logo logo

Mary Doesn't Know Science:
On misconceiving a science of consciousness.

Nigel J.T. Thomas Ph.D.
To email me, click here
California State University, Los Angeles.

[Read at the annual meeting of the Pacific Division of the American Philosophical Association, March 26th 1998.]

The so called "Knowledge Argument" of Frank Jackson (1982, 1986)1 claims to show that there is something about the human mind that must inevitably escape the grasp of physical science: "There are truths about . . . people ( . . . ) which escape the physicalist story" (Jackson, 1986). In effect, materialism is false, and science, as opposed to metaphysics, cannot hope to attain to an understanding of consciousness.

In its best known, most arresting, form, the argument concerns the story of a brilliant scientist of the future called Mary. We are asked to suppose that Mary knows all the physical science relevant to human color vision "in a wide sense of 'physical' which includes everything in completed physics, chemistry, and neurophysiology, and all there is to know about the causal and relational facts consequent upon this, including of course functional roles" (Jackson, 1986). However, for some reason, Mary been imprisoned all her life in a color-free environment. Although her visual system is normal, she has never seen anything colored. (This may not be a true practical possibility - consider phosphenes - but it seems to carry the requisite conceptual and metaphysical possibility.) On being released from her imprisonment, Mary sees colors for the first time and consequently, it is claimed, she learns something new: namely, what colors look like to people. Jackson then lays out the argument as follows:

(1) Mary (before her release) knows everything physical there is to know about other people.
(2) Mary (before her release) does not know everything there is to know about other people (because she learns something about them on her release).
(3) There are truths about other people (and herself) which escape the physicalist story.
(Jackson, 1986).

There have been many ingenious attacks mounted against this argument, but they seem to have done little to diminish its intuitive appeal, and it still finds sophisticated defenders. One suspects that many people hope that it is unsound, but are secretly worried that it may not be. I will argue that it is indeed unsound, but that most of the criticisms that have formerly been made of it have been misdirected. Furthermore, a proper understanding of the strengths and defects of this argument may tell us something significant about how a science of conscious experience should be conducted.

The Knowledge Argument has generally been accepted as formally valid. Defeating it will thus involve defeating one or other of its premises, and in fact nearly all attacks to date have focused on premise (2). It is claimed that in knowing all the science, all the physical facts, Mary must really already know (in some, disputed, sense) all about the experience of color during her imprisonment, and thus she does not really learn anything new on release. It may perhaps be admitted that she acquires 'a new perspective' on knowledge she already had, or an ability to represent it to herself in a new way, but not that it is actually new knowledge2. Van Gulick (1993) reviews several of the best known critiques of the Knowledge Argument; they all turn out to be attempts to undermine premise (2) in some way. I wish to suggest, however, that the real problem lies in premise (1). Both Jackson and his critics are working with an impoverished picture of what it is to have scientific knowledge. As the example is set up, it is impossible for Mary to know "everything there is to know about the physical nature of the world" (Jackson, 1986).

Knowledge, I take it, implies understanding, and, to put it bluntly, Mary (pre-release) does not truly understand color terms, so she cannot fully understand the physics and physiology, let alone the psychology, of color. She may know that certain wavelengths and mixtures of wavelengths of light striking the eye produce a sensation in normal people that is called "blue", and she may also be able to tell you what goes on in the eye and the brain when this happens, but unlike normal physicists and physiologists she cannot attach its ordinary meaning to the word "blue", so she cannot fully understand basic sentences of color science like "Wavelength xyz causes people to have an experience of blue"3. This claim may have a ring of naïveté about it to philosophers steeped in meaning holism, but, of course, color terms are paradigm cases of 'observation' or 'periphery' (Quine, 1951) terms4, and therefore, on pain of linguistic idealism, are particularly closely tied to experience. Holism may be what tempts Churchland (1985) and Dennett (1991) to seem to suggest that if Mary really did have complete theoretical knowledge of a completed future science, she would somehow be able to figure out what the experience of blue is like5. After all, holism seems to imply that she would be able to reach an understanding of the meaning of "blue" that is considerably richer than that available to us poor scientific primitives, and, of course, knowing what "blue" means entails knowing what blue is like! But this suggestion is best regarded as a new reductio of a type of holism (one that focuses on the theoretical and linguistic and slights the underpinning of experience) that has long since found its definitive reductios in Rortyan and Feyerabendian relativisms6. The truth is that all of Mary's scientific learning, no matter how complete in other respects, will not provide her with an understanding of the referent of "blue", which is something that other people can have without any scientific education at all.

