From Science Fiction to Science Fact
From: Minds and Machines
As any marketing professional will tell you, knowing how to properly brand an idea is essential for its success. Indeed, all too often, it can be even more important than the actual idea itself.
The field of phrenology is a good case in point. A deliberately constructed combination of two heavyweight Greek concepts — phren (mind) and logos (structured knowledge) — its very existence argued strongly for its inherent validity. After all, how could a field dedicated to amassing a structured knowledge of the mind be anything less than rigorously scientific?
Well, it turns out that it can. Phrenology had its day, of course. And to be fair, when developed by Franz Joseph Gall in the late 18th century, it represented a significant advance in our understanding, emphasizing as it did the importance of explaining mental states through the neurophysiology of the brain rather than through the previous window of religious or philosophical abstraction.
But by the time people started engaging in detailed measurements of skull sizes to determine which one of the 27 arbitrary so-called “brain organs” was most responsible for someone’s personality, it was clear that the field had descended to the depths of pseudoscience from which it would never recover.
Yet aspects of its legacy persist. According to Duke University neuroscientist Miguel Nicolelis, Gall’s general framework of dividing the brain into regions that control separate aspects of human behaviour, “morphed into one of the key dogmas of twentieth-century neuroscience”. A dogma, as it happens, that Prof. Nicolelis is firmly convinced is, if not flat-out wrong, at least deeply misleading.
What he can’t accept is the pre-eminence that classical theories of the brain give to a sense of locality: that the brain is partitioned into strict divisions of different types of neurons responsible for different types of information processing. For him, the fundamental thinking unit isn’t a neuron at all, but rather populations — clusters — of neurons that are themselves distributed throughout many different areas of the brain. And when the brain acts, he believes, it does so by integrating all these distributed clusters of neurons.
“It is like the population of neurons is voting at each moment in time and the real outcome depends on this voting. It doesn’t depend on a specific class of cells or a cluster of neurons. It depends on this distributed representation.”
Well, people have argued about how the brain works for centuries. What separates Miguel from his many predecessors is that he has managed to construct a good many convincing experiments that back up his hypotheses.
For much of his research career, he has distributed detectors throughout the brains of rats, monkeys and humans to directly measure their electrical signals. This, in itself, is astounding enough.
But what seems straight out of the realm of science fiction is that he also managed to control these signals and relay them to machines to enable the animals to independently control them just by thinking. And not just once or twice, mind you: over and over again.
Once he managed to get rats to drink from mechanical arms merely using their brains, he upped the stakes by illustrating how monkeys could be trained to think their way to remotely controlling a computer cursor.
Still, scepticism remained. After all, both experiments were using brain waves that had originally come from signals related to their upper limbs—the rats had been first trained to press a bar with their forepaw, while the monkeys had been moving a joystick with their arms. Perhaps, people thought, it was impossible to fully harness lower limb signals.
Now, frankly, this seems a pretty curious objection to me. After all, once you get to the point where you can manipulate animal thoughts to regularly move robotic devices, it seems you’ve proven your point. But suffice it to say that I’m not in the field. And apparently, for those that were, this upper limb/lower limb distinction was a pretty big deal.
So off goes Miguel, determined once again to show the critics wrong. He trains a Rhesus monkey to walk upright on a treadmill and measures the monkey’s brain waves. Then he sends these signals to a processor that controls a 100 kg automated robot, so that as long as the monkey is preoccupied with walking, those walking-oriented brain-wave signals are converted into propelling the robot into moving as well. Meanwhile while the real-time image of the robot moving is beamed back to a “video screen directly in front of the monkey on his treadmill.
So the monkey, just like many of us in our health clubs, gets used to walking on a treadmill and watching a large video screen. Except weight loss here is not the issue, the goal is simply to keep the robot walking. Because the monkey quickly learns that as long as the video screen shows the robot walking, he regularly gets a shot of his favourite fruit juice.
So what happens?
Well, the treadmill is turned on. The monkey walks, the robot walks. Then the treadmill is turned off and the monkey, strapped to a device and unable to move except on the treadmill, is forced to stop walking as well. But the robot keeps walking. Why? Because even though he is no longer physically walking, the monkey has made the connection: he knows that if he is still thinking about walking he can still make the robot move with those thoughts and still get his fruit juice.
