Essay Archive

What the Disjunctivist Knows

Mon, 2 Jun 2008 14:03:49 +0100

Does disjunctivism about perceptual experience provide a satisfactory response to the sceptic?
I. Introduction
Disjunctive accounts of perceptual experience have been advanced by Hinton (1973), Snowdon (1990) and others (e.g. Martin 2004) as an alternative to traditional accounts of perception that conceive of such experience on the model of an externally caused ‘inner object’ or event. Whilst generally regarded as a thesis concerning the nature and individuation of experience, some philosophers — chiefly John McDowell (1982; 2008) — have emphasised the theory’s epistemological benefits, claiming that it is capable of explaining how we can have knowledge of the external world as opposed to knowledge of mere appearances — an issue that dates back to the work of Descartes (1641), Hume (1739; 1748), Locke (1690), and others. This essay evaluates whether the disjunctivist conception of perceptual experience is able to yield a satisfactory response to such external world scepticism and, if so, what kind of a response it is. In particular, I will focus upon McDowell’s claim that disjunctivism removes a crucial ‘prop’ upon which such scepticism depends, thus placing the disjunctivist on a firmer dialectical footing than his opponent — a claim that is vigourously denied by Wright (1982; 2008). I begin by describing the disjunctivist conception of perceptual experience in Section II, followed by McDowell’s ‘epistemological disjunctivism’ in Section III. Section IV deals with the issue of higher-order scepticism, and I conclude the final section with some remarks concerning the effectiveness of the disjunctivist response.
II. Perceptual Disjunctivism
The central claim of disjunctivism is that the veridical perception of an external object X and a phenomenally indistinguishable hallucination of X do not, as a matter of conceptual necessity, share a common psychological or experiential element. Rather, in the genuinely perceptual or ‘good’ case, the experience is essentially relational, since it relates the subject to an external object; e.g. a zebra (to borrow Dretske’s [2005] example). In the non-perceptual or ‘bad’ case, the experience is non-relational since it only involves there seeming to be a zebra in front of me, and thus has no external object. By application of Leibniz’s Law, a relational and non-relational experience cannot be numerically identical, since they have different properties, thus giving rise to the so-called disjunctive conception of experience. My experience of it looking to me as if there is a zebra in front of me is either a case of my seeing that there is a zebra or of my imagining, hallucinating, etc. that there is a zebra, where the two disjuncts are externally, rather than internally, individuated. Whilst the disjunctivist is not committed to maintaining that the two cases have literally nothing in common, such as the subject’s brain state or how the experience represents the world as being, the claim is that they constitute two different types of experience — or at least that this cannot be ruled out a priori (Snowdon 1990: 129). This may be contrasted with the traditional ‘Lockean’ conception of experience as an inner process or state that is common to both veridical and non-veridical cases, and which in a materialist theory of mind might be identified with or supervene upon a subject’s brain state, for example.1
Perhaps the greatest advantage of such ‘perceptual disjunctivism’ is its ability to explain how perception can make external objects available for demonstrative thought (Snowdon 1990: 143). On the Lockean conception of experience, it is unclear what makes it possible for me to think of that particular zebra, even when it is right in front of me, since the psychological experience that I enjoy is non-relational, and therefore logically and metaphysically independent of the object itself. In order to bridge this gap, some further (e.g. causal) condition must be satisfied since the very same experience would be compatible with there being no such external object. The disjunctive conception of perceptual experience, on the other hand, conceives of the subject as standing in some particular relation (e.g. that of seeing) to an actual object, from which the appearance cannot, so to speak, become ‘detached’, thus making the object available for demonstrative thought. The relationality of perceptual experience makes it essentially object- or world-involving, as opposed to hallucination, which is a purely internal phenomenon.2
Before moving on to consider the epistemological significance of disjunctivism, it is worth noting its connection with some broader philosophical themes. The disjunctivist thesis essentially amounts to a claim about the scope of experience. The traditional conception of experience is that it has narrow scope, with experiences being individuated phenomenologically by virtue of their representational content or phenomenal character. The disjunctive conception, on the other hand, claims that experience has broad scope with experiences being individuated epistemologically on the basis of factors external to the subject. Just as externalism about meaning (e.g. Putnam 1975; Burge 1979) or representational content (Tye 1993) is intended to overcome problems concerning the individuation of thoughts or sensations by taking external contextual factors into account, perceptual disjunctivism is intended to explain how the nature of experience can differ even where the content of those experiences is phenomenologically indistinguishable. Corresponding to these notions of broad and narrow experience are two different notions of justification. Let us say that an individual is subjectively justified, or S-justified, if and only if they have good reason to believe X and are ‘epistemically blameless’ in the sense that they have done nothing wrong in their method of acquiring the relevant belief. Alternatively, a subject is all-things-considered justified, or A-justified, if and only if, were they in possession of all the facts, they would be justified in believing X.3 External world scepticism can be described as attempting to drive a wedge between S- and A-justification, claiming that we are only ever entitled to claim the former, whilst only the latter is capable of yielding knowledge. McDowell’s response to this challenge is both subtle and radical since it attempts to undermine the plausibility of the sceptic’s position by reversing the direction of explanation for what is required for knowledge, as described below.
III. Epistemological Disjunctivism
McDowell’s response to external world scepticism has two distinct phases, or parts, each of which I shall consider in turn. Firstly, the disjunctive conception of experience is used to explain how we can have direct, non-inferential knowledge of the external world. This is intended to secure the possibility of first-order knowledge in the face of sceptical doubt. Secondly, a form of transcendental argument is employed to undermine the plausibility of the sceptic’s own account of knowledge. This is intended to shift the burden of proof away from the common-sense notion of knowledge that we ordinarily employ, and which McDowell’s account supports, and onto the sceptic, who must then give good reason why we should call our everyday knowledge claims into doubt. The issue of second-order knowledge, i.e. our knowing that we know, is not directly addressed by McDowell, but I will argue that a similar strategy may be employed as for first-order knowledge using the disjunctive conception of experience to justify our second- and higher-order knowledge claims.4 It is important to note that at no point does McDowell profess to have refuted the sceptic, but rather his intention is ‘diagnostic’ (McDowell 1992: 378). Indeed, he agrees with Stroud that ‘the worst thing one can do with the traditional question about our knowledge of the world is to try to answer it’ (Stroud 1980: 56, in McDowell 2008: 384). His strategy does not therefore aim to convince the sceptic of the falsity of his argument on some mutually acceptable grounds, but rather to gain dialectical advantage by undermining the coherence and apparent attractiveness of the sceptical position. This will become relevant when evaluating the success (or otherwise) of McDowell’s epistemological disjunctivism.
McDowell (2008: 378–9) maintains that the initial plausibility of external world scepticism (hereafter simply referred to as ‘scepticism’) stems from the fact that the traditional Cartesian view of perceptual knowledge does not adequately secure the possibility of epistemic access to worldly facts. On the narrow conception of experience described above, perceptual experience warrants belief — and thereby knowledge — only when it stands in the appropriate (typically causal) relation to the relevant state of affairs in the external world. However, since precisely the same experience could be enjoyed in the absence of this state of affairs — in the case of a hallucination, for example — such experience can at best give defeasible warrant for belief, since it cannot yield knowledge unless the relevant external condition is satisfied; e.g. when there really is a zebra in front of me. For McDowell, this places perceptual knowledge on dubious foundations since the ‘gap’ between what experience warrants and knowledge makes it easy for the sceptic to claim that the necessary conditions may not have been satisfied due to some as-yet-unconsidered factor, such as our being deceived by a malicious demon (Descartes 1641: 15), by being a ‘brain in a vat’ (Putnam 1981), or through the presence of some cleverly disguised mules (Dretske 2005). This amounts to the claim that, since perceptual experience only involves its seeming to us that p, such experience can only warrant disjunctive beliefs of the form: either we are perceiving that p or we are imagining (hallucinating, etc.) that p. This places the sceptical scenario on precisely the same footing as that of veridical perception. Expressed in the terms defined above, when conceived as having narrow scope, perceptual experience can only yield S-justification, and never A-justification, since no possible evidence is sufficient to rule out the possibility that we are being systematically deceived.
By rejecting the Lockean view in favour of a disjunctivist account of perceptual experience as individuated in terms of its epistemological — rather than phenomenological — properties, McDowell aims to prevent the first-order sceptical argument from gaining a foothold. On this view, veridical perceptual experience is relational by definition and so involves a subject standing in a particular relation to some external world object.5 The presence of the object is intrinsic to the nature of the experience, since on the disjunctivist view without it the subject could no longer be said to have had the same experience. In the ‘good’ case, perceptual experience warrants the belief that there is, for example, a zebra in front of me, since the experience itself entails that the object is present. In the corresponding ‘bad’ case — e.g. in the presence of a cleverly disguised mule — it merely seems that such an object is present, and so any resulting belief that there is a zebra would be unwarranted. In contrast to the Lockean picture where non-disjunctive experience only warrants disjunctive (i.e. sceptical) belief, we now have two disjuncts, one of which warrants belief (or knowledge) and the other which does not, but both of which are phenomenologically indistinguishable to the subject. On McDowell’s picture, however, the fact that the subject may not know which disjunct their experience falls under does not undermine their epistemic standing in the ‘good’ case. At worst, they may form an erroneous (i.e. unwarranted) belief by mistaking a hallucination or fake for the real thing, but this is different to saying that their warrant is defeasible. Indeed, it is arguably the notion of defeasibility, which arises from the narrow conception of perceptual experience, that enables the sceptic to claim that we only ever have S-justification, since we cannot know whether the relevant external conditions have been satisfied. For McDowell, on the other hand, we can have both S- and A-justification in virtue of our direct epistemic access to the facts through our faculties of perception and higher-order cognitive abilities.
It is crucial to McDowell’s account that the reason or justification which warrants perceptual knowledge is not something over and above or presupposed by experience, but perceptual experience itself. Thus, instead of perceptual justification (e.g. seeing a zebra-looking object) being distinct from the beliefs that it warrants (i.e. that there is a zebra), my seeing that there is a zebra directly warrants my belief, since the experience already ‘takes in’ the relevant fact (McDowell 1982: 472). This closely matches ordinary linguistic practice, since when prompted to justify a given piece of perceptual knowledge we typically respond by saying that we just see that it is so, rather than offering any independent justification in support of this fact or the reliability of our perceptual faculties.6 Wright (1982: 345), however, rejects this proposal, claiming that it presupposes precisely what the sceptic denies, i.e. that there is an external world, making it unsuitable for use in an argument that purports to grant knowledge of the external world since it is effectively question-begging. McDowell (2008: 384–5) responds by noting that ‘the point of the disjunctive conception is not to improve our resources for such arguments’, but rather to ‘remove a prop’ (Wright’s own phrase) upon which such scepticism depends; i.e. the assumption that perceptual experience cannot by itself ‘reveal how things are’ (ibid.), but instead stands in need of independent justification. By rejecting this assumption, the disjunctivist effectively blocks the possibility of first-order doubt as to whether such justification is available, since it is no longer separable from the experience itself, instead forcing the sceptic to concentrate upon the second-order issue of whether the experience is in fact perceptual or not, as discussed below. Consequently, Wright’s (2008: 398) insistence that a disjunctive conception of experience fails to warrant knowledge or belief is beside the point, since it is not the disjunction as a whole to which warrant is attached, but to each of the disjuncts (McDowell 2008: 386). This insistence demonstrates the pull of the precisely the narrow conception of experience that the disjunctivist denies, effectively conflating the issue of whether a subject has warrant with the separate question of whether the warranting status of an experience is accessible to the subject, where the latter is a matter of second-order knowledge.
The model of knowledge that McDowell presents is one in which the subject is genuinely ‘open to the world’ and the facts within it without requiring the satisfaction of further (e.g. causal) conditions. Due to its relational nature, the broad conception of perceptual experience is essentially world-involving, and so there is nothing to come between the subject and their knowledge of the world in the ‘good’ case from which the sceptic can construct his counter-argument. Although this may show how perceptual knowledge of the world is possible, it falls short of banishing sceptical doubt altogether since there is as yet no way for the subject to tell whether her experience is in fact perceptual or not, since (by stipulation) the ‘good’ case is subjectively indistinguishable from the ‘bad’.7 At this point in the dialectic, the sceptic would seem to have two options. He can either accept McDowell’s account of perceptual experience and redeploy his argument against the possibility of second-order knowledge, or else reject the disjunctivist conception of experience upon which it is based. However, since it cannot be assumed a priori that disjunctivism is false (Snowdon 1990: 129), it is not clear that the latter option is available, and so the sceptic must directly confront the legitimacy of second-order knowledge claims.8
IV. Second-Order Scepticism
Having secured the possibility that we have perceptual knowledge of the external world, we now turn to the second part of McDowell’s argument, which concerns the issue of whether we can know (or claim) that we have such knowledge. Rather than attempting to argue for this from a neutral starting point that is acceptable to the sceptic — a task that he acknowledges to be futile (McDowell 2008: 386) — McDowell instead employs a transcendental argument aimed at undermining the apparent coherence and plausibility of the sceptic’s own position (ibid. 382). The argument proceeds from the premise that, given the importance and centrality of the concept of knowledge in our everyday experience as a bearer of objective significance, we are rationally bound to adopt a model of knowledge that can make sense of this fact. After all, if our concept of knowledge were completely without foundation then we would be unable to make sense of many of our everyday practices, including perception, learning, scientific investigation, and so on. McDowell claims that the sceptic both grants that our knowledge aims to have ‘objective purport’ whilst simultaneously denying that it can do so in practice (ibid. 380). This position is unstable since it undermines the very idea of knowledge that the sceptic is presupposing, as well as failing to make any of the above practices intelligible. The disjunctivist, on the other hand, can plausibly claim that in the ‘good’ case, experience does indeed have objective purport, and so the distinction between knowing and not knowing is well-founded, thereby rendering our practices intelligible. All other things being equal, we should prefer a model of knowledge that is able to do this, thus making the disjunctivist’s account of knowledge more plausible than the sceptic’s.
This argument is all well and good, but even if the sceptic were to grant McDowell his conclusion, it is still possible (although perhaps unlikely) that all of our experience falls under the ‘bad’ case. After all, this is precisely what the machinations of the Cartesian demon or vat-scientist were supposed to illustrate: that it is entirely consistent with the content of our experience (narrowly construed) that all of our beliefs could turn out to be false for a reason that it is beyond our powers of discrimination to discount. Whilst McDowell appears reluctant to engage with this possibility, a suitable McDowellian response may perhaps be given along the following lines. Just as our first-order perceptual abilities place us in a direct relation to worldly objects and facts, we also have second-order cognitive abilities that enable us to tell from the surrounding context and situation whether these first-order experiences are in fact veridical.9 Indeed, most if not all perceptual experiences contains some indication of their own veridicality, which is why we are inclined to take some experiences at face value and to doubt others; e.g. a fleeting glimpse, unconvincing facsimile, or dream-like visual impressions. This is not to say that such second-order abilities are infallible any more than our perceptual abilities are infallible since we do not know in all cases whether we are being deceived. However, the McDowellian conception of knowledge is perfectly able to explain how it is that we have direct epistemic access to the fact that we can and do perceive veridically in a wide variety of cases. Just as having a particular (broadly constituted) perceptual experience can itself warrant first-order belief or knowledge, the very same experience, along with the appropriate contextual and cognitive factors, can warrant the corresponding second-order knowledge claim. As with first-order knowledge, such claims do not require the existence of any independent justification or warrant over and above the initial perceptual experience along with our previous training and experience in recognising instances of ‘good’ and ‘bad’ cases. Thus, the claim is not that we always — or even most of the time — know whether or not we perceive veridically, but that in the ‘good’ case, we can know that we know since our higher-order cognitive abilities are able to ‘take in’ this higher-order fact. Of course, there is still the possibility that in the ‘bad’ case we may be mistaken, but the possibility of such false knowledge claims does not justify the sceptical conclusion that none of our knowledge claims are warranted, since the disjunctivist’s conception of experience has already shown how we can have direct epistemic access to the facts in the ‘good’ case.
The structure of the above argument exactly parallels that of McDowell’s argument for first-order knowledge. Once again, the sceptic cannot rule out the possibility that the acquisition of such higher-order knowledge is possible, but will seek to cast doubt upon its reliability, shifting the argument to an even high-order form of scepticism. Surely, he will claim, any malicious demon or evil scientist worth their salt would set things up so that it would seem to us as if we had knowledge even where we have none? Here it may seem as if the disjunctivist response does little more than to set up an infinite regress in which both sides perpetually claim victory by invoking ever higher-order cognitive abilities and higher-order sceptical doubts. However, at this stage the McDowellian disjunctivist can justifiably claim that their model of knowledge offers a more plausible basis for our everyday use of the term than that of the sceptic, which seeks to render the very possibility of knowledge unobtainable, thereby rendering the concept meaningless.10 The disjunctivist account of first-order perceptual and higher-order cognitive abilities is therefore better placed to make sense of our everyday talk and practices surrounding knowledge whilst acknowledging that we may be more or less fallible in individual cases. What it denies is that such fallibility requires us to engage with the sceptical argument on its own terms, since both first- and higher-order knowledge are, so to speak, already within our grasp.
V. Conclusion
Whether disjunctivism is capable of supporting a satisfactory response to external world scepticism largely depends upon the issue of who such a response is required to satisfy. McDowell’s epistemological disjunctivism will certainly be unsatisfactory to any committed sceptic, who will accuse him of rejecting the Lockean conception of experience out of hand, or to those who will settle for nothing less than a full-blown refutation of the sceptic’s argument. On these views, perhaps the best that could be said about McDowell’s account is that it re-describes the facts to show how knowledge might be possible, but falls short of giving any decisive response to the sceptic, and ultimately fails to engage with the argument. Such a conclusion would, however, be unwarranted since it both underestimates the resources available to the disjunctivist about perceptual experience as well as misreading the whole tenor of McDowell’s approach which was to avoid the need to engage with scepticism on its own terms — a task that arguably never had any prospect of success in the first place.
Perhaps a better way of understanding McDowell’s anti-scepticism is as a form of Wittgensteinian ‘philosophical therapy’ (McGinn 1997: 23), designed not to refute the sceptic, but to relieve us of an anxiety concerning the demands that knowledge places upon perceptual experience. As such, we need only be satisfied that first- and higher-order knowledge is available to us in the ‘good’ case, and that the sceptic’s argument, whilst acknowledging its initial appeal, need not prove compelling provided that we are willing to accept that justification for both types of knowledge can arise from having a particular kind of externally individuated perceptual experience. By conceiving of knowledge in terms of direct epistemic access to the world, or facts, rather than attempting to build it up from subjective justification plus the satisfaction of some further external condition, McDowell rejects the sceptical premise that we require independent justification over and above perceptual experience — our primary means of knowing about the world. This in turn underpins an attack on the sceptic’s model of knowledge, which fails to account for our everyday ‘forms of life’, thereby shifting the burden of proof back onto the sceptic. In this way, McDowell seeks to reassure us that we should not be seduced into letting the logic of the sceptic’s position undermine our confidence in our everyday practices and knowledge claims. When expressed in these terms, disjunctivism supports a powerful and effective remedy against what tempted us towards scepticism in the first place, thus putting our sceptical doubts back into perspective, rather than placing them at the forefront of our epistemological concerns.
——————
1 Disjunctivism does not entail that experiential states do not supervene upon brain states, only that they do not merely do so.
2 Martin (2004: 42–3) argues that disjunctivism also has the benefit of presenting the least revisionary account of perceptual experience, since both sense-datum theory and traditional empiricism require that we are mistaken as to the nature of the objects of experience.
3 Neither kind of justification entails the other, since it is possible for a subject who is A-justified to have good evidence for the falsity of their belief, thus undermining S-justification, and vice versa.
4 For brevity, I will treat knowledge claims as a form of second-order knowledge. Nothing in the argument turns on this distinction.
5 Here, I will talk only of objects, but the argument applies equally to facts.
6 Compare, for example, the absurdity of responding with the disjunctive justification of ‘either I see that it is so or I am hallucinating it, and I have good reason to believe that my perception is reliable’.
7 Indeed, Wright (2008: 401) argues that Cartesian scepticism was always an issue of second-order not first-order knowledge, and that McDowell’s argument simply misses the point.
8 Whether it would be sufficient for the sceptic to remain agnostic about which model of perception is correct is discussed below.
9 Millar (2007, 2008, forthcoming) calls such capacities ‘recognitional abilities’.
10 An alternative strategy would be to ‘relativise’ the concept of knowledge to whatever epistemic situation it turns out we are in, as per Putnam (1981), but this would concede too much to the sceptic.

Bibliography
Burge, Tyler 1979: ‘Individualism and the Mental’. In Studies in Metaphysics, P. French, T. Uehling and H. Wettstein (eds.), Minneapolis: University of Minnesota Press.
Descartes, René 1641: Meditations on First Philosophy. Reprinted in J. Cottingham (ed. & trans.), Cambridge: Cambridge University Press, 1986.
Dretske, Fred 2005: ‘Is Knowledge Closed Under Known Entailment?’. In Contemporary Debates in Epistemology, M. Steup & E. Sosa (eds.), Malden, MA: Blackwell. pp. 13–26.
Haddock, Adrian and Fiona Macpherson (eds.) 2008: Disjunctivism: Perception, Action, Knowledge. Oxford: Oxford University Press.
————— 2008: ‘The Disjunctive Conception of Experience as Material for a Transcendental Argument’. In Haddock & Macpherson (eds.), pp. 376–89.
Hinton, J. M. 1973: Experiences. Oxford: Clarendon Press.
Hume, David 1739: A Treatise of Human Nature. Reprinted in D. F. Norton and M. J. Norton (eds.), Oxford: Oxford University Press, 2000.
————— 1748: An Enqury Concerning Human Understanding. Reprinted in T. L. Beauchamp (ed.), Oxford: Oxford University Press, 1999.
Locke, John 1690: An Essay Concerning Human Understanding. Reprinted in P. Nidditch (ed.), Oxford: Clarendon Press, 1975.
Martin, M. G. F. 2004: ‘The Limits of Self-Awareness’. Philosophical Studies, 120, pp. 37–89.
Millar, Alan 2007: ‘What the Disjunctivist is Right About’. Philosophy and Phenomenological Research, 74, pp. 176–98.
————— 2008: ‘Perceptual-Recognitional Abilities and Perceptual Knowledge’. In Haddock & Macpherson (eds.), pp. 330–47.
————— forthcoming: ‘Disjunctivism and Scepticism’. In The Oxford Handbook to Scepticism, J. Greco (ed.), Oxford: Oxford University Press.
McDowell, John 1982: ‘Criteria, Defeasibility, and Knowledge’. Proceedings of the British Academy, 68, pp. 455–79.
————— 2008: ‘The Disjunctive Conception of Experience as Material for a Transcendental Argument’. In Haddock & Macpherson (eds.), pp. 376–89.
McGinn, Marie 1997: Wittgenstein and the Philosophical Investigations. Oxon: Routledge.
Putnam, Hilary 1975: ‘The Meaning of “Meaning”.’ In Mind, Language and Reality: Philosophical Papers Volume 2, Cambridge: Cambridge University Press.
————— 1981: ‘Brains in a Vat’. In Reason, Truth and History. Cambridge: Cambridge University Press, pp. 1–21.
Snowdon, Paul 1990: ‘The Objects of Perceptual Experience’. Aristotelian Society Supplementary Volume, 64, pp. 121–50.
————— 2005: ‘The Formulation of Disjunctivism: A Response to Fish’. Proceedings of the Aristotelian Society, 105 (1), 129–41.
Michael Tye 1993: ‘Visual Qualia and Visual Content’. In The Contents of Experience, T. Crane (ed.), Cambridge University Press, pp. 158–76.
Wright, Crispin 1982: ‘Anti-Realist Semantics: The Role of Criteria’. In Idealism: Past and Present, G. Vesey (ed.), Cambridge: Cambridge University Press, pp. 225–48.
————— 2008: ‘Comments on John McDowell’. In Haddock & Macpherson (eds.), pp. 390–404.

