For many people, the phenomenon of consciousness is the best evidence we have that there must be something important missing in our basic physical description of the world. According to this worry, a bunch of atoms and particles, mindlessly obeying the laws of physics, can’t actually experience the way a conscious creature does. There’s no such thing as “what it is to be like” a collection of purely physical atoms; it would lack qualia, the irreducibly subjective components of our experience of the world. One argument for this conclusion is that we can conceive of collections of atoms that behave physically in exactly the same way as ordinary humans, but don’t have those inner experiences — philosophical zombies. (If you think about it carefully, I would claim, you would realize that zombies are harder to conceive of than you might originally have guessed — but that’s an argument for another time.)
The folks who find this line of reasoning compelling are not necessarily traditional Cartesian dualists who think that there is an immaterial soul distinct from the body. On the contrary, they often appreciate the arguments against “substance dualism,” and have a high degree of respect for the laws of physics (which don’t seem to need or provide evidence for any non-physical influences on our atoms). But still, they insist, there’s no way to just throw a bunch of mindless physical matter together and expect it to experience true consciousness.
People who want to dance this tricky two-step — respect for the laws of physics, but an insistence that consciousness can’t reduce to the physical — are forced to face up to a certain problem, which we might call the causal box argument. It goes like this. (Feel free to replace “physical particles” with “quantum fields” if you want to be fastidious.)
- Consciousness cannot be accounted for by physical particles obeying mindless equations.
- Human beings seem to be made up — even if not exclusively — of physical particles.
- To the best of our knowledge, those particles obey mindless equations, without exception.
- Therefore, consciousness does not exist.
Nobody actually believes this argument, let us hasten to add — they typically just deny one of the premises.
But there is a tiny sliver of wiggle room that might allow us to salvage something special about consciousness without giving up on the laws of physics — the concept of downward causation. Here we’re invoking the idea that there are different levels at which we can describe reality, as I discussed in The Big Picture at great length. We say that “higher” (more coarse-grained) levels are emergent, but that word means different things to different people. So-called “weak” emergence just says the obvious thing, that higher-level notions like the fluidity or solidity of a material substance emerge out of the properties of its microscopic constituents. In principle, if not in practice, the microscopic description is absolutely complete and comprehensive. A “strong” form of emergence would suggest that something truly new comes into being at the higher levels, something that just isn’t there in the microscopic description.
Downward causation is one manifestation of this strong-emergentist attitude. It’s the idea that what happens at lower levels can be directly influenced (causally acted upon) by what is happening at the higher levels. The idea, in other words, that you can’t really understand the microscopic behavior without knowing something about the macroscopic.
There is no reason to think that anything like downward causation really happens in the world, at least not down to the level of particles and forces. While I was writing The Big Picture, I grumbled on Twitter about how people kept talking about it but how I didn’t want to discuss it in the book; naturally, I was hectored into writing something about it.
But you can see why the concept of downward causation might be attractive to someone who doesn’t think that consciousness can be accounted for by the fields and equations of the Core Theory. Sure, the idea would be, maybe electrons and nuclei act according to the laws of physics, but those laws need to include feedback from higher levels onto that microscopic behavior — including whether or not those particles are part of a conscious creature. In that way, consciousness can play a decisive, causal role in the universe, without actually violating any physical laws.
One person who thinks that way is John Searle, the extremely distinguished philosopher from Berkeley (and originator of the Chinese Room argument). I recently received an email from Henrik Røed Sherling, who took a class with Searle and came across this very issue. He sent me this email, which he was kind enough to allow me to reproduce here:
Hi Professor Carroll,
I read your book and was at the same time awestruck and angered, because I thought your entire section on the mind was both well-written and awfully wrong — until I started thinking about it, that is. Now I genuinely don’t know what to think anymore, but I’m trying to work through it by writing a paper on the topic.
I took Philosophy of Mind with John Searle last semester at UC Berkeley. He convinced me of a lot of ideas of which your book has now disabused me. But despite your occasionally effective jabs at Searle, you never explicitly refute his own theory of the mind, Biological Naturalism. I want to do that, using an argument from your book, but I first need to make sure that I properly understand it.
