I've already sent Paul comments on some minor matters in the current version, so I won't dwell on those issues here (though may flag one or two in passing in case anyone wants to pursue them in the comments).
The key question is what exactly are "drives" for Nietzsche. Katsafanas (hereafter "K.") says (I think rightly) that Nietzsche uses Trieb and Instinkt interchangeably (K. does not comment, however, on where Affekt fits in, though it turns out to play an important role in his exposition). In Section 1.1, K. argues against interpreting drives as homunculi, i.e., as little agents within the human agent. This strikes me as correct, and so I will not dwell on it here.
In section 1.2, he considers dispositional views of drives. Bear in mind that K's own preferred reading is a kind of dispositional view: "Nietzschean drives are dispositions that induce affective orientations in the agent...[T]hese affective orientations can be understood as evaluative orientations" (p. 10, end of 1.2). This is closely related to the view Richardson defended originally in Nietzsche's System, though K's criticism of the view at p. 9 strike me as not wholly satisfactory. It is true that in later work Richardson tries to treat the dispositions of drives in selectionist terms (there are pertinent criticisms of that general interpretive strategy here and here). As I've already pointed out to Paul, there is no reason to think that natural selection will necessarily produce ascetics "disposed...to engage in sexual activity," but all he needs is the possibility of an ascetic who is both disposed towards having sex and disposed towards being an ascetic (natural selection is irrelevant). Thus K's objection: "There are cases in which values and selected dispositions appear to diverge [such as the ascetic in question]....Despite the fact that the agent is strongly disposed toward sexual activity, we would typically say that the agent does not value sexual activity" (p. 9). But all this requires us to say is that the drives (i.e., the dispositions in question) differ in strength, so that the ascetic disposition (and associated valuation) simply domiante the sexual disposition. Indeed, I take it such a solution is compatible with K's preferred view.
Section 1.3 takes up the question of how to interpret N's remarks expressing skepticism about what we know about our drives and their role in the genesis of action (K. scratches the surface of the kinds of remarks N. makes in this regard at p. 11, but the quotes he chooses are certainly representative). K. notes my view that N. thinks that "we do not have epistemic access to what the causally effective motives really are" (NOM, 104), but then glosses that as something else, namely, "the claim that we are often [emphasis added] mistaken about our motives" and says this "is just a platitude" (p. 13). But even the weaker claim is not a platitude, and the fact that Kant was ready to bite the bullett on the possibility that no one ever acts morally (because it might be that no one ever acts for the right kinds of reasons) hardly show this to be a platitude. The Cartesian picture of the self, of its essential transparency, reverberates throughout modern philosophy, all the way to the present, though since Freud's triumph, probably more thinkers would be prepared to sign on to the skeptical view.
Ultimately, K. wants to argue (in sections 2 & 3) for a particular reading of N's skepticism about what we know about our drives, according to which when a drive brings about an action A, we may know we are A'ing, but not know the aim (of the drive) that led us to A (p. 18). (I let pass in silence the strange discussion of what wolves and flies know about their actions at p. 17, since I don't think anything ultimately turns on these claims being correct--what matters is that the account works for humans.)
I accept this as a more precise way of stating the point I was making in NOM, i.e., that what we do not know when we do not know the "motive" for our action is the ultimate aim of the drive that is the causal genesis of the action. This may well be the right way to gloss N's skepticism about the sources of agency.
Section 3 ("The nature of Nietzschean drives") is the core of the paper, and quite illuminating, until p. 32 (in section 3.3), when the argument, it seems to me, goes off the rails.
3.1 wants to explain how "the affective orientation induced by a drive can be understood as an evaluative orientation (p. 21). (This is the point at which I wonder about the connection between drives and affects.) The discussion is highly illuminating, emphasizing the way in which drives/affects influence "perceptual salience" (p. 22), not only the features of a situation that come to the fore for the agent (because the drive focuses attention on them), but by influencing "the content of experience itself" (p. 22) by affecting the interpretation of the sensory stimuli themselves. (K's discussion resonates with recent work by a former student, Neil Sinhababu, on the Humean theory of motivation, which emphasizes the role of desire in affecting salience: vide his recent Phil Review paper. Neil, as readers of the blog will know, is also interested in the similarities between the Nietzschean and Humean views.) Here is K's summary statement at the end of 3.1: "Drives manifest themselves by coloring our view of the world, by generating perceptual saliences, by influencing our emotions and other attiitudes, by fostering desires" (p. 26).
3.3 appeals to Freud's account of drives, arguing, plausibly to my mind, that it illuminates N's view (Freud may well have gotten the view from N., of course). There are two important features of drives on the Freudian view. First, drives have a kind of constancy that other desires do not--"they arise, with some regularity throughout the individual's life" (p. 30). The music you desired to listen to in your 20s may no longer appeal in your 40s; but the hunger drive keeps coming back whether you are 20 or 40. Second, drives do not depend on an external stimulus to be aroused. "Drives arise independently of external stimuli and once they have become active, they will seek discharge" (p. 31). External stimuli can give rise to a desire to eat or to have sex, to be sure, but those same deires can simply arise in the absence of any stimuli. (Sometimes K. overstates the point, e.g., at p.31, where he says that "drives...do not arise in response to external stimuli." But that can't be right: surely they can be triggered by external stimuli. What allegedly distinguishes them is that they can arise on their own, without any externl provocation.) It is useful to distinguish, as Freud does, between the Ziel (aim) of the drive (e.g., sex, eating) and the Objekt of the drive (e.g., this woman, this bit of food). Insofar as a drive is aroused not by an external stimulus, it will then seek out an object for its realization--and in so doing (I assume this is K's point) impose a "valuation" on the object.
