RATIO, Vol. XVII no 1 (March 04)
LOAR’S DEFENCE OF PHYSICALISM
Brian Loar believes he has refuted all those antiphysicalist arguments that take as their point of departure observations about what is or isn’t conceivable. I argue that there remains an important, popular and plausible-looking form of conceivability argument that Loar has entirely overlooked. Though he may not have realized it, Saul Kripke presents, or comes close to presenting, two fundamentally different forms of conceivability argument. I distinguish the two arguments and point out that while Loar has succeeded in refuting one of Kripke’s arguments he has not refuted the other. Loar is mistaken: physicalism still faces an apparently insurmountable conceptual obstacle.
1. Antiphysicalist arguments and intuitions
In his paper Phenomenal States, Brian Loar makes the following claim:
Antiphysicalist arguments and intuitions take off from a sound intuition about concepts. Phenomenal concepts are conceptually irreducible in this sense: they neither imply, nor are implied by, physical-functional concepts. . . The antiphysicalist takes this a good deal further, to the conclusion that phenomenal qualities are themselves irreducible, are not physical-functional properties.[i]
According to Loar, it is in taking this second step that the antiphysicalist goes wrong: ‘We have to distinguish between concepts and properties’.[ii] The independence of our concepts of, say, pain and C-fibre firing – the fact that there is nothing conceptually problematic about the suggestion that pain might exist without C-fibre firing and vice verse – does not entail that these two concepts do not pick out one and the same property.
Of course, the independence of a pair concepts is not usually taken to entail that the corresponding properties are non-identical. After all, the independent concepts heat and molecular motion pick out the same property, as do alcohol and CH3CH2OH. Yet antiphysicalists believe the independence of the concepts pain and C-fibre firing entails that pain cannot be identical to C-fibre firing. Why so? According to Loar, because the concepts pain and C-fibre firing pick out the properties they do by way of essential features of those properties. On Loar’s understanding, at least implicit in the anti-physicalist’s reasoning here is the kind of argument that is explicitly set out in Kripke’s Naming and Necessity. That argument runs as follows.
2. The Kripkean argument
The concepts heat and molecular motion pick out the same property, yet they are independent concepts. What explains this independence, and thus the fact that heat = molecular motion is not an a priori truth, is the fact that the concept heat picks out the property in question by way of what Loar calls a contingent mode of presentation. That is to say, heat picks out the property it does by way of a merely contingent feature of that property. We think of heat as that property that feels like this, and of course it is contingent that heat feels to us the way it does: it might have felt differently. It is because we can’t know a priori the contingent fact that molecular motion feels like this that we can’t know a priori that molecular motion is what heat essentially is.
In the case of pain = C-fibre firing, on the other hand, each of the concepts involved picks out the property it does by way of a non-contingent mode of presentation. In particular, while we think of pain as the property that feels like this, feeling like this is an essential property of pain. But then we cannot similarly explain why, if pain = C-fibre firing, this isn’t a priori. Kripke concludes that pain is not identical with C-fibre firing. If these properties were identical, the identity would be knowable a priori. It isn’t.
3. The implicit assumption
Loar points out, correctly, that Kripke’s argument, as expounded by Loar, relies on the following implicit assumption:
The only way to account for the a posteriori status of a true property identity is this: one of the terms expresses a contingent mode of presentation.[iii]
Call this principle (P). It is this principle that Loar attacks. Loar agrees that one way in which one might explain the a posteriori character of a property identity would be to point out that one or both concepts involve a contingent mode of presentation; however, Loar denies that this is the only possible explanation. Thus, while Loar agrees with Kripke that the concepts pain and C-fibre firing both pick out the properties they do via non-contingent modes of presentation, he denies that this, in combination with the fact that pain = C-fibre firing is not a priori, entails that pain is not identical with C-fibre firing. Loar explains as follows.
[A] phenomenal concept rigidly designates the property it picks out. But then it rigidly designates the same property that some theoretical physical concept rigidly designates. This could seem problematic, for if a concept rigidly designates a property not via a contingent mode of presentation, must that concept not capture the essence of the designated property? And if two concepts capture the essence of the same property, must we not be able to know this a priori? These are equivocating uses of ‘capture the essence of’. On one use, it expresses a referential notion that comes to no more than ‘directly rigidly designate’. On the other, it means something like ‘be conceptually interderivable with some theoretical predicate that reveals the internal structure of’ the designated property. But the first does not imply the second. What is correct in the observation about rigid designation has no tendency to imply that the two concepts must be a priori interderivable.[iv]
Loar has sketched out in more detail how concepts that ‘capture the essence of’ the same property may nevertheless fail to be conceptually convertible; he suggests[v] that phenomenal concepts like pain and theoretical physical concepts like C-fibre firing are realized in very different parts of the brain. Phenomenal concepts get their distinctiveness from their close connections with the experiential parts of the brain. Theoretical concepts like that of C-fibre firing, on the other hand, have their origins in other, quite different concept-forming faculties. Thus the two concepts are unlikely to be conceptually convertible anyway, even if they do pick out the same property. So it is plausibly just a psychological fact about us that the concepts pain and C-fibre firing are not conceptually convertible. Pace Kripke, no deep metaphysical conclusions about the nature of pain can legitimately be drawn from the existence of this psychological fact. So there is, after all, no conceptual obstacle to our accepting physicalism.