Searle says this of consciousness: it is caused by neuronal processes and realized in neuronal systems, but is not ontologically reducible to these; consciousness is not just a word we have for something else that is more fundamental. He uses the following analogy to visualize his description: consciousness is to the mind like fluidity is to water. It’s a higher-level feature caused by lower-level features and realized in a system of said lower-level features. Of course, for his version of consciousness to escape the charge of epiphenomenalism, he needs the higher-level feature in this analogy to act causally on the lower-level features — he needs downward causation. In typical fashion he says that “no one in their right mind” can say that solidity does not act causally when a hammer strikes a nail, but it appears to me that this is what you are saying.
So to my questions. Is it right to say that your argument against the existence of downward causation boils down to the incompatible vocabularies of lower-level and higher-level theories? I.e. that there is no such thing as a gluon in Fluid Dynamics, nor anything such as a fluid in the Standard Model, so a cause in one theory cannot have an effect in the other simply because causes and effects are different things in the different theories; gluons don’t affect fluidity, temperaturs and pressures do; fluids don’t affect gluons, quarks and fields do. If I have understood you right, then there couldn’t be any upward causation either. In which case Searle’s theory is not only epiphenomenal, it’s plain inaccurate from the get-go; he wants consciousness to both be a higher-level feature of neuronal processes and to be caused by them. Did I get this right?
Best regards,
Henrik Røed Sherling
Here was my reply:
Dear Henrik–
Thanks for writing. Genuinely not knowing what to think is always an acceptable stance!
I think your summary of my views are pretty accurate. As I say on p. 375, poetic naturalists tend not to be impressed by downward causation, but not by upward causation either! At least, not if your theory of each individual level is complete and consistent.
Part of the issue is, as often happens, an inconsistent use of a natural-language word, in this case “cause.” The kinds of dynamical, explain-this-occurrence causes that we’re talking about here are a different beast than inter-level implications (that one might be tempted to sloppily refer to as “causes”). Features of a lower level, like conservation of energy, can certainly imply or entail features of higher-level descriptions; and indeed the converse is also possible. But saying that such implications are “causes” is to mean something completely different than when we say “swinging my elbow caused the glass of wine to fall to the floor.”
So, I like to think I’m in my right mind, and I’m happy to admit that solidity acts causally when a hammer strikes a nail. But I don’t describe that nail as a collection of particles obeying the Core Theory *and* additionally as a solid object that a hammer can hit; we should use one language or the other. At the level of elementary particles, there’s no such concept as “solidity,” and it doesn’t act causally.
To be perfectly careful — all this is how we currently see things according to modern physics. An electron responds to the other fields precisely at its location, in quantitatively well-understood ways that make no reference to whether it’s in a nail, in a brain, or in interstellar space. We can of course imagine that this understanding is wrong, and that future investigations will reveal the electron really does care about those things. That would be the greatest discovery in physics since quantum mechanics itself, perhaps of all time; but I’m not holding my breath.
I really do think that enormous confusion is caused in many areas — not just consciousness, but free will and even more purely physical phenomena — by the simple mistake of starting sentences in one language or layer of description (“I thought about summoning up the will power to resist that extra slice of pizza…”) but then ending them in a completely different vocabulary (“… but my atoms obeyed the laws of the Standard Model, so what could I do?”) The dynamical rules of the Core Theory aren’t just vague suggestions; they are absolutely precise statements about how the quantum fields making up you and me behave under any circumstances (within the “everyday life” domain of validity). And those rules say that the behavior of, say, an electron is determined by the local values of other quantum fields at the position of the electron — and by nothing else. (That’s “locality” or “microcausality” in quantum field theory.) In particular, as long as the quantum fields at the precise position of the electron are the same, the larger context in which it is embedded is utterly irrelevant.
It’s possible that the real world is different, and there is such inter-level feedback. That’s an experimentally testable question! As I mentioned to Henrik, it would be the greatest scientific discovery of our lifetimes. And there’s basically no evidence that it’s true. But it’s possible.
