Mary’s Room and stuff
Frank Jackson interviewed by Richard Marshall.
Frank Jackson is the go-to guy on philosophy of mind, epistemology, metaphysics and meta-ethics who invented the Mary’s Room thought experiment and changed his mind and who thinks all the time about the epiphenomenalism of dualism, qualia, conceivability, the defence of conceptual analysis, why Kripke, Gettier and Putnam were doing xphi really, on laptops and quarks, on David Lewis, Quine, Kripke and influence, on networks and theoretical terms and their meanings, on threats to circularity and on beating the drum for reductionism. This one swings a haymaker! Kaboom!
3:AM: What made you become a philosopher?
Frank Jackson: My parents were both philosophers, so I knew a bit about the subject before I went to university. I liked what I knew about it and that led me to take some philosophy subjects as part of my maths/science degree at Melbourne University. I enjoyed the philosophy subjects more than the maths/science subjects. As a result, I took out a pass degree in Science and completed an honours degree in Arts, with philosophy as my main subject.
3:AM: You once thought that there’s more to reality than physical stuff. The Mary’s Room thought experiment is what made this view lively for us. But now you’ve changed your mind about this haven’t you? So can you say what the original thought was and what happened to make you rethink the original position? Are your views now fixed on this matter, or could they change yet again as new ideas occur?
FJ: I started out, right from when I was a first year undergraduate in philosophy, thinking that there’s more to reality than appears in the physicalists’ picture. More particularly, I believed that the sensory side of mental life – the qualia, as it is often put – were additions to anything to be found in that picture. But, as someone taught by David Armstrong and someone who was at times a colleague of Jack Smart’s, I was very aware that I needed a powerful argument to counter their arguments. The knowledge argument seemed to me back then to be exactly what I was looking for. I knew about the zombie argument to the same conclusion – the ‘there’s something more’ conclusion –(mainly through Keith Campbell’s ‘imitation man’ version of it) but the knowledge argument seemed to me to be the stronger argument (of course some dualists take the opposite view). The Mary’s Room thought experiment was my way of bringing out the force of the knowledge argument. What I sought to contribute to the debate was a compelling thought experiment. Actually, the original paper (“Epiphenomenal Qualia“) contains another thought experiment – the one about Fred who can see one more colour (hue) than we can. He sees two reds – I call them red1 and red2 – each of which is as different from the other as is blue from yellow. This thought experiment is described before I give the Mary’s Room one and I have always thought that it carries more weight than the Mary’s Room one. I occasionally meet people who agree with me about this, but we are in a minority it seems.
Why did I change my mind? The Australian materialists had convinced me that dualists had to be epiphenomenalists and, in the last section of “Epiphenomenal Qualia”, I argue that we can live with epiphenomenalism. I came to be deeply sceptical about this. That’s why I changed my mind. A conversion to representationalism about the phenomenal side of mental life also played a role, and is why I do not think I will be changing my mind yet again.
I should though say two things about the arguments in the last section of “Epiphenomenal Qualia”. First, I still think that we should allow that there may well be properties we can know nothing about because they fail to causally register with us. What I find hard to believe is that among these properties are the properties philosophers have in mind when they talk of the sensory side of the mind. Secondly, I should emphasise that, in my view, the key problem for epiphenomenalism about phenomenal properties is not so much that it makes them causally inefficacious but that it means that happenings in the physical world, including our behaviour, fail to carry information about phenomenal properties. In my view, the key challenge posed to dualism by the success of the physical sciences in providing causal explanations of happenings in the physical world is that this success makes it very hard for dualists to allow that the movements of our limbs and the words that come from our mouths carry information about phenomenal properties.
3:AM: An issue that this raises is the role of thought experiments and conceivability. Was the Mary’s room thought experience a case of an appearance of conceiving something (but it was only an appearance), or was it a case that shows that being able to conceive of something doesn’t make it possible?
FJ: I think everyone should agree that it is possible that Mary in the black and white room lacks knowledge of features of the world we inhabit. This is because we should all agree that physicalism might be false (and most philosophers of mind do agree about this). It follows that we should all agree that Mary might, on leaving the room, become acquainted with the relevant extra properties (the one’s physicalists don’t believe are instantiated in our word). So I think – for the Mary thought experiment – that we should ask a different sort of question. The case is both conceiveable and possible but – and here’s the question – were it to be realised in our world, would it be the case that Mary would become acquainted with new properties – new, in the sense of properties she could not have known about while in the room – on leaving the room?
