On Consciousness and Free Will
On this blog, we’ve recently tackled religion and the nature of existence, but we’ve left out one huge chewy topic that people tend to lump into this philosophical category, and that’s consciousness. You need only look at a site like Closer to Truth in order to see just how tightly coupled these ideas are in the public imagination. It’s also a topic of significance to me, as it was through writing on this subject that I first found myself exploring digital physics many years ago.
One of the great defenders of the specialness of human consciousness in the physical realm has been Roger Penrose, the man who proposed that consciousness was non-computable because it was founded on non-computable processes in nature. A lot of this blog has been about demolishing that idea. So on Penrose’s hypothesis, digital physics is pretty clear.
However, there are plenty of other ways you might integrate consciousness with discrete reality. For instance, take the essays submitted for the 2011 FQXI prize, on the subject ‘Is Reality Digital or Analog‘ (probably the highest profile public discussion forum on this subject in the last five years), and you’ll see the word consciousness showing up within the first four titles. What about these other models? Do they have anything to add? Here’s my answer:
Digital physics has nothing to do with consciousness, because consciousness has nothing to do with physics.
The notion that consciousness has any bearing on quantum mechanics, and therefore physics at large, is, to my mind, a lamentable side-effect of the times in which QM was first formulated. Poor old Neils Bohr had the unfortunate fate of hanging out with a bunch of logical positivists, who were sort of trendy at the time, and he tracked some of that muck back into physics along with him.
Enthusiasts on the topic of quantum consciousness point to the fact that observation of a QM event affects how it will play out. However, as we’ve seen in previous posts, we can generate identical effects in a simulation by simply asking the question–is information leaving the system or not? If it is, then an observation has taken place, if it isn’t, then an observation hasn’t happened yet.
In other words, particles are non-committal. They’ll hedge their bets and be everywhere until you force them to make up their mind. And forcing them to decide often has the side effect of forcing a bunch of their friends to decide too. In this regard, particles are rather like teenagers trying to decide where to go on a Friday night. They’re no more strange and magical than sixteen year-olds. (Though, admittedly, sixteen-year-olds are pretty strange.) Ask any self-respecting working particle physicist about the role of consciousness in QM, and they will struggle not to roll their eyes at you. This is why.
So if we can rule out consciousness having an impact on quantum mechanical events, and we can rule out its dependence on smooth symmetries of nature, is there anywhere left for the specialness of consciousness to hide? At this point, we invoke the principle of minimal complexity which we used to unpack the idea of god, and we ask ourselves if the universe is more or less complex if we have to carve out some special extra room for sentience in physical law. The answer, I’d argue, is that it’s more complex, and therefore massively unlikely. Nice though it might be to cogitate on, then, consciousness arises naturally out mechanistic physical processes, just like everything else.
But what about free will? Given that quantum mechanical events are completely unpredictable, isn’t there at least enough room left for that? Not in this model of the reality, there ain’t.
To describe the universe completely, we need to treat the rules that run nature and the data that they run on as a single closed system. Otherwise we haven’t finished describing them yet. Thus, if we find that a huge pile of random numbers are necessary for the universe to work, then they belong as part of our model–as a giant list of lottery tickets printed at the beginning of time and slowly spent.
Of course, a huge pile of random numbers that lasts for the length of the universe is a really, really awful implementation. The principle of minimal complexity strongly suggests that reality is better than that.
Free will, does it exist, then (at least in the sense of something special outside of logic)?
Sorry, no luck, as it were, so to speak, if you’ll pardon the pun, etc.
How to Build A Universe 
Your argument seems tautological to me. If we assume that the universe is a formal system operating mechanically on a set of (presumably finite) rules, then it follows that there will be no free will. Therefore it follows from the assumption that free will, and hence consciousness, is an illusion generated by the system.
But I would like to turn this around. I declare that I am conscious, and that I have free will and freedom of choice. If I am right in my observation on myself, then (proof by contradiction), the universe cannot be a formal system operating mechanically on a set of rules.
So to maintain your thesis, how are you going to prove that I’m not conscious and endowed with free will?
