Print This Post Print This Post
Thinking About Reasons
By Antti Kauppinen

Expressivist accounts of normative judgment typically (always?) begin with all-things-considered verdicts: Hurrah (helping old ladies cross the road)! Boo (getting your little brother to murder)! But of course, many normative thoughts are not all-things-considered. I think there is some reason for me to go to bed early, and some reason for me not to do so. When I deliberate, I try to figure out which of these is stronger, and so arrive at an all-things-considered judgment.

Here is a partial list of things that an account of thoughts about reasons should explain:

1) Pro tanto judgements: It is possible to think that A has reason to φ, although A ought not φ, all things considered.
2) Comparative judgments: It is possible to think that r is a stronger reason to φ than r’ is to not φ.
3) Non-additivity: It is possible to think that r is a reason to φ and think that one has more reason to φ in the absence of r.
4) Lack of isomorphism: It is possible to think that A has reason to φ, although A ought not be at all motivated to φ. (First-person corollary: It is possible to think I have reason to φ without being either akratic or at all motivated to φ.)
5) Generality: It is possible to have reasons to act without deliberation or decision, as well as reasons to feel in a certain way.

Jonathan Dancy has argued, both in Ethics Without Principles (OUP, 2004) and ‘What Do Reasons Do’ (in Horgan and Timmons (eds.), Metaethics After Moore, OUP 2006), that expressivists fail to account for 1, judgments about pro tanto or ‘contributory’ reasons, even if they can explain overall ought-judgments. The ultimate explanation for this is that it is in general not possible to explain contributory reasons in terms of what we ought to do. An expressivist will have to assume otherwise, since she is bound to work with attitudes of all-out approval or disapproval, like accepting norms for guilt and anger (Gibbard).

To see the problem, consider a simple case. George has promised John to help him move, but has to abandon this plan in order to save Ringo from drowning. I think George does the right thing, though his promise gave him some reason to help John instead. What does my thought consist in, on the expressivist account? It certainly cannot be that I accept a norm requiring George to feel guilt for not helping John – since I think he did what he should have done, I don’t think he should feel guilty. Nor should anyone be angry with him for failing to keep his promise, in these circumstances. Quite the contrary – he would have acted wrongly, had he kept his promise. Yet I continue to think he had a reason to help John. What, then, is the attitude that constitutes such a thought? In Dancy’s words, “The difficulty is to make sense of the idea that one might approve of S’s φ-ing to some extent while being entirely against it.” (Dancy 2006, 57)

Gibbard has an answer to this: “To say that R is some reason for S to φ in C is to express acceptance of a system of norms that directs us to award some weight to R in deciding whether to φ in C” (Gibbard 1990, 163). Putting this directly in moral psychological terms, when I think that the fact he promised so is some reason for George to help John, I accept a system of norms that directs George to award some weight to having promised in deciding whether to help John. More simply, I approve of George giving weight to the promise in deciding. So here we indeed have an all-out attitude, approval or acceptance of norms, not toward an action but rather toward awarding weight in deciding to act.

Dancy rejects this analysis for a number of reasons. Here’s a reconstruction of one argument:

Dancy’s Dilemma
a) To award weight to r in deciding whether to φ is either to think that r favours φ-ing or give consideration to whether r in deciding to φ
b) If the former, the analysis is uninformative (it makes appeal to thinking that r favours φ-ing, which is precisely what we’re trying to understand)
c) If the latter, the analysis if false, because it is possible to award weight in this sense without thinking that r is a reason to φ: “I might think, for instance, that I should always consider the question whether R if I am deciding whether to φ in C, but suppose nonetheless that on many occasions the question whether R or not will be in fact irrelevant to how I ought to act.”
d) So, we can’t understand reason judgments in terms of approving awarding weight in deciding.