In answering this question, we have to start from the common ground that Mary would have new sorts of perceptual experiences on leaving the room. The debate then turns on the nature of these new sorts of perceptual experiences. Dualists insist that they involve the instantiation of properties that Mary did not know about while in the black and white room. Physicalists deny this. I now think that the best way to make this latter denial plausible is to embrace representationalism about experience, a view which is independently plausible.
3:AM: You defend conceptual analysis in your book “From Metaphysics to Ethics“. Given recent worries about this from the likes of Tim Williamson in his “The Philosophy of Philosophy” book do you still hold to these initial claims that not only is conceptual analysis central to philosophy but that it is still misunderstood?
FJ: A full answer to this question would be, at the least, a long article containing detailed replies to the various objections people have raised to the role I (and many) see conceptual analysis as playing in philosophy, in metaphysics especially. This is not the place for such an exercise. I will simply make three observations to give a sense of why I am unpersuaded by the critics.
First, suppose I ask you whether or not 73 is a prime number. How could you possibly answer that question without possessing the concept of a prime number (where what I mean by possessing the concept is knowing what it takes to be a prime number – not everyone understands possessing a concept in this way, of course)? You might object, what about someone who doesn’t possess the concept but gets the answer by asking someone who does? I reply that what they know is something about what the person they ask uses the term “prime number” for, and that’s not knowing whether or not 73 is a prime number. It is instead knowing something about words. Now what it takes to be a prime number is exactly what you find in maths text books and is the sort of thing I mean by an analysis – of the concept of a prime number, in this case. So the message is that the analysis is need if one is to be able to answer the question. I have given you just one example, but surely there is a general moral to be drawn. Coming to sensible opinions about whether or not so and so’s exist requires, at least in very many cases, having an opinion about what it takes to be a so and so, and what we are doing when we debate one or another analysis of being a so and so is addressing this very question.
The second observation is that we rightly take it for granted that words are wonderful sources of information about the world we occupy. This requires us to have a fair degree of confidence in which situations are covered and which situations are not covered by the words we utter, read and hear. This in turn means that we should not be too sceptical of our ability to say which situations, described in a language we understand, do and do not fall under some proffered classification. Many who have doubts about conceptual analysis have them because they are sceptical of our ability to reliably use the method of possible cases, as it is sometimes called. What I am saying is that I don’t think we should be too sceptical. We should have a fair degree of trust in our intuitions about which cases are and which cases are not, say, cases of knowledge or acting intelligently or pain. How else could words like “It will hurt”, “She knows the answer” and “That wasn’t an intelligent thing to do” be such a valuable source of information about our world?
The third matter concerns a difference I have with many about what we learn from the example of water and H2O. Many say something like the following. It is impossible for water not to be H2O. This impossibility was established experimentally. It is not something that is a priori available. But the program of conceptual analysis is committed to facts of this kind being a priori available. So much the worse for that program. However, when we reflect on whether or some proposed state of affairs is or is not possible (knowing that p when our belief that p is correct only because of some incredible fluke; acting freely when our actions are determined by a past outside our control; etc.), what we reflect on are not the words as such but what they signify. I think “water’s not being H2O” can be read in two ways. On one, what it signifies is the situation of H2O’s not being H2O; on the other, it signifies the situation of the watery stuff not being H2O. And of course the impossibility of the first and the possibility of the second are both a priori available.
3:AM: So xphi has an approach that questions whether presuppositions about folk beliefs in conceptual analysis by philosophers (say about belief and value) are accurate or not. You discuss xphi in your paper “On Gettier Holdouts” and your review of the Knobe and Shaun Nichols book “Experimental Philosophy.” You make a case for thinking that what Kripke, Putnam and Gettier were doing was sort of xphi don’t you? Can you explain your thoughts about all this? And as a result, do you see a point in burning your armchair from time to time as well as sitting in it?