Jackson might well agree with much of the above. Knowledge of what blue is like will not be acquired from the textbooks and black and white TV lectures that he asks us to imagine as the sources of Mary's vast knowledge. She will learn something new on leaving her prison. Some of Jackson's critics also agree that she will. Nemirov (1980, 1990) and Lewis (1988) have argued, in effect, that although Mary does gain new abilities, abilities to recognize and imagine colors, when she sees colors for the first time, this does not amount to new knowledge of facts or propositions. Thus, there are no facts that the complete scientific account (that she had before her release) leaves out.

Jackson himself accepts that Mary acquires new abilities of the sort Nemirov and Lewis suggest, but he insists that this will entail her gaining new factual knowledge as well (Jackson, 1986). That may well be so, but what I want to insist is that her scientific knowledge of color (or, indeed, anything) cannot itself be complete in the absence of extensive recognitional and imaginative abilities of just this sort: abilities that can only be gained through direct 'hands-on' experience. Far from being the essentially theoretical web of words that meaning holists tend to depict, connected to experience only at its periphery7, scientific understanding depends through-and-through upon praxis and perceptual skill. Both Jackson and his critics seem implicitly to regard scientific knowledge as entirely propositional, as a matter of knowing a lot of empirical and theoretical facts (and being able to draw inferences from them). It is the sort of thing you could learn from a diligent reading of appropriate textbooks, or close attention to lectures. Jackson (1986) says (and his critics do not demur) that: "it is plausible that lectures over black-and-white television might in principle tell Mary everything in the physicalist's story". I reply: it is not plausible at all. Indeed, even color television will not cut the mustard. Science is not just a matter of theoretical, textbook knowledge. To achieve the sort of depth of scientific knowledge that is attributed to Mary, you have to get your hands dirty! Perhaps philosophers are tempted to forget this because philosophical work (unlike scientific work) is almost entirely a matter of working with words; perhaps it is the lingering spell of Plato's dream of rendering everything explicit. Anyway, enough work, philosophical, historical, and sociological, has been done by now to definitively establish the fact that science is not, and cannot be, entirely a matter of explicit fact and theory.

Scientific understanding intrinsically involves what Polanyi (1958), an accomplished scientist himself, called "tacit knowledge" (Polanyi, 1958): knowledge that cannot be acquired from books but only through experience with what Kuhn (1970), thinking in the same vein, called "exemplars": concrete scientific achievements or techniques. Likewise, Ravetz (1971) and others have explained just how much of scientific expertise is skilled craftsmanship rather than book learning, and more recently, Collins (1990) has documented in some detail the impossibility of making such practical expertise fully explicit8. It is not just that scientists in the laboratory rely on rules of thumb that they find difficult to justify and articulate, but that they depend on perceptual (recognitional/imaginative) and practical skills that are not explicitly or consciously represented at all, but are part of the unconscious, learned wisdom of the body. Although philosophers may sometimes forget this point, I do not think many now seriously dispute it. It is only reinforced by the work of philosophers like Hacking (1983) and Franklin (1986), who emphasize how experimental practice does not serve just to verify or falsify preconceived theories, but is inherently involved in the whole scientific understanding of nature. This is why so much of a scientific education (unlike a philosophical one) takes place in the laboratory or the field, rather than just the lecture hall and the library. Science educators are emphatically not wasting taxpayers' money or students' time when they insist that this must be so.

