Does Quantum Mechanics Create Consciousness? (Patreon)

Nobel laureate Roger Penrose is widely held to be one of the most brilliant living physicists for his wide-ranging work from black holes to cosmology. And then there’s his idea about how consciousness is caused by quantum processes. Most scientists have dismissed this as a cute eccentricity—a guy like Roger gets to have at least one crazy theory without being demoted from the supersmartypants club. The most common argument for this dismissal is that quantum effects can’t survive long enough in an environment as warm and chaotic as the brain. Well, a new study has revealed that Penrose’s prime candidate molecule for this quantum activity does indeed exhibit large scale quantum activity.  So was Penrose right after all? Are you a quantum entity?

—-

The story starts with Roger Penrose as a young man attending a lecture on the Gödel incompleteness theorems. These are the mathematical arguments by logician Kurt Gödel, which prove the limits of proof. The first incompleteness theorem says that, for any consistent formal mathematical system capable of expressing basic arithmetic, there are true statements in that system that cannot be proven with the rules of that system. 

Here’s an example of an unprovable statement: This statement cannot be proved true in this language. Now imagine there’s an algorithm that can search through all combinations of expressions in the natural language in question until it successfully found the one that proved this statement true. That very act would render the statement false, thereby showing that the system is inconsistent. But if the algorithm couldn’t find such an expression it would demonstrate that the statement—which is that this statement can’t be proved—is true but unprovable. So either the system is inconsistent or contains true but unprovable statements—it's incomplete. Natural languages are certainly inconsistent, but mathematics is designed to be perfectly self-consistent, therefore, mathematics is incomplete. And since any algorithmic system is generally built on mathematics, like classical computation, they too must be incomplete, according to Kurt Gödel.

But Penrose argues that humans CAN “prove” the unprovable. For example, mathematicians seem able to be pretty sure of the truth value of certain conjectures without any formal mathematical proof—and sometimes those formal proofs may not exist. According to Penrose, that means our conscious process of knowing does not come from a process limited by Gödel incompleteness—it’s not algorithmic or computational. Penrose goes further to say that this implies consciousness itself can’t emerge from purely computational processes.

Similar ideas were articulated by philosopher John Lucas, so this whole “consciousness isn’t classical computation because Gödel incompleteness” thing is called the Penrose-Lucas argument.

There are plenty of objections to this argument—for example, that it’s a mistake to equate our sense of knowing to a formal proof. After all, it’s possible to think you know something but be wrong. Or that even if we do have super-Gödel reasoning abilities, it’s a stretch to connect that to consciousness. We’ll leave it to you to check out Penrose’s books “The Emperor's New Mind” and “Shadow Of The Mind” for further details if you’re interested.

OK, let’s get to the weird part. If we’re not classical computers, what are we? Penrose argues that the only place we might possibly find a type of computation or information processing or whatever that’s free from the limits of Gödel incompleteness is … maybe you guessed it, quantum mechanics.

Quantum mechanics describes the world of the tiny. Quantum objects like subatomic particles or even molecules can have very weird behaviors. Their properties like their location or speed tend not to be well defined. They can exist in multiple states or even places at once—in what we call a quantum superposition. And these fuzzy quantum properties can also be correlated with other quantum objects in a mysterious way called quantum entanglement. This is all stuff we’ve described before.

It’s not hard to force a quantum object to give up its quantum fuzziness and take on well-defined values. All you need to do is measure it. Measurement causes an object in a superposition of multiple states to choose one of those states randomly and eliminates any entanglement. We say that measurement “collapses the wavefunction”--where the wavefunction is the mathematical object that defines the distribution of possible results of that measurement. After the measurement, one result is chosen and the wavefunction collapses to that single value.

Exactly what causes this collapse or even what precisely defines a measurement is not known, even 100 years after the discovery of quantum mechanics. We call this mystery the measurement problem. 

One thing we know about wavefunction collapse is that it appears to include a truly random factor—measurement results are chosen by the roll of the dice, albeit dice weighted by the shape of the wavefunction. So, if an event is truly random then almost by definition there is no algorithm that can perfectly determine its outcome. Therefore, Penrose argues, the outcome of any computation or information processing performed by a quantum system involving wavefunction collapse is non-algorithmic, in the sense that the outcome cannot be predicted within the framework of any Gödel-esque mathematical system—and therefore is not necessarily subject to Gödel incompleteness. Therefore maybe conscious reasoning has a quantum component.

