Naturalism Reading Group: McDowell's 'Naturalism and the Philosophy of Mind'

The dialectical structure of McDowell's paper is the following (missing out a few side points against Rorty and some linking of the argument to the history of philosophy).

(1) Lay out the distinction between the logical space of reasons within which knowing and thinking operate and the logical space of law within which the natural sciences explain the phenomena that constitute 'nature'. Accept this distinction as a genuine intellectual achievement, but refuse to identify 'nature' with the realm of law.

(2) Give motivation for this 'liberal naturalism': '[T]hinking and knowing are aspects of our lives' (p94) as rational animals, and so need to be included within the domain of nature.

(3) Point out that liberal naturalism accepts the 'sui generis' nature of activities that occupy the logical space of reasons, against 'restrictive naturalism' which does not, wanting instead to characterise such activities in terms of the realm of law.

(4) Explain the aim of isolating the realm of law, according to liberal naturalism: to understand its 'mode of intelligibility' so that it can be distinguished logically from the space of reasons. But also, to see how easy it is to conflate the realm of law and the realm of nature and thus to understand the root cause of our modern anxiety over the place of mind in nature.

(5) Diagnose the roots of this anxiety in the Cartesian philosophy of mind that attempts to include the special normative features of thinking within the realm of law: that is, such philosophy misunderstands the sui generis nature of the logical space of reasons, thinking of it as a demarcated region of the realm of law with special characteristics. Conversely, liberal naturalism properly understands the sui generis character of thought, and crafts a conception of nature that goes beyond the realm of law and is thus able to accommodate it.

(6) Show that modern 'restrictive' naturalistic philosophy of mind and language, such as that of Ruth Millikan, simply repeats the basic mistakes of the Cartesianism it purports to reject. Moreover, because of this it cannot explain the phenomenon of thought as it patterns our lives, so is to be rejected (such patterning is equated with Frege's conception of intentionality). Note, this last point is not that restrictive naturalism endangers some aspect of our folk psychology to which we are particularly wedded. Rather, it apparently endangers the very idea of thought.

I will do my best here to assess McDowell's case for (6). I will then say something about why 'restrictive' naturalists might not find his case against Millikan completely convincing, and why this hangs on what one's fundamental philosophical motivations happen to be.

McDowell takes up Millikan's rejection of what she calls Frege's 'meaning rationalism'. This is the idea that 'sameness and differences in elements of thought-content are transparently available to a rational subject.' (p100) Millikan finds fault with (amongst other things) Frege's assumption that intentionality is 'semantic rationality', which is (amongst other things) the thesis that grasping a sense is being intentionally directed towards an object (towards a Bedeutung). Thus grasping a sense is semantically rational in that 'it is a matter of what, for instance, can and cannot be true together.' (p101) Millikan takes this assumption to be 'substantive and indefensible' (ibid.). That she thinks it substantive is because - according to McDowell - she conceives of rationality as inhering in an 'intact mind' that is in good mechanical order. And this means that Frege (on this conception of the mind) is committed to the internal machinery of the mind being in such mechanical order that it could track the requirements of semantic rationality. But, Millikan points out, certain forms of semantic rationality presuppose directedness at determinate objects. 'And there are modes of directedness at determinate objects - for instance, those expressible by perceptually-based demonstratives - that we cannot get into our picture without appealing to environmental circumstances - circumstances external to the intra-organismic machinery that Millikan thinks would have to be meant by the 'intact mind'.' (ibid.) So Frege's conception of intentionality fails for Millikan because it conceives of rationality in semantic terms that are too strong to be accounted for by any properly (mechanically) functioning 'intact mind'.

But surely, McDowell argues, this is bad, for if sense and reference don't hang together in propositional attitudes as Frege takes them to, we lose partial grip on the rationality of thinking - we lose the ability to explain why a thinker is not irrational if she holds two conflicting propositional attitudes (say, belief and disbelief) toward the same object. Moreover, why not contrapose and reject Millikan's concepton of the 'intact mind' as properly functioning internal machinery, precisely because it can't account for semantic rationality? McDowell states 'to have an intact mind just is to be semantically rational.' (p102) The point being, only semantic rationality can account for the kind of normativity that structures the space of reasons. Millikan's mistake is to construe Frege's notion of sense as characterising 'the internal mechanics of having one's mind on objects.' (ibid.) But this is to treat the notion as characterising a special portion of the realm of law, and this is what her repetition of the Carteisan mistake consists in. Moreover, Millikan's allergic reaction to the idea of semantic rationality is that it would commit her to the 'spooky' idea that 'meanings move the mind directly' (p103). But this is only because she conceives of the mind as falling in the logical space of law. If it is seen as falling within the logical space of reasons, then the idea of meanings moving the mind becomes unproblematic: it becomes the idea that grasping senses describes 'patterns in our lives'. (ibid.) The basic mistake is trying to conceive of such 'grasping' in terms of mechanical understanding rather than the semantic rationality operative in the space of reasons.

