared to Markowitz's portfolio allocation theory, was quite astonishing.
No comments:
Post a Comment