Neutral Monism as a Response to the Conceivability Argument

Sat, 31 May 2008 19:04:37 +0100

I. Introduction
The conceivability argument (hereafter abbreviated as CA) is an argument against any kind of materialism that takes mental and physical events1 to be numerically identical, but only contingently so. Such arguments have been given by Descartes (1641: §4), Kripke (1980) and Chalmers (1996) and concern the conceivability, and thereby the possibility, of disembodied minds and mindless bodies, the latter of which are often referred to as philosophical ‘zombies’ (ibid. 94) — beings that look and behave exactly like us and yet have no ‘inner life’ or conscious experience. Standard responses to these arguments include a priori materialism (Stoljar 2000), substance or property dualism (Chalmers 1996), and contingent identities between ‘non-rigid designators’ (Jacquette 1987), as described in Section II. Perhaps a less well-known response, however, is based upon the metaphysical doctrine of neutral monism — the view that the universe is neither fundamentally physical nor mental, but composed of ‘neutral’ elements from which all other substances and properties are comprised. This view was famously held by Russell (1921; 1927), Mach (1886) and James (1904), amongst others, although here I shall concentrate less upon the details of neutral monism itself and more upon the structure and plausibility of the response to CA that it facilitates. In particular, I will draw upon recent work by Maxwell (1978) and Stoljar (2001a; 2001b) who, despite describing their accounts as physicalist, could be equally described as holding a form of neutral monism, since the structure (although not their interpretations) of these accounts matches that given by neutral monism.
I begin by describing Kripke’s version of CA in Section II, along with some possible responses to it. In Section III, I present a neutral monist interpretation of Maxwell and Stoljar before highlighting some potential problems for this approach in Section IV. Finally, I conclude that the neutral monist response to CA is successful, but may still be problematic due to the way that it accounts for the metaphysics of perceptual experience.
II. The Conceivability Argument
The type of materialism that CA is opposed to claims that in some, but not all, possible worlds, each mental event can be identified with some corresponding physical event such that they are one and the same thing. This ‘type-B’ materialism (Chalmers 1996: 165) does not mandate that this is the case in all possible worlds — i.e. the entities in question are not necessarily identical — but merely that such identities hold in the actual world, along with a suitable selection of close possible worlds. Moreover, CA applies to both type-type and token-token identity theories, the first of which posits identities between mental and physical event types, e.g. ‘pain = C-fibre firings’, and the latter between particular mental and physical events, or tokens, e.g. ‘my headache at 2pm = these activations of nerve endings within my brain at that time’. For the purposes of this essay I will concentrate mainly upon the latter, but similar considerations apply to both. It is important to note that the relationship that materialism posits is one of identity, and not mere correlation. Consequently, certain pains just are C-fibre firings (to borrow Kripke’s example) and so must possess precisely the same properties — something that is a potential problem for any identity theorist (see Section IV: the grain objection).
Kripke’s version of CA proceeds from what he takes to be a basic intuition about the connection between mental and physical phenomena. Kripke claims that we can imagine a particular pain (e.g. my headache at 2pm) existing without the particular pattern of nerve activations that the materialist claims it is identical with. Had the relevant portions of my brain been replaced with identically functioning pieces of silicon, for example, then it seems that I could have had the same pain in the absence of these particular nerve activations. More generally, we can conceive of minds that exist independently of any body, in which case their ‘pains’ and other mental phenomena would be independent of all physical activity. Of course, the existence and status of such intuitions is highly contentious, and many philosophers (e.g. Stich 2004) question their significance. Even if such do intuitions exist, they may be inaccurate or misleading due to our lacking knowledge of the relevant mental and physical domains (e.g. the laws of physics). Alternatively, they may be the result of exposure — direct or otherwise — to the works of such philosophical luminaries as Descartes and Kripke, or traditional religious beliefs, rather than indicating an innate proclivity to countenance the possibility of immaterial spirits or ‘zombies’. For the sake of argument, however, I will accept that such intuitions are widespread and that we can and do have knowledge of modal facts on the basis of them since, in the absence of this link between conceivability and possibility, it is difficult to see what else could underpin genuine modal knowledge.2
Kripke’s argument also draws upon the modal semantics that he himself created (Kripke 1963). He defines a rigid designator as any linguistic term that identifies the same individual in every ‘possible world’ in which they exist. This category includes most proper names as well as some demonstratives, such as ‘that pain’. Conversely, a non-rigid designator picks out different individuals in different possible worlds, as is generally (although not always) the case with definite descriptions, such as ‘the author of Naming and Necessity’. This picks out different objects in other possible worlds, since it is not an essential property of Saul Kripke that he wrote this particular work.3 Due to the indiscernibility of identicals (i.e. Leibniz’s Law) and the way in which objects in possible worlds are ‘stipulated’ rather than ‘discovered’ (Kripke 1980: 44), all identity statements between rigid designators turn out to be a posteriori necessities. Whilst the detailed arguments for this thesis lie beyond the scope of this essay, the necessity of identity is central to both Kripke’s modal semantics and to CA, as well as being widely accepted by contemporary philosophers and logicians.
Putting all this together with the above intuition concerning the apparent contingency of mind–brain identity statements, and assuming that the relevant terms are in fact rigid designators, Kripke presents the materialist with the following dilemma. Either:
(1) The equivalence between mental and physical events is only contingently true, i.e. true in just some possible worlds, in which case such correlations are not identity statements, since this would contravene the necessity of identity, and so type-B materialism is false, or
(2) Such identities are necessarily true, i.e. true in all possible worlds, in which case the materialist must provide some explanation for their apparent contingency, as established by the above intuitions.
Kripke takes it that proponents of materialism must take the latter route if they are to vindicate their theory, although this is by no means the only option. One could, for example, accept the necessity of identity and question the validity of the relevant modal intuitions (although neither Kripke nor Chalmers finds this option plausible). Alternatively, one could reject the semantic framework that gives rise to the problem, claiming that necessary of identity is merely an artefact of the way in which Kripkean possible worlds are specified. In the absence of an alternative modal semantics, however, this option is unattractive as it calls much recent philosophy into doubt; e.g. Chalmers (2004) and Stalnaker’s (2001) two-dimensional semantics. A more promising strategy might be to deny that the relevant mental and/or physical terms are in fact rigid designators. On this view, ‘that pain’ or ‘those C-fibre firings’ might be definite descriptions that only contingently pick out the relevant events, rather than doing so necessarily. In this case, however, Kripke (1980: 331) argues that the relevant non-rigid terms may simply be replaced with suitable rigid designators, and the argument run again to show that these are not identical. A determined materialist might resist this move on the basis that it is question-begging, but this would seem to undermine the materialist’s own claim that pains, nerve activations etc. are natural kind terms, since we do normally think of terms such as ‘water’ and ‘H2O’ as picking out the same thing in all possible worlds. This response does, however, highlight a major assumption in Kripke’s argument, thus creating a point of departure for the kind of ‘non-materialist physicalism’ that is discussed in Section III.
In response to the above dilemma, Kripke considers the attempt to explain the apparent contingency of mind–brain identity statements by analogy with other comparable scientific identities, such as those between heat and molecular motion, or water and H2O. In these cases, Kripke argues that the apparent contingency arises from the presence of a suitable phenomenological counterpart, namely the sensation of heat or the observable properties of water, by which the left-hand term is normally picked out. He then argues that we are liable to mistake the contingency of the causal relationship between this counterpart and the right-hand term for the contingency of the original identity. For example, although it is conceivable that in some possible world the sensation of heat is caused by something other than molecular motion — such as green light, for example — this does not show that heat is not identical to molecular motion, but merely that the sensation by which we detect heat is contingently caused by it. The resulting ‘illusion of contingency’ leads us to falsely conclude that such identities are contingent, when in fact they are necessary, as Kripke’s modal semantics predicts. In the case of pain and other phenomenal states, however, no such phenomenal counterpart is available, since pain (what it’s like to see red, thinking, etc.) is already a phenomenal state. Consequently there is no possible world in which ‘pain’ is distinct from the sensation of pain, since pain simply is the sensation of pain, or else such a world would fail to explain the apparent contingency of ‘pain = C-fibres firing’, since there is no reason why we should identify the non-phenomenal counterpart with the sensation pain. In the absence of any other explanation for the apparent contingency of mind–brain statements, Kripke concludes that such identities must be false and that, by implication, materialism is false.
Some philosophers (e.g. Hill 1997) have responded to this argument by attempting to provide an alternative explanation for the apparent contingency of mind–brain identity statements. Others (e.g Jacquette 1987) have sought to redefine the relevant mental and physical terms in a way that makes the contingency (apparent or otherwise) explicable. Throughout the remainder of this essay, I shall concentrate on one such approach that arises from the theory of neutral monism (Stubenberg 2005), a metaphysical doctrine which posits that the fundamental constituents of the universe are neither physical nor mental, but rather ‘neutral’ elements from which both the mental and the physical are comprised, or whose properties they supervene upon. My account is largely based upon Maxwell (1978) and Stoljar (2001a), but whilst they class their theories as physicalist, I claim that they are characteristically neutral — although since the structure of both theories is substantially equivalent, the difference turns out to be largely terminological.
III. Neutral Monism
The neutral monist response to CA is based upon two key assumptions. The first is that physical science does not tell us everything there is to know about the nature of the universe (Stoljar 2001a: 313; Maxwell 1978: 395). This is not simply due to the incompleteness of current physical theories, but that they can only provide us with a certain type of information about the world, namely its ‘causal structure’ (ibid. 346). Nevertheless, physics does (according to Maxwell) posit the existence of some underlying substance or elements that instantiate the structural properties that we can investigate empirically, such as the observable properties of matter and energy.4 These purely formal properties may be contrasted with the intrinsic properties of the underlying ‘causal network’ (ibid.), which are distinct from, but nevertheless give rise to, all of the causal properties described by physics. It is a logical consequence of this view that it would be possible for a given causal (i.e. physical) structure to be instantiated by a number of different causal networks. This makes the causal properties of any given part of the network (e.g. a C-fibre firing) a contingent, rather than necessary, property of the underlying network, since it does not possess this property in all possible worlds. The second assumption is that both physical and mental events are comprised of, or supervene upon, a single kind of entity which I shall refer to as a neutral element. Whilst the structural properties of these elements give rise to everything we observe in the physical world, as described above, their intrinsic properties correspond to the nature of ‘inner’ mental phenomena, such as sensations, thoughts, consciousness, and so on.5 Neutral monism is thus able to account for the intimate connection between the mental and physical domains by explaining both types of phenomena as being comprised of a single common or ‘neutral’ element.
At this point, neutral monism parts company with Maxwell and Stoljar, who describe the underlying elements as being essentially physical, rather than neutral in character. However, since both acknowledge that physical science only studies the structural (i.e. causal), rather than intrinsic (i.e. mental), properties of these elements, this assumption seems somewhat ad hoc since there is no reason to think of the physical realm as extending beyond the domain of causal phenomena. Stoljar’s (2001a: 313) appeal to the physical properties of objects (‘o-physical’) as opposed to those of physical theory (‘t-physical’) is similarly problematic since such properties are either causal, and therefore relate to structural and not intrinsic properties, or else they are mental, in which case they cannot obviously be classed as physical. Such ‘non-materialist physicalism’ may be contrasted with Russell’s overtly neutral monist account in which the underlying entities are described as being neither mental nor physical in nature, which is the approach I have taken here.6 Terminological differences aside, however, the structure of all of these accounts is substantially equivalent since they each posit that the intrinsic properties of the underlying elements are unknown to physical science, and that these correspond — either individually or in appropriate configurations — to mental phenomena.7 Although positing the existence of an entirely novel type of ‘substance’ about which we know very little might seem to count against the theory, this is offset by the considerable philosophical advantages of the position in explaining the nature of the connection between mental and physical phenomena (although presumably we do know something about its intrinsic nature, since this constitutes the basic elements of experience).8 In support of the position, many thinkers such as Lewis (forthcoming) and Jackson (1988) have found it plausible that the intrinsic nature of the universe cannot be accounted for by the laws of physics alone, thus presenting a prima facie case for the adoption of a neutral monist position.
Expressed within the framework of neutral monism, then, Maxwell’s argument proceeds as follows. Being a particular phenomenal event — e.g. seeing a red patch or feeling a particular pain — is an intrinsic, and therefore essential, property of the relevant neutral element(s). Conversely, being a particular physical event is a structural property of the relevant neutral element(s), and is therefore contingent, since causal structures are not intrinsic to these elements, but multiply realisable through different causal networks which possess the same structural properties. Now the analogy with heat and molecular motion runs precisely in reverse to Kripke’s example, since being a particular brain event is a contingent physical property of the neutral elements which instantiate the relevant brain event in the actual world, whilst being a particular pain is an intrinsic, and therefore necessary, property of the very same neutral elements. This preserves the necessity of identity since such identities hold between two different ways of designating the same neutral elements in this, the actual world, with the physical term (e.g. ‘C-fibres firing’) picking them out by means of a contingent causal property, and the mental term (‘pain’) via a necessary property. The apparent contingency of such identities is explained by the fact that the physical property of being a brain event can come apart from the relevant neutral elements in other possible worlds that have the same causal structure — and therefore the same physical properties — but different intrinsic or phenomenal properties. Or, to paraphrase Maxwell (1978: 380), ‘a given brain event [qua neutral element] might not have been a brain event [qua causal role]’ (original emphasis). Thus there is a possible world in which the neutral element(s) that constitute each pain in the actual world play a different causal role, and so no longer constitute the same — or indeed any — brain event, although the identity between the pain and the actual causal-physical role of these elements remains a necessary one, since they are identical in all worlds where both exist.
Neutral monism therefore explains the apparent contingency of mind–brain identity statements in a manner that exactly parallels the approach that Kripke discounts. However, instead of explaining the ‘illusion of contingency’ in terms of mental events being picked out via one of their ‘accidental’ properties — something that both sides agree is impossible, or at least very implausible (ibid. 375; Kripke 1980: 330) — it is brain events that are picked out by an ‘accidental’ property; namely the property of being a brain event, which is contingent to both the mental and brain events in question (a possibility that Kripke finds ‘self-evidently absurd’ [ibid.]). The metaphysics of neutral monism explains how both of these terms can pick out precisely the same neutral element(s), making it a true mind–brain identity thesis. Physicalism, on the other hand, still turns out to be false, since the underlying entities are (pace Maxwell and Stoljar) not physical but neutral in character. This amounts to a rejection of Kripke’s assumption that physical descriptions are rigid designators since, according to neutral monism, they pick out different entities in different possible worlds by means of their contingent causal role, thereby escaping CA.9
IV. Objections
The plausibility of the above response rests largely upon the plausibility of neutral monism generally. Perhaps the most serious objection to the view, and to identity theses in general, is what Maxwell (1978: 392) calls the ‘grain objection’. If mental events are identical with brain events (or the corresponding neutral elements that comprise or subvene them) then, according to Leibniz’s Law, the properties of any brain event must be identical with the properties of the neutral elements, since they are one and the same thing. But whilst ‘the occurrence of a smooth continuous patch of red in our visual experience’ (ibid.) exhibits both unity and smoothness, it is difficult to imagine how the relevant neutral element(s) or brain event could also possess these properties, since they may be comprised of discrete, logically separable elements. Stoljar (2001a) presents a convincing response to this objection by drawing a distinction between the properties of the contents of an experience, or how that experience represents the world as being, which may be unified and smooth, and those of the experience itself, which may not.
Another potentially more pressing concern for the neutral monist is whether mental events are the sort of thing that can be comprised of intrinsic, non-relational properties at all. Whilst neutral monism does not necessarily entail the existence of so-called ‘free-floating’ qualia — one of the common historical objections to it (Stubenberg 2005) — it seems plausible that many types of experience, and therefore many mental events, are intrinsically relational in a manner that neutral monism fails to capture. Perceptual experience, or the kind of switching scenarios considered by Burge (1979; 1988), show that the contents of thought or experience cannot be captured internalistically, but incorporates elements of the surrounding environment or context. In order to accommodate this insight, the neutral monist would either have to posit that the underlying neutral elements are also externalistically individuated, which would make it difficult to identify them with a particular physical event via its causal role, or else invoke Stoljar’s distinction between representational content, which would be relational, and phenomenal experience, which would not. Both views are problematic and, whilst it lies outside the scope of this essay to explore these problems in greater detail, either may threaten to undermine — or at least complicate — the neutral monist account.
V. Conclusion
I have argued that the metaphysical doctrine of neutral monism provides a successful response to Kripke’s version of CA concerning the apparent contingency of mind–brain identity statements. Furthermore, the structure of this response closely mirrors Kripke’s own analogy with other scientific identity statements, such as ‘heat = molecular motion’ and ‘water = H2O’, but reverses the direction of the analogy such that ‘pain’ is analogous to ‘molecular motion’ and ‘C-fibre firing’ to ‘heat’, rather than (as Kripke assumes) the other way around. This explains both the apparent contingency of the resulting identity statements, and their necessity, as required by Kripke’s modal semantics. The resulting mind–brain identity thesis is, however, non-physicalist, since the bearers of the relevant properties are not physical in any ordinary sense of the term, although the structure of this account closely resembles Maxwell and Stoljar’s ‘non-materialist physicalism’. Its plausibility, however, depends upon that of neutral monist metaphysics as a whole, and whether this adequately reflects the relational nature of experience and representational content. In spite of these concerns, however, neutral monism offers a powerful and flexible framework for explaining the relation between mind and body — a problem that has been a recurrent theme in philosophy ever since Descartes’s Meditations.
——————
1 For brevity, I shall talk only of events, but the argument should be taken to apply equally to processes, states and other mental or physical entities.
2 Chalmers (2006) argues more explicitly for this link.
3 Although depending upon how philosophical works are individuated, it may be an essential property of Naming and Necessity that it was written by Saul Kripke.
4 The philosophical position known as structuralism denies that there are (or that we have knowledge of) any such underlying entities (Ladyman 2008).
5 This is not to say that every neutral element constitutes a phenomenal property, merely that some do — perhaps as a result of multiple elements combining in the appropriate manner (cf. Mach 1886). This avoids Chalmers’s charge of panpsychism, which is by no means a necessary consequence of neutral monism (Stubenberg 2005).
6 Russell’s adoption of ‘events’ as the fundamental ontological category is not essential to the position.
7 I will remain agnostic on the question of whether percepts are individual or groups of neutral elements, as per Russellian and Machean monism, respectively (see Stubenberg 2005: §5.1).
8 The suggestion that such knowledge is unattainable corresponds to Colin McGinn’s (1989) ‘mysterianism’.
9 Whether this argument allows for the existence of ‘zombies’ largely depends upon how the relationship between mental events and neutral elements is spelled out. However, such creates are at least a logical possibility on the neutral monist view.

Bibliography
Burge, Tyler 1979: ‘Individualism and the Mental’. In Studies in Metaphysics, P. French, T. Uehling and H. Wettstein (eds.), Minneapolis: University of Minnesota Press.
————— 1988: ‘Individualism and Self-Knowledge’, Proceedings of the Aristotelian Society, 85 (11), pp. 649–63.
Chalmers, David J. 1996: The Conscious Mind. Oxford: Oxford University Press.
————— 2002: Philosophy of Mind: Classical and Contemporary Readings. Oxford: Oxford University Press.
————— 2004: ‘The Foundations of Two-Dimensional Semantics’. In Two-Dimensionalism, M. Garcia-Carpintero and J. Macia (eds.). European Review of Philosophy. URL = <consc.net/papers/foundations.html>
————— 2006: ‘The Two-Dimensional Argument Against Materialism’. In The Character of Consciousness. Oxford: Oxford University Press.
Descartes, René 1641: Meditations on First Philosophy. Reprinted in J. Cottingham (ed. & trans.), Cambridge: Cambridge University Press, 1986.
Hill, Christopher S. 1997: ‘Imaginability, Conceivability, Possibility, and the Mind-Body Problem’. Philosophical Studies, 87 (1), pp. 61–85.
Holman, Emmett L. 2004: ‘Maxwell and Materialism’. Synthese, 66 (3), pp. 505–14.
Jackson, Frank 1998: From Metaphysics to Ethics. Oxford: Oxford University Press.
Jacquette, Dale 1987: ‘Kripke and the Mind-Body Problem’. Dialectica, 41 (4), pp. 293–300.
James, William 1904: ‘A World of Pure Experience’. Reprinted in W. James, Essays in Radical Empiricism. Lincoln: University of Nebraska Press, 1996, pp. 39–91.
Kripke, Saul 1963: ‘Semantical Considerations on Modal Logic’. In Reference and Modality, L. Linsky (ed.), pp. 63–72.
————— 1980: Naming and Necessity. Oxford: Blackwell.
Ladyman, James 2008: ‘Structural Realism’. The Stanford Encyclopedia of Philosophy, Spring 2008 Edition, E. N. Zalta (ed.). URL = <http://plato.stanford.edu/archives/spr2008/entries/structural-realism/>.
Lewis, D. forthcoming: ‘Ramseyan Humility’. In The Canberra Programme, D. Braddon-Mitchell and R. Nola (eds.). Oxford: Oxford University Press.
Mach, Ernst 1886: Die Analyse der Empfindungen und das Verhältnis des Physischen zum Psychischen. Fifth edition translated as The Analysis of Sensations and the Relation of Physical to the Psychical, New York: Dover, 1959.
Maxwell, Grover 1978: ‘Rigid Designators and Mind-Brain Identity’. In Perception and Cognition, pp. 365–404. Minneapolis: University of Minnesota Press. Reprinted in D. Chalmers (ed.) 2002, pp. 341–53.
McGinn, Colin 1989: ‘Can We Solve the Mind-Body Problem?’. Mind, 98, pp. 349–66.
Russell, Bertrand 1921: The Analysis of Mind. London: Allen and Unwin.
————— 1927: The Analysis of Matter. London: Kegan Paul.
Stich, Stephen P. 2004: ‘Semantics, Cross-Cultural Style’. Cognition, 92, pp. B1–12.
Stoljar, Daniel 2000: ‘Physicalism and the Necessary A Posteriori’. The Journal of Philosophy, 97 (1), pp. 33–54.
————— 2001a: ‘Two Conceptions of the Physical’. Philosophy and Phenomenological Research, 62, pp. 253–81. Reprinted in D. Chalmers (ed.), 2002, pp. 311–28.
————— 2001b: ‘The Conceivability Argument and Two Conceptions of the Physical’.  Noûs, 35 (s15), pp. 393–413.
Stalnaker, Robert 2001: ‘On Considering a Possible World as Actual’.  Aristotelian Society Supplementary Volume 75 (1), pp. 141–56.
Stubenberg, Leopold 2005: ‘Neutral Monism’. In The Stanford Encyclopedia of Philosophy, Spring 2005 Edition, E. N. Zalta (ed.). URL = <http://plato.stanford.edu/archives/spr2005/entries/neutral-monism/>.
Tye, Michael 1993: ‘Visual Qualia and Visual Content’. In The Contents of Experience, T. Crane (ed.), Cambridge University Press, pp. 158–76.

Is Red a Recognitional Concept?

Sat, 31 May 2008 18:55:47 +0100

Jerry Fodor and Christopher Peacocke on the theory of concepts
I. Introduction
One might think that if any concept could be considered recognitional — i.e. individuated at least in part by its recognitional properties — then the concept of red could. Not so, according to Jerry Fodor (1998a; 1998b), who rules out the existence of such epistemically individuated concepts in favour of his own informational semantics based account. This essay evaluates an alternative account of recognitional concepts by Christopher Peacocke (1992; 2000; 2005), one of the main proponents of the epistemic approach, along with some of Fodor’s objections to it. In Section II, I set out the main points of Peacocke’s theory, including the possession conditions that he specifies for <red>.1 In Section III, I evaluate Fodor’s main criticism of this account, namely that it fails to account for compositionality; i.e. our ability to grasp complex concepts like <red apple> by understanding their constituent concepts, i.e. <red> and <apple>, due to the difference between an object’s forming part of the extension of a complex concept and being a ‘good instance’ of it (Fodor 1998b: 10). Finally, in Section IV, I present Peacocke’s response to these criticisms before finally concluding that <red> is, after all, a recognitional concept, subject to some further philosophical and metaphysical concerns.
Before addressing the details of Peacocke’s theory, it is necessary to clarify the relevant notion of a concept. In contrast to the everyday usage of this term, the philosophical notion of a concept is that of a basic constituent or element of thought or language from which propositional (and possibly perceptual) content is comprised. This covers not only types, such as <apple>, <fish>, <water> etc., but connectives, qualities, relations, demonstratives, indexicals and proper names, to mention but a few. Concepts may be simple, e.g. <animal>, or complex, e.g. <pet fish>, in virtue of being linguistically and semantically analysable into their constituent concepts, e.g. <pet> and <fish>, plus a suitable mode of composition — in this case predication.2 In addition to accounting for such systematic compositionality, Fodor and Peacocke agree that any theory of concepts must respect Evans’s (1982: 100–5) Generality Constraint; i.e. that any subject who is capable of thinking that a is F, where F indicates some concept possessed by the subject, must also be capable of thinking that a is G, where G is any other concept of which the subject can conceive. For example, nobody who is able to grasp the sentence ‘the car is red’ would be unable to grasp ‘the car is green’ or ‘the apple is red’, provided that they possess the concepts <green> or <apple>, respectively (cf. Peacocke 1983: 1). Conversely, any loss of the ability to understand these concepts would result in a corresponding systematic loss of the ability to entertain propositions within which they feature (Evans 1982: 102).
Fodor (1998b: 1) defines a recognitional concept as any concept that (i) ‘is at least partly constituted by its possession conditions’ and (ii) whose possession conditions include ‘the ability to recognize at least some things that fall under the concept as things that fall under the concept’. In other words, a concept which is partially defined or individuated by the sensitivity of subjects to recognise instances of that concept. I will examine how this definition applies to Peacocke’s account of <red>, whose possession conditions involve the exercise of precisely such an ability, in Section III after setting out the main features of his account below.
II. Peacocke’s Theory
According to Peacocke, concepts are individuated by their possession conditions. That is, a concept’s identity conditions, i.e. what makes it that concept rather than another, are fixed by the conditions that determine whether a subject possesses the concept, rather than — as has traditionally been assumed — the other way round (Peacocke 2000: 328). In many cases, including the logical constants (&, ∨, →, etc.), these conditions simply list the set of inferences that the subject finds ‘primitively compelling’ (Peacocke 1992: 6), thus explaining the connection between a subject’s possessing a concept and the rational (i.e. truth-preserving) inferences that it permits them to make (Peacocke 2005: 170). Indeed, Peacocke rightly considers this to be an important task for any theory of concepts and rationality. Possession conditions may, however, involve conditions other than a concept’s inferential role (Peacocke 1992: 7). Peacocke specifies the possession conditions for <red>, for example, in terms of a thinker’s having the following dispositions:
1. ‘[T]o believe a content that consists of a singular perceptual-demonstrative mode of presentation m in predicational combination with C when the perceptual experience that makes m available presents its object in a red′ region of the subject’s visual field […] in conditions he takes to be normal, and [where] he takes his perceptual mechanisms to be working properly.’
2. To form the appropriate belief as to why m is so presented.
3. To believe of other objects that are capable of causing the kinds of experience necessary for (1), but which are not presented to the thinker, that they are C.
where C is satisfied by the abstract object <red> (ibid. 7–8).
The first of the above conditions does the bulk of the work by specifying <red> in terms of its combination with the Fregean ‘mode of presentation’ that arises when the subject sees an object in such a way that the experience exhibits the phenomenal property of redness (red′) within their visual field. At first glance, this mention of red′ in (1) might appear to be circular, since it defines the concept red in terms of a subject’s having a visual experience of red. However, since the analysans (<red>) is a concept and the analysandum (red′) a phenomenal property, Peacocke escapes the obvious circularity by defining the concept red in terms of the property red′, which he takes to be conceptually and experientially prior to <red>. Indeed, Peacocke argues that provided the concept being analysed does not appear within the scope of a ‘that’ clause specifying the content of a propositional attitude held by the subject, it is perfectly legitimate for its definition to mention that concept without fear of circularity (ibid. 9). He justifies this by pointing out that it would be unnecessary for a subject to already possess the concept of red in order for them to meet the above possession conditions, since red′ is taken to be a feature of normal human experience (ibid. 8). Whether there are any such phenomenal properties and whether they can be individuated without reference to the corresponding concepts, however, remain important questions for Peacocke’s account, as Fodor (1995: 32) points out, although this issue does not form the focus of the dispute between them. Peacocke’s theory is, in any case, flexible enough to allow the phenomenal property red′ to be replaced by a suitable reference to the metaphysical property, red, in keeping with Fodor’s (1998a: 73) own information theoretic approach.
Another notable feature of (1) is that it references a set of normal conditions, including the proper functioning of the subject’s visual faculties. This rules out someone’s possessing <red> as a result of, for example, direct electrical stimulation of their brain, or by sensations experienced in non-standard lighting conditions. Interestingly, Peacocke’s account does not actually require the conditions to be normal, but merely that the subject takes them to be so. This is presumably to avoid any circularity that would result from the concept’s possession conditions referring to the experience of a typical subject, which would itself require the same condition to be satisfied, and so on ad infinitum. However, the fact that the experience must be one of red′ is already sufficiently normative to avoid the generation of bizarre counterexamples that take advantage of this discrepancy. That what constitutes ‘normal conditions’ is an epistemic rather than metaphysical condition, and so may vary between one concept and another, will become important when considering how such conditions apply to cases of compositionality.
Conditions (2) and (3) qualify (1) by requiring that the subject be disposed to identify (where appropriate) the cause of his experience as being due to seeing something ‘red’, even for objects that are not, or may never be, visually present. Whilst Peacocke does not spell out the precise content of the relevant belief in (2), it presumably relates to the subject’s having seen a red-looking object, or perhaps the properties of the object that cause it to give rise to the phenomenal experience of red′. Here, as in (1), circularity would be avoided by mentioning the concept via the variable C, since any further use of <red> would fall foul of Peacocke’s restriction upon the concept appearing within the scope of a propositional attitude clause, as described above. In the case of subjects who are not capable of forming such beliefs since they lack the requisite cognitive faculties, e.g. young children or lower animals, Peacocke argues that it is sufficient that they would be disposed to form such beliefs were they to acquire the requisite faculties — as is the case with older children, for example. This is something of a moot point in the case of animals since there is no way to test the hypothesis, but it serves to emphasise the fact that Peacocke’s account of <red> is essentially dispositional, and does not require the subject to form the relevant belief in every case, or even to be capable of representing predicate structure (cf. Peacocke 1992: 8). Condition (3) caters for the fact that unobserved objects may also be red, since causing red′ sensations is a sufficient, but not necessary, condition for redness. If this condition were omitted then it would lead to the obviously implausible conclusion (pace Berkeley) there are no unobserved or unobservable red objects as a matter of a priori necessity (ibid. 20).
According to Peacocke, (1) through (3) constitute the complete set of possession conditions for <red>, and thereby individuate the concept. In order to determine whether something is red or not, a subject must exercise the epistemic capacities specified within these possession conditions, such as judging whether an object is presented in a red′ region of their visual field, thus enabling them to assign the appropriate semantic value (e.g. true or false) to propositions that involve <red>.3 The way in which the semantic value of a concept may be derived from its possession conditions is known as its ‘determination theory’ (ibid. 17). Although the details of this theory may vary according to the concept in question, it will in each case be both consistent with, and permit the subject to form true beliefs or inferences in accordance with, that concept’s possession conditions (ibid. 19).4 The correct exercise of the epistemic capacities specified by the determination theory for <red>, along with the state of the world at some particular time, will thus yield a true judgement as to, for example, whether or not some particular object is red. In this way, the semantic and empirical content of a concept is defined by a combination of its possession conditions plus the appropriate determination theory (ibid. 17).
The main strength of the above theory is undoubtedly its ability to explain the close connection between the possession a concept and the truth-preserving inferences that it enables a rational subject to make, since the two are, on Peacocke’s view, fundamentally interrelated through the concept’s possession conditions (see Peacocke 2000: 332). This in turn explains how the acquisition of a new concept enables a subject to carry out a wide range of valid inferences involving that concept, as the Generality Constraint requires, as well as the essential role that concepts play in our general ability to think and reason. However, Peacocke’s theory ties the possession conditions and determination theories of most (if not all) concepts to the epistemic or recognitional capacities of a rational subject. It is this aspect of the theory to which Fodor takes exception, causing him to rule out the existence of any such recognitional concepts — including <red>.
III. Fodor’s Critique
Before addressing the main thrust of Fodor’s critique, it will be useful to examine precisely how the above account relates to the definition of a recognitional concept given in Section I. Since the possession conditions for <red> involve only dispositions to form <red>-beliefs, whereas Fodor’s definition requires the ability to recognise things that fall under the concept as being red, it could be argued that <red> is not a recognitional concept, since (i) a subject may fulfil the relevant possession conditions without actually forming the belief that some object is red, and (ii) subjects need not possess the cognitive capacity required to form the relevant beliefs, as described above in the case of animals or young children. In general, however, the disposition to form <red>-beliefs under the circumstances specified by (1) does presuppose the ability to recognise red objects in virtue of the phenomenal property that they cause within the subject’s visual field. Since Fodor is opposed to any reference to epistemic capacities in explaining what ‘constitutes’ such concepts (Fodor 1995: 31), Peacocke’s definition clearly falls within the spirit, if not the letter, of the above definition.
Turning to the the main point, both Peacocke and Fodor take it that compositionality — i.e. the ability of rational subjects to grasp complex concepts simply in virtue of grasping their constituent concepts — is an essential (although not sufficient [Peacocke 1983: 55]) feature of concepthood, and something that any successful theory of concepts should explain. Fodor claims that Peacocke’s theory is unable to do this for the following reason. A subject who grasps <red> and <apple>, along with the predicational mode of composition, should thereby be capable of grasping the complex <red apple>. Conversely, a subject who grasps <red apple> must, ipso facto, be in possession of its constituent concepts, namely <red> and <apple>.5 Fodor argues that this cannot be the case for recognitional concepts since the relevant epistemic capacities do not compose in the required manner. What it takes to recognise a ‘good instance’ of <red apple>, he argues, may well differ from what it takes to recognise ‘good instances’ of <red> and <apple>, meaning that the ability to recognise instances of the former concept cannot be ‘inherited’ from an understanding of the latter (1998b: 10). It would therefore be possible for a subject who grasped both <red> and <apple> to nevertheless fail to satisfy the possession conditions for <red apple>, since they lack the relevant recognitional capacity, thereby contravening the principle of compositionality.6
As Peacocke (2000: 337) points out, Fodor’s own example of <pet fish>, whilst dialectically appealing, does not constitute a counterexample to Peacocke’s theory of recognitional concepts since neither <pet> nor <fish> is plausibly recognitional. Nevertheless, the objection generalises since every putatively recognitional concept may be combined with other non-recognitional concepts to form other complex concepts — e.g. <red hair> — to which a similar objection would apply. Although the resulting concept may be non-recognitional, its possession conditions arguably do not include the recognitional ability for red′, since red hair is not literally red, but rather reddish — or, to put it another way, red for hair (cf. ibid. 338). As such, the principle of compositionality appears to have been violated unless another explanation for the discrepancy can be found.
Peacocke responds to this objection by citing two kinds of cases in which the above situation can occur, but which nevertheless do not constitute a counterexample to the compositionality (ibid.). The first involves cases where the possession conditions for a non-standard mode of composition — in this case <for> — must also be taken into account. In order for a complex concept to be recognitional, each of its components must also be recognitional. Since <for> is not a recognitional concept, any complex concept, such as <red hair>, that involves it will also be non-recognitional, even if all of its other constituents (i.e. <red> and <hair>) are recognitional. In such cases, any additional epistemic content comes from applying the determination theory of the relevant mode of composition (i.e. <for>), and not from the fact that semantics ‘runs a barber shop’, as Fodor (1998b: 10) mischievously suggests. The possession conditions for <red hair> will, therefore, include the relevant elements of (1), but in a modified form due to their having been ‘multiplied out’ (Peacocke 2005: 173) — i.e. combined according to their semantic content, rather than simply being listed as a set of criteria — by the possession conditions for <for> and <hair>. The resulting non-recognitional concept can therefore be grasped by the possession conditions of its constituent concepts, as compositionality requires, even though its instances will not necessarily be recognisable this way.
The second way in which Peacocke suggests that complex concepts and recognitional abilities can come apart concerns what constitutes the normal conditions for recognising instances of each concept. Even in cases where the mode of composition is straightforwardly predicational, as with <red apple>, the conditions conducive to recognising, for example, red things (or apples) may plausibly differ from those conducive to the recognition of red apples. This may sound like a mere restatement of Fodor’s original objection in the <pet fish> case in that it involves a situation where the possession of a recognitional ability for each of the constituent concepts appears to be insufficient for recognising instances of the complex concept. However, this problem does not arise at the level of possession conditions, since the relevant conditions, e.g. (1), do not specify precisely what constitutes the normal conditions for recognising instances of that concept, but rather appeal to the independent knowledge of the thinker, as previously noted. Moreover, it may — in principle, at least — be determined in advance whether the conditions for recognising instances of the relevant constituent concepts are in fact compatible, and thus whether the resulting complex concept will be recognitional or non-recognitional. In the former case, where the normal conditions are compatible, the compositionality of both possession and recognition conditions will be respected, and so the relevant epistemic capacities will compose. In the latter, the resulting concept will be non-recognitional, and so does not present a counterexample to the principle of compositionality for recognitional concepts (Peacocke 2000: 339).7
Peacocke’s response to Fodor is therefore that, barring special cases that arise from the use of non-predicational modes of composition and conflicting recognition conditions, all simple recognitional concepts, when combined with other recognitional concepts or modes of composition to form complex recognitional concepts, obey the principle of compositionality (ibid. 340). Furthermore, even in these special cases, the possession conditions of such concepts (including the relevant mode of composition) combine in such a way that the compositionality of the relevant concepts’ possession conditions is respected. It is therefore essential to Peacocke’s theory that there are no cases in which a complex concept’s possession conditions do not contain each element of its constituent concepts’ possession conditions, since such a case would represent a counterexample to the theory. This is, perhaps difficult to evaluate in the absence of specifications for the possession conditions of many everyday concepts — something that Peacocke can justifiably be criticised for failing to provide, and which Fodor is doubtful could ever be achieved in practice (Fodor 1995: 31). Whilst this may count against Peacocke’s theory of concepts in general, in the absence of any convincing counterexample, it seems at least plausible that this should be the case, particularly with regard to recognitional concepts, such as <red>, for the reasons given above.
IV. Conclusion
I have argued that <red> is a recognitional concept since, on Peacocke’s account at least, its possession conditions plausibly involve the exercise of epistemic capacities, such as the ability to recognise red′ regions of one’s visual field. One consequence of this account is that it requires the existence of such phenomenal properties or sensations as red′, along with those ‘primitively compelling’ inferences that define the logical constants and other non-recognitional concepts. This may seem metaphysically extravagant since the properties and inferences in question mirror the very concepts that such a theory might itself be expected to explain. This objection is not, however, decisive since even rival non-epistemic theories, such as Fodor’s own account, must posit the existence of suitable metaphysical properties to which each of our concepts refer, and so have comparable (although different) ontological commitments.
Fodor’s criticism of the ability of Peacocke’s theory to account for compositionality is unsuccessful since the possession conditions that individuate recognitional concepts such as <red> do in fact compose, provided that the relevant mode of composition is taken into consideration (although whether the same applies to other non-recognitional concepts remains undecided). The theory also explains our ability to apply new concepts in conjunction with, for example, demonstrative concepts in accordance with Evans’s Generality Constraint, which is itself derivable from Peacocke’s theory that concepts are individuated by their possession conditions (Peacocke 1992: 48). Indeed, such demonstrative concepts arguably form an important class of recognitional concepts in their own right. Perhaps a more damaging and fundamental objection to this theory arises from Quine and Fodor’s rejection of the analytic–synthetic distinction (Fodor 1995: 33–4). However, since this issue lies beyond the limited scope of this essay, I conclude — for present purposes at least — in favour of the intuitively plausible view that <red> is, after all, a recognitional concept.
——————
1 Concept names are enclosed within angle brackets such that <red> denotes ‘the concept red’. The terms red′ and red refer to the corresponding phenomenal and metaphysical properties, respectively (adapted from Peacocke 1992: 7).
2 This differs from the Fregean notion of a concept as the referent of an ‘unsaturated’ predicate; e.g. ‘______ is red’ (Frege 1892).
3 Indeed, for Peacocke, a subject’s possessing a concept is just a matter of their knowing what it is for something to be its semantic value — an equivalence he refers to as ‘the Identification’ (ibid. 23).
4 Elsewhere Peacocke (2005: 171) talks of a concept’s semantic dimension (i.e. its truth values) being ‘recoverable’ from its cognitive dimension (i.e. its possession conditions), and vice versa.
5 Note that a subject could possess a simple concept with the same extension as <red apple> — call it <rapple> — without satisfying the possession conditions for <red> and <apple>. However, since <rapple> is arguably a different concept from <red apple>, which, by stipulation, has complex semantic structure, such cases do not constitute counterexamples to the principle of compositionality.
6 Fodor’s (1998a: 74) own view is immune to this objection since he takes it that relations between concepts occur at the metaphysical level of properties in the world as opposed to the epistemic level of possession conditions, as on Peacocke’s account.
7 One might object that even in these cases, compositionality must be preserved, but provided that the possession conditions make is possible to grasp the complex concept in virtue of its constituent concepts, the subject need not be able to recognise instances of a non-recognitional concept.

