The Gap in the Knowledge Argument

Abstract

Alter (The Matter of Consciousness: From the Knowledge Argument to Russellian Monism, GB: Oxford University Pres, 2023) argues for something surprising: despite being widely rejected by philosophers, including Frank Jackson himself, Jackson’s knowledge argument succeeds. Alter’s defense of Jackson’s argument is not only surprising; it’s also exciting: the knowledge argument, if it’s sound, underscores the power of armchair philosophy, the power of pure thought to arrive at substantial conclusions about the world. In contrast, I aim to make a case for something unsurprising and unexciting: that the knowledge argument does not succeed, or, even less far-reaching, that Alter’s defense of it is not persuasive. Mine is a classic file-drawer thesis, but what it has going for it is that it’s true, or so I think, and hope to illustrate why you should too.

In The Matter of Consciousness: From the Knowledge Argument to Russellian Monism, Torin Alter argues for something surprising: despite being widely rejected by philosophers, including Frank Jackson himself (1994), Jackson’s (1982, 1986) knowledge argument—in essence, that not all facts are physical facts because you cannot learn what the experience of seeing color is like from black and white information—succeeds. Alter’s defense of Jackson’s argument is not only surprising; it’s also exciting: the knowledge argument, if it’s sound, underscores the power of armchair philosophy, the power of pure thought to arrive at substantial conclusions about the world. In contrast, I aim to make a case for something unsurprising and unexciting: that the knowledge argument does not succeed, or, even less far-reaching, that Alter’s defense of it is not persuasive. Mine is a classic file-drawer thesis, but what it has going for it is that it’s true, or so I think, and hope to illustrate why you should too.

The knowledge argument moves from the inadequacy of black-and-white information to convey what it is like to see color, to a conclusion about the falsity of the metaphysical doctrine of physicalism. But how does it make this move? according to Alter, the argument, in its Sunday best, proceeds roughly as follows: Mary, without ever having color experience, “learns the complete physical truth”(Alter, 5) and, because she has perfect powers of deduction, from this truth, she can “deduce any phenomenal truth that is deducible from the complete physical truth” (Alter, 15), yet since she learns something new upon seeing red for the first time, not all truths are deducible from the complete physical truth, which, assuming that physicalism is the view that all facts are deducible from the complete physical truth, means that physicalism is false.¹

What is my objection to this argument? Let me start with what might seem a mere quibble—a point some may think ought to be locked in the file cabinet once and for all—but it’s a quibble that prevents me from understanding the aim of the argument: I can’t make sense of what Alter means by “the complete physical truth” (Alter, 19).² Alter tells us that to know the complete physical truth is to know all the fundamental facts of physics (Alter, 19). But how should we understand these? If we understand them as the fundamental facts of today’s physics, there will be many things about the world Mary doesn’t know—for example, she won’t know why the universe contains far more matter than antimatter—and will not know even when she escapes from her imprisonment since today’s science has yet to explain them. If the physics in question is simply some future physics, this physics may also be incomplete. Does Alter mean by “the complete physical truth” the truth about the fundamental nature of the world that a true and complete physics will offer? On what for me (following Quine, 1981; Van Fraasen, 1996) is the natural understanding of what it means to be a true and complete physics, he can’t because on that understanding a true complete physics is just that: true and complete, or, in other words, one that correctly accounts for the fundamental nature of everything. Perhaps we will never achieve a true and complete physics. However, given that consciousness exists, then this physics would, by definition, account for the fundamental nature of consciousness. Thus physicalism, understood as Alter seems to, that is, as the thesis that everything, or basically everything—he excludes negative as well as indexical and related facts (Alter, 20–21)—is either part of or grounded in fundamental physics, would be true from the outset of the argument; if Mary knows the fundamental physical fact, which reveals the fundamental nature of everything, and has perfect powers of deduction, then, by definition, she knows everything.³ So if the knowledge argument is to have any hope of questioning physicalism, “the complete physical truth” must mean something other than the theory that accounts for everything.

