Papineau's Causal Argument for Materialism is Invalid

I’ve been thinking about Papineau’s causal argument for materialism again, since I’ve been putting together some lecture notes for a phil mind class I’m teaching. You can find the argument in chapter 1 of his book Thinking about Consciousness (Googlebooks preview here). Here’s how the argument goes:

(1) Conscious mental occurrences have physical effects.

(2) All physical effects are fully caused by purely physical prior histories.

(3) The physical effects of conscious causes aren't always overdetermined by distinct causes.

(C) Materialism is true

A couple of concerns occurred to me while thinking about this argument again. I’ll share one of those concerns in this post, and another one in a later post.

Here’s the first concern: is this argument even valid?

Papineau seems to think so. After laying out the premises, he says: “Materialism now follows.” I've read several responses to Papineau's argument, but I haven't found one questioning its validity. (If you know of one, please let me know!)

Let me flesh out my concern that this argument is not valid. It seems that we can describe a coherent view on which Papineau’s premises are true but his conclusion is false. I’ll first describe the view, and then I’ll explain why it isn’t a form of materialism.

Here’s the view. It’s pretty weird, but all I claim is that it is possible. Suppose nearly all conscious mental occurrences are irreducibly non-physical and non-functional, but one is not. The taste of a dirty papaya, let’s say, just is a brain state. That’s the only mental state that is also a physical state. All the other ones supervene on their corresponding physical states with at most nomic necessity. So this isn't a form of functionalism. (Like I said, it’s a weird view. But possible.)

Add to the view that all conscious mental occurrences have physical effects (so premise 1 is true). Add also that all physical effects have prior sufficient physical causes (so premise 2 is true). Therefore, on this weird view we’re considering, most of the time the physical effects of conscious causes are overdetermined. But not always: when the taste of a dirty papaya causes you to frown as you recall your youthful misadventures off Jamaica's beaten path, there is no overdetermination. On this view, the taste of a dirty papaya just is part of the physical cause of your frown. And so premise 3 is true on this weird view I’m describing.

Alright, so all the premises of Papineau’s argument come out as true on the view I’m describing. Now how about the conclusion? Would materialism be true on this weird view I’ve described? Insofar as we can have intuitions about semi-technical terms like “dualism” and “materialism,” I’d think it’s pretty clear that this view is not a form of materialism. It sure looks like a weird form of dualism.

But we need not rest our case on such intuitions. Though it's pretty hard to say exactly what materialism is, any adequate definition should at least entail this modal supervenience claim: the mental supervenes on the physical&functional with something stronger than nomic necessity. But that supervenience claim is false on this weird view. There are physically&functionally identical possible worlds that nevertheless differ with respect to mental facts. Sure, the facts about the taste of a dirty papaya can’t vary unless the physical facts vary, but that doesn’t hold with respect to all the other mental states on this view.

So, that’s the problem for Papineau. The premises of his causal argument could be true even while the conclusion is false, and so the argument is invalid.

Let me know if I’m wrong about this, but it looks like the argument would be valid were we to strengthen premise 3 to something like this: The physical effects of conscious causes are NEVER overdetermined by distinct causes. But that’s a substantially stronger claim than Papineau’s original premise 3, and no argument is given for this stronger version of premise 3. And while you might be repulsed by the suggestion of rampant overdetermination, it's more difficult to get worked up about the suggestion that it sometimes happens. That is, the stronger version of premise 3 has much less going for it.

So it looks like this particular argument for materialism doesn’t have teeth.

Eleven Questions and One Question

1) What is it for something to be physical?

2) What is it for something to be a part of nature?

3) What is it for something to be a mental state?

4) What is it for something to be dependent upon something else?

5) What is it for something to be a cause of something else?

6) What is it for something to be a known fact?

7) What is it for something to be an empirical question / an empirical fact?

8) What is it for something to be scientifically established / scientifically explained?

9) Is there just one rationally acceptable answer to each one of the preceding questions?

10) Is there just one possibility to answer all of the questions 1) - 8) in a rationally coherent way?

11) Is there a single best possibility to answer all of the questions 1) - 8) in a rationally coherent way?

If you have done your best to answer these eleven questions, does materialism (or physicalism) seem inevitable to you?

