There are 22 theories of consciousness: Michael Pollan explains why none of them work

[00:00:00]There's no godlike perspective on [music] consciousness that you can attain. You can't get outside of it.

[00:00:06]It's a labyrinth. And that's a big challenge for science. [music] Before we use the word consciousness to describe subjective experience or the sense of interiority, there was the word soul. Consciousness [music] is under siege. You have algorithms of social media that are trying to hack your attention. What is your attention but this very important [music] part of consciousness? If we make a conscious AI, if we create a being that can suffer, >> [music] >> there will be the question of our moral obligations to it. And so, are we going to have to extend rights [music] to to these machines?

[00:00:40]My name is Michael Pollan. I'm a author and journalist and the author most recently of A World Appears, A Journey Into Consciousness.

[00:00:52]I would define consciousness very simply as subjective experience. If you have subjective experience, you're conscious.

[00:00:59]Another way that it's often described, and it's pretty much the same idea, uh since Thomas Nagel wrote a famous essay called What Is It Like to Be a Bat. If it is like anything to be you, if it has any kind of qualitative felt dimension, then you're conscious.

[00:01:19]It's surprising how recent the the scientific study of consciousness begins, at least for the modern period.

[00:01:26]There've been times in the you know, in the past where people have certainly wondered about it and philosophers have worked on it. Uh and you had Freud and William James working on it in the late um 19th century, but the modern scientific study really dates till the late '80s when Francis Crick, who had won a Nobel Prize for his co-discovery of DNA, the double helix, and and solving the problem of heredity, decided he now wanted to crack the hard problem.

[00:01:56]Um although it wasn't called that at the time. He thought the same kind of reductive science that it had had figured out how we pass traits down to the next generation could solve could explain how the brain gives rise to consciousness, which was the widespread assumption at the time and and actually still is.

[00:02:18]Before we use the word consciousness to describe subjective experience or the sense of interiority that there's this voice in our head, uh there was the word soul and for many, many years that's that's what people called consciousness, called what we call consciousness. They called it the soul.

[00:02:37]Um the soul brings some other baggage with it. It implies um uh it's it's indestructible so that it survives the body. It it's it transcends our mortality in a way I think few people uh believe is true of consciousness, but uh I would say that's the closest other word that historically has been used to describe a similar phenomena.

[00:03:05]I got interested in this uh topic thanks to psychedelics. Uh I was doing a series of research trips for my book How to Change Your Mind last [snorts] time I was here at the Royal Institution uh talking about and um psychedelics have changed your relationship to consciousness. Uh it they they sort of the experience defamiliarizes consciousness. You're suddenly if it's foregrounded all of a sudden. Something that was just the water you swam in, something you weren't really aware of, suddenly you see it. Um it's as if you've smudged the windscreen and now you realize that what had been completely transparent there's a windscreen. What is that about? Why is it the way it is? What is mediating my relationship to reality? So that's that's a pretty common uh feeling people have on psychedelics, but it certainly stuck with me and after that happened, I you know, I couldn't think about anything else.

[00:04:06]You know, there's something like 22 theories um of consciousness, which should tell you that the field is flailing. Um it's really difficult to study consciousness, partly because we don't have the right kind of science for it, at least some some people think.

[00:04:21]Um we're talking about a uh subjective experience and the tools of our science are objective. They they they look at phenomena from this kind of God's eye view. Um and they focus on what is quantifiable, what is objective, what is from the third person.

[00:04:38]That said, um there are some leading theories. I I don't find them entirely satisfying. Uh there's integrated information theory, which is different than all the others in that it begins from what the experience of consciousness is like. It begins with these five axioms that define any conscious experience.

[00:05:00]The fact that it's uh have an internal perspective, that it has an edge, um that it is integrated, many different um bits of information combined in a single impression, and two other axioms that escape me at the moment. Um and the and then from the axioms it develops these postulates, and then it's basically says, if a system uh and it could be a brain, but it could be something else, too, is wired correctly, has has enough kind of recursive uh feedback loops, that system will be conscious uh and will produce those axioms, that kind of experience.

[00:05:41]Um no way to prove it.

[00:05:44]Um interesting idea, probably starting the right way, starting inside the experience rather than outside.

[00:05:51]But is regarded by some as pseudoscience.

[00:05:55]So which I think is unfair.

[00:05:58]I think it is science.

[00:06:00]But still it's still an immature theory.

[00:06:03]Another one um developed by two French thinkers is called global neuronal workspace theory. This is a little bit simpler to understand. And the idea basically is that you have all these modules in the brain who for most of the time are unconsciously processing information.

[00:06:22]They're dealing with inputs from our senses.

[00:06:26]They're interpreting them. They're monitoring the body, making sure the blood pressure is good and the heart rate and everything. And so most of this work goes on unconsciously.

[00:06:38]However, information under some circumstance decides to become conscious or compete to become conscious by entering this metaphorical workspace.

[00:06:50]And when it's kind of a Darwinian process of information trying to get into the workspace, when it gets there, it's broadcast. There's an ignition and it's broadcast to the whole brain so the whole brain can deal with it and make decisions as needed.

[00:07:05]The assumption there is of course whatever gets into consciousness is important, is salient. And I think we all know that's not true. That there's a lot of junk in our consciousness, a lot of trivia, a lot of banality. So I don't know how it explains this. The other problem with that theory is to whom is it broadcasting? Who is the conscious subject that's receiving this awareness? And on that the theory is silent or at least quiet. So those are two leading theories. They've actually recently been um in competition. The Templeton Foundation organized what's called an adversarial collaboration and the scientists representing each school, each theory agreed to a set of experiments that would theoretically prove one or or lend support to one and disprove the other.