Great question. And you’re right. This post was, perhaps, a little redundant. Please note, though, that I’m not asserting anywhere that people aren’t *conscious*. I’m conscious, and I imagine that others are. My assertion is that we should not expect there to be special laws of the universe for the manifestation of consciousness.
Having said this, the answer to your question here depends on what you consider to be a proof. What we almost certainly can prove is that a finite, deterministic system can maintain the interior sensation of consciousness and free will. Given such a model, we can, in principle, make it arbitrarily large, such that its computing power is greater than that of you or any other individual human. And from there we can ask, what’s different between you and it? However, this alone may not meet your criteria.
If you want examples of experiences in which you think you’re exercising free will, but evidence can be supplied that you did not, psychology and neuroscience can provide those. If you want examples of instances in which you recall having been conscious, but were not, they can also be provided. So whether any moment of self-knowledge is valid can also be thrown into question.
But if these kind of examples don’t constitute a proof, we have to ask what would, in order to make sure that you haven’t set up your challenge in a non-falsifiable way. Admittedly, one can solipsistically assert free will, along with any number of other notions in a way that is hard for anyone else to contest without access to your interior experience. However, solipsistically derived statements tend to be neither provable nor disprovable, and so don’t have a role in science.
I know you’re taking Dennett’s line here, with the comments about our self being more fragmented than we think, and our assessment of incoming information being much less than we realise. I can also accept that we are fooled into believing that we have much more free will than we actually have. But regardless of that, I still hold that I have some sort of sense free will, and although they may be fragmented and multiple, I have some sort of sense of self (or selves) and being. And it doesn’t matter whether I have only a smidgen of free will, or the consciousness that gives rise to that free will only infrequently kicks in – the fact remains that if there is even a trace of free will and consciousness, they exist.
And then you are stuck with a dichotomy if you assume that the universe runs on deterministic system – either free will isn’t free will, or deterministic systems can give rise to free will, at which point they are no longer deterministic.
So if you take the former stance, then we’ll have to agree to disagree. And if you take the latter, then we have to accept that sufficiently complicated deterministic systems can generate entities that can change the rules arbitrarily. I don’t imagine the latter will sit comfortably with your world-view though.
(And yes, I’m not looking for a mathematical proof, and even if one was available – well, read Lakatos’ Proofs and Refutations. Mathematical proof doesn’t do what most people think it does.)
Okay. I should be clear about where I stand on free will here, because we’re getting into some interesting waters.
What I intended to object to in my original post was the idea that you needed some kind of special non-determinism in nature in order to have free will. And that expecting non-determinism was unscientific.
However, it seems to me that the principle of free will is *fine* so long as it’s encompassed by the overall description of the universe. Because function that computes nature may depend on the total state of the universe, it’s legitimate to propose that in order to determine the next thing that you or I do, you would have to model the total state. From inside the universe, you can never do that, because you don’t have enough bits to represent your model *and* make a prediction. Hence, it’s consistent to say that individual’s choices can be completely unpredictable, and that the universe is finite and deterministic at the same time.
So here’s the thing. As a person living with apparently self-observed free will in a finite universe, how can you differentiate between this kind of free will and any other. I don’t think you can. And if you can’t, why assert unnecessary non-determinism?
The dictionary I just saw defines free will as `freedom of humans to make choices that are not determined by prior causes or by divine intervention’. I think this kind of free will cannot exist (from a global standpoint) if we live in a deterministic universe.
However, I doubt we will ever know for sure that the universe is deterministic, and so if somebody claims to have free will the only way to prove them wrong (that I can think of) is to predict what they will do. The thing is though, even if the universe is deterministic, the persons behavior could (and probably would) be so complex that such prediction is fundamentally impossible. If the brain’s processes correspond to an irreducibly complex computation then (from a local standpoint) we might as well have free will.
On the subject of ‘might as well have free will’, I’m in complete agreement. It makes sense to treat ourselves and others as unpredictable entities. As for knowing for sure whether the universe is deterministic or not, I agree that it’ll probably turn out to be impossible to say. But science always has to make the assumption that a system is predictable in order to advance. Thus, if someone wants to assert that there’s true randomness that can never be removed from our description of nature, the burden of proof is on them. An uncertain model of nature will always have higher complexity, because of all the spare numbers it has to store in its initial conditions.