Bibliography
Fodor, Jerry A. 1995: ‘Review of Christopher Peacocke’s A Study of Concepts’. In In Critical Condition. Cambridge, Massachusetts: MIT Press (1998), pp. 27–34.
————— 1998a: Concepts: Where Cognitive Science Went Wrong. Oxford: Oxford University Press.
————— 1998b: ‘There Are No Recognitional Concepts; Not Even RED’. Philosophical Issues 9: 1–14.
Frege, Gottlob 1892: ‘On Concept and Object’. In The Frege Reader, M. Beaney (ed.), Malden, MA: Blackwell Publishing, pp. 181–93.
Peacocke, Christopher 1983: Sense and Content: Experience, Thought and their Relations. Oxford: Clarendon Press.
————— 1992: A Study of Concepts. Cambridge, Massachusetts: MIT Press.
————— 2000: ‘Fodor on Concepts: Philosophical Aspects’. Mind and Language, 15 (2), pp. 327–40.
————— 2005: ‘Rationale and Maxims in the Study of Concepts’. Noûs, 39, pp. 167–78.
Recanati, François 2002: ‘The Fodorian Fallacy’. Analysis, 62 (276), pp. 285–89.

Direct Realism and Visible Figure in Reid

Wed, 16 Apr 2008 15:04:32 +0100

 This essay has been temporarily removed site as it is currently under revision for submission to a professional journal. An updated version will be made available as and when this becomes possible, subject to relevant copyright restrictions. Sorry for any inconvenience!

Realism and Conceivability

Wed, 16 Apr 2008 14:57:53 +0100

Nagel’s argument against Davidson in Chapter VI of The View From Nowhere
I. Introduction
Whether we can adequately conceive of every aspect of reality is a deep and perplexing issue which philosophers have traditionally divided into two separate questions. First, the metaphysical question as to the ultimate nature of reality. Second, the semantic and broadly epistemological question of how thought and language are related to the world. The link between these questions is itself a difficult and complex issue that depends upon the particular metaphysical and semantic theories which are adopted. A metaphysical idealist will, for example, consider reality to be essentially dependent upon thought and language. Thus, all existent entities, such as facts, would be necessarily conceivable, since they have no existence independent of thought, making the metaphysical question subordinate to the semantic one. A common-sense realist such as Devitt (1984), on the other hand, might take these two questions to be more or less independent since, for him, the order of dependence is precisely the reverse. On this account, the notion of an inconceivable fact seems perfectly coherent since reality, being independent of thought and language, may outstrip both.
In this essay I will examine an argument by Thomas Nagel from Chapter VI of The View From Nowhere (Nagel 1986; hereinafter vfn) against metaphysical idealism. In particular, Nagel takes issue with a paper by Donald Davidson (1984; hereinafter vics) in which Davidson rejects the notion of a ‘conceptual scheme’ and endorses the intertranslatability of truths between all languages — a thesis that Nagel claims is incompatible with metaphysical realism. I aim to show that Nagel’s argument against Davidson is unwarranted and that the two positions need not come into conflict — or at least not in the way that Nagel appears to think they do. My argument centres upon Nagel’s presupposition that any semantic theory that rules out the existence of inconceivable facts necessarily entails that reality is mind-dependent; i.e. metaphysical idealism. I claim that Nagel’s justification for this presupposition is inadequate as it involves an equivocation concerning the notion of conceivability. In the absence of further independent argument, Nagel’s case for linking the above metaphysical and semantic questions is thereby undermined.
In order to clarify the debate it will first be necessary to define a few key terms. In this, I will follow Devitt in defining realism as the doctrine that ‘the world exists independently of the mental’ (Devitt op. cit: 14). This claim may be further subdivided into the claims that (a) reality is not in any way dependent upon thought and language (the ‘independence dimension’; ibid: 12–15), and (b) that the entities we ordinarily take it to contain — tables, people, electrons, etc.1 — really do exist (the ‘existence dimension’; ibid: 15–21).2 Idealism, on the other hand, rejects (a), since it takes reality to be dependent upon thought to the extent that it is constituted by it, and reinterprets (b) in terms of mind-dependent existence. The idealist thus denies that objects literally exist in the way that the realist supposes they do (Wright 1992: 2). This leaves room for a third position, or anti-realism, which opposes one or both of the realist’s claims, but that does not construe thought as being wholly constitutive of reality, as per the idealist (e.g. Dummett 1978). Despite Nagel’s terminology, I take it that his argument is directed not only at idealism, as defined above, but at any anti-realism that denies the existence of inconceivable facts. Finally, it is necessary to differentiate the notion of ‘reality’ or ‘the world’ from both the phenomenal ‘life-world’3 of human experience and the physical universe studied by the natural sciences. In keeping with the nature of Nagel’s enquiry, I assume that ‘reality’ is an all-encompassing term covering both physical and mental realms, whilst leaving it open as to whether it is to be identified with either (although neither Nagel nor Davidson finds either possibility, i.e. physicalism or immaterialism, plausible).
II. Nagel versus Davidson
Nagel’s argument is directed against the anti-realist view that ‘what there is must possibly be conceivable by us’ (vfn 93). Such a view rules out the existence of facts that are inconceivable by us — perhaps upon the basis that our conception of reality cannot extend beyond those categories of things with which we are already acquainted (vfn 94). Whatever the motivation, the anti-realist denies that there is a gap between what exists, i.e. reality, and what we can conceive of (where the terms ‘we’ and ‘conceive’ have yet to be defined). Nagel, on the other hand, claims that our conception of reality extends beyond not only what we can currently conceive of, but beyond what we could possibly conceive — or as he puts it, ‘beyond the reach of our minds’ (vfn 90). To deny this is, according to Nagel, to accept a form of ‘idealism’ in which the proper order of dependence between mind and world is reversed (vfn 92, 108–9). Although Nagel never explicitly spells out his grounds for making this link between realism and conceivability, it is nevertheless possible to discern them from the body of his argument, which is the approach that I will be taking below.
It is important to note that Nagel does not argue that there are any actual facts that we are unable to grasp, but merely that their existence cannot be ruled out in advance, as the anti-realist attempts to do. Similarly, whilst acknowledging that the dispute concerns the nature of thought, Nagel does not profess to have an alternative account to offer (vfn 95). Rather, he claims only that ‘realism makes as much sense as many other unverifiable statements [that] present fundamental philosophical mysteries to which there is no solution’ (ibid.). In other words, his argument represents a purely negative case against anti-realism, rather than a positive case for realism (although presumably, if the argument is successful, realism will be seen as all the more plausible as a result). Furthermore, since realism is the ‘natural picture’ (vfn 92), the burden of proof is supposed to lie with the anti-realist, who must demonstrate that the notion of an ‘inconceivable fact’ is incoherent if his position is to hold sway (vfn 93). Thus Nagel does not seek to refute anti-realism per se, but merely the claim that the notion of an ‘inconceivable fact’ is incoherent. Despite this, however, Nagel does think it probable that such facts exist, although he is (almost by definition) unable to prove it. He justifies this on the basis of humility (ibid.) and through an analogy that I shall examine in the following section. However, I will first attempt to establish precisely what Nagel takes the connection between inconceivable facts and idealism to be, and how this relates to Davidson’s argument.
The paper by Davidson with which Nagel takes issue is primarily concerned with refuting the notion that truth is relative to a ‘conceptual scheme’, as claimed by Kuhn (1962) and Feyerabend (1962), amongst others. In it, Davidson rejects the idea that we have any grasp of the idea of truth independently of language, and specifically of the notion of translation (vics 16). Even Tarski’s Convention T, which is widely accepted as a minimal condition for truth (although not necessarily as a definition of it), employs a form of translation from an object language to a metalanguage, as in the sentence:
(1) ‘Snow is white’ iff snow is white.
vics 16
(although, as Devitt [op. cit: 169] points out, even recognising the truth of this seemingly trivial fact already presupposes a great deal of theoretical and linguistic knowledge). On Davidson’s view, the notion of an ‘inconceivable fact’ is incoherent since whatever is inconceivable falls outside the bounds of language, and so the notions of truth and falsity, which are based upon translation, simply do not apply. Conversely, anything that can be stated in one language can, in principle at least, be translated into any other language — or at least anything that we would call ‘language’. Care must be taken to differentiate this thesis from the closely related, but obviously false, claims that (a) there is a word or term for every concept in every language, and (b) that any reasonably competent speaker of the target language could readily understand such a translation. Indeed, it may be incredibly difficult and complex to translate certain sentences between one language and another since the relevant concepts may be lacking and require further detailed explanation. Davidson’s point, however, is that in order to be capable of truth or falsity, a statement must be translatable into some recognisably linguistic form, no matter how simple or complex the resulting translation may turn out to be.
Davidson’s position clearly rules out the existence of inconceivable facts, making him a clear target for Nagel’s argument. It is less clear, however, that it can properly be termed a form of idealism. Firstly, Davidson’s thesis is a primarily semantic one concerning the nature of truth and the distinction between a ‘conceptual scheme’ and its ‘empirical content’ — a distinction he calls the ‘third dogma’ of empiricism (vics 11).4 In rejecting this distinction, Davidson rules out the idea that there is ‘something neutral and common that lies outside all schemes’ (vics 12), such as the raw content of human experience, or some ‘theory-neutral reality’ (vics 17). However, this does not mean to say that he gives up on the notion of empirical content or reality altogether, but merely that these should not be conceived as being relative to a conceptual scheme. Davidson concludes his 1984 paper by saying:
In giving up the dualism of scheme and world, we do not give up the world, but reestablish unmediated touch with the familiar objects whose antics make our sentences and opinions true or false.
vics 20
Thus, far from endorsing a form of idealism, Davidson affirms the mind-independence of ‘familiar objects’, which are sufficiently distinct from our thoughts to play a role in determining their truth or falsity.5 Indeed, his rejection of conceptual relativism arguably places him in a better position to affirm the existence of everyday objects than Nagel, who implicitly endorses this doctrine. Elsewhere, Davidson also asserts the objectivity of both thought, which is grounded in Tarskian truth conditions, and of knowledge, which ‘is of an objective world independent of thought or language’ (Davidson 2001: 138). The relationship between thought, truth and reality is, for Davidson, a complex one in which belief connects truth with meaning, but is not constitutive of reality in the strong sense indicated by the idealist (ibid: 189). It therefore seems incorrect — or at least misguided — to characterise him as a metaphysical idealist.
Based upon the above readings, the disagreement between Nagel and Davidson centres upon the coherence of the notion of an ‘inconceivable fact’, and the nature of language, truth and reality. Thus, Davidson does not so much claim that ‘certain attempts to form significant thoughts [e.g. that there are inconceivable facts] fail’ (vfn 94), but that truth is not a concept that can apply to anything of which we cannot possibly conceive (indeed, the very notion of a ‘thing’ is arguably inapplicable in such contexts). In order to further elucidate the basis for the disagreement, we must therefore examine the notion of ‘conceivability’.
III. Knowledge, Understanding and Conceivability
Much of Nagel’s argument against Davidson is based upon a single analogy in which he envisages a race of beings whose mental abilities are, by stipulation, no greater than those of an average human nine-year old (vfn 95–7). Since adult humans can conceive of things — Maxwell’s equations, the general theory of relativity and Gödel’s theorem are Nagel’s examples — that are beyond the comprehension of such beings, it seems reasonable to assume that the same thing might apply to ourselves with respect to yet more complex or sophisticated facts. Nagel argues that the congenital nine-year olds would therefore be incorrect to claim that reality is limited to what they themselves can conceive of, since this contradicts what we already know to be true, namely that the general theory of relativity et al. exist. By analogy, we (and specifically Davidson) would be wrong to conclude that only what falls within the limits of our own comprehension is real for similar reasons. In this way, Nagel infers that there may be things that are conceivable by beings whose minds are more sophisticated or subtle than our own, but which are nevertheless permanently beyond our limited comprehension — i.e. ‘inconceivable facts’.
Nagel’s analogy raises important questions concerning the nature of conceivability and precisely who it is that such facts are deemed to be conceivable by. Whilst nobody would deny that there are things that we, either individually or collectively, do not know, it is less clear that Nagel’s analogy addresses the issue of what we could not know. There are, in effect, several different ways in which we might be ignorant of a fact. The limited extent, or finitude, of our (actual) knowledge gives rise to what might be called epistemic ignorance — i.e. not knowing something that we could, were it to be explained to us or made perspicuous by means of a scientific experiment, for example, come to know. Knowledge of certain facts, however, may require the acquisition of new concepts, and it is a characteristic of natural language that concepts may be created, extended and altered in this way, thus helping to overcome what we might call our conceptual ignorance. Nagel’s analogy, however, alludes to an entirely different kind of ignorance, which I will call cognitive inadequacy. This involves the existence of facts and concepts that we (or the race of congenital nine-year olds) are wholly unable to grasp due to our level of intelligence, or the incompatibility of these facts (or the concepts that they employ) with human thought. However, it is arguable that such constraints are not in fact fixed quantities. They may, for example, be overcome from time to time by certain extraordinarily gifted individuals, such as a Maxwell or Einstein, or, in the longer term, through the development of new technologies, techniques, or even through further evolution of the species. We must also take care to distinguish between our practical ability to understand a fact, and its expressibility in a language that we can (or could ever) acquire. There are many facts — Maxwell’s equations, for example — that are incomprehensible to the majority of human adults due to their complexity and the level of background knowledge required in order to understand them fully. This does not, however, make them semantically inexpressible in the sense that Nagel’s analogy seeks to establish, and that is rejected by Davidson.
With these distinctions in mind, whilst there are undoubtedly things that we (or the race of congenital nine-year olds) are currently unable to understand — i.e. cases of epistemic ignorance — this does not mean that they cannot be conceptualised or expressed in a way that we do, or potentially could, understand. The notion of ‘conceivability’ can thus be seen to possess two separate but related components: (i) a cognitive or broadly epistemic component, i.e. understanding, and (ii) a semantic component, i.e. expressibility. If we read Nagel’s analogy as a claim about the congenital nine-year olds’ (or our own) ability to understand complex facts then it constitutes a relatively unobjectionable empirical claim concerning the first component: understanding. In order to carry this conclusion over to the second component, however, there must also be some fixed limit to the ability to acquire new knowledge and concepts such that the relevant epistemic and/or conceptual ignorance could never be overcome, thus creating a case of cognitive inadequacy. Unfortunately for Nagel, it is far from clear that any such limit exists, even in the case of the congenital nine-year olds. This is especially true if we take future evolutionary and technological developments into account, since even they may be able to extend their cognitive abilities through education and technology, for example. Furthermore, it seems that even complex facts such as Maxwell’s equations could be expressed in (i.e. translated into) the nine-year olds’ language, regardless of their current ability to understand them.6 Indeed, this is precisely the point of Davidson’s argument, which relates to the semantic component of conceivability. Thus, Nagel’s analogy fails to demonstrate that there may be inconceivable — in the sense of unconceptualisable — facts. Rather, it only warrants the much weaker conclusion that we can never be certain whether our ignorance of certain facts is due to a lack of knowledge (i.e. epistemic) or of practical ability (i.e. cognitive), since the two are indistinguishable from the perspective of the thinkers in question. Given that the one does not entail the other, the crucial question concerning the existence of unconceptualisable facts remains unanswered.
At this point we might be tempted to extend Nagel’s analogy to include truths that may be inconceivable to any finite being. However, once again we run into the same problem since it is impossible to say whether such truths are truly unconceptualisable, or whether it is merely the case that no thinking being has yet managed to grasp them. This does not mean that such facts could not, given sufficient time and effort, be understood, but in the absence of any further evidence, the argument merely results in a stand-off.
In order for his analogy to work, Nagel also requires that the notions of truth and falsity transcend our conceptual abilities, but it is difficult to see how the concepts of ‘truth’ and ‘fact’ can possibly apply to such cases. Discounting the possibility of epistemic ignorance, the purported facts must exist in a way that is totally inaccessible to all rational enquiry. This prompts the question of whether they can really be said to be part of reality at all, since anything that can affect us, whether directly or indirectly, can in principle be detected, and therefore subjected to analysis and conceptualisation. The same goes for any entity that supervenes upon known or knowable facts. A genuinely unconceptualisable fact must therefore possess a mode of existence that is not just divorced from the everyday life-world, but one that is independent of the very reality that we inhabit, since it cannot affect us in any way. Its ‘truth’ must be independent of any existent or possible language and yet still (in some presumably unfathomable way) be objectively determined by reality. At this point the notion of an ‘inconceivable fact’ begins to seem less like a philosophical thesis and more like mystical doctrine, since it is no longer clear in what sense such entities can be said to exist at all.
In summary, then, Nagel attempts to conclude from the finitude of our actual knowledge that there are limits upon possible knowledge, as expressed though language, rendering his argument invalid. His analogy appears to gain its plausibility from interpreting ‘inconceivability’ in terms of practical understanding, whereas Davidson’s argument addresses the semantic issue of conceptualisability. As such, the analogy does not represent a counterexample to Davidson. Indeed, even if we were to grant Davidson his conclusion that there are no inconceivable (i.e. unconceptualisable) facts, then this would not entitle us to conclude that whatever we cannot currently conceive of does not exist. Rather, in the absence of any fixed limit upon our cognitive abilities this remains an open question (thus addressing Nagel’s concerns for humility and anti-parochialism). What Nagel requires, but that his argument fails to provide, is some independent grounds for the existence of such a fixed limit upon cognition. In the absence of this, his case against Davidson is at best incomplete, and at worst unsuccessful, since the two accounts address different aspects of the notion of conceivability.
IV. Intertranslatability and Anti-Realism
In the final section of this essay I wish to consider the prospects for a compatibilist account of Davidson’s principle of intertranslatability and metaphysical realism of the kind that Nagel endorses. As we have seen, the Davidsonian may hold both that truths are translatable into any language (including that of the congenital nine-year olds) and that speakers of that language may, as a contingent matter of fact, be unable to understand them without rendering such facts ‘inconceivable’.7 Nagel alleges that this in itself represents a form of metaphysical idealism, since only entities that may be ‘idealised’ (i.e. conceptualised) are taken to form a part of reality (vfn 91). This is correct in so far as it attributes an anti-realist element to Davidson’s thought, but mischaracterises the nature of his position. As I argued in Section II, Davidson rejects the idealist claim that reality is constituted by our ideas and beliefs, although he does claim that belief has a role in determining the truth and falsity of our statements. What Davidson rejects is not realism as such, but the empiricist doctrine that mental entities — and specifically the mind itself — ‘mediate’ the connection between the world and knowledge (Evnine 1991: 149). In the absence of this doctrine, it can no longer be said that reality is wholly independent of our beliefs, but neither is it entirely dependent upon them since what it takes for a sentence to be true also involves our beliefs about the truth conditions of that sentence (Davidson 2001: 189–90). Whilst a detailed exposition of Davidson’s views lies outside the scope of this essay, it certainly seems possible that such an account of truth and reality can be given without falling into the kind of extreme subjectivism to which Nagel so strongly objects, and to which Davidson himself is opposed (ibid: 185).
Once we put Nagel’s central analogy to one side, his case for the link between conceivability and realism appears to rest on the contention that it is a necessary condition for the mind-independence of reality that it exceeds our capacity to conceive of it (vfn 91). However, it is far from clear why this should be the case since in order for reality to be ‘independent of the mental’, to use Devitt’s phrase, it is arguably sufficient that there are things that we do not know — i.e. cases of epistemic ignorance — regardless of what it is that we can know.8 This still allows the world to extend ‘beyond the reach of our minds’ (vfn 90), but not in the strong cognitive sense that Nagel’s argument demands. Furthermore, on the Davidsonian view, it is not that our being able to conceive of something is what ‘makes it real’ (vfn 92). Rather, it is the role that the world plays in constraining our beliefs, not independently of but in combination with language, that constitutes ‘reality’. This principle is never called into question by Davidson’s thesis of intertranslatability,9 although his rejection of empiricism does makes the relationship between thought and reality rather more difficult to define. Thus, the claims that everything is (at least potentially) conceivable and that the truth and falsity of our beliefs are not entirely dependent upon our epistemic powers do not, contrary to Nagel’s argument, appear to be entirely incompatible. The difference between the two positions lies not in the nature of reality per se, but rather in the relation between thought, reality and truth since for Nagel, thought and language are relatively independent of reality, whereas for Davidson the relationship is more complex.
V. Conclusion
Nagel’s argument for the link between the semantic and metaphysical questions with which I began this essay rests upon two main points, both of which turn out to be false. The first relates to his analogy of the congenital nine-year olds which, as I have argued, moves from an unobjectionable claim about the finitude of human knowledge to a more controversial and unsupported conclusion regarding the limits of conceivability. The second concerns the nature of conceivability itself, which I have argued has two different aspects. The source of the dispute between Nagel and Davidson regarding ‘inconceivable facts’ can be traced back to an equivocation between these two aspects — an error which also contributes to the apparent plausibility of Nagel’s central analogy. On Davidson’s interpretation of conceivability, however, the thesis of intertranslatability is not incompatible with the kind of realism that Nagel advocates — albeit with quite different implications for the relation between language, truth and reality. On this basis, Nagel’s argument fails to establish that conceivability and realism are in fact linked in the way that he supposes, making his objection to Davidson flawed, if not entirely misguided. This does not, however, mean that Nagel is wrong to think that reality must exceed our (potential) cognitive abilities any more than it means that Davidson’s view is correct. Rather, it merely shows that Nagel’s argument does not establish the falsity of Davidson’s thesis, and that the question concerning the link between metaphysics and semantics remains undecided.
——————
1 We may of course be mistaken about the nature or existence of some of these entities, e.g. electrons, but by and large we can truthfully say that most such entities exist (ibid: 22).
2 The mere claim that something exists — e.g. the Kantian ‘noumena’ (Kant 1988: A235/B294) — may be characterised as a kind of ‘weak realism’, since it accepts the former claim but rejects the latter (Devitt, op. cit: 15).
3 Lebenswelt (Husserl 1969).
4 The first two being the analytic–synthetic distinction and reductionism (Quine 1951).
5 Although not to the extent that there is any particular ‘thing’ that can be said to do so (ibid: 70).
6 This is arguably true of any language in which the basic concepts of number and logical argument can be expressed. Nagel himself considers this possibility at vfn 97.
7 The mere fact that such truths can be formulated in a language that is translatable into ours is sufficient to make them conceivable by any beings with sufficient cognitive ability to grasp the relevant statements and concepts, regardless of whether we are able to do so.
8 Or what Crispin Wright calls ‘objectivity of judgement’ (Wright 1993: 6).
9 Indeed, he goes to some lengths to deny it (Davidson 1990: 305).

