The Reductionist fallacy

Ard: I remember when I first learned basic statistical mechanics and how to calculate ideal gas law from the interaction between the atoms, and then I began to calculate more complicated gas laws and then equations of states. It’s a very powerful and beautiful thing.

GE: Yeah, it’s beautiful.

Ard: And so it’s not so surprising, sometimes, that if you’ve had that experience, you think this is the best way of understanding the world.

GE: Yeah. And in my view, the scientifically legitimate thing is to try and extend that as far as you can, so you can see the widest domain that you can explain by it. The problem comes if you try to say this is the only thing which is happening; it’s the only kind of causation that is possible. That’s where the problem comes.

And so I have this definition of a fundamentalist. My definition of a fundamentalist is someone who takes a partial truth and claims it’s the total truth, and that’s exactly the problem here.

If you say that statistical physics explains a lot of stuff, there’s absolutely no problem. If you claim it explains everything, including life, it’s simply not true. And I think a lot of the problem here is that the people who are wanting to say that all we’ve got is the bottom-up causation... they are just… one should pursue bottom-up causation as far as one can, and it’s very powerful, but it’s not the only thing in existence.

In fact I make quite a strong claim there. If you want to create life, there’s a level, a ceiling that you can reach by bottom-up assembly of molecules and so on, and you can’t get any further for one very simple reason: after a certain time, adaptive selection has to come and you’ve got to start adapting to your environment. Now when that happens, the environment has to feed in signals to this organism which will change either its structure, or its behaviour, or both.

Now, that’s top-down causation from the environment into the structure of the organism at the macro level in your organism and down to the electrons and protons in the organism. If that top-down causation does not come into existence, you can’t get life, and so I make a strong statement that purely bottom-up causation will not bring life into existence.

David: Ah, that’s very interesting. Because you are saying it feeds information in, and you have said that before: information is the new thing.

GE: Yeah, what’s DNA about?

David: Life is information: it’s coding. And information is a real thing, but it’s not physically real is it?

GE: No.

David: Because the information is different from the way that it’s encoded.

Ard: And the information is coming in from the environment.

David: That’s fantastic.

GE: The whole point of DNA, the wonderful discovery of DNA, is that it can, like a Turing machine, it can encode any protein structure you like, any one at all, and the physics doesn’t determine which comes out. So if you’re a tele-physicist, predict for me what the DNA sequence is going to be. You’ll shrug and say, ’I can’t do it’, because this ain’t physics, it’s biology. It’s a different logic, as it were.

David: So is that an emergent level then?

GE: Yes, absolutely. But why people believe in reductionism is the unbelievable success. And you experience it, as you were saying, with the kinetic theory of gases. It works absolutely beautifully. The atomic structure of matter has been a huge success. Then there was the physical structure of chemistry. The wonderful, wonderful discoveries of Linus Pauling and all these people of how chemical structure emerged, the quantum physics. Wonderful success.

Then there was the molecular biology revolution with Crick and Watson and so on, and life is molecules interacting with each other, and there’s this DNA coding and so on. Wonderful.

And then there was the whole thing of neurons and how you could understand the mechanism of the brain in terms of electronic impulses going down axons and dendrites according to laws which are perfectly understandable in terms of the underlying physics.

And so the reason people pursue it is because it’s been so incredibly successful in a most extraordinary kind of way. But, in each of these areas, there’s always been a counter-reductionist thing, like in the case of statistical physics. The arrow of time is still a worm at the bottom of the thing in the case of statistical physics. In the case of chemistry, chemistry is a success in principle but not in practice.

In the case of the genes there’s been the whole epigenetic thing coming back and saying causation is not only bottom up. Epigenetics is the solid statement there’s a mass of top-down causation. And in terms of the mind, you find you cannot understand the mind in bottom-up terms only.

TOP-DOWN CAUSATION

Ard: We’re going to switch topics. That was great. That was really helpful.

David: Reductionism?

Ard: So why don’t you ask about reductionism because that’s the one you’re… So we’re interested in emergence and reductionism.

