I’ve just been reading Christine
Korsgaard’s new book Self-Constitution – Agency, Identity, and
Integrity. It’s a marvelous
book, one of the best I’ve read in long time. It presents an
appealing picture of actions, agency, practical deliberation,
normativity, morality, and interpersonal relations, it draws from the
history of philosophy in a rich way (not just from Kant, but also
from Plato, Aristotle, Hume, and many others), and it’s wonderfully
written. I do have a problem though. I don’t understand her
argument against standard moral realism (which she calls dogmatic
rationalism). I would be very thankful if someone could help me with
this. It doesn’t matter if you haven’t read the book yet, I’ll
try to give the argument in full.
This
argument is in section 4.2 “The Rationalist Account of Normativity”
(pages 64–66). The realist/rationalist Korsgaard has in mind
believes that there are facts about which actions are right and about
what counts as a reason for what. She emphasises that her objection
to this view is not based on the idea that we could not be motivated
by such facts but rather on the idea that such facts could not be
sufficiently normative for us.
The argument begins from the
assumption that, for the realist, acting on the basis of these moral
facts requires applying them (or the knowledge of them) in particular
instances. We furthermore apply the knowledge of the moral facts of
what is right by choosing those actions.
Korsgaard then claims that the first
problem with this view is that it fails to account for the
normativity of the hypotetical imperative. The only alternative,
according to her, is for the realist to claim that there is a
normative fact according to which one must take the means to one’s
ends. This is rationally required and what one has reason to do. And,
here is the problem:
“How does he act on this
recognition? How does he apply it to the case at hand? The problem is
that the extended version of the hypothetical imperative is itself…
the principle of practical application. But we cannot explain explain
how we are motivated to act on the hypothetical imperative, much less
how we are bound by it, by appealing to the hypothetical imperative
itself. We would have to say that an agent’s end is to do what is
rationally required, and that he sees that actions conforming to the
hypothetical imperative are rationally required. Taking the means to
his ends is therefore itself a means to his end of performing
rationally required actions, and he chooses to conform to the
hypothetical imperative under the influence of – the hypothetical
imperative itself. The point is that the hypothetical imperative
cannot be a normative truth that we apply in practice, because it is
the principle in accordance with which we are acting when we apply
truths in practice”.
According to Korsgaard, this is an
incoherent way of thinking about the normativity of the hypothetical
imperative. It would be binding only if we were already committed to
doing what is rationally required [I’m not sure I see where this
follows from]. And, this creates allegedly another problem:
“[T]he realist picture works no
better for moral principles than it does for the hypothetical
imperative. For even if we know what makes an action morally
required, so long as this is just a piece of knowledge, that
knowledge has to be applied in action by way of the hypothetical
imperative. But the hypothetical imperative cannot itself bind us to
do what is morally required. There would have to be some further
oblivation to take the performance of morally required actions as our
end. And this further oblivation could not just be another piece of
knowledge, for the same problem would the arise once again. What
obligates us to apply it?”
This leads her to conclude:
“So long as we think of the
principles of rational action as something that we apply – and
therefore as something that we may or may not apply – when we
deliberate about what to do, then we cannot really be obligated to
perform morally required actions. For either we are obligated to
apply the standard for morally required action or we are not. If we
say we are not, morality is not normative, whereas it we say that we
are, because the standard is known to hold, the whole argument starts
all over again – why must we apply this piece of knowledge?”
I have real difficulties in
understanding this argument and its point. As far as I can, it seems
to claim that moral realism leads to a vicious regress of
justification. Say that I am morally required to do x and that ying
is a means to do x. In order to do x, I must thus apply the
instrumental principle of taking the means (y) to the required end
(x). Call this application of the instrumental principle (the
mental?) act z. Why should I do z? Because doing z is a means to y
which is means to x. Doing z thus requires applying the instrumental
principle of taking the means to the required end again. Call this
second application of the instrumental principle (the mental act?) p.
Why should I do p. Because, doing p is means to doing z which is
means to doing y which… And, now we see where this is going.
That’s the best I can do to
understand the argument. Whatever the realist says, there will be
always a new why question to which she has to give an answer. And,
whatever answer she gives to this question, it’ll always gives rise
to a new why question.
