We speak, not only of X having reason to A (or of R being a reason for X to A), but also of X having reason to A rather than B (or to A as against B). According to Justin Snedegar, we should always read ‘rather than’ or ‘as against’ in such constructions as meaning but-not. So if I have reason to write this post rather than start scrolling down my Facebook feed, this means that I have reason to take the former option and do not have reason to take the latter option.
And this leads to the following strange result in certain ordinary-enough cases like the one below. (Adapted from Snedegar. The main difference between his case and mine is that I put the [X having reason to A] feature as against the [R is a reason for X to A] feature front-and-center, roughly in line with Daniel Fogal’s framing or reasons and having-reason.) Here is the case:
I’m deciding between running, biking or driving to work and there is one salient goal: my avoiding wearing myself out (as I am out of shape). (We’ll suppose that the goal of getting to work, or doing so in a timely-enough manner, is equally well promoted by each option.) Now suppose that (a) I have reason to drive rather than bike. Reading ‘rather than’ as indicating but-not (and supposing that the driving will not wear me out but the biking will) we should infer that (i) I do not have reason to bike. And suppose as well that (b) I have reason to bike rather than run (again due to the wearing-me-out factor). From this, trivially, we should conclude: (ii) I have reason to bike. And now we have an apparent contradiction.
We might think we can get out of the contradiction by denying, contra Snedegar, that ‘rather than’ in these constructions always indicates but-not. I’m pretty sure that Snedegar is wrong here — but I think ‘rather than’ does in some contexts does indicate but-not and we have stipulated such a situation for [a] (and I have filled in the case to show how and why this is so).
Snedegar suggests that we solve the problem here by treating [i] and [ii] as shorthand versions of [a] and [b], versions that simply focus on one of the two options in each contrast. And this reflects the general idea that having-reason (and reasons) is, in all cases, relative to a contrasting alternative (or set of alternatives) — “contrastivism” regarding reasons (or having-reason) is what Snedegar calls this. The more standard, non-contrastivist view of having-reason (and reasons) looks to be stuck with the contradiction.
Snedegar is wrong (or at least his case does not work). He is wrong because, in cases such as this, what appears to be one salient outcome (value, goal) across [a] and [b] (and [i] and [ii]) might be two distinct, though closely-related, outcomes. In the above case, for [a] (and [i]), the relevant outcome is binary, something like (I) not getting worn out beyond a certain amount. But this same outcome is not of the right sort to get the contrastive having-reason feature for [b]. What we need is some gradable quantity like this: (II) minimizing the amount that one is worn out. So then we get (i*) I do not have reason, relative to the outcome of not getting worn out beyond extent n, to bike and (ii*) I have reason, relative to the outcome of minimizing the extent to which I am worn out, to bike. And there we are: non-contrastivist framing, no contradiction. In this way, we can save the more natural explanation of contrastive having-reason features: they are explained by more basic, non-contrastive having-reason features that options have entirely apart from being related to some other option (or set of options).
Of course, one might try to come up with a case like the above, but in which it is clear that both (a) some one outcome really is the sole outcome that is salient to each contrastive normative feature and (b) both of the statements about the contrastive having-reason (or reason) are true. I suspect that this challenge will be hard to meet. If so, the puzzle here goes away — and with it the big problem for the non-contrastivist view of having-reason (and reasons). If I’m right here, the correct positive task here is coming up with a detailed, convincing account of (outcome- or other response-object-specific) contrastive normative valence (for some option or between options) in terms of (outcome- or response-object-specific) non-contrastive normative valences for individual options.
*****
Snedegar on contrastive reasons:
https://www.amazon.com/Contrastive-Reasons-Justin-Snedegar/dp/0198785933
Fogal on ‘there is reason’ (or ‘having reason’) and something being a reason:
Policy of Truth 
Good post. But can’t you save Snedgar’s view by relativizing reasons to times and/or circumstances? I don’t know if he’d do this, but here’s what I’d do.
Suppose that S faces the choices of driving, biking, or running to work, as stipulated in your post. If reasons are relativized to times and circumstances, then suppose that T1 and C1 obtain. Then if S has reason to drive to work at T1/C1, he lacks any reason to bike or run at T1/C1. To give him a reason to bike or run, you have to change the T/C variables (maybe one, maybe both). But described this way, no contradiction arises. S has reason to drive at T1/C1, but does not (at all) have reason to bike. So there’s no way to stipulate that he has reason to bike and generate the contradiction you describe.