A possible objection might arise at this point from those cognitive scientists who regard all mental activity as somehow linguistic in character. I am referring to those who follow Fodor (1975) in regarding the brain as a computer which manipulates the symbols of an in built, unconscious, language-like representational system, dubbed "mentalese". They might object that although many of the perceptual and practical skills acquired by scientists in the course of their training may never in fact get consciously and explicitly represented, and although explicit instruction might never be a practicable way of acquiring them, nevertheless they must in fact be encoded in representations in mentalese, and this mentalese could, in principle, be translated into natural language, and so made explicit. In fact (they might say), full and explicit descriptions of these representations would surely be in the textbooks of the completed cognitive science that Mary is supposed to know. She would indeed know all about practical scientific skills.

This argument does not move me much because I have never found this 'Language of Thought' hypothesis in the least plausible; it seems to fly in the face of everything we would expect from our knowledge of evolution and neuroscience. Even so, it still has a lot of sophisticated supporters, and I certainly will not attempt to refute it here. Fortunately that will not be necessary, because, of course, knowing about skills and abilities, in whatever detail, is simply not the same as having them. This is obvious enough when we think in physiological terms: knowing what the strength of your muscles consists in, and knowing how to strengthen them, does not in itself make you strong: you have to actually train and exercise. Likewise, knowing the neurophysiological details of the embodiment of some skill will not in itself endow you with that skill. You still have to practice it.9

The point may not be quite so obvious within the Language of Thought picture of things, because its proponents have usually thought of mentalese as being in some sense the programming language through which the mind is written into the neural computer, and there is a temptation to think that if only we knew how to read and write that sort of code, we could reprogram ourselves. There is a concomitant temptation to see programming as a purely intellectual exercise, quite unlike working-out in the gym: as something that Mary could do just by hard thinking. But in fact programming is not purely intellectual. Even granting the picture of brain as programmed computer, we must notice that the physical facilities for direct reprogramming have not been provided. The brain has no keyboard or floppy disk slot. If the brain has a program, it is firmware, encoded in our DNA, and its behavior is changed not by rewriting the code, but by inputting new data: by what we call practice and experience.

So, even if Fodor is right about representation, it remains true that real science cannot be learned just from lectures and textbooks. Being able to parrot the physics textbook perfectly does not amount to a knowledge of physics. Of course, in the real world, if I learn my physics text by heart, I will learn something about physics. I know everyday English, and most of the words found in the text are either ordinary English words, or are used in ways very close to their everyday meanings. I would be able (and likely) to make many inferential connections amongst the statements in the book, and between them and statements that I believe on other grounds. I would also be able to answer a lot of questions about physics, and perhaps even be able to pass a written examination in the subject. However, I hope you would think twice before letting me loose in a physics laboratory or, a fortiori an engineering workshop.

What is more, much of the everyday experience that enables me to have this sort of superficial understanding of the physics text (or the Scientific American article), is of a piece with the sort of ordinary color experience that Mary is supposed to lack. Although it might get me through the pop quiz, I do not know Newton's second law just by being able to parrot the formula "F=ma", or even by being able to do calculations with it: I have to know what the letters stand for, and, ultimately, I have to know what the words "force", "mass" and "acceleration" mean, and those meanings have to be rooted in my experience. There are two aspects to this: I can feel forces on my body, see examples of acceleration, and also feel masses (at least indirectly, through the inertial forces they generate when I manipulate them); if I also have some real, practical training in physics, I will be able to solve not only textbook, but also real world problems using the formula, applying its terms appropriately to things in my environment. A Mary kept numb and motionless throughout her life would not be able to understand mechanics; no more can a Mary kept color blind understand color.

Jackson's original thought experiment can actually be modified somewhat to allow Mary to acquire many of the perceptual and practical skills that underlie the scientific understanding of real scientists. If instead of thinking of her cell simply as having an austere black and white decor we think of it as illuminated with monochromatic light, then we can allow her something much more like a proper scientific education. Instead of merely reading or hearing about brains, and seeing black and white pictures of them, she will be able to see, touch, smell and dissect them, and she will be able to look at real brain tissue under a microscope and thus learn, by experience, at least some of the perceptual skills by which the histologist recognizes cellular structures in what, to the unschooled eye, appears as a confused jumble or blur. Mary might indeed become a genuinely competent neurophysiologist, and quite competent in other branches of science too, all without exposure to a normal range of color experience.

