Can We Solve the Mystery of Consciousness?

It was one of the great puzzles bedevilling late-19th-century scientists: how could it be that some neurons in the human skull produce the live-at-courtside-and-glad-to-be-here experience we call consciousness? “It is just as unaccountable, mused biologist and Darwin cornerman Thomas Huxley, “as the appearance of the genie when Aladdin rubbed his lamp.”

What’s wilder still is that it remains a mystery almost a century and a half later. In 1995, the philosopher David Chalmers coined a term that launched a thousand dissertations – the ‘hard problem of consciousness.’ The mechanics of how a brain works, Chalmers noted, is the easy stuff: if science doesn’t have all the answers yet, it at least knows the path to them. But some mystery in the machine is producing a result that’s more than the sum of its parts. Whatever that ingredient is that makes us know we’re alive – awash in lust and pride and the smell of popcorn and the urge to do right by our parents – that has eluded our ability to even study it in an organized way. We still don’t know what’s animating the genie.

But never has there been so much interest in getting to the bottom of it. As fascination with happiness, meditation and psychedelic therapy has soared, so has the public’s curiosity about consciousness itself. And now of course there’s AI. The prospect of machines that might one day rival – or even surpass – human intelligence has moved consciousness from the philosophy seminar into the public conversation. (Most researchers agree that AI is not conscious; the debate is over whether to throw in a “yet.”)

There has been an equal surge of interest in consciousness on the academic side. Neuroscience’s maddening inability to make much headway in the matter, despite having produced a London-cabbie map of the brain at this point, has left the door open to researchers from other domains. And so, psychologists, neuroscientists, mathematicians and computer scientists have rushed into the philosophers’ salon to hash this all out. Andrew Lee, an assistant professor of philosophy at U of T Scarborough, says there are dozens of competing theories of consciousness: “At the moment, it’s kind of the wild west.”

Where the field of consciousness studies is now, Lee says, is a bit like where physics was before Isaac Newton came striding into the picture and said, Guys, there are actually some math rules that explain the way things move, and they apply on Earth, in the heavens – everywhere. The universe is actually fairly … predictable.

“Newton’s theory pulled together basic physical phenomena and gave us a formal framework,” says Lee. “It gave us mathematical tools for modelling and making predictions about the physical world. Maybe someday we’ll reach a similar stage with consciousness research.”

Lee, who will never be accused of a lack of ambition, is working on a unified model of consciousness where conscious experiences can be described with mathematical precision. What makes consciousness so hard to explain, he says, is that no equation can fully capture what it feels like to smell cinnamon or experience pain. At present, the best neuroscience can do is hunt for correlations between the physical properties of the brain state – the pattern of neurons firing when cinnamon toast touches your tongue – and your subjective experience of “Yum!” But it’s loose and messy work. “In order to find those correlates, we need some understand ing of the structures of conscious experiences.” Lee believes there are basic units of consciousness that, if not yet understood, can at least be modelled.

Lee’s conviction that consciousness can be described objectively has roots in the work of two of his philosophical heroes, David Chalmers and Thomas Nagel – the latter of whom sat on his PhD committee. At the end of his famous paper on how humans will never be able to know what it feels like to be a bat, Nagel quite suddenly moved out of the fog and pounded a stake into the ground. He suggested we need a way of describing conscious experience that could be understood regardless of who – or what – was having it. The most natural Esperanto, Lee figures, is mathematics. “We can build on Nagel’s idea by developing mathematical descriptions of conscious experience,” Lee says. “We need to be sure we all understand and agree on what we’re talking about. I want to make the theoretical frameworks really secure and precise.”

Lee believes the study of consciousness has reached a turning point. He’s been drawing up an audacious initiative to capitalize on the buzz and energy surrounding the field. He’s proposing a kind of global superhive for advancing research on consciousness in the age of AI, with U of T at the centre of it. This fall, he is launching a three-year pilot project called the Artificial Consciousness Initiative. He would like to see it morph into a permanent institution – a Centre for Consciousness – doing cutting-edge interdisciplinary research involving top minds in cognitive science and related fields. But in these early days of a nascent discipline, it would actually be the philosophers doing the heavy lifting. “Every science starts out as philosophy – even physics started out as natural philosophy,” Lee says. “Once we figure out how to do it, it becomes its own field.”

