After listening to the first half of the “Donald Hoffman with The Harrises’” podcast I wrote down a bunch of thoughts here. I found little to disagree with, because so much sounded like something between mainstream ancient philosophy (we don't have direct access to ontological reality) and the work of David Deutsch in "The Fabric of Reality". So in the second half I was looking forward to hearing something controversial. It did not disappoint. What I want to say at the outset is that although I ultimately disagree in places on points of epistemology and science and even whether Donald's theory even constitutes science, I think his overall approach is interesting and there’s no easy way to refute what he is saying. But this is just to say: almost any theory of consciousness is not easy to refute. Panpsychism is not easy to refute. As we have no theory of consciousness nor even a good working definition of what it even is, all approaches are more or less welcome at this point. So I want to say upfront: I think this is useful work. It’s worth trying. Almost anything is worth trying. That said, let’s get into trying to explain what he is saying and what criticisms one might conjure in response. Also: I actually read a couple of Donald’s papers to supplement his podcast. At times he seemed to deviate (especially under questioning from Annaka) with what he has said in published material. I’m going to stick with his words rather than try to argue with “the two Donalds” as I see them (spoken and written Donald). For what it's worth I read http://cogsci.uci.edu/~ddhoff/Chapter17Hoffman.pdf and also link.springer.com/content/pdf/10.1007%2FBF03379572.pdf and may get to commenting more on those in more detail at some other time.
Donald wants to find “a mathematically precise theory of consciousness” – and by this he means something like looking at all the ways consciousnesses interact (note the plural there: so we are conscious of course, but it’s not clear “how far down the phylogenetic tree” Donald sees consciousness as running. At times he even speaks as though he agrees with panpsychism and Sam basically seemed to put him in the panpsychist box at times – and I tend to agree) – but whatever the case he wants a way of mathematizing the relationships between individual instances of consciousness. So he talked about, for example, big data and how we might use that to analyze how humans interact online and everywhere else. If he could analyse these interactions or model them in some way then he could have some sort of mathematical description of what human consciousnesses are doing. This was an interesting approach to my mind because he was contrasting it to the “usual” way of doing things which, as he explained, was basically to look “in” matter for consciousness. So most people begin with neurons and try to work their way to consciousness. They start with the physical and try to get to the mental. Donald said he was trying to take the opposite approach. He’s doing this, he says, because he’s motivated to find a “dynamics” of consciousness. But for this he needs a mathematically precise definition of consciousness. He says he has such a precise definition. Now I should say: definitions are one thing – but people have tried defining something like “the electron” again and again and it’s no easy task. And, along with Popper, definitions are possibly misleading and always fallible anyways. What we need are explanations. So right away Donald is sort of in an instrumentalist camp. We will return to this. Whatever his "mathematically precise" definition we never actually hear what the rough-in-English definition is.
Whatever the case – at the 1 hour and 14 minute point in the podcast I hear the first remarks I can really begin to raise my eyebrows at – although not disagree with out of hand. What he claims is that his notional theory of consciousness – his “conjecture” one might say – is that consciousness is the most fundamental thing in the universe. So it is more fundamental than quantum field theory and general relativity and evolution by natural selection and indeed everything else we basically know of science. So his theory, if correct, must be able to generate quantum theory, general relativity and evolution by natural selection and everything else. And indeed he'd be right that the latter follows from a theory purporting to be the most fundamental of all. He says at one point “All of these scientific theories are wonderful tools but they have only been the science of our interface, not science of objective reality.” On this view even space-time is part of our “user interface” that he defined in the previous hour.
