The other week, we read an interesting and intriguing paper by Joe Levine and Kelly Trogdon on “The Modal Status of Materialism” (Phil.Studies), arguing that the contingency of physicalism clashes with the necessity of realization and physicalist supervenience. The upshot isn’t that such a clash can be avoided, but that we shouldn’t think of physicalism as contingent. I was a bit perplexed by the argument. For one thing, I think it’s fair to say it’s a bit on the convoluted side (the reconstruction below, if correct, suggests so). For another, some premises struck me as unobvious at best. I wasn’t the only one to be perplexed, but some of us thought the argument was fine and I failed to put my finger more precisely on what worried me. So I’ll try again. Since I still don’t really have any good grasp of the argument, I’ll limit myself to trying to reconstruct it, and outline some of my puzzlement here and there.

It starts with an explanation (including some motivations) for the two claims that are supposed to clash.

WHY IS PHYSICALISM CONTINGENT?
The authors give the following explanation:
(1)    some physical facts are brute (=x is brute if “there is no illuminating explanation for why it is x”)—they cite some physical magnitudes such as the speed of light as brute.
(2)    brute metaphysical necessities don’t make sense: if something is brute, it can be actual but it “could have been otherwise” (if it can’t have been otherwise, it can’t be brute).
(3)    Unless they are grounded in logic (logical truths are necessary but not brute, since there is an explanation for their necessity), there can’t be brute metaphysical necessities.
(4)    Hence, physicalism is contingent: if it weren’t, it’d be committed to unacceptable brute metaphysical necessities.

There’s a couple of things that already seem strange to me here:
i) I just don’t understand (2). For one thing, if there really are brute facts, it seems to me there can be brute metaphysical necessities. For instance,

There are some possible worlds.

That’s necessarily true since it’s true at every possible world (though not true of any possible world: it’s a bit like atemporal truths that are true at every time, though not of every time—to borrow an idea from Kit Fine). But it doesn’t seem to be a logical truth—certainly not one the necessity of which can be explained in the usual way in terms of its syntax via a truth-table. And it seems to be brute, or at least as deserving of the title as anything else. For another, I don’t really have the intuition that actual brute facts are ok, but necessary ones aren’t. Brute facts strike me as puzzling and undesirable in one’s metaphysics, but I don’t really see how their necessity should add to their puzzlingness and undesirability.

ii) more importantly, I don’t understand how (4) is supposed to follow from (1) and (2). The traditional way of spelling out physicalism is in terms of physicalist supervenience: the physical facts determine all the facts so that any difference in the non-physical facts comes with a difference in the physical facts (that’s one way to phrase it). No matter how you phrase it, this is supposed to capture what’s the crucial claim in common between different kinds of physicalists. Physicalists also usually agree that there is some explanation of some kind for why the supervenience thesis is true, but they largely disagree about what the form of that explanation will be (or are agnostic).

But now, suppose physicalist supervenience isn’t contingent, but necessary. Since most physicalists think there is a further explanation for why that claim is true, physicalist supervenience can be necessary, but there’s no reason to think it’ll be brute. So, even granting (2), it’s hard to see how this entails (4). It’s also hard to see how physicalist supervenience + (1) and (2) would entail (4). There may be brute physical facts (1), but (even if physicalism is necessary) no physicalist would be tempted, it seems to me, to think that all physical facts are necessary. In other words, physicalism is not really a thesis about physical facts as such (certainly not about any particular physical fact). It’s a claim about the relation between physical facts and non-physical facts—so (1) in no way suggests that such a relation is brute.

