What Mary Never Did Know; or, How Kant Was Right

A well-known argument, due to Frank Jackson, goes as follows. (You can read the short version here.) The brilliant genius Mary has complete knowledge of physical reality. All the sciences, physics, chemistry, neuroscience, etc., have been completed—there is nothing more to add—so that the fundamental physical constituents and causes of all phenomena are known, together with everything that supervenes on them, and Mary has mastered all of this. But although Mary thus knows everything about the physical world there is to know, she does not know everything there is to know. For, a peculiarity about Mary is that she has lived her entire life in a black and white room and has never been permitted to view anything except in black and white. Thus, on the day when she finally leaves her room and sees, say, a red object, she will learn something she didn’t know before. She will say, “Ha! So that is what seeing red is like.” If this is correct, then, apparently, red, or the experience of seeing red, is not part of physical reality.

The “Mary” argument is just one of several ways to bring out what is really an old, classic problem with any sort of reductionistic physicalism. It is this. The fundamental elements of physics—the elementary entities and their properties, whatever they might be—do not include sensory qualities like colors, sounds, tastes, the feel of warm and cold, and so on. For example, say the elementary entities are subatomic particles like electrons and protons, with their properties of spin, charge, mass, and whatever else, and electromagnetic radiation with its properties of frequency and amplitude and whatever. Notice that these fundamental entities and properties don’t include qualities like red or sweet. And yet modern science is a reductionistic system: everything is supposed to be in effect constructed out of the fundamental entities and properties. For example, all the properties of a number 2 pencil are supposed to be explainable by its physical constitution: the arrangement of its atoms, their different internal structures, the forces binding them together into larger structures, and so forth. And for the properties of a pencil, you can see how this works. We really can explain all the structural and functional properties of a pencil in terms ultimately of its physical constitution.

But how are you going to explain the quality of being red (“what seeing red is like”) this way? What arrangement of particles or atoms brings red into existence? How can charge, mass, and spin (or whatever you like) be combined to construct red? To grasp the question seems to be to realize that one can’t begin to answer it. There seems to be no way to answer this question, even in principle.

This problem is as old as atomism itself. Democritus, one of the architects of the original Greek atomism, recognized the problem. In his ontology, nothing exists but atoms and the void. Atoms are the smallest units of matter, indestructible, impermeable, having shape, size, weight, and position, and that’s all. He recognized that although you can construct properties like hard and soft out of atoms and their interactions, you can’t do that with sweet, bitter, warm, cold, color, etc. Later, Galileo, in reviving atomism, encountered the same problem (in The Assayer). Newton likewise (in the Optics).

It is sometimes known as the problem of primary vs. secondary qualities. The primary properties are the ones of fundamental physics. Back in the day, that meant the properties of atoms: size, shape, weight, and position. Today we have a more sophisticated set, but the problem remains the same. The fundamental properties don’t include the “secondary qualities” (sweet, bitter, warm, cold, colors, etc.), and there seems to be no way even in principle to build secondary qualities out of the primary ones.

What people have typically said in response to the problem is that the secondary qualities are mental. Thus, Galileo: “I think that tastes, odors, colors, and so on are no more than mere names so far as the object in which we locate them are concerned, and that they reside in consciousness. Hence if the living creature were removed, all these qualities would be wiped away and annihilated.” And Newton: “For the rays, to speak properly, are not colored. In them there is nothing else than a certain power and disposition to stir up a sensation of this or that color.” But this is cheating, of course. For, what is the mind? Is it something outside the physical realm? Maybe that’s what these writers thought. But if you think the universe is completely physical, then you just recreate the problem all over again, this time in the brain rather than in the external objects. How do you build a “sensation of red” out of fundamental physical entities and their properties? If we couldn’t answer this question before, when it was about qualities of objects, we won’t be able to answer it now, when it is about sensations in brains. Putting sensory qualities in the mind just relocates the problem; it doesn’t solve it.

So this is why the problem can be taken to be an argument for dualism, and many philosophers today seriously advocate it as such. For instance, this was Jackson’s aim when he formulated the “Mary” argument.