Bibliography
Devitt, Michael. 1984. Realism and Truth. Oxford: Basil Blackwell.
Davidson, Donald. 1984 (vics). ‘On the Very Idea of a Conceptual Scheme’. In Inquiries into Truth and Interpretation. Oxford: Clarendon Press, pp. 183–98.
————— 1990. ‘The Structure and Content of Truth’ (The Dewey Lectures 1989). Journal of Philosophy, 87, pp. 279–328.
————— 2001. Subjective, Intersubjective, Objective: Philosophical Essays, Volume 3. Oxford: Oxford University Press.
Dummett, Michael. 1978. Truth and Other Enigmas. London: Duckworth.
————— 2006. Thought and Reality. Oxford: Clarendon Press.
Evnine, Simon. 1991. Donald Davidson, Cambridge: Polity Press.
Feyerabend, Paul. 1962. ‘Explanation, Reduction, and Empiricism’. In Scientific Explanation, Space, and Time: Minnesota Studies in the Philosophy of Science, 3, pp. 82–.
Husserl, Edmund. 1969. Die Krisis der Europäischen Wissenschaften und die Transzendentale Phänomenologie. Hamburg: Verlag.
Kant, Immanuel. 1989. Critique of Pure Reason. P. Guyer and A. W. Wood (trans. and eds.). Cambridge: Cambridge University Press.
Kuhn, Thomas S. 1962. The Structure of Scientific Revolutions. Chicago: University of Chicago Press.
Nagel, Thomas. 1986 (vfn). ‘Thought and Reality’. In The View From Nowhere. Oxford: Oxford University Press, pp. 90–109.
Quine, W. V. O. 1951. ‘Two Dogmas of Empiricism’. The Philosophical Review, 60, pp. 20–43
Wright, Crispin. 1992. Truth and Objectivity. Cambridge, Massachusetts: Harvard University Press.
————— 1993. Realism, Meaning and Truth. Oxford: Blackwell.

Making Sense of Indexicals

Wed, 16 Apr 2008 14:47:14 +0100

Does our understanding of indexical expressions pose insuperable problems for Frege’s account of sense, reference and the relations between the two?

I. Introduction
In this essay I will consider whether our understanding of indexical terms such as ‘I’, ‘here’ and ‘now’ cause insuperable problems for Frege’s account of ‘sense’ (Sinn) and ‘reference’ (Bedeutung). I will begin by briefly outlining the key aspects of Frege’s account in order to show why indexicals are a problem for Frege, and what his own response to the problem was. This will be followed by a consideration of the constraints to which a genuinely Fregean response to the problem should adhere, and why these are not met by some of the alternative approaches. Finally, I will evaluate a proposal by Wolfgang Künne (1997) that purports to solve this problem in a manner closely matching Frege’s own approach, followed by some closing remarks by way of conclusion. For brevity, I shall restrict myself primarily to a consideration of the first-person pronoun, ‘I’, although much of what follows may also be applied, mutatis mutandis, to other indexicals such as ‘here’ and ‘now’.1
II. The Problem
In Frege’s later works, the notion of ‘conceptual content’ (Frege 1879: 49) gives way to those of Sinn and Bedeutung, or sense and reference, as they are commonly translated (Beaney 1997: 36–46). Under this account, the linguistic meaning of an expression is comprised of precisely two components: (i) an object that is being presented (its referent), and (ii) the mode or method by which the object is presented (its sense). The referent of an individual word or concept expression is typically some object or aspect of the world to which the term is said to ‘refer’. The referent of an entire sentence, or proposition, is its truth value. Frege thus concludes that every proposition refers to one of two objects, which he terms ‘the True’ and ‘the False’ (Frege 1906a: 297). The Fregean sense of a proposition is therefore a ‘mode of presentation’ (Frege 1892: 152) of its truth value, just as the sense of an individual word — a noun, for example — is the method by which the corresponding object is presented.2 Two different terms for the same object, e.g. ‘the person on the left’ and ‘the tall man’, will have the same referent (i.e. the object) but different senses (i.e. modes of presentation), as per a single location on a map that is picked out by two different routes (cf. ibid.). Moreover, Frege held that the sense of a sentence or proposition is determined by the senses of its constituent parts such that if these were replaced with other words possessing a different sense, the overall sense — and therefore the linguistic meaning — of the sentence would change, even if its truth value, i.e. its referent, did not (ibid: 155–7).3
Frege’s concern in developing the above linguistic apparatus was to render the meaning of words objective, as opposed to earlier ‘phenomenalist’ accounts of language which portrayed meaning as a primarily subjective phenomenon (Frege 1897). This has important consequences for Frege’s account of indexicals. Since the linguistic meaning of indexical propositions can vary — compare my utterance of ‘I am hungry’ with your utterance of the same sentence, for example — their sense cannot be determined by the form of words alone, but must be sensitive to some aspect of the context within which they are uttered, such as the identity of the speaker. Frege states that in order to express ‘the same thought’ today and tomorrow, where a ‘thought’ is the sense of some particular proposition, I would have to add a ‘time-specification’ (ibid.); e.g. the word ‘today’ or ‘yesterday’ or the date upon which the proposition was uttered (Frege 1919: 343–4). Consequently, ‘[o]nly a sentence [that is] complete in every respect, expresses a thought’ (ibid.). In his treatment of indexicals within the same essay, Frege states that ‘the knowledge of certain conditions accompanying the utterance […] is needed for us to grasp the thought correctly’ such that ‘[t]he same utterance containing the word “I” in the mouths of different men will express different thoughts’ (op. cit: 332). Thus, only once all of the indexicality has been fully cashed out with respect to time, place and identity can it be said that the following propositions express the same objective thought:
(1) I am hungry [uttered by KW]
(2) You are hungry [uttered by someone else about KW]
cf. Newen 1997: 111
Frege’s account of indexicals raises the following question: what exactly is the sense that these sentences express? (1) cannot have the same sense as
(3) KW is hungry
where I, as a matter of fact, am KW, since I could conceivably that KW is hungry without believing that I am hungry — if I was suffering from amnesia, for example, and had forgotten who I am (cf. Lingens in Stalnaker 1981). As Perry (1979) points out, in order to explain my subsequent behaviour of going out to buy some food, for example, in addition to believing (3) I must also believe that
(4) I am KW
However, it seems implausible that I do have to believe (4) in order to think that I am hungry since this (1) something I come to believe solely on the basis of my own feelings of hunger, and not through identifying with some particular individual ‘from the outside’, as it were (Newen op. cit.). Furthermore, (3) and (4) fail to offer a reductive analysis of the indexicality of (1), since (4) reintroduces the indexical ‘I’ into the explanans, and is therefore circular.
In order to avoid the above problems, Frege held that ‘everyone is presented to himself in a special and primitive way, in which he is presented to no one else’ (Frege op. cit: 333). This makes the Fregean sense of each first-person indexical expression private and unique to each individual, and therefore inaccessible to anyone else (Perry op. cit: 14–15), thus presenting a quite different problem. On the Fregean view, communication takes place when two or more individuals ‘grasp’ the objective sense of a proposition. However, if the sense of an indexical proposition containing the word ‘I’, e.g. (1), is inaccessible, then how can anyone else possibly understand it? At best, the listener might grasp a different proposition that is sufficiently similar to the original thought to enable them to respond appropriately — (3) perhaps, or ‘the person who said “I am hungry” is hungry’. However, they can no longer be said to have grasped the same thought since, as has already been shown, (1) and (3) have different senses. Moreover, if we (pace Frege) take a similar position concerning temporal indexicals such as ‘now’, ‘today’ and ‘yesterday’, it would be impossible to express one and the same temporally-indexed thought at different times, since each time would have a unique mode of presentation that is inaccessible from within the context of another time (Beaney op. cit: 32). This would lead to a profusion of senses and, since the truth of most propositions is in some respect sensitive to time, an almost total lack of communicability, thus undermining the objectivity of language that Frege sought to establish.
III. Towards a Solution
Despite the above difficulties, it seems that we are able to understand one another’s use of indexicals and express the same thought across time. If Frege’s account of sense and reference is to hold, a more detailed account of the meaning or sense of indexical propositions must therefore be given. Such an account should explain both (a) the way in which the senses of (1), (2) and (3) are similar enough to convey what can be recognised by different individuals as ‘the same thought’, and (b) how (1) differs from (3) such that it explains KW’s behaviour in a way that (3) cannot. Furthermore, if the account is to be a Fregean one then it must some satisfy some additional constraints arising from Frege’s account of sense and reference as follows:
1. Such an account should explain precisely what the sense of indexical terms (‘I’, ‘now’, and so on) is, since according to Frege, only the senses of words and not objects can contribute to the sense of a proposition (Frege 1969: 203f in Künne op. cit: 54).4 This rules out treating indexicals as devices of ‘pure reference’ that introduce an object into the truth conditions of a sentence but do not have any sense in their own right (e.g. Wettstein 1979 and Marti 1995). Such accounts cannot properly be classed as Fregean since they reject the applicability of Frege’s notion of sense to indexical expressions, but rather are representative of the so-called New Theory of Reference in the style of Kripke and J. S. Mill’s accounts of proper names (ibid.).
2. The sense of any indexical expression should be objective such that it (or some appropriately derived aspect of it) can be grasped by others in order that genuine communication may take place. This is arguably true of John Perry’s (op. cit.) account in which different individuals are said to hold the same belief in different ways, by virtue of being in different ‘belief states’ (ibid.); e.g. in the indexical and non-indexical cases (1) and (3). However, the definition and individuation of such states is itself problematic and lacks independent justification in the philosophy of mind, making this an unattractive solution.
3. In order to meet the tests that Frege lays down for sameness of sense (Frege 1906b: 299; Beaney 2005: 35), any reductive analysis of indexical terms should be intersubstitutable for the original salva veritate5 such that it would be immediately apparent to any competent language user who understood both that they have the same truth value.6 In possible worlds based accounts, such as Stalnaker (1981), the content of an indexical proposition such as ‘It is now 5 o’clock’ may differ between speakers who already know which possible world they are in (i.e. they know what time it is) and those who don’t, since only the latter — Stalnaker’s ‘diagonal proposition’ (ibid: 138–9) — will range across possible worlds (Newen op. cit: 110). It is therefore unclear whether the truth of one proposition can be determined from the other as is required in order for them to have the same sense. However, even if this were to be the case, Stalnaker’s account also suffers from the difficulty that it requires first-person propositions like (1) to range across all possible referents, picking out the particular possible world in which KW is hungry. As described above, this is implausible since I do not have to believe that I am any particular person in order to believe (1), let alone to doubt which possible world I inhabit (ibid: 111).
As has already been established, indexical expressions are unusual in that their linguistic meaning varies systematically with the context of utterance. Their Fregean sense — if indeed they have any — must therefore incorporate some salient aspect of this context whilst retaining an underlying structure or form that can be understood and ‘tracked’ (through time, for example) by substituting the appropriate token-reflexive terms (‘today’, ‘yesterday’, etc.). Any attempt to simply replacing the indexical element with some proper name or definite description will fail on the basis that there are circumstance in which the name or description can misidentify the referent (ibid: 113; Künne, op. cit: 54); e.g. Perry’s (op. cit.) supermarket example, or the amnesiac Lingens in Stalnaker (op. cit.). Even the definite description ‘the thinker of this thought’ has a plausible if somewhat surprising counterexample in Newen’s (op. cit: 115) case of the schizophrenic whose thoughts are not always presented to her as her own.7 Nevertheless, the unique relation between the way in which a first-person thoughts is presented (its sense) and the thinker (its object) forms the basis for the Fregean account of the indexical ‘I’ by Wolfgang Künne (op. cit.) that I shall be considering throughout the remainder of this essay.
IV. Ego-Modes of Presentation
Central to Künne’s account is his notion of an ego-mode of presentation. This is a Fregean sense, symbolised by α, that is both ‘simple’, i.e. not analysable in terms of other, more primitive senses, and which meets the following condition for a given subject, y:
(EMP) ⃞(∀x)(∀y)(x is presented to y by α → x = y)
Künne, op. cit: 55
In other words, an ego-mode of presentation is a primitive mode of presentation whose object (its referent) is necessarily the same as its subject (i.e. the individual thinker or speaker). Thus, the sense of the word ‘I’ for subject y is defined as the ego-mode of presentation in which the referent of the indexical is necessarily identical to y herself, and so private and unique to that individual as per Frege’s specification. This forms the basis for Künne’s account of indexical belief, according to which the situation outlined in proposition (1) may be restated from a third-person perspective as:
(1*) (∃α)(KW = the object presented by α & α is an ego-mode of presentation & ‘α is hungry’ is the content of one of KW’s beliefs)
cf. ibid: 61
This may be differentiated from the more straightforward logical form of (3) on the basis that it involves a distinct mode of presentation, or sense, denoted by α. This sense is none other than the ego-mode of presentation that is contributed by the indexical ‘I’ and by which KW and KW alone can refer to himself, as defined above. Moreover, since the propositional content of the belief ascribed to KW (‘α is hungry’) neither names nor describes any particular individual, it is immune to counterexamples arising from misidentification, since the identity of the subject and object is built right into the definition of an ego-mode of presentation. Furthermore, by defining the indexical ‘I’ in terms of the ‘special and primitive’ mode of presentation by which an individual is presented to himself and no other, it is no longer necessary that the subject believe that he is in fact KW in order to explain the behaviour that results from his belief.8 This addresses Perry’s point concerning the explanatory inadequacy of (3) as an analysis of (1), as well as accounting for the difference between (1) and (3) in terms of the content of the beliefs in question.
Having establishing what is common to all ‘I’ beliefs, namely that they involve an ego-mode of presentation (ibid: 57), Künne’s account must also explain how thoughts like (1) can be communicated, since the relevant mode of presentation is ‘logically private’ to each individual (ibid.) and so cannot be grasped or expressed by a third party. According to Künne, indexical thoughts such as (1) are ‘propositionally opaque’ (ibid: 59) in that they do not fully reveal their propositional content and are thus, in the strict sense, of the word, inaccessible. However, the fact that the thought expressed by (1) employs an ego-mode of presentation can be grasped by a third party by virtue of its containing the first-person pronoun, ‘I’. Similarly, the predicate, ‘… is hungry’ may be grasped from the overall form of the sentence. All that remains, therefore, in order to yield an understanding of all three conjuncts of (1*) is the fact that KW is the object being presented, as is revealed by the context of utterance, since it is KW who is expressing his belief.9 Thus, it is true that the thought grasped by the listener, (1*), is different to the sense of the original utterance, (1), by virtue of its having a different logical form. However, since (1*) varies enables all of the important content of (1) — namely the object of presentation (KW), the type of mode of presentation (the ego-mode), the predicate (being hungry) and a belief ascription — this does not necessarily stand in the way of communication. On the contrary, in understanding the indexical proposition, (1), the listener is effectively able to ‘collapse’ the indexicality, revealing the sense of the corresponding third-person statement, (3). Moreover, the listener is able to identify which first-person proposition was being uttered, which is arguably all that is required for successful communication (Künne, op. cit: 63–4).
To return to the constraints that I laid down in section III, Künne’s account grants indexical expressions as having a Fregean sense involving, in the case of ‘I’, an ego-mode of presentation described by (EMP). It also makes the content of such propositions sufficiently objective that they can — in part, at least — be grasped by any third person whilst at the same time retaining Frege’s notion of a ‘special and primitive’ mode of presentation that is particular to each speaker. Whether the proposition that is communicated, e.g. (1*), has the same sense as the original, i.e. (1), however, is questionable since the first-person indexical, ‘I’, cannot be presented in precisely the same manner (i.e. as an ego-mode of presentation) to the listener as was expressed by the original speaker, even though they may be able grasp its logical form, or type. However, far from being a disadvantage, this is exactly what we would expect from an account of the first-person indexical. What a listener would understand in hearing (1) is not the same as what they themselves express by the words ‘I am hungry’, but rather that ‘KW is hungry’ (or perhaps ‘KW believes that he is hungry’), and so in a strictly formal sense they do understand a different proposition to the one originally uttered. Indeed, Frege himself admits as much in his discussion of the word ‘I’ (Frege 1919: 332), but this is now backed up with a detailed account of the sensitivity of first-person propositions to their context in terms of ego-modes of presentation and the logical structure of (1*). So, whilst (1*) and (3) are not intersubstitutable salva vertitate, neither should we expect them to be. It is sufficient that any competent speaker who understands both (1) and (1*), including the indexical nature of the ego-mode of presentation, is able to immediately infer the truth of one from the truth of the other, and vice-versa, and so cannot hold one to be true and the other false without contradiction. Künne’s analysis of the sense of the indexical ‘I’ therefore meets Frege’s crucial test for sameness of sense.
One possible problem for Künne’s account is the fact that, according to (EMP), the Fregean sense of a first-person proposition uttered by subject s is conditional on the existence of s (Künne, op. cit: 55). This might be thought to undermine the objectivity of such senses, since the resulting thoughts would literally not exist in the absence of the thinker. However, as Künne (ibid.) himself points out, Frege’s account only requires that the sense of a proposition can be grasped by a thinker independently of their thinking it; i.e. that it is not a subjective ‘idea’ in the mind of the agent. The fact that the sense of an indexical proposition cannot exist in the absence of the individual whose ego-modes of presentation it involves does not show that it is subjective, but rather emphasises the point that the sense of a proposition depends upon the existence the objects to which it refers, just as with any other non-indexical expression.
V. Conclusion
Künne’s account shows how indexical terms such as ‘I’, ‘now’, here’, etc. can have both an objective sense and a determinate reference in a given context whilst also contributing to the sense of the sentences that contain them in a systematic and communicable fashion.10 The plausibility of his account hangs on whether it is possible to make sense of Frege’s notion of a ‘special and primitive’ mode of presentation by which each of us is presented to ourselves and no-one else in a way that does not simply beg the question. Nevertheless, if we grant the existence of such ‘ego-modes’ of presentation, and are willing to let go of the assumption that communication requires both speaker and listener to grasp precisely the same proposition (ibid: 63), then our understanding of indexicals need not present any insurmountable problems for Frege’s account of sense and reference.
——————
1 I shall leave it open whether demonstratives — ‘this’, ‘that’, etc. — should be treated in a similar manner, and where the boundary between the two lies.
2 The term ‘mode of presentation’ is problematic in that not all linguistic terms refer, in which case there may be a sense but no referent, and so nothing being ‘presented’. For this reason, it may be better to regard Fregean senses as ‘methods of determination’, where determination may conceivably fail to obtain (ibid: 23–4).
3 Conversely, according to Frege’s ‘context principle’ (Beaney op. cit: 13), words only have meaning within the context of a proposition, bringing these two principles into tension.
4 Objects may, however, contribute to expressing a proposition as part of the surrounding context (Milne 2007).
5 Without loss of truth.
6 Alternatively, it would be contradictory for one to be true and the other false (Frege 1906c: 305).
7 In the case discussed, some of the subject’s thoughts are presented to her as being those of television personality Eamonn Andrews (ibid.).
8 The references to KW in (1*) are required for third-person ascription of the belief only.
9 Note that the object need not be the name of the speaker, but can be any term that picks out the individual in the relevant context, such as the second-person indexical ‘you’, thus explaining the close relation between (1) and (2) (ibid: 66).
10 The sense of other indexicals such as ‘now’ and ‘here’ may be explained in a similar manner via the so-called ‘nunc’- (Künne, op. cit: 64; Newen, op. cit: 118) and ‘hic’-modes of presentation (Newen, op. cit.), respectively.

Bibliography
Beaney, Michael (ed.) 1997: The Frege Reader. Malden, MA: Blackwell Publishing.
Frege, Gottlob 1879: ‘Begriffsschrift’. In The Frege Reader, M. Beaney (ed.), pp. 47–78.
————— 1892: ‘On Sinn and Bedeutung’. In The Frege Reader, M. Beaney (ed.), pp. 151–71.
————— 1897: ‘Logic’. In The Frege Reader, M. Beaney (ed.), pp. 227–50.
————— 1906a: ‘Introduction to Logic’. In The Frege Reader, M. Beaney (ed.), pp. 293–8.
————— 1906b: ‘A Brief Survey of My Logical Doctrines’. In The Frege Reader, M. Beaney (ed.), pp. 299–300.
————— 1906c: ‘Letters to Husserl, 1906’. In The Frege Reader, M. Beaney (ed.), pp. 301–7.
————— 1919: ‘Thought’ (Der Gedanke). In The Frege Reader, M. Beaney (ed.), pp. 325–45.
————— 1969: Nachgelassene Schriften. In Frege: Posthumous Writings, B. McGuinness (ed.). Oxford: Basil Blackwell.
Künne, Wolfgang 1997: ‘First Person Propositions: A Fregean Account’. In Direct Reference, Indexicality, and Propositional Attitudes. W. Künne, A. Newen, M. Anduschus (eds.), Stanford: CSLI Publications, pp. 49–67.
Marti, Genoveva 1995: ‘The Essence of Genuine Reference’. Journal of Philosophical Logic, 24, pp. 275–89.
Newen, Albert 1997: ‘The Logic of Indexical Thoughts and the Metaphysics of the “Self”’. In Direct Reference, Indexicality, and Propositional Attitudes. W. Künne, A. Newen, M. Anduschus (eds.), Stanford: CSLI Publications, pp. 105–31.
Perry, John 1979: ‘The Problem of the Essential Indexical’. Noûs, 13, pp. 3–21.
Stalnaker, Robert C. 1981: ‘Indexical Belief’’. Synthese, 49, pp. 129–51.
Wettstein, Howard K. 1979: ‘Indexical Reference and Propositional Content’. Philosophical Studies, 36, pp. 91–100.