iii) what’s also strange about L&T’s reconstruction of the motivation for the contingency of physicalism is that it’s not the motivation its defenders appeal to (L&T cite Lewis and Jackson). The explicit motivation (amongst others) seems to be one the authors mention at the end of their paper: dualism is not incoherent, so there must be some possible worlds in which it’s true. Also, a certain kind of epistemic humility about possible worlds may be at play: we don’t know all the possible worlds there are, and some may be quite different from the actual world. Finally, I suspect, there may be strategic motivations too: the contingency of physicalism, which, for Lewis and Jackson, comes with some restrictions on the sets of worlds in which global physicalist supervenience is supposed to hold, provides more room to manoeuvre against various anti-physicalist arguments, including the conceivability and possibility of zombies (if physicalist supervenience is true only in those worlds that are minimal physical duplicates of our world—i.e., same physical properties, same distribution (Lewis’83 and Jackson’ 98 differ a tiny bit here), same laws, and no alien properties—then zombies can be conceivable and possible, provided zombie-worlds aren’t minimal physical duplicates and/or have some alien property: thus, when you claim to be able to conceive of zombies, can you really rule out that you’re not conceiving of a world that has a tiny physical difference or some tiny alien property in it?).

Now to the second claim that’s supposed to be clashing with the first.

REALIZATION, SUPERVENIENCE, AND NECESSITY
L&T spell out the claims and its motivation along the following lines:
(5)    for physicalists, physical facts realize mental facts.
(6)    Realization has to be more than just lawful correlation, because dualists like Chalmers accept lawful correlations between physical facts and mental facts.
(7)    Realizers must necessitate their realizees.

So far, so good. The only worry here is dialectical. Physicalists who take physicalism to be contingent formulate physicalist supervenience as a global supervenience thesis about worlds (and not the properties of individuals in those worlds, as in strong supervenience), where the sets of worlds where such a thesis is supposed to be true is restricted to nearby possible worlds that are minimal physical duplicates of our world, rather than being “about all possible worlds”, as L&T suggest.

In contrast, the authors use strong (unrestricted) supervenience: “if my pain is realized by my brain state, then in no possible world can there be a creature in a type-identical brain state who isn’t in pain”.

The worry (a question rather) is this. First, to see whether there really is a clash between the contingency of physicalism and physicalist supervenience, it seems it would help to consider what contingent physicalists think about supervenience—namely, use their version of physicalist supervenience. Second, those physicalists who use strong supervenience may not accept (or care much about) the contingency of physicalism—I don’t know whether or not they do. L&T cite Shoemaker and Melnyk and I’m ignorant of the details of their respective views, sadly. But it’d better not be the case that, in arguing that there’s a clash between the contingency of physicalism and physicalist supervenience, one pits one kind of physicalist against another. After all, it isn’t news that there must be a clash between (a) some of the claims contingent physicalists hold and (b) some of the claim non-contingent physicalists hold (if there are such actual or possible people). And the authors do nothing to rule out this possible but uncharitable interpretation of their argument.

In this respect, note that if you have a version of physicalist supervenience restricted to a set of nearby worlds (minimal physical duplicates with no alien properties), then, by the authors’ own lights, such supervenience thesis could be brute, since it doesn’t hold across every possible world. If so, the argument below has little traction: if you restrict (8) below, you don’t need to worry about (9), and so you don’t need (10), (11), …

Now, here’s how these two claims are supposed to clash:

THE ARGUMENT
Seems to go something like this:
(8)    Realizers necessitate their realizees.
(9)    But metaphysical necessitation cannot be brute.
(10) however, if the relevant necessity is logical, it’s not brute (because the necessary truth of logical truths can be explained).

Which leads them to introduce what they call the ‘identity doctrine’:
(11) if mental state m is realized by physical state p, p does not logically entail m, but if there is another description f of m, such that m = f, which logically follows from p, then it’s not a brute fact that p necessitates m.
(12) f must be logically entailed by the physiological facts and the laws and the only candidate description meeting these constraints is a functional characterisation specifying a causal role.

And then, the final step, supposed to show that realization clashes with the contingency of physicalism:
(13) but if m = f (mental properties are functional/causal roles), it’s “hard to see how they could be basic in any possible world” (in the sense, according to L&T, that the dualist claims mental properties to be basic, i.e., not realized).
(14) so if physicalism is the view that mental properties are realized (not basic), then physicalism is, if true, necessarily so.