However, my own take is that the problem does not force us to be dualists. I think the problem arises because the only properties we have scientific access to are (basically) causal. We know objects and their properties only through their effects. Ultimately, this means by their effects on our senses, but also by their effects on each other. Think of our knowledge of electromagnetic forces, or of particles like electrons. We know them only by what they do. They have certain systematic effects on certain objects, which we learn to formulate by mathematical equations in terms of constructs, such as “mass” and “charge,” which are ultimately defined functionally by their role in these very equations. (This is why fundamental physical entities can be defined by Ramsey sentences.) Thus, what anything is in itself we never observe or know; what we know about all things is what they do. Still, there must be some intrinsic properties. It can’t be relations or causes all the way down; there must be relata and things on which the causes operate. Our experience is in part of just such intrinsic properties, namely the sensory qualities such as colors. Experience has to include such intrinsic properties, because experience can’t be only of relations any more than reality can consist only of relations. There must be the things that are related.

Notice in this connection that the quality red doesn’t do anything. It has no effects (except by our psychological reaction to it; but that’s not something it does). The quality helps to constitute things that are related. It is not itself a relational or causal property; it is intrinsic.

So red may be physical after all, and it may exist in the world as well as in our experience. But though it may be physical, it will never be part of physics. The intrinsic character of physical reality is epistemically inaccessible to us.


9 thoughts on “What Mary Never Did Know; or, How Kant Was Right

  1. How was Kant right? I see nothing about that after the title.

    Emergent properties? Hydrogen and oxygen as separate elements aren’t wet, nor is one molecule of H2O, but water is.


  2. Emergent properties fall under the pencil example. In fact, a pencil can be seen as a bundle of emergent properties, such as being five inches long, having a certain shape, having a part that leaves a smear of itself when rubbed against a surface, having a core of one material surrounded by another, and being disposed to snap in two under a certain amount of force. None of the pencil’s constituent atoms individually has any of these properties. But these properties can be predicted from knowledge of the constituent atoms, their properties, their arrangement, and the laws governing their interactions.

    Similarly in the case of water. As a liquid, water has various emergent properties, such as being fluid, maintaining its volume under compressive forces, having a certain surface tension, having a certain specific gravity, evaporating at a certain rate under certain conditions, boiling and freezing at certain temperatures and atmospheric pressures, and so forth. All these properties are predictable from the properties of water molecules (and air molecules, etc.) and from the laws governing their bonding and other interactions. And notice, by the way, that they are all behaviors; they are things water does. That, actually, is just why they are predictable from the properties of the molecules and from the laws governing them. Those properties and laws are precisely in the business of describing and predicting the behavior of the molecules in various combinations and arrangements and conditions. By the same token, the sensory quality of redness will never be predictable from such properties and laws, because it is not a behavior. For this reason, where the sensory quality of redness comes from, and what it is, are beyond the reach of science.

    With regard to Kant, sorry, maybe I should have been more explicit. The connection is that both this doctrine and Kant hold that the way things are in themselves—at least as regards what I’m calling their intrinsic properties—are forever epistemically forbidden to us. Of course, the two views diverge in that mine holds that many properties of things, and much the most important ones—the causal and structural ones—are quite accessible. So we can know what we need to know to get along in the world. Nevertheless, since the so-called intrinsic properties include all the sensory qualities, even the sensory quality of space (as opposed to its relational structure), it means that our sensory image or conception of the world has no necessary connection with, and is probably utterly different from, the way it really is. And we have no way of knowing whether it is, one way or the other. We are cut off from the intrinsic nature of reality! That is, to me, a very unsettling notion.


  3. It seems right that perceptual experience represents redness to be a property of objects that is both intrinsic in the sense of ‘not relational’ (on the object side of things) and intrinsic in the sense of ‘not relative to the process of being aware of it’ (as relations as well as non-relational properties might be – or are). But that perceptual experience itself is ultimately composed of basic entities and their intrinsic (in the sense of ‘not relational’) features seems like an entirely separate issue. The experience of seeing an object as red-colored is certainly, in part, composed of red qualia. If red qualia were non-relational features of the world, then we might be able to apply something like Kant’s point to avoid much pressure in the direction of dualism (pressure toward taking red qualia to “really be” just as they appear – something like that). But I don’t see why red qualia (and the like) need to be intrinsic features that compose experiences of seeing-something-as-red (or the like). Or maybe I’m misunderstanding?