The difficulty of understanding what counts as physical for the purpose of formulating the thesis of physicalism—a problem I refer to as the “body problem” (Montero, 1999) is an old problem (one finds it in Chomsky, 1995, Hempel, 1980, and as I argue in Montero 2015a, even in Hegel 1830/1971), and it’s one that Alter does address towards the end of the book, suggesting that, following Robert J. Howell (2012), we may be able to understand physical properties as properties that, in Howell’s words, have “implications for the distribution of things in space over time” (342). I don’t think Howell’s suggestion works. Not only might physicalists want to leave open some physicists’ speculations, however outré, that spacetime may be an illusion (Unzicker, 2020) without this showing that there are no physical properties, but also, if interactive dualism is true, the Cartesian mind has “implications for the distribution of things in space over time.”⁴ And if the thesis of physicalism is going to be inconsistent with any thesis, it should, I would think, be inconsistent with the thesis of Cartesian dualism.⁵

Alter, however, is not committed to Howell’s circumvention of the body problem. Jessica Wilson (2006), he thinks, also has an adequate proposal for understanding what it means to be physical. According to Wilson, to be physical is to be “treated, approximately accurately, by current or future (in the limits of inquiry, ideal) versions of fundamental physics” and “is not fundamentally mental” (quoted by Alter, 215). I am sympathetic to this view, however, as I’ve argued in Montero (2001) in response to a similar proposal by Papineau (1993), I see no reason for the first conjunct: if what is at issue is what is approximately accurately treated by current or ideal physics, then, if ideal physical is the theory that accurately treats everything, this requirement provides no additional restrictions on what is to count as physical that are not also present in the “not fundamentally mental” requirement.

Should Alter understand “physical” in the thesis of physicalism as whatever is not fundamentally mental?⁶ Although I think such a proposal—sometimes referred to as the “via negativa” (Gillett & Witmer, 2001; Montero & Papineau, 2005; Worley, 2006)—still has flaws (Montero, 2001, 2006, 2012, Montero & Brown, 2018), I also think that it offers at least a rough idea that one might use upon entering the debate over physicalism. However, Alter doesn’t like this approach since, although I was simplifying before, he thinks that the correct conclusion of the knowledge argument is not simply that physicalism is false, but rather that either physicalism is false or Russellian monism is true, where Russellian monism can be as a version of physicalism (Montero, 2010, 2015b). Now, if Mary knows all the fundamental facts that are not mental and all facts that are reducible to or can be derived from fundamental non-mental facts, then what she doesn’t know is going to be something irreducibly mental, which for via negativaists is something nonphysical. Thus, identifying the physical, in the knowledge argument, with that which is not fundamentally mental would prevent Alter (given his other assumptions) from deriving his preferred disjunctive conclusion to the argument.

Alter has the prerogative, of course, to advocate for the conclusion of his choosing. But how should he understand the complete physical truth in a way that leaves open his disjunctive conclusion? One suggestion might be to understand the complete physical truth as the concatenation of all structural truths, where the structural, following Chalmers (2010), is roughly that which can be fully characterized using only “logical, mathematical, nomic, and perhaps spatiotemporal vocabulary” (Alter, 29). And, although Alter doesn’t define the complete physical truth as the complete structural truth, he does understand physics in structural terms. Thus, it resonates with his approach.

Is the structuralist solution to the body problem viable? I think it might at least give us something to go on. Nonetheless, as is my wont, I have trouble understanding the nature of the structural.⁷ The answer that Alter offers is that the structural is that which can be fully expressible in exclusively logical, mathematical, nomic, and perhaps spatiotemporal terms. But how are we to understand these? Since the “perhaps” qualification is there to exclude the temporal properties of the Cartesian mind and other fundamentally mental phenomena (Alter, 33), and if we take the mathematical and logical to cover the non-spatiotemporal (something Alter could push back on), I am left feeling that the structural devolves into roughly the non-spatiotemporal and the spatial and/or temporal as long as it’s not fundamentally mental, or, in other words (since the non-spatiotemporal and the spatial and/or temporal would seem to everything), devolves into a via negativa account of the physical as the fundamentally not mental.⁸

I also wonder how we are to understand the mathematical component of the structural. The structural understanding of the complete physical truth grounds the physical world—the world of sticks and stones—partially in mathematics, but others, such as Maddy (1990), have grounded mathematics in the physical world. Which direction is the dependence? Furthermore, If you understand mathematical truths as grounded in our psychology, as do Locke (1690) and Mill (1846), mathematics is grounded in the mind, yet presumably one does not want the theory of physicalism to imply that the mental is grounded in the logical, mathematical, nomic and perhaps spatiotemporal and the mathematical is grounded in the mental. Fictionalism about mathematics, a view favored by Rosen (2001) and Leng (2010), among others, might not be such an attractive option either, since not only is the fiction one we create, but it might also seem to skirt too close the view that the aspect of the physical world that is supposed to be grounded in the mathematical is actually grounded in nothing.