However, she would still be missing the essential experiences that would allow her to fully understand the science of color. She would be in a somewhat similar position to the Norwegian visual scientist Knut Nordby, who, through a rare genetic defect, was born without cone cells in his eyes, and has thus never been able to perceive colors. Consider his testimony:

Although I have acquired a thorough theoretical knowledge of the physics of colours and the physiology of the colour receptor mechanisms, nothing of this can help me to understand the true nature of colours. From the history of art I have also learned about the meanings often attributed to colours and how colours have been used at different times, but this too does not give me an understanding of the essential character or quality of colours.
(Nordby, 1996)10

Nordby, it is clear, has thought about color vision much more than most of us probably do, and, as a scientist, he has already made contributions to our understanding of it. No doubt he will continue to do so. Indeed, he is probably particularly well placed to make contributions towards our understanding of certain aspects of the problem. However, I would submit that he could never be the scientist to give us the definitive scientific account of color experience. He could not even fully understand such an account because, as he says, he would not truly understand some of its key terms. Mary, although she may know many more facts than are available to Nordby today, will nevertheless be in the same position until she escapes her jail.

The Knowledge Argument fails, therefore, because it is impossible that someone could have a complete understanding of physical science without having had the relevant experiences. Feyerabend (1969) was wrong: science without experience is impossible11. Physicalism is vindicated, not because Mary would fail to learn something new when she first saw colors, but because the story told about her is incoherent: you cannot completely understand the science of color without the recognitional and imaginative skills that allow one to know what color terms mean, and you cannot get those skills without experiencing colors. Indeed, quite generally, recognitional and imaginative skills are an essential part of scientific understanding. For most scientists this can, and often does, go without saying, but when we are trying to do the science of consciousness, of human experience itself, it is surely something we should keep explicitly in mind.

[Click on note numbers to return to your place in the main text]

1. Jackson (1986) credits Nagel (1974) and Robinson (1982) with giving versions of the argument independently.

2. Horgan (1984) allows that Mary acquires new "ontologically physical information", but this turns out to mean information about physical entities, and relations between such entities, that she already knew all about under another description.

3. She could understand "Wavelength xyz causes people to have an experience that they call 'an experience of blue'", but that is not at all the same.

4. Of course, Quine himself would insist on talking about periphery statements, not terms, but it would seem nevertheless seem to remain fair to say, for current purposes, that certain terms, such as color names, are particularly likely to turn up in periphery statements rather than elsewhere, and thus do have a special association with the periphery. I see no reason why the argument of the current paragraph could not be reformulated, with some circumlocution, so as to be about statements rather than terms.

5. If the suggestion is merely the much more plausible one that Mary's scientific expertise would enable her to find a way to somehow experience blue (despite her captors' utmost efforts) through self stimulation or imagination, then, as Jackson (1986) points out, it is no threat to the Knowledge Argument: it will still be the case that Mary only finds out what blue is like after experiencing it.

6. It is surely no coincidence that Churchland (see especially his 1979) is profoundly influenced by Feyerabend. However, Churchland, quite unlike his mentor, then seems to have made the necessary existential leap to a deep love of the scientific.

7. The locus classicus for this picture, or at least this terminology, is Quine (1951), but it is not hard to see its roots in the sharp distinction drawn by the Logical Empiricist movement between the theoretical and the observational, where the latter functioned merely to verify or falsify the former. For some philosophers, the holist move to efface this distinction, and to question the possibility of definitive verification or falsification, seems to have pushed the experiential aspect of science even further into the background, assimilating it to the theoretical.