4. A counterexample to Principle (P)
Loar is surely right to deny (P). Indeed, it is not difficult to think of counter-examples (though Loar does not himself provide any).
Take, for example, the property of being a tiger. On certain fairly plausible assumptions about the essentiality of both some unique genetic feature and some unique piece of evolutionary history to the species tiger, one may conceive of the property of being a tiger as:
Property A: the property of belonging to the animal species with genetic feature G
(where G is a genetic feature unique and essential to the tiger), or as:
Property B: the property of belonging to the animal species with evolutionary history E
(where E is a piece of evolutionary history unique and essential to the tiger). These conceptions pick out the very same property. They do so by way of essential features. Yet these two conceptions are not a priori interderivable. It would take an empirical investigation to establish that Property A = Property B. As principle (P) denies this, (P) is false.
The key point to recognise here is that while a conception may pick out a property via an essential feature, it need not be that property’s only essential feature; nor need all that property’s essential features be conceptually interderivable.
5. The no-fool’s-pain argument
I accept both that Kripke’s argument as presented by Loar rests on (P) and that (P) is mistaken. However, a quite different argument is also suggested by what Kripke has to say about physicalism. As I explain below, while this alternative argument also turns on considerations about conceivability, it does not depend upon principle (P). Indeed, the argument is immune to Loar’s criticism. I call this alternative argument the no-fool’s-pain argument.
First, note there is no conceptual obstacle to supposing that, though it seems to one that one is clearly in pain, ones C-fibres are not firing. Whether or not this is a genuine metaphysical possibility, it is at least not ruled out a priori.
Now consider the identity theory that pain is necessarily identical with (say) C-fibre firing. This theory entails that necessarily: if someone’s C-fibres are not firing, then they are not in pain.
But then the identity theory plus the conceptual coherence claim together commit the identity theorist to the conceptual possibility of fool’s pain. If there is no conceptual obstacle to supposing that, though it seems to one that one is in pain, ones C-fibres are not firing, then, if one supposes that pain is identical with C-fibre firing, there is no conceptual obstacle to supposing that, though it seems to one that one is in pain, one isn’t really in pain (i.e. because ones C-fibres are not firing). To accept the identity theory is, in effect, to allow that fool’s pain should at least be a conceptual possibility. But fool’s pain is not a conceptual possibility. The concept of pain itself demands that if it seems to one that one is in pain, then one is. It follows that the identity theory is false.
6. No-fool’s-X arguments
This is, of course, a familiar form of argument. It runs as follows:
(i) Something that appears to be X but not Y is at least a conceptual possibility
(ii) If X is identical with Y, then Y is a necessary condition of X
(iii) Therefore, if X is identical with Y, in conceiving of something that appears to be X but isn’t Y one is conceiving of fool’s X (i.e. something that appears to be X but isn’t because a condition necessary for X is unfulfilled)
(iv) But fool’s X is not a conceptual possibility
(v) Therefore X is not identical with Y
In fact one finds arguments of this form used against a whole raft of identity claims.
Take, for example, the claim that colour properties are identical with certain microstructural properties (such as the property of reflecting light of such-and-such a wavelength). In this case, too, it seems that our conception of one of the two properties alleged to be identical won’t allow for the conceptual possibility of the kind of gap between appearance and reality required if the property identity is to hold.
To explain: ‘fool’s gold’ – stuff that has the appearance of gold but isn’t – is a conceptual possibility. Our concept of gold allows for the possibility of gold having a ‘real essence’: a deep microstructural property possession of which is necessary if something is to qualify as a sample of gold. We can envisage gold-like stuff that, because it lacks whatever is the essential microstructural property in question, isn’t really gold. Because ‘fool’s gold’ is a conceptual possibility, so it is conceptually possible for it to turn out that the property of being gold and the property of having such-and-such a deep microstructural constitution are identical.
