I very much love Michael Smith’s recent paper “A Constitutivist Theory of Reasons: Its Promise and Parts” (Law, Ethics and Philosophy 2013, also available on his website). One thing that strikes me about this paper is that, whilst discussing Parfit’s views on practical reasons, Smith seems to also create a powerful objection to the buck-passing views of value, which at least to me seems original and something that hasn’t been discussed in the vast buck-passing literature before. So, what I want to do below is to outline this argument briefly and then introduce some of the ways this argument could be resisted. For what it’s worth, my own commitment to buck-passing might be getting weaker because of this argument.
Here’s the rough outline of the argument:
Premise 1: No equivocation thesis.
We are not equivocating when we talk about reasons for beliefs and reasons for final desires. Corollary: if we give a reductive account of the concept of reasons in either one of the domains, then we have to accept the same reductive account of concept of reasons in the other domain.
Premise 2:
Humean/Lewisian view of reasons for belief: the concept of reasons for belief can be reductively analysed in terms of the concepts of entailment and truth. Hume himself thought that a fact is a reason for belief only if that fact entails the truth of the proposition believed, which would mean that there are only deductive reasons for belief. But, following Lewis, we can also understand inductive and abductive reasons in a somewhat similar fashion. We can think that the fact that p is a reason for S to believe q reduces to the fact that, in those possible worlds in which S believes that q on the basis of p, the fact that p removes all of the other possibilities except for q that the subject isn’t properly ignoring.
Conclusion 1:
Premises 1 and 2 mean that the concept of finally desiring something must be analysable at least in some way in terms of entailment and truth, or otherwise we would be equivocating.
Consequence of Conclusion 1:
The Inheritance Thesis. It seems like the only relevant truths, in terms of which we could understand the concept of reasons for finally desiring things, are truths about what is finally good. This would mean that a consideration is a reason for finally desiring something in virtue of being a consideration that entails the truth of the proposition that the thing in question is finally good. Smith calls this the inheritance thesis because the considerations that are reasons for final desires inherit their reasonhood from supporting the truth of the theses that the objects of these desires are finally good.
Conclusion 2:
Inheritance thesis rules out buck-passing. The buck-passing theory of value must be false because it is incompatible with the Inheritance Thesis. Buck-passers want to understand final value in terms of the reasons there are for desiring the given object. If this is the case, then the reasons for the final desires for the object could not be reduced to being considerations that entail the truth of the proposition that the object is finally good (in the way Inheritance Thesis claims) because as Smith puts it ‘this would get the order of explanation wrong’. And so, because we have reason to believe the Inheritance Thesis on the basis of Premises 1 and 2 we have to give up the buck-passing account of value.
What can we do in response to this argument? It seems to me initially that the premise 1 seems plausible and it would be a significant cost to assume ambiguity here. It also looks like the argument has a valid form
I am less certain about the premise 2. It means that all reasons for belief are epistemic reasons – reasons that count in favour of the beliefs’ truth. If there are pragmatic reasons for belief, then all reasons for beliefs cannot be reduced in the Humean/Lewisian fashion. Smith himself says that the price of giving up the premise 2 for the buck-passer would be that then you would have to think that we cannot explain what it is for a consideration to be a reason for belief, which would beggar belief. But, can there be no other explanations of reasons for belief that would accommodate pragmatic reasons for belief and be compatible with buck-passing?
I also have some concerns about the Inheritance Thesis. What if we accept with Smith that reasons for final desires have to be analysed in terms of some truths, but deny that these truths must be truths about what is of final value? Could we not rely on oughts here? What if to be a reason for a final desire is to be a consideration that supports the truth of the thesis that you ought to desire that thing? In this case, buck-passers could still understand value in terms of reasons.
I finally have some weaker worries about the Conclusion 2 and the incompatibility of the Inheritance Thesis and Buck-Passing, but I’m less certain about how to formulate these concerns. I do accept that explanations are asymmetric. If p explains q then it cannot be that q explains p. If reasons are explained in terms of value, then value cannot be explained in terms of reasons.
However, what if there is no explanatory or conceptual priority here? What if reasons and values are on a par (in the same way as perhaps truth and meaning are)? In that case, it seems to me like something like the inheritance thesis (A consideration is a reason for final desire for X consists of this consideration supporting the truth that X is finally good) and the buck-passing view (For X to have final value consists of there being considerations that are reasons for finally desiring X) could be true at the same time. Both would be just indications of the same close relation from different directions. In this case, we would need to give up any claims of explanatory priority and think that these claims both illustrate how these concepts are related. This would fit the lessons from wrong kind reasons debates where it seems difficult to give a non-circular account of what the wrong kind of reasons are.
In any case, I feel like I am clutching at straws. Smith’s argument seems like a powerful new objection to the buck-passing view to me. I would be very interested in hearing what the readers of PEA Soup make of this argument.
Hi Jussi, I haven’t had a chance to read the paper yet, but why can’t buck passers deny the inheritance thesis by claiming that reasons for desires should be understood with reference to the relevant good-making features, rather than goodness itself? Imagine that pleasure is the only good thing. Buck passers could then say that a reason to desire X is something that entails that X is pleasurable (or similar).
Jussi, I don’t think Premise 2 could be right.
A typical reason to believe that p isn’t a conclusive reason to believe that p. For example, the fact that the class in room 223 is a philosophy class is a reason for you to believe that the instructor is male. But, the worlds in which you believe the instructor is a male on the basis of the class being a philosophy class are not worlds in which the fact that the class is a philosophy class rules out the instructor’s being a woman.
That’s not just a quibble. I doubt there is any way to finish the analysis without including normative terminology.