I believe Gibbard would deny a) (he does something of the sort in Thinking How To Live), and with good reason. But rather than trying to extract a rival interpretation of Gibbard, I’ll rather offer an alternative expressivist formulation that is designed to deal with all of 1-5 above and avoid Dancy’s challenge. I will assume, for the sake of argument at least, that expressivists can make sense of all-out normative judgments. Here goes:

Expressivism about reason judgments
To judge that r is a pro tanto reason for A to φ in C is to approve of A’s awareness of r having causal influence in the direction of φ-ing on the mechanism(s) responsible for A’s φ-ing or not in C.

This is a permissive formulation. For judgments about peremptory or requiring reasons, substitute ‘disapprove of A’s awareness of r not having causal influence’. It is a nice feature of the account that it allows for making a distinction between thoughts about how demanding reasons are.

Note that I take awareness to be factive: if r is not the case, A can’t be aware of r.

How does this work with the desiderata above?
1) Pro tanto judgments: On the expressivist view, let’s say, to think A ought to φ is to disapprove of A’s failing to φ. (Again, we’re assuming an acceptable account of all-out approval and disapproval.) To think A ought not φ is to disapprove of A’s φ-ing. So, to think George ought not help John is to disapprove of George’s helping John. As desired, this is compatible with approving of George’s awareness of the fact that he promised to help John having causal influence in the direction of helping John on the mechanism responsible for whether he helps John or not. Let’s say the mechanism is his deliberative system. If I think George has some reason to help John, I approve of his knowing he promised influencing in the direction of helping, perhaps by creating deliberative pathways that remain blocked in the actual situation (by the more urgent reason), but would lead to motivational pull in a host of nearby possible worlds. I might even disapprove of George if his promise made no difference to the way he arrives at the choice to save Ringo instead.

2) Comparative judgments: The account extends naturally to comparison. To think that r is stronger reason to φ than r’ is to not φ is to approve of the following scenario: awareness of r influences the mechanism responsible for φ-ing in the direction of φ-ing to a greater degree than awareness of r’ influences the mechanism in the direction of not φ-ing. If the mechanism is deliberation, it is to approve of knowledge of r having a stronger causal influence toward deciding to φ than knowledge of r’ has toward deciding not to φ. To think that the fact that Ringo is about to drown is a stronger peremptory reason for George than the fact that George has promised to help John is to disapprove of awareness of Ringo’s drowning not having more influence on George’s decision than awareness of his promise. To think that George has most or overall moral reason to jump in the water is to morally disapprove of some set of considerations not having decisive influence in leading George to jump in the water – in short, to morally disapprove of George not jumping (or not trying to jump) in the water. We’ll want comparative judgments to lead to overall judgments in some such way.

3) Non-additivity: Dancy argues that there are cases in which something is a reason to φ, though in its absence there would be even more reason to φ. His example is along these lines: The fact that someone is my friend might be a reason for me to give him a hand getting up – but if he wasn’t my friend, I might have even more reason to give him a hand getting up (it would be more admirable etc.). This rules out accounts on which thinking that r is a reason to φ is approving of φ-ing more on account of r than otherwise. (Such an account would meet desideratum 1, since it is possible to disapprove of φ-ing all things considered while nevertheless approving of it more in the presence of r than otherwise). This is not a problem for the present suggestion, however. To think that the fact that Ringo is George’s friend is a reason for George to help him is to approve of awareness of it influencing George’s decision to help. To think that were Ringo a stranger, George would have even more reason to help, is to approve of the awareness that Ringo is a stranger having more influence on George’s decision in counterfactual scenario than the actual awareness has in the actual world.

4) Lack of isomorphism: I take it that it’s possible for awareness of r to have causal influence on a mechanism responsible for φ-ing in the direction of φ-ing in other ways than giving rise to motivation to φ. Influencing the structure of deliberation, for example, can count as relevant influence without counting as motivation. This way we avoid having to think that for every reason, we must expect at least a tiny bit of motivation.

5) Generality: The account is formulated in terms of mechanisms responsible for φ-ing or not to include emotions and non-deliberate actions. The fact that I got a job interview can be a reason for me to feel happy. To think so is to approve of my awareness of this fact causally influencing my mood mechanism, say, in the direction of happiness.

