Interview

MYSTERY OF THE MICROTUBULES

ENLIGHTENNEXT: You’re best known as one of the world’s leading proponents of a quantum-physics-based theory of the mind. How did you first become interested in the mystery of consciousness?

STUART HAMEROFF: I got interested while in college in the late 1960s. Studying mostly science and math, I took a course called Philosophy of Mind and was intrigued with how difficult was the problem of explaining how conscious experience arises from the pinkish-gray meat we call the brain. And I remained interested through medical school, being drawn toward fields having to do with consciousness—psychiatry, neurology, neurosurgery. But one day, while doing a research project in a cancer lab in the early 1970s, I was looking at cells dividing—mitosis—under a microscope, observing how the DNA-containing chromosomes were separated and pulled apart into perfectly equal mirror images of each other. The tiny strands and little machines moving the chromosomes were called microtubules and centrioles (which were themselves composed of microtubules). The dance of the chromosomes had to be perfect, because if they divided unequally, abnormal cancer cells could result.

Most of my research colleagues followed the trail of the chromosomes and went into gene-based research. But for some reason, I became fixated on how these little molecular machines knew exactly what to do and how they were choreographed. I wondered how they were organized and guided, and whether there was some intelligence, if not consciousness, at that level. Around the same time, it was discovered that these same microtubules existed in all cells—especially neurons—as major components of the cell skeleton, or structural scaffolding. Being highly asymmetrical, brain neurons are just full of microtubules. So it occurred to me that microtubules, which seemed to display some kind of intelligence or consciousness in cell division, might have something to do with consciousness in brain neurons. Maybe in addition to being the cell’s structural support, microtubules were also the cell’s on-board computer.

After medical school in Philadelphia I gave brief consideration to a full-time research career, but decided to take a clinical internship in Tucson, Arizona, to figure out what I wanted to do next. I was leaning toward neurology, but then met the chairman of anesthesiology at the new University of Arizona medical school hospital who was in need of residents for his fledgling program. He was a Boston-trained Texan named Burnell Brown, and he showed me around the operating rooms at the new hospital, explaining how anesthesiology was important, paid well, and could be fun. As he came to know my interests, he told me that if I really wanted to understand consciousness, I should figure out how anesthesia works, because anesthesia selectively erases consciousness while sparing other brain functions. He also showed me a paper written by a colleague of his in 1968, suggesting that if you apply the gases used in anesthesia to microtubules, they depolymerize—they fall apart. So there was a theory that anesthesia worked by causing brain microtubules to fall apart. It turns out, fortunately, that that’s not true. You need about five times the amount of anesthesia for microtubule depolymerization than you need to cause somebody to lose consciousness. But it showed that anesthetics do affect microtubules, which further suggested that microtubules might have something to do with consciousness.

EN: What, exactly, is a microtubule?

SH: First and foremost microtubules are the rigid structural support defining the shape of all animal cells, but continually moving and rearranging. The rearrangements account for all cell growth, development, movement, synaptic regulation—pretty important stuff. Each microtubule is a molecular assembly, a cylindrical polymer composed of many versions of a single, peanut-shaped protein called tubulin. Each of these tubulins can flex into alternative conformational shapes, and can also have genetic and other types of diversity, but are overall similar. The tubulin proteins self-assemble into hollow cylinders whose walls are elegant lattices which are both hexagonal and helical, the helical winding patterns having beautiful Fibonacci geometry. I became somewhat obsessed, enchanted really, with the structure of microtubules. These self-assembling and unassuming cylinders somehow accounted for cell growth, movement, and function. Their actions reminded me of the “Indian rope trick” where the Fakir tosses up a rope, climbs it, and then disappears. Except there’s no Fakir, just self-assembling proteins forming the cytoskeleton, the bone-like structural support or scaffolding, inside all animal cells. Like a building assembling itself, brick by brick. And the more asymmetrical a cell is, the more it needs the structural support. So neurons with their long axons and dendrites have lots of microtubules. If you look inside a single neuron, you see hundreds of microtubules composed of something like one hundred million tubulin protein subunits. You could say that neurons are actually made of microtubules. So I just figured that if microtubules were organizing complex activities during rudimentary cell division, then they might be doing something similar in brain neurons related to consciousness.

EN: Interesting! Most people think that consciousness arises from activity between brain cells, or neurons, but you’re saying, well, no, it may actually be these extraordinarily tiny structures within neurons that provide the real physical basis for consciousness.