With regard to Millikan's 'externalist' argument against Fregean intentionality, McDowell accepts that the environment plays a role in determining thought contents, but denies that we need displace thoughts from the rational animals that have them, as Millikan does, when she conceives of rationality as 'a norm effected in an integrated head-world system under biologically ideal conditions.' (ibid.) The problem with this form of externalism is that the thing that thinks is no longer identified with the rational animal that does the thinking.

Ultimately then, McDowell complains that Millikan fails to see his liberal-naturalist alternative because of her implicit Cartesianism (her conceiving of the mind as a special part of the domain of law). This is disastrous, because we end up losing our grip on thinking as something we do, which presumably was what the restrictive naturalist originally wanted to explain.

Now, I certainly think that if Millikan presents Frege's conception of thought in the psychologistic terms of properly functioning 'internal machinery', then she is wrong to do this (I have to say, I doubt that someone as smart as she is would do this). And it certainly seems a counterintuitive result to relocate thought outside the rational animal that thinks. But I have two worries about McDowell's argument here. First, does Millikan really conceive of the mind in terms of the realm of law? For she thinks of it in terms of 'proper biological function under ideal conditions', and that seems like a teleological and thus normative notion to me. Granted, this notion does not exhibit the same normativity that's operative in the 'space of reasons', but neither does it exhibit the non-normativity of the realm of law. Second, and more important, McDowell's motivation for refusing the assimilation of nature to the realm of law is to avoid the anxieties we experience in trying to account for thought without supernaturalism. Fair enough. But what if one is motivated by the idea that naturalism is driven by the aim of a unified scientific account of the world? Accepting the sui generis phenomenon of thought into the realm of nature is going to look ad hoc to someone so motivated, representing a failure of naturalism rather than a success. To that person the anxiety McDowell can't countenance is just an acceptable by-product of working towards such an account of the world, hopefully to be overcome in some future version. Don't we just reach a stand-off of competing philosophical motivations here? Perhaps McDowell can win the stand-off by demonstrating that restrictve naturalism leaves us with the 'grotesque' (as he puts it) result that thoughts are not wholly possessed by thinkers, or that restrictive naturalism can't account for the normativity of thought? But presumably, Millikan's conception of 'proper biological function under ideal conditions' is precisely supposed to handle the normativity of thought. Given her commitment to a unified scientific account of nature, perhaps partially extricating thoughts from thinkers heads is a price worth paying to achieve this goal? How do we mediate here? Can we appeal to the common sense of McDowell's neo-Aristotelianism? One thing is sure, simply slinging mud won't help, as McDowell does at the end of the paper by calling restrictive naturalists 'scientistic'.

In Mind and World, McDowell takes the line that he is just providing an alternative account of naturalism and intentionality, one where we don't have to stay on the seesaw of Myth of the Given/Coherentism (he uses the language of showing up a blindspot, an unseen possibility). So there he admits that the 'therapy' he offers is available to those who want it, but he doesn't claim that it is mandatory for all. Here, he seems to take a tone where rather than simply providing us with an alternative possibility, he's actually closing off the possibility of restrictive naturalism. But the arguments don't seem to warrant this stronger line to me.

Reading Group: Huw Price, Naturalism without Representation

Price argues against a view he calls 'object naturalism', whilst defending a view he calls 'subject naturalism'. The former is the view that, "all there is is the world studied by science...all genuine knowledge is scientific knowledge." (73). The latter is the view that, "Science tells us that we humans are natural creatures, and if the claims of philosophy conflict with this view, then philosophy needs to give way." (73).

As I understand it, the case against object naturalism involves the following claims:

(1) If object naturalism is to stand a chance of being well motivated, it must assume a substantial account of semantic relations, such as reference.

(2) An object naturalist needs to offer a naturalistic, that is empirical, account of the semantic relations themselves.

(3) An empirical investigation into the semantic notions must treat 'semantic irrealism' as a live empirical option - semantic irrealism being the claim that, for example, 'reference' fails to refer.

(4) It is impossible to make sense of the empirical possibility of this sort of semantic irrealism, since in asserting the irrealism one has employed the semantic notion of reference.

Price says of this argument that, "it exposes what is inevitably a nonnaturalistic presupposition in the leading contemporary conception of what is involved in taking science seriously in philosophy." (81). Interestingly, it looks rather like a sort of transcendental argument - naturalism is undermined by the fact that it must make a nonnaturalistic presupposition about reference.