So I don’t think downward causation is of any help to attempts to free the phenomenon of consciousness from arising in a completely conventional way from the collective behavior of microscopic physical constituents of matter. We’re allowed to talk about consciousness as a real, causally efficacious phenomenon — as long as we stick to the appropriate human-scale level of description. But electrons get along just fine without it.
The proof that early mathematical philosophers would probably attempt for the statement 1/2 +1/4 +1/8 + … = 1 might be something like this:
I am in seat #1 in an infinite auditorium which is inside an infinitely-large cave, waiting to hear a philosophy lecture. The lecturer is late, so someone is using the overhead projector to make hand-puppet shadows on the wall. I take out my last stick of gum. My neighbor in seat 2 says, “Pardon me. Could I please have some of your gum?”, so I cut the stick exactly in two halves with a razor, keep one half, and give him the other. His neighbor in seat 3 then asks him to share the gum, and so on. This will produce the exact same series if it goes on forever, and at every step along the way the sum of the pieces is one stick of gum.
In practice however, we would run out of gum. Gum is not infinitely continuous. However, even if it were, it would take us forever to complete the proof.
In effect, what mathematicians now do to sum an infinite series, is guess the result (or look it up in a table, or use the geometric series equation) and use reasoning similar to the above to convince themselves they have the right answer. Which is okay, but they have avoided Zeno’s real issue. We can’t physically add up infinite pieces of infinitely-continuous gum. How does nature do it? Maybe the gum is not really infinitely continuous.
At the microphysical level, ‘We can affect the past’ is just as correct as the statement ‘We can affect the future’. (Which opens up a new kind of free will compatibilism – but I digress.) At the macroscopic level, our familiar level, it’s generally not, due to all the entropy gains involved. And it’s fine if causality isn’t specified in the quantum equations – it’s an emergent property, ultimately only of interest to agents like us. Everything we know and love about consciousness remains true in the “block universe” picture. Some overgeneralizations about time and causality fall away, but we didn’t need them anyway.
JimV
Matter is not continuous, space and time are likely not continuous, computation is certainly dicontinuous, so the infinitesimal is banished to a Hilbert Space which is a mathematical construct.
PT
I think quantum irreversibility was always the correct position – certainly Schrodinger and many since have viewed QM as thermodynamic. Now we seem to have experimental proof, e.g. Batalhao et al, 2015.
The block universe of itself does not account for timelike (causal) trajectories. Time we have, the arrow of time we experience, and it should be remembered that the thermodynamic argument was explored to explain the arrow of time, not to prove that it exists. A correlation cuts both ways – why should not entropy be a consequence of time?
Suppose there is an observer who can perceive of time as being what it really is, namely a fourth component of “space”, in other words someone who is not shackled to the perception of time as being something that flows from one instant to another. To him the entire spacetime history of the universe is laid out in the same sense that space is laid out for us. So from this vantage point, everything from the beginning of time to its end is eternal, it has been existing forever.
Whither then, free will?
It’s called Laplace’s demon, and yes, no free will from that theoretical point of view.
But the point is exactly that. This point of view is exclusively theoretical and cannot be achieved in the reality as we know it. It has a mandatory requirement: observing from the outside (which is not possible, so it can as well be considered just a fantasy).
If Laplace’s demon is inside the system he himself he’s observing, then his model needs to recursively include him in the description. And the description can never be complete, same as Godel demonstrated in mathematical terms. A system is only complete when you observe it from the outside. Recursive, self-referent systems are always incomplete.
“A correlation cuts both ways – why should not entropy be a consequence of time?”
I agree with this question, if that even makes sense.
The arrow of time and entropy are two ways to view the same thing. But just because ‘our only past’ has low entropy does not mean that that we understand time in any other direction. The same thing as ‘time’ should exist in the other directions, but we can’t observe it because there is no arrow in those directions. So… it just looks like space.