FJ: I think that what Kripke, Gettier and Putnam were doing was pretty much classic xphil. They gave us what experimental philosophers and social scientists often call vignettes, and invited our responses to them, which is exactly what happens in xphil surveys. There are two points of difference however. First, they focussed on the responses of fellow philosophers, whereas experimental philosophers typically address their surveys to wider audiences. But of course when we tell our students about the thought experiments (vignettes) of Kripke, Putnam and Gettier, the audience consists of people who are not professional philosophers. The second point of difference is that when Kripke, Putnam and Gettier first presented their vignettes, they took it, I take it, that there is exactly one correct response to the vignettes, namely, the one they themselves had. (But here I am speculating to some extent.) But in my view part (part) of what we learn from responses to vignettes is something about how we use some given word, and I see no reason to think that there is always one correct answer to that question for a given word in, say, English. It does not have to be the case that what I use “knowledge” for is the same as what you use “knowledge” for.
I am not in favour of burning the armchair but I am in favour of getting out of it. One should leave the armchair to collect the needed empirical information – be it about the results of an xphil survey, the latest information about how the brain represents the environment, the best way of making sense of quantum mechanics, etc. – however, once you have collected the information, you need to digest it and consider what, if anything, it tells one of interest. It is nice to have an armchair when doing that.
More seriously, there are two important points to be made here, as many have urged. First, there is a crucial distinction between collecting experimental data and interpreting it – and interpreting data is not in itself doing an experiment, though it may suggest the need for more experimentation. Secondly, we should distinguish between how we, the folk, or maybe our colleagues, use one or another term, and how it would be good to use that term if we want to speak truly or want to make a distinction that is fit to play a role in one or another theory. Surveys do not in themselves address these two questions.
3:AM: Some metaphysicians seek to resist the idea of having a fundamental framework for discussing reality and say that we can talk about laptops and quarks as both being real so long as we understand the different scales of reality that are in play. This is a kind of Huw Price, Carnapian project that claims frameworks are all delivering on reality and are consistent. These are what philosophers label M-Worlds aren’t they? You say M Worlds are useless don’t you? Why?
FJ: I think laptops and quarks are both real. My difference with Carnap and Huw Price is that I am a Quinean in the sense that I am unable to make sense of different scales or levels of reality, of the whole way of thinking that lies behind the Carnapian project as Price conceives of it. On the Quinean way of thinking, the difference between the existence of quarks and of laptops lies in the difference between quarks and laptops, not between the sense in which they exist. Something similar is true for properties (though, as far as I know, Quine never talks about this). The difference between someone’s having the property of being intelligent and their having the property of being French lies in the difference between the properties, not between the sense in which the properties are instantiated.
I do though make sense of some elements of reality being more fundamental than others, but only in the relatively mundane sense that some things are made up of other things, and their nature supervenes on the bits that make them up and the way they are arranged. Thus, one might say that atoms are more fundamental than molecules. This is not to say that atoms are more real than molecules or that molecule talk is second grade or anything like that.
Whether or not there is an absolutely fundamental level seems to me an open question. There may be a set of entities such that everything that exists either is one of them or is composed of a number of them, but that isn’t something that is settled as far as I know.
I am not sure what you have in mind when you cite me as holding that M worlds are useless. Maybe the following. I do think, and have said, that the Carnapian framework doesn’t help us answer questions about cases where the account we accept in terms a Carnapian would say belong to one framework has implications for the account we accept framed in terms a Carnapian would say belong to a different framework. Take a simple example. We have accounts of people in terms of the language (and concepts) of psychology – someone is happy, believes in god, is intelligent, is sentient, and so on. We also have accounts of people in terms of the operations of their brains, the way their brain states control the way their limbs move, and so on, where these accounts are framed in terms of the language (and concepts) of neuroscience and of the physics and chemistry of the body. I take it that the Carnapian would class these two accounts as belonging to different frameworks. Now the exact relationship between the two accounts – the one in the language of psychology and the one in the language of neuroscience and the physics and chemistry of the body – is notoriously controversial, but we can say this much without courting too much controversy. Sometimes there are major implications for what the account to be given in the language of psychology about a subject, X, arising from the account to be given in the language of neuroscience, physics and chemistry about X. For example, if an investigation of X’s brain reveals that their intelligence-displaying responses to the challenges of life depend on a receiver implanted in their brain that is receiving radio signals sent by someone on Mars (apologies to Chris Peacocke), and it is the nature of these signals which is responsible for these responses, we would rightly infer that, despite the ‘display’ of intelligence, X was not in fact intelligent. It is very hard to see how the Carnapian way of thinking could handle these kinds of cases in a principled way.