Now here is where I may part company with him. What is the role of science then? And explanatory knowledge generally? If we are trapped inside this user interface – a sort of Platonic Cave – then science becomes about us - a parochial arm of psychology. It's about what we are doing and how our brains (or minds, I'm not clear) are doing. But I'm not sure on his thoughts about the relationship between brain and mind (see below). Even general relativity, on this view, is mainly about us – and not a separate space time. It seems on Hoffman’s view space time is not “out there” but rather “in us” in some way - as the manifestation of an interface with objective reality or an interface between conscious creatures, or something. Now I am of the view: general relativity is a theory of stuff out there – it describes a really existing space time. Now could it be the case that general relativity gets refuted? Not only do I think this could be the case, I think it must be the case. In some ultimate sense it has to be false because (1) it disagrees with quantum theory so there must be a deeper theory than both (2) as a matter of principle all theories are improveable – we cannot have a final theory. If general relativity is the final word we must ask: WHY? And if there is no answer to that question we can find, progress stops and we are doomed because some problems are insoluble not because a physical law says but because our capacity to understand the universe is bounded by a brain that is unable to grok something about physical laws. This would mean our mind is not universal (now I just as an aside admit here that in some places – namely here: http://cogsci.uci.edu/~ddhoff/Chapter17Hoffman.pdf) that in Hoffman straight up admits: the mind is not what the brain does (see page 11). I am unconvinced and he never grapples with the universality of computation anywhere - not in the podcast and not in his two papers that I read. This is a deep problem for his thesis, so far as I can see it. And his quick dismissal of actual, literal, realist quantum theory is probably also disqualifying. But then, few physicists take quantum theory as being literally true. We’ll come back to this also.
The idea that spacetime is part of us, or of “consciousness’ user interface” seems to me to be the mistake of thinking that a scientific theory is at best a fiction – perhaps a useful model – but not really connected with reality. This is an epistemological claim. A realist says: no – although ultimately false, scientific theories are not merely fictional though useful models. They are accounts of what is actually there. There is a connection, therefore to ontology. Hoffman seems to be discounting this. Even if some consciousness theory is deeper than all other scientific theories this does not mean those other scientific theories never have a basis in objective reality. Consider by analogy any other demonstrably false theories. I’ll go to my favorite example: Newtonian Gravity. It says that there exists a force between particles that varies inversely as the square of the distance between them. Now that theory has been shown by experiment to be false. But is it utterly false? A complete fiction? No. There really is a way that particles influence each other in space (so that's correct) but it is via the spacetime between them. Einstein’s general relativity replaces Newton’s theory but it does not say of it: utter fiction. The motion of things in space does indeed approximately obey the inverse square law. It’s not a random guess. Approximations are approximations to something: and they cannot be approximations to simply other complete fictions. They have to be getting to something if the predictions being made are reliable to some extent (more than random guessing).
Whatever the case: Hoffman argues that because his theory of consciousness means consciousness is more fundamental than space and time that it means consciousness itself is outside space and time. Well…that’s rather neat. I don’t think there is the slightest bit of evidence refuting the claim that consciousness is IN space and time, but there we have it for now. We need an experiment (and more precise theories too) in order to decide between the two claims: consciousness is inside space time vs consciousness is outside space time.
Annaka asked a question about “The Fundamental Nature of Reality” and this comes up more than once. Hoffman responds at this point that his theory of consciousness would indeed answer to that label. So, ultimately, Hoffman believes in an end-to-science: he thinks there can be a final theory and that he is on the way to finding it. It won’t explain “everything” – he concedes that – but within the realm of fundamental physics – he would have a final theory. It would be non-improveable (he doesn’t say this: I’d like to ask him – but his language is telling “final” and “fundamental nature of reality” and so on). So while he doesn’t think quantum theory or general relativity or natural selection can be the final word (quite right) he does think his theory could be.
But once we have that theory, why can we not ask: why isn’t it some other theory? Like if someone came up with an alternative – why must it be ruled out a priori? Will his theory explain why all other theories are not possible? And if so, will it then explain why his mathematics is infallible? Why he cannot possibly have made a mistake and so on? These claims to “absolute final theories” all run into these questions. The alternative is: however deep your theory goes, that is no guarantee you’ve reached the bottom. There may be no bottom. There may just be better and deeper places to delve.