Abalieno said
‘And the description can never be complete, same as Godel demonstrated in mathematical terms. A system is only complete when you observe it from the outside. Recursive, self-referent systems are always incomplete.’
Yes. Haven’t followed this too carefully but we are never ‘outside’ of course.
Logico said,
“A correlation cuts both ways – why should not entropy be a consequence of time?”
Anyone got a decent answer?
Correlation rather than causation could perhaps be a better definition of what we find in science?
Coming up to 400 comments – why haven’t we cracked this ‘consciousness’ thing? 😉
Look, if someone could just come up with a simple computer program that seems like its doing something we could reasonably call ‘consciousness’, that would go a long way to settling the issue.
The vast majority of the human brain’s neurons are surely not needed for consciousness, and we also know that probably very simple organisms have *some* degree of consciousness (you don’t need to be a panpsychist to believe that).
So if we look for the most basic possible ‘unit of consciousness’, assuming that consciousness is computational, then there should be a simple computer program we could write that represents this ‘minimum unit’ of conscious experience. So what is that program?
zarz,
“zarz,
“in truth, only atoms and the void.”
Maybe there is a difference between consciousness and what it is conscious of. What we are conscious of are distinctions, differences, relations, etc. So naturally reality appears quantified.
In the above example, is the void, i.e. space, truly quantified, or is it only the atoms? What is the network tying all these nodes together? If even space is quantified, then what actually ties all these quanta together? Some say it’s math tying everything together, but that is a bit anthropocentric.
Does function follow form, or does form follow function?
Maybe consciousness is only apparent as process, not entities it seemingly collects. There is no stable state of consciousness. It only exists in this complex state between the coalescing order of the past and expanding chaotic probabilities of the future.
When something is fully defined, then it is past.
Logico,
I’m not sure what moral to draw from Batalhao et al, 2015. The laws of microphysics aren’t supposed to be T-reversible, only CPT-reversible.
Also, what Moe said. The arrow of time and entropy/2nd Law are the same thing, in my view.
PT
“At the microphysical level, ‘We can affect the past’ is just as correct as the statement ‘We can affect the future’. ”
“The laws of microphysics aren’t supposed to be time reversible.”
That could confuse a stupid person.
It seems to help to appreciate the vast complexity of the systems we are talking about. A hammer and a nail are two polycrystalline metals whose description defies description based on the standard model. No one can predict the plastic deformation of a nail that results from a hammer hitting it on the basis of the core theory. In ideal conditions we have some success in understanding crystalline defect dynamics that are the core feature of ductility in metals, but you would have to know the precise boundaries of an enormous number of domains along with impurity locations and more to predict how the metal will deform. But even those questions are simple compared with predicting the velocity of the hammer or the timing of the blow, both of which depend on the physiology and intentions of the person wielding the hammer, and ultimately on the structure she/he is trying to create.
It seems to me that theoretical physicists and philosophers are often committed to simplifying the world beyond a level that is adequate for these questions. Even the idea of different levels of description for which upward or downward causation can be defined is a simplification made for convenience. In fact, electrons, protons, etc have been interacting over billions of years to evolve planets and humans that can consciously choose to build houses and swing hammers that deform nails while driving them in. It is a tightly coupled system. You can’t simplify it to a few degrees of freedom associated with ‘consciousness’ or the small number of fermions and bosons in simple statistical ensembles that we physicists are able to cope with.
When you consider the entire coupled system, there are emergent features that we conceptualize with shorthands like ‘living’ and ‘conscious’. Conceptualized this way, there seems to be something like downward causation. Meaning humans choose to build things like LHC and it causes events to occur that would never have been expected on a planet like earth unless you knew that there were conscious agents on the planet. Whether all the events obey some underlying physical theory is essentially irrelevant because it is completely impossible to predict very many useful things about the design of LHC or most of what they measure on the basis of only the underlying theory. Only a few statistical features of a few kinds of events are predicted on that basis. To predict the rest, you have to know the noise sources, economic constraints, politics, and human errors etc. that went into the whole system. So humans cause the bits stored in the data set as much as Higgs bosons do. You could never predict that data set without both kinds of causes. But they are both short hand ways to capture parts of the coupled system that is far too complicated for us to describe. Once the focus turns to the practical question of predicting specific observed phenomena, it becomes clear that it is the limitations of our conceptual and predictive tools rather than any fundamental feature of reality that need to be understood to make progress in this area.