SH: Yes, exactly. Most views consider the brain-as-computer, with neuronal firings acting as “bits.” Neurons are seen as simple fundamental components of brain information processing, able to perform simple logic functions. But I began to think the mechanisms for consciousness went deeper. A couple of other things helped lead me in this direction. The first was that I looked at single-celled organisms like paramecia. A paramecium is one cell and therefore has no neurons or synapses. But it swims around, finds food, avoids obstacles and predators, finds a mate, has sex, and can learn. It seems to have some intelligence. Not necessarily consciousness, but the single-celled creature definitely has cognitive functions—“cognition” meaning sensory processing, control of behavior, and so forth. It has intelligence and yet no neurons nor synapses. It does, however, have microtubules, and organelles called cilia composed of microtubules which act as both sensors and motors. This suggested to me that a paramecium might use its microtubules to process information and organize its behavior. And if paramecia did so, why wouldn’t neurons?

The second thing was that, around the time I learned about microtubules, I also began to read about computer switching matrices, lattices, and networks. The structure of microtubules consisted of a cylindrical lattice of tubulin proteins. Each of these could switch between different conformational states, and be programmed by genetics and other factors, and be influenced by lattice neighbors, like gates and switches in computers. This was more support for the notion that microtubules might be acting not only as bone-like support, but also as molecular-scale computers, the intelligence system inside cells.

EN: So you basically started to realize that there’s actually a lot more activity—and maybe even conscious activity—going on inside the brain than most people imagine?

SH: That’s right—I saw more intelligence at a deeper level inside neurons, specifically in microtubule computation. Most views saw the brain as a computer with one hundred billion simple, dumb neurons interacting together to produce something intelligent and conscious. I thought each neuron at the level of its microtubules had significant information processing and intelligence. I had a hunch that microtubules were “Nature’s computers.”

So I started working with engineer and physicist colleagues to model and simulate tubulin states in microtubule lattices. We assumed each tubulin could be in two alternative states correlating with its dipole, and that neighboring dipoles interacted, or computed according to rules set by the microtubule geometry—very much like cellular automata, self-organizing computers. We also assumed the computational interactions were synchronized by coherent excitations on the scale of nanoseconds. Based on these assumptions, I worked with my colleagues Rich Watt, Steen Rasmussen, Jack Tuszynski, et al., and showed that microtubules were well suited to be efficient computational devices. Based on about ten million tubulins per neuron and nanosecond-range computations, we calculated that microtubules within each neuron in the brain could perform roughly 1016 operations per second. That was twenty years ago. Recent evidence has shown slightly slower coherent microtubule excitations of about one-hundred nanoseconds, and ten times more tubulin per neuron, so a revised estimate would be about 1015 operations per second per neuron for microtubule information processing. So instead of each neuron registering as a single bit in the computer of the brain—a one or a zero, firing or not firing—there was another layer of microtubule processing deeper inside each neuron, raising the potential computational complexity of the brain tremendously.

This was in the 1980s and early 90s, and I was going to a lot of artificial intelligence and neural network conferences where they were trying to model and simulate the brain as many simple neuronal switches. The Singularity people are still trying to do that. Considering brain computation strictly at the level of neuronal synaptic interactions, they estimate one hundred billion neurons per brain, each with up to one thousand synapses per neuron, and up to one-hundred operations per second per synapse. This gives roughly 1016 operations per second for the entire brain. Using Moore’s law for the minaturization and speedup of computer components, they were forecasting brain equivalence—and hence consciousness—in a few decades.

That 1016 was familiar. It was what we had calculated for one neuron at the microtubule level. I was saying, “No. Each of your simple switches is incredibly complicated. You have to take into account this added computational complexity. Each neuron has 1016 operations per second. The computational capacity of the brain is squared!” They didn’t like that very much. If correct, it pushed their goal of simulating a human brain way down the road. So I became kind of unpopular among that crowd.

But then one day someone said to me, “Okay, let’s say you’re right. Let’s say each neuron has all this enormous added computation going on. How would that explain conscious experience? How would that explain why we have feelings, why we see red, why we feel pain? How does that explain consciousness?” And I realized I didn’t have an answer to that, which brings us to what the Australian philosopher David Chalmers famously dubbed the “hard problem” of consciousness research.

EN: The question of how we get mind out of matter.

SH: Exactly.

 

ROGER PENROSE & SCHRÖDINGER’S CAT

SH: Fortunately, someone suggested that I read a book by the English mathematical physicist Sir Roger Penrose called The Emperor’s New Mind. So I did, and it was really amazing. The book’s title was intended as a slap in the face to artificial intelligence theorists because they maintained that if you had sufficiently complex computation in a computer, it would be conscious. But Roger argued—in a somewhat obscure mathematical direction called Gödel’s theorem—that consciousness involves something noncomputable, that understanding, or awareness, is not a computation. Something else is involved. So after ruling out the idea that consciousness was strictly a computation, Penrose then offered a mechanism for consciousness that involved something so far out of left field that most people considered it—and still consider it—rather bizarre. And that has to do with quantum physics, and in particular, quantum gravity.