3:AM: So what do you say is the relationship between, say, laptops and quarks?
FJ: Quarks are parts of laptops but laptops are not parts of quarks.
3:AM: In your paper with David Braddon-Mitchell on “The Teleological Theory of Content” you discuss the issue of purpose in biology in ways picked up by Jerry Fodor in his recent attack on the coherence of the view as used in contemporary literature regarding ‘natural selection.’ He thinks that ‘selection for’ is just impossible without minds and that Mother Nature is mindless. What do you think about this?
FJ: Our paper argues that content properties aren’t selectional properties in the sense of selectional properties current in the literature on teleological theories of content. As far as I can see, this issue is orthogonal to the one Jerry Fodor is addressing.
3:AM: You have written admiringly of David Lewis who is well known in philosophical circles but is probably obscure to most people. You make the claim for him that he is one of the rare few who has reshaped the way philosophy is done and you talk about him alongside Kant, Hume and Plato. Can you say why you find Lewis so outstanding? Is it fair to characterize you as being a Lewisian philosopher corrupting the youth of Australia?
FJ: I think that the three pre-eminent figures of recent analytical philosophy – or perhaps I should say, of those areas that have especially interested me – are W. V. Quine, David Lewis and Saul Kripke. I guess they are all relatively unknown to the public at large. Although they all write very clearly, their work is by and large devoted to questions that interest professional philosophers in the analytic tradition, which limits their audiences.
What makes them so special? The scope of their work, its influence, and the way it sets agendas (and in good ways).
I take it that the reference to the youth of Australia is an invitation to comment on influence. The three big influences on philosophy in Australia since about 1960 have been Jack Smart, David Armstrong and David Lewis – or at least on philosophy in the analytic tradition. They are in broad agreement on many topics, most famously on the nature of mind and the centrality of metaphysics. They also write with clarity and directness without sacrificing penetration and rigor, and – in my view – an awful lot of what they say is true. Of course, I would speak of enlightenment and not corruption.
3:AM: Do you broadly follow David Lewis in thinking that theoretical terms get their meanings from the networks they inhabit and that for this reason you think that folk theories about the mind and also of morality are likely to be endorsed? Can you say something about this and how it ends up denying eliminativism?
FJ: I think this is true for lots of theoretical terms, not for all. Among those that I think it is true for are mental state terms and moral terms, as you note.
Why do I think that mental state terms get their meanings from their role in a network? I think that the most obvious fact about mental states is the role they play in explaining and predicting the ways we interact with the environment. We predict and explain what Hilary Clinton or the person ahead of us in the supermarket queue will do in terms of what we think they believe, desire, hope, are fearful of, etc. I could go on and on about this, but of course it is a theme in the writings of philosophers as diverse as Elisabeth Anscombe, Paul Churchland, David Lewis, Jerry Fodor, David Armstrong and Robert Stalnaker, e.g. I think this is – or should be – uncontroversial. I also think – and this is more contentious – that the correct response to this observation is, first, to think of mental states as the posits of an explanatory theory of behaviour, where the posits are defined by their connections to each other and to behaviour and the environment, and, second, to think of the terms for mental states as referring to these states.
To give a rough idea, one clause in the explanatory theory might read “belief is a state that gets together with desire to cause behaviour that satisfies the desire when the belief is true” (a fancier version would mention strength of belief and of desire, and expected utility), and the view about the terms “belief” and “desire” is that are names for the states that are so related. Here I am following many, including Armstrong, Lewis and Churchland.
What are implications for eliminativism? If you accept a network analysis of mental state terms, the question of eliminativism becomes the question of whether or not there are states standing in the posited relations, near enough. An analogy. How come there is a Prime Minister of Australia but not a President of Australia? The reason is that there is someone who plays the role definitive of being the prime minister of Australia but there is no-one who plays the role definitive of being the president of Australia. The converse is true of the United States, obviously. So the network analysis of mental state terms tells us that mental states exist just if the relevant roles are filled. I happen to think, but won’t argue here, that the relevant roles are ones we can be certain are filled. That’s why I reject eliminativism about mental states. I take it that Paul Churchland’s view is that the relevant network, at least as it applies to the propositional attitudes, is such that we cannot as of now be certain that the relevant roles are filled (but I emphasise that this is my way of understanding his position).