Introduction, Part III

Beyond the three forms of Reductive Materialism, there is an altogether different form of materialism, namely, Eliminative Materialism, according to which there simply are no mental properties—or, at least, no instantiated mental properties. It turns out, however, that there are extremely few full-blown Eliminative Materialists. Most philosophers who identify themselves as eliminative materialists do so simply because they reject some central subcategory of mental property. For example, Paul and Patricia Churchland reject propositional-attitude properties, but they nevertheless accept that there are experiential properties (regarding which they adopt a certain form of reductionism).v Moreover, although they deny that there is a propositional attitude of knowing, they hold that there is knowledge. Another radical view is that there is no consciousness whatsoever (and so, in particular, no conscious experiential properties); but among the few philosophers of mind who have held this view, most have accepted that there are at least nonconscious propositional-attitude properties.vi The fact is that it is difficult to think of any major philosopher today who is thoroughgoing eliminativist, who holds that there are absolutely no (instantiated)mental properties—no knowing, no experiencing, no consciousness.

Besides Reductive Materialism and Eliminative Materialism, there has been one further rallying point for materialists, namely, Supervenience. In the setting of the Mind-Body Problem, this is the thesis that, whether or not mental properties are identical to physical properties, they in any case supervene on them (if only as a brute fact). Approximately, and in slogan form, mental properties supervene on physical properties if and only if it is necessary that any two objects that are alike in their physical properties are also alike in their mental properties (i.e., it is necessary that any two objects that differ in their mental properties also differ in their physical properties). Since the Mind-Body problem concerns the metaphysics of mind, the relevant modality here is metaphysical necessity, not mere nomological necessity; when we speak of supervenience, we will always mean metaphysical supervenience. Naturally, there are other notions of metaphysical supervenience besides the one just articulated in slogan form, some stronger and others weaker; and associated with each of these notions is a corresponding supervenience principle (more below).

We will say (relative to a chosen notion of metaphysical supervenience) that mental properties logically supervene on physical properties if they not only supervene on them but do so as a logical consequence of relevant definitions (including perhaps a posteriori scientific definitions). Logical Supervenience is the thesis that mental properties logically supervene on physical properties (relative to some chosen notion of supervenience). The appeal of Logical Supervenience is that (if correct) it guarantees that there is an explanation of why the underlying supervenience principle holds: it holds because it is a logical consequence of certain relevant definitions. For example, if Reductive Materialism is correct, mental properties will have reductive definitions that yield, as a logical consequence, a supervenience principle for each supervenience notion, regardless of strength. In this way, Reductive Materialism (if correct) provides explanations of why these supervenience principles hold. This, of course, is no surprise. But suppose that Reductive Materialism fails. This does not rule out the possibility of other instances of Logical Supervenience. For instance, Minimal Functionalism (discussed two paragraphs above) provides second-order definitions of mental properties that yield, as logical consequences, certain weak supervenience principles, for example, principles relativized to the actual world (more below); therefore, even though Minimal Functionalism is not a form of Reductive Materialism, it would (if correct) guarantee that there is an explanation of why these weak supervenience principles hold. And there are still further forms of logical supervenience. Suppose, however, that Minimal Functionalism is incorrect; and suppose, more generally, that all forms of Logical Supervenience fail. It would nevertheless be at least coherent to maintain Brute Metaphysical Supervenience (Brute Supervenience, for short): the thesis that not only do mental properties supervene on physical properties but they do so as a brute synthetic necessity (or as a consequence of brute synthetic necessities which are as much in need of explanation a this supervenience itself). Of course, Brute Supervenience comes at a price: supervenience would then be an unexplainable mystery (this will be relevant to the issue of complexity in the next section).

As already indicated, for Reductive Materialists, Supervenience is just a trivial corollary of their view. Likewise, for Eliminative Materialists: Supervenience is a trivial corollary since it is vacuously true on their view. But for Nonreductive Materialists—that is, materialists who deny both Eliminative and Reductive Materialism (in most cases, these materialists are advocates of the sort of Logical Supervenience associated with Minimal Functionalism or they are advocates of Brute Supervenience)—Supervenience has some promise of filling a crucial gap in their materialism. For suppose some supervenience principle (or a cluster of supervenience principles) provides a sufficient condition for materialism. In this case, our Nonreductive Materialists would be in a position to give a characterization of materialism that does not require the truth of Eliminative Materialism or Reductive Materialism and so does not trivially exclude from the start their own form of materialism. The envisaged characterization may be put as follows: materialism is the doctrine that one of Eliminative Materialism, Reductive Materialism, or Supervenience Materialism holds. For this strategy to succeed, the requisite supervenience principle or principles must meet two requirements: (1) they must be strong enough to provide a sufficient condition for materialism, and (2) they must be weak enough to avoid easy refutation. In service of goal (1), some materialists have gone beyond ‘intraworld’ supervenience (i.e., the sort of principle articulated in slogan form two paragraphs above) and have proposed instead a stronger ‘interworld’ supervenience principle. Others have gone beyond this and have proposed an outright entailment principle. The problem with both proposals is that they seem to threaten goal (2). That is, they appear open to possible counterexamples: for example, the metaphysical possibility of a disembodied being—a possibility that is accepted by many self identified materialists (e.g., David Lewis, Frank Jackson, Jerry Fodor, and many others). To lessen the threat of such counterexamples, the most common strategy has been to propose certain weaker supervenience principles, namely, principles that are relativized to the actual-world. For example, Jackson has proposed the following principle: if a world is a minimal physical duplicate of the actual world, it is a duplicate of the actual world simpliciter and so, in particular, is a duplicate in all mental respects. (A minimal physical duplicate is a world that (a) duplicates all the actual physical facts and (b) contains nothing else beyond what it must contain in order to satisfy (a).) But, then, the worry is that these principles might now run into trouble with goal (1): that is, they might be too weak to provide a sufficient condition for materialism and hence fail to provide the desired characterization of materialism.