There are plenty of criticisms of this—for example, quantum computation is still algorithmic in a sense, even if not classically algorithmic. Also, it’s not clear how the injection of a random process frees one from the constraints of Gödel in the way that Penrose needs.

Another common criticism is that Penrose has fallen for the Holmsian fallacy. To quote the great Sherlock Holmes—When you have eliminated all which is impossible, then whatever remains, however improbable, must be the truth. In fact Penrose himself has quoted Holmes to justify the reasoning that if consciousness can’t be accounted for by things we understand, then it must be accounted for by the other big thing we don’t yet understand—quantum mechanics and the measurement problem. But that assumes that Penrose has comprehensively eliminated, or is even aware of, all other candidate mysteries, which feels presumptuous. On the other hand, this is Roger Penrose we’re talking about, so maybe he really has taken a complete census of all mysteries and is justified in Sherlocking consciousness.

A final objection, and the one that will finally get us to the new result, is that quantum behavior is generally only observable in the most pristine conditions. Typically for individual or smallish collections of subatomic particles, and carefully isolated from the environment. Quantum states decay extremely quickly unless in a vacuum and/or at near absolute zero temperature. And the more particles involved in a quantum state, the more easily the state gets destroyed. That’s why quantum computers are so hard to build.

The inside of the brain is far from a pristine environment—it’s warm and gooey and seems thoroughly macroscopic and classical. In order to do the sort of quantum information processing that the Penrose-Lucas argument demands, a coherent quantum state needs to be maintained for timescales much longer and involve far more particles than should be possible in our meat computers.

Penrose first published his idea in the Emperor’s New Mind, but at the time had no idea how brains might do quantum processing. But then along came Stuart Hameroff. Hameroff is an anesthesiologist who had developed a fascination with consciousness from a young age, and subsequently with a molecular structure inside cells called microtubules, which he suspected may be involved. 

Microtubules are these tiny tubes that play many roles in every cell in your body. They’re a major part of Eukariot cell’s skeleton—the cytoskeleton—stabilizing its shape. They act as conveyor belts, moving proteins around. They even play a key role pulling chromosomes apart when a cell divides. They’re constantly being assembled and disassembled and adapt to what the cell needs at any given moment. A given cell might have billions of microtubules.

Microtubules also have an extremely regular structure—almost crystal-like. They are made of alternating tubulin molecules of two different types. Those molecules each have a polarity which can point one way or another. This structure got Hameroff to thinking maybe microtubules could in some way work as molecular computers, or at least molecular information storage. Add to this two more facts: microtubules are more abundant and differently structured in neurons than any other cell. Also, there’s evidence that anesthetics may act by disabling microtubules—thereby “disabling consciousness”. So it’s not so crazy that Hameroff wondered whether these molecules might actually have some fundamental role in consciousness. 

And then Hameroff read the Emperor’s New Mind and reached out to Penrose with the idea that microtubules could be a candidate for this quantum information processor inside neurons. That was in the early nineties. Since then Penrose and Hameroff and others have developed some fairly involved scenarios for how this might work. The basic idea is that information gets stored across one or many microtubules, perhaps even across multiple neurons. The information in the form of quantum bits—qubits—could be stored in a variety of ways, for example the polarization direction of individual tubulins. 

So you have these quantum states that are networks of entangled qubits, and they are in superposition. That means many possible configurations of the 1s and 0s in the qubits exist simultaneously across microtubules. And then something happens to collapse this superposition—some type of measurement causes a single state to be chosen from the many. And it’s that moment of collapse that Penrose thinks is a sort of proto-conscious moment. So our conscious experience would then be the sum of these moments happening all the time across the brain.

By the way, Penrose also has a mechanism for this wavefunction collapse. It’s called objective reduction, and it happens when the superposition of different spacetime curvatures corresponding to the different quantum states reaches a certain threshold, causing one quantum state to be chosen. We did an episode on objective reduction if you want the details.

The overall theory is called orchestrated objective reduction, and the idea is that the brain uses this wavefunction collapse as part of its information processing mechanism, and this leads to conscious experience and also the ability to transcend classical computation.

Orchestrated objective reduction has been around now for over 30 years, and frankly very few people took it seriously—and I’d guess most still don’t. Some don’t agree with the Penrose-Lucas argument, and fewer believe that coherent quantum states can persist for long enough inside the messy environment of the cell interior. 

