Revisiting Ross on the immateriality of thought

The late James Ross put forward a powerful argument for the immateriality of the intellect. I developed and defended this argument in my essay “Kripke, Ross, and the Immaterial Aspects of Thought,”which originally appeared in American Catholic Philosophical Quarterly and is reprinted in Neo-Scholastic Essays. Peter Dillard raises three objections to my essay in his ACPQarticle “Ross Revisited: Reply to Feser.” Let’s take a look.

Ross’s argument

Before doing so, let me summarize Ross’s argument. The basic idea can be put in the form of the following syllogism:

1. All formal thinking is determinate.

2. No physical process is determinate.
3. Thus, no formal thinking is a physical process.

Naturally, the significance and justification of the premises of this syllogism need spelling out. I do that at length in my ACPQ essay, and readers who have not read it might want to do so before proceeding here. For present purposes I will merely review some key points.

First, for those totally unfamiliar with this debate it might be necessary to point out that the determinacy and indeterminacy in question have nothing at all to do with causal determinism, quantum mechanics, free will, etc. They have instead to do with the semantic determinacy and indeterminacy in view in some famous twentieth-century philosophical thought experiments like W. V. Quine’s “gavagai” example from Word and Object and Saul Kripke’s “quus” example from Wittgenstein on Rules and Private Language.

Something is “determinate” in the sense in question here if there is an objective fact of the matter about whether it has one rather than another of a possible range of meanings – that is to say, if it has a meaning or semantic content that is exact, precise, or unambiguous. It is “indeterminate” if it does not, that is to say, if there is no objective fact of the matter about which of the alternative possible meanings or contents it possesses.

Now, Ross argues, first, that at least some of our thoughts and thought processes do have a content that is entirely determinate or exact. The sort of thinking involved in mathematics and formal logic is the example on which he focuses. There can, for instance, be an objective fact of the matter that I am adding, specifically, or reasoning according to the inference rule modus ponens.

To deny that any of our thoughts has any determinate content, argues Ross, would be not only bizarre but incoherent. For example, you have unambiguously to grasp what it is to add or to apply modus ponens in the very act of denying that we ever unambiguously grasp what it is to add or apply modus ponens. You have unambiguously to apply formal rules of inference in the very act of giving an argument for the conclusion that we never unambiguously apply any formal rules of inference.

In defending the second premise, according to which no physical process is determinate in the relevant sense, Ross makes special use of Kripke’s “quus” example. Kripke defines the “quus” function as follows:

x quus y = x + y, if x, y < 57;

= 5 otherwise.

When carrying out addition, 2 and 2 will give you 4, 52 and 3 will give you 55, and 68 and 100 will give 168. But when carrying out “quaddition,” though 2 and 2 will still give you 4, and 52 and 3 will still give you 55, 68 and 100 will give you 5.

Now, suppose you had never added using any numbers higher than 57. Then all the past behavior you exhibited when it seemed that you were adding would be equally consistent with the hypothesis that you were really quadding rather than adding. Of course, most people have added using numbers larger than that. But we could always define quaddition instead using some number higher than 57. For example, we could use the number 3,998,702, and for any number we chose there would always be some higher number we could choose instead. Hence we could always define “quaddition” in such a way that for anyone who has ever apparently been adding, the behavior he exhibited in doing so would be equally consistent with the hypothesis that he was really quadding.

Of course, we would all say that, even if our outward behavior is consistent with this weird “quadding” hypothesis, we know that we have always really been adding and not quadding. But this is where Kripke introduces a famous skeptical scenario. First of all, anything we say about the way we use symbols like “+” and words like “plus” can be said about anyword. Just as some person you are observing might in fact be quadding rather than adding when he speaks or writes sentences like “Two plus two equals four,” so too, when he says things like “Oh, I’m really adding and not ‘quadding,’ whatever that is!” what he might really mean, for all you know, is that he is quadding rather than adding. It might be that every utterance he makes can be given an alternative interpretation in a way that is consistent with the hypothesis that he is using quaddition rather than addition.

But second, what is true of our interpretation of the words and behavior of other people is (so the argument goes) true also of our interpretation of our own words and behavior. Maybe you have yourself always been quadding rather than adding. And if you say “But sometimes I have just entertained the sentence ‘I am really adding and not quadding’ within the privacy of my own mind rather than speaking or writing it,” the trouble is that that sentence, which you only entertained mentally, might really have the meaning that you were actually quadding and not adding. So, Kripke’s imagined skeptic says, you can never really know what anyone’s words mean, not even your own. Every linguistic expression, whether spoken or written or even just existing in the form of mental imagery, is indeterminate in meaning or semantic content.

