Thursday, May 1, 2008

Katsafanas on Nietzsche on Consciousness

References are to Paul Katsafanas, “Nietzsche’s Theory of Mind: Consciousness and Conceptualization,” European Journal of Philosophy 13 (2005): 1-31. Some other references are to David Rosenthal, “Consciousness and Its Function,” forthcoming in Neuropsychologia (I cite to the MS version).

Katsafanas, as I’ve acknowledged in my “Nietzsche’s Theory of the Will,” Philosophers’ Imprint 7 (2007), is plainly correct to criticize me (and Deleuze) for claiming that Nietzsche views consciousness simpliciter as epiphenomenal. That is not consistent, as Katsafanas shows, with a variety of claims Nietzsche makes, and in retrospect this strikes me as the most serious mistake in my 2002 book Nietzsche on Morality. (There are other interpretive points I would put differently now, but the treatment of epiphenomenalism is the one issue that I now think in error.) Yet it is clear that Nietzsche thinks some aspects of consciousness (e.g., our conscious experience of will) are epiphenomenal: the challenge is to specify the parameters of the epiphenomenalism, and give some principled theoretical account for those parameters. That is not my concern here, though Katsafanas’s paper has interesting suggestions on that score that deserve attention.

Instead, I want to consider critically Katsafanas’s own proposal regarding how Nietzsche demarcates the “conscious” and the “unconscious.” It is an intriguing and subtle discussion, but having taught it recently in my seminar, it strikes me as problematic, both textually and philosophically.

Early on, Katsafanas dismisses the view that the hallmark of consciousness is that it involves “awareness” (2-3). This can’t be right, he says, since there are “unconscious perceptions” (3), and since “a perception is a type of awareness of the world” (3), it follow that unconscious states can involve awareness. That seems right as far as it goes, but it elides a more pertinent proposal (that travels under the general heading of the Higher-Order-Thoughts [“HOTs”] account of consciousness) according to which the hallmark of consciousness is not awareness simpliciter, but rather awareness of being in a particular psychological state, i.e., the one that ergo is conscious. Here is Rosenthal, a leading proponent of the view: “A psychological state is conscious…if one has a thought, distinct from the state itself, to the effect that one is in that state” (15). The HOT need not itself be conscious, indeed, most often it probably is not—unless there is another even higher-order HOT about the original HOT. The details of the view may not matter for our purposes. The point here is that Katsafanas has dealt far too quickly with the intuitive idea that consciousness has something to do with “awareness”: it’s not, contra Katsafanas, awareness of the world that’s at issue, but rather awareness of the state of perceiving or thinking or desiring that we count as conscious. We’ll return to this, below.

Katsafanas dismisses the “awareness” account without citing Nietzsche. But when he turns to claims about Nietzsche’s own view he, quite correctly, assumes a textual burden, as well as a philosophical one. The key passage on which he relies is GS 354, and the key bit (cited at his p. 3) is this (I follow the translation Katsafanas uses, which seems fine for our purposes):

Man, like every living being, thinks continually without knowing it; the thinking that rises to consciousness is only the smallest part of all this—the most superficial and worst part—for only this conscious thinking occurs in words, which is to say signs of communication, and this fact uncovers the origin of consciousness. In brief, the development of language and the development of consciousness…go hand in hand.

This passage comes fairly late in GS 354, and thus obscures the fact that the primary thesis of the section is that, as Nietzsche puts it, “consciousness in general has developed under the pressure of the need to communicate,” which arose for basic evolutionary reasons, i.e., at some point the human animal “needed help and protection.” We will return to this point shortly. Let’s focus, instead, on the passage that Katsafanas makes central to his reading.

What Katsafanas wants to take from GS 354 is the following argument: (1) there is no conscious thinking without language; (2) to think in language is to think conceptually; (3) therefore, conscious thinking is necessarily conceptual or “conceptually articulated.” Here is Katsafanas:

for Nietzsche words and concepts go hand in hand; to think in words is to think by means of concepts. Accordingly, in writing [in GS 354] that conscious thinking occurs in words, Nietzsche is claiming that conscious thinking is conceptually articulated…[I]t follows that unconscious mental states do not have conceptually articulated content; unconscious state must have a type of nonconceptual content (3).

It is one thing to say that “words and concepts go hand in hand,” it is another thing to claim that words are essential to conceptual articulation, which is what Katsafanas needs to support his strong concluding claims, namely, that “It follows that unconscious mental states do not have conceptually articulated content.” Does he have any evidence? Certainly not GS 354. BGE 268, which he cites, tells us that words express concepts, but that is not enough, since it may be that conceptual content can be expressed in other ways (e.g., through images). More on point is a Nachlass passage, WP 506, in which Nietzsche says, in passing that “concepts, possible only when there are words,” though the rest of the passage is silent on that idea and its import. Strikingly, the prior section (WP 505) says, “Consciousness is present only to the extent that consciousness is useful,” which is more in line with GS 354. At the same time, WP 505 makes no claim about consciousness requiring conceptualization or words.