    (If Mary knows all the physical facts and physicalism is true, then she does after all know what it is like to see colored objects or see objects as colored via having red qualia, prior to venturing out of the room. So that the set-up is coherent presupposes that physicalism is false. The case appears to, but does not, go beyond the strong intuition that mental properties, and qualia in particular, are not simply complicated relations between physical entities and properties but rather something different in kind. We should just start with this premise in an honest way. Perhaps this “anti-physicalist appearance” of the mental or qualia in particular is non-veridical – I think this is right – but this appearance, and the evidence confirming or disconfirming it, is the right place to start.)


    • I don’t think we can escape the problem by alluding to perceptual experiences, presumably as structures in the brain, as opposed to what is represented in perceptual experiences. Joseph Levine has a pretty well-known argument that addresses this. He says, imagine we had a thorough account of the neural structures and processes that go to produce a sensation green. This might make an elaborate story involving electro-chemical activity, ion flow, cascades of activation flowing through ensembles of neurons, and so forth. But what would there be–what could there be–in this story that determines that the sensation produced is one of green? Couldn’t the very same story be used for yellow? If so, we don’t have an explanation of the sensory quality, only of the processes that produce a sensation of some quality.

      Of course, this argument is another intuition pump, just like the “Mary” argument (you’re quite right about that). But it makes the point that talk about the functional or other aspects of the perceptual experience does not address the problem that we have no scientific understanding or explanation of intrinsic sensory qualities like colors and tastes.

      This is why I favor my own analysis in terms of the “intrinsic” character of these qualities. (This is my own modest contribution to this debate, in my dissertation.) These arguments are typically thought of as being about qualia, as being about the mind and dualism, but that’s a mistake. Fundamentally, what drives these arguments is that the qualities in question have no causal or structural properties that are essential to them. They don’t do anything, and this puts them beyond the reach of science.


  4. We have discovered coolness/warmth (within a range shy of sensation of freezing or burning), to be a gauge, or measure, of rate of heat flow into or out of our body, which measure is a function of (health and) level of heat in external object in comparison to level of body heat and a function of the thermal conductivity of the external object. Does this sort of understanding of coolness or warmth render them any less an intrinsic sensible quality than redness? I gather not.

    Thinking of the senses as instruments is a good insight, but the ways in which they are not like the instruments we invent “to extend the senses” is something I’d like to think on and something perhaps illuminating that certain je ne sais quoi of Mary.


  5. Now it is again a Sunday, and many turn especially to thoughts of God. David—not the one anointed at Hebron, but the David who writes to us here—says in the setup for Mary that she “has complete knowledge of physical reality. All the sciences . . . have been completed . . . the fundamental physical constituents and causes of all phenomena are known, together with everything that supervenes on them, and Mary has mastered all of this.” A constitution like this is included in some long-running constructs of the thing called God.

    The divine being knows all that stuff too, though He or It is not physical at all and is not passing time. Rather, It and Its knowing and willing is, by some human accounts, in an eternal instant, perhaps like you or me before or after we lived, which is to say, perhaps like anything that does not exist. I should object to the conception eternal-instant knowing, but also to a conception of knowing all things physical by black-and-white Mary or by any possible intelligence at all.

    Physical reality has a past and a future (nervousness of Nicholas or Hume notwithstanding). It is not possible to have knowledge of all particulars of the past, for most all of them have had all their traces erased. The complete detailed history of any carbon atom in my body, for example, has had all record for most all those particularities erased. Most past particulars are unknowable, and not just due to any peculiarities of human intelligence or any other real intelligence. Rather, due to the nature of the physical world. Further, by the nature of some chaotic processes, and other physical considerations, there are not enough bits in the universe to compute either all the particularities of each of its components in its past or in its future. Most particulars are in principle unknowable.