He goes on to argue that his own subject naturalism fares better is it needn't assume a substantial account of reference. Thus it's countenancing the possibility that science will not force us to employ semantic notions is not the denial that 'reference' refers, but the refusal to say anything about reference.


I want to say something about (1) and (2). First (1):

(1) is the claim that object naturalism is inconsistent with a deflationary account of the semantic notions. Price suggests that this follows from the view that the 'data' for a naturalistic account in a given domain is linguistic: "we note that humans (ourselves or others) employ the term "X" in language, or the concept X, in thought. In the light of a commitment to object naturalism...we come to wonder how what these speakers are thereby talking or thinking about could be the kind of thing studied by science."(76).

Price spends some time defending this linguistic conception of the 'data' of naturalism. But I'm interested in the move from the linguistic conception to the claim that naturalism presupposes a substantial account of reference. His argument is that if a deflationary account is accepted, then there really is no difference between taking the 'data' to be linguistic practice or non-linguisitic reality. This is because, on the deflationary view, saying that the sentence, "Snow is white" is true, is just to say that snow is white.

It seems to me that this argument is only convincing if our reasons for believing in the linguistic conception of the 'data' are simultaneously reasons for disbelieving in the non-linguistic reality conception of the 'data'. But are they? I won't go into the discussion of Lewis and the Canberra Plan, but it strikes me that the point about non-cognitivism loses some of its force. Price argues that since some naturalistic views are views about linguistic practice (e.g. non-cogntivism), then they will simply be ignored if we take the 'data' to be non-linguisitic reality. But this seems wrong, if the reason for taking the data to be non-linguistic reality is a deflationary account of reference. This view wouldn't ignore non-cognitivism, it would presumably consider itself to have some reason for thinking it false, and that is a different matter.

Now (3). (3) is the claim that an object naturalist needs to offer a naturalistic, that is empirical, account of the semantic relations themselves. This seems to be something that Price takes to be too obvious to really argue for. But I wonder whether a naturalist need accept it. For, Price's definition of object naturalism includes both ontological and epistemological naturalists, let's say, physicalists and empiricists. But what reason is there to think a that a physicalist need accept (3) unless he or she is also committed to empricism?

Price says that, "Naturalism of this kind [object naturalism] seems committed to the empirical contingency of semantic relations. For any given term or sentence, it must be to some extent an empirical matter whether,and if so to what, that term refers" (81). But this doesn't seem to be an obvious conseqence of a committment to physicalism. So, it strikes me that someone who thinks that "all there is is the world studied by science", even if 'science' is read as 'physics' (note, quite a lot hangs on whether psychology counts as a science, and if so which branches it it), need think that the proper investigation of semantic notions is empirical in the above sense.

But even if one thinks that, "all genuine knowledge is scientific knowledge" (73), there may still be room to reject (3). For we are owed an account of what counts as scientific knowledge. Here's a thought. A great deal of scientific knowledge is achieved via the method of inference to the best explanation. Perhaps the best explanation of certain aspects of our competence with semantic, mathematical and logical terms is that we possess some knowledge the canonical justification of which is not experiential, i.e. is apriori. Such a view will not, of course, be an empiricism. But given its reliance on inference to the best explanation, one might be tempted to accord it the status 'scientific'. Maybe, maybe not. What does seem true to me is that (2) is in need of more support that Price gives it in this paper.

Reading Group: Hilary Putnam, 'The Content and Appeal of Naturalism'

Putnam has two criticisms of (scientific) naturalism:

(I) Naturalism cannot give stable accounts of semantic content so that central naturalistic claims about language imply views that are not consistent with an overall naturalistic perspective (anti-realism or even supernaturalism).

(II) Naturalism has anti-supernaturalistic appeal, but goes too far in its reductionist fervour -- Putnam attributes this to fear of the normative.

The essay starts with a survey of naturalistic attitudes towards semantic issues. As is well known, naturalism privileges the formalised languages of science, in particular physics, over language as we know and use it. Language, meaning and reference in the ordinary sense may or may not be part of the naturalistic picture of language. At any rate, there is a distinctive hierarchy between first-rate conceptual systems with contents that pertain to the absolute conception of the world sought by physics. Only statements of this (fairly rarefied) kind are bona fide statements. If less formal statements are considered at all, naturalism strives to reduce their contents to those of first-order statements. Putnam stresses, of course, that some areas of discourse resist this reduction to what pertains to an absolute conception of the world -- talk of desire, belief, meaning or reference. Naturalism may even accept this. But it is bound to take the irreducibility of certain areas of discourse as a symptom for their groundlessness. This should raise suspicions about this reductionist approach, or so Putnam argues.