Now we come to the point where the argument and exposition seem to me to go off the rails. At p. 32, K. claims that "drives operate by influencing the agent's perception and reflective thought, so that the agent sees a certain activity as warranted" (emphases added). K. made a good case for how drives might influence perception, but where did talk about "reflective thought" come from and what does it mean? Nothing in the argument so far, or in N's texts, invites interjecting this mechanism. The crux of what K. is getting at, I think, is his claim that "drives affect the agent's perceptions of reasons" (p. 32). Here is K's example:
But the "reasons" here are simply causal product of the drive, and the fact that the aggression seems "warranted" is itself just an artifact of the causal umph of the drive. Where is the "reflective thought" coming in? What is reflection doing?The aggressive drive does not just produce a blind urge that causes the agent to act aggressively. Rather, the aggressive drive manifests itself by producing desires, affects, and perceptual saliences that jointly inclidne the agent to see aggression as warranted by the circumstances. (p. 32)
Perhaps K. means to concede this latter point? He writes: "being moved by a drive and being moved by reflective thought are not distinct processes. Drives move us by directing and influencing our reflective thought" (p. 33). But is "directing and influencing" the same as "determining"? This is the crucial ambiguity. The only claim justified by the argument so far would be that what we take to be reflective thought is, in fact, determined by our drives. But talk of "directing" and "influencing" might suggest that there is still some causal work for "reflective thought" to do.
In 3.4, K. acknowledges the obvious (or what, by now, I've forced everyone to admit is really obvious!), namely, that "N. is notoriously skeptical of reflective agency" (p. 33). But what does this skepticism amount to? In 4.1, K. claims that "it is by now a commonplace that N. rejects the libertarian concept of choice" or free will, another 'commonplace' I am happy to take credit for making common. (Joking aside, I do think it is a mark of the maturity of Nietzsche studies over the last decade that one can now make arguments--demonstrating, e.g., N's skepticism about reflective agency or his rejection of libertarian conceptions of free will--and other scholars actually register the arguments and their conclusions, and then development their own interpretive views based on these 'results'. Anyone reading the Nietzsche literature twenty years ago knows how utterly rare this used to be.)
As K. admits, rejecting the libertarian conception of free will leaves a lot open. I have also argued (in both NOM and in "Nietzsche's Theory of the Will") that Nietzsche can not be a compatibilist about "choice" or free will either, though K. does not consider those arguments (to be fair, his immediate concern is not responsibility but the causal contribution of reflective thought, so perhaps the omission doesn't matter). K. concentrates instead on my claim that "Nietzsche argues that choice is epiphenomenal: 'there is no causal link between the experience of willing and the resulting action' (Leiter, 2007, 13)" (p. 34).
Two quick points about this gloss on my argument in the "Nietzsche's Theory of the Will" paper. First, the argument there concerned the causal status of the experience of willing, not everything that might be denominated a choice. When I went to the refrigerator, I chose a diet pepsi, rather than a diet coke: was my "choice" epiphenomenal? It seems a tad odd to think it is, and one might hope nothing in N. committed him to thinking so. But it seems equally clear that what I will call "mundane choices" do not often involve the experience of willing, that is, of choosing to make something happen: these are mostly automatic, unrefelctive behaviors, that are casually characterized as choices post-hoc.
Now this may not matter--perhaps we should just stipulate that "choice" for K's purpose just means any action that is preceded by the "experience of willing" that is the focus of N's analysis and my reconstruction in the 2007 paper. But a second complication may warrant notice. In the 2007 paper, I noted that there are two kinds of passages in N.: one kind invites the epiphenomenalist interpretation that K. calls attention to and that seems to be the favored interpretation of the experience of free will in some of the psychological literature (notably Wegner); but the other kind of passage (think the Cornaro case) is different. It suggests that the experience of willing is part of the causal chain that results in the action, but it is not an "ultimate" or "primary" cause: that is, an explanation of the action that stopped with the experience of willing would be like an explanation of the invasion of Iraq that stopped with Bush's decision to send in the troops--that decision was causally important, but it does not really explain the Iraq War. In the 2007 paper, and in my recent work generally on Nietzsche, I have taken the view that if the text support two different readings of his position, we should opt for the one that seems the most promising as a matter of empirical psychology. Right now, this seems to favor the epiphenomealist reading, but perhaps the empirical tide will turn in favor of the second reading--what I called "the Will as Secondary Cause" reading.
