Disagreements about Morals, Conditionals, and Epistemic Modals

A distinguishing feature of the evidential roles introduced in section 8.1 was that one believes an evidential role only in virtue of having one’s credences distributed over the precise (i.e.

In the three cases we mentioned earlier, it is natural to explore theses, such as those below, in which beliefs about certain suspect kinds of propositions are identified with an attitude towards an ordinary non-suspicious proposition. Such theses can be seen as ways of articulating (an important component of) the expressivist picture: attitudes towards the lightweight propositions are derivative on your attitudes towards the factual.

Conditionals: To have a certain credence in a conditional proposition, that if p then q, is just to have certain credences in non-conditional propositions. In particular it is just for you to have credences in p ∧ q and p that have that ratio.

(Adams [1], Edgington [38], and others) Epistemic Modals: To have a certain credence in the proposition that it might be the case that p is just to have a certain credence in p.

(Schulz [129]) Moral Propositions: To have a certain credence in a moral proposition, that it is good that p, is just to have a certain non-doxastic attitude towards the non-moral proposition that p. For example, it is for your credence that it is good that p to be identical to the degree to which you desire that p. (Seethe ‘Desire as Beliefviewcriticizedin Lewis [88]. A similar thesis identifying attitudes towards moral propositions is endorsed in Gibbard [64]).^[125]

In each case, we have a thesis about the role that a certain type of proposition plays in thought replacing a metaphysical claim about the status of those truths. To believe that the cradle fell if the bough broke (to some degree, or fully) is to do nothing over and above believing, to certain degrees, non-conditional propositions about the bough breaking and the cradle falling.

At this juncture one might worry that the literal identifications of these attitudes are too strong. Take, for example, Gibbard’s thesis that deciding what one ought to do is deciding what to do. It seems quite clear that one can decide that something is what one ought to do without deciding to do that thing. Perhaps a rational person will always make that last step, but it is a step that it seems as if one could fail to make. Similarly, it is natural to think that one could be fully confident that it’s not raining, while also being confident that it might be raining. Of course, it would be completely irrational to do this, but it is not metaphysically impossible, just as it is irrational, but not metaphysically impossible, to believe that it’s raining and it’s not raining.

No doubt one could tell a special story about attitude reports that accommodates these intuitions.

The difference between moral, conditional, and epistemic matters now reduces to a thesis about rational disagreements. In general, two people can rationally agree about one subject matter whilst disagreeing about another. Arguably, such disagreements can persist even when the evidence is shared. On the other hand, the expressivist will maintain that this is not so when it comes to moral, conditional, and epistemic propositions: in such cases two subject matters can be intimately tied together. For example, suppose two people agree about the categorical matters—suppose, in particular, that they agree about how likely it is that a given coin was flipped, and about how likely it was that it was flipped and landed heads. Then according to the expressivist about conditionals, they cannot rationally disagree about the conditional matter of whether the coin landed heads if it was flipped. Note, however, that the proposition that the coin landed heads if it was flipped is logically independent of the propositions that it was flipped and the proposition that it was flipped and landed heads: indeed the conditional could be true or false consistently with the coin having not been flipped.^[126] So agreement about the non-conditional forces agreement about conditional matters, even though such agreement is not forced by logic or conformity to the probability calculus. Similar agreement principles can be formulated for more familiar versions of expressivism. For example, a certain kind of moral expressivist might maintain that if two people agree about the non-moral propositions—they agree about how many lives each charity will save, say—and care about the same non­moral propositions—they care about saving lives—then they must agree about which charity they ought to donate to.

In each case, we have what I'll call the ‘rational supervenience' of attitudes in one kind of proposition on another.

Supervenience of Hypothetical Beliefs on the Categorical Beliefs:

If two rational agents assign the same credence to every categorical proposition, they will assign the same credence to every hypothetical proposition.

This supervenience principle has a tight-knit connection to the principle Condition­als. Conditionals tells us what it is to believe a conditional to a certain degree purely in terms of our degrees of belief in non-conditional propositions—so credences in simple conditional propositions are straightforwardly determined by credences in the categorical. A shortcoming of Conditionals, however, is that it only tells us what it is to believe a simple conditional proposition, and not an arbitrary hypothetical proposition. I take it that hypothetical propositions also include arbitrary disjunc­tions, conjunctions, and negations of conditional propositions, as well as conditional propositions with hypothetical antecedents and consequents. To get the full force of the supervenience principle we need principles that tell us what it is to have a credence in one of these extended hypothetical propositions purely in terms of our credences in categorical propositions.^[127]

Regardless of the details, the takeaway message is that these versions of expressivism about each subject matter entail a kind of rational supervenience thesis. In the hypothetical case, for example, these effectively guarantee that all disagreements about hypothetical propositions derive from disagreements about the categorical.

Could these rational supervenience theses give precise cash value to the idea that hypothetical propositions are second-rate or derivative, while categorical propositions are not? The driving intuition here seems to be that while you can have genuine disagreements about categorical matters, all disagreements about the hypotheticalboil down to disagreements about the first-rate, categorical facts.

It is, however, unclear to me whether this strategy will succeed in completely capturing non-factualism about hypothetical facts. One reason for scepticism is that supervenience is not an asymmetric relation: it could turn out that categorical beliefs also supervene on the hypothetical in the sense that once you know what a rational agent's credences in the hypothetical propositions are you can work out their credences in the categorical propositions.^[128] This symmetry problem is also a problem for the stronger thesis that simply identifies a credence in a hypothetical fact with a certain distribution of credences in categorical facts: the assertion of the identity of these states alone can't capture expressivism, because it could equally be taken to be a reduction of conditional categorical beliefs to unconditional hypothetical beliefs.

In light of all this, I’m not quite sure what to say on behalf of the expressivist. However, whether or not the supervenience theses (and the identity theses) fully capture the idea that certain facts are second-rate, they do help distinguish these views from a pretty stark kind of realism about conditional facts. According to the stark realist the truth of hypothetical propositions doesn’t depend in any important way on the categorical facts. Even once you have formed opinions about all the categorical facts, there may still be a large number of opinions to be had about the hypothetical facts left open to you. If we are merely worried about distinguishing ourselves from the stark realist then the supervenience thesis does a good job of doing that.

8.3