The Grammar of Psychological Concepts

Tue, 29 May 2007 14:10:29 +0100

What do Wittgenstein’s grammatical investigations reveal about the nature of psychological phenomena?
I. Introduction
In the Philosophical Investigations (2001) and elsewhere (Wittgenstein 1967, 1980, 1982),1 Ludwig Wittgenstein undertakes a detailed study of what he calls the ‘grammar’ of psychological concepts. This covers a wide range of phenomena, including ‘understanding’, ‘knowledge’, ‘sensations’, ‘emotions’, and so on — many, if not all, of which are often taken to refer to mental states or processes. In this essay I will show how, through his method of grammatical investigation, Wittgenstein challenges this preconception, which is present in much contemporary philosophy, psychology and science of the mind. In particular, I will concentrate upon a number of grammatical features of these concepts, namely: (i) the differences and ‘family resemblances’ between various psychological concepts, (ii) the asymmetry between their first and third person use, and (iii) the emptiness of the notion of a ‘private mental object’, which Wittgenstein claims arises from a mistaken application of the ‘picture’ of the inner and the outer. I will also consider various objections to Wittgenstein’s position, including materialism and behaviourism, before drawing some conclusions concerning the implications of his views for both philosophical and scientific investigation of the mind.
II. The Notion of a Grammatical Investigation
Throughout his later period, Wittgenstein emphasis the nature of his task, and that of philosophy in general, as being a ‘grammatical investigation’ (McGinn 1997: 12). The aim of such an investigation is not, as with conventional philosophical and scientific analyses, to ‘penetrate phenomena’ (pi §90), or reduce them to a set of necessary and sufficient conditions, but to ‘[clear] away misunderstandings’ ‘concerning the use of words’ (ibid). Wittgenstein’s goal is to identify and eliminate the root causes of philosophical confusion, which lies in our tendency to be ‘bewitched by language’ (pi §109) and ‘held captive’ by metaphorical ‘pictures’2 (pi §115) of how concepts operate, but that do not fit their actual use. Central to this technique is the abandonment of explanation in favour of description (pi §109), and a rejection of philosophical theorising and abstraction in general in favour of careful examination of particular cases and examples, and how they fit into ‘the stream of life’ (lw1 118). This approach, influenced as much by Goethe’s work on the morphology of plants (Goethe 1790) as by any philosophical tradition (Monk 1991: 303), is summed up in pi §66 with the instruction: ‘don’t think, but look!’.
On first sight, one might question the appropriateness of such a method for investigating psychological concepts, such as feelings, sensations, etc. After all, these are generally taken to be empirical phenomena, as studied by the a posteriori of the sciences of the mind, and not mere grammatical constructs (Goldfarb 1989). However, to simply assume this from the outset would be to commit exactly the sort of philosophical error that Wittgenstein’s method is intended to guard against. To take the structural similarity between the two sentences ‘I have a pain in my arm’ and ‘I have a handkerchief in my pocket’ as prima facie evidence of a deeper parallel between the two cases, i.e. that they describe physical possession of some kind of object is, according to Wittgenstein, to mistake the nature of the very phenomena that we seek to describe (‘[t]he decisive move in the conjuring trick’ which is ‘the very one that we thought quite innocent’ [pi §308]). Of course, it is conceivable that the notion of mental states and processes may have some basis in scientific fact. Even so, it is questionable whether these are the entities to which our psychological concepts refer, and to simply assume that this is the case is question-begging and potentially unsound. All that can be assumed from the outset is that we are dealing with a variety of distinct but related concepts, and not that these must as a matter of metaphysical necessity share any particular underlying nature or structure (as states or processes within the brain, for example). The main task of a grammatical investigation of psychological phenomena is therefore to understand the similarities and differences between these and other concepts (pi §130), and to remind us how they are used in practice (pi §117) in order to gain a clear, or ‘perspicuous’ (pi §122), view of language.
Wittgenstein’s non-standard use of the word ‘grammar’ also requires some explanation, since it clearly does not refer to the system of verbs, nouns, and so on that normally go by this name. In pi §664, Wittgenstein states that ‘[i]n the use of words one might distinguish “surface grammar” from “[deep] grammar”’.3 Adopting this terminology, we might say that ‘surface grammar’ is constituted by those syntactic rules and relationships that are internal to language, such as the familiar verb-noun or subject-verb-object structure of English sentences. On this notion of grammar, the sentence ‘[c]olourless green ideas sleep furiously’ (Chomsky 1957: 15) would be considered perfectly grammatical. Wittgenstein’s notion of ‘deep grammar’, however, covers what traditional philosophy of language would call semantics and pragmatics, or the way that our use of language is embedded within our various ‘forms of life’ (pi §241). As such, deep grammar reaches ‘outside’ of language, connecting words with their uses — i.e. their meanings — as well as to non-linguistic practices such as physical gesture, tone of voice and facial expressions (pi §21, §433, §537). A consideration of the deep grammar of the above sentence reveals it to be nonsensical since it has no use (as opposed to it being false due to there being no ‘colourful green ideas’ that are ‘sleeping furiously’).4 By observing the subtle differences between how the different ‘regions’ of our language operate, a grammatical investigation can clarify the various types and uses of concepts, replacing the Fregean ‘model of “object and designation”’ (pi §293) with a much richer and more flexible conception of the role of language within human affairs.
III. The Psychological Language-Game
Having outlined Wittgenstein’s methodology, I will now examine how it is applied with regard to psychological phenomena. In Zettel §472, Wittgenstein sketches out a ‘[p]lan for the treatment of psychological concepts’. This identifies many of the main characteristics of concepts that we class as psychological, including:
‘Psychological verbs characterized by the fact that the third person of the present is to be verified by observation, the first person not’
‘The first person of the present akin to an expression’
‘All have genuine duration’
‘Possibility of … simultaneous occurrence’
‘All have degrees and qualitative mixtures’
‘Place of feeling in the body…’
(ibid.)
Before considering these points in greater detail below, it is worth noting that Wittgenstein is not attempting to lay out a schema as to how psychological concepts are to be defined, but rather offering a ‘synoptic view’ (z §464) of the many and varied ‘family resemblances’ (pi §66) that hold between them. These are, in Wittgenstein’s sense of the word, the criteria by which we identify concepts as psychological, rather than an exhaustive ‘taxonomy’, as might be provided under a behaviourist or materialist account (Hilmy 1995: 236).
In the Philosophical Investigations, Wittgenstein cites the concept of a ‘game’ as being paradigmatic of terms that are understood by all competent language users, and yet are seemingly impossible to define in terms of necessary and sufficient conditions (pi §66).5 Instead, the meaning of ‘game’ is determined by a complex series of similarities, interconnections and ‘family resemblances’ that hold between activities that we call games, along with the use of paradigm cases or samples — team sports, games of skill, games of chance, etc. — around which these relations are clustered (pi §16, §69). Similarly, psychological concepts do not form a unified and consistent genus, but are united by a complex and overlapping series of similarity relations that hold between them. The concept of ‘knowledge’, for example, exhibits a different grammar to that of ‘pain’, which is a sensation concept and so has duration, intensity and physical location (z §485). Pain is itself differentiated from other sensation concepts, such as feeling warm or cold, which have no particular ‘characteristic expression’ (z §483), and from joy, which is difficult (although not necessarily impossible) to locate at a particular point in the body (z §486). For Wittgenstein, it is these grammatical distinctions that unite or differentiate psychological concepts, both from each other and from other concepts, and not necessarily any commonality in their underlying physiology (although neither is such commonality ruled out). This is true even of sensory modalities, i.e. touch, taste, hearing, sight and smell, whose similarity is justified in terms of their ‘analogies and connections’ (z §474), and not just the ‘logical criterion’ that ‘they give us knowledge of the external world’ (z §477). By highlighting these distinctions, Wittgenstein forces us to attend to the details of the role that they play within the psychological ‘language-game’, weakening the grip of the idea that there must be something common that ‘lies behind’ all of these concepts in terms of them fitting into the familiar model of states and processes (Donagan 1968: 349). For Wittgenstein, what ‘lies behind’ is not a mental state or process, but the practice of using a concept in a particular way.
This concept of a ‘language-game’ (Sprachspiel, pi §7) is referred to throughout Wittgenstein’s later philosophy and combines the family-resemblance structure of a game with Wittgenstein’s distinctive notion of grammar as an integral part of the ‘forms of life’ from which language takes its meaning. Thus a single sentence, such as ‘I am in pain’, may be used in many different ways: as an expression of pain substituted for an inarticulate cry (pi §244); to report a symptom — to a doctor, for example (Donagan op. cit. 326–7); as a form of pretence, as when lying to someone in order to elicit sympathy; and so on. Each of these constitutes a different, but interrelated, language-game within which the concept of ‘pain’ plays a part. As such, not just the meaning of the word but its context must be taken into account in order to determine the role that it plays within the language-game — and in many cases its meaning will depend upon its context.6 Some commentators (e.g. Mounce 2002: 77) have accused Wittgenstein of over- or misapplying the notion of a game, since not all concepts conform to this model. However, it is clear that Wittgenstein’s game analogy is not intended to be taken as a comprehensive model of how concepts operate, since natural language does not on his account have the ‘formal unity’ of a logical calculus (pi §107, §81).7 Rather, Wittgenstein is offering an alternate picture of how certain concept words function, and thus an antidote to the ‘crystalline purity’ (pi §108) of logic, through which many philosophers (notably Russell) have approached the philosophy of language.
IV. Mental States and Processes
In order to demonstrate the application of Wittgenstein’s method to the investigation of psychological phenomena, I will consider two examples: (i) understanding, which is often characterised as a kind of mental state or process, and (ii) pain, which is typically thought to refer to a specific bodily state, such as C-fibres firing (Kripke 1980: 98). Underlying both of these characterisations is the notion that thoughts, feelings, sensations, and so on, take place in a hidden ‘inner realm’ that is separate and distinct from the ‘external world’ of physical objects. This conceptual divide, which continues to exert considerable influence upon both the philosophy and science of mind, may be traced back at least to Descartes (1641), where it emerges in the form of the familiar mind–body problem. Wittgenstein seeks to ‘dissolve’ both this problem and the dualism to which it gives rise not by rehabilitating mental phenomena within the physical world, as per physicalism or materialism, but by questioning the aptness of the ‘picture’ of the ‘inner’ and the ‘outer’ phenomena in relation to psychological concepts (McGinn ibid. 27–9).
Wittgenstein first raises the question of understanding with reference to the primitive language-game of the builders in pi §2, asking ‘am I to say that [ostensive definition] effects an understanding of the word?’ (pi §6; my italics). The suggestion that understanding is not a mental process, but tied to the performance (or disposition to perform) an appropriate action is later taken up in the discussion of rule following in pi §§138–55. Here, Wittgenstein considers whether understanding could be something that occurs ‘in a flash’ (pi §138), such as when we suddenly grasp the meaning of a word, for example. Whilst the description accurately captures the phenomenology of the experience (at least in certain cases), Wittgenstein questions what precisely is it that comes before our minds in such a situation. It cannot be the use of the word, since this is extended in time, and not momentary (ibid.). And if the use of words is determined by their meaning, then how can what we grasp ‘in a flash’ ‘fit a use’ (pi §139), since the grammar of mental states is not something that can ‘fit or fail to fit’ in this way (ibid.)? The idea that a certain picture, application or interpretation of the word comes to mind is similarly problematic. Whilst understanding may often be accompanied by a picture or particular interpretation — a formula, for example — this cannot in itself constitute the understanding, since a picture can support many different applications (pi §141). Conversely, the picture or formula may be absent in another case where a person understands by different means, or where understanding is not accompanied by any picture or formula and yet they are still able to perform the appropriate action or use the word correctly (pi §151).
Here, Wittgenstein shifts away from considering the ‘hidden’ aspects of understanding to what is publicly displayed and accessible: the criteria of understanding (pi §143, 146). We say that someone has understood a word when they go on to use it in the usual way — i.e. in accordance with a practice. In cases where they, despite having felt that they understood, misuse the word (again, relative to our own practices and expectations) then we are ‘tempted to say that [they have] understood wrong’ (pi §143). Even when a person does go on to do what was expected, we may later find that they diverge from this norm, and so we would say that they had not understood after all (pi §145). Thus, it is always possible that what we took to be understanding might turn out otherwise, and vice versa, so attributions of understanding — like those of knowledge, whose grammar is similar (pi §150) — are always potentially defeasible.
The above considerations call into question the notion of understanding as a mental process or state, since processes and states can always be said to occur at a particular time and place and in a particular manner; e.g. the process of making tea, or the state of sleep. This is not obviously true of understanding, which cannot be said to start and stop along with the actions that it precipitates (pi §148). Also, as Wittgenstein points out, we have no ‘knowledge of the construction of the apparatus’ (pi §149) — i.e. the mind or soul — within which the process or state is supposed to take place, and yet this criteria for the use of state and process concepts does not appear to be present in (or relevant to) the case of ‘understanding’ (pi §153). What this shows is that the grammar of ‘understanding’ does not fit that of an inner state or process, but rather describes a set of circumstances to which we attach significance (pi §155). Whilst the criteria for attributing understanding may include the feeling of being able ‘to go on’ (pi §151), a flash of insight, or the presence of a formula or mental picture, there are possible circumstances in which each of these is lacking and yet we would still say that understanding was present. When we say ‘I thought I understood, but I was mistaken’, the decisive factor in determining whether understanding was in fact present is not to be found ‘in the head’ in the form of a state or process, but in the world in the form of the distinctive pattern of circumstances that constitute understanding.8 These are entirely public and not hidden from view in some inner realm, as we might otherwise suppose.
In Zettel §§608–11, Wittgenstein goes on to question the assumption that psychological concepts can necessarily be correlated with any kind of brain process. In z §608, he observes that in some cases, the relevant facts for the attribution of psychological phenomena may lie in the history of a subject, as opposed to their current physiological or neurological state. As such, it would be impossible to simply ‘read off’ (ibid.) their psychological state from their physical brain or body. A materialist might counter this with the suggestion that psychological concepts supervene upon a wider range of facts than just the subject’s current physiological state, thus taking into account historical and other contextually relevant facts. However, the range of facts that these concepts would need to supervene upon would be very wide indeed, extending to the linguistic practices of the speaker and their community, as well as to the ‘fact’ that the subject is capable of advanced cognitive feats — i.e. that it is a living being (cf. pi §250, §284; z §§520–1). Furthermore, as Goldfarb (1989: 641–2) suggests, even if we were to discover some neurological state that was correlated with, for example, the subject’s understanding of the word ‘manatee’, then it would be impossible for us to identify this with the relevant psychological phenomena due to the potentially infinite number of factors (contextual, historical, ruling out special cases such as lying, and so on) that would also need to be taken into account. Conversely, if the criteria for understanding were present without the neurological state, then we would still be inclined to say that the subject understood, thus demonstrating the inadequacy of the proposed identity relation. On this view, the relevant facts are just not amenable to ‘the models we have of states and processes in physical science’ (ibid. 642; cf. pi §308), and so no reductive account of psychological concepts — let alone a physiologically reductive one — can be given. This makes a materialist explanation of our everyday use of psychological concepts highly implausible due to the extreme complexity of such an account, even if were theoretically possible.
V. Privacy and Pretence
The grammar of ‘understanding’, which is a primarily cognitive and dispositional concept, may be contrasted with that of ‘pain’, which is primarily sensory and occurrent. The concept of pain covers a wide range of cases, including various kinds of bodily sensations and emotions, and is associated with certain characteristic forms of expression — crying out, flinching, etc. (z §483). Unlike understanding, pain is always located in a particular region of the body (z §482) and has both intensity and duration (z §472). Wittgenstein’s ‘[p]lan for the treatment of psychological concepts’ also notes the asymmetry between first and third person uses of sensation concepts. In the first person case, the use of such concepts is described as being ‘akin to an expression’, whilst in the third person, such concepts may be used to provide ‘information’ (ibid.). Wittgenstein’s notion of an ‘expression’ may be understood as a way of expressing pain — the verbal equivalent of a cry, for example (pi §244). On this view, the sentence ‘I am in pain’ may be a more articulate substitute for other more primitive forms of ‘pain-behaviour’, such as crying (ibid.), and is no more descriptive of a mental state than the word ‘ouch’. The same sentence may also be used to report symptoms, or tell a lie (pi 161e), as described above, each of which has its own distinctive context and criteria of application. Even the third person sentence ‘she is in pain’ need not be read as an ascription of a mental state since we might also use this as a description of pain-behaviour; e.g. when someone cries out unexpectedly, or to report a fact about another person, such as a doctor reporting a patient’s symptoms to a colleague. Together, these form an entire family of language-games involving the use of the word ‘pain’, the substantial nature of which has yet to be determined.
The asymmetry between first and third person use of psychological concepts arises due to the possibility of doubt, which may be present in the third person case, but not in the first person, since one cannot legitimately doubt whether one is in pain, regardless of how (or whether) one expresses it (pi §288). Furthermore, Wittgenstein rejects the idea that one has knowledge (pi §246) of one’s own pain, since in order for the concept of ‘knowledge’ to apply it must be possible that we could conceivably not know; i.e. there must be the possibility for discovery or doubt (pi §251). This does not apply to the first person use of the word ‘pain’ — ‘except perhaps as a joke’ (pi §246) — since it makes no sense to say ‘I am unsure if I am in pain’, or ‘I thought I was in pain, but I was mistaken’, in the way that I can be mistaken about the weather, for example (pi §288). Such statements are, in Wittgenstein’s sense of the word, grammatically incorrect (although such cases must be clearly distinguished from those in which one masks or ignores one’s pain). Conversely, I can know that another person is in pain precisely because it is possible for me to doubt it — although even this may be difficult to do ‘in a real case’ (cf. pi §303). Furthermore, in the third person case, it is possible that someone might intentionally set out to deceive others by enacting characteristic forms of pain-behaviour. This may range from a straightforward lie (‘That hurts!’) to an elaborate exhibition designed to fool the observer into thinking that the subject is in pain. However, on Wittgenstein’s account, ‘[l]ying is a language-game that must be learned like any other’ (pi §249). In particular, this cannot be taken as evidence for a private mental state or process in the case of genuine pain since the language-game of pretence is possible in most (if not all) human interactions. Indeed, without the possibility of concealing or feigning our thoughts, sensations and emotions, we would scarcely be tempted to think of them as inner states or processes at all (z 557).
The ‘surface grammar’ of sentences like ‘I have pain’, coupled with the possibility for error or pretence, tempts us to think of pain in terms of privileged knowledge of one’s ‘inner state’, as if ‘pain’ were the name of some kind of private mental object (pi §244, §257). On this account, someone who is feigning pain is thought of as exhibiting all of its outward signs (i.e. pain-behaviour) whilst lacking the requisite ‘inner state’ for genuine pain. Wittgenstein challenges this use of the object–designation model by drawing our attention to the distinctive grammar of the word ‘pain’ compared to that of non-mental states and processes. For example, the ‘criterion of identity’ (pi §253) for pain is such that I can have the same pain as you (indigestion, for example), and yet it would make no sense to say that you have my pain (ibid.). Moreover, we cannot learn pain through ostensive definition — the model of learning exemplified by Aquinas’s recollections in pi §1 — since we cannot ‘point’ to pain, but only to pain-behaviour. In order for a child to acquire the concept of pain, it must therefore master the language-game of which it forms a part, as opposed to learning what ‘pain’ refers to, since there is no such public referent. We should be equally careful of attributing any prior familiarly with a private mental object that is called ‘pain’, since in the absence of this concept and the grammatical distinctions that it entails, there is nothing to unite the various bodily processes and states under this banner.
The above emphasis upon outwardly visible behaviour and ‘criteria’ as opposed to inner states has led some philosophers to read Wittgenstein as denying the existence of such states altogether, or as defining ‘pain’ in terms of pain-behaviour, or dispositions towards it (cf. Donagan op. cit. 335). When Wittgenstein’s interlocutor raises the issue of behaviourism in pi §305 and §307 in the context of a discussion about remembering, Wittgenstein responds that ‘[w]hat we deny is that the picture of the inner process gives us the correct idea of the use of the word “to remember”’ (pi §305; my italics), and ‘[i]f I do speak of a fiction, then it is of a grammatical fiction’ (pi §307; original italics). The ‘grammatical fiction’ to which Wittgenstein refers is the notion that the word ‘pain’ need refer to an actual state or process in order to fulfil its role in the language-game. This is merely the application of a picture whose grammar has been shown not to fit our actual use of language. As Wittgenstein’s ‘beetle in a box’ analogy concludes, the object (i.e. the mental state) ‘drops out of consideration as irrelevant’ (pi §293) since it does not matter what each person has ‘in their box’ (i.e. their mind) in order for them to participate in the language-game. Indeed, the box may even be empty provided that the word continues to be used in the correct way — i.e. according to accepted practice. On this reading, Wittgenstein does not in any way deny the existence of mental processes, which he agrees are ‘not a nothing’ (pi §304), but rather is claiming that they are not required in order to play the language-game — i.e. they ‘are not a something’ either (ibid.). The only reason that we feel them to be necessary is because of our prior commitment to the ‘picture’ of inner states and processes to which we take our psychological concepts to refer. Once we set this picture aside, we find that all that is required for the practice of talking about ‘pain’ to function is the concept of pain (pi §301), and that even the language of ‘states and processes’ is inappropriate, since we have little or no conception of the underlying mechanisms to which the metaphor alludes (cf. pi §149).
VI. Conclusion
Wittgenstein’s grammatical investigation reveals a number of important features of psychological phenomena. Firstly, the relevant concepts exhibit a complex series of ‘family resemblances’, both within and between the various subcategories of knowledge, sensations, emotions, and so on, as opposed to comprising a single unified genus. Secondly, our use of these concepts does not entail the existence of any particular mental states and processes to which they correspond (although Wittgenstein does deny that such states and processes exist). Finally, the notion of a ‘private object’, to which we sometimes take our cognitive and sensation concepts to refer, does no work within the ‘language-game’ that comprises our everyday use of these concepts. Rather, it is ‘forced upon us’ by the application of certain ‘pictures’ of the way that language and psychological concepts are thought to operate. Chief among these are the ‘model of an “object and designation”’ (pi §293) and the picture of the ‘inner’ and the ‘outer’, which we mistakenly take to be supported by the asymmetry between first and third person uses of psychological concepts, and the possibility of pretence.
Perhaps some of the anxiety that we feel about these conclusions may be relieved when we consider that they do not rule out the existence of a purely causal link between mental states or processes and our behaviour (although this is not the relation of an object to its referent). Such states and processes undoubtedly play a major role in guiding and shaping both our actions and the constantly evolving ‘forms of life’ within which our language-games take place. Without them, we would have little or no cause to use such expressions as ‘understanding’ or ‘pain’, but this does not entitle us to infer that they are the entities to which our psychological concepts refer. In this respect, Wittgenstein draws a clear line between the purely physical, or causal, concerns of the sciences of the mind on the one hand, and the purely conceptual — i.e. grammatical — concerns of philosophy on the other. This dramatically reduces the extent to which the results of any a posteriori scientific investigation of mental phenomena can bear upon our understanding of psychological concepts, since there is no longer a clear point of contact or correspondence between the two.
——————
1 Hereafter abbreviated as pi, z, rp and lw, respectively.
2 The term ‘picture’ refers to the models, metaphors and assumptions that lie behind and guide our use of language, such as the model of an ‘object’ and its ‘designation’ (pi §293) or the notion of ‘inner’ and ‘outer’ phenomena. For Wittgenstein, such pictures are prior to language use, although in many cases, they — or the temptations to adopt them — are themselves are embedded within language (cf. pi §115). This underlines the need for careful cross-referencing and examination of multiple cases in order to prevent linguistic habits and prejudices from obscuring genuine grammatical distinctions.
3 Anscombe’s translation of Tiefengrammatik as ‘depth grammar’ is somewhat eccentric since this is the antonym of Oberflächengrammatik, making ‘deep grammar’ the more natural choice.
4 Of course, even this sentence could be given a use under certain circumstances.
5 Although an arbitrary definition of the term can always be stipulated (pi §68).
6 Meaning is, for Wittgenstein, defined by use ‘in a large class of cases’ (pi §43).
7 The view to which Wittgenstein (1922) was committed in the Tractatus.
8 This is comparable to Putnam’s (1975) ‘Twin Earth’ thought experiments in which the meaning of a term is shown to vary according to external factors, except that here the relevant factors are distributed in both time and space.

Bibliography
Chomsky, Noam 1957: Syntactic Structures. The Hague/Paris: Mouton.
Descartes, René 1641: Meditations on First Philosophy. In Cottingham (ed.), Cambridge: Cambridge University Press.
Donagan, A. 1956: ‘Wittgenstein on Sensation’. In G. Pitcher (ed.), pp. 324–51. London: McMillan.
Egidi, Rosaria 1995: Wittgenstein: Mind and Language. Dordrecht: Kluwer Academic Publishers.
Goethe, J. W. von 1790: Versuch die Metamorphose der Pflanzen zu Erklären. Germany: Gotha.
Goldfarb, Warren 1989: ‘Wittgenstein, Mind, and Scientism’. The Journal of Philosophy, 86 (11), pp. 635–42.
Hilmy, S. Stephen 1995: ‘Wittgenstein on Language, Mind and Mythology’. In Wittgenstein: Mind and Language, R. Egidi (ed.), pp. 235–48.
Kripke, Saul 1980: Naming and Necessity. Oxford: Blackwell.
McGinn, Marie 1997: Wittgenstein and the Philosophical Investigations. Abingdon: Routledge.
Monk, Ray 1991: Ludwig Wittgenstein: The Duty of Genius. London: Vintage.
Mounce, H. O. 2002: ‘The Inner and the Outer’. Philosophical Investigations, 25(1), pp. 67–78.
Putnam, Hilary 1975: ‘The Meaning of “Meaning”’, Philosophical Papers, Volume II: Mind, Language, and Reality. Cambridge: Cambridge University Press.
Wittgenstein, Ludwig 1922: Tractatus Logico-Philosophicus. London: Kegan Paul, Trench & Trübner.
—————  1967: Zettel (z). G. E. M. Anscombe and G. H. von Wright (trans.). Oxford: Blackwell.
—————  1980: Remarks on the Philosophy of Psychology (rp), Volume 2. G. H. von Wright and H. Nyman (eds.), C. G. Luckhardt and M. A. E. Aue (trans.). Oxford: Blackwell.
—————  1992: Last Writings on the Philosophy of Psychology (lw1 & lw2), Volumes I & II. G. H. von Wright and H. Nyman (eds.), C. G. Luckhardt and M. A. E. Aue (trans.). Oxford: Blackwell.
—————  2001: Philosophical Investigations (pi), Third Edition. G. E. M. Anscombe (trans.). Malden, Massachusetts: Blackwell.

