Monday, June 22, 2009

Reasons as Facts or Propositions?

I've been re-reading Miller's "Motivation in Agents" along with Dancy's Practical Reality and I'm puzzled by a few things.

I think Dancy (2000: 131) is wrong to say that we can give a non-factive explanation of an agent's actions. While Dancy seems to think this is coherent, this strikes me as contradictory:

(1) Mustard’s reason for running down the hall was that the murderer was chasing him, but of course he was mistaken about that.

It doesn’t sound like a contradiction to say things like, ‘His reason for Φ-ing was that p, but of course he was mistaken about that because owing to self-deception he thought that his reason was q.’ Dancy’s suggestion, however, is that (1) could be true and it could provide us with a non-factive explanation (i.e., a successful explanation that lacks a true explanans) when it is not true that the murderer was chasing him, not when the agent has a mistaken belief about his own motivations. If (1) is contradictory, then it seems (2) could be true only if there was indeed a murderer chasing Mustard:

(2) Mustard’s reason for running down the hall was that the murderer was chasing him.

If it's part of the story, however, that Mustard was mistaken in thinking that the murderer is chasing him and (1) is false, then (2) is false.

If (2) is false, you might opt for the view that says that Mustard’s motivating reasons are either mental states or the contents of his states. Miller, seems to opt for this sort of view. According to Miller, "considerations such as she loves me or I ought to keep my promise could ... be among the kinds of considerations which might motivate me to act if I happened to believe them" (2009: 250). He raises this objection to the view that identifies motivating reasons with facts:
For unless we are infallible about what facts there are, there will be plenty of instances in which we invoke motivating reasons in our practical deliberation and yet at the same time are quite mistaken about the existence of the facts to which they make putative reference (2009: 229).

When Miller says above that motivating reasons are considerations which might motivate an agent to act, he's taking considerations to be propositions rather than facts. Sticking with Mustard, I take it that on Miller's view (2) could be true even if:

(3) No one was chasing Mustard.

Again, it seems (to me) that (1) is false. It also seems to me that (1) is entailed by (2) and (3). Doesn't Miller's view face pretty much the same problem that Dancy's view faces?

Bracketing this problem, I'm not entirely sure what Miller's objection in the passage above is supposed to show. Mustard is fallible. That's obvious. He believes he's being chased, but he's not. I take it that Miller thinks that Mustard's fallibility concerns matters of fact but he's not mistaken about what his motivating reasons are. (If Mustard were mistaken about both matters, there's no objection to treating reasons as facts in the passage above.) Why should we assume that Mustard's beliefs about his motivating reasons are true even when the beliefs that figure in practical deliberation are false? It doesn't seem to me that Mustard can say truthfully, 'My reason for running down the hall is that there's a murderer after me'?


Mike Almeida said...

That's an interesting problem! Suppose S and S' are identical twins. Smith is in love with S, but not S'. Smith says to S' (believing he speaking to S),

1. The reason I asked you to marry me is because I love you.

But that can't be the reason. He does not love S'. But consider the suggestion that the real reason is Smith's belief that he loves S'. If that were the reason in the case of S', it would also be the reason in the case of S. But it isn't. Consider (2), where Smith is now talking to S,

2. The reason I asked you to marry me is because I believe I love you.

That's not the reason he asks S to marry him! He does it because he loves S, not because he believes he loves S.

Maybe the reason for action in each case is the following:

3. Smith loves S and he believes (mistakenly or not) that the person in front of him is S.

In that case his motivating reason is true, despite his misidentification in case (2).

In the case you describe, the reason Mustard is running because he is being chased, and he believes (mistakenly or not) that the chaser is a murderer.

ADHR said...

Just with regards to Dancy, you have to keep in mind that, at the point he brings in non-factive explanations, he's already rejected psychologism about reasons and rejected "hybrid" views (which sometimes count psychological states as reasons, sometimes other than psychological states count as reasons). Since he's swept all that off the table, I'm not sure he has anywhere to go but to non-factive explanations, which involves a certain amount of bullet-biting. (Well, one alternative, I suppose, is Bittner's: that there just is no explanation.)