Finally, the proposal avoids Dancy’s Dilemma. To approve of r having causal influence in, say, deliberation, is not merely to approve of considering (entertaining?) r when deciding whether to φ, so an analogue of c fails to apply. Nor is it to approve of think of r as a favouring consideration, so an analogue of b fails to apply.

So what’s wrong with it?


  1. 1. Posted by John Turri | September 15, 2009 2:13 am

    Hi Antti,

    Is your account consistent with (a) there being a reason for you to φ that you’re presently unaware of, or (b) there being a reason for you to φ that isn’t presently causally influencing the φ-relevant mechanism?

  2. 2. Posted by Clayton Littlejohn | September 15, 2009 2:55 am


    This is an interesting suggestion:

    “Expressivism about reason judgments
    To judge that r is a pro tanto reason for A to φ in C is to approve of A’s awareness of r having causal influence in the direction of φ-ing on the mechanism(s) responsible for A’s φ-ing or not in C. ”

    I don’t have any clear sense as to how these sorts of cases are described, but suppose that r is a pro tanto reason for A to do X in C but there’s another reason, r’, that is a stronger reason for A to do Y in C. I guess I don’t see that I’m approving of r having some (but not sufficient?) causal influence in the direction of X-ing in C. Part of this is probably that I don’t quite get what it means to say that r has some causal influence towards X-ing when the agent Y’s rather than X’s. Part of this is probably based on the intuition that if it is just obvious that A ought to Y rather than X I guess I don’t know that I’d think favorably of A’s being causally influenced by r to do what is obviously the thing that A oughtn’t do. Yet, I feel like I do have some sense of why I’d think of r as a pro tanto reason to act in those circumstances. Maybe if you spelled out this idea of r having causal influence to do what the agent ends up not doing. Is that just the agent’s feeling the tug towards doing what r counts in favor of?

  3. 3. Posted by Jussi Suikkanen | September 15, 2009 7:28 am

    I think there’s bound to be problems if the analysis does not take into account the universality of reasons judgments. So, imagine that A sees a snake and this makes him jump back to protect himself. Unfortunately by jumping back he steps on a land-mine. Imagine also that I don’t like A at all and thus want him to get killed.

    In this situation, I approve of A’s awareness of the snake having a causal influence in the direction of jumping on the mechanism that is responsible for whether A jumps or not. However, I have not judged that awareness of the snake is a reason for A to jump.

    So, I think you have to add something like ‘approves of anyone’s in awareness in A’s situation…’. Given that I don’t approve of the causal influence in my own case, in this case the example would not go through.

    I’m also worried about deviant causal chains and the lose formulation of the mechanism. Remember the study where people who crossed a scary bridge were more likely to ask out the research assistant who was asking the subjects random questions than the people who hadn’t crossed the bridge. In this case, the awareness of the bridge may causally influence the mechanism that is responsible for asking the assistant or not to the direction of asking out. And, we may also approve of this. But, we don’t think that the hight of the bridge is a reason to ask the assistant out. So, somehow the analysis should specify the correct deliberative mechanism and how the awareness of r should affect it.

  4. 4. Posted by Antti Kauppinen | September 15, 2009 9:39 pm

    Thanks for all the quick comments – and apologies for the slowness of the blog, which resulted in some delay before I or anyone could see them.

    John, short answer: yes. At least I take it I can approve of your φ-relevant mechanism being influenced by your awareness of r even though you’re not in fact aware of r. If that sounds odd, I’ll need to add ‘influenced by A’s (possibly counterfactual) awareness of r’ as a qualification.

    Clayton, I certainly don’t want to say the agent needs to feel a tug for every pro tanto reason – that would be going the isomorphism way rejected in 4. (I would reject even the weaker version that Mark Schroeder defends.) If we stick to the agent’s deliberative system as the paradigm φ-relevant mechanism, I think it can be influenced in the direction of φ-ing by other things than motivational states like desires (which themselves need not be felt, as many have emphasized). Say that the fact that it’s sunny is some reason for your to play tennis after teaching. At the same time, the fact that tonight is an important deadline is a much stronger reason for you to stay in your office all night. And not only are they competing reasons for you, but I think they are.