Time, Tense and Adverbial Change

Tue, 29 May 2007 12:58:12 +0100

Does adverbialism about time solve the problem of A-series changes in events?
I. Introduction
In ‘The Unreality of Time’, J. Ellis McTaggart (1908) describes two different conceptions of time, concluding that, since neither can account for the existence of substantial change, time must be unreal. McTaggart claims that such change involves the progression of events from the future, through the present and into the past — a set of determinations that McTaggart calls the A-series. The A-series is inherently ordered and directional, since events assume these determinations in strict succession. The B-series, on the other hand, is a linear and directional sequence of times or events that are ordered according to the relations of ‘before’ and ‘after’. Since B-series relations are fixed and immutable, McTaggart argues that only the A-series can account for the existence of genuine change, and thus for the reality of time. However, he goes on to argue that since the A-series requires events — and by extension times — to assume the contradictory determinations of past, present and future, it cannot be real. In this paper, I describe a potential solution to the paradox presented by E. J. Lowe (1998: 130; 2002: 312–9), which involves viewing tense as a changing series of adverbial relations between a perceiver and events. In particular, I will evaluate whether and how Lowe’s account is able to overcome the alleged contradiction in the A-series without reducing his adverbial conception of tense to the unchanging relations of the B-series. I will then examine a number of potential objections to adverbialism concerning the treatment of complex tenses before concluding with a diagnosis of the source of the disagreement between Lowe and McTaggart on the nature tensed facts.
II. McTaggart’s Regress
The contradiction in the A-series arises as a result of events changing their location within the series as they move from future to present to past with the passage of time. For simplicity, let us consider just the three A-series determinations of past, present and future. An event e that has yet to occur — the rising of the sun tomorrow, for example — starts off as being future, which may be symbolised as Fe. Since past, present and future are mutually exclusive terms, ‘e is future’ also entails ‘e is not-present’ (¬Ne) and ‘e is not-past’ (¬Pe). As time passes and the event occurs — i.e. the sun rises — e is then present (Ne), before finally receding into the past (Pe). Thus, each and every event is:
(1) Present when it is occurring, and not-present when it is past or future
(2) Future when it has yet to occur, and not-future when it is present or past
(3) Past when it has occurred, and not-past when it is present or future
Since each of the above conjunctions constitutes a contradiction, they cannot be true of one and the same event. The obvious response to this is that neither (1), (2) nor (3) is true of the same event at the same time, since each conjunct only holds of a given event at different times. However, in order to show that the A-series view of time is coherent, we cannot simply help ourselves to the notion of time, and must further unpack this claim in a way that shows it to be internally consistent, but without reducing it to the static and unchanging relations of the B-series.
McTaggart considers how the obvious response might be developed by introducing a more finely-grained series of predicates — FF, FN, FP, NF, NN, NP, PF, PN and PP — that may be applied to each event. Now, instead of three mutually incompatible determinations, we have nine partially compatible determinations, out of which any given event will assume at least one from the first set of three, at least one from the second set, and at least one from the third set. Thus, a presently occurring event, ‘will be past’ (FPe), ‘is now present’ (NNe) and ‘was future’ (PFe) [Mellor 1998: 74]. Since these three predicates are expressed in different tenses, they are perfectly compatible, and so the immediate contradiction is avoided. However, a similar problem now arises when we consider what happens after the event has occurred. In this case, the event may be described as FPe, NPe and PNe, since it now lies entirely in the past. This conflicts with the above determinations since NPe entails ¬NNe, which contradicts NNe above. Thus each event must still assume more than one of the incompatible determinations, leading to a contradiction. Any attempt to resolve this by introducing yet more finely-grained predicates corresponding to the 27 three-place combinations of past, present and future (i.e. FFF, FFN, FFP, FNF, etc.) will fail in a similar way. Although it resolves the immediate conflict between the original nine predicates, certain combinations — e.g. FFF and PPP — are still incompatible, and since every event must at some point assume the incompatible determinations, there is no escape from the regress. Thus, McTaggart concludes that the regress is vicious, the A-series unreal, and since changes in tense are essential to the concept of time, time is unreal.
There are many possible responses to this argument. The first is to deny the viciousness of the regress, as per Lowe (1987a: 64), who claims that the contradiction within A-series terms is indeed resolved through the introduction of the additional temporal predicates at each level. Another response is to argue that all talk about events may be reduced to statements about objects undergoing change (Prior 1968). This side-steps the contradiction on the basis that only objects change, and not events, since events are changes, although this does not address the problem of temporary intrinsics described below. A third option is to agree with McTaggart about the unreality of the A-series, but to argue in favour of a tenseless B-series conception of time (e.g. Mellor 1998). However, this requires an account of substantial change that is based on the essentially static and unchanging relations of the B-series — a position that McTaggart (1927: 264) considers untenable. Finally there is the possibility of an adverbial response, which is the option that I will be examining throughout the remainder of this essay.
III. Adverbialism about Objects
In order to demonstrate how an adverbial account of time might address the apparent contradiction within the A-series, I will first examine an adverbial response to the distinct, but related, ‘problem of temporary intrinsics’ (Lewis 1986: 202). This concerns the contradiction between intrinsic (i.e. internal non-relational) properties of objects that change over time, such as a banana that changes from being green to being yellow as it ripens (Hawley 1988: 211). Just as ‘past’, ‘present’ and ‘future’ are mutually exclusive determinations of events, so a single object, a, cannot be both green-all-over and yellow-all-over without contradiction, since the statement ‘a is green-all-over’ entails ‘a is not yellow-all-over’, which is incompatible with ‘a is yellow-all-over’. Lewis identifies three possible solutions to this problem, which involve temporally qualifying the property, copula and object, respectively (Lowe 1988: 73). The first of these options describes properties as relations to times, e.g. green-at-t1 and yellow-at-t2, which unlike green-all-over and yellow-all-over are perfectly compatible since a is green-at-t1 only entails ¬(a is yellow-at-t1) and not ¬(a is green-at-t2). Lowe and Lewis object to this solution on the grounds that it makes temporary properties relational, which is highly counterintuitive as it is unclear how ‘having a property’ can be construed in terms of an object’s relation to a time (cf. Hawley op. cit.). The second option, which is favoured by Lewis, incorporates the temporal reference in the object such that each object consists of multiple ‘temporal parts’ (Lewis op. cit. 202), which are the bearers of its (temporary) properties.1 Accordingly, it is not the object (a) that is both green and yellow, but a-at-t1 (an earlier temporal part of a) that is green and a-at-t2 (a later temporal part) that is yellow. Just as there is no contradiction between having one of our spatial parts (e.g. our hands) pointing left while the other points right, different temporal parts of an object may possess conflicting properties without contradiction. However, this solution requires positing the existence of an entirely novel class of objects, namely temporal parts, with which we are otherwise unfamiliar, and so could be accused of being ontologically extravagant or ad hoc (Lowe 1998: 134). Finally, there is the adverbial account, which temporally qualifies the relation between an object and its properties by means of an adverbial modifier, which indicates its tense. Just as ‘possibly green’ and ‘actually yellow’ are compatible modal claims, ‘a is-at-t1 green’ and ‘a is-at-t2 yellow’ do not conflict, since is-at-t1 and is-at-t2 are distinct ways in which an object can possess the properties of being-green or being-yellow.2
Philosophical opinion on adverbialism as a solution to the problem of temporary intrinsics is divided. Lewis rejects the adverbial account as a mere variant of the relational account, claiming that it involves a three-place relation between objects, properties and times (Lewis 1988: 66 fn.). Haslanger (op. cit. 121) and Lowe (op. cit. 74) deny that any such three-place relation is required and defend adverbialism on the grounds that it alone supports the intuition that objects are fully present throughout their lifetime, i.e. they endure, as opposed to Lewis’s ‘perdurance’ theory, in which only one ‘temporal part’ of an object is present at any given time (Lewis op. cit. 202). Regardless of the merits of adverbialism as a solution to the problem of change in objects, however, a similar account may also be given regarding the changing properties of (or relations between) events. This seems particularly appealing given the close relationship between adverbial modifiers and the everyday notion of tense, which may be understood in terms of adverbial modification of the copula. In the following sections, I will examine how one such account (Lowe 1998; 2002) purports to overcome the contradiction within McTaggart’s A-series, along with some potential objections to it.
IV. Adverbialism about Tense
Lowe’s adverbial account of tense closely resembles McTaggart’s initial response to the contradiction between the incompatible A-series determinations of ‘past’, ‘present’ and ‘future’. Like McTaggart, it involves the introduction of the tensed relations of was, is now and will be. However, unlike McTaggart, Lowe denies even the appearance of a contradiction, since the three tenses are conceived of as adverbially modified ways in which an event may possesses its (temporal) properties. This enables claims such as ‘e was future’ and ‘e is now past’ to be made of one and the same event without contradiction, provided that they involve different A-series adverbial modifiers, or tenses. Specifically, Lowe (1998) claims that:
(1) For any event e, (i) it either was, is now, or will be true to say ‘e has happened’, and (ii) is either was, is now or will be true to say ‘e is happening now’, and (iii) is either was, is now or will be true to say ‘e will happen’.
(Lowe 1998: 91; my italics)
On this account, the statement ‘e is now past’ says nothing more than ‘e has happened’. Similarly, ‘e was future’ may be analysed as ‘e has happened or is happening now’, since an event that was (but is no longer) future must be either past or present. As per the adverbial solution to the problem of temporary intrinsics, these adverbially modified relations enable the same event to possess the conflicting properties of pastness, presence or futurity in different ways, thus overcoming the initial contradiction within McTaggart’s A-series. However, in order to avoid the subsequent regress, Lowe must also explain how it is possible for the same event to be successively past, present and future (Lowe 2002: 316), but in a way that does not invoke the notion of ‘times’, which would reduce his account to the static and unchanging relations of the B-series.
Lowe explicitly rejects the idea that we should ‘reify moments of time’, i.e. ‘to treat them as real entities, to be included in our ontology along with events and persisting objects’ (ibid. 315; original italics). Instead, he introduces the A-series adverbs pastly, presently and futurely, which adverbially modify the manners in which events can be past, present or future.2 This gives nine possible combinations such that
(1) Every event is such that it is (i) either pastly past or presently past or futurely past and (ii) either pastly present or presently present or futurely present and (iii) either pastly future or presently future or futurely future.
(ibid. 318; my italics)
Accordingly, the statement that or ‘e is future at some time in the past’, may be unpacked as ‘e is-pastly future’, which is compatible with ‘e is-presently past’ and ‘e is-futurely past’ due to the different manners in which pastness or futurity are predicated. In this way, Lowe reduces all statements concerning times to statements involving A-series adverbs, since tense is, on his account, conceptually prior to time, and not the other way around.
According to Lowe (2002) then, any event that is-presently future will also be futurely present, since the two terms are coextensive, and similarly for the other adverbial locutions, mutatis mutandis. This gives rise to an obvious objection concerning the use of is and will be in such expressions, since if these terms are themselves considered to be tensed then this gives rise to a regress, as in McTaggart (ibid. 315). However, Lowe argues that ‘is-presently future’ is equivalent to ‘is future’, where the ‘is’ is present-tensed, and ‘will-futurely-be present’ to ‘will be present’.3 In other words, the tense of the copula does no additional work in these statements — it is only there to make the grammar consistent — and so no regress need arise. If the copula is considered to be tenseless, on the other hand, then there is no need to reintroduce the notions of pastly, presently and futurely in order to explain its meaning, and so either way the regress is avoided.4
A more substantive objection to Lowe’s adverbial account is that it seems not to have solved the original problem of how the same event can be both ‘presently future’, ‘presently present’ and ‘presently past’, since all events must still successively assume these conflicting determinations due to the passage of time. The claim ‘e is-presently present’, which may also be expressed as ‘e is happening now’, entails ‘not (e is-presently future)’, or ‘it is not the case that e was happening’. Nevertheless, it will subsequently be the case that ‘e was happening’, and so it seems that we must say of e that it is both happening and not happening, which is contradictory. Lowe denies that this argument has any force on the basis that it equivocates over either statements or times, the latter of which he has already excluded from his ontology (ibid. 316). For Lowe, e is either (i) ‘presently present’ and ‘pastly future’, or (ii) ‘presently past’ and ‘pastly present’, but never (iii) ‘presently present’ and ‘presently past’, or (iv) ‘pastly future’ and ‘pastly present’, which would be incompatible. Of course, it may be the case that e was-presently present and is-now-presently past, but since both of these claims involve different A-series adverbs, or tenses, there is no contradiction between them, and so no cause for regress.5
V. Complex Tenses
So far, I have considered only the simplified A-series determinations of ‘past’, ‘present’ and ‘future’. This is sufficient for dealing with expressions like ‘I had finished eating’, which may be analysed as ‘my finishing eating was-pastly past’. However, the situation becomes more complex when we consider statements such as
(1) The Battle of Hastings was future, was present and was past.
whose components are true of the years 1065, 1066 and 1067, respectively. A naïve adverbial analysis of (6) along the lines described above would yield
(1) The Battle of Hastings is-pastly future, is-pastly present and is-pastly past.
which is contradictory, since all three determinations are predicated in the same manner, i.e. pastly. In order to avoid this problem, Lowe employs more finely-grained A-series adverbs, as in
(1) The Battle of Hastings is-pastly future, is-more-pastly present and is-still-more-pastly past.
This may be expressed in a more general form as
(1) The Battle of Hastings is X-ly future, Y-ly present and Z-ly past.
where X, Y and Z are A-series adverbs corresponding to the relative ‘degrees of pastness’ (Lowe 1998: 92) of the claims in question, such as 942-years-past, 941-years-past, and so on. Here, Lowe’s idea seems to be that terms such as ‘pastly’, ‘more-pastly’ and ‘still-more-pastly’ quantify over the A-series such that the ordinary English terms, was, is now and will be, identify distinct regions of the series that are relevant to the claim in question (ibid.). Since (9) states that the conflicting properties of pastness, presence and futurity are held in different adverbially modified manners (X-ly, Y-ly and Z-ly), the contradiction is avoided. Furthermore, due to the token-reflexive truth-conditions of tensed sentences (op. cit. 90), the relevant A-series adverbs for a given event will constantly change as it becomes ‘less future’ and ‘more past’ with the passage of time (Lowe 2002: 317), causing X, Y and Z to range over different regions of the A-series depending upon when this statements is asserted.
This modified adverbial locution, however, gives rise to two problems. Firstly, the notion of ‘quantifying over the A-series’ has yet be fully explained. Secondly, in order for (9) to intersubstitutable for (6), it is necessary to add the constraint that X is more-past-than Y, which is more-past-than Z, thus requiring comparison between A-series adverbs. To illustrate the first of these problems, consider:
(1) John had already eaten when Mary arrived.
(ibid.)
On a B-series view of time, this would be understood to mean that there is a time before now (an indexical expression) at which Mary arrives and that is after (a B-series relation) the time of John’s eating. By substituting the references to times for A-series adverbially modified relations, we obtain
(1) There is an adverb, X, such that John’s eating is X-ly past and Mary’s arrival is X-ly present.
It is tempting to read both (9) and (11) as quantifying over, or ‘reifying’, times, thus undermining the primacy of the A-series, and reducing Lowe’s tensed A-series locutions to their tenseless B-series equivalents. However, X, Y and Z are not times, but A-series adverbs and so, on an adverbial account, conceptually prior to times. Moreover, A-series adverbs are not existent entities, but ways in which the relation between an event and its properties may be modified; i.e. they are ‘predicate modifiers’ (Lowe 1998: 94). Thus, Lowe need not be taken to be reifying the terms of the A-series, other than to say that they are the series of temporal modifiers that affect the way in which events possess their properties, and so are themselves neither actual nor ‘in time’ (although they do of course give rise to times).
The second problem raised by (9) relates to the nature of the ‘second-order’ relations that hold between X, Y and Z concerning their relative ‘degrees of pastness’. By invoking the notion of more-past-than, it again seems that Lowe is again either quantifying over times or reducing the A-series to the B-series relations of ‘before’ and ‘after’. Lowe can perhaps escape this objection by suggesting that since the A-series is inherently ordered and directional, then A-series adverbs are comparable to the elements in a numeric series — the real numbers, for example — and the notions of ‘more-past-than’ and ‘more-future-than’ analogous to the mathematical operators, ‘less than’ and ‘greater than’, which compare the position of terms within the series. This would not require the existence of any additional entities, such as times, and does not reduce the A-series to the B-series, since only the relative position of A-series adverbs can be compared in this way, with the more fundamental notions of ‘past’, ‘present’ and ‘future’, which have no B-series equivalents, remaining irreducible.
How the notion of quantifying over adverbial modifiers can be represented in formal logic lies outside the scope of this essay, but the basic concept may perhaps be clarified by considering the parallel case of degrees of probability. The fact that such degrees may be ordered in terms of their likelihood does not require us to reify probabilities corresponding to the various adverbial modifiers in question (although it does perhaps require that they are quantifiable in some way — numerically, for example). If this kind of quantification can be made to work, then Lowe’s adverbial account offers the possibility of supporting arbitrarily complex tenses without the need to iterate or ‘stack up’ adverbial modifiers (ibid.), since each additional clause simply specifies further constraints upon the ‘second-order’ relations between the A-series adverbs in the sentence, thereby escaping McTaggart’s regress.
V. Conclusion
I have argued that Lowe’s adverbial account of time does overcome the contradiction within McTaggart’s A-series, but with two important caveats. Firstly, it must be possible to quantify over A-series adverbs without reifying its terms, or B-series ‘moments of time’. Secondly, Lowe rejects McTaggart’s implicit assumption that it is possible to provide a ‘complete description of reality’ (Dummett 1960: 356) irrespective of one’s temporal perspective. Instead of making tenseless claims about one and the same event at different points throughout its lifetime, as McTaggart attempts to do, and which leads to a contradiction, Lowe accepts that all contingent facts are essentially tensed (Lowe op. cit. 90) and so are only true or false from a particular temporal perspective. Whilst we can evaluate the truth or falsity of statements from different temporal perspectives, there is, for Lowe, no question of being able to ‘step outside time’, and, since the sum total of all temporal facts is always consistent, the alleged contradiction in the A-series never arises. On this account, McTaggart’s paradox is founded upon a misconception, since it attempts to do just that (Lowe 2002: 319). For Lowe, we simply cannot avoid the need for tense — which is perhaps unsurprising given that on an A-series account of time, the notions of past, present and future are taken to be essential and irreducible.
——————
1 Although not all adverbial modifiers afford this kind of compatibility; quickly and slowly, for example.
2 Is-pastly, is-presently and is-futurely in Lowe (2002) are logically and semantically equivalent to was, is-now and will be in Lowe (1998).
3 In some cases, ‘will-futurely-be present’ will be taken to mean ‘is-more-futurely present’, as considered in section V above.
4 This explains Lowe’s use of the disjunctive ‘is, was, or will be’ in (4) above.
5 Indeed, for Lowe, the truth of both of these statements is determined by the same A-series fact: whether or not e has happened.

References
Dummett, Michael 1960: ‘A Defence of McTaggart’s Proof of the Unreality of Time’. Philosophical Review, 69, pp. 497–504.
Haslanger, Sally 1998: ‘Endurance and Temporary Intrinsics’. Analysis, 49 (3), pp. 119–25.
Hawley, Katherine 1998: ‘Why Temporary Properties are Not Relations Between Physical Objects and Times’. Proceedings of the Aristotelian Society, 98 (2), pp. 211–6.
Lewis, David 1986: On the Plurality of Worlds. Oxford: Blackwell.
—————  1988: ‘Rearrangement of Particles: Reply to Lowe’. Analysis, 48 (2), pp. 65–72.
—————  2002: ‘Tensing the Copula’. Mind, 111 (441), pp. 1–13.
Lowe, E. J. 1987a: ‘The Indexical Fallacy in McTaggart's Proof of the Unreality of Time’. Mind, 96 (381), pp. 62–70.
—————  1987b: ‘Lewis on Perdurance versus Endurance’. Analysis, 47 (3), pp. 152–4.
—————  1988: ‘The Problems of Intrinsic Change: Rejoinder to Lewis’. Analysis, 48 (2), pp. 72–6.
—————  1998: The Possibility of Metaphysics. Oxford: Oxford University Press.
—————  2002: A Survey of Metaphysics. Oxford: Oxford University Press.
Mellor, D. H. 1998: Real Time II. London: Routledge.
McTaggart, J. Ellis 1908: ‘The Unreality of Time’. Mind, 17 (64), pp. 457–74.
—————  1927: The Nature of Existence, Volume II. Cambridge: Cambridge University Press.
Prior, Arthur 1968: ‘Changes in Events and Changes in Things’. In The Philosophy of Time, R. Le Poidevin and M. MacBeath (eds.), pp. 35–46.

Heidegger and the Phenomenology of Time

Tue, 29 May 2007 12:00:00 +0100

Does Heidegger’s Being and Time Resolve the Contradiction within the A-series?

Contents
I. Introduction
II. The Dimensional Conception of Time
III. The Phenomenology of Time
Zuhandenheit and Vorhandenheit
Authentic temporality
Public time
IV. McTaggart Revisited
Duration and change
The authentic past, present and future
Events and agency
V. Conclusion
I. Introduction
In Being and Time, Martin Heidegger (1996) presents a detailed analysis of the nature of, and relationship between, these two most fundamental concepts of philosophy. In the context of investigating the ‘meaning of being’ — a question Heidegger claims that has been neglected since Aristotle — time emerges as ‘the horizon1 of the understanding of being’ (bt 17) or, to put it in Kantian terms, the precondition for all human knowledge and experience. Heidegger’s primarily phenomenological account may be contrasted with more conventional analyses of time as being an objectively present quality, property or dimension along which events are arranged, or series of relations between objects. However, according to an argument by J. Ellis McTaggart (1908), such accounts become incoherent when we consider the issue of changing tense, which McTaggart takes to be essential to the concept of time, since no event can be both ‘past’, ‘present’ and ‘future’ without generating a contradiction.2 McTaggart concludes that, since neither the tensed ‘A-series’ account of time, nor the tenseless ‘B-series’ is sufficient to account for time, then time must be unreal.
In this essay, I present an interpretation of Heidegger’s phenomenology that aims to defuse this sceptical paradox by rejecting certain key assumptions in McTaggart’s account, chiefly the notion that time may be described in terms of a series, or dimension. I begin by giving an account of this ‘dimensional conception of time’ and its relation to McTaggart in section II below. In section III, I sketch out the structure of Heidegger’s phenomenological account, beginning with the ‘existential meaning’, or significance, of time, followed by the phenomenon of authentic temporality and the emergence of public time. In section IV, I show how this account highlights the presuppositions that give rise to the contradiction in McTaggart’s A-series, and how this may be resolved by rejecting the dimensional conception of time in favour of Heidegger’s existential account. Finally, I conclude that the contradiction in the A-series arises a result of taking time to be an objective, as opposed to existential, notion and, as such, is based on a misguided attempt to view time from an atemporal perspective.
II. The Dimensional Conception of Time
Since Aristotle (1984: 218b 21ff), the conventional view of time has been that of a ‘series of nows’ (ibid.). In it, each moment of time is a complete and self-contained ‘instant’, devoid of any change or activity, and ordered relative to previous and successive instants by the temporal relations of before and after. This gives ontological priority to the phenomenon of the ‘now’, or present, which is defined either as a distinguished moment that traverses the series, and so is ever-changing (as in McTaggart’s A-series), or as an indexical term denoting the temporal ‘location’ of the observer (as per the B-series) [cf. Mellor 1998]. Regardless of whether time is regarded as relational (Leibniz 1956) or absolute (Newton 1689; Einstein and Minkowski3 in Barbour 1999: 142), what these views have in common is that they attempt to spatialize time; i.e. to represent time as being analogous to a spatial dimension (tb v). This ‘dimensional’ conception of time (bpp 248) is deeply entrenched in our ways of talking and thinking about time, particularly in the natural sciences, where time is often characterised as a ‘timeline’ or dimension along which events may be placed (Figure 1). Similarly, our systems of measuring and representing time — e.g. sundials, calendars and clocks — all involve the spatial projection of temporal relationships, making it quite natural for us to think of time as a kind of space ‘within which events take place’ (ct 3).4
However, there are also many dissimilarities between space and time. Unlike space, time — at least as we experience it — is asymmetrical due to differences between the past, which we think of as being immutable and certain, and the future, which is generally thought of as largely uncertain and unknown. Time also has an inherent directionality. We are able to move up, down, left, right, forwards or back within space, but — despite theoretical claims regarding the possibility of time travel — are only ever aware of a single ‘direction’ of time, which advances in predictable fashion, with future events becoming present and then past. Despite the prevalence of spatial metaphors, such as ‘movement through time’ or the ‘flow of time’, there are clear differences between time and space, and so we should not be surprised if the analogy between them breaks down at some point.
McTaggart’s A-series, which consists of the temporal determinations that are assumed by events as they ‘move’ from the far future, through the present and into the distant past, is a dimensional account par excellence. Without wishing to provide a detailed description of it here,5 the contradiction that it presents concerning the incompatible properties of events is symptomatic of the difficulties inherent in any such account. As such, McTaggart’s conclusion — that both the A-series and time are unreal — may also be read as a reductio of the dimensional conception of time that gives rise to this contradiction, and so favours the adoption of an alternative conception of time and temporality in which the problem does not arise.
III. The Phenomenology of Time
Being and Time is a primarily phenomenological, as opposed to primarily metaphysical, study of time. That is to say, Heidegger aims to determine the existential meaning, or significance, of time, rather than to directly analyse its ‘essence’ or nature. In particular, Heidegger’s account calls into question certain presuppositions that are central to McTaggart’s conception of the A-series, and therefore of time, namely, (i) that ‘past’, ‘present’ and ‘future’ are mutually exclusive determinations, (ii) that time may be characterised as series of homogeneous ‘instants’ or locations that stand in various relations to events, and (iii) that time is entirely objective and independent of human agency. In order to show how these presuppositions give rise to the contradiction within the A-series, I will first give an overview of Heidegger’s phenomenological account, starting with the notion of ‘to-handedness’ and objective presence.
(i) Zuhandenheit and Vorhandenheit
According to Heidegger, the being of everyday things inheres in their Zuhandenheit (‘to-handedness’; bt 71).6 That is, they are ontologically grounded in familiar forms and functions that are so integrated in our everyday lives that we barely even notice their presence. Ordinary objects such as tools, clothes, signs, and much of the natural world form what can be described as a network of significance — i.e. the web of meaningful relationships by which we navigate and manipulate the world. This phenomenal world, as distinct from the physical universe described by the natural sciences, is primary for us, and we grasp it and everything contained within it in terms of the possibilities that they present. The Zuhandenheit of a table, for example, is constituted by the possibilities it presents for gathering around it for meals, its use as a work surface, being broken up to use as firewood, and so on (Inwood 1997: 33). Together, these comprise its meaning for us as a phenomenal object (as opposed to a physical object, which presents only the possibility for repeated intra-personal observation [Beyer 2007]). Since Zuhandenheit is essentially related to our projects and purposes, it is primarily future-directed. It also incorporates an essential reference to the human subject; i.e. it is not objective, but neither is it purely subjective, since an object may possess the same phenomenal qualities for many different subjects. This is summed up in our fundamental attitude towards the world which is, for Heidegger, that of Sorge (‘care’), or active and self-concerned engagement.
Time, as distinct from temporality, also has its place in the network of significance. The day, delineated by the sunrise and sunset, is the most basic and natural (although still not entirely primitive) measure of time, and understood in terms of the possibilities that it presents for wakeful activity during the hours of daylight, and for sleep or rest at night (bt 412). Similarly, there are times to eat, times to go out, times to plant and harvest crops, and so on.7 Thus, the everyday concept of time exhibits the significance of Zuhandenheit, making each moment not just an isolated point or ‘location’ in the timeline, but a ‘time for . . .’ that has meaning in its own right (bt 414). Time is therefore also subject-dependent, but not subjective since ‘it is not a characteristic of the subject, but rather a mode or way in which the subject is in the world’ (Gelven 1989: 218; original italics).
Heidegger contrasts the significance of Zuhandenheit with Vorhandenheit, or ‘objective presence’ (bt 61). When confronted with a being outside of its usual context, such as a tool that is broken or missing, for example, or when we study it scientifically, it becomes ‘levelled down’ to a mere thing. As such, it is stripped of its usual significance — its ‘worldliness’ (bt 63) — and seen solely in terms of ‘brute facts’, such as might be amenable to logical or mathematical analysis. In moving from the engaged significance of Zuhandenheit to the dispassionate and disinterested mode of Vorhandenheit, we turn Being from a ‘how’ (i.e. a meaningful set of possibilities that relate to some future end) into a ‘what’ (i.e. a mere thing) [ct 12]. Thus, for Heidegger, the natural sciences do not address ordinary being as such, but a secondary or ‘deficient’ (bt 73) mode of being that is derived from — and therefore secondary to — the world of Zuhandenheit. Although Heidegger does not denigrate the notion or study of objective presence, which is essential to both science and human creativity (e.g. in the design and manufacture of artefacts), the ‘levelling down’ that this entails causes the resulting concepts to lack the significant relations of everyday phenomenal experience. Thus it follows that science is, for Heidegger, secondary to phenomenology, and not the other way round.
(i) Authentic temporality
The distinction between authentic and inauthentic existence is central to much of Heidegger’s philosophy (bt 43). ‘Authenticity’ is characterised by ‘resolute’ self-awareness and pursuit of deliberately chosen goals and commitments, informed by individual and historical circumstances. ‘Inauthenticity’, on the other hand, is characterised as being ‘swept along’ (bt 348) by the passive group-mentality that is dominated by the pursuit of novelty, ‘busyness’ (bt 178) and ‘idle talk’ (bt 167). Perhaps the most striking aspect of this distinction may be found in the individual’s attitude towards death. As with all existent beings, Da-sein8 is essentially temporal and ‘stretched along’ (bt 373) in time. The life of Da-sein — and therefore its time — is necessarily limited as it gradually ‘uses itself up’ (bt 333) until the inevitable moment of its death, beyond which Da-sein no longer ‘is’ except as a lifeless and objectively present thing — a corpse. Thus, in order to live authentically, i.e. in accordance with its nature as a self-determining and self-interpreting being, Da-sein must choose which of its possibilities to realise in the limited time available to it.
However Da-sein’s choices are limited or constrained by the situation into which it is ‘thrown’ (bt 135) as a result of previous actions and other contingent aspects of its existence. This ‘having-been’ (bt 326) reveals both how Da-sein came to be in its present situation and what its possibilities for action are, and is the ontological basis for the phenomenon of ‘the past’. The authentic past is not an exclusive determination, but rather just one aspect, or ekstasis (‘standing out’; Gelven op. cit. 182), of an essentially unified and coherent whole, since the past can only be understood relative to the current projects, goals and possibilities of Da-sein, which are futural. Similarly, the dimensional ‘future’ is replaced with the projection of future possibilities, or ‘futural projection’, which retains an essential reference to the past, since it is through past experience that we grasp the possibilities that objects, or the life of Da-sein, have to offer. The perception of a chair, for example, as distinct from a mere jumble of materials, involves our having-been familiar with chairs and their usage in the past, as well as the projection of its possibilities — e.g. for sitting upon — into the future. Past, present and future are therefore, for Heidegger, very much bound together and permeate every ‘moment’ of time (Dostal 1993: 156), rather than being external to one another, as on the dimensional conception.
Heidegger’s shift from talking about time to talking about temporality reflects the ontological priority that he gives to the subject and its manner of being in the world. Since time is not a being, it is not in the world, but rather the means by which the world is manifest to us (bt 365). Heidegger rejects the question of what time is as misguided since it reifies time, treating it as a ‘mere thing amongst things’ (bt 26 in Sinclair 2006: 140), and instead concentrates upon ‘how [Da-sein] “has” its time’ (bt 410). Da-sein’s temporality (and therefore time) is manifested through its ‘temporalizing’, which may be authentic or inauthentic, according to its manner of being. Inauthentic temporality, which is constitutive of everyday being, is characterised by the process of ‘making present’. In this mode of temporalizing, ‘the present’ is not an objective state of affairs, as with the dimensional conception of time, but a continually manifested and manifesting totality that arises from the juxtaposition of past experience (‘thrownness’) with future possibilities (projection). It therefore combines elements of both past and future in a single unified whole: the ‘e[k]static unity of temporality’ (bt 408).
Whilst inauthentic temporality is primarily centred upon the phenomenon of the present, with past and future characterised by the passivity of ‘forgetting’ and ‘awaiting’, respectively, authentic temporality is primarily directed towards the future and the possibility for agency. Acknowledging the certainty of its own death, authentic Da-sein resolutely pursues its projects and commitments, thereby actively ‘anticipating’ and shaping its own future. The authentic present, or ‘Moment’ (Augenblick; bt 338), combines both past and future in the possibility for action, which is primarily future-directed. For Heidegger, it is this threefold structure of authentic temporality, or ‘datability’, that makes engaged concern possible (bt 328), and is what gives time its distinctly temporal, as opposed to spatial, structure (Figure 2). Thus, on Heidegger’s account, the past, present and future are not mutually incompatible determinations, as they are for McTaggart, but essentially unified aspects of a single phenomenon: the authentic ‘temporalizing of temporality’ (bt 406–7). Although the balance between these three aspects, or modes, is constantly shifting, it is impossible to conceive of one without the others. Consequently, for Heidegger, the past, present and future describe only isolated aspects of this temporal unity, and are necessarily derivative and incomplete (bt 326).
(i) Public time
The final element of Heidegger’s account of time is his description of the emergence of the ‘vulgar concept of time’ (bt 404). This emerges when we move from a subject-dependent view of time to the purely objective conception of ‘public time’ or ‘clock time’. As described above, this involves a ‘levelling down’ or loss of significant meaning, fragmenting the essential unity of time into a series of objectively present elements: the inauthentic past, present and future of the dimensional account. Time is ‘public’ because it is accessible to everyone, and may be measured by dividing it into discrete, recurring intervals using devices such as sundials or clocks (bt 416). However, the movement from the Zuhandenheit of everyday time to the existence of objectively present ‘times’, which are purely formal and so inherently meaningless, reveals the essential inauthenticity of public time (Gelven op. cit. 217). Each moment of clock time becomes an idealised point-like instant, or ‘now’, independent of the other ‘nows’ with which it stands in temporal and causal relations. The characteristic asymmetry, directionality and advancement of time are replaced by a single homogeneous ‘dimension’. Moreover, we can no longer locate ourselves within this temporal continuum since each ‘now’ is merely objectively present, as opposed to bearing an essential reference to the experiencing subject. Consequently, the notion of ‘the present’ either becomes redundant (as in McTaggart’s B-series), or something that ‘pass[es] along’ the sequence (as in the A-series; McTaggart op. cit. 463). However, it is no longer clear precisely what is is moving or why this should be the case since the subject is longer ‘stretched along’ in time, but standing ‘outside’ it, looking in. It is this disembodied view that, according to Heidegger, gives rise to the paradoxes and contradictions within our philosophical understanding of time, including McTaggart’s A-series, as described in section IV below, since ‘[o]nce time has been defined as clock time then there is no hope of ever arriving at its original meaning again’ (ct 18–9).
Because the purely logical relations that hold between times can always be extended to encompass still previous and subsequent times, public time is also effectively infinite (bt 424). This ‘covers over’ the inevitability of Da-sein’s death, and is contrary to experience, which gives us no reason to believe that time is unlimited, even when the existence of multiple subjects is taken into account.9 For Heidegger, this is erroneous, since it is only from the perspective of an embodied subject within the world that we can make sense of such worldly phenomena as being and time.
IV. McTaggart Revisited
Having established the basic structure of Heidegger’s phenomenological account, I will now consider how this addresses the contradiction within McTaggart’s A-series. The contradiction arises because ‘past’, ‘present’ and ‘future’ are conceived as mutually incompatible intrinsic or relational properties of events. Consequently, one and the same event cannot possess more than one of these determinations. However, since the passage of time requires every event to do just that, an event that was future, is then present, and will be past. The obvious response — that no event is required to assume more than one of these determinations at the same time — is ruled out since each ‘time’ must also successively assume the incompatible determinations, and so is similarly contradictory. Attempting to resolve this second-level contradiction by reiterating the notion of ‘at the same time’ sets up a regress, which McTaggart (op. cit. 468) takes to be vicious. Similarly, the notion of tense fails to resolve the issue since not all tenses are compatible; e.g. a single event cannot be both ‘presently present’ and ‘presently future’ (cf. Lowe 2002: 318). The same is true of complex tenses, such as ‘will be presently past’ and ‘will be presently future’, which are contradictory when applied to the same event over time, and so do not avoid the paradox.10
Authentic temporality, as described by Heidegger’s existential concepts of having-been, the Moment and futural projection, is fundamentally tensed, and so is ostensibly an A-series account. However, as previously noted, these concepts are not the mutually exclusive determinations of past, present and future that McTaggart describes, but rather unified aspects of the authentic temporalizing of temporality. Heidegger’s central claim with respect to time is that these phenomenological concepts yield an account of time and temporality that is both internally consistent and in keeping with our everyday talk and experience of time. Conversely, the inauthentic (i.e. dimensional) conceptions of past, present and future cannot fully account for the phenomenology of time (its significance, asymmetry, datability, etc.), and so is inadequate and incomplete. I will now examine the validity of Heidegger’s claim and its consequences for the A-series in detail, starting with the notion of duration, which Heidegger, like Kant, takes to be irreducible.
(i) Duration and change
For Heidegger, all being is temporal and ‘stretched along’ in time, as opposed to being reducible to a series of point-like instants or ‘nows’. This applies both to our own mode of being, which is ‘used up’ throughout our lifetime, and the being of other things, such as objects and events. Regardless of the temporal scale at which we consider the world — lifetimes, days, minutes, picoseconds — every object and event will be spread out in time, thus occupying a distinct ‘span’ or duration (bt 409). This commits Heidegger viewing time as continuous, rather than discrete, and is consistent with both scientific observation and Kant’s First Critique (op. cit. A208–11/B253–6). Since each event is ‘stretched along’ in time, it cannot properly be classed as either past, present or future while it is occurring, since the beginning and end of the event will also lie in the past and future, respectively. Thus, for Heidegger, a presently occurring event is also past and future.
An obvious objection to this argument is that the start of a presently occurring an event lies wholly in the past, and so is not present or future, and its end lies wholly in the future, and so is not past or present. However, this requires defining duration as ‘the interval between two points in time’, which is precisely the view that Heidegger rejects. Rather, for Heidegger, it is duration that is fundamental, and from which we abstract the notion of individual instants, or ‘points in time’, and not the other way round (bt 391; tb 12). Thus, even though we can always subdivide a given event into ever-shorter periods, at each level an event that is currently occurring will not just be present, but also past and future, thus exhibiting the fundamental unity of the three ekstasies of authentic temporality.
The above argument challenges McTaggart’s claim that past, present and future are mutually contradictory determinations, since all three are, at a fundamental level, coexistent on Heidegger’s conception of occurrence (bt 375). If we reject this claim, then the alleged contradiction in the A-series does not arise, since there is no longer any conflict between an event’s assuming all three determinations due to the passage of time (the original source of McTaggart’s paradox). However, this conclusion is too quick as it equivocates between the dimensional past, present and future and their authentic equivalents. In order to establish the true relation between these concepts, we must therefore consider their ontological grounds in Heidegger’s existential notions of having-been, the Moment and futural projection.
(i) The authentic past, present and future
For Heidegger, the past, present and future, although temporal concepts, are not themselves ordered in time, i.e. ‘[t]he future is not later than the having-been, and the having-been is not earlier than the present’ (bt 350; original italics). Rather, they are aspects, or modes, of our temporalizing that reflect time’s underlying structural unity. Considering each mode individually, we find that it contains elements of both of the others. The authentic future, constituted by Da-sein’s projection of its past situation (its having-been or ‘thrownness’) onto its present possibilities for action, brings both past and present ‘into’ the future. Similarly, the authentic past, as constituted by the present interpretation (a mode of understanding, which is primarily futural [bt 336)) of Da-sein’s having-been, again contains aspects of both present and future. Finally, the authentic present, or Moment, is a constantly shifting convergence of having-been and futural projection that gives rise to Da-sein’s possibility for action, and so again encompasses elements of both past and future.
It is important to note that these three modes do not represent different relations between an existent past, present and future, as per the dimensional account. Rather, the authentic past and future are partially constitutive of the authentic present, and similarly for the other temporal modes, mutatis mutandis. Consequently, previous and subsequent ‘times’, or instants, do not bear merely external relations to one another, but are internally related since it is impossible to conceive of one without the other — i.e. without the future, there would be no past or present. Conversely, the dimensional, or inauthentic, past, present and future form an externally related series of elements that could conceivably exist independently of one another, and of the experiencing subject. It follows from this that the authentic past, present and future of Heidegger’s phenomenological account are not mutually incompatible determinations, as they are for McTaggart, but are always present together in the authentic temporalizing of temporality.11
Since all three temporal modes contain elements of each of the others, the difference between them lies in the precedence and structure of the ontologically prior notions of having-been, the Moment and futural projection, respectively. These existential modes are described as ‘equiprimordial’ (i.e. equally primitive and without conceptual priority) and constitutive of the being of Da-sein. The various relations between them give rise to the secondary determinations of past, present and future as the ways in which Da-sein has its time. For example, an event’s ‘pastness’ is constituted by its forming part of Da-sein’s (present) having-been in the context of the (future) possibilities that it discloses. Similarly, ‘futurity’ is grounded in (present) futural projection, as constrained by (past) having-been. Finally, ‘presence’ is characterised by the juxtaposition of (past) having-been and (future) possibility, creating the possibility of authentic self-awareness and agency (the Moment). Thus ‘past’, ‘present’ and ‘future’ represent different structural manifestations of having-been, the Moment and futural projection in which one of these three modes dominates the others. They, in turn, are not contradictory determinations, but rather complementary modes of being that are constitutive of authentic temporality, and therefore of the being of Da-sein.
On this view, past or future events are not external to the present, as on the dimensional account, but are already here, albeit in different manners, as befits their relative pastness or futurity. As time advances, future events become more present, and present events more past, reflecting the shifting balance between possibility and actuality in the world of Da-sein. This allows Heidegger to avoid the contradiction in McTaggart’s A-series since no event is exclusively past, present or future, but instead possesses a constantly shifting combination of all three as a result of the way in which it is apprehended, or ‘temporalized’, by Da-sein (bt 406–7). As such, all events are, were and always will be past, present and future. Of course, Da-sein cannot simply choose to temporalize an event as past, present or future any more than it can arbitrarily choose other aspects of its world that are grounded in its having-been. However, in order to understand why this is the case, we must first consider the nature of the experiencing subject, or agent.
(i) Events and agency
Instead of describing time in terms of its constituent ‘parts’ — i.e. the dimensional past, present and future — authentic temporality aims to elucidate the fundamental unity of temporality, and so address the phenomenological question of how Da-sein has time and engages with the world. As such, events do not bear relations to times, since times are not beings, but to an experiencing subject, or agent, which is essentially embedded in both time and space. For Heidegger, as for Kant, without subjectivity, there is no time, since it is the inherent temporality of the experiencing agent that gives rise to the phenomena of time and ‘the world’ (bt 365). Time is therefore neither subjective nor objective, but an a priori condition of experience (bt 419; tb viii). Furthermore, to regard the agent as being merely objectively present at some moment ‘in time’, such as the present (in McTaggart’s A-series), or at some other point in the sequence (as per the B-series), severely distorts its nature as an experiencing subject (bt 328). Similarly, the subject cannot be placed outside the temporal sequence, since it is essentially temporal, and thus ‘stretched along’ through past, present and future.
The above considerations raise two important questions for Heidegger’s conception of authentic temporality: (i) what makes the ‘objective’ phenomenon of public time possible, and (ii) how can one and the same event be characterised as predominantly future, predominantly present and predominantly past for the same agent (the Heideggerian equivalent of McTaggart’s paradox). Whilst Heidegger does not deny that all existence is grounded in some kind of pre-conceptual reality (bt 412), he is remarkably vague as to how this reality should be regarded. It is certainly not to be identified with either the phenomenal world of experience. Nor can it be the physical universe of the natural sciences, which is a purely theoretical construct that ‘levels down’ all first-person subjectivity in an attempt to create a fully objective third-person perspective,12 and thus is in no sense conceptually prior to experience. Perhaps, like Kant’s ‘noumena’ (op. cit. A235/B294), such ultimate reality is beyond the reach of Da-sein and language, although this seems to leave little room for the possibility of either genuine self-knowledge (through phenomenology) or the systematic investigation of reality (i.e. science). However, perhaps a more satisfying answer can be given in terms of Heidegger’s grounding the being of Da-sein in its ‘having-been’ (bt 328).
For Heidegger, the difference between future and past events lies in their manner of being; i.e. the way in which they are temporalized by Da-sein. Whereas future events exist as possible occurrences, and are experienced as futural projection, past events have already become actual, and so form part of Da-sein’s ‘having-been’, as characterised by the authentic past. This emphasis upon possibility explains both the asymmetry of time, and is the reason why we can change the future, but not the past, since it already part of our being. Thus it is not that events exist ‘in’ the past or future, as if they were presently happening ‘somewhere else’, but that they exist here as part of the possible or actual structure of the world. The authentic present, or Moment, on the other hand, is characterised by the possibility for resolute action, or agency, in which Da-sein embraces both past and future in order to realise its future possibilities; i.e. to make them real. In this way, authentic Da-sein makes what was merely possible, i.e. futural, actual, i.e. part of what Da-sein presently is, and therefore its having-been. In acting, Da-sein brings about a change both in the being of events (from possible to actual) and in itself, as it creates — or, in the inauthentic case, is created by — the temporal unfolding of its possibilities.13
But what about events that are not the result of any action of Da-sein, such as the rising and setting of the sun? For Heidegger, these too involve changes in the experiencing agent, since events are only ‘past’ or ‘future’ in so far as they are already (or have the capacity to become) part of the being of Da-sein (e.g. by altering the range of possibilities that are available to it). As such, ‘public’ events like the movements of the planets are constitutive of the being of all Da-sein, and thus form the basis of a framework within which public time may be measured — such as when we implicitly reference the rotation of the Earth by looking at a clock, for example (bt 71). On this view, it is not that the intrinsic or relational properties of events undergo changes, but that their manner of being changes as a result of them becoming manifest to the agent in different ways, corresponding to the actualisation of possibilities within the world of Da-sein. The paradox of McTaggart’s A-series may then be restated as the question of how is it possible for one and the same event to be both possible and actual, since on the dimensional account, these determinations are also conflicting. For Heidegger, however, it is precisely Da-sein’s capacity for agency, as exhibited in the threefold structure of Sorge, or ‘care’, that enables such changes to take place, since changes in events are merely changes in the way in which the world becomes manifest to Da-sein through its temporalizing of temporality.
V. Conclusion
Heidegger’s phenomenology of time addresses the contradiction inherent in McTaggart’s A-series in two ways. Firstly, by replacing the dimensional conception of ‘past’, ‘present’ and ‘future’ with the conceptually and ontologically prior notions of having-been, the Moment and futural projection, Heidegger argues that these three aspects, or ekstasies, of time are essentially unified, thus avoiding the appearance of a contradiction in his own account. Secondly, by showing how the dimensional or ‘vulgar’ conception of time may be derived from this account, but not the other way around [bt 426], Heidegger identifies the source of the problem as lying in the attempt to view time as an ‘objectively present’ phenomenon, independent of our own existence. By leaving out any reference to the subject and its capacity for agency — both of which are essential to Heidegger’s account — McTaggart’s ‘levels down’ the significance of authentic time into a series of homogeneous and inherently meaningless ‘locations’. This transforms time into a series of determinations against which individual objects and events are measured, thus assuming the linear characteristic of a ‘timeline’ or dimension. This ‘spatialization’ of time obscures its fundamentally perspectival nature, concealing the finitude and unity of authentic temporality, and ultimately leading to the contradiction inherent in McTaggart’s conception of the A-series.
What Heidegger’s account shows is that time can only be understood from the perspective of an essentially embodied and temporal agent that is already in the world. As soon as we try to ‘step outside’ time and consider it sub specie aeternitatis, as on the dimensional account, we necessarily lose our grip upon its fundamental structure, as constituted by the three internally related modes of temporality, or Sorge (‘care’). Externalised in this way, each moment of authentic time is transformed from a significant ‘then’ into a meaningless ‘there’, and the ‘how’ of Da-sein’s authentic having-been into the ‘what’ of mere objective presence (ct 12). For Heidegger, the contradiction in McTaggart’s A-series arises not because time or the A-series is unreal (although he agrees with McTaggart to the extent that neither has any substantial being since it is not a thing in any meaningful sense [tb 3]), but because we are asking the wrong question. On Heidegger’s view, it makes no sense to ask what time is, or whether it is real, but rather: ‘how do we have time?’ and ‘in what manner are the past, the present and the future in the world?’. In this way, the temptations that led us to view time as a dimension in the first place may be removed, and the paradox of McTaggart’s A-series no longer arises.
——————
1 A term attributable to Husserl (Beyer 2007).
2 A detailed account of this argument may be found in my essay, ‘Time, Tense and Adverbial Change’ (Wilson, 2007).
3 Four-dimensional ‘spacetime’ is a more sophisticated variant of the same basic conception.
4 Whether Aristotle himself held this view is, however, debatable (Sinclair 2007: 156).
5 See Wilson (op. cit. 2–4), and section IV below.
6 I have retained the original German terms where these are more descriptive than the common English translations.
7 Heidegger’s use of rustic, down-to-earth examples reflects his aim to reveal the nature of everyday being, and not some theoretical version of it.
8 ‘There-being’; the mode of being that we ourselves possess.
9 This echoes Kant’s First Antinomy (2002: A426/B454), where he argues that time is neither knowably infinite nor knowably finite.
10 For a more detailed discussion of McTaggart, see Wilson (ibid.).
11 Indeed, Heidegger argues that if past, present and future were conceived of as external to one another, as per the dimensional account, then it would be impossible to attach any significance to the concepts of ‘past’ and ‘future’ whatsoever, since both would be defined entirely by absence and so could not affect present experience (Gelven op. cit. 180; Dostal op. cit. 147).
12 Thomas Nagel’s (1989) ‘view from nowhere’.
13 Here, ‘action’ denotes any kind of engagement with the world. In this sense, even inaction is a form of action (cf. ‘Resolute, Da-sein is already acting.’ [bt 300]).