There are good reasons to think that this strategy for characterizing materialism by resorting to the indicated actual-world relative supervenience principles does not succeed; in particular, there are reasons to think that such principles do not provide a sufficient condition for materialism. Here is one style of counterexample. The envisaged principles do not on their own rule out (i.e., they are consistent with) the existence of possible worlds (remote from ours) in which there are disembodied beings. For the same reason, these principles do not on their own rule out your existing in one of those remote worlds in a wholly disembodied state. Thus, these principles do not on their own rule out there being a difference between the modal properties possessed in the actual world by you and those possessed in the actual world by your body (after all, your body cannot exist in a wholly ‘disembodied’ state in any world); and this is all that is needed to establish a thesis of substance dualism. Since substance dualism is paradigmatically anti-materialist, none of these supervenience principles on its own provides a sufficient condition for materialism. This is just one style of counterexample; there are several others.

These considerations indicate that the suggested Supervenience approach faces an in principle problem. Our Nonreductive Materialists are committed to the falsity of Reductive Materialism; so, for them, there must be possible counterexamples to the biconditionals associated with the reductive definitions proposed by Reductive Materialists. In the case of the Identity Theory, for example, there must be either a possible failure of the necessity condition (e.g., a disembodied being, in the extreme case) or a possible failure of the sufficiency condition (e.g., a zombie, in the extreme case); there are far less extreme possibilities that suffice for the same purpose. But once such possibilities are admitted, they may then serve as counterexamples to the strong supervenience principles described above, thus forcing our Nonreductive Materialists to adopt
the proposed ‘local’ actual-world relative supervenience principles. In that context, however, the same possibilities may then be used (as they were in the preceding paragraph) to construct new, ‘distant-world’ counterexamples to the claim that the ‘local’ supervenience principles provide sufficient conditions for materialism.

The situation is even worse for advocates of Brute Supervenience. Consider an analogy between this thesis and a broadly Moorean brute supervenience thesis about aesthetic properties. According to the latter thesis, aesthetic properties (being beautiful, elegant, etc.) do not reduce to physical properties (they are neither first-order nor second-order physical properties) and, more generally, they do not logically supervene on physical properties; instead, they brutely supervene on them. No one in Moore’s day, however, would have said such a view is a form of materialism about aesthetic properties. Quite the contrary, this brute supervenience thesis was universally considered an instance of antinaturalism. But, according to virtually everyone—both in Moore’s day and today—materialism is just a special case of naturalism. In view of this, it difficult to see what could justify counting the above wholly analogous Brute Supervenience thesis about mental properties as a form of materialism.

Taken together, these considerations cast doubt on the above strategy for building Supervenience into the account of what materialism is, thus lending support for the view that there is only one coherent notion of materialism, according to which materialism is the doctrine that either Eliminative Materialism or Reductive Materialism holds. We find this view quite compelling even if we are not ready to endorse it here. A number of contributors to this volume, however, do endorse this view. That said, most of the contributors to the volume are willing, at least for sake of argument, to count various supervenience theses as forms of materialism. The majority of these contributors believe, however, that there are convincing arguments against such supervenience theses; for this reason, they are simply much less concerned about whether such theses really should be counted as forms of materialism.

Thus, ‘materialism’ is used in two main ways in this volume, one stronger and one weaker. According to the stronger use, materialism is the doctrine that either Eliminative Materialism or Reductive Materialism holds. According to the weaker use, materialism is the doctrine that one of Eliminative Materialism, Reductive Materialism, or Supervenience Materialism holds.xv (The best policy for readers is to refer to the individual papers to understand how the author is
using the term.)