Here’s why I think this argument
fails. There are both vicious regresses and non-vicious regresses.
The difference is that in the vicious regress, on every level, a new
consideration must be introduced to answer the objector’s
challenge, and yet this answer allows the objector to push the same
question. Thus, one must give another answer that is based on
something else, and there will be a new challenge to this answer
again. As a result, there is always an explanatory burden which one
fails to dissolve. Yet, in the benign regresses no such explanatory
burdens are left over because one has a standard answers to the
challenges on each level. And, I’m not sure why the realist cannot
hold this view here. There is some fact that such that it is a reason
for doing x. It, as a reason, makes x morally required. Why cannot
the realist insist that this very same consideration is also a reason
to do y which is a means to doing x and a reason to use any
principles of deliberation one needs to apply to get do y in order to
x. That someone will not be hurt is a reason not to drive
dangerously, to take driving lessons which help one not to do so, to
think about the ways in which one can become a safer driver, to use
these means, and so on. I don’t see any Why question left
unanswered here. But, I’m afraid that I haven’t understood
Korsgaard’s argument and would much appreciate any help.
Dear Jussi,
I think Korsgaard’s reply to your suggestion that the regress is benign would be that what you’ve achieved is a kind of draw: the moral realist can keep giving the same answer again and again but the moral skeptic can keep asking the same question again and again—with neither of them doing anything obviously defective. But a draw isn’t good enough for Korsgaard. To explain why something is a reason, you have to convince the normative skeptic that it’s a reason. One might think this standard is too high, but I’m not so sure. What I think is going on here is that she thinks you can’t move from facts (is-statements) to practical decisions (ought-statements perhaps) without taking ought-statements as bridge principles, and the way you show this is to imagine someone who doesn’t endorse any ought-statements at all. If you can’t convince him, then your explanation of why something ought to be done isn’t complete. (It may convince other people but only because they are assuming background ought-statements.)
I think this is a newer version of the argument in Sources of Normativity when she talks about the Open Question argument (1.4.8 I believe). So looking at that may be helpful.
Also, Mark Schroeder has an article on the kind of argument I think this is, its history, and some of its problems: http://www.jesp.org/PDF/JESP%20Cudworth.pdf
Best,
Mark Schroeder’s Cudworth paper (if I remember it from the talk version I heard way back when) seems to me to be exactly on point. At what stage in the realist’s argument for her position is Korsgaard’s objection to get a grip? Suppose, crudely, that the realist takes herself to have successfully shown that what makes moral-ought statements true is some complex property M’s being instantiated and what makes claims about rationality true is some property R’s being instantiated. To take a moral example, suppose that some act (or, more plausibly, act together with circumstances and facts about contextually selected possible actors) in fact instantiate M. How does the regress get going here? The instantiation of M makes the relevant moral ought-statement true. Her question surely can’t be ‘how can a true, moral-ought statement be normative for one of the relevant possible actors?’ Her argument looks to run through some claims about what’s involved in applying moral principles, but I don’t see how the realist needs to address that issue to answer the underlying challenge. I guess the issue turns on what she means by “sufficiently normative”. Admittedly, I approach these issue sideways, more as someone trained in phil language and metaphysics than ethics, but I don’t see what “normative” could mean such that a ‘standard realist’ needs to capture that and that isn’t captured just by saying that the relevant moral-ought claim is true. Enlightenment welcome.
Jussi,
I’ve not gotten to Korsgaard’s book yet, but I’m pretty familiar with her earlier work. So here’s my conjecture as to how the argument is supposed to work (and I think this is consistent with Kenny’s suggestion that she means to answer the normative skeptic):
Her complaint against moral realism is that moral facts are not “sufficiently normative” to account for the rational obligatoriness of morality. (So she assume at the outset an internal relation between moral obligation and reason.)
So begin with the hypothetical imperative argument. The HI says to take the best means to our chosen ends. But if, in a particular instance, Y-ing is the best means to our chosen end X, what makes it rationally required that we Y? By itself, the normative instrumental fact — that Y-ing is the best means to X — does not make Y-ing rationally required. In order for Y-ing to be rationally required, we must take HI itself as rationally required and apply it to the case at hand. It is thus the further mental act of applying HI, and thereby treating HI as rationally required, that makes the instrumental fact sufficiently normative for Y-ing to be rationally required. And since one might reasonably ask why it’s rationally required to abide by HI, neither the instrumental fact by itself nor HI can serve to explain the normativity of choosing the best means to our ends.