Logico,
Huh? People interact with various charges and polarities, so if you interact in a certain way with a left handed electron, and in a corresponding way with a right handed positron, whatever effect you have on the future of one, you’ll have on the past of the other. That is, to whatever extent “effect” is a meaningful word at this level. Maybe causal talk should be reserved for irreversible processes, so that we could make “A caused B” asymmetric by definition. As long as we don’t then mistake that definition for a discovery about causality.
PT
My point was that you made two contradictory assertions. Can’t do anything with that.
Your new point is that backward causation is possible BECAUSE an electron travelling from A to B (interaction with me) to C (some causal effect of B) can be whimsically interpreted as a positron travelling from C to B to A in negative time, on the reverse-everything CPT symmetry principle. (Feynman diagrams are calculating aids, not veridical models of reality.) Unfortunately you forgot to change causation to ‘uncausation’ when you reversed things. We can’t affect the past. That would become very obvious if you looked at something like beta decay.
PT
Or try this. My bank account is in credit at A when my regular income was last deposited. At B an unexpected emergency got me overdrawn. At C the next pay cheque put me in credit again. I could construct a Feynman type cash flow diagram which simply showed periods of debt as time-reversed periods of credit, and for all I know it might even be easier to follow than double entry bookkeeping. But you can’t spend negative credit, and my interaction at C definitely did not cause the debt.
ganv
I agree broadly with your perspective on complexity and the viability of ‘final analysis’.
It takes decades for engineering companies to get to a working understand of areas like metal pressing, corrosion resistance, induction hardening, etc etc, and the vast majority of it is either simplistically modeled, or just held as communal experience by a body of engineers. While classical engineering allows fairly representative finite element analysis, for example, many areas of material behaviour are still subject to major unpredictability.
The questions here on this blog are more philosophical than practical, but practical experience can help a person to see just how complex our world is and just how ‘short-range’ and simplistic our understanding and modelling really are. In addition, our ability to handle transitions out of micro-physics is subject to conceptual uncertainties as well as profound state definition and computational limitations.
Still you seem to have a basic faith in some form of deterministic, naturalistic emergence.
‘In fact, electrons, protons, etc have been interacting over billions of years to evolve planets and humans that can consciously choose to build houses and swing hammers that deform nails while driving them in.’
There is an anthropomorphic causality in there, or at least in your use of words. Easy enough to justify if there is Someone outside our big system.
ganv
Certainly if you start with the presumption that consciousness is emergent/secondary, that’s the only answer you get out: then playing the emergence card a second time, you can conclude that agency (free will) must also be emergent – to quote Leonard Cohen, ‘a card that is so high and wild you’ll never need to deal another.’ But the point of one of the two main threads in this discussion is that if consciousness is indeed too deep to understand, maybe that depth is ontological, not epistemological. This is not a spiritual position. It is Wheeler’s It-from-Bit, matter derivative from information processing, a computational universe of some sort.
Personally I don’t believe in qualitatively novel emergence. When Benard cells ’emerge’, microscale spatial relations generate macroscale spatial relations – we expect/require spatial relations; when CERN/LHC is built across a national boundary, the political geography of countries led to the geography of LHC – but some geography/site was expected/required; when the economics of the funding bodies translated into design constraints, we expected/required that building LHC would involve a cost, and there is no need to describe the particular/contingent details of resourcing all the way back to solar radiation, Gibbs Free Energy and the Big Bang in establishing the concept. Proof of concept is a different animal from specific prediction.
On a lighter — but relevant — note, I think you might enjoy my fantasy serial novel, which has fun with the notion of consciousness. If you read it, I’d like to know what you think. https://glassgrimoire.wordpress.com/fantasy-serial-novel/prologue/
Logico,
There’s a debatable interpretation yes, contradiction no.
