Thursday, February 14, 2008

Is Kant's Moral Psychology Implausible? (I): A Reply to Leiter and Knobe

In a cursory examination of Brian Leiter’s Nietzsche blog, I stumbled upon a small exchange that involved Leiter insulting Kant scholars (perhaps justified, since it was in response to the claim that Allen Wood insulted Nietzscheans). I was going to let this go, but then Leiter also suggested that Nietzsche’s moral psychology is more plausible than Kant’s, and I couldn’t quite let this go. The result was that people started arguing with me and giving me reading suggestions, and it felt wrong to keep responding to them on Leiter’s thread, which was on a completely different, Obama-related topic. So instead I promised to blog about it here. What I mainly want to do here is defend Kant in response to the sources that were suggested to me—Leiter and Knobe’s [L&K] article, "The Case for Nietzschean Moral Psychology", and (briefly) Paul Katsafanas’s Nietzschean arguments against Kantians.

Let me start with a preliminary: In keeping with the theme raised on that thread, my question will be whether Kant’s moral psychology is any less plausible than Nietzsche’s. The plausibility is to be assessed (at least in part) in light of current empirical knowledge of human decision making. I don’t want to claim that Nietzsche’s moral psychology is less plausible than Kant’s; I am very fond of it, in fact (or at least the little I learned back in the day). What I want to argue is that there are no good grounds provided in L&K’s paper for thinking that Kant’s moral psychology is any less plausible. And the main reason for this is that I think empirical evidence largely leaves Kant’s important points untouched; the parts that seem to support Nietzsche, on the other hand, strike me as largely compatible with Kant. I will get to this point in the next post; first I want to question the assumptions made by L&K, specifically the assumption that they are somehow criticizing Kant’s views. This post, then, relates more to meta-considerations about the debate, and particularly to the role of moral philosophy, than to Kant’s moral psychology itself.

Now my response is, in part (but I hope only in part), something Leiter will no doubt consider a cop-out. And this is because my response (again, in part) does tend to follow the lines of a long-familiar argument that L & K start out by dismissing:

Certain Kantians might say: “Kant’s theory is not intended as a psychological hypothesis. It should be understood rather as a statement of the conditions of possibility of moral agency. Hence, if we find that no one actually meets the conditions set out by the theory, we should not conclude that the theory itself was mistaken. Instead, we should conclude that no one ever truly is a moral agent.” Let us call philosophers who adopt such a posture Above-the-Fray Moral Philosophers. [AFMPs] Such philosophers are indeed invulnerable to empirical results: they tell us how moral agents ought to be, and they are indifferent to how moral agents actually are or can be. We reject such an approach in this essay. We assume that ought implies can is a reasonable aspiration in moral psychology; indeed, that ought implies realistically can is an even better aspiration… We assume with most moral philosophers (including many Kantians) that there are agents who perform morally valuable acts, and thus the question for moral psychology is not merely a question about the possibility conditions for such psychology, but how this psychology actually works. (2-3)

So, let me comment on this. If L&K use such a claim to preface what they take to be an attack on Kant, then they are being intellectually dishonest. And this is (partly) because they are failing to distinguish Kantians from Kant. Their objections in the paper do, I think, pose challenges to many contemporary Kantians. They do not—as I will show—pose any challenge to Kant. One reason is that Kant is, to a large extent, what L&K call an AFMP. L&K may well disagree with such a position, but an attack on Kant that assumes this disagreement cannot reasonably be taken to be an attack on Kant. I do not, to repeat, disagree that the paper poses serious challenges to many contemporary Kantians; but I think Kant himself poses equally serious challenges. To be sure, attacking Kant by inaccurate proxy is a time-honored tradition in 20th century Anglophone moral philosophy. But someone who works on a figure in the history of philosophy (Leiter) ought to know the difference between attacking Kantians and attacking Kant. (What would Leiter say to the people who attack Nietzsche based on a reading of Heidegger and Deleuze?)