    Now, suppose that you know it’s sunny. Perhaps this awareness doesn’t give rise to a desire to play tennis, but gives rise to some wistful tennis-related thoughts, or better yet, causes playing tennis to appear as an option in the background options file in your deliberative system. (The background options file is what you reach for if your current plan reaches completion before you’ve scheduled anything else.) I think these are ways of influencing the mechanism in the direction of φ-ing without counting as motivation to φ, and certainly not a felt pull towards it. If I approve of your awareness of r having such influence on your deliberative system (in this case), I regard r as a reason for you to φ – even if I wouldn’t approve of your actually φ-ing, since I think you have much stronger reason to not φ (in this case finish the paper by the deadline). Could I consider the fact that the sun is shining is some practical reason for you to play tennis if I didn’t approve of some such influence? The expressivist is bound to answer in the negative, and that doesn’t seem so implausible to me.

    Jussi, interesting points. I wanted to take care of generality by placing conditions on approval (more precisely, on its characteristic aetiology). Very very roughly, the approval would only count for reason judgments if any unbiased and informed observer would approve, either in the position of those affected (in the moral case) or in the position of the choice-actualized future self of the agent (in the prudential case). That would take care of the snake example – to make a prudential reason claim, I’d have to approve of awareness of the snake causing the agent to jump back after having adopted the perspective of his informed future self. More traditional expressivist views of approval would also handle it – you wouldn’t accept a norm in Gibbard’s sense, or disapprove of others not approving of the influence, as on Blackburn’s story.

    In the shaky bridge case, if I do approve of awareness of the scariness of the bridge influencing asking out – and let’s bear in mind that the relevant sort of approval is not just the absence of disapproval, but the presence of a complex set of attitudes that constitutes a normative stance – am I really not thinking that the fact that the bridge is scary is a reason for the guy to ask the girl out? I personally don’t, in this sense, endorse the causal influence, and I also don’t think that the scariness is a reason for the guy. Presumably, and this is of course the point of the experiment, the guy wouldn’t approve of the influence himself – this is why the study is used by skeptics about rational control. If he did approve of it, wouldn’t that amount to him endorsing the scariness as a reason for asking out?

    If this isn’t convincing yet, take the stronger formulation: would you disapprove of the awareness not having the influence it does? Surely not. But suppose you did, and on top of that as a result of at least trying to consider the matter from an informed and unbiased perspective, which would explain why regarded your attitude as carrying some authority. Wouldn’t you then necessarily consider the scariness of the bridge as a reason for the guy to ask the girl out?

  5. 5. Posted by Jussi Suikkanen | September 17, 2009 9:44 pm

    That seems like the right sort of move to make about the snake case even if I worry that this might no longer be an expressivist account.

    I’m not sure about the bridge though. That sounds still somewhat counterintuitive. The scariness of the bridge seems to be a wrong kind of consideration for being a reason no matter how much a approve of its causal influence. Don’t seem to count in favour of asking the girl out. I’m sure we can tell a story where it’s clearer that we would have the approving attitudes – the guy we are talking about would never have the courage to ask the girl out without the adrenalin, they would live happily ever after and so on.

    Also, while I was away, I thought that your account would face the classic surprise party objection.

    I happen to think that the fact that there is a surprise party at Ben’s house for him is a reason for Ben to go home. But, surely I don’t think in any way, positive or negative, that the fact that there is a party at Ben’s house should play any role in Ben’s deliberation about whether to go home or not. That would spoil the surprise.

  6. 6. Posted by Antti Kauppinen | September 18, 2009 2:00 pm

    Ah – the surprise party! I was waiting for someone to bring it up. Now, the answer is quite straightforward in the normal case, in which I think people have reason to go to parties for them, whether they’re a surprise or not. Then I endorse awareness of the party making a difference to whether Ben goes home to Carrboro or not. Of course, I know that if he’s aware of it, he won’t go, but to me that’s just another case of failing to take one’s reasons into account.