One final terminological point. Among the philosophers of mind who reject Reductive Materialism, Eliminative Materialism, and Supervenience Materialism, many believe that the instantiation of mental properties is nevertheless determined by the instantiation of physical properties, where the hypothesized determination relation is a contingent relation—for example, a contingent causal or contingent nomological relation (in which case either the physical events would cause the mental events, or it would be nomologically necessary that, if the physical facts are such as they are, the mental facts would be as well). These views, however, are not positions in the metaphysics of mind; they are instead contingent scientific theories and as such are not versions of materialism, at least not on the primary use(s) of ‘materialism’ in traditional philosophy of mind. (Dualists from René Descartes to the present have held just such contingent-determination views of sensory experience, for example.) In any case, this is how we are using the term when we speak of the waning of materialism.

Introduction, Part II

MATERIALISM

Although the term ‘materialism’ has been used in diverse ways in philosophy, it traditionally has had a comparatively precise use within philosophy of mind. In this context, materialism is a certain view, or family of views, in the metaphysics of mind. Specifically, materialism is a certain view, or family of views, on the Mind-Body Problem, which concerns the ontological status of, and fundamental metaphysical relationship between, the mental and the physical—between, for instance, mental properties and physical properties, mental relations and physical relations, mental events and physical events, people and their bodies. (For simplicity, we will hereafter focus primarily on mental and physical properties (and relations); understanding their relationship arguably provides a key to resolving the entire problem.)

Historically, materialism was just the reductionist position that mental properties are identical to—and in that sense are nothing but—physical properties. (Idealism was the competing reductionist answer to the Mind-Body Problem, reducing physical properties to mental properties.) Throughout most of the history of philosophy, materialism took the form of what today we call the Identity Theory, according to which mental properties are identical to internal bodily properties, whether they be the properties associated with Democritean atoms, Hobbesian motions in the body or, in our period, electrochemical interactions at the neurological level. (Of course, nothing prevents such a theory from incorporating environmental factors in order to accommodate content externalism; for us, this kind of extended theory would still count as a materialist ontological reduction.) In the first half of the Twentieth Century another form of materialist reductionism emerged, namely, Behaviorism, according to which mental properties are identical to behavioral properties (dispositions of the body to behave in certain ways in certain circumstances). In the 1960s and ’70s a third form of reductionism gained prominence, namely, functionalism, according to which our standard mental properties and relations (being conscious, thinking, etc.) are identical to (and hence reducible to) second-order properties: specifically, mental properties are held to be definable in terms of the characteristic interactions of their first-order ‘realizer’ properties with one another and the external environment—where in the actual world, and perhaps all possible worlds, these first-order realizer properties are physical properties (presumably, the sort of physical properties invoked by the Identity Theory).2 On a strong version of this view (hereafter called ‘Functionalism’), the realizers of mental properties are necessarily first-order physical properties, from which it follows that mental properties are necessarily second-order physical properties and therefore belong to the general ontological category of physical property. Like the Identity Theory and Behaviorism, Functionalism qualifies as a form of Reductive Materialism.

There is a weaker version of functionalism according to which, even though mental properties are reducible to second-order properties and even though their realizer properties in the actualworld are physical, it is not necessary that the realizer properties be physical. If this view were correct, however,mental properties would not belong to the ontological category of physical property. To see why, consider a world in which the realizer properties are not physical (a possibility implied by this version of functionalism). Plainly, the inhabitants of such a world would be mistaken if they were to assert that mental properties belong to the ontological category of physical property. Therefore, since properties cannot change ontological category, it follows that it would, in the actual world, likewise be a mistake for us to assert that mental properties belong to the ontological category of physical property; on the contrary, mental properties would need to belong to an entirely different ontological category. Given this, this weak version of functionalism does not count as a form of Reductive Materialism, unlike the strong version described in the previous paragraph. There is another weak version of functionalism that is like this one except that it simply remains neutral on the question of whether it is necessary or contingent that the first-order realizers of mental properties be physical properties. This version does not on its own count as a form of Reductive Materialism (but only in conjunction with the independent thesis that it is necessary that the first-order realizers of mental properties be physical). We will call these two weaker versions of functionalism Minimal Second-order Functionalism, or Minimal Functionalism for short.