Nor, if mental imagery along with speaking, writing, and other bodily behavior, is all there is that could determine semantic content, is this merely an epistemological result but also a metaphysical result. It’s not just that you couldn’t know what you or anyone else really means. It’s that there would be no objective fact of the matter at all about what you or anyone else really means.

Now, Ross adapts Kripke’s example to his own purposes. He argues that if all the facts there are to go on are the physical facts, then Kripke’s imagined skeptic would be right. There would be no determinate meaning at all, no fact of the matter about the semantic content of what anyone ever says or thinks. Material or physical processes are inherently indeterminate in precisely the way Kripke’s example describes. But, again, formal thought processes do in fact have a determinate content. Hence formal thought processes cannot be material or physical. (As I argue at length in my essay, the way to see how this is possible given what Kripke says about mental imagery and the like is sharply to distinguish between intellect on the one hand and imaginationand sensation on the other, i.e. between strictly conceptual thought and the mere having of sensations, mental pictures, auditory imagery, and so forth.)

A lot more could be said, and I say it in the essay. So if there is some point in the argument you don’t understand or some objection you think hasn’t been considered, give the essay a read, because I think you’ll find I address it there. But this summary should suffice to provide context for my discussion of Dillard’s objections.

Metaphysics or just epistemology?

According to Dillard’s first objection, Ross is not in fact entitled to a metaphysical conclusion, but only an epistemological one. In particular, the most Ross can say is that you cannot know from the physical facts what you or anyone else means. But it doesn’t follow that the physical facts don’t suffice to make it the case that our thoughts and utterances actually have some determinate meaning. For all Ross has shown, maybe our thoughts are purely physical but nevertheless do have some determinate meaning, even if we can’t know what that meaning is. Hence (Dillard concludes) Ross hasn’t really established that formal thinking is not physical.

Dillard defends this claim by appealing to what he takes to be a parallel example. Consider the mitosis that a cell undergoes, and an imaginary Kripke-style parallel process which Dillard labels “schmitosis.” Schmitosis is just like mitosis, except that:

“[S]chmitosis”… yields nuclei containing an exact copy of the parental nucleus’s chromosomes for the first 10ⁿcell divisions but an entirely different set of chromosomes for any cell divisions > 10ⁿ. No matter how many mitotic divisions the cells undergo, their behavior will also conform to an incompatible, non-mitotic process. (p. 140)

What we’ve got here, Dillard suggests, is a scenario in which it is indeterminate from the lower-level physical facts whether a cell is undergoing mitosis or schmitosis. But it doesn’t follow that there is no objective fact of the matter about whether a cell is really undergoing mitosis, and it doesn’t follow that, if there is a fact of the matter, something non-physical is happening here. What we should say instead, in Dillard’s view, is that mitosis is irreducible to the lower-level physical facts, but is still itself physical.

Now, so far I am happy to agree with what Dillard says, at least for the sake of argument. Indeed, though he doesn’t put it this way, he is essentially making a very Aristotelian claim. For it is part of the Aristotelian theory of substantial form that a true substance has properties and causal powers that are irreducible to those of its parts. And that has nothing essentially to do with immateriality. The causal powers and properties of a dog or a tree are irreducible to those of their parts, but a dog and a tree are still purely material substances. The causal powers and properties of water are irreducible to those of a mere aggregate of distinct parcels of hydrogen and oxygen, but water is still a purely material substance. And so forth.

So, if so far Dillard has made a point that an Aristotelian like Ross or me would agree with, how is it supposed to pose a problem for Ross’s argument? The answer is that the irreducibility of mitosis to lower-level physical facts has, Dillard evidently thinks, only epistemological rather than metaphysical significance. And this in his view supports the judgment that the indeterminacy of meaning too has only epistemological significance. Just as mitosis is a purely material process even if irreducible to lower-level physical facts, so too might formal thinking be purely material even if its content is indeterminate from the physical facts.

