Schroeder is offering more of a general structure for an expressivist account than a fully-worked out one, and one of the points he’s fairly vague on is what descriptive predicate should typically follow the ‘is for’ attitude. For the purposes of the paper, he adopts a proposal of Gibbard’s, which analyses disapproval (a technical term for the expressivist) in terms of being for blaming for; so the idea is that ‘Jon thinks murder is wrong’ should be rendered as ‘Jon is for blaming for murdering’.
(Note that we can’t just adopt the ‘is for’ proposal without any descriptive predicate: ‘is for the non-occurrence of’ because this collapses two readings we want to keep distinct; the non-occurrence of not-murdering is the same as the occurrence of murdering, while not blaming for not murdering is not the same as blaming for murdering.)
Taken literally, it looks like there are counterexamples to the analysis in terms of blaming. There are surely cases where we think something is wrong, but are against blaming anyone for it, perhaps because we think that apportioning blame at all would be unhelpful. Similarly, the suggestion that we should use ‘avoiding’ falls foul of cases where we think something is wrong, but are not for avoiding it, because all of the alternatives are worse.
Of course, these observations rest on ordinary usage of ‘blame’ and ‘avoid’. If ‘blaming for’ is a technical term with a stipulative meaning, like ‘disapproval’ and ‘tolerance’ have traditionally been for the expressivist, then perhaps the problem can be nullified. So I’d suggest resurrecting the old Blackburnian terminology of ‘booing’ and ‘hooraying’, and saying that we have tacit knowledge of the meaning of these terms in virtue of our competence with moral discourse.
We can take either of these as primitive, and define the other in terms of it; for example, booing x is equivalent to hooraying not-x, and hooraying y is equivalent to booing not-y. The advantage of this is that it doesn’t seem to be vulnerable to the same kinds of intuitive counterexamples as any of the candidate descriptive predicates that Schroeder mentions. The disadvantage is that we then require two primitive notions in our expressivist semantic, rather than one (the being for relation).
Another thought we had was that the ‘being for’ proposal seems to lose some of the distinctive thought behind expressivism, that moral judgments consist in some attitude to the act whose morality is called into question. On Schroeder’s proposal given in terms of blaming for, thinking murdering is wrong doesn’t involve having some attitude to murdering; rather, it involves having some attitude to blaming for murdering. This allows us to ask for an explanation of why someone has their particular attitude to blaming for murdering. For Schroeder’s kind of expressivist, no explanation is possible; but for a moral realist, an explanation is easily available – it’s because murdering is wrong.
The worry, then, its that the demand on Schroeder’s expressivist for explanation of why someone has some particular attitude to blaming for x seems rather more pressing than the demand on the traditional expressivist to explain why they have some attitude to x itself. This can be thought of as a dilemma for the expressivists – either they have a working semantic theory without sufficient motivation, or they have a well-motivated theory which cannot explain logical validity for moral arguments. Schroeder, who is no expressivist himself, would presumably be happy with this dilemma.
One nice thing we noticed about the proposed analysis is that it disambiguates apparently distinct claims which the normal expressivist view runs together. Where a traditional expressivist would say ‘Jon strongly disapproves of murdering’, Schroeder’s expressivist can disambiguate this as either ‘Jon is strongly for blaming for murder’ or ‘Jon is for strongly blaming for murder’.