by Michio Kaku
In the short term, this is nothing less than miraculous for people who are totally paralyzed. One day, they are trapped, helpless, in their bodies; the next day, they are surfing the Web and carrying on conversations with people around the world.
(I once attended a gala reception at Lincoln Center in New York in honor of the great cosmologist Stephen Hawking. It was heartbreaking to see him strapped into a wheelchair, unable to move anything but a few facial muscles and his eyelids, with nurses holding up his limp head and pushing him around. It takes him hours and days of excruciating effort to communicate simple ideas via his voice synthesizer. I wondered if it was not too late for him to take advantage of the technology of BrainGate. Then John Donoghue, who was also in the audience, came up to greet me. So perhaps BrainGate is Hawking’s best option.)
Another group of scientists at Duke University have achieved similar results in monkeys. Miguel A. L. Nicolelis and his group have placed a chip on the brain of a monkey. The chip is connected to a mechanical arm. At first, the monkey flails about, not understanding how to operate the mechanical arm. But with some practice, these monkeys, using the power of their brains, are able to slowly control the motions of the mechanical arm—for example, moving it so that it grabs a banana. They can instinctively move these arms without thinking, as if the mechanical arm is their own. “There’s some physiological evidence that during the experiment they feel more connected to the robots than to their own bodies,” says Nicolelis.
This also means that we will one day be able to control machines using pure thought. People who are paralyzed may be able to control mechanical arms and legs in this way. For example, one might be able to connect a person’s brain directly to mechanical arms and legs, bypassing the spinal cord, so the patient can walk again. Also, this may lay the foundation for controlling our world via the power of the mind.
MIND READING
If the brain can control a computer or mechanical arm, can a computer read the thoughts of a person, without placing electrodes inside the brain?
It’s been known since 1875 that the brain is based on electricity moving through its neurons, which generates faint electrical signals that can be measured by placing electrodes around a person’s head. By analyzing the electrical impulses picked up by these electrodes, one can record the brain waves. This is called an EEG (electroencephalogram), which can record gross changes in the brain, such as when it is sleeping, and also moods, such as agitation, anger, etc. The output of the EEG can be displayed on a computer screen, which the subject can watch. After a while, the person is able to move the cursor by thinking alone. Already, Niels Birbaumer of the University of Tübingen has been able to train partially paralyzed people to type simple sentences via this method.
Even toy makers are taking advantage of this. A number of toy companies, including NeuroSky, market a headband with an EEG-type electrode inside. If you concentrate in a certain way, you can activate the EEG in the headband, which then controls the toy. For example, you can raise a Ping-Pong ball inside a cylinder by sheer thought.
The advantage of the EEG is that it can rapidly detect various frequencies emitted by the brain without elaborate, expensive equipment. But one large disadvantage is that the EEG cannot localize thoughts to specific locations of the brain.
A much more sensitive method is the fMRI (functional magnetic resonance imaging) scan. EEG and fMRI scans differ in important ways. The EEG scan is a passive device that simply picks up electrical signals from the brain, so we cannot determine very well the location of the source. An fMRI machine uses “echoes” created by radio waves to peer inside living tissue. This allows us to pinpoint the location of the various signals, giving us spectacular 3-D images of inside the brain.
The fMRI machine is quite expensive and requires a laboratory full of heavy equipment, but already it has given us breathtaking details of how the thinking brain functions. The fMRI scan allows scientists to locate the presence of oxygen contained within hemoglobin in the blood. Since oxygenated hemoglobin contains the energy that fuels cell activity, detecting the flow of this oxygen allows one to trace the flow of thoughts in the brain.
Joshua Freedman, a psychiatrist at the University of California, Los Angeles, says: “It’s like being an astronomer in the sixteenth century after the invention of the telescope. For millennia, very smart people tried to make sense of what was going on up in the heavens, but they could only speculate about what lay beyond unaided human vision. Then, suddenly, a new technology let them see directly what was there.”
In fact, fMRI scans can even detect the motion of thoughts in the living brain to a resolution of .1 millimeter, or smaller than the head of a pin, which corresponds to perhaps a few thousand neurons. An fMRI can thus give three-dimensional pictures of the energy flow inside the thinking brain to astonishing accuracy. Eventually, fMRI machines may be built that can probe to the level of single neurons, in which case one might be able to pick out the neural patterns corresponding to specific thoughts.
A breakthrough was made recently by Kendrick Kay and his colleagues at the University of California at Berkeley. They did an fMRI scan of people as they looked at pictures of a variety of objects, such as food, animals, people, and common things of various colors. Kay and colleagues created a software program that could associate these objects with the corresponding fMRI patterns. The more objects these subjects saw, the better the computer program was at identifying these objects on their fMRI scans.
Then they showed the same subjects entirely new objects, and the software program was often able to correctly match the object with the fMRI scan. When shown 120 pictures of new objects, the software program correctly identified the fMRI scan with these objects 90 percent of the time. When the subjects were shown 1,000 new pictures, the software program’s success rate was 80 percent.
Kay says it is “possible to identify, from a large set of completely novel natural images, which specific image was seen by an observer …. It may soon be possible to reconstruct a picture of a person’s visual experience from measurements of brain activity alone.”