Hoffman speaks of his theory of consciousness as describing consciousness as a “field” not unlike the fields in a physical theory. I suppose he might think – though he didn’t say this – that particular instances – your mind and mine – are excitations of that field rather like the way electrons and photons are excitations of the quantum field. I’ll need to read more about this to find out but that could be a neat idea. Annaka rightly observed that this means this would be a field outside fields – so a field outside spacetime. But Hoffman says he is using the word field in a different sense to that in quantum field theory but admits the conscious field will be outside spacetime (see the 1 hour and 18 minute mark).
Soon after Sam interjects with a question about panpsychism. Incidentally Sam says a thing he often has: consciousness is the one thing that cannot be an illusion (following Descartes). Hoffman is no fan of panpsychism and rejects the label because he says unlike the panpsychists he is not starting with the material world as an assumption. So Sam later suggests he is an idealist. Hoffman also rejects that label. On the "cannot be an illusion" thing, I used to think the same. I was talked out of it by David Deutsch. My own way of putting this now is: what's not an illusion? Consciousness? What about that string of letters? Does it infallibly label what you think it does? Is that an illusion? If it could possibly be, then what exactly is not an illusion? And given we know that this moment actually is being experienced in the future in some sense (because neurons take time to fire), then consciousness is already a complicated thing that could be illusory. But I digress.
Hoffman makes some interesting remarks to further explain his “user interface theory”. He says: when you look in the mirror you see skin, hair, teeth, a face and so on and yet you know there are emotions, sensations, hopes, aspirations there. So the face is like a portal – it’s an icon so to speak – which gives you access to consciousness. So too with other people. Now I quite like that. But he then goes onto say that a cat is an icon for consciousness too although it’s “dimmer” (his word) – he thinks this as he can tell his cat likes some food but not others. Hoffman then goes even further: a rock or a glass of water is an icon of consciousness. Now at one moment he rejects the idea that the rock or water are themselves conscious – instead he says the consciousness is “behind” those icons in some way. And some of those consciousnesses are greater and lesser than ours. This is where he lost me. He seems to be saying that what we see as rocks or glasses of water and so on are the means and ways in which we conscious entities interact. At least that’s my steelman version of what he is getting at. After all – were there no physical objects out there, what point would there be in consciousness and what would we have to interact about?
Sam rightly suggests that this is all panpsychism but it’s in virtual reality.
Hoffman dismisses the quantum multiverse for poor, sadly incoherent reasons revealing a level of ignorance about quantum theory. He says, for example, that it “does not explain the measurement problem”. I do not have the time here and now to explain entirely why that’s wrong but here is a brief sketch:
The so-called “measurement problem” is the question as to why before a measurement is taken in quantum theory, the mathematical formalism describes a system (say a single electron passing through a slit) as taking all physically possible paths. Upon measuring it on the other side of the slit (the observation) we see only ONE path having been taken. But the very point of the multiverse is to take seriously all those paths described by the formalism as equally real before, during and after the measurement. They all exist. But you only ever see one because you only occupy the one universe (in a sense, speaking loosely). But quantum mechanics describes many many universes. This is really not that mysterious. It’s rather like how even Kepler’s laws describe an orbit as consisting of many many points in space…and yet Earth is only ever observed at some point in time to be at one of them. The others really exist though – but we’ll never observe the Earth at all of them. Sure: we can take photos, perhaps when it’s somewhere on January 1stand then 2ndand so on. But we can do the same for the electron – repeating the experiment and noticing it really does take all the possible paths. Whatever the case, Hoffman fundamentally – deeply misunderstands what quantum theory says and how the multiverse solves these misconceptions people have. (Look forward to an upcoming ToKCast for this in just two episodes when we go through “The Multiverse” chapter of “The Beginning of Infinity”…or if you can’t wait – read it yourself now!). For anyone playing along I've written a number of short articles on The Multiverse like this one.