Two views that remain regarding how to understand the mathematical nature of the world for the purpose of defining physicalism are structuralism and platonism. Should Alter’s physicalism embrace structuralism or platonism? Structuralism would seem to be a reasonable answer since the idea is, after all, that physicalism grounds consciousness in structure. But what type of structuralism? Some structuralists about mathematics (Hellman, 1989) see structure as grounded in the concrete world. Yet adopting this understanding of structuralism would, for Alter’s purposes, again seem to get the dependence relation backwards. Others (Benacerraf, 1965) are anti-realists about structure, resulting in a physicalism that teeters again somewhat too closely towards the view that everything arises out of nothing.⁹ This leaves understanding structure as abstract (Linnebo, 2017), which means that if we take the structural route of understanding mathematics, the result is a kind of platonism. Is this a problem?

According to Schneider (2017), physicalism is inconsistent with the existence of abstract entities. And some physicalists, rather than taking the most fundamental physical properties as abstract, exclude abstracta from the physical base in part because they take physicalism as a theory about the concrete world (see Montero, 2017 for discussion). Furthermore, many, following Benacerraf (1973), argue that mathematics cannot be abstract since our knowledge of math would be threatened.¹⁰ I, however, think it is possible to respond to views along the lines of Schneider’s and Benacerraf’s (Montero, 2017, 2022a). So I don’t think these concerns necessarily pose a problem for thinking of the physical in in terms of the structural. Can the higher-level facts about such things as chemical bonding and photosynthesis be deduced from fundamental structural facts? Perhaps it’s just me—and I’ve been in this business long enough to know that it is frequently just me—but I don’t have much confidence that these higher-level facts could be logically deduced from purely structural facts. Are there examples of deductions of one domain of knowledge from a distinct domain, deductions in which the conclusion follows with logical necessity from the premises? For example, can we deduce facts about, say, orchidology, from structural facts? Even one example of a deduction—written out in full detail so each step is logically or mathematically justifiable—from purely structural facts to facts about chemistry or botany would go a long way towards helping me to see how the deducibility claim could be true.

In fact, I tend to question whether there are many fully reductive explanations in science at all.

The typical examples of successful reductions in science that philosophers offer—lightning is reducible to electrical discharge, water to H2O, and the gene to DNA—are, I think, not true reductions. For example, as I argue in (Montero, 2022b), there are many cases of electrical discharge—the spark between your finger and the doorknob, the sparks generated by a dust storm or a Van de Graaff generator, the reddish-orange flashes in the upper atmosphere, called “sprites”—that are not cases of lightning. And drinking water does not exclusively contain H2O; furthermore, if you were served ultrapure water—highly purified water used, for example, in pharmaceutical manufacturing—when you requested a glass of water, it would make perfect sense for you to say, this is not what I requested— “I wanted water not ultrapure water” would be a reasonable thing to say (especially since drinking ultrapure frequently would not provide you with important electrolytes and, if it was your only source of liquid hydration, could lead to serious health complications whereas drinking normal amounts of water is not, one would think, supposed to be harmful to your health). And, as for the identity between the gene and DNA, long sequences of DNA lie between genes. To be sure, proponents of reduction would likely respond that in claiming that the gene is reducible to DNA, they are merely speaking in an abbreviated manner and that the correct analog would be much more complex. Yet once the complexities are spelled out, internecine squabbles may arise. For example, scientists debate whether the gene reduces exclusively to the sequence of DNA transcribed to RNA or to all of the DNA involved in this transcription process, including enhancers and promoters that might be completely separated from the sequence being transcribed.¹¹ Perhaps such reductions can be carried out, but right now, to me, the claim that they can is a leap of faith.

In sum, I feel I do not have enough evidence from successful reductions in science to make the leap from the fact that there are reductions in science to the conclusion that all higher-level facts—even excluding those concerning conscious experience—are deducible from lower-level ones, especially lower-level ones stated in exclusively structural terms. Will there always be phenomena, even outside the realm of consciousness, that we do not know how to reduce? I side with Niels Bohr, who was fond of the old Danish saying, “it is difficult to make predictions, especially about the future.”