Oh yes, I forgot one little thing: Miguel and his colleagues at Duke decided to put the robot the monkey was controlling in Kyoto.
“We got a 5 kilogram monkey to control a 100 kilogram, 150 centimetre robot in Japan by physically generating movement out of brain waves. And we showed it could be done, it could be walking — it doesn’t have to be just upper limbs anymore. And it could be done across the planet very efficiently. In other words, we scale space.”
All very remarkable, you might think. But so what? Controlling robots by brain waves is not generally something that most humans need to worry about doing either. But that, of course, is profoundly missing the point.
“I think that in the short run, over the next few years, the main impact of this thing we call ‘brain-machine interface’ will be in medical rehabilitation. No doubt about it. Patients who are paralyzed will benefit from this possibility of bypassing the lesion and using brain activity to control prosthetic devices of a huge variety: single limb, lower limb, whole body, upper limb. There is now a huge diversity of potential devices. Then there’s communication. For people who cannot communicate, they will be able to use their brain activity to communicate. It’s not only paralysis.
“We’re working here in the lab on prosthetic devices for Parkinson’s disease that take advantage of the basic science that we’ve discussed regarding this new model of the brain. They would never work if the brain operated on the classic model.”
For those of us anxious to focus on the boundless medical possibilities, it’s easy to gloss over that last point. But it’s hugely significant as well, because Miguel is convinced that his research not only directly paves the way for a significantly enhanced quality of life for vast numbers of people, it also provides emphatic evidence that our core understanding of how the brain works needs to be dramatically overhauled.
He doesn’t deny, of course, that there is a high degree of specialization of brain processing that occurs in specific regions. But his main point is that just looking at our brains as localized regions of specific activity is missing the big picture, which is all about the broad connecting circuits that are distributed throughout the entire brain.
“There is a degree of specialization, no doubt about it. But it is not as strict as we were led to believe, and is not phrenology, not by a long shot.
“This is still a tough debate, particularly in the vision community where there’s long been a natural focus on a localized approach. But the evidence is growing so much. Here’s another point to mention: we can find visually-driven neurons in the tactile cortex, in the motor cortex. So, it’s not just in one place. And I think many people would agree with me on that right now.”
Since brain-machine interfaces naturally involve combining a wide variety of different sensory inputs with other areas of brain processing, it seems the perfect arena to emphatically illustrate Miguel’s central thesis of how our brains fundamentally operate, a perspective which unsurprisingly has the broadest possible relevance imaginable.
“I truly believe that we’re going to see other ways of brain-machine interfaces, not only for patients with neurological disorders or psychiatric disorders. We’ll see this technology advance using non-invasive methods: so, no sensors inside the brain but from the outside. This will allow us to have a completely different experience of interacting with our computers and anything that is digitally controlled.
“In a couple decades we will very likely be a part of our desktops, our laptops, our iPads or whatever: we will be a part of the operating system in the sense that we will be interacting by thinking, with icons, with applications and so forth. And we’ll get feedback. We are going to assimilate the operating system.”
It sounds a bit out there, I have to admit. But underestimating Miguel Nicolelis strikes me as decidedly unwise: he’s definitely earned the benefit of the doubt.
This is the introduction written by Howard Burton of the book, Minds and Machines, which is based on an in-depth, filmed conversation between Howard and Miguel Nicolelis, Professor of Neurobiology, Neurology, Neurosurgery, Biomedical Engineering, Psychology and Neuroscience and Orthopaedic Surgery and Co-Director of the Center for Neuroengineering at Duke University. The book is broken into chapters and includes questions for discussion at the end of each chapter. The book is also available as part of the 5-part Ideas Roadshow Collection called Conversations About Neuroscience.
Visit the dedicated page for our conversation with Miguel Nicolelis on Ideas On Film: https://ideas-on-film.com/miguel-nicolelis/. On our Ideas On Film YouTube channel you can watch a clip from the filmed conversation: https://youtu.be/nAIDabkjDdQ.