David: We’ve gone around asking everyone about the two and trying to get them to define both.

GE: And you’ve got 150 different answers.

David: You said it George. So would you like to add one more?

David: And I should stress, we’re not trying to be in any way anti-reductionistic. We’re just saying, what are they? What are its strengths? What are its weaknesses? Is it the answer to everything or not?

GE: Reductionism is the question: can you, in principle, completely explain what happens at the higher levels in terms of what’s happening at the lower levels?

David: And what do you think?

GE: Well my answer is absolutely not.

David: Why?

GE: I think a very good place to start is digital computers because digital computers are something which we can actually understand fully. Now a digital computer, it’s made up of all of these gates, transistors joined together by wires and then put together with a keyboard and a display and so on, and you can look at it at the different levels. At the bottom-most level it’s electrons flowing through the gates in a particular way, and that’s what controls what happens at the screens.

But given that hardware – there’s the hardware sitting there – it doesn’t do a thing until you load it up with a program, and what the identical hardware does is completely different depending on the software you load. So if you put in a word processing program, a music program, a paint program or something, you have completely different outcomes from the identical hardware. So the hardware, per se, does not determine what happens. What determines it is the program you load in. So now the question is, what is the ontological nature of a digital computer program?

David: When you say ontological…?

GE: What kind of existence does it represent? And now the hard-line reductionist will say that it’s nothing but electron states in these gates. Now that is completely wrong, and a useful side-step here is to think about what is the ontological nature of a book.

Now in the old days we all thought of a book as paper with printing on it and covers, and that’s a book. Now the question is, is it the same book if it’s got a different size print? Is it now still the same book? Is it the same book if you print it in italics or not? And I think we all agree it’s the same book. It’s different instantiations of the same book.

Nowadays we know that the book can be given to you in electronic form. Is the electronic form still the same book? And my answer is yes. The book is the abstract quantity which can be instantiated in those different ways. Now a computer program is the same: it’s an abstract entity; it’s not a physical thing. And so some people say, ‘But it’s here: it’s the CD disc.’ But it’s not the CD disc. Is it the excited states in the gates? That’s an instantiation of the computer program. And actually a computer has got layers of structure: there’s the machine language, the assembly language, there’s the operating system language, there’s the high-level language, and what is the key thing here is that there’s a logic in the computer program.

That abstract logic then gets written into a high-level code. What happens in the computer is absolute magic. The high-level code gets typed in and then interpreters or compliers write it down in to the lower-level languages.

Exactly the same logic is present at every level, but it’s represented in a different way with different rules, and at the bottom level it gets turned into instructions at the gate, and it gets turned into electronic states in the gates. Now, what is the computer program? It’s the equivalent class of all of these representations: it’s an abstract thing. Does it have causal power? Yes, it causes things to happen.

And very, very similar to this is the rules of chess and the rules of football. Those are abstract agreements and they can change with time. Now, the rules of chess determine what happens on the chess board, and I like to imagine this Martian coming and watching chess – after a while he works out these different motions, and then he picks up the rook and he looks underneath it. ‘Is something causing it?’ And then he can’t figure that out, and so he invents a force field which acts on rooks and only enables them to move in different force fields…

David: Looking for a physical explanation?

GE: Looking for a physical explanation. It’s a mental explanation. It’s a set of rules, abstract rules, and again they’re abstract, they’re not physical things. And they can be realised in a computer program or in your mind, or they can be realised in a book and so on. And again this multiple realisation, whenever you’ve got multiple realisation of some concept, it’s telling you that there’s causation from the concept to all of these realisations.

David: So you’re saying an abstract logic has this power?

GE: An abstract logic has physical outcomes in the real world through being what is implemented in a computer. So then, of course, the old philosophers of mind would say, ‘but you’re talking a dualist position.’ My answer is, ‘yes I am.’ A computer is a dualist machine: there’s the hardware and the software.

David: So you’re saying there’s a difference between the mind and its ideas, and the brain?