I’ve got nothing against defining “causality” in a time-arrowed way, using irreversibility, in light of the local entropic gradient . In which case we can distinguish “uncausing” in the backward time direction from “causing” in the forward. But then we need to admit that this conceptualization depends on large ensembles of microphysical states in order to get the probabilities to good-enough approximation to one, to ensure the irreversibility. In small ensembles considered on their own, causality will be undefined. In which case at the microphysical level “we can affect the future” will be just as false as “we can affect the past”. Either that, or: come up with observations that indicate a more fundamental arrow of time than the entropic one. Show me.
I’m more convinced than ever that consciousness is a fundamental property of reality, and ‘qualia’ is a unit of time. I’ve been doing my research on neuroscience, and it’s the ‘time sychronization’ and ‘signal modulation’ (time sequences) going on in the brain that are really leaping out at me.
You know one of the earlier theories about consciousness was that it was associated with time sychronization where large coalitions of neurons were fired in sychrony, and this was mediated with 40Hz gamma waves. This theory sort of went out of fashion, and newer theories became more popular – the ‘self modeling’ and ‘information integration’ theories are the most popular today.
But I’ve really gone back to the earlier theory of time sychronization and I’m starting to suspect that the newest theories are red herrings.
It would be fun to be able to see the universe from outside of the constraints of time and yet be able to understand what it is like to experience our constraints of time.
PT
How could the two assertions, which I say are conradictory, both be true?
“Thermodynamics” may apply to very small groups of quantum particles, as I already said. Neutron decay into proton plus electron plus antineutrino is readily observed. The reverse process has never been observed, since the chance of bringing three particles together simultaneously is remote.
I am well aware that by replacing a particle on one side of the equation by its antiparticle on the other, a valid and well observed reaction (eg K capture) can be symbolized and then translated into a Feynman diagram; but that is not an actual time reversal. If it had been, it would allow time reversed causation. As Feynman wrote in 1976: “The science fiction writers who have interpreted my view of the positron as an electron going backward in time have not realized that that theory is completely consistent with causality principles, and in no way implies that we can travel backward in time.”
But that is in any case an entirely material, particles-and-fields concept of reality. The It-from-Bit view is a concept of computation as fundamental: computations are not generally reversible. I pointed out earlier that 2+2=x is entirely determined; 4=x+y is not.
Ben,
“Consider Sean’s favorite example of a cup of coffee with cream. The low-entropy state is very orderly and simple: coffee on the bottom, cream on top. But the high-entropy state is even simpler: creamy coffee everywhere.”
This is a very incorrect way to think about entropy. A cup with coffee and cream totally seperated is not simple at all. It is the most complex cup possible.
The highest entropy describes the simplest state. At maximum entropy, you could not say where you were in the universe. It would all be identical. You could be anywhere.
The lowest entropy describes the most complex state. At the lowest entropy, you could pinpoint your location with very high accuracy.
If instead of mixing just two liquids, you carefully layered 20 liquids, this is an even more ordered state. You could be sure of the location of any molecule up to the volume taken up by that type of molecule within the whole.
This is the key point. Entropy is meaningless without a comparison to the surroundings. You have to have a ‘space’ for different possibilities. Isolated points in space have no entropy at all, and that’s why they are not bound by time.
Time is the result of there being options.
Moe
Read up on Third Law of Thermodynamics. (You will be embarrassed.)
“Time is the result of there being options.”
Time is emergent, like temperature, color, etc. Contra Smolin.
It is not that the laws evolved over time, but that they are descriptions of processes that cycle through stages of complexity/simplicity/expansion/consolidation/order/chaos/stasis/dynamism, etc. and there is no objective perspective on this, because taken as a whole(absolute state), it cancels/whites out.
People distill out ideals, from gods, to dimensionless points, to laws and then assume them to be absolutes. An ideal is epitome, while an absolute is essence. An ideal is definition, not explanation.