Non-Nietzschean Decision Making

Abstract

Two ways have been suggested to protect evidential decision theory¹ from the kinds of counterexamples which have given rise to causal decision theories.² One, which I shall call the ‘Cartesian defence’, relies on using the agent’s privileged access to his own reasons for acting in order to render the counterexamples innocuous, and has been presented in a number of papers by Ellery Eells.³ The other, known as the ‘ratifiability defence’, has been described by Richard Jeffrey,⁴ and consists in ratifying that a preferred act will be preferred even when that act is actually chosen — i.e. having decided that an act is optimal, there is no other act which would then be preferred, conditional on that decision having been made. Both of these defences require that restrictions be placed on the set of situations within which it is appropriate to use decision theory. Ingenious and inventive as these defences are, the required restrictions are, I believe, unjustifiable as normative criteria for rationality. Furthermore, if implemented, they would so restrict the application of decision theory that its interest as a guide to life would be almost completely erased.

It is a great pleasure to contribute an essay to a volume honouring Wes Salmon. For many years his work in explanation, probability, and causality has served as the inspiration for much of my own work in those areas and his influence on the literature as a whole has been widespread and beneficial. He has also had a longstanding interest in Bayesian reasoning. This last subject is not quite as far from the first three as one might think, for it is exactly the kind of probabilistic causal relations underpinning explanation and objective probabilities which turn out to be required for a satisfactory decision theory. I do not know if he would agree with what follows but even if he does not, it is irrelevant, because Wes is one of those rare people whose interest lies in getting things right rather than insisting that he has got it right. Would that we were all as careful and honest.

Previous versions of this paper were read at the Center for Philosophy of Science, University of Pittsburgh; the Pacific Division of the American Philosophical Association; and Virginia Commonwealth University. I am grateful to Brad Armendt, Ellery Eells, John Heil, Mark Overvold, and Nicholas Rescher for helpful discussions on issues connected with this paper. It should not be assumed, of course, that they concur with the conclusions of this paper.