In short, the textual basis for the view Katsafanas ascribes to Nietzsche is exceedingly thin, consisting of a sentence fragment from the Nachlass, which has no analogue I am aware of (Katsafanas cites none) in the published corpus. The absence of real textual support is significant, however, primarly because the view about consciousness Katsafanas wants to ascribe to Nietzsche is extremely implausible on the merits. Since this implausible view is not required by the texts, one should probably not ascribe it to Nietzsche.

Let us now review the details of the view of consciousness that Katsafanas attributes to Nietzsche.

According to Katsafanas, a state has conceptual content when “first, the content is composed of simpler parts, namely, concepts; second, these concepts are structured or composed in a certain way in order to constitute the content” (4). The belief “the cat is white” is “conceptually articulated; the content appears to be composed of two concepts, CAT and WHITE, which are structured in a certain way, namely in a subject—predicate fashion, in order to form the belief” (4). But one could perceive a white cat without having the belief that the cat is white. But to have the belief, you must have the concepts of CAT and WHITE, as well as a grasp of the syntactic form.

Conceptual content is not, however, the only way for a mental state to have determinate content on Katsafanas’s view. Once again, he uses perceptions as the example. “Perceptual content would be conceptualized if the perceived object were represented as an instance of some concept, that is, as a token of some type.” Nonconceptual, but determinate perceptions “represent their objects in a definite way, but do not represent them as instantiating concepts” (7). (To my knowledge, those who think perceptions can involve nonconceptual content do not think this has anything to do with whether or not they are conscious.)

Katsafanas also emphasizes, plausibly, that while both perceptions and concepts require discriminatory abilities, having a concept requires more than this (8). “For Nietzsche, concepts are classificatory abilities; possessing a concept involves the ability to classify various objects as falling under the concept” (8). Thus “concepts are systematically related to other concepts, and concepts can be employed in non-perceptual contexts” (9).

Two points here.

First, the textual evidence proferred for saying this is Nietzsche’s view is misleading in the extreme: footnote #16 (accompanying the last quote) cites BGE 20 as saying that “concepts…grow up in connection and relationship with each other” and involve an “innate systematic structure and relationship” to other concepts. But Nietzsche is not discussing concepts per se—a subject on which, unsurprisingly, he has no views at all as far as I can tell—but rather “philosophical concepts” (philosophischen Begriffe), Katsafanas simply having dropped “philosophical.” In context, it is quite clear that he is making claims about distinctively philosophical concepts—the Cartesian “I think” and the Schopenhauerian “I will”—and not any point at all about Begriffe, at least as philosophers today would understand that idea. So Katsafanas has described a reasonable view of concepts, but there is no reason to say it is Nietzsche’s view.

Second, the fact that “concepts are systematically related to other concepts” has nothing to do with whether they are conscious. Rosenthal has effectively made this point against the view that consciousness is essential for both rationality and intentional action. For the rationality or intentionality of mental states is a matter of their intentional contents, not whether they are conscious; so, too, we might suppose with “systematic relations.” Here is Rosenthal on rationality:

Thoughts and desires are rational in virtue of their having causal connections that reflect rational connections among intentional contents….[S]ince the intentional content of thoughts and desires occur independently of whether those states are conscious, rational connections among them will tend to occur independently of whether they are conscious (9).

And here is Rosenthal on intentional actions:

Actions are intentional when they are initiated by volitions to do those things. And volitions tend to cause the particular actions they do in virtue of the intentional content of those volitions. As with cognitive states, volitions can occur without being conscious; so the property of a volition’s being conscious is independent of its intentional content….So even though we are aware of our own actions as being intentional only when the relevant volitions are conscious, the consciousness of the volitions is not necessary for an action to be intentional (10).
This gets us to the very heart of the difficulty with Katsafanas’s account of consciousness. It may be true, as Katsafanas writes, that “conscious perceptions involve a classifying awareness, whereas unconscious perceptions involve only a discriminatory ability” (9); to be clear, I am not sure it is correct either philosophically or empirically. But even if it is, that does not get us very far in the case of Nietzsche. For Nietzsche, like Freud after him, surely thinks there are unconscious beliefs and desires which figure in the best explanation of observable behavior. But unconscious beliefs and desires must surely have determinate intentional contents! Yet how can they have such contents without their being conceptually articulate contents? How can, per Freud, my unconscious wish to sleep with my mother be anything other than a mental state that deploys the concepts of SLEEP and MOTHER together with some syntactic connectors? The issue of conceptually articulate content seems entirely orthogonal to whether or not the mental state is conscious or unconscious.