You also ask about morality. In work with Philip Pettit, I have argued that moral terms are defined by their place in a network, a network we might call folk morality. Again, this view says nothing in itself about whether or not one should be eliminativists about morality. It tells us rather what the issue turns on. Are there or are there not properties that stand in the needed relationships? The matter is complicated by the fact that, in my view anyway, folk morality is evolving; we are still negotiating its key clauses. I am however an optimist. I think we will end up with a theory – mature folk morality, I call it – which is such that there are properties that stand in the needed relationships to vindicate the theory. Some of my colleagues wonder at my optimism.
3:AM: How do you deal with the criticism that your explanations are threatened by vicious circularity? I think your theory of causality was developed to head off such worries. Can you tell us something about this issue?
FJ: I avoid the threat of circularity by, as one might put it, tying the networks down via their connections to matters outside the networks. Let’s do a couple of simplified worked examples to give the idea.
Take first the earlier example of belief-desire psychology. Suppose I told you that belief was a state that got together with desire to cause behaviour that satisfied the desire when the belief was true. You might well worry about circularity. I’ve told you what belief is in terms of what it does in tandem with desire, and what desire is in terms of what it does in tandem with belief. How, in that case, might one differentiate belief from desire?
However, there is way to differentiate belief from desire. As it is often put, belief and desire differ in their direction of fit to the world. Belief seeks to fit the world; desire seeks to get the world to fit it. Add those two clauses to the simple network analysis I offered above and there is no longer the threat of circularity. Here the ‘tying down’ is done by the difference in causal connections to the environment.
Here’s the second example. Take, to start with, the traditional conception of marriage. Wives and husbands form networks: wives are married to husbands, and husbands are married to wives. All the same, we can specify what it is to be a wife without circularity, and likewise for being a husband. I have inserted the specifications immediately below.
x is a wife if and only if there is a y such that x is married to y & y is male & x is female.
y is a husband if and only there is an x such that x is married to y & y is male & x is female.
There is no circularity in either of these specifications.
In this example, the ‘tying down’ is done by gender. That naturally raises the question of how we might specify being a wife and being a husband in a way that applied to same sex marriages. In these cases, another way of making the wife-husband difference will have to be employed, and if none is forthcoming, there would be a circularity problem. But that’s no objection to the network analysis. It is the correct result. If one talks to same sex couples, they say one of two things – that the distinction between being a wife and being a husband makes no sense inside their relationship, or that it does, but it is marked in a different way from the traditional one in terms of gender.
I don’t have a theory of causality; wish I did. But I did appeal to causation to meet the charge of circularity when discussing belief and desire – they were differentiated by the different way they are causally connected to the world – and maybe that is what you had in mind.
3:AM: Why shouldn’t we worry about the Swampman?
FJ: I am not quite sure what question you have in mind here. A quick answer might that the probability of Swampman is so small that there’s nothing to worry about in practice. We will never meet him, so we do not need to worry about hurting his feelings or whether we should give him the vote. Perhaps your question is rather, why I am I confident that Swampman would have thoughts and feelings? My reason is that I think this is very much the most plausible thing to hold, and that the arguments to the contrary are weak ones.
3:AM: In your paper on “The Autonomy of Mind” you make a cool point in the first section “How not to win the Nobel Prize” by imagining the bafflement of Karl von Frisch not being able to make sense of the notion of an autonomous mind. It strikes me that this issue is a warning against all sorts of reductionism, not just in the philosophy of mind. Is this also the idea of Galen Strawson’s Panpsychism which is about properties emerging from nothing more than aggregation. But if aggregation produces, say, powerful cars out of aggregated non-powerful parts, why doesn’t this mean that the temptation to talk about the autonomy of powerful cars should be resisted. Similarly, how does this prevent minds being reduced too?