I also think that physicalists do not even need to be committed to the metaphysical supervenience of the higher-level properties on the lower-level ones (Montero, 2013, and Montero & Brown, 2017). Alter rejects my view because he thinks that “traditionally, physicalism is conceived as a variety of monism,” and the rejection of supervenience would lead to a world “in which pluralism is true” and, also, that when we are concerned about whether the mind is physical, it is irrelevant whether, for example, botany supervenes on chemistry and chemistry on physics (194). I am unmoved. If one thinks that physicalism must be a species of monism, a world in which there are no supervenience relations between physics and chemistry and chemistry and botany can still be thought of as a monistic world since, if my proposal is correct, everything in it could still count as physical. To say that the failure of the chemical to supervene on the physical indicates that the chemical is not physical is to beg the question. More generally, I think that counting how many kinds of things there are is relative to one’s interests (Montero, 2022a). How many kinds of things are in my fruit bowl? Is the answer “one,” since every item is a fruit? Or should I say “three,” since I have apples, oranges, and a grapefruit? Or “two”: citrus and pome? Does the air count? The bacteria? It depends on what matters to you. As the structural is the mathematical, the logical, the nomic, and perhaps spatiotemporal, one could just as easily call this a type of pluralism. And as to Alter’s claim that the non-supervenience of the chemical on the world of physics is it is irrelevant to the discussion of whether the mind supervenes on the brain, I disagree. If chemical bonds fail to supervene on quarks, leptons, and so forth, then physicalists would not (and I think it would be reasonable for them to not) say chemical bonds are immaterial. Rather, they would and could reasonably claim that the supervenience of the chemical on the stuff of physics is not required for the chemical to be physical, and if it’s not required in that domain, it should not be required in the mind-brain domain either.

But let’s say that the complete physical truth can be understood along via negativa or structural or some other reasonable lines and that Mary learns it and that all higher-level features of the world (features that are supervenient on or perhaps reducible to lower-level features of the world) are deducible–in the relevant a priori sense–from the fundamental physical fact (understood along via negativa or structural lines) and that Mary has perfect logical powers of deduction, and, moreover, that she’s surprised upon seeing red for the first time. Is a physicalist position—a non-Russellian physicalist position—consistent with these premises? I have argued that it is (2007). Even if all the higher-level facts are deducible from the relevant lower-level facts (which, we can take to imply that physicalism is true), and even if Mary has perfect powers of deduction, she still might not be able to deduce what it is like to see red in her black and white environment since in order to understand the conclusion of her deduction—the conclusion that seeing red is like this, where “this” refers to the conscious experience of seeing red¹²—she may need (and I think it is reasonable to think that she does need) to have experienced what it is like to see at least some colors.¹³ What is it like to see red would be a priori deducible in the sense that from a bunch of information about what is going on in the brain it a priori follows—in Alter’s sense that “the justification of each step will make no essential appeal to experience”—that experiencing red is like this, even if understanding what the “this” requires having experienced red. Similarly, I might be able to deduce a priori from the fact that the Antikythera mechanism is smaller than a breadbox, a breadbox is smaller than an elephant, and the smaller than-relation is transitive, that the Antikythera mechanism is smaller than an elephant, however, I won’t be able to understand what I’m saying unless I have some sense of what the terms mean. Of course, I can learn what an Antikythera mechanism is without ever having experienced one: it’s an Ancient Greek mechanical model of the solar system. But the point still holds if we substitute in color squares for the mechanism and elephant: This red patch is smaller than that blue patch, and that blue patch is smaller than that yellow patch, therefore, the red patch is smaller than the yellow patch. Mary can a priori deduce this conclusion from the premises–the justification of each step of her reasoning will not require experience– yet she won’t fully understand her conclusion without ever having seen colors. And that she can’t, has no bearing on whether physicalism is true; or at least it doesn’t on the via negativa or the structural account: the world could still be at its most fundamental level entirely nonmental or structural. As Kant points out, some knowledge, although rightfully a priori, involves concepts that “can be derived only from experience” (Critique of Pure Reason, B3).

Maybe Alter thinks that what I’m calling Mary’s a priori deduction—a deduction arriving at a conclusion she can’t understand unless she has had color experience, is not rightfully a priori. Kant argues, for example, that “every change has a cause,” is a priori—it is what he refers to as “impure a priori” in contrast to “pure a priori”(Critique of Pure Reason B3)—even though to understand what a change is one needs to have had the experience of change, and I want to say that, similarly, the fact that seeing red is like this could be a priori deducible from some statement about the neurological processes in Mary’s brain (on the assumption, which I questioned earlier, that, roughly put, the divisions between the sciences are connected by reducibility relations) even though in order to understand what the “this” refers to, Mary will have had to have experience red.¹⁴ Thus, I think that one can accept that Mary learns something new upon leaving her black-and-white environment, that she knew all the lower-level features of the world that were not mental (or were structural), that she has perfect powers of deduction, and that physicalism is true even in the strong sense that there is nothing more to the world than lower level nonmental (or structural) phenomena and all higher-level phenomena that are reducible to this lower level phenomena.

I’m done for now. I hope to have persuaded some of you of my perspective on the knowledge argument, and I look forward to Alter’s responses to my comments—for I’ve been in this business long enough also to know that there are always responses.