The goal of this approach is to create a “dictionary of thought,” so that each object has a one-to-one correspondence to a certain fMRI image. By reading the fMRI pattern, one can then decipher what object the person is thinking about. Eventually, a computer will scan perhaps thousands of fMRI patterns that come pouring out of a thinking brain and decipher each one. In this way, one may be able to decode a person’s stream of consciousness.
PHOTOGRAPHING A DREAM
The problem with this technique, however, is that while it might be able to tell if you are thinking of a dog, for example, it cannot reproduce the actual image of the dog itself. One new line of research is to try to reconstruct the precise image that the brain is thinking of, so that one might be able to create a video of a person’s thoughts. In this way, one might be able to make a video recording of a dream.
Since time immemorial, people have been fascinated by dreams, those ephemeral images that are sometimes so frustrating to recall or understand. Hollywood has long envisioned machines that might one day send dreamlike thoughts into the brain or even record them, as in movies like Total Recall. All this, however, was sheer speculation.
Until recently, that is.
Scientists have made remarkable progress in an area once thought to be impossible: taking a snapshot of our memories and possibly our dreams. The first steps in this direction were taken by scientists at the Advanced Telecommunications Research (ATR) Computational Neuroscience Laboratory in Kyoto. They showed their subjects a pinpoint of light at a particular location. Then they used an fMRI scan to record where the brain stored this information. They moved the pinpoint of light and recorded where the brain stored this new image. Eventually, they had a one-to-one map of where scores of pinpoints of light were stored in the brain. These pinpoints were located on a 10 × 10 grid.
(photo credit 1.3)
Then the scientists flash
ed a picture of a simple object made from these 10 × 10 points, such as a horseshoe. By computer they could then analyze how the brain stored this picture. Sure enough, the pattern stored by the brain was the sum of the images that made up the horseshoe.
In this way, these scientists could create a picture of what the brain is seeing. Any pattern of lights on this 10 × 10 grid can be decoded by a computer looking at the fMRI brain scans.
In the future, these scientists want to increase the number of pixels in their 10 × 10 grid. Moreover, they claim that this process is universal, that is, any visual thought or even dream should be able to be detected by the fMRI scan. If true, it might mean that we will be able to record, for the first time in history, the images we are dreaming about.
Of course, our mental images, and especially our dreams, are never crystal sharp, and there will always be a certain fuzziness, but the very fact that we can look deeply into the visual thoughts of someone’s brain is remarkable.
Reading thoughts via EEG (left) and fMRI (right) scans. In the future, these electrodes will be miniaturized. We will be able to read thoughts and also command objects by simply thinking. (photo credit 1.4)
ETHICS OF MIND READING
This poses a problem: What happens if we can routinely read people’s thoughts? Nobel laureate David Baltimore, former president of the California Institute of Technology (Caltech), worries about this problem. He writes, “Can we tap into the thoughts of others? … I don’t think that’s pure science fiction, but it would create a hell of a world. Imagine courting a mate if your thoughts could be read, or negotiating a contract if your thoughts could be read.”
Most of the time, he speculates, mind reading will have some embarrassing but not disastrous consequences. He writes, “I am told that if you stop a professor’s lecture in midstream … a significant fraction [of the students] are involved in erotic fantasies.”
But perhaps mind reading won’t become such a privacy issue, since most of our thoughts are not well defined. Photographing our daydreams and dreams may one day be possible, but we may be disappointed with the quality of the pictures. Years ago, I remember reading a short story in which a man was told by a genie that he could have anything he could imagine. He immediately imagined expensive luxury items, like limousines, millions of dollars in cash, and a castle. Then the genie instantly materialized them. But when the man examined them carefully, he was shocked that the limousine had no door handles or engine, the faces on the bills were blurry, and the castle was empty. In his rush to imagine all these items, he forgot that these images exist in his imagination only as general ideas.
Furthermore, it is doubtful that you can read someone’s mind from a distance. All the methods studied so far (including EEG, fMRI, and electrodes on the brain itself) require close contact with the subject.
Nonetheless, laws may eventually be passed to limit unauthorized mind reading. Also, devices may be created to protect our thoughts by jamming, blocking, or scrambling our electrical signals.
True mind reading is still many decades away. But at the very least, an fMRI scanner might function as a primitive lie detector. Telling a lie causes more centers of the brain to light up than telling the truth. Telling a lie implies that you know the truth but are thinking of the lie and its myriad consequences, which requires much more energy than telling the truth. Hence, the fMRI brain scan should be able to detect this extra expenditure of energy. At present, the scientific community has some reservations about allowing fMRI lie detectors to be the last word, especially in court cases. The technology is still too new to provide a foolproof lie-detection method. Further research, say its promoters, will refine its accuracy. This technology is here to stay.
Already, there are two commercial companies offering fMRI lie detectors, claiming a more than 90 percent success rate. A court in India already has used an fMRI to settle a case, and several cases involving fMRI are now in U.S. courts.