Here’s the parallel she’s drawing with moral realism. Let’s say that in some situation Z-ing is morally required by the moral fact F. Are we rationally obligated to Z? Just as with the normative instrumental fact that Y-ing is required for end X, a further principle is required to show that Z-ing is rationally required. In the case of instrumental rationality, HI is the principle. In the case of morality, the principle would have to specify how F renders Z-ing rationally required. But then, it’s the mental act of endorsing and applying this principle, not F, that renders Z-ing rationally required. And if it can then reasonably be asked why we should endorse and apply *that* principle, then neither F nor that principle account for the normativity (i.e., the rational obligatoriness) of doing what is morally required.
So either the normative facts account for rational requirements on their own or they need some further principle (as Kenny called them, bridge principles) to do that, but the rational obligatoriness of any such principle is open to reasonable doubt by, for instance, a normative skeptic. What we need, then, is to account for the rational obligatoriness of certain acts in terms of requirements internal to rational deliberation as such. That’s the aim of her constructivist/constitutive program in ethics.
A small point: If Korsgaard treats moral realism as a view not only about moral facts but (as you say) a view “about what counts as a reason for what,” then the force of the argument might be blunted, since F’s being a reason to Z seems more immune to the skeptic’s question “but what reason have I to do Z?”. But it can’t fully answer the argument since the skeptic might then question the principle that shows that Z-ing is not simply rational but rationally required.
I hope that’s on track…
Thanks everyone for helpful comments!
Kenny,
a draw with her would be nice. You are right – I would take it to be an unreasonable demand that one has to be able to convince the normative sceptic to be able to explain why something is a reason. I’m not sure even whether it makes sense to ask why some things are reasons – they just are. Explanations have to come to an end somewhere. Also, I don’t think moving from reasons-statements to ought-statements is akin to moving from is-statements to ought-statements.
I’ll have to check Sources of Normativity again.
Janice,
I’m with you here.
There are couple of relevent things in Mark’s paper which really is a great one. In the end, there is a discussion of Korsgaard. Here, a slightly different argument is being discussed. The idea is that, according to Korsgaard, unless you are a realist (who is foot-stomping in insistance about moral facts) or a Kantian, you’ll have to be a voluntarist who faces the threat of Cudworth’s arguments. However, Mark points out that you can be a reductive naturalist whose views are not threatened by Cudworth’s challenge. Yet, the argument above addresses the moral realist rather than the voluntarists or the reductivists.
There is an interesting point in the paper though in which Mark explains how Cudworth and Price can avoid their own challenge. If I get this rigth, the idea is that that something is morally required, this cannot change the moral status of the means for doing what is morally required. Taking the means is never something one ought to do qua being the action it is. Rather taking the means is an instance of doing what it was that one was initially morally required to do. This seems to be what I had in mind.
Michael,
that’s probably a good understanding of her argument. There are problems with this argument. First, you might think that rationality is a question of internal coherence between moral judgments and intentions and actions. In this case, violating moral requirements would not need to be irrational even if it was what had best reasons not to do. And, I’m not sure why you need a principle to specify how F renders Z-ing rationally required. If F is a reason, then this could be what reasons do in general – they make actions required. Finally, even if there was a principle that specified how F renders Z-ing rationally required, I don’t see why that principle could not make Z-ing rationally required and not the mental act of endorsing and applying that principle.
This is really helpful, but there are a couple of things I’m still not getting.
Jussi, by a “foot-stomping realist” do you mean a realist who is not a reductionist? Reductionists are realists, too, but maybe not the target of her argument. In any case, I would have thought that what I said would apply to a non-reductive view as well. A non-reductive realist thinks there are moral properties too. The puzzle about such a view shouldn’t be located here: ‘how could what makes a true moral claim true be normative?’ but in something like Mackie’s question ‘how could there be primitive moral facts?’ (Forget what I said about Mark’s paper; think it is as an independent argument.)