“Whatever that ingredient is that makes us know we’re alive … has eluded our ability to even study it.”

Some of the biggest questions at the heart of this AI century concern the ‘moral status’ of AI systems, and the ethical and practical risks raised by increasingly capable AI – and these are in philosophy’s wheelhouse. Philosophy students who have had trouble selling the practicality of their degree might want to call their parents into the room at this point. Promising applications keep swimming into view.

For instance: in medicine. A little more than a decade ago, Adrian Owen, a neuroscientist at the University of Western Ontario, found that a patient who had been in a vegetative state so deep his brain showed no relevant neural activity could, upon waking, repeat back what had been said in the room. Clearly, a light was on that no one could see and no scientific instrument could detect. “Turns out it’s actually kind of important, in the ICU, to know that this person who’s in a vegetative state is actually conscious,” says William Seager, a U of T professor emeritus of philosophy. It’s one thing to debate consciousness in a philosophy seminar. It’s another to realize that someone’s life may depend on whether doctors can recognize it. This is why some philosophers don’t just say “consciousness matters.” They say, “consciousness makes things matter.”

Other questions won’t remain theoretical for long. Can we defend ourselves from AI’s silver-tongued proclamations of how much it loves us? And if a machine one day really does experience love – or pain, or anything at all – how will we know? After more than a century of wondering what consciousness is, we may soon need an answer. 

What if consciousness is everywhere?

Most theories assume consciousness emerged once brains became sufficiently complex. Panpsychism starts from the opposite premise: consciousness was here all along.

Some force is in play that we don’t yet understand, says William Seager, a UTSC professor emeritus of philosophy. He compares consciousness to gravity before Isaac Newton. Gravity shaped the universe long before anyone could explain it; perhaps consciousness is another fundamental property of reality that science has yet to understand.

This is the central idea of “panpsychism.” Human brains, in this view, didn’t create consciousness from nothing – they evolved the capacity to combine fundamental bits of consciousness into the complex form we experience. “The panpsychist does not say that your coffee mug is conscious,” Seager says, “but the panpsychist does say that the elementary parts of the universe have a little bit of consciousness in them.”

The concept is wacky enough to invite ridicule but nuanced enough to seem reasonable when a really smart person explains it. Panpsychism has attracted support from prominent thinkers, including philosopher Philip Goff, physicist Roger Penrose and neuroscientist Christof Koch – not because it has solved the mystery of consciousness, but because more conventional explanations haven’t solved it either.

Because consciousness can’t yet be measured directly, Seager argues that understanding it may require more than science alone – tools such as literature or psychology or philosophy or religion. “When your theory doesn’t explain the world, that should make good scientists think, ‘What have we missed?’” 

How would we know if AI became conscious?

If an AI insists it’s conscious, should we believe it?

That’s a much more difficult question than it sounds because scientists still don’t know how to define consciousness in humans, let alone in a machine. “There’s no clear answer – philosophy’s tough like that,” says Karina Vold, an associate professor at U of T’s Institute for the History and Philosophy of Science who specializes in the ethics of AI development. “All we have are proxy indicators, such as behaviours associated with consciousness.”

One example is playfulness. When crows have been observed picking up a wrapper, flying to a snowy roof and sliding down, scientists infer they’re doing it for no obvious survival purpose but for enjoyment.

Are we seeing similar behaviour from AI? “I’d be very cautious about concluding we are,” Vold says. If AI appears whimsical, autonomous or wise, we need to remember that it’s just remixing what it’s been fed – stories by us. “A better test would be giving it a clean data set with no human narratives to draw from. That’s not been done yet,” she notes.

Until scientists truly grasp what consciousness is, they can’t build a test for recognizing it in AI. Which means that if a machine ever does become conscious, knowing that it has may prove to be the most dif–ficult question of all.