Katsafanas, I fear, has obscured the implausibility of the account of consciousness he has ascribed to Nietzsche (an account that no one, as far as I can tell, defends in trying to explain consciousness) by concentrating on the perceptual case—though even here, I don’t take it that those who think perceptions can involve nonconceptual contents think this has anything to do with whether they are conscious or not. But at least in the case of perception, the idea of determinate but non-conceptual content makes some sense. But does it make any sense at all in the case of intentional content that is unconscious? Either Katsafanas has to explain unconscious intentional content without reference to conceptual articulateness or he must claim (how could he claim this?) that there is no such thing as unconscious intentional content.

I am not, needless to say, a philosopher of mind (even though, once upon a time, I published two peer-refereed papers on mental causation!), so it’s possible I’ve made some obvious error in construing the issues about content. I hope Paul Katsafanas or some philosopher of mind with a side interest in Nietzsche (the only kind likely to be reading this!) will set me straight accordingly! I also want to be clear that the only reason I have bothered to write about Katsafanas’s paper is because it is of significantly higher quality than is the norm in Nietzsche studies. I do think one of his central theses is mistaken, but it is interestingly mistaken, and it reflects a degree of scholarly and philosophical seriousness that is far too rare in Nietzsche studies.


Anonymous said...

Thanks for your comments and criticisms. I'd like to respond to a few

Your chief criticism is this: unconscious beliefs must have determinate
intentional contents. You write,

"For Nietzsche, like Freud after him, surely thinks there are unconscious
beliefs and desires which figure in the best explanation of observable
behavior. But unconscious beliefs and desires must surely have determinate
intentional contents! Yet how can they have such contents without their
being conceptually articulate contents?"

I agree that unconscious mental states must have determinate intentional
contents. But I'm at pains to argue, in the article, that these two
claims are distinct:

(a) a mental state has determinate intentional content
(b) a mental state has conceptually articulated content

I argue that (a) does not imply (b).

Now, I do point out, in the article, that many philosophers assume that
(a) implies (b). I think this is a mistake, and not one that Nietzsche
makes. I note that there are contemporary accounts of the way in which a
mental state could have determinate, yet non-conceptual content, such as
Stalnaker's "What Might Nonconceptual Content Be?" I also point out that
there are historical precedents for distinguishing (a) and (b):
Schopenhauer argued, explicitly and at length, that while non-human
animals have mental states with determinate contents, only human beings
have mental states with conceptual contents. In other words, Schopenhauer
explicitly endorses a distinction between (a) and (b).

So, while I agree with you that unconscious mental states must have
determinate intentional contents, I do not think this implies that
unconscious mental states must have conceptually articulated contents. (I
do agree with you, though, that there are philosophical puzzles concerning
how a state can have non-conceptual, yet determinate, intentional content.
I don't think that Nietzsche has a fully worked-out view on this topic,
though I make some suggestions in the article.)

Two additional, unrelated points:

- You note that most philosophers who are concerned with the
conceptual/non-conceptual distinction think it has nothing to do with the
conscious/unconscious distinction. I agree: I think that Nietzsche’s view
is novel in this respect. (It's worth noting, though, that some
philosophers who discuss the conceptual/nonconceptual distinction do
associate it with the personal/subpersonal distinction.)

- In GS 354, Nietzsche says that conscious thought, and only conscious
thought, is linguistically articulated. I take this to mean that it is
conceptually articulated. You criticize this interpretation on the
grounds that I don't establish the connection between language and
concepts. I take your point: I should have said more, here. However, I
just want to note that part of the motivation for my interpretation of
Nietzsche is that it accounts for, and indeed implies, some of Nietzsche's
puzzling claims about conscious thought. Nietzsche repeatedly claims that
conscious thoughts are generalized, superficial, and distorted versions of
unconscious states. In the article, I argue that this surprising and
abstruse claim is a consequence of the view that conscious mental states
are conceptualized versions of unconscious mental states. So I think my
interpretation of Nietzschean linguistic mental states as conceptual
mental states can be supported by considering how it entails other claims
that Nietzsche makes about consciousness.

Thanks again for your criticisms and thoughts. All the best,
Paul Katsafanas

Brian Leiter said...

Thanks for your response. Two quick thoughts in reply:

1. I understand that you argue that a mental state can have determinate content without being conceptually articulated. The difficulty is that all your examples involve perception, for which this is plausible. But how could a mental state like 'belief' or 'desire' be determinate without involving conceptual articulation? That's the puzzle.

2. There's no doubt that if N held the view you attribute to him it would be 'novel.' The worry is that it is also false, which is too high a price to pay for novelty!

Justin Tiehen said...

Here are a few thoughts from a philosopher of mind who’s trying to learn more about Nietzsche.