Second, L&K’s characterization of AFMPs certainly makes such an approach sound like silly, outdated, armchair stargazing nonsense. But it is a lot more appetizing if we stop to think about what Kant was trying to do instead of setting up a straw man for the fire (or whatever one does with straw men). We can approach this, I think, by noticing that L&K seem to simply conflate (as they also do later in the paper) moral theory with moral psychology. The first (roughly) being that branch of philosophy that tells us what principles of action or character traits or goals are good (I am trying to be neutral here with regard to moral theories; Kant’s own formulation of the moral question is: “What ought I to do?”), the second being the study of how human beings actually comport themselves with regard to morally relevant concepts or principles. By conflating the two, L&K essentially assert that morality must be utterly subservient to moral psychology. There is certainly reason for a Nietzschean to take such a position, but again: it kind of prejudices the argument from the outset. We can, of course, with Hume affirm that one cannot derive an ought from an is, that philosophy can only study what is, and that we can therefore throw the ought out the window. But doing that involves taking a stance within moral theory, and an argument against Kant has to occur at this stage, not after the stance has already been taken. True, L&K's “ought implies can” reference may be intended to forestall just such a criticism, but their move there is so fast and unsupported, that it cannot justify using empirical moral psychology to question Kant’s moral theory.

Third, there are good reasons for Kant to be an AFMP. Because what Kant was trying to do was, precisely, to (1) figure out how we ought to act, and (2) give the conditions of possibility for so acting. And the problem with the “ought implies can” principle is that either you base your moral philosophy entirely on moral psychology (people want x & y, therefore they should do p & q; alternatively, people have the psychological traits a & b, therefore they should do or can be expected to do p & q), or you figure out your morality independently of empirical data. Only the second approach allows you to say what we ought to do, rather than just what we should do, or what it would be best for us to do given what we are like as natural beings. And, in fact, you can only figure out what we ought to do by refusing to start out with how we are by nature. And this position becomes more tenable still, I think, if you ask yourself why we should reject psychologism in mathematics and logic, but maintain it with regard to moral philosophy. (Imagine: “Some philosophers claim that the square root of 2 has a determine value, regardless of whether human beings can calculate it in their head. I maintain that to have a determinate value is to have a value determined by actual human capacities.”)

Fourth, “ought implies can” is obviously a principle Kant takes seriously. It is, in fact, an a priori truth for him. Kant explicitly uses the principle to argue from the fact of the moral law to the human ability to follow it. Since the ought, for Kant, is derived from reason and not from nature, unsurprisingly the question of whether or not human beings can follow it turns out to be a metaphysical, not an empirical question. I agree with Kant that this is a step that has to be taken by anyone who wants to differentiate what we ought to do from what we, on some description, happen to want to do. Of course the metaphysics must be such that it is not contradicted by empirical evidence. And it will turn out that the empirical facts, as Kant sees them, are in no way contradicted by L&K. And this is partly because Kant does his best to make his empirical psychology as neutral as possible on moral questions. What he rejects, however, is the approach that limits all human knowledge to the empirical, and this is what allows him to posit a moral law in the first place. Kant’s grounds for rejecting the limitation of all human knowledge to empirical knowledge may, of course, be questioned. Almost all contemporary Anglophone philosophers reject them. But—and this is a very different discussion—his challenge to naturalism has yet to be answered directly (to my knowledge). And, in any case, the notion that philosophy is just there to clarify empirical claims and not to make ought-statements largely went out the door with logical positivism. Or did it?

Fifth, we should, especially if we care about the ought question, take a second look at the conflation issue. The AFMP is characterized as saying that, “if we find that no one actually meets the conditions set out by the theory, we should not conclude that the theory itself was mistaken. Instead, we should conclude that no one ever truly is a moral agent.” And L&K reply that such philosophers “tell us how moral agents ought to be, and they are indifferent to how moral agents actually are or can be.” But L&K seem to be confusing three things here: (1) whether there actually are any agents that live up to the demands of moral theory, (2) whether actual human beings have the psychological capacities that might allow them to live up to those demands, (3) whether, given actual human capacities, human beings are capable of living up to the demands of morality. These are obviously distinct questions. If human beings really lacked the capacity to be moral, Kant tells us, there would be no morality. But this goes back to the previous point. In any case, the conflation is important because the point of the ought, once again, is to set an ideal above and beyond the goals we have by nature, so the moral psychology demanded by the moral theory will be at least in part non-empirical. At least one—though major—reason for this, is that the goal set by morality must be capable of overriding all other demands, and this means that it must transcend them. On this picture, the question is whether or not we are something more than merely natural beings; but it should be obvious that this is not an empirical question. Moreover, it would not be surprising, given such a picture of morality, if no one lived up to it; in fact, if anyone did live up to it, we could have no empirical evidence for this! Morality is, for Kant, very different from making a tasty pot of chowder or even getting a perfect score in a bowling match. Morality is something to strive for infinitely. But even if no one meets the conditions set out by Kant’s moral theory, this does not imply that no one meets the conditions set out by his moral psychology.