    But suppose I think people (or maybe just Ben) only have reason to go to a party if it’s a surprise to them (that is, if they’re not aware of it). This is a real challenge. It is partly for analogous reasons that I didn’t formulate the proposal in terms of deliberation: some things might only be worth doing spontaneously. So let us be careful. What exactly is the reason I think Ben has in this situation? It seems that it is the fact that there is a party for him at his house that he’s not aware of. And my proposal says that when I think this is a reason, I endorse the influence of his (possibly counterfactual) awareness of the fact that there is this very party for him he’s not aware of on his homegoing-relevant mechanism (which in this case may be just habit – another reason not to talk of deliberation).

    Now, it seems impossible for one to be aware of p that she’s not aware of. (Similarly, you can’t know a fact you don’t know – though you know there are some facts you don’t know.) But were it possible to be aware of something that you’re unaware of, I would approve of your awareness of the surprise party influencing you. So maybe the proposal could be saved with the addition of a ‘perhaps per impossibile’ clause. Otherwise things get really tight for the expressivist.

    It is perhaps worth noting that here what we might call a comparative approval model would give a more straightforward answer. It seems natural to say that given the fact that there is a surprise party for Ben, I endorse his going home more strongly than in the absence of the surprise party. This approach, as I pointed out, fails to account for desideratum 3, non-additivity. It also requires us to be able to make sense of very fine-grained differences in the level of (dis)approval of actions.

    As to the bridge case, I still think we don’t approve of the influence of the awareness that it’s dangerous in the relevant sense, which is more like endorsement. We don’t criticize the guy for being influenced by it, to be sure, and we may think he’s quite lucky to have been influenced by it, if it leads to a happy marriage. It turns out he did have plenty of reason to ask the girl out, though the fact that they met in a dangerous place wasn’t among them. As I suggested, the difference between approving of something as a lucky influence and as a reason to act will be clearer if we go negative. We wouldn’t disapprove of the guy if he wasn’t influenced in the direction of asking the girl out as a result of being aware of the dangerous situation.

    Incidentally, I’m not sure if the case is correctly described as guy being influenced by his awareness that the bridge is scary or high or dangerous. Rather than the awareness, the influence is hormonal – he needn’t be aware of why for a burst of adrenaline to take place. Whether we approve of such hormonal influence is neither here nor there from the perspective of the proposal.

  7. 7. Posted by John Turri | September 18, 2009 8:51 pm


    Something else occurred to me about your proposal.

    To judge that r is a pro tanto reason for A to φ in C is to approve of A’s awareness of r having causal influence in the direction of φ-ing on the mechanism(s) responsible for A’s φ-ing or not in C.

    Suppose I think that recognition is the first step on the road to rehabilitation. You must first become aware of how certain factors influence your behavior, in order to become a good person again.

    Now suppose I’m considering A’s behavior. As is his habit, he’s once again doing something for a bad reason: he’s φ-ing for reason r again. And while I don’t in the least think that r is a pro tanto good reason to φ, I do approve of A’s awareness of r’s influence on A’s φ-ing (for the general reason indicated in the previous paragraph).

    Doesn’t your view give the wrong result in this sort of case?

  8. 8. Posted by Jussi Suikkanen | September 18, 2009 9:57 pm

    I was thinking of the second kind of party case. I’m worried if the expressivist view would be based on impossibilities.

    I think there is a more serious theoretical problem behind the surprise party situation. As you know, there’s two types of internalisms; judgment-internalism and reasons-internalism (or existence-internalism). Much of the motivation for expressivism is seems to be based on the judgment-internalism in Blackburn and Gibbard. Now, a lot of people think that these two internalisms go hand in hand, and your expressivist take on reasons seems to support this. I’m not sure though why we should think this. The Williams-styled reasons-internalism seems to be a substantial view about what reasons there are, and I don’t really see why a semantic, metaethical view should commit one to such a substantial theory about normative reasons. I think I’ve seen Daniel Elstein give a paper where he argued that expressivists should not do this and this seems right to me.

