Do animals possess consciousness? Can consciousness be uploaded into a computer? Can we measure objectively whether someone is conscious or not?
Those may sound like deep, imponderable questions — but in a newly published book, “The Feeling of Life Itself,” neuroscientist Christof Koch actually lays out some answers: Yes, no … and yes, scientists are already testing a method for measuring consciousness, with eerie implications.
Along the way, Koch addresses brain-teasing concepts ranging from the Vulcan mind melds seen on “Star Trek” to the kind of brain-computer interface that billionaire Elon Musk is backing through his Neuralink venture.
Koch has studied the workings of the brain for decades as a Caltech researcher and now as chief scientist and president of Seattle’s Allen Institute for Brain Science. This is the third book he’s written about the consciousness conundrum — and you might think he’d stick to the view, popular among the technorati, that being conscious is just a matter of having sufficiently complex software.
“In particular, this means that computers, if they are sufficiently complex and if they begin to resemble our cognitive abilities and memories and speaking, then they too will become conscious,” Koch told GeekWire. “Consciousness is all about doing. Consciousness is just a special form of computation. It’s just a particular algorithm. So, consciousness is just a hack away.”
Contrary to expectations, Koch takes a radically different view: He says consciousness has to do with how the brain’s hardware is intrinsically structured and interconnected.
“It’s a measure of how complex systems are. It’s a measure of how usable a system is. It’s a measure of how much causal power the system has upon itself,” he said. “So, in principle, you could measure it.”
The scale of consciousness
In his book, Koch lays out a concept for consciousness known as integrated information theory, which was first proposed 15 years ago by a collaborator of his, Giulio Tononi of the Center for Sleep and Consciousness at the University of Wisconsin at Madison. The theory states that, by analyzing the structure of a system’s circuitry, it should be possible to derive a numerical measurement of the system’s cause-and-effect power, known as phi (that is, the Greek letter Φ).
“If phi is zero, then the system doesn’t exist for itself,” Koch said. “A system that has phi that’s positive, 10 or 20 or 5 million, exists for itself. The larger this number, the more conscious it is.”
Koch devotes much of the book to explaining the concept of phi — and experimental efforts to measure levels of consciousness in a wide variety of individuals using brain-wave monitors.
When a person’s conscious brain is zapped with a strong magnetic field, the response produces a complex sweep of brain waves that reverberate like the ringing of a bell. But the response of a non-conscious brain — say, the brain of someone who’s in dreamless deep sleep — is more like a cracked bell, without reverberation in different brain regions.
This “zap-and-zip” technique of using a strong magnetic field and electroencephalography is already being used in clinics as a “consciousness-meter,” Koch said, and the readings seem to correlate roughly with the apparent state of patients who are minimally conscious as opposed to being in a vegetative state.
One eerie aspect of the studies is that some patients who were thought to be in a vegetative state exhibit strongly reverberating brain activity. “If the theory is correct, it implies that these patients are conscious yet unable to communicate with the world,” Koch writes in his book.
And that’s not all: The theory suggests that consciousness is not a binary, on-or-off phenomenon. Looking beyond humans, animal brains show a lesser degree of cause-and-effect, but the effect is still there. Koch has come around to the view that all forms of life — from apes, dogs and dolphins all the way down to microbes — possess a modicum of consciousness.
This concept, known as panpsychism, has transformed Koch’s life. “I’ve turned into a complete vegetarian,” he said. “That is one of the implications [of the view] that consciousness is more widespread than we assume.”
Minds vs. machines
Another implication is that it would be impossible for computers to become conscious, at least if they’re merely advanced versions of current-day machines.
Thanks to deep learning and other tricks of the artificial intelligence trade, it’s possible to train a computer to pass an eighth-grade science exam, or even pass itself off as a human. But because of the linear way in which the circuitry is structured, Koch argues that even the most advanced AI systems lack cause-and-effect power.
“Yes, they can mimic the human brain, including mimicking the program in the brain for waking up and saying ‘I’m conscious,’ ” Koch said. “But it’s all going to be the ultimate deepfake. It will not feel like anything to be these machines.”
Koch admits there’s a catch: If future computers are modeled to reflect the highly complex, self-referential way in which neurons are connected in living brains, the question of machine consciousness could be revisited. As an example, Koch pointed to IBM’s True North project, which is developing neuromorphic computer chips.
All this implies that we won’t be uploading our brains into the cloud anytime soon. But how about merging our brains with machines? This concept is what motivated Elon Musk’s investment in Neuralink. The way Koch sees it, hooking up a computer to your brain would be analogous to having a search engine in your head.
“It’s cool what they’re doing, no question,” Koch said of Neuralink. “But we’re very far away from anything of the sort that Elon is talking about. And of course, unlike cars and rocket technology, just getting the FDA to approve the surgical devices is going to take many years.”
Merging different brains, as the fictional Mr. Spock was able to do on “Star Trek,” is a different matter. Studies of surgically split brains and conjoined twins suggest that brain-to-brain mergers are possible.
The effect seems likely to vary depending on the magnitude of the merger, Koch said. If integrated information theory is correct, there could come a point when the brain bridging is so complete that two minds merge into one.
“Your conscious experience of the world vanishes, as does mine,” he writes. “From your and my intrinsic perspective, we cease to exist. But our death coincides with the birth of a new amalgamated über-mind. It has a Whole extending across two brains and four cortical hemispheres.”
Will we all be assimilated? If so, you heard it here first.
Rise of the mini-brains
In another case of science fiction intersecting with real-life neuroscience, researchers reported last month that artificially grown bunches of human brain cells, known as cerebral organoids, exhibited electrical activity similar to the brain waves of newborn babies.
The findings are still tentative. “We don’t have any good evidence that it’s developed anything [that] you would confuse with the electrical activity in a conscious mouse or in a conscious dog,” Koch said. “But you know, sooner or later, in the next five or 10 or 15 years, we’re going to get there.”
He said the use of cerebral organoids in research is likely to raise deep questions.
“What it would feel depends on the exact wiring, but probably primarily it would have a feeling of a notion of spatial expanse. Probably it wouldn’t be pain or pleasure. It would feel like something … maybe very undifferentiated. Nothing like the highly sophisticated consciousness that you and I have, or that a 2-year-old has,” Koch said. “But, you know, this is the beginning of the road.”