That’s the argument, and a couple of things puzzle me.
i) Obviously, (13) is crucial, but why think it’s true? They consider a couple of ways in which it could be false, which they reject (see below), but no positive reason is offered.

ii) also, the funny thing is that, if, as a physicalist, you think mental states are functional/causal roles, then (13), provided it does the work it’s supposed to, does it alone. That is, given (13) it’s the functionalist account of mental states that clashes with the contingency of physicalism. The necessity of physical supervenience or of realization doesn’t do any (direct) work here. In particular, if (13) is true, it’s not the ‘bottom up’ (as it were) thesis that realizers necessitate their realizees, which entails the necessity of physicalism: it’s the reverse ‘top down’ thesis that mental properties, qua functional roles, have to have realizers (I take it, that’s what (13) amounts to). Third, if all contemporary physicalists are functionalists of some sort (not sure about the ‘all’, but most seem to be), then the worry, if it’s a worry, arises regardless of the necessitation of realization, of physicalist supervenience, and of worries about brute necessities—in other words, given (13), steps (1) to (12) are unnecessary, it seems. So, I’m a bit unclear as to why all this extra stuff was needed (perhaps, the point was to convince non-functionalist physicalists who also worry about brute necessities that they should be functionalists in some way?).

I now turn to L&T’s interesting responses to responses—most of which focus on (13). They consider 6; 2 seem particularly important (and puzzling).

ALIEN MENTAL PROPERTIES
Against (13), one might claim that if actual mental properties are realized (and are realized in every world in which they are instantiated), there could be alien mental properties, which are not realized, and hence basic. Perhaps, the suggestion is that, whilst those mental properties instantiated in our world are functional, it’s possible that there are alien mental properties which aren’t. In which case, we have no reason to think (13) applies to all mental properties: even if it’s true about actual mental properties, it doesn’t follow it applies to alien mental properties. (This could combine with another response L&T consider—ectoplasmic realization—so that, in some dualist worlds where there are alien mental properties which are basic, these basic mental properties realize, in those worlds, the mental properties that are also instantiated in our world.)

To understand their response to this, there are two further assumptions one needs (assumptions L&T put on the table in responding to the possibility of ectoplasmic realization, to the effect that ectoplasms are in fact compatible with physicalism). First, to ensure that alien mental properties (or ectoplasms) are mental, they suggest that mental properties are either conscious or intentional (or both). Second, they suggest that, without going into issues about the nature of the physical, physicalism can be defined as the view that only non-mental properties can be basic (not realized), but all mental properties must be realized. (It’s this definition of physicalism that allows them to claim that either ectoplasms are themselves mental properties which are realized, in which case they are compatible with physicalism, or if they are basic mental properties, then (13) applies. The suggestion that ectoplasms may be alien properties blocks the application of (13).)

L&T’s response seems to be going like this:
(15) if alien mental properties are mental, they must be determinates of the determinable D (conscious and/or intentional). [from first assumption]
(16) if physicalism is true about the actual world, all actual determinates of this determinable D are non-basic (realized) and the determinable itself is actually non-basic. [from second assumption]
(17) if the determinable D is non-basic, then it is necessarily non-basic (its instances are realized in every world in which it is instantiated).
(18) if the determinable D is non-basic in all possible worlds, then all its determinates (including its alien determinates) are non-basic too.
(19) hence, alien mental properties must be realized too.
(20) hence, dualism (the claim that mental properties are basic) is impossible.

Now, as someone pointed out, this seems to be THE master argument in L&T’s paper—the one establishing that physicalism can’t be contingent. Note also that, again, this argument establishes (if it works), rather than relies upon, the claim that realization of mental properties is necessary. So the argument doesn’t really show that contingent physicalism and the necessity of physicalist supervenience or of realization (à la (8)) are incompatible. Rather, it argues for the claim that all mental properties must be realized in every world in which they are instantiated, thus making dualism necessarily false.

Two things about this argument seem strange to me (if I have reconstructed the argument accurately):
i) I just don’t see why premise (17) should be true. It seems to be a particular instance of the following claim:

if property F has G, then necessarily F is G.