If you're strongly tempted to the idea that it's a (mistaken) belief that's the reason, then you're really getting off Dancy's train a lot sooner in the argument.

Clayton said...


I agree. In the paper I'm presenting at RoME, my beef with Dancy is that he doesn't take his view to what I think of as the view's logical conclusion. Look, if motivating reasons are the sorts of things that could be normative reasons and normative reasons are facts, the thing to say is that (1) and (2) are false because there's nothing we could describe as Mustard's motivating reasons. What Dancy should say (I say) is that in the good case we can explain an agent's actions in terms of motivating reasons and in the bad case we can't.

Of course, someone could object saying that it's just obvious that we can explain Mustard's conduct in terms of motivating reasons, but I suspect that this is because the conception of motivating reasons they work with is closer to something like reasons-why and those needn't be identified with the considerations that showed the agent's action in a positive light. On this conception, it's not at all surprising that such reasons aren't the sort of things that could turn out to be normative reasons. So, we either ought to follow those like Smith and (I believe) Stratton-Lake in saying that motivating and normative reasons belong to different ontological categories or go a bit more radical than Dancy and say that when the subject's beliefs about the practical situation are false, we can't explain the subject's actions in terms of reasons.

Hey Mike,

If I'm reading you right, are you suggesting that Smith's reason is a fact he believes?

I couldn't quite get the description you gave of Mustard's reason. Are you of the view that when nobody at all is running after Mustard, Mustard's reason for running is that someone's chasing him? Or, were you supposing something like this: a non-murderer is chasing Mustard and Mustard's reason for running is the fact that he's being chased coupled with the belief that the chaser is a murderer (and one who is not trying to return his wallet).

ADHR said...


I see what you're saying. Have you read Bittner's Doing Things for Reasons? He advocates pretty much that solution, i.e., that in the case where the belief is mistaken, then there's not really a reasons-explanation for the action. He does hedge a little, though, and refers to a sort of family resemblance kind of explanation. That is, while we can't explain Mustard's action, we can explain non-deviant cases -- cases where Mustard's belief is right -- and use that explanation in the case at hand as a "sort of, but not quite" explanation.

That said, I'm not sure that there's anything illogical about what Dancy is saying. It's weird, of course. But I suspect he'd say the weirdness arises from our pre-theoretical commitments to factive explanations. So, we could look at Dancy as presenting us with something of a challenge: if non-factive reasons-explanations aren't a legitimate solution, what is it about non-facticity that's problematic? Dancy might be able to get some support from issues in scientific explanation, where explanations can be advanced that either involve frankly fictional or highly approximate explanans.

Clayton said...


I haven't read the Bittner book, but it looks like I should.

My impression (which is probably based on nothing) is that Dancy relies on non-factive explanations to address a problem, but is not entirely in love with the notion. I'm critical of (1), but only because I pretty much buy everything else that Dancy says and don't quite see the need to say something as weird as (1). (Moreover, I don't see the advantage of Miller's view. If (1) isn't all that strange, what possible reason could there be to prefer factive explanations w/ propositions rather than facts to non-factive explanations with facts rather than propositions?)

ADHR said...


The Bittner book's fun. And, it's a pretty short and punchy read.

I agree the propositions solution is unworkable. As you say, if it's not weird to admit what looks like a non-factive explanation, then invoking propositions isn't an obviously necessary move. To my eye, it looks like some ad hoc jury-rigging the notion of facts in order to force the explanation to be factive -- which, if I'm right, would be begging the question against Dancy.

I'm not sure whether it's weirder to say some actions cannot be explained by reasons (or, perhaps, can't be fully explained by reasons) or it's weirder to say that some actions are non-factively explained by reasons. I think here the causalist intuition becomes a problem. Whether or not explanations are all causal, there does seem to be something to the idea that any action which happens is something we could explain somehow. So, there'd have to be some careful parsing of the difference between reasons-explanations (which can't always be offered) and causal explanations (which can be).

Clayton said...

