Varieties of the Manifest

Mark Johnston argues by reductio that water is not H₂O:

(1) Water = H₂O

(2) If water = H₂O then ice = H₂O and water vapor = H₂O

(3) Water vapor = ice = water

On this basis Johnston concludes:

I. Denying (1) is the only way to avoid the absurdity of (3).

II. Denying (1) implies that the term ‘water’ refers to a manifest kind, “whose instances we identify and re-identify on the basis of their manifest properties” (565). Water is constituted by H₂O, as a statue is constituted by chunks of matter.

III. I. & II. (denying identity & accepting constitution) cancel the semantical difference between ‘pain’ and ‘water’ because:

1. The identity with the physical base is false in both.

2. A certain constitution claim is a priori in both.

I question I-III. There are two senses of the manifest, weak and strong. III.1., I argue, presupposes a notion of a strongly manifest kind (SMK). However the puzzle supports II only if water is taken to be a weakly manifest kind (WMK). Moreover, even accepting that water is a SMK, instances of water could be ice, so claim I is false: there must be another way to avoid the absurdity of (3). Most importantly, III.2 is false under both notions: taking water to be a manifest kind, weakly or strongly, does not cancel the semantic asymmetry between it and pain or red. Even in the domain of the manifest – not all general terms are on a par.

1. Two senses of the manifest

A kind is strongly manifest if we identify its instances only on the basis of manifest properties. The term ‘water’ refers to a SMK if and only if it refers to the stuff on Twin Earth (TE). The manifest properties are necessary and sufficient for identification of this kind. It feels like water? It looks like water? It tastes like water? It is not poisonous? So it is water.

A kind is weakly manifest if its manifest properties are necessary for identification of its instances. Pace Kripke, we would not call a pink solid ‘water’ even if it were H20. But the manifest properties are not sufficient: we refuse to apply our term ‘water’ to the stuff on TE when experts tell us about the difference in inner structure. Though water is not identical to H₂O if it is WMK, it is still necessarily constituted by it. The semantics of ‘water’ is dependent on the semantics of a natural kind term (NKT), a` la Putnam.

Does Johnston intend water to be a SMK or a WMK? If water is only a WMK there can be fool’s water, namely twater. Kripke’s intuition is exactly that fool’s pain is impossible (“what seems like pain is pain”), but fool’s water is possible (on TE). Johnston’s III.1 purports to cancel this asymmetry between water and pain, so it presupposes that ‘water’ refers to the stuff on TE. So water must be SMK. However, the claim that no instance of water could be ice (I’ll refer to it as ‘Johnston’s intuition’) is neither necessary nor sufficient for the claim that water is a SMK.

2. Johnston’s intuition is not sufficient for the claim that water is a SMK

One can agree with Johnston that no instance of water could be ice, but hold that twater on TE is not water, and tice on TE is not ice, since they are constituted by XYZ. Accepting Johnston’s intuition is not sufficient for denying that water is necessarily constituted by H₂O. That water is a SMK doesn’t follow from the premise that no instance of ice could be water.

Of course, the objector might reply, it doesn’t follow logically. But it seems rather plausible that it follows by something like inference to the best explanation. After all, the move from identity to constitution is supposed to be a move from necessary to contingent connections. Constituting matter is relevant to identity of particulars, but not to general terms. What’s the motivation to allow two different general terms, ‘water’ and ‘twater’, which denote qualitatively identical kinds, once we give up the identity of water and H₂O?

I disagree. The controversy is semantical: how do we use the term ‘water’. So the best reply should maximize the intuitions of the normal speakers. That water is necessarily constituted by H₂O allows us consistently to accept both Johnston’s intuition and the TE intuition that we would not call ‘water’ the stuff on TE. So the best explanation may be that water is a WMK. Johnston’s intuition is not sufficient for the claim that water is a SMK.

3. Johnston’s intuition is not necessary for the claim that water is SMK

One can reject Kripke-Putnam’s natural kind semantics but hold that instances of water could be ice. From the fact that water is not identical to H₂O it does not follow that instances of it could not be ice.

Johnston disagrees: “one thing that is really implausible is to identify some liquid with the solid ice it yields … . For then we should have no choice but to identify each of these with a quantity of the underlying chemical kind (ibid, 572).” Again, he may have in mind something like an inference to the best explanation: if you already dismiss inner constitution as irrelevant to fixing the reference of water, why take it as relevant for identifying instances of water with instances of ice?

But I’ll argue that identifying some water with “the solid ice it yields” is not due to a common inner constitution, but due to the manifest lawlike connections between water and ice. So it is reasonable to deny that water is H₂O, while still accepting that an instance of it could be ice. This would be to think of water and ice as we think about changes in animals and plants. From the facts that

(2’) Young are persons; and Older are persons;

there is no way to deduce the absurdity of

(3’) Young are older.

This is because the ‘are’ in (2’) is predicative. Young and Old are phase concepts. Similarly, ‘water’ has an inclusive use which refers to that stuff which we drink, find in lakes and in the freezer, etc. Even one who has no idea whatsoever about chemistry, I claim, understands this inclusive use. Such an ignoramus would take ice and water to be states of the same strongly manifest stuff, not kinds of substances themselves. So Johnston’s first claim is inconclusive: there is another way to avoid the absurdity of (3), by showing that it is not implied by (2), since the ‘is’ in (2) is predicative, as is the ‘are’ in (2’).

Of course, where I see changes of substance Johnston would see destruction of it (the water no longer exists in the freezer). I don’t have a decisive argument for the view that ice, snow etc. are states of substances rather than kinds of substances. My only point is that each alternative, mine no less than Johnston’s, is equally compatible with water being SMK.

Is ‘water’ a NKT, WMK or SMK? I’ll argue that decision is not urgent, since even as a SMK term there is an asymmetry between ‘water’ and other general terms like ‘red’ or ‘pain’.

4. Why III.2 is false

Water, pain or red have at the appropriate basic physical level some physical base that accounts for their manifest causal powers, and by definition of constitution, (C) is true:

(C) The constituting matter of a thing is what accounts for its characteristic causal powers. (582).

So (C) applies a priori to ‘red’ or ‘pain’ as it does to ‘water’.

As against this, I’ll argue that (C) is a priori only when it applies to ‘water’, not to ‘red’ or ‘pain’, because only water falls under the general category of ‘material stuff’ – in virtue of its manifest properties.

How do the manifest properties of a thing provide for its conception as material kind? Here is a beginning of an answer. I suspect that my gold ring is a forgery. Someday, I know, it will lose its splendor. What explains my belief? Nothing about abnormal conditions of perception will do. This belief must be explained by appeal to the inner constitution of gold, what makes it part of the semantics of gold. Contrast the case of red. If I suspect that something is not red though it looks that way, I’ll check either my vision or the conditions (lights, glass, and reflection).

I’ll develop this difference in order to show that the category of the material is itself manifest, so (C) applies a priori only to things that fall under it, namely, water, not red or pain. Someone who does not have any idea of the atomic theory of matter, who thinks that matter is continuous, would still classify ‘water’ as material, but not ‘red’ or ‘pain’.