I know I seem to be making a mountain out of a molehill, and I certainly seem to be missing L&K’s point. The truth is, I am quite sympathetic to the project of bringing actual psychology to bear on issues of agency and ethics, and I find their paper very interesting. But if philosophy is to be something other than just psychology, or a tool for clarifying psychology (see my previous post, on Appiah and x-phi), then it is reasonable that it should have components that do no take their directives from the empirical evidence, but instead provide us with the directives for evaluating that evidence in the first place. Taking is questions to make important and valuable contributions is quite proper and healthy for philosophy. But taking is questions to provide counter-evidence to ought questions (and, really, taking ought questions to simply be varieties of is questions) often simply involves philosophical confusion. (See Michael's post criticizing the idea that moral psychology can replace moral theory.)

In the next post, I will address the actual issues of moral psychology in question, and argue that they do not cast doubt on the plausibility of Kant's moral psychology.


  1. I should say that in the Leiter thread I invoking the work of Katsafanas -- the publication of most of which is pending -- as a suggestion of where one might find the materials for a Nietzschean challenge to your apparent view that substantive psychological claims play only a modest or negligible role in the working out of Kant's ethical system.

    Having had the recent pleasure of reading his draft "From Moral Psychology to Ethics" (kindly made available upon request), I'm quite excited about how his pending publications will bear on this issue (among others) -- particularly his elucidation of Nietzsche's "drive psychology" in conjunction with constitutivism.

    For more on Katsfanas:

  2. You must be PK's PR agent! Really, I found his page about half a year back, while checking to see what Korsgaard's students are doing, and it looks extremely interesting. From what I can tell from his summaries, I am largely in agreement with many of his themes. I just want to reiterate again (a point I will clarify significantly in the next post) that the moral psychology of most Kantians is not that of Kant. Kant did not insist that deliberation is somehow crucial to moral action, for example. Nor does anything in Kant oppose a drive psychology, as far as I know.

    As for constitutivism, it is an interesting recent approach to normativity, and one with which I am sympathetic. But it is clearly not Kant's approach, and arises (in Korsgaard's work, at least) from an approach that Kant explicitly repudiates, namely: the Groundwork's deduction of the moral law from freedom.

  3. Oh, I'm simply an avid and insatiable consumer of quality Nietzsche scholarship; and since I'm not a producer of it I try in my modest and occasional way to raise awareness of the good stuff I've encountered. Though the situation has improved dramatically over the past decade or so, I think it remains the case that far too little of the scholarly output on Nietzsche is worth the time it takes to read.

    I'm far too pitifully unqualified to comment in any interesting way on your post except to say that (probably owing to my ingrained obtusity) I remain puzzled by the claim that "you can only figure out what we *ought* to do by refusing to start out with how we are by nature."

    This recalls for me a remark Nietzsche makes about Thucydides (in TI, "What I owe to the ancients ") when he praises him for "finding reason in reality," as if Thucydides represents a precursor to the project of drawing ought from is, grounding normativity in how human beings really are. I think Nietzsche's point (which finds echo in Katsafanas' "From Moral Psychology to Ethics") might be that how we actually are is normatively illuminating in a way that is obfuscated by the Kantian formal insistence upon the metaphysical nature of *ought*.

  4. You are right that "you can only figure out what we *ought* to do by refusing to start out with how we are by nature" is perhaps a puzzling claim. Let me try to make it a little clearer by starting with a premise you mention: that we can find reason in nature.

    What does it mean to find reason in nature? Well, we could say that nature makes sense; and making sense involves exhibiting rational connections. Hume takes roughly the following path here: we know nature through our perceptions, which exhibit certain relations to each other. But he is completely silent on how these relations might be connected to each other, or what their source is. If you take this path, then, you might acknowledge that nature has a rational structure, but this structure is utterly mysterious. We can rely on it--in fact, we must rely on it, if we are to be able to act in the world--but philosophy has no purpose over and beyond clarifying these connections.