This survey leads to Putnam’s summary critique of Quine’s view that reference is indeterminate and inscrutable. Quine is opposed to ordinary ideas about word-world correspondence on the grounds that there is no ostentative distinction between an object, undetached object parts or object stages. Therefore it cannot be determined whether ordinary sentences refer to the objects and relations between objects we take them to speak about literally or whether they refer to parts or stages of objects. On a different presentation of ontological relativity, it cannot be determined whether our ordinary statements refer to what we take them to speak about literally or to complement events. Maybe it should be added at this point that from a naturalistic point of view only highly formalised languages are equipped to deal with ontological relativity. This is in part because they cover a limited and very specific range of objects, in part because there are appropriate (scientific) tools for individuating what the statements of these languages refer to. In their case, reference is in fact reduced to an aspect of scientific method.

Putnam’s main point seems to be this: If word-world correspondence is abandoned along with a sui generis concept of reference, it is not clear what language does. But it would be highly counterintuitive to deny that language makes a determinate contribution to our accounts of the world. Naturalists may reply that language is still connected to objects on their view, albeit causally and not in terms of reference. But in Putnam’s view this amendment only conceals the problem that the naturalistic position on language is ant-realist. He believes that this endangers the overall realism of (scientific) naturalism. What is left of realism if utterances have no semantic relation to an appropriate part of reality, he suggests, is too thin to support naturalism. If propositions do not refer to reality in a determinate sense, preference for the (scientific) naturalistic view that they transmit or organise neural stimulation over, say, the notion that language helps creatures like us to cope with the world (Rorty) is a matter of predilection. Putnam seems to believe that this ambiguity seals the fate of scientific naturalism as an overall realist position.

Turning to (II), like Stroud and Dupré, Putnam wants to engage with the acclaim that naturalism appears to enjoy today. He does this from his own perspective: ‘”Naturalism”, I believe, is often driven by the fear that accepting conceptual pluralism will let in the “occult”, the “supernatural”.’ (66) To keep the occult at bay, naturalism adheres to two broad rules:

(1) All things are mereological sums of physical objects.
(2) The only admissible explanations are complete explanations in physical terms.

Failure to respect (1) means that outer-worldly entities can be included in one’s ontology. Failure to respect (2) means allowing events to be accounted for by miracles. However, according to Putnam, these rules create more problems than they solve.

In the case of (1), claiming that X is a mereological sum of its physical parts is unproblematic as long as I just mean that X is never disconnected from its parts. If I assume that this means the logical identity between X and the mereological sum of its physical parts, however, I could, for instance, create mereological sums of my parts which predate me by several centuries. Evidently, these sums wouldn’t be me.

Putnam argues against (2) that naturalism is in fact selective with regard to what it takes to be a pernicious and a benign incomplete explanation. The basic vocabulary of geology, for instance, does not map onto that of physics. Naturalists don’t believe this to be obscure. What they do not tolerate, however, are normative terms like true, justified, adequate, referring. According to Putnam, naturalism appears to imply that if these normative terms cannot be reduced to non-normative ones, it is unavoidable for us that we posit a supernatural realm of values. Contrary to the parsimonious understanding of language in naturalism, Putnam’s conceptual pluralism states that a wide variety of indicative sentences, including normative ones, are bona fide statements. These normative statements are not descriptions of reality. So there is no question of them relating to that occult realm. The naturalist fear is unfounded.

One of the things I don’t understand is how Putnam gets from instability of content (I) to horror of the normative (II). More specifically, I don’t see why Putnam’s conceptual pluralism is the most compelling solution to the semantic instability of naturalism. There seem to be other solutions to the semantic instability of naturalism. One of these is naturalistic in the sense attacked: to seek a different first-order conceptual system so that reference is not unstable (Millikan). Another solution is mentioned by Putnam: if we follow Rorty, language does not consist of bona fide statements to start with. So there is no particular issue with the inclusion of normative statements. But there is no special weight to them either that would suggest that those singling out specific bona fide statements commit another mistake (anti-normativism) in addition to the mistake of ascribing special status to certain statements.

Then, I wonder whether Putnam’s blanket accusation of normative horror might have profited from some reference to emotivism, expressivism or prescriptivism. Though his accusation is more concerned with normative terms like truth, justification rather than those of ethics, emotivism et al are broadly naturalistic attempts to say something about normativity without making it disappear. They talk about normative claims that are not descriptions of reality. The main difference to Putnam’s position is that emotivism et al do not treat normative claims as bona fide statements. In particular, Putnam might have explained vis-à-vis emotivism etc why his internal realism gives us a better idea than emotivism et al about claims that are not descriptions of reality. This might have prompted him to restate his blanket accusation of normative horror as a critique of non-cognitivism about norms. And this in turn might have incited him to reveal more about the systematic architecture in the background of his critique.