In his Modal , Brian CheUas raises two objections to standard deontic logic. I shall be concerned here with the first objection: that standard deontic logic is inadequate because it lacks the resources to express conditional obligation sentences. I shall argue, first, that Chellas' objections tell against conditional obligation sentences because those sentences are conditional, not because they pertain to obligation. The problems he raises, therefore, afflict standard accounts of 'if ... then ...' whether or not deontic notions are involved. Second, I shall contend that there are other significant differences between Chellas' and standard systems which reflect a fundamental difference in interpreting statements of obligation in general. Only by recognizing these differences can one make intelligent choices among competing deontic principles.
CONDITIONALS
Chellas represents the conditional obligation of B, given that A, by 'O(B/A)'
(e.g., that you ought to apologize to Peter if you have hurt his feelings). Standard deontic logic, as Chellas deFmes it, contains no such complex operator. We might think that the notion can nevertheless be defined by a standard formula: 'A-+OB', 'O(A ~B)', or 'D(A-+OB)'. But, as Chellas points out, none succeed in expressing conditional obligation. The first comes out true whenever A is false or 'OB' true; the second, whenever either 'O-A' or 'OB' is true. On all three, 'O(B/A) ' implies 'O(B/A & A')'. The alleged definitions thus variously legitimize arguments such as 'You are not in Scotland; therefore you should commit a murder if you are in Scotland', 'You shouldn't hurt Peter's feelings, so, if you do hurt them, you ought to kill yourself and Peter too', and 'You ought to apologize to Peter if you have hurt his feelings; hence you should apologize to Peter if you have hurt his