8. I do not say that Collins' work proves in principle impossibility, but it certainly renders the point vivid and plausible (if it was not already). I should also say that despite the fact that this tradition in the philosophy and sociology of science has been strongly associated with epistemological relativism, it is not obvious that accepting an irreducibly tacit element in science need inevitably push one that direction. Indeed, I think the association is largely adventitious. Polanyi, in fact, was vehemently anti-relativist, and Kuhn spent most of his career disavowing the apparent relativistic implications of his early work (without, I think, giving up the stress on the tacit). In fact it was Kuhn's associate Feyerabend who really embraced relativism, but Feyerabend continued to talk of scientific theories (rather than "exemplars" and "paradigms"), thus continuing to stress the explicit rather than the implicit aspects of science, and his relativism was clearly founded upon his holism about the meaning of scientific terms.

9. Conceivably, in the utopian scientific future where Mary resides, it might be possible to acquire a skill without practice, through undergoing some sort of micro-neurosurgical procedure, and Mary might know how to do such things to herself. But the point remains that you do not have a skill just in virtue of knowing about how it is embodied.

10. I am grateful to David Chalmers for alerting me to this article.

11. In my view, the reason for this is not that the artificially intelligent scientist that Feyerabend envisages is an impossibility, but rather that a machine with such capabilities necessarily would have experiences. However, it is beyond both the scope and the needs of this paper to argue for that interpretation.


Churchland, P.M. 1979 Scientific Realism and the Plasticity of Mind. Cambridge University Press.

Churchland, P.M. 1985 "Reduction, Qualia, and the Direct Introspection of Brain States," Journal of Philosophy (82) 8-28.

Collins, H.M. 1990 Artificial Experts: Social knowledge and intelligent machines. MIT Press.

Dennett, D.C. 1991 Consciousness Explained. Little, Brown.

Feyerabend, P.K. "Science Without Experience," Journal of Philosophy (66) #22.

Franklin, A. 1986 The Neglect of Experiment. Cambridge University Press.

Fodor, J.A. 1975 The Language of Thought. Thomas Crowell.

Hacking. I. 1983 Representing and Intervening. Cambridge University Press.

Horgan, T. 1984 "Jackson on Physical Information and Qualia," Philosophical Quarterly (34) 147-152.

Jackson, F. 1982 "Epiphenomenal Qualia," Philosophical Quarterly (32) 127-136.

Jackson, F. 1986 "What Mary Didn't Know," Journal of Philosophy (83) 291-295.

Kuhn, T.S. 1970 The Structure of Scientific Revolutions (2nd edition). University of Chicago Press.

Lewis, D. 1988 "What Experience Teaches," Proceedings of the Russellian Society University of Sydney. [Reprinted in W.G. Lycan (ed.): Mind and Cognition. Blackwell: 1990. pp. 499-519.]

Nagel, T. 1974 "What is it Like to be a Bat?" Philosophical Review (83) 435-450.

Nemirow, L. 1980 Review of: Mortal Questions, by Thomas Nagel. Philosophical Review (89) 473-477.

Nemirow, L. 1990 "Physicalism and the Cognitive Role of Acquaintance," in W.G. Lycan (ed.). Mind and Cognition. Blackwell. pp. 490-499.

Nordby, K. 1996 "Vision in a Complete Achromat: a personal account," published on the World Wide Web at: http://consc.net/misc/achromat.html

Polanyi, M. 1958 Personal Knowledge. Routledge & Kegan Paul.

Quine, W.v.O. 1951 "Two Dogmas of Empiricism," Philosophical Review (60) 20-43.

Ravetz, J.R. 1971 Scientific Knowledge and its Social Problems. Oxford University Press.

Robinson, H. 1982 Matter and Sense. Cambridge University Press.

Van Gulick, R. 1993 "Understanding the Phenomenal Mind: are we all just armadillos?" in: M. Davies & G.W. Humphreys (eds.): Consciousness: Psychological and Philosophical Essays. Blackwell. pp. 137-154.


To contact the webmaster, click here
Mail me

Return to Home Page:
Imagination, Mental Imagery, Consciousness, Cognition:
Science, Philosophy & History.

Imagination, mental image, consciousness, cognition