GE: Yes, I take the completely unpopular position: I’m a dualist. There’s the mind and the brain, and the mind inhabits the brain… or thoughts, thoughts inhabit the brain and thoughts are not physical things. Thoughts are abstract things which get represented in a physical way.

And again we do not understand how this happened, but the brain has a hierarchical structure. Thoughts have a hierarchical structure, and in the computer you can see these different levels. You can understand them, and you have got these interpreters or compilers which do it. I think eventually when we understand the brain enough, we will see exactly the same kind of structure happening in the brain.

The logic is going top-down from the top level down to the bottom and it is the logic which is controlling what happens at the bottom level. Abstract entities are driving the physics at the bottom level. The physics is not controlling what happens.

David: So that is exactly the opposite of when people say the reductionistic picture is always once you’ve understood these things at the bottom, they are what cause things to happen?

GE: Yep.

David: The cause always runs from the bottom up?

GE: Well it goes both ways, and there’s a large literature on what is called supervenience.

David: What is supervenience?

GE: Supervenience is if I knew everything about the bottom-level state – and let’s take the computer as an example – so I know the wiring and the chip; if I knew at an instant all of the states of the electronic gate, and then I reproduced it over there in another computer in exactly the same way, then it would give you exactly the same high-level stuff there. So at that level the reductionist picture is correct.

But the question is, how did those bottom-level states get to be in the excitation states that they are? And that can only happen top-down from the logic controlling what happens at the bottom level. And something similar like that will be happening in the mind.

David: I was just thinking about why so many scientists are really wedded to reductionism, and is it because they think the only thing that you could add into the universe, that isn’t what the reductionist story has in it, matter, that they think you’re going to import the supernatural? Whereas you’re importing something that is not supernatural. You’re saying there’s a realm of ideas, which isn’t supernatural.

GE: No, it’s not supernatural in that sense. The ideas don’t reach down and act of themselves.

David: No.

Ard: Like ghosts.

David: No, true, okay.

COMPLEMENTARITY

David: Running through our conversation has been this sense, listening to you, that ideas, in your example ideas, that you find beautiful, are very powerful: they can guide what you think and what you do next. Do you think that ideas really do have that power? That they are as much a force in the world as the basic forces you are thinking about?

Because non-scientists tend to think physicists are always going to say that everything that happens is just these particles bumping into each other. And here you’ve been talking about it as if thoughts themselves shove the particles around: it was the thought which led you to the next idea. Do you see what I mean?

FW: Yes, I think I do.

David: It’s saying the radical reductionism that says everything that happens is determined by the bumping of the elementary particles, taking Laplace to the extreme…

FW: There’s a concept that Niels Bohr elaborated that I’ve fallen I love with that I think is very profound: it’s called complementarity. It’s something that’s just a true fact about quantum mechanics but is much more general. It’s the idea that you can have different descriptions of the same thing that are both valid, and both important for answering different kinds of questions, but they are mutually incompatible. It’s closely connected to wave particle duality. There are different ways, mathematically, of processing the wave function. I think that’s a much more general principle. Well I know it’s a much more general principle, and I think that it even applies to things like the problem of free will versus determinism.

There’s one description based on physics that tells you that, for practical purposes, in principle, the brain, for instance, is a pretty nearly deterministic, noisy, but deterministic system. Whereas if we want to deal with our own experience, if we want to interact with other people, if we want to have sensible systems of law, we need to use this concept of free will. And they are both valid descriptions, but they are meant to address different aspects.

David: But you think they are both genuine? Because listening to you, it was on the tip of my tongue to say there’s the rejoinder which says, ‘Yes, but that stuff about free will: that’s just a place-holder until we’ve really worked out the physics, and really, ultimately, there’s no such thing.’

I’ve talked to scientists who have said there is no such thing as free will; there is no such thing as the self; there’s no such thing as consciousness. I’ve had all those things said to me in the last five or six years, and the answer is always the same: ‘I know you think you’re conscious, but actually it’s just the chemicals in the air and they’re buzzing around and that’s why you’re laughing, it’s not because anything is funny.’