Reading The Emperor’s New Mind, I was floored with the breadth and subtlety of Penrose’s knowledge, much of which I didn’t understand. I did know that anesthetic gases exert their effects by quantum forces, so consciousness having something to do with quantum physics made sense to me. And I had this gut feeling that he was onto something. He at least had a mechanism for consciousness. It was based on a particular type of quantum computation in the brain having something to do with quantum gravity—the fabric of spacetime geometry. I learned that quantum computation required information, e.g., “bits” of 1 or 0, to exist for a time in quantum ‘superposition’ of coexisting possibilities—quantum bits, or “qubits” of both 1 and 0. After interacting/computing the qubits then reduce, or collapse to bits, e.g., 1 or 0 as the answer.  Roger was proposing a new explanation for the reduction, or collapse based on quantum gravity, and associated with consciousness. But on the brain side, he didn’t have a strong candidate for a qubit, suggesting possibly superpositions of neurons both firing and not firing.

I said to myself, well, maybe tubulins are qubits and microtubules are the quantum computers Penrose is looking for. So I wrote to him and we soon met in his office at Oxford.

Roger is a gentle, unassuming man, despite being incredibly brilliant and highly acclaimed. He mentioned he was going to a consciousness conference at Cambridge, and had me do almost all the talking. So I just started talking about microtubules and showed him the 1987 book I’d written on the subject. He listened intently, asked questions and was particularly taken by the Fibonacci geometry of the microtubule lattice. After several hours, he finally said, “Well, that’s very interesting.” I said goodbye and didn’t think anything was going to come of it. But about two weeks later, I was having dinner with some friends in London and they said, “Guess what? We were at this conference at Cambridge and Roger Penrose was talking about you and your microtubule stuff.” Soon after that, I received an invitation to a conference in Sweden that Roger was attending, and we struck up a friendship and decided to start developing a formal model of consciousness based on his theory of quantum gravity and the possibility of quantum computation in microtubules in the brain. I also invited him to speak at the first Tucson conference, Toward a Science of Consciousness in 1994.

EN: Pretend I don’t know anything about quantum physics. Could you explain what a quantum superposition is? And how it relates to consciousness or microtubules?

SH: Quantum means, literally, the smallest fundamental unit of energy, like a photon—an indivisible unit of light. But behavior at the quantum level is bizarre. It’s so bizarre, it’s like another world. In fact reality seems to be divided into two different worlds—the classical world and the quantum world. The classical world is our everyday familiar world, in which Newton’s laws of motion, electromagnetism and other basic physics pretty much describe everything very well. If you throw a ball, its trajectory, speed, location and so forth can be easily predicted. But as we go to smaller scales—let’s say, for argument’s sake, atoms and smaller—we enter a world where completely different physical laws apply, and predictions become a lot more difficult. For example, at the quantum level particles can be in two or more places or states at the same time. Instead of being either here or there, particles can be both here and there simultaneously, more like smeared-out waves than particles, and governed by a quantum wave function. And when pairs of superpositioned particles are separated, they remain somehow connected. This is called entanglement, or what Einstein called “spooky action at a distance.” But we don’t see this other world. And some say that’s because quantum superpositions collapse, or reduce to classical systems—the wave function collapses—only when consciously observed.

EN: This means that a human observer is required to collapse a state of superposition?

SH: In one interpretation of quantum physics, yes. The Danish physicist Niels Bohr popularized this model, which became known as the Copenhagen interpretation. Early experiments seemed to show that if a machine measured a quantum system, the results in the machine remained in superposition until observed by a conscious human, that consciousness “collapsed the wave function.” This put consciousness outside science, but allowed Bohr and others to continue experiments without worrying about any deeper meaning for reality or consciousness.

If you take Copenhagen to its extreme, you might suppose that if you’re sitting in a room and there’s a picture hanging behind you, then the picture may be smeared out in multiple places at once until you turn around and look at it. In other words, anything unobserved would be in a wave-like state of quantum superposition. That idea is pretty bizarre, however, and Erwin Schrödinger, another early quantum physics pioneer, thought it was downright silly. So he came up with his famous thought experiment, called Schrödinger’s Cat, to try to demonstrate how nonsensical it was.

But the question raised by Schrödinger’s thought experiment remains: How big can a quantum superposition get? Can isolated quantum systems be amplified so that something as large as a cat can be in two states simultaneously? There’s still no answer to that, but the question has led physicists to come up with alternatives to the Copenhagen interpretation—different models of wave function collapse that don’t necessarily require a conscious observer.