Now back to K. He doesn't, by his own admission, argue against the epiphenomenalist reading, he just (1) notes that the reading has critics, citing in n. 31, Clark & Dudrick and papers by Gemes and Janaway that argue against my interpretation of N. on free will (Janaway's adds nothing, except a head count, so we can ignore it here); and (2) claims that this interpretation "has textual costs" (p. 35). I understand how limitations of space go, but given what seems to be the centrality of the issue to K's argument, this hand-waving seems quite unsatisfactory. I've dealt with the problematic arguments by Clark & Dudrick and Gemes elsewhere, though even Gemes concedes a lot to my position.
As to (2), K. cites (p. 35) some passages I've deflated elsewhere (like the GM passage on the 'sovereign individual') or passages from Twilight about those who resist immediate stimuli and those who do not. But a well-trained dog has the characteristics at issue here, and the question then is whether that is enough for K's real thesis, namely, that "N. claims that some individuals have the capacity to control their behavior." A well-trained dog can "control" its behavior, e.g., it can resist mounting the nearest female, or chasing after another dog, or running from its master. Is this the "capacity to control" behavior that's at issue for K.? If so, then we have no quarrel. But the faint odor of Kant and Korsgaard is in the air, so I'm not sure.
Let's be clear about what is at stake. On my reading of N., drives causally determine action, perhaps via the mechanism of conscious mental states (per the Will as Secondary Cause reading) and perhaps not (per the Epiphenomealist reading). K., as I read him, thinks that N. thinks that there is something left over for the "will" or the "self" to do, but his only evidence, noted in the preceding paragraph, is inapposite.
Section 4.2 introduces K.'s preferred resolution to these issues. He opts, I think, for my Will as Secondary Cause reading, though he does nto put it quite that way. On K's picture, choice causes action, but "N's drive psychology problematizes the connection [the causal connection?] between the agent and choice" (p. 36). When he sums up his view at the end, he puts it in terms that sound like the Will as Secondary Cause reading from my 2007 paper: "choice may control action, but agents do not control choice." Or put a bit differently: the will may cause an action, but the will is not within the causal control of the person (the self), it is, itself, causally determined by something else (e.g., type-facts, something unconscious, etc.).
I would be happy if that is what K. means, but I am not sure. Also on p. 38 he summarizes his point as follow:
The reflective agent is, in one sense, different from the unreflective
agent: after all, the reflective agent deliberates, thinks about reasons
for acting, and examines her motives, whereas the unreflective agent does none
of this. But in another sense, the reflective agent is not so different
from the unreflective agent: while the reflective agent supposes that she
is escaping the influence of her drives, she is mistaken. The
influence of the drive has simply become more covert. (emphasis added)
Everything turns on what that last line means, and on the ambiguity that attaches to "influence." If "influence" means "causally determines," then K.'s account is a version of the Will as Secondary Cause reading. But the footnote attached to this paragraph suggests this is not what he means. In n. 35, K. says that "N. maintains that reflection and self-understanding enable an agent to counteract the effects of particular drives." This, of course, seems in tension with D 109 (which K. discusses earlier), unless "reflection and self-understanding" are just themselves the causal product of countervailing drives. But that doesn't seem to be what K. means, for he continues (in the footnote): "this seems to be one reason why N. emphasizes the importance of self-understanding: self-understanding enables one to counteract the effects of certain drives, and thereby renders the agent increasingly in control of her action." No text is cited, and the reference back to n. 28 seems to be a mistake, since n. 28 does not appear pertinent. It is unfortunate that these absolutely crucial (and, to my mind, quite implausible) claims are buried in a footnote.
That K. really wants to save autonomous agency is suggested by some additional remarks in the conclusion. He says: "the deliberating agent's thoughts and actions are guided, sometimes decisively, by her drives" (p. 39). I take it "decisive" guiding is equivalent to my "causally determined," in which case it is clear that K. wants to carve out a space where non-drive-determined thoughts and actions make a causal contribution. (But "a thought comes when 'it' wants, not when 'I' want"!) So too: "there is some sense in which the agent acting under the influence of drives may be a passive conduit for the drives; however, N. also suggests that there is some way of acting that avoids this problem" (p. 40). As far as I can tell--I hope Paul or other readers will correct me--not a single text from N. has been adduced to support this claim. In the preceding paragraph, K. refers to a few snippets and phrases that are, I guess, supposed to suggest that N. believes in "genuine agency" (p. 40, K.'s term, not N.'s), but considered in context, as I have argued at length elsewhere (see esp. pp. 15 ff.), they simply do not support anything like K's claims.
So to sum up: K's account of drives, and the ways in which they structure perception and evaluation, is illuminating and plausible, as is his incorporation of the Freudian aim/object distinction to make sense of N's skepticism about what we know about why we act. But the entire picture is then compatible with what I had called the Will as Secondary Cause reading in the 2007 paper (though not the epiphenomenalist reading). But K. wants more than that, it seems, he wants some autonomous causal role for conscious reflection and deliberation in agency. Might this be satisfied simply by instrumental reasoning, the supplying of information about means to drive-given ends? I sense K. wants more, but it is both unclear whether he does and implausible that N's texts would support such a reading in any case (or, in any case, K. would really need to make the textual case).
UPDATE: The discussion of K's paper continues here.