What Michael says is helpful here. Implicit in his formulation of the argument, I take it, is the assumption that to be normative, it’s not enough for a fact to be a moral one, it must also generate a requirement of rationality. This is a super-Kantian assumption that the realist doesn’t need to accept. If there are moral requirements, they are normative if anything is.
Sobel tells me that Parfit gives a very similar argument against Korsgaard’s earlier work in his paper “Normativity” in one of Shafer-Landau’s volumes. He says the relevant discussion starts a little more than halfway through the paper.
Yes – I think Korsgaard’s foot-stomping realist must be a non-reductionist as she seems to think that reductivists face something like the Cudworth’s problem (which they don’t according to Mark). It’s true that Mackie’s arguments might be a problem for the realist (which I’m not sure about) but at least they are not the same argument as Korsgaard’s. I think you are right about Michael’s comments.
It’s also true that Parfit has a discussion of this on OSME 1 (section 6 onwards in his paper). I didn’t remember that Korsgaard has a similar argument in her Normativity of Instrumental Reason paper in the Cullity and Gaut volume.
Here’s very roughly how the dialectic goes.
According to the realists, ‘rationality is a matter of conforming the will to standards of reason that exist independently of the will, as a set of truths about what there is reason to do… The difficult with this account … exists right on its surface, for the account invites the question why it is rational to conform to those reasons, and seems to leave us in need of a reason to be rational (Korsgaard)’.
Now, the realist could say that being rational just consists of wanting and doing what we have most reason to do and want.
Apparently, Korsgaard’s problem with this is that it makes realism trivial.
Parfit’s response is that it would be trivial in a bad way if this would be a stipulative redefinition. But, he claims that it isn’t – it is an obviously true analytic truth. And, it doesn’t make realism trivial because the truths about what there is reason to want are not trivial.
He also worries that Korsgaard’s objection seems to assume that under realism there might be no (motivating) reason to do what one has most (normative) reason to do. She then seems to be requiring that motivating reasons are required for normativity rather than normative reasons. And, this seems to assume that beliefs about normative reasons are not enough by themselves to motivate us which is something that the realist can also reject (as she can reject the connection between normativity and motivation).
Parfit also has an answer to the regress worry which seems to be the same as mine. He thinks that she is ignoring the realist’s view. She is assuming that any reason must be derived from a principle which we have adopted. This means that to end the regress one must find a principle of which it cannot be asked why we have chosen it. But, Parfit says, the realist thinks that we can here appeal to normative truths about the reasons – they are not principles or based on principles in the first place. There are questions about why believe such truths (as Mackie would contest) but that’s not a normative question of the reasons or requirements themselves. And, if the reasons exist, then they should be able to stop the regresses.
Thanks, that is super-helpful. This seems to me to be exactly on point:
“[Parfit] worries that Korsgaard’s objection seems to assume that under realism there might be no (motivating) reason to do what one has most (normative) reason to do. She then seems to be requiring that motivating reasons are required for normativity rather than normative reasons.”
Jussi, I worry that neither your reading nor Parfit’s is charitable enough to Korsgaard. Insofar as I understand her on this, I take her to be essentially playing the role of the tortoise in Carroll’s “What the Tortoise Said to Achilles” (1895). The point is supposed to be, I think, that the dogmatic rationalist makes a mistake by describing someone’s having a reason as a factual truth, merely adding another premise to the argument, as it were. What they need to make the argument go through is not another true premise, but a principle of (practical) reasoning. So I don’t think the point is primarily one about motivation, although it does raise a question about motivation. It’s a point about what kind of thing a norm is: on her view, a kind of commitment rather than a kind of truth.
Does this help?
Thanks Simon. hmh. I know she discusses that analogy in the following section. Unfortunately it confuses me even more. One standard reply to the Carroll’s challenge is that the reason why one doesn’t have to add the modus ponuns principle to the premises is that the premises are already good enough reasons to draw the conclusion. What the principle does is to make this explicit. In the same way, the rationalist is going to say that, whatever the good reason for your end is, that’s already a good enough reason for taking the means without there being a need to add the instrumental principle as an additional premise. What the instrumental principle does is again just to make this explicit. The point you mention is supposed to be a conclusion of her argument against rationalism – and not its premise (on a threat of being seriously question-begging). (Hope you are well!)