    Now the Kantian response is that this is unsatisfactory. Philosophy cannot limit itself at what we simply find in reality, because as human beings we need not only to know the principles of nature, but also to be able to justify our reliance on them. To shorten the account, I'll put it like this: in understanding the order of nature, we rely on principles, which we connect through various inductive and deductive forms of reasoning. But insofar as we must use reason to understand the structure of nature, we cannot stop with nature. The same principles of reason that we use there, require further justifications for our premises, and further justifications for those, and so forth. In Kant's terms: reason aims at the unconditioned, i.e., the point at which we have an absolute justification. And since everything in nature is conditioned by something else, this means that reason as such goes beyond nature. So: reason can be found in nature, but it is not constrained by nature.

    There is a psychological claim involved: human beings are rational. But this does not mean that human beings always seek to find an ultimate justification--that would be evidently false. What it means is that reason as such requires that it's processes continue tracing premises all the way back to the unconditioned. We human beings might stop short at any point, because we get distracted, or bored, or find further inquiry too difficult, or get excited about a premise at a particular step and are not incline to go further. But this shows not that reason does not aim at the unconditioned, but that we reasoning natural beings have psychological processes that generally lead us to stop short.

    Something similar happens in the domain of practical reason. Reasons aims at the unconditioned, once again: a principle that requires no further justification. And this principle is given by the moral law which, as unconditioned, gives us an unconditional "ought." We may never follow this "ought"; we may, in fact, reliably act for all sorts of other, non-moral reasons, which moral psychology can bring out. But, again, this does not show that the aims set by reason are invalid; it shows only that we, imperfectly rational beings that we are, have other aims, other goals, and other springs of action.

    And if reason goes beyond nature, this means that if we are looking to understand something unconditioned--like moral obligation--then we must look to reason, and not to nature.

    Lastly, I fully agree that how we are--as revealed in part by psychology and other sciences--can be normatively illuminating. It will, ultimately, determine how we tackle moral problems in concrete application. But it cannot alter the standard according to which norms themselves are measured, i.e., the principle by which our norms are not simply explained, but justified.

    Does any of this make sense?

  5. This very clear, and helpfully so. It represents what seems to be the invulnerable default position: the normative can never be outstripped by the descriptive because the former is infinitely capable of absorbing challenges from the latter as merely further material upon which to operate.

    I'm probably guilty of an excessive enchantment with how (in the second Treatise of the Genealogy) Nietzsche presents a naturalistic account of the origins of normativity in a way that seems to clarify rather than undermine it. It's as if the spade has turned on how matters, at an appropriate level of description, are -- resulting in constraints upon theorizing that are, well... normative.

  6. Wow, you make the Kantian position sound so stodgy and conservative!

    I admit that my Nietzsche is rusty, but I've never felt like the gap between him and Kant was all that wide. I'll have to go back to the Genealogy to say anything meaningful here (which I actually will be doing in a few days--going back to the Genealogy, I mean, not saying something meaningful). But it's always seemed to me that the invention/discovery distinction is rather slippery when talking about human nature; more than that, the distinction can tend to vanish under some pressures. Here's what I mean:

    Human beings are responsible animals and see themselves as responsible animals. (And, as I understand it, Nietzsche does not deny that we are responsible animals, only that we are such by nature. How did this come to be? How did we become responsible animals?

    One way of telling the story is by giving a genealogical account of the invention of responsibility: telling the story of its genesis out of primitive social interactions and giving an account of how the outcome is shaped by self-interest, ressentiment, the desire of the weak to be strong, and so forth. Another way of telling this story is to say that, if human beings could come to be responsible in the first place (which is undeniable, since here we are), then this possibility belonged to the species in the first place, and the human being has always been a responsible animal, only this potential had to be uncovered for us.

    These are different stories, of course. But it is only by rejecting the second one--that is, rejecting human nature in its current historical form--as having value that we can think that the first story is the last word. The entire Nietzschean genealogy could be told in a more positive light, and at the moments when Nietzsche grinds his teeth and grudgingly acknowledges the cleverness of the priests, admitting, really, that responsibility and memory are a form of genuine progress, well, at those moments I think you see that the negative story he tells about morality is itself structured by his underlying normative convictions.

    As for the normative being "infinitely capable of absorbing challenges from" the descriptive, I think there is a level at which it's trivially true: descriptions are normative, inevitably so, because normativity is necessary to make sense of anything whatsoever. So--putting the question of specifically moral normativity aside--the normative as such necessarily outstrips the descriptive, because it is its condition of possibility.