Bibliography
Aristotle 1984: Physics, Book IV. In J. Barnes (ed.).
Barbour, Julian 1999: The End of Time. London: Orion Books.
Beyer, Christian 2007: ‘Edmund Husserl’. The Stanford Encyclopedia of Philosophy, Spring 2007 Edition. E. N. Zalta (ed.), <http://plato.stanford.edu/archives/spr2007/entries/husserl/>.
Dostal, Robert J. 1993: ‘Time and Phenomenology in Husserl and Heidegger’. In C. Guignon (ed.), pp. 141–69.
Gelven, Martin 1989: A Commentary on Heidegger’s Being and Time, Revised Edition. DeKalb, Illionois: Northern Illinois University Press.
Guignon, Charles 1993: The Cambridge Companion to Heidegger. Cambridge: Cambridge University Press.
Kant, Immanuel 2002: Critique of Pure Reason. P. Guyer and A. W. Wood (eds.). Cambridge: Cambridge University Press.
Heidegger, Martin 1972: Time and Being (tb). J. Stambaugh (trans.). New York: Harper & Row.
—————  1982: The Basic Problems of Phenomenology (bpp). A. Hofstadter (trans.). Bloomington: Indiana University Press.
—————  1992: The Concept of Time (ct). W. McNeill (trans.). Oxford: Blackwell.
—————  1996: Being and Time (bt). J. Stambaugh (trans.). Albany: State University of New York Press.
Inwood, Michael 1997: Heidegger: A Very Short Introduction. Oxford: Oxford University Press.
Le Poidevin, Robin and Murray MacBeath 1993: The Philosophy of Time. Oxford: Oxford University Press.
Lowe, E. J. 2002: A Survey of Metaphysics. Oxford: Oxford University Press.
Leibniz, G. W., S. Clarke and I. Newton, 1956: The Leibniz-Clarke Correspondence, Together with Extracts from Newton's Principia and Opticks, H. G. Alexander (ed.), Manchester: Manchester University Press.
Lewis, David 1986: On the Plurality of Worlds. Oxford: Blackwell.
McTaggart, J. M. E. 1908: ‘The Unreality of Time’. Mind, 17, pp. 457–74.
—————  1927: The Nature of Existence, ch. 1, ‘Time’. In R. Le Poidevin and M. MacBeath (eds.), pp. 23–34.
Mellor, D. H. 1998: Real Time II. London: Routledge.
Nagel, Thomas 1989: The View From Nowhere. Oxford: Oxford University Press.
Newton, Isaac 1689: Scholium to the Definitions in Philosophiae Naturalis Principia Mathematica, Book 1. A. Motte (trans.), F. Cajori (ed.), Berkeley: University of California Press, 1934. pp. 6–12.
Sinclair, Mark 2006: Heidegger, Aristotle and the Work of Art: Poeisis in Being. London: Palgrave Macmillan.
Wilson, Keith 2007: ‘Time, Tense and Adverbial Change: Does Adverbialism about Time Solve the Problem of A-series Changes in Events?’, <http://www.keithwilson.org.uk/philosophy/essays/Entries/2007/5/29_Time,_Tense_and_Adverbial_Change.html>.

Connectionism and Representation

Tue, 29 May 2007 11:04:52 +0100

Are connectionist networks capable of representing the structure of thought?
I. Introduction
Connectionist, or parallel distributed processing (PDP), networks are typically very good at pattern recognition and responding to previously encountered stimuli. However, it is less clear that they are suited to the kind of symbol manipulation that is often taken — although not known a priori (pace Davies 1991) — to be a prerequisite for abstract thought and language. According to Fodor (1975), such abilities are only explicable in terms of a computational architecture that is based upon a symbolic ‘language of thought’ (ibid.) whose structure is substantially similar, if not identical, to that of natural and formal languages. Furthermore, Fodor (1988: 8) claims that the intimate connection between syntax (i.e. formal structure) and semantics (meaning) exhibited by such languages is essential to explaining the causal structure and systematicity of human thought. This commits Fodor to a view of consciousness in which beliefs, desires, knowledge, etc. are not just convenient approximations, or ways of talking about underlying neurological processes, but literal descriptions of the causal structure of the human mind. A vivid illustration of this can be found in the following quote, in which the fictional Sherlock Holmes recounts the means by which he arrived at his conclusion:
I instantly reconsidered my position … [this] gave rise to the suspicion that the rope was there as a bridge for something passing through the hole … The idea of a snake instantly occurred to me, and when I coupled it with my knowledge that the Doctor was furnished with a supply of the creatures from India I felt that I was probably on the right track.
(Fodor 1987: 13–14)
For Fodor, Holmes isn’t simply reconstructing his thought processes, but giving a literally accurate account in which the terms ‘suspicion’, ‘idea’, ‘knowledge’, and the feeling ‘that [he] was probably on the right track’ refer to actual states or processes that took place within his conscious mind as the relevant events took place.
Whilst a full consideration of Fodor’s systematicity argument lies outside the scope of this essay, I will challenge his claim that the language of thought is ‘the only game in town’ on the grounds that connectionist networks are also capable of representing the kind of structures necessary for thought and language. I will begin by examining the nature of the representations contained within such networks, and whether they exhibit the features that Fodor supposes are required for linguistic ability, such as compositionality and context independence. I will then examine some strengths and weaknesses of the connectionist paradigm, including the use of complex network architectures. Finally, I will conclude that connectionist networks are capable of representing the structure of thought and language, but are currently at too early a stage of development to carry out advanced cognitive processes, such as abstract reasoning. There is, however, no philosophical reason why such capabilities could not be achieved without needing to implement the sort of ‘language of thought’ that Fodor et al. describe.
II. Local and Distributed Representation
All representation must necessarily be of something. In the case of mental representations, the relationship is one of intentionality, or ‘aboutness’, and causal connection. That is, the representation must be caused — directly or indirectly — by what it represents; i.e. its content. Additionally, a representation may also resemble its content in terms of its structure or appearance. This may not be immediately obvious since the relevant information may be encoded in any number of different ways — spatially, temporally, and so on. However, as Putnam’s (1981: 1–5) Winston Churchill-drawing ant demonstrates, mere resemblance does not suffice for representation. Similarly, whilst resemblance in the form of isomorphic co-variance may be characteristic of certain classes of representation — visual images, for example — it is not necessary that a representation should resemble its content. A word, for example, may represent a complex proposition whose structure is in no way ‘contained within’ its form. Rather, it is the fact that it has content, i.e. that it is used as a symbol, that makes it representational. When it comes to representing complex statements, propositions or thoughts, however, it seems reasonable to expect the form of representation to possess a complex internal structure like that of sentences, which are built out of simpler component parts, such as words and phrases. The classical symbol processing view of the mind, hereafter abbreviated as CSP, takes such forms of composition to be essential to human cognition. Furthermore, Fodor claims that this is only explicable through CSP, since it alone is based upon a system of interrelated symbols that stand in structural relationships to one another, thus building up more complex representational forms.
Since internal structure and causal connection are taken to be necessary for representing thoughts, I will first examine whether connectionist networks exhibit either of these characteristics. The simplest kind of ‘first generation’ (Clark 2001: 68) networks contain arrays of ‘input’ and ‘output’ units — essentially very simple signal processors — that are connected together in such a way that the outputs of one set of units feed the inputs of other units. By applying a mathematical function to the weighted sum of its inputs, each unit controls the activation of processing units further along in the network, which may be ‘trained’ to recognise salient features of some external stimuli by repeated exposure to a suitable training corpus (ibid. 69). Each pattern of ‘input’ activations registers a distinct pattern of activation in the ‘output’ units, with optional layers of ‘hidden’ units contributing intermediate processing capabilities. By a suitable method of ‘back propagation’ (Clark 1990: 210), connectionist networks can be trained to detect features of an input signal that are not apparent from its surface structure; e.g. radar echoes indicating the presence of an underwater mine (Churchland op. cit. 157–162). Simple first generation networks tend to exhibit a strong degree of correlation between the presence of such features and the activation of individual units within the network, resulting in a ‘localized representation scheme’ (Rowlands 1994: 486). This is comparable to the way in which CSP represents symbols as distinct and identifiable tokens within the computational system (although there are also important differences, as described below).
Localised activation, however, is far from typical in connectionist networks, especially in more complex network architectures, which tend to distribute information across multiple processing units. This results in ‘locally distributed representations’ (ibid.), which involve a number of different physical units in the recognition of each input feature. Furthermore, in ‘fully distributed’ (ibid.) schemes, the ‘representational’ aspects of the network are spread out across its entire structure, rather than being localised to any single unit or group of units. Since individual units are no longer dedicated to processing particular stimuli, but instead participate in many diverse and physically overlapping sub-networks, it is no longer possible to simply ‘read off’ the network’s representational content by inspecting its fine-grained structure. This poses a problem for the claim that connectionist networks are representational since there is no longer any clear separation between individual representations, or between representations and the underlying computational architecture (the traditional ‘software–hardware’ distinction). Such ‘representations’ are, at best, functional states of the whole network or, at worst, a confused amalgamation of interacting and overlapping elements with no clearly defined representational content.
In analysing connectionist networks, cognitive scientists typically differentiate between their logical and physical structure. Although ‘fully distributed’ representations employ different physical parts of a network to encode what a symbol-based system might represent with a single localised token, the network can also be described as having a particular logical structure. This is comprised by the hidden structures and correlations between different regions of the network, and can be only retrieved by carrying out a detailed mathematical analysis, such as principle component analysis (PCA). This identifies not only significant clusters of network activity, but also how certain states ‘can promote or impede movement into future states’ (Clark, op. cit. 71–2), taking into account second- and third-generation networks’ sensitivity to temporal sequence. This moves even further away from the static, localised representations of CSP, towards a more dynamic, continuously changing sequence of states in which no individual element can be said to ‘represent’ a particular object, but in which the network as a whole is able to encode information relating to — and indeed caused by — the appropriate stimuli. Consequently, provided that it contains sufficient logical structure to represent the syntactic and semantic relations necessary for language, there is no reason why a connectionist network cannot represent such content in physically distributed form.1
III. Causal Structure and Compositionality
Having established that connectionist networks satisfy the conditions necessary for representation, I will now turn to the question of whether they contain the kind of structure that is required to represent human thought and language. Fodor argues against this on two counts. Firstly, connectionist representations (of sentences, for example) are not necessarily composed of representations of the relevant component parts (i.e. words and phrases), as they would be in a symbolic system. Consequently, the productive capacity of thought does not arise as a matter of necessity, or ‘psychological law’ (Fodor 1990: 184), out of the computational architecture, as with CSP, but must be explained by some other factor — something that Fodor considers to be highly implausible and lacking from the connectionist account. Secondly, the tokens within a CSP system, which Fodor takes to be straightforwardly correlated with language and psychological concepts (Holmes’s beliefs, suspicions, and so on), play an essential role in any causal description of the system’s behaviour. However, in a connectionist system, such symbols are irrelevant since its behaviour is only explicable in terms of the functioning of individual processing units. Furthermore, Fodor claims that the only way that such symbols would play a causal role is if the connectionist network were to implement a classical ‘language of thought’ architecture.2 As we will see, this argument turns upon the precise meaning of the term ‘implement’, but it also leaves open the possibility of networks that merely approximate CSP, but whose causal structure is fundamentally connectionist, as discussed in section IV.
As Smolensky (1988) has argued, the activation states of a connectionist network may also be analysed in terms of ‘tensor product representations’, which are mathematical vectors that describe constituent components of each state. Multiple tensor products may be summed using vector arithmetic to produce a composite state, e.g. Vblack + Vcat = Vblack-cat, much as symbolic representations may be combined to create composite representations, e.g. ‘a black cat’. However, unlike CSP tokens, tensor products typically exhibit a high degree of context sensitivity (Rowlands op. cit. 487). For example, the representation of a cup of coffee minus the representation of a cup would not give a representation of coffee, but that of coffee in the context of a cup (Smolensky 1990: 207). Coffee in the context of a flask, for example, might be represented quite differently by the network, which may have no single context-independent way of representing coffee, but rather a whole family of otherwise unrelated representations of coffee as it appears in different contexts (coffee beans, coffee shop, etc.) [ibid. 209].3 The use of context-neutral symbols (i.e. words) to represent content is highly characteristic of human thought and language and, Fodor claims, essential to the representational scheme that underlies it. Once again, Fodor’s objection is not that connectionist networks cannot generate context-independent symbols — they can — but that they do not do so as a matter of nomological necessity (Fodor 1990: 202).4 So, whilst advanced connectionist networks undoubtedly contain significant representational structure, it is not necessarily of the kind that facilitates abstract thought.
In order to overcome Fodor’s objection, the connectionist has to show that (i) there is some other reason or mechanism behind the emergence of complex semantic structure, and (ii) that connectionist networks possess comparable combinatorial ability to CSP (the systematicity argument). The first of these may be explained in terms of environmental selection and other systemic pressures that favour the development of context-neutral representations over context-dependent ones; e.g. for survival or reproductive success. Combined with exposure to an appropriately diverse range of stimuli (existing networks are typically trained for very specific tasks or functions), this may in itself be sufficient to explain the predominance of context-neutral representation in human thought under the connectionist model. However, other approaches are possible. Although the basic architecture of a connectionist network does not necessarily lead to context-neutral representations, certain types of connectionist networks do. For example, a network that is organised in terms of ‘roles’ and ‘fillers’ (Smolensky op. cit. 212) will exhibit similar computational abilities to CSP, especially when combined or ‘layered’ with conventional context-sensitive networks, as described below. Such an organisation mirrors the capabilities of a primitive Von Neumann machine (Copeland 2006) in which symbols are held within a small number of memory locations, or ‘registers’, and may be manipulated and combined in various ways, thus addressing the second point above regarding connectionism’s adequacy for representing and manipulating semantic structure.
IV. Architectural Concerns
The possibility that a connectionist network might be able to emulate CSP raises the question of whether such an arrangement would constitute a truly connectionist model of computation, or whether it is really a ‘language of thought’ implemented in connectionist hardware (Fodor 1988: 49). The answer to this question hinges upon whether the logical structure of such a network can be described solely in terms of its symbolic structure, or whether the underlying connectionist network merely approximates a CSP architecture. This may seem like a minor distinction, but if the resulting system cannot be explained solely in terms of its symbol processing abilities, then it cannot be said to implement a ‘language of thought’ (Smolensky op cit. 204, 210, 216). This point cuts both ways. Conventional connectionist networks cannot be explained in purely symbolic terms since the only causally efficacious entities they contain are the individual processing units, and not the symbols which are merely approximated by the network. Any description of the network that makes use of such symbols — including Smolensky’s tensor product representations (ibid. 210) — is merely a convenient approximation, or at least only one possible view, of the network’s causal structure.
Fodor takes this to be a knock-down argument against connectionism (Fodor op. cit. 203), but it may also help to explain another important aspect of human cognition, namely, the ability of humans and other animals to act upon poorly defined, vague or incomplete representations. On the connectionist view, such abilities can be explained in terms of the network’s ability to make use of ‘sub-symbolic’ elements representing subtle aspects of the network structure that are not fully articulated (or are unarticulable) in symbolic form. Sherlock Holmes, for example, might draw upon these in the form of the ‘hunches’, intuitions and tacit knowledge that helped lead him to his conclusion, but that later play no part in his dramatic reconstruction of events, which is fully symbolic. Fodor rightly claims that symbolic architectures can also employ finer levels of granularity than conventional words or concepts (Fodor 1988: 5 fn.), but this fails to address the question of why some symbols should be accessible to conscious introspection and may be described in natural language, whilst others remain seemingly elusive and unavailable to consciousness. With connectionism, this distinction falls out as a natural consequence of the architecture, much as the systematicity of natural language falls out of the ‘language of thought’ hypothesis, since only the symbols that are approximated by the system are translatable into natural language, whilst the remaining content remains hidden at the ‘sub-symbolic’ level.
A striking demonstration of connectionism’s ability to capture and make use of such ‘sub-symbolic’ structure can be found in a groundbreaking piece of research by David Chalmers (1990). In it, a form of connectionist network known as Recursive Auto-Associative Memory, or RAAM (Pollack 1988), is trained to recognise sentences worded in the passive voice with a reasonable degree of accuracy (over 80%). This is wired directly into a second ‘transformation network’, which translates the resulting sentences back into the active voice — a task for which the original RAAM network has no competency. The transformation network achieved a success rate of 65% with sentences that the RAAM network had not previously been trained to recognise, and 100% with those it had. However, since translation proceeded directly from the distributed representations within the RAAM to the output stage without passing through any intermediate stages, at no point was any internal symbolic representation or ‘language of thought’ present or required. What is crucial about this experiment is that it demonstrates the ability of connectionist networks to make use of the hidden ‘sub-symbolic’ structure within a distributed representation to perform the sort of transformations that are usually associated with CSP, but without having to implement any formal symbol processing mechanism. In other words, it gives the appearance of symbol processing without the need for actual symbols.
Fodor might reject this as a counterexample to the language of thought hypothesis on the basis that such translation may be conceived a one-stage, rather than two-stage process, and so no intermediate representation is required. However, this objection is unconvincing since what the RAAM example shows is that there is no reason in principle why cognitive processes cannot be modelled in this way, thus undermining Fodor’s claim that beliefs, desires, knowledge etc. need be explicitly represented within the network provided that the required behaviour is produced. Chalmers’ result is particularly remarkable given that connectionism is still at a relatively early stage of development and, by any account, far from being able to exhibit the kind of generalised and highly abstract processes that are characteristic of human thought (Smolensky 1990: 202). It also strongly suggests that the ‘layering’ and interconnection of multiple networks, each specialised for a particular task, may offer a much more powerful and flexible model of human cognition than conventional single-purpose networks, highlighting the need for further research in this area.
V. Conclusion
Connectionist networks are capable of encoding the sort of complex structures necessary for representing thought and language. Unlike CSP architectures, which typically employ highly localised forms of representation, these may be distributed across the entire physical structure of the network. Moreover, their logical and causal structure is only approximated, and not literally described, by the kind of symbolic representations envisaged by Fodor’s ‘language of thought’ hypothesis. Instead, the syntactic (i.e. causal) structure of a connectionist network is related to its physical structure, whereas its semantic structure (i.e. meaning) resides at a higher level of abstraction (Smolensky 204–5). Connectionist networks are therefore able to mimic the abilities of conventional CSP architectures without literally implementing a ‘language of thought’, throwing serious doubt upon Fodor’s claim that only CSP can explain human cognitive ability. On the connectionist view, the folk psychological description recounted by Sherlock Holmes in the above quotation is not a literal description of his thought processes, but a post hoc reconstruction, or approximation, of the causal processes that took place within his brain. Although such approximations necessarily fail to capture the precise causal structure of a connectionist system, they may nevertheless be sufficiently accurate for use in everyday life (although a significant amount of activity may also take place at a lower ‘sub-symbolic’ level).
This leaves open the question of whether the human mind is itself a kind of connectionist network — an issue upon which the current scientific evidence is inconclusive. To answer this question, we should look less to structural differences between connectionist and CSP architectures, and more to computational differences. These include whether the human mind implements the kind of ‘fetch’ and ‘store’ operations that are required for symbolic computation, and whether its operational data and algorithms — the ‘rules’ of the system — are inextricably intertwined, as per the connectionist approach, or distinct, as with CSP. By differentiating the algorithms from the architecture, an empirical analysis of the strengths and weaknesses of both accounts may be carried out, thus answering the a posteriori question as to the structure of the mind and its methods of representation.
——————
1 Fodor (op. cit. 40) claims that CSP can also be implemented in physically distributed system, although whether classical symbol processing algorithms can be suitably decomposed to enable the sort of ‘massive parallelism’ that is possible with connectionism is debatable.
2 Fodor’s (1988: 38–46) claim that such an arrangement would exhibit many of the benefits of connectionism (increased parallelism, graceful degradation in the event of damage to the network, tolerance of noise and partial input data) combined with the representational ability of CSP is again highly questionable due to the nature of the algorithms involved.
3 Note that similarities and interrelationships between representational states may, however, still exist without these having to be explicitly coded within the network (Clark 1993: 9 in Garson 2007: §6).
4 Although one could level the same objection against symbolic representation, since there is no reason why the symbols for cup of coffee, coffee beans and coffee house must be composite, as opposed to simple — e.g. they could instead be labelled ‘1’, ‘2’, ‘3’ and ‘4’ — it is true that symbols are naturally combinatorial in a manner that tensor products are not.