  9. 9. Posted by Jussi Suikkanen | September 18, 2009 10:02 pm

    Now your answer in the bridge case also seems to be going to right direction. You are saying something about the right kind of causal influence. It has to be direct and not mediated by the mediating effects of an adrenalin rush. I’m not as sure about the first point though. I don’t think it can be a condition for thinking that someone has a reason that you would criticise the person for overlooking the consideration especially when we are talking about the pro tanto.

  10. 10. Posted by Antti Kauppinen | September 19, 2009 3:38 pm

    Thanks for the continued interest, John and Jussi! A few more replies.

    Jussi, why do you think the account commits to internalism of either sort? I can’t see it at all. I can endorse awareness of r influencing you in the direction of φ-ing regardless of what’s in your actual subjective motivational set. And when I do so, I don’t necessarily endorse the awareness motivating you – that’s the point about the lack of isomorphism.

    I guess adding a ‘in the right way’ clause wouldn’t make much difference. Everybody waves their hand in that direction anyway. But I’d just rather say that in the adrenaline case, it is the hormone rather than the awareness that’s influencing the agent (note that you could get the adrenaline rush without being aware of the height).

    Lilian pointed out to me that there’s an alternative response to the surprise party cases, which is flat out denying that the agent has a reason. After all, it’s not an implausible constraint on reasons for action that the agent should be at least in principle able to act on them. And this type of ‘reason’ you couldn’t act on. We might want to say that ‘There is a reason for A to φ’, but this could amount to saying that φ-ing would be good for the agent (to some degree), which would be a different kind of judgment. (This kind of response would indeed draw on what’s plausible about Williams’s internalism.)

    John, I think we have to be careful about what the object of awareness, and thus the reason, is in these cases. I think in your type of case we have a second-order awareness, so to speak, whose influence you approve of. So, Michael is a heavy drinker. His awareness of the fact that the Liquor and Guns Warehouse has a deal on Night Train causally influences his going there. This you don’t approve. But after attending some therapy as part of his parole program, he becomes aware of the detrimental influence that thinking about bargain spirits has on him. What you do approve of is this second-order awareness making a difference to his behaviour, most likely by way of putting such thoughts out of his mind, or by way of focusing on the long-term damage that acting on those thoughts has.

  11. 11. Posted by John Turri | September 20, 2009 6:49 am


    I agree that you could deal with cases involving higher-order awareness that way. But in the case I was imagining, I approve of A’s first-order awareness of r’s causal effects on A’s behavior (the same sort of awareness featured in your thesis statement). It continues to seem to me that your view gives the wrong verdict in this sort of case.

  12. 12. Posted by Jussi Suikkanen | September 20, 2009 9:26 am

    The internalisms? Well, the something akin to the judgment-internalism I take it is built into the attitude of approval. To make a judgment about reasons is to approve (which is desire-like in the direction of fit) of something.

    The reasons-internalism is of course different as it is third-personal. I didn’t think that anything exactly like Williams’s internalism would follow. It seems right that on your account the awareness of p can have a causal influence by creating new desires and motivations. But I was thinking that the account limits the reasons to things the person whose reasons we are talking about can be aware. This is something the view shares with internalism and something that externalists will want to deny.

    I know that the surprise party case is just a silly case in which you say what Lilian says. But, it does illustrate the structure of the problem. There are though more serious cases of with the same structure (I take it that we are talking here about the conditional fallacy). Ignorance and learning cases are another one where this comes up (I wish I had Robert Johnson’s paper here). The idea is that facts such as that I don’t yet know that Obama has dropped the missile defense system can be a reason for me to read today’s paper even if it would be odd to approve of that fact having a causal influence on me deciding to read the paper. If I were aware of that fact, there wouldn’t be a reason for me to read the paper. And, so it goes for all facts that I am unaware of. On this view, my ignorance of them could never be a reason.

Post a comment

Name: (required)

Email Address: (required) (will not be published)



(Spamcheck Enabled)

This work is licensed under a Attribution-NonCommercial-NoDerivs 3.0.