In the pilot episode of 22nd Century, we propose an audacious concept: that someday our brains could be hooked up to the Internet the way our computers are today. Physically hooked up, that is, with numerous electrodes in the brain recording brain activity. Thousands or even millions of electrodes, in fact, threaded into the brain by way of the capillary system. With such a system, you could monitor brain activity with such a high degree of resolution that with the appropriate software for analyzing the data, it might become possible to know what a person is thinking or feeling based on their neural activity. And conversely, to “force” the brain to think/feel particular things by sending the appropriate electrical triggers to the neurons.
What challenges would have to be solved to make such a thing possible?
We already know that wires inserted into the brain through the skull can read certain kinds of information. A research team at the University of Pittsburgh has inserted wires into the parts of monkeys’ brains that control the motor system – that is, the limbs. By reading the way the neurons fire in that area of the brain, they can make a robot arm move as if it was the monkey’s own. With training, lab monkeys have gotten good enough to reach out with the robot arm, grasp a piece of food, and bring it to their mouths. (See this article on their research here.)
To make a World Wide Mind feasible, we’d have to learn how to thread millions of wires into the brain through the capillaries. At present, we have no technology that can guide large numbers of wires into an equally large number of convoluted capillaries. While they’d be small – the diameter of a red blood cell – we presently have no way of ensuring that they won't get tangled, or cause blood clots. And there are subtler problems as well. Because of the physics of how metal conducts current in a wet environment, a wire that is thin enough to fit into a capillary without blocking it will have a very limited data carrying capacity. You can insulate it, but then you have a much thicker wire.
Let's assume for the moment that that physical obstacles can be solved. There might, for example, be some nanotechnological system that could do it. But the computational requirements of such a system are equally formidable. Experiments done on epileptics fitted with just 60 implanted electrodes have yielded terabytes of data. A terabyte is a thousand gigabytes, which is roughly 10 times the capacity of an average hard disk today. Imagine writing software that would have to comb through all of the data on every computer in your city, over and over again, looking for patterns.
The challenges go well beyond the wiring and computing. Foremost among them is that we presently know little to nothing about how the brain enacts a thought or feeling in terms of neural activity. One widely held theory of cognition holds that thought is distributed throughout the brain in such a way that any individual neuron can be participating in multiple events simultaneously. (A single neuron can receive inputs from tens of thousands of other neurons, and send its output to a neuron that may be inches away, an enormous distance on a cellular scale.) According to that theory, it’s the pattern of neural activity that matters most. If that's the case, we can try to discover the meaning of specific patterns of neural activity by brute-force pattern matching, e.g. by inviting the user to have a specific thought or feeling several times and trying to discern a consistent pattern of neural activity out of the flood of data. This is what Phillip Kennedy does in training his computers to identify which of 26 letters a patient is mentally “speaking aloud.”
But words are much more complex than individual letters, and feelings are more complex still. There is almost certainly no single area of the brain that is individually responsible for a feeling. And the neural representation of a feeling may be enormously difficult to identify. You don't just feel sad, you feel sad about something, and that feeling is impossible to separate from your reactions to the environment around you. Many parts of the brain may be involved in generating it, and those parts will be working on other things at the same time – say, looking for a tissue to cry into.
On the other hand, it's known that some neurons will fire only in response to specific events such as seeing Jennifer Aniston's face. That lends support to the rather peculiar idea that individual neurons are responsible for specific concepts. If that was actually true, it would be impossible to neurally convey the idea of Jennifer Aniston directly from one brain to another if the receiving brain had no “Jennifer Aniston” neuron. (See this article here.)
Of course, the brain is very likely both a distributed and centralized system. But even if the neural code is fully understood someday, it will be a huge challenge to communicate an image/sentence/emotion by evoking a similar state in another brain. We know that all human brains have the same organs in the same places -- everyone’s got an amygdala, everyone’s got a hippocampus -- but that mapping may or may not extend down to a neural level. The developing brain does considerable pruning of its neural structures at an early age, and that pruning is shaped by life events. A pattern that corresponds to a certain kind of anger in one brain may be nothing but gibberish in another. It might be necessary to have a computer “translate” the output of one brain into a neural coding that would evoke an analogous state in another brain. That's a huge challenge, too. Even now, computers are not very good at translating one language into another, and that's a much simpler task by comparison.
Perhaps the deepest challenge of all is answering the question of what one would communicate if the World Wide Mind concept was possible. What would it allow us to do that we cannot do now with speech, facial expressions, and touch? We know that we can infer an extraordinary amount of information about another person with the senses that we have had for millions of years. Rodolfo Llinas suggests that it would allow us actually to feel what another person is feeling, sense what another person is sensing. But even then, we would still have to interpret such information through our own brain’s beliefs, preconceptions, and ideas. What would we actually gain? It’s very likely that it would allow for entirely new kinds of communication rather than simply reproducing old ones.
So, the challenges can be summed up as these:
- Physical insertion of the electrodes
- Managing enormous amounts of data in real time
- Cracking the neural code of mental activity
- Translating the output of one brain into equivalent input for another brain
- Developing modes of communication with the technology that are superior, or at any rate usefully different, to the ones we have now.
For these reasons, some scientists are skeptical that the World Wide Mind idea is possible at all. Gerald Loeb, a professor of biomedical engineering at the University of Southern California, said in an interview,
For me, the interesting reasons to think about a World Wide Mind are not the technological challenges in achieving it (daunting as they are), but rather the neurophysiological, linguistic and philosophical concepts that would have to be true for it to be possible at all. From my perspective, these are actually null hypotheses that modern neuroscience has been systematically eliminating. I think "information theory" is getting further from a World Wide Mind even as the technology for "data transmission" seems to make it conceivable.
Answering skeptics such as Dr. Loeb will require numerous breakthroughs in biology, engineering, and cognitive science. But as we all know, equivalent breakthroughs have enabled technologies that our ancestors could not even have imagined. We’re like Alexander Graham Bell, the inventor of the telephone, trying to imagine television in the 1870s and predicting how it would change society. He really wouldn’t have been able to do it.
Take the cochlear implant that I myself have. I'm a totally deaf man who can use a cellphone. Many eminent scientists and engineers considered the concept of cochlear implant ridiculous in the 1970s. They objected that it wouldn’t be possible to get electrodes into the inner ear, that no computers existed that could do the processing in real time, and that no one understood the neural coding of the ear. Do those objections sound just a little bit familiar? They were solved with decades of patient research and engineering. Perhaps the World Wide Mind idea will take a similar trajectory.
Michael Chorost is a science writer living in San Francisco. His first book, Rebuilt: How Becoming Part Computer Made Me More Human, was published in 2005, and went on to win the 2006 PEN/USA Book Award for Creative Nonfiction. More on the book and the author can be found at http://www.michaelchorost.com.