I realize that people in the reasons literature would at this point accuse me of conflating “all in reasons” with “reasons,” but I would contest that distinction precisely on the grounds that it yields the contradiction you attribute to Snedgar. I agree with Snedgar’s claims about contrastivism, or at least the one that heads your post:
You say:
I don’t see why (a) and (b) are any harder to meet than identifying gradable quantities that underlie all of practical reasoning, in order to save the thesis that when S has the options of driving, biking, or running to work, the case in which driving is best and available is still a case in which he has some residual reason to bike.
Given my work schedule, it’s going to be awhile before I can respond to anything you write, but I thought I’d throw that comment out there.
LikeLiked by 1 person
Thanks for wading through that and for your thoughts. Just a couple of quick points (for you to get to or not as you can or want — I know you are busy):
(a) JS’s main position is that R being a reason for X to A (and I think as well X having reason to A or what I call normative valence) is a relationship that includes an argument-place (‘rather than B’) for a contrasting option or set of options (from the or some relevant complete set). He agrees (as do I) with the standard view that there are sometimes-omitted argument-places for times and circumstances.
(b) The dialectic here is this: even though it is natural to explain explicitly contrastive reasons (or having-reason) in in terms of non-contrastive, it turns out that we can construct cases in which this framing or analysis yields contradictions (e.g., X has non-contrastive reason to A and X does not have non-contrastive reason to A). What I’ve tried to do is block this move in the very specific sort of case he has constructed. I haven’t here raised any trouble for contrastivism (other than green-lighting going with intuitions that favor explaining the contrastive in terms of the non-contrastive). So his view does not particularly need to be saved against any points that I make in my post.
(c) Here, for illustrative purposes, is how we might flesh out a non-contrastive view of what is going on in a somewhat similar drive/bike/run case. Suppose the only salient goal is my trying to avoid wearing myself out (say, not exerting myself more than x amount). And suppose that driving successfully avoids this, biking has a 90% chance of avoiding this, and running has no chance of avoiding this. On the non-contrastivist view, the basic normative elements here are something like this: I have reason of quantity x to drive (goal will be achieved), reason of less quantity than x to bike (goal will probably be achieved), and no reason to run (goal will not be achieved) (ignore that a more revealing description might put matters here in terms an outcome to be avoided and my having reason against that negative outcome being achieved). So, based on this, I have reason to bike rather than run (where ‘rather than’ is read to indicate but-not). But also, intuitively, I have reason to drive rather than bike. Maybe this is because in this case ‘rather than’ indicates ‘more than’. Or maybe ‘rather than’ here indicates only what the conversationally salient competing option is (and thus does not indicate some other-option-relative normative having-reason feature at all). JS does also believe that ‘rather than’ always indicates genuine normative contrast or but-not (even though, confusingly, the most natural reading of this constrastive structure pushes in the direction of explaining the contrastive in terms of the non-contrastive!). I disagree with him on this point as well.
(d) In these sorts of “teleological” cases (desired or valuable outcomes to be promoted to generate having-reason and being-a-reason features), it is natural to think of an option promoting the right sort of outcome as a not-relative-to-other-options matter. And, correspondingly, so is the agent having reason (to some degree) to take that option. Snedegar’s more theoretical reason for being a contrastivist about reasons (and having-reason or an option having normative valence) is that he takes promotion itself to be an other-option-relative matter. For example, if you take option A promoting outcome O to be taking A raising the probability of O obtaining, you need to ask “raising the probability of O obtaining compared to what baseline?” If the baseline is taking some other option (e.g., doing nothing, which looks to be an option), then A promoting O is always some-other-option-relative. These arguments get rather technical, but I think they go against what is, for me, a pretty strong intuition in favor of options promoting outcomes being rather self-contained affairs. I suspect that the right way to vindicate this intuition, at the level of an analysis of promotion, is to analyze promotion as a kind of causation and analyze causation — or causation of the relevant sort — in terms of production (as against the straight counterfactual analysis of all causation). Or one might keep the counterfactual analysis, but relativize to standing conditions prior to choice (this being subtly different from relativizing to the option of doing nothing).
(e) Options themselves are relative to their option-sets and how they get individuated (roughly, by something like (i) availability to the agent and (ii) salience to various conditions that demand response (e.g., outcomes to be promoted) in the context). And so, through the identity of the option and its set, an option’s normative valence (relative to some single factor, say an outcome to be promoted) is relative to its entire set and the conditions that define that set or make it the or a set that the agent faces. But I take this to be a different issue from that of whether there is an argument-place (‘rather than B’) for contrasting options in X having reason to A (or R being a reason for X to A).
I’m hoping some of these points address some of the points you made!
LikeLike