Which brings us to the reason we’re doing this episode now. Evidence is emerging that microtubules may exhibit interesting quantum behaviors after all. In 2013 evidence was found that microtubules display large-scale quantum resonance, giving them unusual long-range electrical conduction properties and potentially allowing them to work as memory switching elements. And just a few months ago, a paper came out adding weight to this idea. It’s by Nathan Babcock and Phillip Kurian from Howard University and their collaborators, and is titled “Ultraviolet Superradiance from Mega-Networks of Tryptophan in Biological Architectures”. Those meganetworks of tryptophan are basically microtubules. But to grok the rest of that title we need to understand superradiance.

So, normally, if you radiate a bunch of atoms or molecules with a bunch of photons, each might absorb a photon, bumping an electron to a higher energy level, and then radiate another photon as the electron loses that energy. When the emitted photon has lower energy then the process is called fluorescence. The de-excitation of many excited electrons takes a bit of time as they decay randomly one by one, so fluorescent materials glow for some time after being illuminated. 

Superradiance is a bit different. It can occur when a group of atoms or molecules get into excited states as an entangled group rather than individually. That can happen, for example, if the spacing between excited particles is smaller than the wavelength of the incoming photons, then we can’t say which particle absorbed which photon so we have to treat the collection of particles as a single quantum superposition state of all possibilities of the absorption matchups. 

In any case, if the system of particles is in a collective excited state then it tends to decay collectively, and so photons are emitted together. That results in a much brighter and shorter spike of radiation than regular fluorescence. Another way to think of this is that the emission of photons by one electron in the ensemble prompts many identically excited electrons to decay, amplifying the first electromagnetic wave. Which, by the way, is also exactly how lasers work. Superradiance is sort of a scaled down example of the laser phenomenon.

So, superradiance is a very quantum effect, and these researchers claim to have observed it in actual microtubules. More precisely, they claim to see superradiance in the tryptophan amino acids within tubulin molecules when radiated with ultraviolet light.

If this is right it would require large-scale quantum entanglement of tryptophan molecules across a microtubule. Although the researchers of the paper didn’t directly measure that entanglement or its source, according to their models the amount of light emitted was 100s of times that expected from simple fluorescence, and so could have only been possible through a superradiant process.

They also did a bunch of quantum simulations to explore the effect, and found that these entangled, excited states can extend a long way along microtubules—far longer than would normally be expected in an environment like the brain. This implies that microtubules could at the very least serve a role in cell signaling. Finally, they found that this coherence was more stable the larger the microtubule and microtubule network, which gives some hope that this effect might actually appear in a real neuron. On the other hand, Penrose and Hameroff estimate that a good fraction of all microtubules in all the neurons in the brain need to be entangled to generate human consciousness—and that level of warm, wet, macroscopic entanglement is pretty hard to believe.

So what are the implications of this? Well forget consciousness—even if microtubules contribute to cognition, we may be way further from artificial general intelligence than some believe. Brains have around 10^14 synapses. AGI optimists have said we’ll be able to simulate human intelligence when we have computers capable of simulating that many connections … which Moores law tells us will be pretty soon. However, if neurons are also doing internal computation with microtubules, and there are a billion microtubules per neuron computing maybe a million times faster than the neuron firing rate, then we may need to wait quite a bit longer to make our first AGI. And at any rate, if orchestrated objective reduction or anything like it is right, no AI we make in the near future has any hope of human-like consciousness. If we want that, and I’m not sure we want that, we’re going to have to build our AIs on quantum computers.

After a good mix of science, philosophy, and speculation, it's appropriate to finish with a little disclaimer. What the microtubules superradiance really shows is that there are potentially quantum processes happening in the brain. There is of course no claim in the recent paper that this relates to consciousness at all. Consciousness only comes into play if you accept the arguments of Penrose, Lucas, and Hameroff. We’ll leave it up to you to take a philosophical stance there. But I will say that I’ve always been pretty dismissive of Penrose’s idea, but this has me looking at it a bit more closely—and the same is surely true of others. A lot of work needs to be done to establish that actual quantum information processing is happening in microtubules and to even establish a concrete mechanism by which this phenomenon could generate our conscious experience. On the other hand,  don’t you kind of feel like you’re a massively entangled quantum state collapsing under the buildup of superposed spacetime.