But Dillard is making two mistakes here. The first is in supposing, without argument, that irreducibility has only epistemological significance. He seems to think that if a higher-level feature is physical despite being irreducible to lower-level physical features, then the gap between the levels cannot in any way be metaphysical and thus must be epistemological. But the Aristotelian hylemorphist denies that. There are, on the Aristotelian view, metaphysical gaps within the material world itself and not just between the material and the immaterial. For example, there are the traditional Aristotelian distinctions between the inorganic and the organic and between merely vegetative and animal forms of life. So, unless he provides some argument for the supposition he is making, Dillard is here begging the question against the Aristotelian. (My fellow Thomists will take note that for ease of exposition I am here using the term “metaphysical” in the broad sense in which it is typically used by contemporary analytic philosophers, not the narrower sense in which it is traditionally used by Thomists.)

More importantly, the issue doesn’t really have anything to do with irreducibility per se in the first place. Dillard is essentially conflating questions about indeterminacy and questions about irreducibility, and thereby misunderstanding Ross’s argument. Ross isn’t arguing that thought is irreducible and therefore immaterial. Again, as an Aristotelian he would not make such an inference. Rather, he is arguing that thought has a determinate semantic content and is therefore immaterial. So, the mitosis/schmitosis example is simply not relevantly parallel to Ross’s examples, because there is no semantic content involved in mitosis.

In developing his objection against Ross, Dillard makes some further mistakes. He attributes to me the thesis that “forms do not actually exist in the material world but only as idealized universals abstracted by the human intellect” (p. 141), and on the basis of this attribution also attributes to me an odd view about causal powers. But I have never said any such thing, and that is not my view at all. Take the substantial form of Socrates. It exists in Socrates himself, and since it is the ground of his causal powers, that ground also exists in Socrates himself. Something similar can be said about the substantial form of Aristotle. Now consider the form humanness. Unlike the substantial forms of Socrates and of Aristotle, this is a universal. And it is qua universal that this form exists only as abstracted by the intellect from Socrates, Aristotle, and other particular human beings. Dillard seems to be confusing (what I said about) the Aristotelian account of universals with an account of the ontological status of forms in general.

Dillard is also, I think, insufficiently attentive to the implications of the Aristotelian distinction between genuine substances and artifacts, but since this seems to be tangential to the main point of his first objection (and the relevant remarks are largely confined to a footnote), I won’t pursue the matter here. Readers interested in that distinction, and in the notion of substantial form and the other key components of hylemorphism, are directed to chapter 3 of my book Scholastic Metaphysics.

Incoherence?

Dillard’s second objection questions Ross’s claim that it is incoherent to deny that our thoughts ever have any determinate content. Here Dillard makes two main points. First, he notes that a Quinean naturalist would express the claims to which he is committed in terms of a formal language, and in summarizing how this would go Dillard speaks of what he calls the “L-sentences” of such a language. Dillard then says:

Whether an L-sentence is logically valid or whether an L-sentence is a logical consequence of other L-sentences has nothing to do with whether there are determinate facts about human thinking, any more than whether ferns in the Smoky Mountains are undergoing photosynthesis has anything to do with determinate facts about Tasmanian devils. (p. 144)

Now, Dillard’s point here, as far as I can tell, is that whether an argument in such a formal language is valid or not is just an objective fact that has nothing to do with what anyone thinks about it. Hence the determinacy or indeterminacy of human thought is irrelevant.

But if this is what Dillard is saying, then it seems to me that he is simply missing Ross’s point. The question isn’t whether there might still, as a matter of objective fact, be logical connections between propositions even if human thought was material (if we understand these objective facts in Platonic terms, say). The question is whether human thought could ever get in contact with these facts. And what Quinean and Kripkean indeterminacy arguments entail, Ross argues, is that human thought could not do so if it were material. For while there might still in that case be a fact of the matter about whether modus ponens is objectively a valid form of inference, there would be no fact of the matter about whether anyone’s thoughts actually conform to modus ponens or to some other, invalid inference form instead. And that’sthe sort of result that generates the incoherence Ross is talking about.