Hi Jussi, I didn’t know that Korsgaard mentioned Carroll’s topic directly in the new book, and unfortunately I’ve just moved home and my stuff’s in a mess so I can’t find the book to check it out at the moment. But my thought was this: the rationalist wants to insist that some set of natural facts give you a reason for, e.g. taking the means to some end. And the way to explain how natural facts can do this, according to the dogmatic ratinoalist, is to say that there is a particular kind of truth. That truth is fundamentally a “pure normative claim” (this is Scanlon’s term from his recent Locke lectures”) of the form “If P were the case then it would be a reason for someone in C to do A”. I understand Korsgaard to be suggesting that there is a difficulty with this answer: namely, that to claim that there is a truth of this form _is_ just to add another premise to the practical syllogism. What we wanted was not another premise, but a way to explain how rational agents can complete the syllogism. The only way anyone can complete the syllogism is by being committed already to the principle that the rationalist’s new premise is meant to capture. If an agent weren’t already committed to it, the new truth the rationalist is asking them to observe couldn’t get any grip on them, either normative or motivational.
The rationalist (like me) are not going to say that there is a reason to take the means to any end whatsoever. This would allow detaching and you get very strange reasons. You might think that not taking the means is irrational but irrationality is not reason-based but rather coherence-based. And, there are many good attempts to explain the incoherence is not taking the means to an end.
But say that there is some natural fact, that someone is in pain, that gives you a reason to pursue some goal, say, to help the person in pain. Now, the same fact, that she is in pain, gives a reason to take the means to this end – to call the ambulance.
The problem only gets started if we think that there is something to explain in how the natural fact can give this reason. Only then do we get to principles and additional premises. But, this is to misconstrue the rationalist position. They are not going to think that there is anything in the need of an explanation at this point. No premises to be added. And, you have a reason to call the ambulance, it seems really odd, as Parfit points out, to say that the reason does not get any normative grip on the agent. And, if the agent forms a belief about the reason, given that the rationalist is going to be a judgment-internalist, it’s hard to see how the reason-judgment could not be motivational either if the agent is rational.
My thought was similar to Simon Rippon’s, though I could never quite see how it would work for Korsgaard.
Maybe the problem is that the distinction needed isn’t between a truth and, well, whatever it is that Simon is contrasting with a truth (‘commitment to a principle’?), but rather between a belief and some other state of mind. (Or maybe I’m just projecting.)
Jussi, I think you may have missed the point I was trying to make. The point is not that the rationalist needs to advert to a set of principles, but that the rationalist does indeed attempt to offer “something to explain … how the natural fact can give [a] reason”, namely that (filling in with Scanlon again) there exists a _pure normative truth_ of the form: If P were the case then it would be a reason for someone in C to do A”. Korsgaard’s question is how a belief about such a thing is supposed to play a role in one’s practical deliberation.
Let me come at this from a different angle. Suppose you present to me the following system of laws, let’s call it Jussi’s Law. It is defined as containing an infinite number of principles of the following form:
L1) Everyone must A
L2) Everyone must obey (1)
L3) Everyone must obey (2)
…
Ln) Everyone must obey (n-1)
Having presented this, you tell me:
– You must A
We may imagine the dialog proceeding as follows:
– Why must I A?
– Because Jussi’s Law says so in (L1)
– I see that, but why must I follow (L1)?
– Well, you must follow (L1), Jussi’s Law says so in (L2)
– I see that, but why must I follow (L2)?
– Well, you must follow (L2), Jussi’s Law says so in (L3).
…
Me:
– Look, we’re involved in an infinite regress here, your principles always close one normative question only by opening another. You’re just not getting a normative grip on me with your so-called Law. Why should I care about it?
You:
– No, this isn’t a _vicious_ regress. The very same fact (that this is the content of Jussi’s Law) provides all the resources for an infinite series of answers to your questions.
I think I have good reason to be dissatisfied with your answer. And I think Korsgaard takes the dogmatic rationalist to be offering an answer to the normative question of essentially the same form, by citing what we might call “Nature’s Law”. It simply doesn’t matter what the content of Nature’s Law is, it just can’t get a normative grip on me all by itself. Why must Nature’s Law be a law for me?