1. The claim that the conceptual/nonconceptual distinction has *nothing* to do with the conscious/unconscious distinction is controversial; there are a number of different ways people try to relate the two. The view being attributed to Nietzsche here – that some states have nonconceptual content, but that to rise to the level of conscious experience a state must have conceptual content – actually sounds quasi-McDowellian to me. Like McDowell’s view, it would seem to entail that animals and young children lacking conceptual sophistication – that is, lacking language, given the proposed link between language and concepts – are incapable of genuine conscious experience. Whether or not that’s an acceptable consequence, I’ll leave as an open question.

2. Today, many philosophers of mind think that there are (at least) two distinct ways of understanding the conceptual/nonconceptual distinction, and that these ways have been run together occasionally. On the first understanding, the distinction is between different types of content entities. More specifically, conceptual contents are Fregean Thoughts, that is, entities literally composed of concepts like MOTHER. By contrast, nonconceptual contents are entities not composed of concepts, like Russellian propositions or sets of possible worlds. On this understanding, could a (non-perceptual) belief have nonconceptual content? Yes, if belief contents are (say) sets of possible worlds. For what it’s worth, this is the understanding of the distinction Stalnaker operates with. He rejects Fregeanism about content, and thus holds that belief contents are always nonconceptual, never conceptual. I’m no Nietzsche scholar, but I would be surprised if this understanding of the distinction is of much relevance to Nietzsche: I doubt Nietzsche has any commitments in the debate between Fregeans and their critics.

3. On the second way of understanding the distinction, the crucial question is whether a given mental state imposes conceptual requirements on subjects. Is it a precondition on entering into a certain mental state that you must possess certain concepts? If so, the state has conceptual content; if not, it has nonconceptual content. So, take the belief that mother is sexually attractive. In order to have that belief, do you need to possess the concepts MOTHER and SEXUALLY? If there is such a conceptual requirement, the content of the belief is conceptual; if not, it’s nonconceptual. I can sort of see how this distinction might be relevant to Nietzsche. On this way of understanding things, though, it’s widely thought that all beliefs do have conceptual content – even Stalnaker concedes as much (in effect). So, it’s going to be pretty controversial to hold that unconscious beliefs have nonconceptual content in this sense, even among philosophers of mind who are otherwise friendly to the idea of nonconceptual content.

4. One last thought: the case might be different for other mental states, like desires, or wishes, or drives. Does a drive impose conceptual requirements on its subjects? It doesn’t strike me as completely crazy to hold that it doesn’t, but that nevertheless drives have determinate intentional (nonconceptual) content.

Brian Leiter said...

Justin, thanks for those very helpful comments. A question about your #4. It's plausible that *drives* could have determinate, nonconceptual content, but could *desires"? What it would mean to have an unconscious desire to sleep with your mother and kill your father if that desire is not conceptually articulated?

Rob said...

So, in DAYBREAK 119, when Nietzsche observes (according to the Hollingdale translation) that dreams are “inventions, which give scope and discharge to our drives to tenderness or humorousness or adventurousness or to our desire for music and mountains,” is he operating with a notion of drives (and analogous unconscious desires) as having conceptual content that is just less specific than Freud would have it? Or is this an instance of drives as having determinate (nonconceptual) content?

The the explanatory role of drives in Nietzsche's work has usually struck me as one of forces which dynamically support, encourage or facilitate more-or-less conscious states with conceptually articulated content, but lacking in themselves such content.

Anonymous said...

Regarding #4 and desires, here's a possible view. (I'm not endorsing this view, just noting its possibility.)

First, a theory of concepts: In order to possess the concept MOTHER, you must be disposed to draw certain rational inferences. E.g., "If X is a mother, then X is a female;" "If X is Y's mother, Y is X's child;" etc. According to this theory, if you lack these inferential dispositions, you don't possess MOTHER -- you don't really know what a mother is. (By the way, this theory makes possession of MOTHER require possession of other concepts, like WOMAN, CHILD, etc., which is the sort of view being attributed to Nietzsche by Kastafanas.)

Now, question: In order to have the unconscious desire that you sleep with your mother, do you need to possess MOTHER? That is, do you need to have the inferential dispositions in question? I don't think it's crazy to say No, you can have such a desire even if you lack the inferential dispositions in question. Unconscious desire just doesn't impose such rational requirements on its subjects, in the way that belief does. If this is correct, then given the theory of concepts in question, unconscious desire has nonconceptual content -- nonconceptual in that it doesn't require concept possession.

Again, I don't necessarily want to defend such a view, or to attribute it to Nietzsche. But, I don't think it's a complete non-starter.

--Justin Tiehen

"Q" the Enchanter said...

What's the difference between conceptual content and conceptually articulated content?

Narziss said...

Having gone through and read articles from both Drs. Leiter and Katsafanas, I've found both lines of thought to be interesting and attractive in different ways. I've tried to articulate my struggles and preferences here:

Consciousness: Conceptualization versus Epiphenomenalism