FJ: I think you may have misunderstood the point I was seeking to make in that section. Let me start by quoting from the section (it is a small fraction of the paper so I don’t think there’s a copyright issue). Here’s the quote:
“Karl Von Frisch was awarded the Nobel Prize in 1973 for decoding the honey bee dance. Imagine, however, that mid-way through his researches he gave up and wrote a letter to his co-workers containing the passage below
Thank you for your work on this demanding project. We have accomplished two things. We have confirmed that the bees get information from the dance, information that allows them to find sources of nectar. This means that there is a property in the dances reliably correlated with nectar locations. We have also shown something important about this property: it is a matter of the geometry and kinematics of the dance, and its orientation with respect to the sun and the vertical. My colleagues in the philosophy department have told me a good way to express this: say that the property supervenes on the geometry and kinematics of the dance, and its orientation with respect to the sun and the vertical.
We have, however, been unable to articulate the crucial property of the dance. Exactly what it is about the geometry etc. of the dance that delivers the location information to the bees has defeated us. We have tried many formulae but each has been found defective in one way or another. At first I thought our failure was due to the complexity of the task and that we might succeed if only we worked a bit harder, were a bit smarter or a bit luckier. However, my colleagues in the philosophy department have opened my eyes to another possibility. They express it in different ways. Sometimes they express it by saying that the relationship between the nature of the honey bee dance and the location of nectar is sui generis. Sometimes they express it by warning against reductionism. Sometimes they express it by talking of the autonomy of the location of the nectar with respect to the nature of the dance. Sometimes they say that the properties of the dance are shapeless with respect to the location of the nectar. But the words they use aren’t crucial. What is crucial for our project is the message: what we were attempting is impossible as a matter of principle, not complexity. The honey bee dance is uncodifiable. This is why I have abandoned the project. (One small consolation is that we can be sure that no other research team will succeed where we have failed.)”
I hoped that readers would find it obvious that anyone who wrote such a letter was seriously confused. If the property of the dance that the bees use (sub-personally) to locate the nectar supervenes on the way the dance is put together and its relationship to the position of the sun and the vertical, it must be possible to specify the property in a way that allows one to a priori derive it from the nature of the dance and its relationship to the sun and the vertical (this is in fact what von Frisch did). The rest of the paper is devoted to arguing that as it is for the property of the bee dance so it is for psychology. In other words, I was beating the drum for reductionism, in one sense of that term. I now realise that the title of the paper is potentially misleading; the paper is about the autonomy of mind in the sense that it is about why we should not embrace it.
You mention Galen Strawson. I believe he and I agree on the meta-philosophy here. We are both suspicious of what is sometimes called brute emergence. We humans are complicated aggregations of physical stuff. Much of science is devoted to studying the nature of the bits and the way they fit together. How then is it that we are conscious? Strawson and I agree that in order to answer this question, we have to show how the nature of the bits and the way they are put together a priori entails that the aggregation is conscious. I think that the way to do this is to embrace analytical functionalism; he thinks that the way to do this is to embrace panpsychism. That’s a big difference between us, but we agree that the wrong way to go is to hold that consciousness emerges brutely, by a sort of magic, from the aggregation. The consciousness must be derivable in the sense of being a priori deducible from the details of the aggregation.
3:AM: Ha! I totally screwed up that last question! It’ll not be the last time! Thanks for clearing it up for me! Now, you appeared as a character (as yourself) in one of David Lodge’s novels. When not doing philosophy, are there any books or films that have been particularly enlightening for you?
FJ: I don’t know about enlightening. My three favourite films are The Third Man, Casablanca, and whichever of the classic westerns catches my fancy at the time of asking. When I am not reading philosophy, I read anything relaxing, often crime fiction, sometimes crime fiction which has genuine literary merit (P. D. James, for example) but not always, I fear.
3:AM: Finally, which five books should we philosophically curious readers be reading to be further enlightened by these issues (not counting your own of course, which we will be dashing away to read straight after reading this!).
Armstrong, David. A Materialist Theory of the Mind.
Crane, Tim. The Mechanical Mind.
Lewis, David. On the Plurality of Worlds.
Ludlow, Peter, Yujin Nagasawa and Daniel Stoljar, eds, There’s Something About Mary: Essays on Phenomenal Consciousness and Frank Jackson’s Knowledge Argument.
Smart, J.J.C. Philosophy and Scientific Realism.
ABOUT THE INTERVIEWER
Richard Marshall is still biding his time.
First published in 3:AM Magazine: Friday, March 7th, 2014.