And that seems clearly false: just like objects can have contingent properties, the same is true of second-order properties (i.e., the properties of properties). The property of being grey has the property of having an instance right here on my shirt, but it could have failed to have that property. The property of being a computer has the property of being such that it allows one to send emails, but it could have failed to have that property.

In premise (17), this claim is applied to determinable D (being conscious and/or mental), so perhaps there is something special about determinables such that, if a determinable has a certain feature, it has it necessarily? But it doesn’t seem so: the property of being red has the property of spanning a spectral reflectance range from, say (approximately), 660 nm to 690 nm, but it could have spun a slightly different range; and it has the property of being seen by humans, but could have failed to. And the property of being a mammal has the property of being instantiated in the actual world, and the property of being instantiated by no flying creature (I think!), but it could have failed to. (note how these are examples of relational properties, just like the property of being realized or basic are).

This isn’t to say that no second-order property is necessary: it may be that red, in order to be the colour that it is, has to have certain resemblance relations to yellow and orange, say. But then the question is: which properties of a determinable are necessary and which ones aren’t, and what sort of reasons do we have to answer such questions? L&T offer no such reason to motivate (17)—in fact, they don’t even try to motivate (17) in any way. It’s hard to see why being realized should be a necessary property of mental properties, especially if physicalism is contingent. In this respect, it’s hard to resist the impression that (17) simply begs the question.

ii) I have a similar problem with (18). It seems to be a particular instance of the claim:

if a determinable D has property F (in all possible worlds), then all its  determinates (in all possible worlds) have F.

Again, that doesn’t seem generally true (in the actual world, or in all possible worlds). Red has the property of spanning a spectral reflectance range from, say, 660 nm to 690 nm, but surely none of its determinates do (even if red necessarily corresponded to that spectral reflectance range). It also has the property of being instantiated in the actual world and in possible world w, but not all its determinates do. And it has the property of triggering mostly cones of type L in the human visual system, but not all its determinates do. And being red or being a mammal have the property of being determinable properties (necessarily?), but surely not all their determinables do. Again, then, if this principle isn’t always true, why think it’s true of the mental determinable (being conscious or intentional) and the second-order property of being realized?

BASIC FUNCTIONAL ROLE PROPERTIES
Another response L&T consider: (a) it’s possible that there are dispositional properties which lack a categorical basis (either they are realized by non-categorical dispositional properties all the way down, or they are not realized at all), and (b) it’s possible that some functional/causal role properties are dispositional. Hence, contra (13), it’s possible that some functional/causal role properties lack a categorical basis, in which case they may either be realized by further functional/causal role properties all the way down, or basic (not realized at all).

To this, they respond that it is a “pyrrhic victory” for the contingent physicalist. So they seem to grant that some dispositional functional/causal role properties may be basic in the sense that they lack a categorical basis. But, they point out, worlds in which such bare dispositions exist need not be Cartesian worlds where mental properties are basic. And, remember, the motivation for contingent physicalism was to allow for the possibility of Cartesian worlds. Hence, the response seems to be, the possibility of bare dispositions fails to capture the motivation for contingent physicalism.

This is a weird response. For one thing, even if the bare dispositional worlds fail to capture the main motivation for contingent physicalism that has to do with dualism, they suffice to block premise (13). In which case, unless there is some argument L&T can offer to think that (13) is true, it’s unclear why we should accept it. For another, even if it’s true that bare dispositional worlds fail to capture the motivation for contingent physicalism (because they need not be dualist worlds), it offers another motivation for contingent physicalism (as L&T think of physicalism): namely, that if functional/causal role properties are realized in the actual world, there may be some worlds in which they aren’t, and ditto for mental properties (even if they are functional/causal role properties). In this respect, it’s weird to demand that one motivation for the contingency of physicalism has to capture another, as L&T seem to. Finally, since some bare dispositional worlds can also be dualist worlds, it’s wrong to say that bare dispositional worlds aren’t Cartesian worlds.

Ok, that’s quite a long list of questions/worries about L&T’s argument. As you can see, I need a lot of help! Now back to grading.

Posted by Philippe Chuard on Friday, December 12th, 2008 at 12:43 pm.