Sorry Simon about misunderstanding your point. I think you may be misunderstanding mine too. You might be right that *some* rationalists try to offer an explanation for how a natural fact can be a reason. I’m not sure I see why they should. What is there to explain? And, I don’t see Scanlon’s princple as explaining reason-hood of a natural fact. Rather, it just seems to describe the fact that something is a reason.
Depends on what you take the Nature’s law to be. If it is that do not kill people, it should take a grip on you because the other person dies otherwise, if the law is that do not hit people, it should take a grip on you because the other person is hurt otherwise and so on.
Of course, these reasons may not motivate you, but I fail to see how they could fail to be normative for you. I know that Korsgaard thinks that normativity just consists of the predictive fact that your will will be necessitated to act according to the laws if it is free, rational, universalisable, and so on. But, why should that have a normative grip on my? What’s a better reason not to hit someone – that she will be hurt if you do or that you wouldn’t hit her if your will had a certain form? I take it to be the former and that’s why I take these first-order facts to be sufficiently normative rather than the form of the willing. I understand what motivational grip is but I don’t understand what normative grip is unless is the question of whether you have a reason or not.
I’m not sure that Korsgaard would say that the latter is your reason to do anything. But if she were to say that (or does in fact), that explanation would be compatible with the former, because it rests on it (that a person will be hurt if you hit them will be part of what explains why your maxim is a non-universalisable one, and thus why you wouldn’t act that way if your will had a certain form.)
On your explantion of how Nature’s Law is supposed to get a normative grip:
You seem to me to be missing out part of the explanation here, on pain of deriving an ‘ought’ from an ‘is’. You need to say that, e.g. the other person dies otherwise and you ought not to cause that to happen, or that would be a Very Bad Thing, or you would have bad character if you permitted such things, or something similar. And the dogmatic rationalist (remember that we are not talking about naturalists here) has an explanation of that latter, normative claim, namely that it is a mind-independent truth built into the normative universe. But that answer just invokes another clause of Nature’s Law.
Compare, e.g.,
-(1) You ought not to ever dance in a circle clockwise.
-Why not?
-Because otherwise (2) circular dancing will occur.
Why is the dogmatic rationalist unlikely to say that (2) explains the normative grip that (1) ought to have on me? Because he thinks there is no TRUTH built into the normative universe that explains why (2)ought not to be permitted, or is a Very Bad Thing, or whatever. What’s missing in this case must be present in the first for the principle “do not kill people” to get a normative grip. The dogmatic rationalist thinks it’s a brute normative TRUTH that’s missing. Korsgaard thinks it is something else, because a brute normative TRUTH, just another clause of Nature’s Law, can’t complete the argument.
I just really really don’t get this. Seems like such an uncharitable understanding of the realist position.
I’m not deriving an ought from is but rather from statement about reasons. It is true that I think that reason-relations are part of the normative reality but I don’t think that’s an explanation for why something is a reason. Neither do I understand why something being a reason should be explained by oughts, values, character or anything alike. I would be inclined to explain those notions by reasons but that’s not really at issue here. And, I certainly don’t want to invoke Nature’s Law. I don’t even know that means.
Also, capitalising TRUTH is not an argument (- it is funny though). It sound fine to me to say that that there will be circular dancing is not a reason not to dance in a circle clockwise. This explains why it’s not the case that you ought not to ever dance in a circle. On the other hand, that people will die is a reason not to kill. This explains why you ought not to kill. Of course you can make this view sound funny by talking about TRUTHS, normative grip, Very Bad Things, Nature’s Law, and so on. But that’s your terminology and not mine. And, as far as I see, when you stick to the normal language nothing unintelligible or inexplicable has been said.
Should also say that the realist position as it been described does not commit one to any funny heavy-weight ontology. It can be run in expressivist terms, fictionalist terms, anti-realist terms and so on.
