| OCR Text |
Show Rationality and the Structure of the Self, Volume I: The Humean Conception 515 Thus we can see the significance of Hodgson's (and, for that matter, Kant's) interest in truth-telling and promise-keeping in general as examples for discussion, rather than the instances of these on which Gibbard and Lewis focus. For this reminds us that the question of whether such general conventions of behavior are possible must be settled before the question of whether any particular instances of them are. To the extent that promisekeeping and truth-telling presuppose regularities of speech behavior and shared expectations about how language is used, it is far from clear that communication of any sort would be possible under such "ideal" conditions. To assume in advance that they would begs the question of whether, in an Act-Utilitarian society, such conventions could ever arise, and this is just what Hodgson implicitly denies. If my account of Hodgson's argument is right, then any attempt to formulate the problem as a choice between two alternatives and to assign weights and probabilities accordingly is bound to fail. For in any confrontation between two Act-Utilitarians, neither can have any valid expectations about what the other will do or say, nor any basis for predicting this on probabilistic grounds. This means that the possible alternatives of action open to two Act-Utilitarians in any situation are fairly unlimited. To the extent that doing x depends on its utility, x's utility on whether x is expected, and whether x is expected on x's utility, there can be no sufficient reason for expecting x, for any x, to be done. So the choice is not between, say, going to the tennis courts and staying home, for which a coordination could indeed be established. The choice is rather between going to the courts, staying home, walking the dog, breaking a window, doing a headstand, and the myriad other possibilities that exist between two Act-Utilitarians who have "agreed" to play tennis. Now sometimes Gibbard seems to talk as though a coordination solution might fortuitously occur and be acknowledged as a solution even if it cannot be expected to occur.33 Such an occurrence might then provide sufficient reason for expecting it as a solution to future problems. Basically, this is Lewis's argument for the origin of a convention, briefly adumbrated above, and the same objection to it is relevant. If the parties could originally conceive of the issue as a limited-alternative coordination problem, perhaps some such solution to it might be forthcoming. But we have seen that they cannot. Because a basis for expectations about others' behavior is lacking, no act possible under the circumstances has greater initial subjective probability than any other. Hence any act that may be performed cannot be regarded as a solution to the problem of whether to do x or y, since the question of what one should do cannot be made determinate in this way. So even if an act x were the solution to such a problem, the parties would not regard it as such 33 Gibbard, op. cit. Note 21, pp. 167-70. © Adrian Piper Research Archive Foundation Berlin |
