William Alston’s essay, "Moral Attitudes and Moral Judgments" (Nous 2:1 (1968), 1-23) is fairly old by the standards of work in metaethics, but one of the ideas he discusses has been gaining some recent popularity, and I’ve never seen his argument discussed. He argues that the most plausible attempt to analyze moral judgment along expressivist lines leads to a dilemma for expressivists: the analysis is either circular, or the expressivist has to give up non-cognitivism.
He says that according to expressivists (all the following numbered propositions are Alston’s, but the numbering is different in the article),
1. ‘x made a moral judgment about O’ =df. ‘x expressed a moral attitude toward O, [and x asserted (implied, presupposed, committed himself to its being the case that P]’.
The problem is providing an analysis of the right-hand side, what a moral attitude is. He offers the following:
2. ‘x has a moral attitude toward O’ =df. ‘(a) x is disposed to act for or against O, (b) x has affective dispositions that naturally go with (a), (c) x is prepared to make a moral judgment about O’.
The third condition, (c), gives rise to the first horn of the dilemma. Moral judgment is analyzed in terms of the expression of a moral attitude (plus, perhaps, something else). But a moral attitude involves the disposition to make a moral judgment, either pro- or con-, about the intentional object of the judgment. But why think that (c) should be part of the analysis of moral attitude?
Alston’s answer is that (a) and (b) can’t provide sufficient conditions. He writes:
I take it to be perfectly clear and uncontroversial that the class of moral attitudes is narrower than the class of "attitudes". My preference for Munster cheese and my desire to be liked by people are clearly not moral attitudes (though of course it is conceivable that I may also morally approve of eating Muster cheese and that I may take it to be my duty to act so as to be liked by people). Having pro-or-con dispositions toward an object is not sufficient for having a moral attitude toward that object (10).
(c) would just be the simplest way to distinguish between moral attitudes and all the other non-moral ones.
What else could replace (c) and help distinguish morally disapproving an object from simply disliking it, being annoyed by it, or thinking it very foolish? Is there a way to make the distinction without having at some point to introduce the notion of a moral judgment? Alston considers the idea that we pick out the moral attitudes by considering the kinds of feelings, emotions, or affective states involved in the attitude. But how do we do that? What’s the phenomenal difference between moral and non-moral emotions? He doesn’t think there is one. The only way to distinguish, e.g., moral indignation from annoyance is to notice that in the former one is prepared to judge the object as morally wrong. So, we’ve only made the circle slightly bigger: moral judgment is analyzed in terms of moral attitudes; moral attitudes are analyzed in terms of moral emotions; but, an analysis of moral emotions invokes moral judgments or the dispositions to make them.
A final attempt to make the distinction leads to the second horn of the dilemma. He suggests that an attitude is a moral attitude if only moral considerations are relevant to its assessment or justification. What makes a consideration or reason a moral one? He offers a simple interest-based account:
3. ‘x gave a moral reason for A’ =df. ‘x attempted to justify A by citing facts about the interests, needs, or desires of other people’.
Given this account, the following replaces (2):
4. ‘x has a moral attitude toward O’ =df. ‘(a) x is disposed to act for or against O, (b) x has affective dispositions that naturally go with (a), (c) only considerations concerning the interests of others are relevant to the assessment of (a) and (b)’.
(4) allows (1) to escape circularity, but it gives up much of what typically makes (1) a kind of non-cognitivism. This is because (4) entails that there are objective criteria for determining whether reasons given for or against moral judgments are relevant. But (4) incorporated into (1) is still recognizable as a form of expressivism:
5. ‘x made a moral judgment about O’ =df. ‘(a) x expressed pro-or-con action and affective dispositions towards O, [(b) x asserted (implied…) that P], (c) what x said, insofar as it goes beyond (b), can be justified only by considerations that have to do with the interests of others’.
So, if the most plausible account of expressivism is to be non-circular, it must build into the concept of moral attitude a set of objective criteria for their fittingness, like (5) does.
There’s lots of things to say, I think. I’ll restrict myself to just one observation here. It’s the analysis of moral judgment in (1) that leads to the dilemma. Alston says that moral judgment is most plausibly analyzed in terms of moral attitudes. He thinks that for the same reason having pro-or-con dispositions toward an object isn’t sufficient for having a moral attitude toward that object, expressing pro-or-con attitudes toward an object is not sufficient for having made a moral judgment about it.
What about the following account? According to expressivism, if x made a moral judgment about O, then x expressed an attitude toward O, [and x asserted (implied, etc.) that P]. Alston would apparently complain that this leaves moral judgment unanalyzed.