Hi Alex
yes – I do think that one main way to avoid the argument is to try to avoid a commitment to the inheritance thesis in the way that Smith has formulated it.
One way you could do this would be to focus on evaluative truths other than truths about final value. These could be truths about thick evaluative properties (maybe pleasurable is such a property) that make things finally good. I think the cost of this would be that, if the argument works otherwise, we could only give a buck-passing view of the final good but not of the other thick evaluative properties. And as Pekka Vayrynen has argued, there will be problems if you go this way.
The other alternative would be to focus on non-evaluative good-making properties. I’m thinking that pleasant might one such property. Here the view would say that the reasonhood (for a final desire) of some consideration reduces to its supporting the truth that the desired state affairs has the given non-evaluative property (that is also good-making). I’ve got few concerns about this. One is that it seems to commit the buck-passers analytically to controversial views about what concrete, substantial first-order reasons there are just because they need a response to Smith’s objection.
Anyway, in principle, this is one line to take, but it will have its costs.
Hi Jamie
as far as I can see, Smith contributes the strong thesis to Hume but distances himself from that view to accommodate inductive and abductive reasons which I take it can be less than conclusive. Here’s Smith’s formulation of Lewis’s view which I just included in a very rough formulation (my apologies for the inaccuracy):
“We might suppose that some fact p is a reason for a subject to believe that q iff and because p is the sort of thing that could give a subject knowledge that p, where this in turn is explained by the fact that, in those possible worlds in which the subject does know that q on the basis of p, the fact that p removes all of the other possibilities except for q that the subject isn’t properly ignoring, where the norms of proper ignoring are semantic norms telling us when someone’s forming a belief in the ignorance of certain facts counts a knowledge.”
I’ve got some difficulties in parsing out the view here. As far as I see, this view allows for non-conclusive reasons and you seem right that we might need normative terminology in specifying what can be properly ignored. So, yes, you are right that my formulation of this theory was sloppy, but do you think that Smith’s proper formulation suffers from the same problem?
It’s not that we’ll need normative terms to say what can properly be ignored, or that ‘proper’ ignoring is already normative. It’s that evidence is normative. Leave all those things aside — those could be semantic or pragmatic norms anyway, not epistemic ones, as Michael notes.
I do think Michael’s formulation suffers from the same problem, although I find it a bit hard to understand so I may be wrong. Suppose, as seems to happen in the typical cases of epistemic reasons, that p is the sort of thing that could give a subject knowledge that q, but only if it were helped along by other stuff (other reasons, some enabling conditions). So in the other worlds in which S knows that q on the basis of p, p doesn’t remove all of the other possibilities at all.
For example, let p be the fact that the class in room 223 is a philosophy class, and q is the proposition that the instructor is male. So p is the sort of thing that could give you knowledge that q, but not all by itself. In those worlds in which you do know that q based on p, you have other reasons too, and maybe some defeater defeaters or other non-reason supplements. (One thing that makes it hard for me to understand is the idea of knowing something based on something; I don’t really understand what that is.)
So, p is a reason to believe q, but p doesn’t rule out the other possibilities, even in the worlds in which you know that q on the basis of p.
Hi Jamie
thanks this is helpful. I need to think about this more. Just two quick thoughts:
1. I can see that there might be a couple of ways to defend the view against this worry. One would be that the reason for belief is never properly just one atomic proposition but rather a full set of propositions that includes the other ‘reasons’ and other supplements. The other perhaps would be to take the subject’s background information to the relevant worlds and say that p removes all the other relevant possibilities assuming that background information. I’m not sure whether all the problems could be solved in these ways.
2. Assume that Smith’s view of epistemic reasons suffers from this problem. In this case, those reasons could not be reduced to pure entailment and truth, but rather only to evidence understood in some richer inherently normative way. I worry that this will still be enough for Smith to run the argument against buck-passing view. So, it’s not clear to me that the objection depends on such an austere reduction as Smith suggests.
Interesting.
Of course, one alternative is that what unifies ‘reason’ talk is what expressivists say unifies it. I know Smith doesn’t like that, but it does solve the ‘No Equivocation’ problem.
Hi Jussi
Its an interesting argument, but I’m not convinced it is entirely consistent.
If I understand the proposal correctly, Premise 2 says (something like) the reasons for belief are truths that entail the truth of the belief. The Inheritance Thesis says that reasons for desires must be the same sorts of things as reasons for belief, and so they must themselves be truths that have some entailment relation. But they can’t entail the truth of the content of the desire, so they must entail something closely related to the desire. The only plausible thing in the vicinity is a belief about what’s good.
This seems to me to involve its own equivocation, and thus it violates Premise 1, unless desires are themselves just beliefs about what’s good. Reasons *for* desire are not like reasons *for* belief. They are related to desire via truths about what’s good. Reasons for desires and desires have a very different relationship than reasons for beliefs and beliefs. If that degree of equivocation is accepted, I don’t see why we should buy Conclusion 1.
Derek: I’m just following Judy Thomson on this point. A single concept of a reason is in play, and that is the concept of a reason for being in a mental state with a correctness condition. A reason for being in a mental state with a correctness condition is a consideration that bears the (now let’s vague it up so that we can put to one side the issue that Jussi and Jamie have been discussing) appropriate truth-supporting relationship to the proposition that is that mental state’s correctness condition. Examples of mental states with correctness conditions are belief and intrinsic desire. The correctness condition of the belief that p is the proposition that p, so the reasons are the considerations that bear the appropriate truth-supporting relationship to the proposition that p. The correctness condition of the final desire that p is the proposition that p is intrinsically good, so the so the reasons are the considerations that bear the appropriate truth-supporting relationship to the proposition that p is intrinsically good. There is therefore no
equivocation.