Well, now I wonder if we’ve just been talking past each other. The dogmatic rationlist is by definition committed to “eternal normative verities” in the most straightforward sense (I’m not going to insist on a metaphysical commitment here, but there must be a commitment at minimum to a kind of full-blooded truth.) If you’re not committing to dogmatic rationalism against expressivism, fictionalism and anti-realism, then you’re not the target of this argument. (But then, if your view leaves you uncommitted between these options, it isn’t going to be much of an explanation of anything.)
hmh. You might be right. This is interesting. I was understanding the dogmatic rationalist to be someone who thinks that (i) ordinary facts are reasons and who denies that (ii) there is no explanation for how they are facts. It’s true that these two claims are associated with metaethical realists. I agree with you that the metaethical realism does not provide further explanation for how something is a reason. But, in writing the previous post I realised that (i) and (ii) are metaethically neutral – I think people like Blackburn would accept them (if you think of the mind-independence material) and people who accept Putnam/Wring sort of anti-realism. But, if that’s right, then Korsgaard’s argument should work equally well against these views. I’m starting to think that the crucial disagreement here is whether there is a question of requiring to explain how reasons are reasons and how we should understand what normative grip is. Each side seems to be pretty well entrenched to their own views about these so there seems to be a lot of question begging.
I’m afraid I don’t fully understand your clause (ii), Jussi. But I don’t think Korsgaard’s argument is supposed to work against anything less than a full-blooded realism. I’ve seen this (rather frustrating) pattern of argument quite often, where the dogmatic rationalist insists that all he’s claiming is that e.g. such-and-such really is a reason, and that there shouldn’t be anything problematic (metaphyscially, epistemologically, normatively, etc) about that. The rub comes when he is challenged to then distinguish his view from its metaethical competitors. I don’t think it’s question-begging to say that what must distinguish the dogmatic rationalist’s view is that it advert’s to Nature’s Law, or something very similar, as the ultimate ground of explanation of normative claims.
I find Korsgaard’s regress argument plausible, especially in the form given to it in The Constitution of Agency, p316: “even if we know what makes an action good, so long as that is just a piece of knowledge, that knowledge has to be applied in action by way of another sort of norm of action, something like an obligation to do those actions which we know to be good.”
In response we may simply postulate the existence of moral rules, and their intrinsic normativity. But both claims beg the relevant questions.
“The trouble with that strategy is that it leaves us with two problems, which in the end come to the same thing. First, it does not tell us why there is such a rule. Nor, if this is a different question, does it tell us why we should conform to the rule.”
One way I interpret this as leading directly to the regress is that a moral norm or maxim takes the form of N1: “if conditions C1 are true, then you ought to A.” But if the norm is itself defined in terms of some conditions C2 about the natural world, then we need a norm like N2: If C2, then accept N1. But then we need to seek out the conditions that make N2 true, and the regress is rolling on.
We might like to stop the regress by saying: but if conditions C2 are true, then we *do* know enough already to know N1, and hence to do A given C1. But simply asserting that C2 exists and gives us a “reason” to adopt N1, or constitutes the truth of N1, isn’t enough, unless we *already* have some knowledge of a norm authoritatively connecting C2 with acceptance of N1. But establishing norms was the initial thing we were trying to do, so assuming such a norm simply begs the question.
I also read these & surrounding passages as saying that the problem with “moral realism” is not so much that it is false, but is misguided and misleading. It treats moral knowledge as a simple kind of acquisition of perceivable or intuitable facts. This suggests that the first thing we do, or need to do, is to get our intuitions/moral perceptions straight.
But in her constructivist view, the first thing we need to do is make sense out of our ability to choose and value things, that is, of our practical agency. If we choose in certain ways, we encounter a contradiction in the will, which in Sources of Normativity she also describes as a clash with our practical identity as agents. Therefore we can’t make complete sense of such choices on their own terms. Now, when this happens and we recognize it, we may intepret this as “having perceived a reason not to do X anymore, to do Y instead,” etc. The error of moral realism (or more properly, “substantive realism”, for she sometimes calls her own view procedural realism) is to think that that’s all there is to it: we perceived something we call a reason, we acted on it, and nothing else happened there. It’s not much different from seeing rain, and getting the umbrella; we see the rain as wet, and also see its reason-to-bring-an-umbrella property, and we’re done.