Hoffman is thus also confused about the relationship between quantum theory and probability. I recommend https://www.youtube.com/watch?v=wfzSE4Hoxbc which, in a strongly competitive field of amazing talks by David Deutsch is right near the top for me. The long and short of it here is that: randomness is always just subjective. In the multiverse everything that can happen really does happen. So it happens with Probability = 1. Probability was invented to explain games of chance but has been repurposed and perhaps even perverted for all sorts of things that it perhaps shouldn't be. Now in the case of the multiverse some things happen more often (with so-called “greater measure”) than others but happen they do. And doing probability on infinite sets doesn’t make sense because we can never “do infinite probability experiments” to check if our theory of probability is actually correct. If the theory is supposed to be about the physical world we need to be able to test it against the physical world. So probability theory runs into this deep problem of being untestable. And if it's not supposed to be science but rather a branch of pure mathematics, then it doesn't apply to the real physical world. Whatever the case: Hoffman says that quantum theory simply does give us objective randomness which is a terrible mistake and a gross misunderstanding of the theory. But these misconceptions are all intimately tied together and I think a conversation between him and David Deutsch on this point would be fascinating. But for now, if you do read this Donald – please watch that video and read David’s book “The Beginning of Infinity”. I say this not because I'm wanting to refute your entire thesis but rather - like Annaka - I think there could be something there which might be improved by fixing up these quibbles I have. Moving on…
Free Will: Hoffman broadly agrees with Sam about free will but at the same time Hoffman uses the word “agent” with rather wild abandon in places JHe is tripped up on this by Annaka and it’s not entirely clear what he means by “agent”. He does say he has a mathematical description of what “choice” is and how choice must arise from unconscious factors and hence these choices cannot possibly be real (free) choices. He goes so far as to say (on his theory): you cannot be conscious of making a choice because the mathematical formalism says as such. But I think most of us refute this by personal experience. I am conscious of making a choice and no matter what your mathematics seems to say: I'm choosing to finish this sentence.
This brings me to my most major gripe from the podcast and Hoffman’s approach:
Hoffman very heavily emphasizes the mathematics of his theory whilst deemphasizing experimental testing. He insists upon the formalism. It’s an interesting “tell” by which I mean: many of us have bemoaned so-called “scientism” – the application of scientific methods to problems that are not science (like purely philosophical problems like the very question: are there philosophical problems?) Scientism is something thought of as the stance that no knowledge is “valid” or some such except for scientific knowledge. Now I know this isn’t a word - and I tend to hate neologisms (a point I've made over and again) but I’ve often accused some of “mathematism” – the claim that mathematical theories have some special status or that without some formalism, the theory has less status. I think this is a mistake. Now I’m not (yet!) accusing Hoffman of making this mistake necessarily – he might think the mathematics just arises naturally. But the issue I have is that a scientific theory is typically able to be explained in natural language at least somewhat well to some crude resolution perhaps. Even string theory is about...well, strings - which can be rather simply described using words. I concede the pioneers of quantum theory often failed at this. But even they were able to say something in English. Hoffman doesn’t say much about what consciousness is exactly in natural language (i.e: simple English) – he just says it’s “mathematically precise”. Now this may be all well and good to some extent. For example: we should take something like The Schrodinger wave equation seriously. And that equation – the equation that describes (for example) the motion of an electron over time as having a wave-type shape – tells us that the electron really does occupy many possible places simultaneously and can take many possible paths simultaneously. It is taking this equation – the mathematics - seriously that leads us inexorably to the multiverse (among other things). I hasten to add here: the electron itself is not a wave: it is a thing that can occupy a “sharp” position in space – but the wave aspect means that it has many (initially fungible) instances that can take all the different possible paths. Now the difference here is that: there exist experiments in quantum theory. Many many experiments and observations. We don’t start with the formalism: we start with certain problems prompted by strange observations. In the case of quantum theory some observations are: patterns generated by certain interference experiments. So then we seek an explanation - a solution to our problem. And