Agreed about the causalist intuitions. Here's a line I've been thinking about.

Paraphrasing Williamson, if someone said 'Explain World War I', the request only makes sense if the context helps us see what it is about WWI that the speaker wants us to explain (e.g., why it occurred then and not earlier, why it lasted so long, why it is so important to you on a personal level, etc...). Similarly, if someone says 'Explain S's A-ing', it's not clear what it is that they want explained, but it's clear that S's A-ing cannot be the explanandum. If we want an explanation as to why an instance of A-ing occurred, we might supply a causal story. If we want to know what it is that S got out of A-ing, we want something else.

Suppose someone said that the thing we want to learn from a reasons explanation is this: what it is that S got out of A-ing.

If this is what we were after, the causal stuff seems not to matter so much but it would matter whether the agent's grasp of the situation was accurate or not. So, if we want to know what it is that Bernard Williams got out of drinking the contents of that glass, we'd mention the gin and perhaps the desire for it.

Suppose I know that Bernie's unlucky twin was served petrol with limes and tonic. Since I know there is no gin in the glass, I know (if my view is right) that Bernie's twin's reason for drinking the contents of the glass is not that he would drink a gin and tonic. So, what I do is take aim at a different explanandum--the bodily movements, the consumption of the petrol, etc...

Here's some evidence that this picture is right. Consider this exchange:

You: Why did he drink that?
Me: Oh, Bernie's twin loves piney drinks. His reason for drinking that was that the glass contained a gin and tonic.
You: But that was petrol!

It seems that your second assertion contradicts mine. I wouldn't then say that Bernie's twin had a reason to drink the stuff, I'd say there was a reason why he drank the stuff. If reasons explanations are factive and motivating reasons are facts, perhaps a speaker's beliefs about the agent's practical situation will naturally guide them to explain why an action of a type occurred in terms of reasons why or what it is that the agent got out of acting in terms of the agent's motivating reasons.

(Obviously, this is all very provisional.)

ADHR said...

I'm with you right until the move at the end, where you introduce that distinction between Bernie having a reason to drink the petrol and there being a reason why Bernie drank the petrol. If reasons are facts, and not psychological facts at that, then how is there a distinction here? For on that view, it seems that Bernie's having a reason to A entails that there is a reason why Bernie A'd.

Thinking about it, I wonder if your point is that the entailment doesn't go the other way: even if there is a reason why Bernie A'd, it may not be the case that Bernie has a reason to A. But, if that's the picture you want to sketch, I'm not clear what makes the reason a reason (rather than, say, a cause). After all, ex hypothesi, it's not anyone's reason; it's sort of floating around on its own.

On an unrelated note, I also wonder about the claim that the difference between a causal-explanation-seeking why-question and a reasons-explanation-seeking why-question hinges on the explanandum, i.e., whether we are being asked to explain A-ing as an event (something that occurred) or as something geared towards an end (done for the sake of something else). I worry it's conceding a lot to the causalists. If I A in order to get B, then I must form an intention C to get B; that intention would then be realized in my A-ing for B. But why not go the causalist route and say C is both the cause and the reason for why I A?

(I'd come up with a more evocative example, but it's 4 am here and I've spent the evening grading!)

Mike Almeida said...

Or, were you supposing something like this: a non-murderer is chasing Mustard and Mustard's reason for running is the fact that he's being chased coupled with the belief that the chaser is a murderer (and one who is not trying to return his wallet).

Sorry for not following up so quickly. Yes, this is what I had in mind. So, the reason for running is the fact that he is being chased and the fact that he believes the person chasing him is a murderer. The reason is not that he is being chased by a murdered. Does that make sense?

Novice said...

Forgive me for my ignorance I'm new to academic philosophy but an alternative theory comes to mind, I don't remember where I heard this.

If our reasons to do something are mental states and are mental states are reflections of reality then are motivations are not facts but based on/directly influenced by facts.

So you would say:
Mustard is running down the hall because he believes he is being chased by the murder.

Normative facts could then influence psychological states without being themselves psychological states.