But Korsgaard is getting at the question: what is that “reason,” where does it come from? And her answer, roughly, is we treat things as reasons by choice, and impose “reasons” upon the world; but that there are objective facts about what kinds of reasons we impose are consistent with a coherent sense of ourselves as agents. We can describe the requirement to treat certain things as reasons to act as norms, and conclude that failure to adopt certain norms constitutes a failure of your attempt to be an agent guided by principled reasons of any kind. Rejection of certain norms, like the value of respecting rational agency itself (in you or others), turns out to make your agency more or less incoherent. Realists might want to interpret this as moral reality forcing itself in on us, but Korsgaard’s picture is that it’s more like reality is just sitting there inert, and it’s our practical reasoning which is trying to make sense of our attempts to aim at certain goals.
The clash of a moral error is not between our mind and the world, but between two parts of our mind, which then prevents us from being a complete agent. We can restore our agency and make sense of what we are doing by selecting different principles/maxims. The corrected principles will indeed tell us to respond to specific worldly conditions in certain ways, treating them as “reasons” to do various things, but it’s the criterion of internal consistency that forced us to do that, not some property of “being a reason to A” that in any sense resides in the external world and was there to be perceived.
Put another way: reasons are not perceived, but assigned; all the same, some assignments of reasons lead to incoherency/fragmentation of our agency, and this is an objective fact–but it’s more a fact about the nature of agency than about the objects we are treating as reasons (substantive realism reverses this). So to restore our full agency we must change our ideas about what counts as a reason for what until we eliminate this internal clash.
Just my 2 cents. If anyone thinks I’m misreading Korsgaard here, and importing my own ideas into her views (which I fear I may sometimes do), let me know.
I think you get Korsgaard right. I don’t find the basic picture very appealing though. Say that you walk by a person who is suffering horribly. She is in excruciating pain. As it happens, you happen to have loads of effective pain killers in your pocket. Now, the moral realists claim that it is easy to see that the suffering of the person is a reason for you to give the pain killer to her, it explains why you ought to do so.
At this point, I really don’t get the questions about what it is for the suffering of this person to be a reason for you, or what makes it a reason. Furthermore, it seems to me to be morally obnoxious to think that the source of normativity for me is not in the suffering of the other person itself but rather in the fact that I will be a fragmented or conflicted agent unless I help the person or consider her suffering as a reason. It seems odd to say that if I could have not helped the person and yet be unified as an agent, then there would not have been a reason for me to help the agent. This makes me think that the source of normativity has to be out there in the patient rather than within the agent. And, whatever norms of practical deliberation you have to use in order to get yourself to help the agent (not many I would assume, definitely not the CI), it seems that the external fact just is enough to deliberate in those ways.
Jussi, I think a comparison with empirical perceptions might help here. Now this is the kind of analogy intuitionists, the naïve realists of morality, often use to support their views; but I think taken seriously enough, the analogy undermines their position, as follows.
It is very tempting to say that perceiving someone’s pain and knowing that I can help simply gives me a reason to help, and no further explanation could possibly be required; just as my looking at a red tomato gives me an excellent reason to think “that is red.” But as psychologists and now neurologists have discovered, there’s a lot of complicated processing going on behind that empirical judgment. Likewise there may be a lot of complicated processing behind the judgment “I should help.”
And of course, none of us this means that the person who sees the person in pain and helps in response, but knows nothing of constructivism (or any other moral theory for that matter) isn’t doing just the right thing, or would be helped any by knowing it; any more than I need to understand neurology to help me correctly judge “that is red.” Likewise (as I think you clearly understand), it is not the avoidance of a contradiction in the will, or a disintegration of personality, which constitutes the reason for helping; rather the pain is the reason. These other factors merely explain why the pain must be such a reason. So perhaps the oddness of what constructivists, or other moral theorists who offer explanations of moral reasoning, can be defused simply by distinguishing between X’s being a reason to Y and the explanation for why X is a reason to Y.
Why, though, do we need the constructivist explanation? I could simply appeal to human curiosity abot how things work, but there are more practical reasons as well. Understanding how empirical perceptions form help explain perceptual illusions, and can help us avoid them. There are of course both moral illusions, and people who try to manipulate us into adopting some “moral” view which would be convenient for them if we adopted it. If they get to appeal to “X just seems right, doesn’t it” (especially when it kind of does, but we have some suspicions…), then it’s harder to argue against them. But once we understand there’s a deeper level of explanation to go to, we have a better chance of understanding our real moral duty in hard cases.