Footnote 2: David Lewis construes his functionalism as a form of first-order Identity Theory. This construal is dependent on his implausible view that our paradigmatic mental expressions are nonrigid designators of mental properties and relations. This view of these expressions fails for all of our core mental verbs and verb phrases: ‘thinks’, ‘believes’, ‘perceives’, ‘experiences’, ‘senses’, ‘feels’, ‘is aware of’, ‘is conscious of’, etc. By applying the operation of relation abstraction to these expressions, we get the following relation-abstracts: ‘the relation of thinking’, ‘the relation of believing’, etc. Such expressions are rigid designators, as Lewis himself acknowledges, and they denote core mental relations (the relation of thinking, the relation of believing, etc.). Analogously for verb phrases such as ‘thinks that human beings exist’: the associated property abstract ‘the property of thinking that human beings exist’ rigidly denotes the property of thinking that human beings exist. Expressions like ‘pain’, by contrast, do not even denote properties. On two core uses of the expression ‘pain’ (the core uses, we believe), ‘pain’ functions as a count noun which applies to pains, and it also functions as an associated mass noun for more or less pain (more or less in intensity or extent) or for some pain (some amount of pain). The mental property associated with the count-noun use is the sortal property of being a pain, and the mental property associated with the mass-noun use is the property of being some pain. The associated property-abstracts ‘being a pain’ and ‘being some pain’ are rigid designators of these properties. On Lewis’s functionalism, therefore, all of the above mental properties and relations (the property of being a pain, the property of thinking that human beings exist, the relation of thinking, etc.) are rigidly designated second-order properties and relations. That is, Lewis’s functionalism is just another instance of functionalism, as it was characterized in the text.

Introduction, Part I

The twenty-three papers in this volume, both individually and collectively, help to show why and in what ways materialism is on the wane. By saying that materialism is on the wane, we do not mean that materialism is in the process of being eclipsed—nor do we mean that materialism is likely to be eclipsed at any point in the foreseeable future. Indeed, there is good reason to think that materialism is a perennial fixture of philosophy (not to mention cognitive science). After all, materialism is a readily intelligible monistic worldview, appealing in its apparent simplicity, and a natural complement to the impressive ongoing successes in the natural sciences.

In spite of this, materialism is waning in a number of significant respects—one of which is the ever-growing number of major philosophers who reject materialism or at least have strong sympathies with anti-materialist views. It is of course commonly thought that over the course of the last sixty or so years materialism achieved hegemony in academic philosophy, and this is no doubt right by certain measures—for example, in absolute number of self-identified materialist philosophers of mind or in absolute number of books and journal articles defending materialism. It is therefore surprising that an examination of the major philosophers active in this period reveals that a majority, or something approaching a majority, either rejected materialism or had serious and specific doubts about its ultimate viability. The following is just a partial sampling of these philosophers, more or less in order of birth.

Bertrand Russell, Rudolf Carnap, Alonzo Church, Kurt Gödel, Nelson Goodman, Paul Grice, Stuart Hampshire, Roderick Chisholm, Benson Mates, Peter Strawson, Hilary Putnam, John Searle, Jerrold Katz, Alvin Plantinga, Charles Parsons, Jaegwon Kim, George Myro, Thomas Nagel, Robert Adams, Hugh Mellor, Saul Kripke, Eli Hirsch, Ernest Sosa, Stephen Schiffer, Bas van Fraassen, John McDowell, Peter Unger, Derek Parfit, Crispin Wright, Laurence BonJour, Michael Jubien, Nancy Cartwright, Bob Hale, Kit Fine, Tyler Burge, Terence Horgan, ColinMcGinn, Robert Brandom, Nathan Salmon, Joseph Levine, TimothyWilliamson, Mark Johnston, Paul Boghossian, Stephen Yablo, Joseph Almog, Keith DeRose, Tim Crane, John Hawthorne, Richard Heck, David Chalmers.

For all the people listed, we have documentation that they either rejected materialism or harbored serious and specific doubts about its ultimate viability. All the living philosophers listed (Putnam, Searle, Plantinga, Parsons, Kim, Nagel, and all those following) have given us explicit permission to include them on the list (under the description used in the sentence preceding this one). Limitations on space prevent us from giving a thorough presentation of citations; in the Bibliography, however, we cite relevant works by many of these philosophers. A comment about Russell and Carnap will be helpful here. Russell espoused, at different times, phenomenalism and robust neutral monism, each of which is antithetical to Reductive Materialism and also to the thesis that physical properties are metaphysically prior to—and hence are a supervenience base for—mental properties. See, e.g., Russell (1956). The young Carnap (of the Aufbau) was a phenomenalist. The mature Carnap (of ‘Empiricism, Semantics, and Ontology’) endorsed a form of anti-realism incompatible with the sort of materialism prominent over the course of the last sixty or so years. Like the young Carnap, Nelson Goodman and Benson Mates were also phenomenalists, not materialists.

Materialism plainly has not achieved hegemony when it comes to philosophers of this high caliber.

Here, then, is one respect in which materialism has been on the wane. We will identify two further respects in a moment. But, first, it will be useful to say a few more words about what we mean by materialism.