THEN we seek to formalize all that. This is science – this is reason. It seems to me that Hoffman has no problematic observations he’s really trying to explain. I mean: he has consciousness but it’s not an observation of the form: if we see that, then this refutes my theory or is not refuted by my theory. Instead he seems to be defining into existence something he calls consciousness and then doing some mathematics. But we don’t merely define electrons (for example) into existence – we see their actual marks on photographic plates and so on. Experiments matter, so we absolutely need some experiments to check Hoffman’s thesis. But what would these look like? I cannot even imagine. And crucially we need so-called “crucial experiments” (see here for more about that). An experiment that decides between at least two theories about what’s going on. If the result goes one way, one is refuted and the other theory is not. But without a good working existing definition of consciousness how can we even begin to know how to test Hoffman’s theory. Test? Test against what? People fundamentally misunderstand the nature of science on this point which is why David’ Deutsch’s original paper on all this is so important. (And again, if you want the crib notes on that, see here). So Hoffman uses the phrase “mathematical precise” theory/definition over and again as though if he could just get this robust, self-consistent mathematical theory he will have explained consciousness and other things. But we must recall: Newtonian Gravity is “a mathematically precise theory of gravity”. It is also demonstrably false (which by the way is a strength of the theory: it makes predictions we can observe in the real world and when we fail to observe what is predicted while general relativity is shown to be accurate, we have refuted Newton). So “mathematically precise” does not mean true. Indeed mathematicians make up mathematics all the time that has nothing to do with reality. Famously, Hardy in his "A mathematicians apology" celebrated doing this. Like so many pure mathematicians, he regarded the subject as more an art than anything else.
Now I hasten to add here that Hoffman is no crank (so far as I can tell!) because he readily admits that because he makes the strong claim that consciousness is more fundamental than quantum theory, general relativity and so on, that if his theory cannot produce quantum theory, general relativity and so on as “limiting cases” then his theory is absolutely false. So that’s something at least.
Hoffman invokes Godel’s incompleteness theorems (always a bit of a red flag to me) but not in a too egregious way. He concedes with prompting from The Harrises that taking LSD or mushrooms might help him because Hoffman says that he has never so much as smoked a cigarette. He has said he has spoken to friends who are “experienced psychonauts” and what they report back are certain visualisations and the feeling of being “outside spacetime” that are in very good keeping with his own theory. But of course: that would be some kind of empiricism: knowledge coming from the external world…wherever that world might be.
The final question from Annaka is perhaps the best one as it reveals something about the thought processes of everyone in the conversation, so I’m just going to transcribe the question and the answer in full and make some remarks afterwards. Italics and boldface are my emphasis.
Annaka: “Doesn’t it seem that where we’re left with all of this is that even if you’re able to make great progress with the math and with the theory and that you’re able to come to some point of feeling that you’re confident that this is correct or likely to be correct, what would constitute proof? It seems to me to be outside the realm of scientific experiment. When we’re talking about consciousness you could tell me there’s some theory that says that there is consciousness in the atoms of the sofa I am sitting on but how would we be able to confirm it just seems to me it’s impossible to confirm, so I’m just wondering if there’s some sense in which this can result in some type of scientific experiment or what would result in proof?”
Donald: Right. That’s a critical question for a scientist and the bottom line for me is that if my theory of consciousness when I project it back – so I have to have theory of consciousness and it’s dynamics that is absolutely precise. I have to have a mathematically precise mathematical function that projects it into our spacetime interface because that’s where we’re going to get any data for our experiments. And then the first thing – the first test of my theory will be – if the dynamics when it’s projected into space time does not give me back all of the consequences of evolution by natural selection, general relativity, quantum field theory…if I cannot do that, I’m wrong. I’m flat out – I’m wrong. So already all the sciences we have right now are going to be an acid test for this deeper theory.
Annaka: Except that your starting with this base assumption that you’re calling this foundational math – you’re saying that it represents consciousness but I’m just wondering if there’s any way that that can be confirmed or will that always be an assumption?
