Salwen, Hakan. Hume's Law: An Essay on Moral Reasoning.

Peterson, John

SALWEN, Hakan. Hume's Law: An Essay on Moral Reasoning. Stockholm: Almqvist and Wiksell International, 2003. 154 pp. Paper, $62.50--Hume's law, that is, that moral claims cannot be inferred from exclusively nonmoral claims, is widely accepted by recent and contemporary philosophers, some exceptions being John Searle and A. N. Prior. Chapter 1 distinguishes three versions of the law: the formal version (HL(F)), the conceptual version (HL(C)), and the epistemic version (HL(E)), all of which, according to Salwen, are true. When "valid inference" means "a sentence is a logical consequence of a set of sentences K iff there is no interpretation under which all the sentences of K are true and x false" (p. 15), then the law is, HL(F): "For all valid arguments, K>X, and all moral expressions O, if O occurs nonvacuously in X, then O appears in K" (p. 17). When "valid inference" means an inference "such that the truth of the premise conceptually guarantees the truth of the conclusion" (p. 18), and when a bridge premise is a conditional sentence whose antecedent is a nonmoral sentence and whose consequence is a categorial norm or moral sentence (pp. 18-19), then the law reads, HL(C): "There are no analytic bridge sentences" (p. 19). Thus, HL(C) implies, as against Searle, that the bridge premise, "If Jones has promised to pay Smith five dollars, then Jones ought to pay Smith five dollars," is false. Unlike either HL(F) or HL(C), HL(E) is about nonmoral reasons for holding moral beliefs. Thus, "HL(E) implies that, say, the acceptance of 'Jones has promised to pay Smith five dollars' is a reason to accept 'Jones ought to pay Smith five dollars' only if it is accepted that promises ought to be kept" (p. 19).

Chapter 2 reviews some arguments both for and against HL(C). Though these have drawbacks of their own, they all share the problem, along with HL(C), of relying on the idea of analyticity. Discussing the tenability of HL(C) is therefore futile unless this idea is clear and justified, unless it is shown to have an explanatory role to play, and unless there are in fact analytic sentences (p. 23). For analytic truths have been called into question by Quine.

Accordingly, chapter 3 considers analytic sentences and Quine's denial of them (p. 46-51). For Quine, that denial follows from indeterminacy about the meaning of theoretical sentences. To counter this, Salwen proffers an account of "analyticity" that avoids Quine's criticism and then uses that account in formulating HL(C). That account is: (A) "A conditional belief held by P is analytic only if it is the case that if P were to reject it, then P's belief would cohere very badly with the rest of P's beliefs" (p. 64). Under (A), and when by "bridge commitment" is meant a belief in a conditional that has a nonmoral antecedent and atomic moral consequent, HL(C) becomes: "For all persons, P, bridge commitments and times t, it is false that if P were to reject the bridge commitment at t, then P's rejection would cohere very badly with the rest of P's beliefs held at t" (p. 63). In other words, no bridge commitment is analytic.

Salwen then supports this restatement of HL(C) by the following indirect argument: assume (i) that P believes or has the bridge commitment that if a certain action n causes pain then n, is wrong. Then, assuming holism and the principle of charity (that is, that there is extensive moral agreement among persons), it is safe to assume (ii) that P holds a set of moral beliefs which include believing it right to obey the laws, to do what God wants, to promote the interests of P's nation, and so forth, and that in some circumstances, at least, P holds these things even when the action in question causes pain. Assume, finally (iii), that P's bridge commitment in (i) is analytic. It is evident that (i), (ii), and (iii) are in conflict. For if (iii) is true, then we could not rationalize P to disbelieve that n causes pain but is not wrong. But then it follows that P does not hold to the set of moral beliefs referred to in (ii). For holding those beliefs is sufficient to rationalize P's rejection of her bridge commitment, n. So, since the assumption of the falsity of HL(C)--that is, (iii) above--conflicts with the requirements of holism and charity, it follows that HL(C) is true (pp. 64-7). In short, assuming holism and the principle of charity, then (A) above is plausible, and (A), coupled with the same two assumptions, implies that HL(C) is true.

Chapter 4 formulates and defends HL(E). HL(E) reads: "A nonmoral belief constitutes an epistemic reason for a person, P, at a certain time, t, to hold an atomic moral belief only if P at t accepts a moral principle connecting the two beliefs" (p. 76). It is argued that HL(E) gains support from a coherentist approach to epistemic justification.

Chapter 5 shows that, being a formal thesis, HL(F) is irrelevant to any metaethical theory. Salwen also argues that even if G. Schurz is right in The Is-Ought Problem in arguing that Hume's law implies that ethical theories cannot be scientifically justified, he has not shown that the law implies moral skepticism (pp. 99-106). He argues further that HL(C) supports the metaethical theory that gives the best explanation of HL(C). Since that theory is neither analytic naturalism, which is incompatible with HL(C) (p.113), nor synthetic naturalism (pp. 116-17), it is implied that HL(C) supports nonnaturalism. Finally, he shows that HL(E) has significance in regard to the attribution of moral beliefs (pp. 117-23).--John Peterson, University of Rhode Island.