    True, but the setup for Mary is her knowledge for kinds or species when their instances are encountered. Here too, I’d question a setup of any intelligence in which it is imagined that all species that have been are known to that intelligence. It’s a pretty big leap to suppose that all past species are in principle knowable by a temporal intelligence, which is to say, by any real intelligence. It’s a big leap to suppose that knowing all the physical species and causes of things entails ability to arrive at all the species of past things. It’s a big leap to suppose there is an ‘all’ to the causes of things or to the physical laws for things. Multiplying those three fractions I’d bet yields a fraction close enough to zero to worry that our Mary is too close to God.

    Still, I think a more constrained, more-us, though more wordy, setup concerning Mary’s physical knowledge is feasible. Perhaps this remodeled Mary could also afford Jackson’s problem, more or less. Maybe even new insights are to be had by following Mary out with the more or less.

    But at this time, I must turn to the more of daffodils. We got 600 bulbs this year, and today is for planting.

    Liked by 1 person

    • Thanks for the prompt to finally reply, Stephen. I meant to reply to your earlier comments, but just never managed to find the time.

      I mainly just want to post some pointers to further information, for anybody who may be interested in the issues I’m raising here. When I was groping my way in my dissertation to the view I’m defending here, I had no idea it had already been invented and has a name and everything. The sole previous writing along these lines I knew of was an Analysis paper (and therefore quite short) from 1990 by Simon Blackburn titled “Filling in Space.” I recall rereading it a couple of years ago and not finding it so luminously clear as I thought it before, but it’s still good—and it’s short—and it gets the ideas out there. The reason I reference it so often in my dissertation is that, at the time, it was all I knew of.

      However, the view turns out to be well-known nowadays in philosophy of science, where it goes under the name “Structural Realism.” For an introduction, see James Ladyman’s Stanford Encyclopedia of Philosophy article, “Structural Realism.” I also found valuable a review article by Roman Frigg and Ioannis Votsis from 2011, titled “Everything You Always Wanted to Know about Structural Realism But Were Afraid to Ask.”

      The first structural realist may have been Henri Poincaré. Another author who gave hints of the view early may be Duhem. But the first clear and well-developed expression of the view that I know of Bertrand Russell’s Analysis of Matter (1927). It is also the single best presentation I know of. Also important is a series of papers published by Grover Maxwell in the 1960s. See “The Ontological Status of Theoretical Entities” (1962), “Scientific Methodology and the Causal Theory of Perception” (1968), and “Structural Realism and the Meaning of Theoretical Terms” (1970).

      Russell and Maxwell—and Potts—arrived at structural realism by following what is now called the “upward path,” from epistemological considerations to what it is possible to know about the external world. That’s the way I have been arguing in this post and comments. But there is also a “downward path,” from a consideration of the historical fate of scientific theories about the external world to a weakening of the ambitions of scientific realism. The downward path responds to the fact that scientific theories are regularly discarded and replaced by claiming that although the material content of theories is discarded, their formal or structural content is not. For example, from the 1700s to the present, the going theory of the nature of light changed from being particles in the void (Newton) to waves in a luminiferous aether (Fresnel) to wave-like changes in a primitive electromagnetic field (Maxwell) to photons obeying quantum mechanical principles. These changes are all in what light is conceived as being in itself. However, what is preserved and cumulated over these changes in theory is the empirical content as expressed in their mathematical laws. Thus, the corpuscular theory could handle only simple reflection and refraction; the wave theory added interference and diffraction; the electromagnetic theory added connections with electrical and magnetic effects; the photon theory added the photoelectric effect; and so forth. Here the key paper is undoubtedly John Worrall, “Structural Realism: The Best of Both Worlds?” (1989). Given your background and interest in science, Stephen, you should read this paper. You may find it convincing where the upward path arguments were not. This is the paper that started the current philosophy of science discussion of structural realism, which had been essentially forgotten.

      600 daffodil bulbs! You’ll have done a good day’s work.

      Liked by 1 person

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s