Your argument above makes great sense to me. We need to look for consciousness elsewhere than in quantum mechanics. How does a system become aware of itself, what thresh hold of factors/complexity make that possible? And it seems that consciousness is inseparable from some form of free will. Isn’t the best measure of consciousness that a system can adapt to its environment in some way. Clearly there are degrees of consciousness in the world and it does seem to correlate with complexity.
How can deterministic rules give rise to free will? Or even the illusion of it? Great mysteries to chew on.
Personally I place my money on the synapse. And I define synapse as any juncture where data has to change state. As in a nerve synapse, as in the connection between two ants communicating, or two people exchanging notes in a blog. Consciousness seems to thrive in systems with higher densities of synapses.
Hey Thaddeus, nice to hear from you!
If you’re interested in how simple deterministic rules can give rise to extremely unpredictable behavior, you might want to look at cellular automata, and Stephen Wolfram’s ‘Rule 30’ as a specific example, if you’re not familiar with them already. It’s amazing how much apparent randomness simple rules can generate.
Re the synapse: I agree with you that the most conscious system we know is also one of the most complex (ie: the brain). However, I am personally cautious about the idea that capacity for consciousness scales linearly with complexity. Otherwise crowds of teenagers tightly coupled with cellphones would behave intelligently. (They have a lot of synapses between them!)
I suspect that consciousness is a very specific function of network feedback that requires information compression, attention management, and a bunch of other features. So while it’s easier to do with a large number of neurons, I think it’s more a function of system structure, rather than size.
I also think that it should be possible to build machines that are conscious while being a lot less intelligent than people. The question is, how do you judge such a system as conscious even if you succeed? Clearly language processing isn’t consciousness, as we have lots of machines that can process language without ‘thinking’. But what does consciousness without language processing look like?
Perhaps crowds of teenagers with cellphones, or perhaps even better, social networking sites, are superconscious, and we as individuals just can’t recognise the fact, just as presumably an anthill doesn’t know that I am conscious.
Perhaps Myspace lost out to Facebook not because of people moving from one to the other, but in some kind of Battle of the Titans instead. Which begs the question, which is going to win the next evolutionary battle – Facebook, Twitter, or Linkedin?
Fair point, though perhaps not super-testable. I’d propose that the base measure of whether a system is capable of planning, and therefore exercising awareness, is whether it can pass the Marshmallow test. Without the ability to exercise self-control, there’s no evidence that a system can defer one kind of behavior in search of another. And is consciousness for, if not to enable us to make conscious choices?
As for which of them will win? Definitely MySpace.
I’m sure you understand about chaotic processes; between any two valid solutions which predict X, there is a valid solution which predicts not X. Just like between any two irrational numbers there is a rational one. Though maybe the cardinalities are not as clear.
The basic argument equates to saying that while the brain and its backwards and forwards history may be computable, it may be literally impossible to get precise enough information to compute it (without at the same time invalidating the existing information).
That’s not how I generally think about it, but as far as I understand, I agree with you. The way I tend to frame what I think you’re saying is covered in comment #4. Basically, you can have a system be deterministic and, at the same time, unpredictable when measured from within that system. That just makes it deterministic but not reducible.
Also thinking that the primary non-physics questions is “What does it mean about our attitudes towards personal responsibility that behavior is computed (though not by us reducible).
Nice question. And I think you’re right. That’s where this topic generally leads. To my mind, the computability of humanity doesn’t mean much at all. Steven Pinker covers this ground nicely in the opening chapters to How the Mind Works.
I don’t tend to measure the brain from within the brain. I use an external device.
Logically the Universe might be deterministic but incomputable, but that doesn’t mean that given perfect knowledge of a subset it does not become computable.
The QM objection is that the fundamental processes are stochastic, however that just reduces to saying there is no free will and the computation produces a weighted range of possibilities.
A simple computable model would be to assert that the QM version is accurate except that the random values are in fact pseudo-random, making precise predictions computable.
Secret Model objection (which seems to be about what you are saying) is that the subset is driven by an as yet unknown and perhaps immeasurable mechanism, making it non-computable due to our not having access to the parameters.