Ordinary lie detectors do not measure lies; they measure only signs of tension, such as increased sweating (measured by analyzing the conductivity of the skin) and increased heart rate. Brain scans measure increased brain activity, but the correlation between this and lying has still to be proven conclusively for a court of law.
It may take years of careful testing to explore the limits and accuracy of fMRI lie detection. In the meantime, the MacArthur Foundation recently gave a $10 million grant to the Law and Neuroscience Project to determine how neuroscience will affect the law.
MY fMRI BRAIN SCAN
I once had my own brain scanned by an fMRI machine. For a BBC/Discovery Channel documentary, I flew to Duke University, where they placed me on a stretcher, which was then inserted into a gigantic metal cylinder. When a huge, powerful magnet was turned on (20,000 times the earth’s magnetic field), the atoms in my brain were aligned to the magnetic field, like spinning tops whose axes point in one direction. Then a radio pulse was sent into my brain, which flipped some of the nuclei of my atoms upside down. When the nuclei eventually flipped back to normal, they emitted a tiny pulse, or “echo,” that could be detected by the fMRI machine. By analyzing these echoes, computers could process the signals, then reassemble a 3-D map of the interior of my brain.
The whole process was totally painless and harmless. The radiation sent into my body was non-ionizing and could not cause damage to my cells by ripping apart atoms. Even suspended in a magnetic field thousands of times stronger than the earth’s, I could not detect the slightest change in my body.
The purpose of my being in the fMRI scan was to determine precisely where in my brain certain thoughts were being manufactured. In particular, there is a tiny biological “clock” inside your brain, just between your eyes, behind your nose, where the brain calculates seconds and minutes. Damage to this delicate part of the brain causes a distorted sense of time.
While inside the scanner, I was asked to measure the passage of seconds and minutes. Later, when the fMRI pictures were developed, I could clearly see that there was a bright spot just behind my nose as I was counting the seconds. I realized that I was witnessing the birth of an entirely new area of biology: tracking down the precise locations in the brain associated with certain thoughts, a form of mind reading.
TRICORDERS AND PORTABLE BRAIN SCANS
In the future, the MRI machine need not be the monstrous device found in hospitals today, weighing several tons and taking up an entire room. It might be as small as a cell phone, or even a penny.
In 1993, Bernhard Blümich and his colleagues, when they were at the Max Planck Institute for Polymer Research in Mainz, Germany, hit upon a novel idea that could create tiny MRI machines. They built a new machine, called the MRI-MOUSE (mobile universal surface explorer), currently about one foot tall, that may one day give us MRI machines that are the size of a coffee cup and sold in department stores. This could revolutionize medicine, since one would be able to perform MRI scans in the privacy of one’s home. Blümich envisions a time, not too far away, when a person would be able to pass his personal MRI-MOUSE over his skin and look inside his body any time of the day. Computers would analyze the picture and diagnose any problems. “Perhaps something like the Star Trek tricorder is not so far off after all,” he has concluded.
(MRI scans work on a principle similar to compass needles. The north pole of the compass needle immediately aligns to the magnetic field. So when the body is placed in an MRI machine, the nuclei of the atoms, like compass needles, align to the magnetic field. Now a radio pulse is sent into the body which makes the nuclei flip upside down. Eventually, the nuclei flips back to its original position, emitting a second radio pulse or “echo.”)
The key to his mini-MRI machine is its nonuniform magnetic fields. Normally, the reason the MRI machine of today is so bulky is because you need to place the body in an extremely uniform magnetic field. The greater the uniformity of the field, the more detailed the resulting picture, which today can resolve features down to a tenth of a millimeter
. To obtain these uniform magnetic fields, physicists start with two large coils of wire, roughly two feet in diameter, stacked on top of each other. This is called a Helmholtz coil, and provides a uniform magnetic field in the space between the two coils. The human body is then placed along the axis of these two large magnets.
But if you use nonuniform magnetic fields, the resulting image is distorted and useless. This has been the problem with MRI machines for many decades. But Blümich stumbled on a clever way to compensate for this distortion by sending multiple radio pulses into the sample and then detecting the resulting echoes. Then computers are used to analyze these echoes and make up for the distortion created by nonuniform magnetic fields.
Today, Blümich’s portable MRI-MOUSE machine uses a small U-shaped magnet that produces a north pole and a south pole at each end of the U. This magnet is placed on top of the patient, and by moving the magnet, one can peer several inches beneath the skin. Unlike standard MRI machines, which consume vast amounts of power and have to have special electrical power outlets, the MRI-MOUSE uses only about as much electricity as an ordinary lightbulb.
In some of his early tests, Blümich placed the MRI-MOUSE on top of rubber tires, which are soft like human tissue. This could have an immediate commercial application: rapidly scanning for defects in products. Conventional MRI machines cannot be used on objects that contain metal, such as steel-belted radial tires. The MRI-MOUSE, because it uses only weak magnetic fields, has no such limitation. (The magnetic fields of a conventional MRI machine are 20,000 times more powerful than the earth’s magnetic field. Many nurses and technicians have been seriously hurt when the magnetic field is turned on and then metal tools suddenly come flying at them. The MRI-MOUSE has no such problem.)