Donald: So…what I will be able to do is to see if I’m wrong. Right? So if there are things that if I cannot do them, the theory is absolutely wrong. What I cannot do is prove that I’m right. But that’s not unique to my theory. No scientist can ever prove that they’re right. The best we can do is get evidence that either disconfirms our theory or reduces our credence in the theory or that in some sense increases our credence in the theory. But it’s sort of an elementary point in the philosophy of science that no scientist can ever prove their theory. And even falsification in the strict Popper sense – you may not strictly be able to falsify a theory – you may, because there are so many auxiliary hypotheses that go into a theory that you don’t know which one is the faulty one. But, my theory can be tested as much as any scientific theory can be* and one test that I would take as absolutely critical is if I can’t get back evolution by natural selection and say quantum field theory when I project into our spacetime interface I would take that as clear evidence I’ve got the wrong theory of consciousness.
My comments: I’m going to leave aside all of the anti-Popperian stuff there more or less: lots of talk about “prove” and “confirm” (anti-fallibilist notions of science or confusion between purely deductive systems like mathematics and explanatory sciences like physics) and “confidence” (reducing epistemology to feelings) and even “credence” – the Bayesian mistake (I explain some about that here). As to “cannot strictly falsify” – this is The Duhem Quine Thesis and is all very well. And we cannot prove it true. Indeed.
But this claim that the test of “getting back” quantum field theory is an experimental test is a little dubious. I mean, he’s strictly correct that if he makes the claim everything else is emergent and his theory is fundamental, that if he cannot produce quantum field theory then he’s wrong. But quantum field theory rests not just upon the fact that classical mechanics is a limiting case of sorts but rather many many real world observations only explicable by quantum theory. So can his *theory can be tested as much as any scientific theory can be*? I don't think so. Consider: in the case of quantum theory alone. It is testable against:the photoelectric effect, Compton scattering, radioactive decay, interference experiments, quantum computation, etc, etc. I mean a deep theory seeks to solve many discrete problems and is tested against novel observations and experiments. At least typically. Even evolution by natural selection: solves the problems about homologous structures, the similarity of DNA, embryos, the fossil record, Galapagos finches, etc, etc, etc. And General Relativity: neutron stars and black holes, the precession of Mercury’s orbit, Eddington’s observations of starlight during an eclipse, the GPS system, etc, etc. But Hoffman’s theory – so far as I can tell – does nothing like this. So it's not as testable as any other scientific theory. Not by a long shot. What are its PREDICTIONS? It defines into existence something he labels consciousness – but he has not told us in the entire podcast what consciousness is in simple terms or what we might observe out there to notice if its on the right track or problematic because it's (say) postulating disembodied consciousness which direct the motion of clouds (or some such). So without a simple definition, explaining WHY that definition is correct except to say: here is some mathematical formalism from which we can generate quantum field theory, general relativity and natural selection and whatever else seems more like a mathematical trick than actual science. What predictions does it make BEYOND just: here is a theory that has as limiting cases all other theories? That’s certainly interesting…but is it consciousness?
General relativity explains what gravity is: the bending of spacetime by mass/energy. And then it predicts things like black holes. Which are observed! And which predicts the bending of light during an eclipse – which is observed. And sure it “gives us back” Newtonian gravity as a limiting case…but that is almost beside the point.
So for Hoffman’s theory to fit this scheme I expect not only for his theory of consciousness to “give us back quantum field theory (and the rest) – it has to tell us what consciousness is exactly (as General relativity tells us what gravity is) AND it has to predict things we hitherto do not know about like black holes and allows us to test this theory by making precise predictions about what we should observe. And if the response is: but that would be about “the interface” and not “objective reality” we have a theory immune from experimental criticism. And that ceases to be science.
But this does not disqualify it because it could be very interesting metaphysics and the beginnings of an actual research project.
I have to read more of his material.
The most valuable thing you can offer to an idea