Which breaks down into two versions: your version that if we could go back to the beginning we could see the secret parameters exposed; or that the secret parameters were never exposed.
There are many routes to a non-anticipatible determinism.
One final version is that the QM model is accurate but our should from the spirit world get to choose some of the “random” values…
Mathematics is full of paradoxes, while the measurable universe is demonstrably not. Xeno’s paradox is only so theoretically. If I walk halfway to the wall and halfway to the wall again and again, in the real universe i do eventually reach the wall at Plank Length. No paradox.
Randomness might as well be God. It is at odds with determinism. The “determination” of which state any given particle is in may make it seem like the science is deterministic, but without a way to calculate precisely which state occurs in which circumstances, we are left with some formula or calculation occurring outside of the control.
The simplest explanation leads us to believe that complexity grows out of simplicity over time? This argument is a form of denial. If consciousness is a side effect of deterministic properties of the universe and it is as common as it seems to be on Earth, then it is almost certainly computable. Conscious entities would inevitably compute new ones. If that is the case, science needs to start seriously considering the possibility of the universe being intelligently designed – out of sheer probability.
It is simpler to say that Consciousness precedes Chemistry. That the universe is a construct of consciousness and that from our perspective within it, the whole truth cannot be determined. Determinism and materialism have become illogical. I see people clinging to those principles simply because they were until very recently synonymous with science (in stark relief with their perceived enemy of religion). Simulated constructs full of consciousness must follow. The “universe” such entities would simulate may well need to be less complicated than their own, simply due to the energy required to compute such a simulation being greater than they could provide. If you follow your argument forward, the odds that this universe is at the “top-level” of the hierarchy is very near zero. In other words, if everything is deterministic and materialistic, then the universe is almost certainly consciously designed. Your argument leads to its own undoing.
Hey,
Thank you for your comment. I think I understand what you’re getting at, but I’d like to make sure. You seem to have three core points. First, you propose that complexity arising out of simplicity is unlikely, compared to complexity arising from some parent complexity. Secondly, you propose that if consciousness is computable, then conscious entities would create other conscious entities. You argue that given these two points, a consciously created universe is a natural conclusion. Furthermore, you point out that given that this is even possible, the likelihood of our universe being at the top of a hierarchy of created entities is vanishingly small. Is that what you meant?
Re your first point, have you ever looked at Langton’s Ant, or a 1D cellular automaton? Documented cases of complexity arising out of simplicity are hugely abundant. Given that we have thousands of examples of this happening, arising from trivially complex rule sets, it seems hard to parse this as denial.
Re your second point, I think we’re largely in agreement. We should certainly expect consciousness to be computable. And we can expect conscious entities to at least occasionally be interested in trying to make other such entities. I’m fine with this so long as we accept the caveat that it’s much harder to make consciousness than it is to exhibit it. There have been a lot of conscious people in the world, and very few of them have created general machine intelligence.
Re your third point, the implications of such a chain of complexity seem pretty extreme. For starters, do we expect a top to this hierarchy of creators? If there is, then once again we have to ask why, and where does this complexity come from. As I have outlined to in other posts, it seems logical to propose that highly complex initial conditions for the universe are vanishingly unlikely. In this scenario, the argument for intelligent design appears to collapse. Do you disagree?
If there is *not* a top to this hierarchy, then we can propose that complexity and intelligence just goes on forever up the chain. And, unless we propose that smarter intelligences are less likely to exhibit creative acts of will than simpler ones, this means that those intelligences are more and more likely to tinker with their creations. This leads us to a kind of many worlds hypothesis on acid, in which we should expect the world to be edited, rearranged, or simply destroyed at any second, given that, at any moment, some creator will intervene. The upshot of this would be that there’d be no point in science, because the universe is guaranteed to have changed beyond recognition before anyone can act, in ways we can never hope to understand.
Presuming that I’ve understood you correctly, this viewpoint is fun, but not scientific, IMO. That’s because it’s not proposing a logical structure for the universe. Instead, it’s tacitly abandoning causation.
If I’ve misread you, or you think I’ve missed something, please feel free to put me straight.
Alex