Dillard’s second move here is to appeal to the “skeptical solutions” naturalist philosophers have proposed to deal with indeterminacy puzzles like Kripke’s and Quine’s. He puts particular emphasis on the Quinean idea that we can take others to mean the same thing we do when our utterances don’t produce in them “bizarreness reactions” like blank stares, eye-rolling, puzzled looks, etc. Writes Dillard:

The austere naturalist grants that we can be said to assert, mean, and understand things in the minimal sense that our utterances and inscriptions which either contain the relevant expressions or are made in response to others’ utterances and inscriptions containing them do not provoke bizarreness reactions. (p. 144)

As far as I can tell, what Dillard is saying here is that as long as your utterances don’t produce such “bizarreness reactions” in others, then, the materialist can argue, you can be said to be adding, applying modus ponens, etc., and the indeterminacy problem is thereby solved. Hence the incoherence problem won’t arise.

But there are several problems with this proposal. First, we need to distinguish (a) the thesis that there is no objective fact of the matter about what anyone means, from (b) attempts to deal with the practical problems this thesis generates by way of appealing to the absence of “bizarreness reactions” or the like. Now, what Dillard seems to be saying is that a materialist could hold that as long as we have (b), then we needn’t worry about the practical problems posed by (a). But this completely misses Ross’s point. Ross is not saying that (a) could in principle be true but that it would pose intractable practical problems for the materialist – in which case the materialist’s appeal to (b) would be to the point. Rather, Ross is saying that (a) is incoherent and cannot in principle be true, so that we never even get to the stage of having to deal with indeterminacy problems by appealing to bizarreness reactions, etc.

(Of course, Ross allows that there would be no fact of the matter about what anyone means if human thought were material. But he does not grant that there might in principle be no fact of the matter about what anyone means full stop. He thinks there is and must be a fact of the matter, which is why human thought has to be immaterial.)

Another problem is that the very idea that the absence of “bizarreness reactions” suffices to solve the indeterminacy problem is simply a non-starter, for several reasons. For one thing, the absence of bizarreness reactions in others is neither necessary nor sufficient for one’s reasoning to count as conforming to a valid logical form. As all logic teachers know, if you present an argument like the following to beginning students:

Either 2 + 2 = 5 or the sky is blue.

It is not true that 2 + 2 = 5.
Therefore, the sky is blue.

that will certainly produce “bizarreness reactions” in them. They will think it a very odd way to speak, and they may even go so far as to say that it is not a logical way to speak. But of course, in fact it is a perfectly valid and even sound argument of the logical form disjunctive syllogism. Arguments that are not sound but still valid also provoke bizarreness reactions in people. If you say:

If the sky is green, then water is flammable.

The sky is green.
Therefore, water is flammable.

you will once again provoke bizarreness reactions in beginning students, and it takes a little effort to explain why, for all its obvious faults, this is at least a valid argument in the sense in which the word “valid” is used in logic.

You can also say and do things that do not produce bizarreness reactions in others, yet do not amount to logical reasoning. For example, if you give an argument like the following:

If it is raining, then the streets are wet.

The streets are wet.
Therefore, it is raining.

many will nod approvingly and think it in no way bizarre. But in fact it is an argument of the invalid form affirming the consequent. And it would remain invalid even if you somehow got all logicians to start agreeing with it. Furthermore, we say and do all sorts of other things that do not produce bizarreness reactions in others – walking, yawning, saying “Have a nice day,” etc. – but which do not amount to valid forms of reasoning, precisely because they don’t involve reasoning of any sort at all.

Then there is the fact that what the Quinean calls “bizarreness reactions” are themselves just as indeterminate in their significance as any utterance is. For example, it is no good to say: “The looks people give me when I say that two and two make four don’t seem to express puzzlement; therefore I must be adding and not quadding.” For just as the Kripkean skeptic can always ask: “But what do you or anyone else really mean when you use words like ‘plus,’ ‘add,’ etc.?” so too can he ask: “But what do you or anyone else really mean when you smile, nod, stare blankly, grimace, etc.?” The Quinean appeal to “bizarreness reactions” doesn’t solve the indeterminacy problem at all, but merely pushes it back a further stage.

Skepticism about other minds?

Dillard’s third objection is that if thought processes are immaterial, but all we ever observe of other people are their bodies and behavior, then we could never know the meaning of anyone else’s thoughts. “Ross’s immaterialism appears to open up an unbridgeable gulf between thinking and behavior,” says Dillard (p. 145).

But there are several problems with this objection. First, what Dillard is raising here is just a variation on the traditional “problem of other minds.” And it is difficult to see why Dillard thinks this is a special problem for Ross. It can be and often is presented as a problem whatever one’s view about the metaphysics of mind, whether dualist or materialist. For on either view there is arguably at least an epistemological gap between bodily and physiological facts on the one hand and facts about the mind on the other.

