It always surprises me when social scientific colleagues express the conviction that in principle, human behavior is predictable.  They seem to make two assumptions:  First, that all human behavior is causally determined, second, that in principle, whatever is causally determined can be predicted.

But even leaving aside the question of whether all human behavior is determined, determinism does not entail predictability.  We can form reasonable expectations about many things in the future, but it would be impossible to predict human behavior with certainty even if determinism were true.  The main reason is that the very act of prediction changes the system which the scientist is trying to predict.  He may think he can predict the effects of his predictions, but this is like the dog’s belief that if only he chases fast enough, he will eventually catch up with his tail.

For example, suppose the social scientist predicts that certain people will do P.  What is to prevent them from doing the opposite of whatever he predicts, just because they are bent on refuting him?  In this case, even though their doing not-P was causally determined by his prediction that they would do P, he was unable to predict it.  Suppose he predicts that on the following day, as he is going about his ordinary activities, his research assistant will shoot him.  Isn’t it likely that he will alter his ordinary activities to keep from being shot?

If he takes his response to his prediction into account, then he must make a different prediction, and then, of course, he must have to take into account his response to that prediction.  Like the dog, he can never catch up.