Bibliography
Chalmers, David J. 1990: ‘Syntactic Transformations on Distributed Representations’. Connection Science, 2 (1 & 2), pp. 53–62.
Churchland, Paul M. 1988: Matter and Consciousness. Cambridge, Massachusetts: Massachusetts Institute of Technology
Clark, Andy 1990: ‘Connectionism, Competence, and Explanation’. The British Journal for the Philosophy of Science, 41(2), pp. 195-222.
————— 1993: Associative Engines. Cambridge, Massachusetts: MIT Press
————— 2006: Mindware. Oxford: Oxford University Press.
Copeland, Jack B. 2006: ‘The Modern History of Computing’. The Stanford Encyclopedia of Philosophy, Summer 2006 Edition. Edward N. Zalta (ed.), <http://plato.stanford.edu/archives/sum2006/entries/computing-history/>.
Davies, Martin 1991: Concepts, Connectionism, and the Language of Thought. In W. Ramsey, S. Stich and D. Rumelhart (eds), pp. 229–57. <http://philrsss.anu.edu.au/~mdavies/papers/lot.pdf> (accessed 24/6/07)
Fodor, Jerry A. 1975: The Language of Thought. Cambridge, Massachusetts: MIT Press.
————— 1987: Psychosemantics. Cambridge, Massachusetts: MIT Press.
Fodor, Jerry A. and Zenon W. Pylyshyn 1988 (preprint): ‘Connectionism and Cognitive Architecture: A Critical Analysis’. <http://citeseer.comp.nus.edu.sg/cache/papers/cs/22408/http:zSzzSzruccs.rutgers.eduzSzpubzSzpaperszSzjaf.pdf/fodor88connectionism.pdf> (accessed 29/3/07)
Fodor, Jerry and Brian P. McLaughlin 1990: ‘Connectionism and the Problem of Systematicity: Why Smolensky’s Solution Doesn’t Work’. Cognition, 35, pp. 183–204.
Loewer, Barry and Georges Rey 1991: Meaning and Mind: Fodor and His Critics. Oxford: Blackwell.
Garson, James 2007: ‘Connectionism’. In The Stanford Encyclopedia of Philosophy, Spring 2007 Edition. Edward N. Zalta (ed.), <http://plato.stanford.edu/archives/spr2007/entries/connectionism/>.
Pollack, J. B. 1988: ‘Recursive Auto-Associative Memory: Devising Compositional Distributed Representations’. In Proceedings of the Tenth Annual Conference of the Cognitive Science Society. Montreal, Canada, pp. 33–9.
Putnam, Hilary 1982: Reason, Truth and History. Cambridge: Cambridge University Press.
Rowlands, Mark 1994: ‘Connectionism and the Language of Thought’. The British Journal for the Philosophy of Science, 45(2), pp. 485–503.
Smolensky, P. 1988: ‘On the Proper Treatment of Connectionism’. The Behavioural and Brain Sciences, 1 (1), pp. 1–74.
————— 1991: ‘Connectionism, Constituency, and the Language of Thought’. In Meaning and Mind, B. Loewer and G. Rey (eds.), pp. 201–28.

How Should de Re Modal Claims be Symbolised?

Tue, 1 May 2007 14:56:38 +0100

The distinction between de re (‘of the thing’) and de dicto (‘of the statement’) claims of necessity is central to the doctrine of essentialism—the view that individuals hold some (or all) of their properties necessarily. In this essay I will examine three different ways in which this distinction may be symbolised in formal logic and assess the pros and cons of each before reaching a verdict as to which is to be preferred, and upon what grounds.
To illustrate the difference between these two types of claim, consider the following proposition:
(1) Socrates is necessarily human.
which admits of two possible readings:
(1a) It is necessarily the case that that Socrates is human.
(1b) Socrates holds the property of being human necessarily.
As Plantinga (1978: 28) points out, the de dicto reading, (1a), may be understood as assigning necessity to the proposition
(1′) Socrates is human.
i.e. (1a) is true iff (1′) obtains in all possible worlds (cf. Kripke 1963). However, on the de re reading, (1b), the necessity relates to the way in which the property of ‘being human’ is held by an individual (Socrates), rather than to the truth of (1′), as per the de dicto reading (Plantinga, op. cit.). Assuming that the name Socrates ‘rigidly designates’ (Kripke 1980: 48) a particular individual, viz., the historical Socrates, then the two readings appear to exhibit something resembling a scope distinction concerning the concept of necessity (Sainsbury 1991: 240). This is relevant to essentialism since the readings differ in their existential commitments, and so the way in which they are formally symbolised will affect the outcome of arguments that contain one or more essentialist premisses. Conversely, if we cannot capture the formal difference between (1a) and (1b), then either our modal logic is ambiguous, or else the distinction between de re and de dicto readings is illusory, neither of which appears tenable.
The simplest symbolisation of (1) is
(2) ⃞Fa
where F is the predicate ‘____ is human’, a is Socrates, and ⃞ is the modal operator for necessity. Under the standard modal semantics (Kripke op. cit.), this states that in all possible worlds, the proposition Fa, or (1′), is true. In order for this to be the case then either (i) Socrates must exist in all possible worlds and be human in each of them, or (ii) we must be able to say of Socrates that is human with respect to worlds in which he does not exist. The former interpretation makes Socrates a necessary being, since in order for (2) to be true, he must exist in every possible world. This is incompatible with the highly plausible claim that Socrates might conceivably not have existed, and so is contingent. On this interpretation, anything that possesses one or more essential properties must itself exist necessarily. Furthermore, if existence can be construed as a property (cf. Frege 1884: 103) that is essential to all actual things, then this commits us to the highly counterintuitive and metaphysically contentious claim that every actual being must necessarily have existed, as represented by the Barcan Formula (Menzel 2007a: 2.1).
The second interpretation of (2) requires the use of positive free logic in which the usual constraints upon names (i.e. that they must refer) are relaxed, enabling true or false statements to be made about non-actual entities. Just as statements about fictional entities are true or false with respect to ‘fictional worlds’ (Walton 1978: 311)—e.g. Sherlock Holmes is a detective who smokes a pipe, solves crimes, etc.—statements about contingent entities may be evaluated with regard to the possible world in which they are uttered, or their ‘context of utterance’ (Williamson 1998: 270). In the case of Socrates, the relevant context of utterance is the actual world, which fixes the sense of the proposition. The ‘context of evaluation’ (ibid.), on the other hand, will vary according to the possible world under consideration. On this view, only worlds in which Socrates exists and is not human would render the statement false, since in worlds where there is no Socrates, this name is understood in accordance with its original context of utterance, in which Socrates is human. Hence we can say that (1′) is true even in worlds where Socrates does not exist. However, such an approach seems somewhat ad hoc, and the possibility of allowing such ‘empty names’ adds complexity to classical logic.
In order to avoid this unwanted baggage, we might instead make Fa contingent upon the existence of Socrates in a given world, i.e.
(3) ⃞(E!a ⊃ Fa)
where E! is the existence predicate (cf. Plantinga 1978: 56). This does not entail that Socrates exists necessarily, as per the first interpretation of (2), but still requires the use of empty names since in order for the conditional
(3′) E!a ⊃ Fa
to be true in all possible worlds, its antecedent and consequent must both assume a truth value (typically false) with respect to worlds in which the name a does not refer. Thus (3) requires the use of negative free logic.
One potential objection to this formulation is that it symbolises the apparently simple modal proposition, (1), as a conditional. Although this may seem surprising and unintuitive, such anomalies are by no means unprecedented,1 and it does not constitute a knock-down argument against this approach. More problematic is an argument by Williamson (2002: 233–4) purporting to show that in order for (3) to hold, (3) must itself be a necessary existent, i.e.
(3″) ⃞T[φ(a)] ⊃ ⃞E![φ(a)]
where T[] is the truth predicate and φ(a) is any statement involving a, including (3) [Efird 2006: 5]. This in turn requires that a necessarily exists, thus yielding a similar problem as per the first interpretation of (2), with the added difficulty that any statement of an individual’s essential properties must itself exist necessarily, which is dubious given that language is a contingent phenomena.2
The third and final option for symbolising de re necessity employs Prior’s (1957) modal logic, Q, which was devised as a logic of contingent entities that would modal logic of the ‘myth that whatever exists exists necessarily’ (Prior op. cit. 48). In Q, de re necessity would be symbolised as
(4) ¬♢¬Fa
where ♢ is the modal operator for possibility, whilst de dicto necessity would be symbolised as per (2). This has the advantage of retaining most of classical logic as it avoids the need for empty names. However, Q’s modal operators differ from the conventional possible world semantics in that they are not duals, i.e. Q denies that ⃞φ ≡ ¬♢¬φ and ♢φ ≡ ¬⃞¬φ. Consequently, (4) is true iff there are no possible worlds in which Socrates both exists and is not human, since unlike the ⃞ operator, ♢ does not have existential import for worlds in which Socrates does not exist.
Prior’s modal logic, however, is not without its drawbacks. With the exception of the sort of essentialist claims illustrated above, Q does not permit any necessary truths about contingent entities. This has the peculiar consequence that even apparently tautologies such as
(5) Pa & ¬Pa
are only contingently (rather than necessarily) true in cases where a denotes a contingent entity, since (5) can only be stated in worlds where a exists, and so is not true in every possible world (Menzel 2007b). Williamson (op. cit.) considers this a result of Prior confusing context of utterance with context of evaluation, but this peculiarity hardly seems worse than Williamson’s own acceptance of the necessity of existence, or the positive free logician’s view that statements may obtain in worlds where the names that they contain fail to refer.
In conclusion, each of the three ways in which the essentialist claim that an individual possesses certain necessary properties de re can be symbolised comes with its own costs and advantages. ⃞Fa entails necessary existence and, along with ⃞(E!a ⊃ Fa), requires the use of free logic, whilst ¬♢¬Fa retains classical logic but rejects the duality of the modal operators. The question as to which of these is correct thus turns upon one’s acceptance of their metaphysical implications, rather than mere notational convenience. Prior’s approach has the considerable advantage that it is able to explain the necessity of existence represented by the Barcan Formula in terms of a property that all existent things hold necessarily (i.e. de re), as opposed to it being necessary that they had to exist (de dicto) [cf. Wiggins 1976]. The resulting symbolisation also accords with Plantinga’s characterisation of de re necessity as representing the thought that ‘the object in question could not conceivably have lacked the property in question’ (Plantinga 1969: 236; original italics). Thus, despite its idiosyncrasies, Q offers perhaps the best method of symbolising essentialist claims of de re necessity without committing oneself to the implausible thesis that such individuals are necessary beings.
——————
1 Russell (1905), for example, symbolises definite descriptions as complex conjunctions.
2 The attempt to symbolise de dicto necessity as ⃞T[φ(a)] exhibits the same difficult, although this can perhaps be avoided by replacing (3″) with the conditional E![φ(a)] ⊃ T[φ(a)].
References
Efird, David 2006: ‘Necessary Existence and the Semantics of Quantified Modal Logic’ (forthcoming).
Frege, Gottlob 1884: The Foundations of Arithmetic. In The Frege Reader, M. Beaney (ed.), pp. 84–129.
Kripke, Saul 1963: ‘Semantical Considerations on Modal Logic’. In Reference and Modality, L. Linsky (ed.), pp. 63–72.
———— 1980: Naming and Necessity. Oxford: Blackwell.
Menzel, Christopher 2007a: ‘Actualism’. The Stanford Encyclopedia of Philosophy, Spring 2007 Edition. E. N. Zalta (ed.). <http://plato.stanford.edu/archives/spr2007/entries/actualism/>.
———— 2007b: ‘Prior’s Modal Logic’ (supplement). The Stanford Encyclopedia of Philosophy, Spring 2007 Edition. E. N. Zalta (ed.). <http://plato.stanford.edu/archives/spr2007/entries/actualism/Prior.html/>.
Plantinga, Alvin 1969: ‘De Re et De Dicto’. Noûs, 3(3), pp. 235–58.
———— 1978: The Nature of Necessity. Oxford: Oxford University Press.
Prior, A. N. 1957: Time and Modality. Oxford: Clarendon Press.
Russell, Bertrand 1905: ‘On Denoting’. In The Philosophy of Language, 3rd Edition, A. Martinich (ed.), pp 200–7.
Sainsbury, Mark 1991: Logical Forms. Oxford: Blackwell.
Walton, Kendall L. 1978: ‘Fearing Fictions’. In Aesthetics and the Philosophy of Art, P. Lamarque and S. H. Olsen (eds.), pp. 307–19.
Wiggins, David 1976: ‘The De Re Must’. In Truth and Meaning, G. Evans and J. McDowell (eds.), pp. 285–312.
Williamson, Timothy 1998: ‘Bare Possibilia’. Erkenntnis, 48. pp. 257–73.
———— 2002. Necessary Existents. In Logic, Thought and Language. A. O’Hear (ed.). Cambridge: Cambridge University Press.

Is the Denial of Bivalence Absurd?

Tue, 1 May 2007 14:37:01 +0100

This is an updated version of a paper I gave at a British Undergraduate Philosophy Society conference in Durham, 2007. My thanks go to my respondent, Ed Grefenstette, Professor Robert Hale and everyone else who offered feedback and criticism. Responsibility for any remaining inaccuracies and omissions are of course my own.

I. Introduction
According to the logical principle of bivalence, all utterances that express a proposition must possess one of two possible truth values: true or false (Williamson 1994: 187). However, in cases of borderline vagueness, it seems intuitive to characterise certain utterances as neither true nor false on the grounds that it is unclear which, if either, of these truth values obtains. The conception of truth that is generally taken to apply to such cases is represented by Tarski’s Convention-T, which may be symbolised as follows:
(T1) T[‘P’] ↔ P
(T2) T[‘¬P’] ↔ ¬P
where T[] signifies the metalinguistic truth predicate (Gómez-Torrente 2006). According to T1, the sentence ‘This rose is red’ is true iff the rose in question is actually red. But if the rose was a borderline case of red and pink, it would not be true that the rose was red, but neither would it be determinately false—or at least not straightforwardly so. If we accept this linguistic intuition at face value, then we must allow that some utterances can be neither true nor false but rather possess a third ‘indeterminate’ truth value, thus contravening bivalence.
II. Williamson’s Reductio
Timothy Williamson (1992) presents a reductio ad absurdum argument against such a denial of bivalence on the grounds that it entails denying the Law of Excluded Middle (LEM) in the metalanguage, i.e:
(W1) ¬(T[‘P’] ∨ T[‘¬P’])
By replacing the two disjuncts with the right-hand sides of the Tarskian biconditionals, T1 and T2, this yields:
(W2) ¬(P ∨ ¬P)
which, by De Morgan’s Law, entails:
(W3) ¬P & ¬¬P
This conclusion is clearly unacceptable to the opponent of bivalence as it would render their position incoherent. The source of this contradiction, however, may be traced back to Williamson’s use of a classical (i.e. bivalent) metalanguage. The assumption of bivalence in the metalanguage is then ‘forced through’ to the object language via Tarski’s Convention-T, with disastrous consequences. If this diagnosis is correct, then in order to refute the argument one must either adopt a non-classical metalanguage in W1, or reject T1 and T2, the basis of Tarski’s semantic conception of truth. In the remainder of this essay, I will argue that the opponent of bivalence can and should overcome the above argument by rejecting both of these assumptions.
III. Multivalence and Assertibility
Pelletier and Stainton (2003: 372) deny that LEM necessarily fails in a three-valued (i.e. true, false and indeterminate) logic provided that the negation operator only ever yields determinate truth or falsity, even in borderline cases.1 On this view, the second disjunct of W1 comes out as true where P is indeterminate, and Williamson’s argument never even gets off the ground. Furthermore, they claim, it is doubtful whether both directions of T1 and T2 hold in any such three-valued logic. Although the left-to-right direction is uncontentious, if P is indeterminate—i.e. neither true nor false—then it is not true that P, and so P → T[‘P’] is either false or indeterminate. Regardless of which option one takes, the right-to-left direction of T1 and T2 will fail in cases of borderline vagueness, and so these biconditionals do not constitute logical truths in the resulting multivalent logic. Thus, Pelletier and Stainton conclude that T1 and T2 ‘already presuppose bivalence, and so they are not the appropriate disquotational schemas for a multivalued logic’ (ibid. 374).
Williamson (op. cit. 268–9) justifies his use of Convention-T on the grounds that any predicates that fail to refer to sharply defined properties fail to say anything at all, and so we should refrain from their use as in other cases of reference failure. Given that most concepts contain at least some element of vagueness, this would effectively render the majority of utterances meaningless, thus constituting a further argument against the denial of bivalence—or else in favour of some sort of linguistic nihilism (cf. Unger 1979). However, Williamson’s claim is too strong. In order for a proposition to say something, it is arguably sufficient that there are cases in which it is determinately true or determinately false. In cases of borderline vagueness, its use might be considered incorrect or inappropriate, but it would not be strictly meaningless (we might, for example, meaningfully assert it as a form of exaggeration, or to make a point, even though the resulting utterance would be literally false). On this view, whilst the meaning of the sentence may still be fixed by its (determinate) truth conditions, its use is governed by a separate set of assertibility conditions—the conditions under which it may truthfully be asserted. In the case of vague predicates, these two sets of conditions come apart, leaving gaps in our assignment of determinate truth and falsity, and resulting in what Pelletier and Stainton (op. cit. 371) call a ‘truth-value gap’ theory. The notion of truth employed by this theory may be symbolised as:
(T1*) T[‘P’] ↔ ∆P
(T2*) T[‘¬P’] ↔ ∆¬P
where the ∆ operator represents the sentence modifier ‘It is determinately true that …’, effectively ‘collapsing’ the indeterminacy. Substituting T1* and T2* into Williamson’s original argument, we obtain the entirely plausible and non-contradictory conclusion:
(W3*) ¬∆P & ¬∆¬P
or ‘it is not the case that determinately P, nor is the case that determinately not-P’—precisely what we would expect an opponent of bivalence to say in borderline cases of vagueness (cf. Williamson 1994: 194).
IV. Higher-Order Vagueness
The use of the determinately operator, however, introduces another problem: at what point does a proposition become determinately true or determinately false? Although we can cite paradigm cases of determinate truth and falsity (e.g. ‘snow is white’, ‘the moon is made of green cheese’), it is unclear exactly where the boundaries lie. In other words, determinate truth is itself a vague concept, thus giving rise to the problem of ‘higher-order vagueness’ (Williamson 2003: 8). As Williamson points out, any analysis of vagueness that cannot account for higher-order vagueness is implausible, since we cannot deny that the concept of vagueness itself permits of borderline cases (ibid.). Furthermore, we must be able to assign some particular meaning to the concept of determinacy that is distinct from our original unqualified conception of truth (Williamson 1994: 194).
One possible solution to this problem, based upon a proposal by McGee and McLaughlin (1995: 229–30), is to explain higher-order vagueness in terms of higher-order instances of the (non-bivalent) metalanguage. On this account, determinacy is a metalinguistic concept, comparable to T[], by which the semantics of the object language are defined. First-order vagueness is governed by the constraints upon the object language along with the allowable models that assign truth and falsity to each of its sentences. Since the object language is vague, there will be many such models, each of which assigns different truth values to the vague sentences in the language. Determinate truth and falsity are just those cases where all of the models agree upon a sentence’s truth value (i.e. the ∆ operator acts as a sort of quantifier over allowable models). Second-order vagueness is defined in terms of allowable models of the metalanguage, with a second-order metalanguage (the metametalanguage) governing the formal semantics of the first-order constraints, including the concept of determinate truth. Since both truth and determinate truth inherit the vagueness of terms in the object language, this yields a series of modified biconditionals including:
(T1**) ∆T[‘P’] ↔ ∆P (ibid.)
each of which define the (vague) semantics of the previous level’s constraints, and whose vagueness is itself resolved at the subsequent level. This pattern may be repeated indefinitely to cope with any number of orders of vagueness without fear of contradiction—or as McGee and McLaughlin put it, ‘[a]s we ascend the hierarchy of metalanguages, we find vagueness all the way up’ (ibid. 230).
V. Conclusion
Williamson’s reductio represents a forceful attack on the denial of bivalence. However, its assumption of a classically bivalent metalanguage, along with the use of Tarski’s Convention-T, are clearly question begging. When combined with an appropriate three-valued logic plus a modified conception of truth that takes both higher-order vagueness and determinacy into account, the alleged absurdity gives way to mere complexity. Thus we can conclude that whilst it is not absurd to deny bivalence, such a move loses much of the simplicity and elegance of classical logic, as well as requiring serious revision to the conception of truth represented by the Tarskian biconditionals. Whether this in itself constitutes an argument against bivalence, as Williamson (1992: 279–80) claims, remains open to question. Although there are obvious advantages to retaining the classical approach in cases where vagueness is not an issue or can safely be ignored, the ‘truth value-gap’ theory is arguably a more accurate reflection of our actual linguistic practices as it avoids the need for implausible assumptions about the precise nature of either meaning (Williamson 1992) or interpretation (Fine 1975), both of which remain irreducibly vague on the non-bivalent account.
——————
1 They call this ‘exclusion negation’ on the basis that only one of the three possible truth values (determinate truth) is being denied or excluded.

References
Fine, Kit 1975: ‘Vagueness, Truth and Logic’. In R. Keefe and P. Smith (eds.), pp. 119–50.
Gómez-Torrente, Mario 2006: ‘Alfred Tarski’. The Stanford Encyclopedia of Philosophy (Winter 2006 Edition), Edward N. Zalta (ed.). <http://plato.stanford.edu/archives/win2006/entries/tarski/>.
McGee, Vann and Brian P. McLaughlin 1995: ‘Distinctions Without a Difference’. Southern Journal of Philosophy, 32 (supplemental), pp. 203–51.
Pelletier, Jeffrey and Robert J. Stainton 2003: ‘On “The Denial of Bivalence is Absurd”’. Australasian Journal of Philosophy, 81 (3), pp. 369–82.
Unger, Peter 1979: ‘There Are No Ordinary Things’. Synthese, 4. pp. 117-54
Williamson, Timothy 1992: ‘Vagueness and Ignorance’. In R. Keefe and P. Smith (eds.), pp. 264–80.
————— 1994: Vagueness. London: Routledge.
————— 2003: ‘Vagueness in Reality’. In M. Loux and D. Zimmerman (eds.). <http://www.philosophy.ox.ac.uk/faculty/members/docs/handbook.pdf> (accessed 14/3/07).