This would be especially true of the non-reductive form of naturalism that Dillard pits against Ross. As we saw above, in his first objection against Ross, Dillard suggests that the meaning of our thoughts might still be physical even if it could not be inferred from physical facts about behavior, brain activity, etc. He claimed, contra Ross, that this has only epistemological rather than metaphysical significance. But in that case Dillard himself is affirming an epistemological gap between thinking and behavior that poses just the sort of problem he thinks Ross’s position lands Ross in.

So, again, there is nothing about Ross’s position that raises the problem of other minds in a unique way.

Second, it would be rather absurd for a materialist who accepts Quinean or Kripkean indeterminacy results to raise this sort of objection against Ross. Ross could respond: “At least given my view there is a fact of the matter about what a person means, even if one could not know what that meaning is except in one’s own case. But if materialism were true, we couldn’t say even that much. There would be no fact of the matter at all, not just a fact of the matter that we couldn’t know about.”

Dillard also suggests that Ross’s position no less than anyone else’s faces “private language” problems of the sort Wittgenstein raised in Philosophical Investigations. He says that the “sui generisacts of thinking” entailed by Ross’s argument (for an explanation of which, see my article) would be problematic as purported anchors of meaning even in the first-person case, in just the way Wittgenstein says that sensations and behavior are problematic even in the first-person case. To be sure, Dillard seems to allow that Ross’s “sui generis acts of thinking” would suffice to determine the content of the thoughts I am having here and now. But he thinks they would not suffice to tell me what the true content of my past thoughts was. He writes:

But since my earlier behavior associated with the “+” sign does not determine whether I was actually adding as opposed to quadding or even thinking of nothing at all, I also have no idea whether yesterday I was adding, quadding, or thinking of nothing at all. (p. 146)

But there are two problems with Dillard’s argument here. First, it is not at all clear whyhe thinks that Ross’s “sui generisacts of thinking” would suffice to determine meaning in the first-person case here and now, but nevertheless would notsuffice to determine what I meant in the past. For, contrary to what Dillard says in the sentence just quoted, I don’t have merely my memory of past behavior to go on. I also have my memory of these past sui generis acts. And if present sui generis acts suffice to determine the content of what I am thinking now, why don’t past sui generisacts suffice to determine the content of what I was thinking then? (True, in theory I could be forgetting what past sui generis acts of thought I actually engaged in. But that’s a different problem, merely a special case of the more general question of how I can know memory is reliable. It has nothing to do with Ross’s account, specifically.)

The second problem with Dillard’s objection is that (depending on how one reads him) he may be overlooking the crucial difference between sensations, behavior, etc. on the one hand and Ross’s “sui generis acts of thinking” on the other. As I explained in my original article, with sensations, behavior, etc., there is a gap in principle between the sensation or behavior itself on the one hand, and whatever semantic content it is associated with on the other. But with the sui generis acts of thinking, there is no such gap. The thought just is its content. Now it is the gap that exists in the former case that is essential to the sorts of problems Wittgenstein raises. But since the gap doesn’t exist in the case of sui generis thoughts, the problems in question don’t apply to them.

In any event, Dillard undermines this entire third criticism of his when, on the last page of his article (p. 147), he endorses Paul Ziff’s solution to the question of how one can know what another is thinking. I will let the reader read and evaluate that solution for himself, because the details don’t matter for the point I want to make about it. And that point is that if Ziff’s solution is correct, then it shows why Ross’s position no more faces a “problem of other minds” than does the sort of view Dillard would favor. In which case it is not clear why Dillard even bothers to raise his third objection against Ross. Again, the problem of other minds is simply not more of a problem for Ross than it is for anyone else.

Not that I think it really is a problem. The so-called “problem of other minds” rests on the presupposition that “zombies” (in the philosophy of mind sense of that term) are in principle possible, and I do not think they are possible. But that is a topic for another time (and one I addressed in another post).

Anyway, I thank Dillard for his article and apologize for not responding to it earlier. I had originally planned to do so in the context of an ACPQ article, but given the various book projects and other commitments that have taken up so much of my time over the last couple of years, I kept putting that article on the back burner. Since it now seems a little late in the day for an ACPQ response, I decided to respond in a blog post.

Viral Trends

Revisiting Ross on the immateriality of thought