By contrast, it seems we can know a priori that colours do not possess such microstructural ‘real essences’. We can know this precisely because ‘fool’s red’ is a conceptual impossibility. It is a conceptual truth that (roughly) if something typically looks red under standard conditions, then it is red. But this entails that the property of being red cannot turn out to be identical with some microstructural property. If such an identity did hold, then, because possession of that property would be a necessary condition of something’s being red, and because we can conceive of something typically looking red without that property, so fool’s red would be a conceptual possibility.[vi]
The no-fool’s-pain argument is analogous to the above. It is a conceptual truth that if it seems to one that one is in pain, then one is in pain. The concept of pain doesn’t allow for appearance and reality to come apart in that way. But this, in conjunction with the claim that the identity of pain with some physical or physical-functional property would be necessary, entails that pain cannot turn out to be identical with any such property – in particular, it entails that pain cannot turn out to have a microstructural ‘real essence’ in the way that gold does. (Notice, by-the-way, that other identity claims are similarly blocked by the observation that ‘fool’s sweet’ and ‘fool’s jaundice’ are conceptual impossibilities.)
Loar has done nothing to discredit this popular form of argument as applied either to the theory that red is identical with a wavelength of light or to the theory that pain is identical with some physical or physical-functional feature of the nervous system. For in neither case does the argument rely on principle (P).
7. Is the no-fool’s-pain argument Kripke’s?
Kripke does offer the argument that Loar attributes to him. However, the three premises required to run the no-fool’s-pain argument – (i) if pain is C-fibre firing then C-fibre firing is a necessary condition of pain, (ii) pain without C-fibre firing is a conceptual possibility, and (iii) fool’s pain is a conceptual impossibility – are also all supplied by Kripke. Although the no-fool’s-pain argument is never explicitly formulated, the premises and conclusion are all there in Naming and Necessity.
Indeed, if the argument that Loar attributes to Kripke is all that Kripke has to offer, one wonders why Kripke goes to the bother of mentioning that there is, as he puts it, no such thing as a ‘qualitatively identical counterpart’ to pain that isn’t pain, i.e. no such thing as fool’s pain. For the fact that fool’s pain is a conceptual impossibility is actually irrelevant to the argument Loar refutes. All that argument requires is that pain be picked out via a non-contingent mode of presentation: namely, that it feels like this. It is not required that this feature be sufficient for pain. Yet the observation that ‘the notion of an epistemic situation qualitatively identical to one in which the observer had a sensation S simply is one in which the observer had that sensation’ is clearly considered by Kripke to be one of his key premises.[vii]
It seems not unreasonable, then, to suppose that the argument Loar refutes is not the only argument Kripke has to offer, and that the no-fool’s-pain argument may also be one that Kripke has in mind.
According to Loar, all antiphysicalist arguments and intuitions take as their starting point the (correct) observation that phenomenal concepts like pain and physical-functional concepts such as C-fibre firing are not interderivable. These arguments and intutions all ultimately rest upon principle (P). Loar points out, correctly, that, while (P) may have a certain superficial appeal (thus explaining our antiphysicalist intuitions), it is in fact false. Thus there is, according to Loar, no conceptual obstacle to our accepting physicalism.
However, as should now be clear, there exists another, powerful antiphysicalist argument that has an entirely different starting point. The argument appeals to a quite different observation: that our concept of pain exhaustively captures the essence of that property (i.e. that nothing more is required for pain than that it feel like this). Thus it is conceptually ruled out that pain might turn out to have a hidden, physical-functional ‘real essence’. Loar has not dealt with this antiphysicalist argument. Crucially, the alternative argument does not rely on principle (P).
What Loar tries to explain is why, if pain is identical with C-fibre firing, the suggestion that pain might not be C-fibre firing isn’t conceptually incoherent. What Loar goes no way towards explaining is why, if pain is C-fibre firing, fool’s pain is conceptually incoherent. Pace Loar, there remains a very considerable conceptual obstacle to our accepting physicalism, an obstacle that Loar has not explained away.
Kripke does offer the argument Loar attributes to him. However, the resources required to construct the no-fool’s-pain argument are also supplied by Kripke. Perhaps Kripke does not himself realize that his writings present, or come close to presenting, (at least) two fundamentally different antiphysicalist arguments. By showing that principle (P) is false, Loar has refuted one of these arguments. He has not refuted the other.
Heythrop College, University of London,
[i] Brian Loar, ‘Phenomenal States’ in The Nature of Consciousness, (eds.) Ned Block, Owen Flanagan and Guven Guzeldere (London: MIT Press, 1997), p. 597.
[ii] Loar, ‘Phenomenal States’, p. 597.
[iii] Loar, ‘Phenomenal States’, p. 600.
[iv] Loar, ‘Phenomenal States’, p. 603.
[v] Lecture to the Sensation and Consciousness conference of the University of London School of Advanced Study Philosophy Programme at Senate House, University of London, Friday 3rd December, 1999.
[vi] Colin McGinn presents this argument in Colin McGinn, The Subjective View (Oxford: The Clarendon Press, 1983), pp.11-12.
[vii] Saul Kripke, Naming and Necessity (Oxford: Blackwell,1980), p.152.