superposition ain't that super
Roger Penrose is on NPR and I just received a bill from the library for his book, so now seems like an opportune time to write about it. I finally finished up Shadows of the Mind, having read it in ridiculously short spurts over the past few months (average session length: 3 metro stops).
I mentioned Penrose before, and JK of C-130 called me out on a dig I made at Penrose. Having just finished Francis Crick's highly unsatisfying "reductionism lite", I decided to have a closer look at Penrose's account.
The man is a mathematician, so it's perhaps not surprising that his explanation starts out with Gödel's Incompleteness Theorem. The theorem says that, given a well-defined, rigorous logical system, there will always be propositions within it that cannot be proven. A commonly glossed-over consequence of this is that mathematicians' work will never be done, so presumably we should always keep a few on staff. Mathematicians understandably consider this to be a very important discovery.
Except, well, Penrose doesn't really seem to agree. He thinks that human ingenuity allows mathematicians a privileged ability to figure out the truth or falsity of every proposition. It's tough to argue with this -- our species has only been at this math stuff for a few millennia now, and the types of unprovable propositions we'd be talking about are probably pretty technical. Plus, as Jeff is fond of pointing out, humans are just so goddamn plucky. If giant spaceborn menaces can't get us down, what chance do a few measly theorems have?
But Gödel is clear: computational systems can't get around this restriction. Penrose is pretty sure humans can, so we must be non-computational. Mathematical ability is neatly conflated with consciousness, and we launch off in search of a noncomputational explanation for how minds work. How to keep it grounded in reality? Look for a noncomputational physics. Quantum physics fits the bill.
That's where most reviews of the book end. Coincidentally, it's where a long and confusing explanation of quantum mechanics begins. I won't pretend I understand QM very well, or that I can even begin to argue with Penrose about the mathematical conclusions he reaches in the first part of the book. I don't buy all of his premises, but clearly he's got a better stake to his ideas than I do -- the guy's unquestionably brilliant. Still, I wanted the biological punchline, so I slogged through. The answer he provides is not very convincing.
The biggest problem facing those wishing to come up with a QM accounting of the brain is that quantum effects tend to only occur at small scales. Meaningful quantum interactions happen when states are superposed, their existence nothing but a counterintuitive mush of probability. When this mush interacts with the outside world its probability state collapses into what we think of as conventional reality. Penrose doesn't think this process depends on conscious observation (it's tough to explain how it could), so he needs to find a way for largeish quantum interactions to occur in a messy, warm environment like the brain -- a very tough place for quantum superposition to last for long.
To his credit, Penrose accepts that the "obvious" activity of our brain -- neurons firing action potentials down axons resulting in neurotransmitters carrying messages across synaptic clefts -- falls pretty comfortably within the realm of classical physics. It's a big, hot soup, and quantum effects seem pretty unlikely. So where does he turn? Microtubules. These are protein tubes that form the cytoskeleton, the all-purpose scaffolding of your cells. The inside of these tubes seems like a pretty well-insulated place -- might be some interesting quantum mumbo-jumbo going on in there!
Well, okay. But why should we care? If action potentials don't utilize quantum effects, and we accept that consciousness must be dependent on quantum effects, where the hell is consciousness coming from?
Microtubules seem to be responsible for maintaining the strengths of synapses and, no doubt, for effecting alterations of these strengths when the need arises. Moreover, they seem to organize the growth of new nerve endings, guiding them towards their connectoins with other nerve cells. (364)So this is Penrose's game: he's going to say that neural plasticity -- the modulation of synaptic strength and the creation or destruction of synapses -- is where consciousness lives.
But there are several problems with this. First, saying that the cytoskeleton is responsible for neural plasticity is like saying that growing hair is a task handled by your circulatory system: it'd be hard to achieve the goal without it, but ascribing it a causal role seems like going a bit far.
More importantly, modulation of synaptic strength doesn't seem to correlate very well with consciousness. Although synapses can alter their strength on short time scales temporarily, their adjustment is primarily accomplished through a metabotropic process called long-term potentiation. LTP simply doesn't happen fast enough to correspond with thoughts. Also, it can be prevented through the inhibition of protein synthesis. I doubt anyone has tried injecting undergraduate brains with protein synthesis inhibitors, but it can be done to animals without knocking them out. Penrose's conception of consciousness via varying synaptic strengths might be compatible with a sort of epiphenomenalism, but it's pretty clear this isn't what he has in mind (see his enthusiasm for "intentionality and subjective experience", p. 420).
But the worst problem here is figuring out how the individual cytoskeletons of your brain are supposed to interact with one another. Lesion studies make it obvious that your mind is the result of your brain working together in aggregate -- so how do cells maintain a superposed quantum state across the gaping maw of the synaptic cleft? The separation is typically from a few dozen to a few hundred nanometers -- like I said, I'm no expert, but even Penrose seems to admit that maintaining a superposed state across such wide, uninsulated gaps would be problematic. In fact, he addresses this:
The unity of a single mind can arise... only if there is some form of quantum coherence extending across at least an appreciable part of the entire brain.That isn't exactly convincing. It boils down to "in order for this theory to produce this effect, it would have to overcome this serious problem. The effect exists, though, so nature must have found a way!"
Such a feat would be a remarkable one -- almost an incredible one -- for Nature to achieve by biological means. Yet I believe that the indications must be that she has done so, the main evidence coming from the fact of our own mentality.
Occam would not be pleased, Sir Penrose. It's true that quantum effects can be observed on large scales (even up to meters), but only in very carefully controlled circumstances. To mimic the environment in the brain you'd have to fill your QM experimental apparatus with cerebrospinal fluid. The only thing this seems likely to prove is that your janitorial staff isn't paid enough.
So overall: not very convincing. Penrose is a brilliant guy, and I suspect he's right when he says we're going to need to write some completely new scientific rules to explain where phenomenology comes from. I'm not sure where we'll find them, but I doubt Penrose is leading us in the right direction.