FW: This reminds me very much of Dr Johnson after hearing Bishop Berkley’s sermon about the unreality of matter, and famously Boswell said that this seems crazy but it’s impossible to refute. And Johnson kicked a stone and said, ‘I refute it thus.’ So when people say there’s no such thing as consciousness, come on!

David: I’m with you on that, but there is that sense that…

FW: Okay, we can illuminate it. It’s like saying there is no such thing as life because we can understand it on a molecular level. Life! There are useful concepts that are necessary in describing large domains of experience that aren’t going to go away. They may have alternative complementary descriptions.

David: But you think they’re real?

FW: Yes, I think they’re as real as anything can be. They are useful. They describe actual things in the world.

David: And ideas would be one of them? That realm of ideas?

Ard: And I think what you’re saying is that they may be contradictory if you try to have them at the same time?

FW: Yes, if you try to apply them at the same time, they can lead to contradictions.

Ard: But, nevertheless, they are both true if you are asking the right kind of question?

FW: They are both useful and necessary, I think, in addressing different kinds of questions.

Ard: In your book you say, for example, objects and persons are complementary.

FW: This is very much relevant to the discussion of free will. When we are trying to predict what they’re going to do, how we should interact with them, we are thinking of them in one way. If we are thinking about them as what their height is; what their mass is; what they are as physical objects; am I going to run into this object? If I’m diagnosing a disease, I want to analyse the chemistry. That’s a different way of thinking.

Ard: And those two ways are complementary to each other?

FW: They’re complementary. They’re both valid, but there’s a lot of tension if you try to apply them at the same time.

Ard: That’s really interesting. That’s a profound point that they’re both valid ways, so the world of the person is the world of intentionality, and the world of ideas as well. And then there is the world of the human being as a wet computer, a bag of chemicals, and they’re both useful. They may even illuminate one another, but if you try and apply them at the same time, or if you say that one is superior to the other, you’re missing something.

FW: Yes, that’s a good way to put it: you’re missing out. It’s not that you’re wrong in some sense, well I think actually that you are wrong, but at the very least it becomes very awkward to express concepts of intentionality, of emotion, of ideas in physical terms.

David: So are you sympathetic then to people like George Ellis who think emergence is something that’s a useful idea?

FW: The general concept, I think, is very much in line with these ideas about complementarity: that different levels of description can be useful, and if you’re describing the same thing in different ways, it’s also important, if you can do it, to make those consistent and get a rich interconnection between the two. So yes, I have a lot of sympathy for that. But where I don’t go is, I think our description of the world based on what Newton called analysis and synthesis – and I would like to call it that, but many people call it reductionism – the idea that the way you build up the description of the physical world is by understanding very small parts thoroughly and getting a complete description of those and then building out from there. That’s been extremely fruitful, extremely successful, and I don’t see any sign that it’s open to influence from higher levels. So specifically, for instance, at an accelerator, I don’t believe that no matter how hard you think about it, you’re not going to change how the protons collide.

David: It does sound a little bit like you’re trying to have your cake and eat it too. Obviously the accelerator, you chose the example because it works for what you were just saying, but then let’s take the mind again: does that mean the next word I’m going to say is forced upon me by the particular configuration of all of those particles? Or was it that I was having a thought and the logic of the thought dictated what I would say next?

FW: If you are discussing thoughts and how they evolve, then the appropriate language is cognitive language. That’s the way that people have found useful and have developed to express and analyse the development of thoughts. If you want to know about brain events, especially at a microscopic level, then you want to use quite a different description. Those can both be valid descriptions.

David: They have to be, in a sense.

FW: I think they are, but for different questions. If you want to use thought patterns to predict how chemical reactions and atoms are going to behave, I don’t think you can get very far. There are many, many different chemical reaction patterns and things that could correspond to the same thought, and conversely only slightly different chemical patterns could lead to grossly different thoughts. So it’s not a stable, simple mapping. It’s very complicated at best, and practically useless. So that’s why you have complementary descriptions of these different phenomena that are used in very different terms and superficially incompatible – and maybe not only superficially but just plain incompatible – but, in any case, appropriate for answering different kinds of questions.