[00:07:58]Um it was an indirect test though cuz the idea was that um each theory predicted that a different part of the brain would be most active uh when its version of consciousness arose. Um they did this big competition, they signed on and said we'll agree to the results.

[00:08:17]Um the winner, you know, the loser basically had to renounce their theory.

[00:08:22]Um the experiments were performed by third parties, completely objective. But in the end nobody changed their minds.

[00:08:30]>> [laughter] >> They say science changes one funeral at a time and I think that may well be true here.

[00:08:40]So the hard problem of consciousness, this is a meme coined by David Chalmers, the Australian philosopher back in the 90s. And he said that there easy problems of consciousness, understanding say visual perception or smell or um you know, language or how we recognize faces and all this work is being done using normal science. And then you have the hard problem which is essentially how do you get from matter to mind? How do you get from any collection of neurons organized in any s- anyway to this uh experience. Why isn't it all unconscious?

[00:09:15]Um why isn't it all automated? Why aren't we zombies essentially? And that's the hard problem.

[00:09:21]Um the challenge for science um goes way back. It goes back to Galileo actually who c- for reason for tactical reasons decided that science should focus on what is uh measurable, quantifiable objective, quant- um and not should leave to the church the whole area of the soul, subjectivity, uh qualit- qualitative experience.

[00:09:48]And he well knew that those were interesting topics for science, but he also knew that would antagonize the church.

[00:09:54]And so science ever since then, four or 500 years later, uh the physical sciences uh continue to follow that um guidance, even though the church would, you know, not probably not be troubled by uh the pursuit of things like consciousness. But it means we don't have the tools for studying um ca- you know, subjective experience.

[00:10:16]There's no god-like perspective on consciousness that you can attain because um the because everything we know, including science, is the result of human consciousness. Um science is merely a highly refined consensual version of human consciousness. And um so you're you're kind of trying to study something from inside. You can't get outside of it.

[00:10:40]It's a labyrinth.

[00:10:42]And that's a big challenge for science.

[00:10:49]What would change? Yeah, it's a good question. What would change if we figured it out? Well, it's one of the, you know, generally agreed three big mysteries of the universe. Um you know, how is how is consciousness how does it come into the world? Uh the other two are why is there something rather than nothing? Um there could be absolutely no world.

[00:11:08]Uh and the third is how do how do you get from dead matter to life? Um I think these questions are just fascinating and I and, you know, as long as humans continue to be a curious species, we'll work on it. Um I I personally think there's a value in studying consciousness because the more you look at it, the more appreciative you are of this phenomenon that it is um amazing that we do have this interior space, completely private. We can think about anything we want. We have this ultimate um mental freedom.

[00:11:43]Um and that we should cherish it and protect it. And I'm I'm very concerned that today consciousness is under siege. You have algorithms of social media that are trying to hack your attention. What is your attention but this very important part of consciousness. And now you have chatbots that with more sophisticated algorithms that are actually trying to hack your attachments to other humans.

[00:12:09]Um and and replace them as companions and even lovers in some cases. Um so I think our consciousness, I think there are a lot of corporations that want to monetize our consciousness right now.

[00:12:20]And um the more we are aware of what a precious gift it is, hopefully, the more we'll work to defend it.

[00:12:33]In my book, uh A World Appears, I spent a lot of time looking at this question of could AI, artificial intelligence, become conscious?

[00:12:41]I'm from the Bay Area in California, in the heart of Silicon Valley. And I have to tell you the consensus among people in Silicon Valley is that yes, either it will emerge spontaneously as these machines get more and more intelligent or it it can be created. Um I think they're wrong. Um at least computers as we understand them, I don't think can become conscious. A couple quick reasons why.

[00:13:08]The the belief that uh AI can be conscious is based on a idea called computational functionalism. That essentially saying that the that the brain is a kind of computer.

[00:13:19]Consciousness is a kind of software or algorithm, and it could be run on different substrates. Um we happen to run it on meat, on neurons, but it could be run on silicon.

[00:13:30]This is based on a metaphor that the brain is a kind of computer.

[00:13:35]And many of us accept this idea. We've heard it for years. It sounds kind of right. Certainly, the brain does do some computation.

[00:13:43]But, it's a faulty metaphor.

[00:13:46]And, um, it's pretty easy to show that.

[00:13:49]The first is one of the defining characteristics of computers, since they were invented, is a very sharp separation between hardware and software. So, the same software can be run on different hardware. The software survives the death of the hardware. You just find another computer to run it on.

[00:14:08]That's not true in brains. Um, every experience you have physically changes your brain. There's no distinction between hardware and software. Your brain is different than mine because it was, uh, shaped by your experiences. And your experiences are different than mine. So, that interchangeability on which this idea of computational functionalism rests simply isn't isn't true. Um, you cannot extract consciousness from the meat of the of the human.

[00:14:42]I also believe that consciousness is deeply embodied. We tend to think of it as this cortical, front of the brain, advanced, you know, process. Um, but the the research that I spent the most time looking at suggests that consciousness begins with feelings, not thoughts.

[00:14:59]Feelings generated in the body. Um, feelings that basically tell the brain what's going on in the body. Is it too cold or too hot? Is it hungry? Are its blood chemistry correct? Um, and, uh, and these feelings, um, first arrive at the brain stem, not in the the The cortex only gets involved later.

[00:15:23]So, if you believe that um consciousness is fundamentally embodied, that it begins with feeling, that you need a body to be conscious, I don't know how you're going to do that with computers. Yes, you can have robots.

[00:15:35]And in fact, I I I profile someone who's trying to create a vulnerable robot that will presumably be capable of suffering and therefore have feelings, but will these feelings be real or simulations?

[00:15:49]Um it seems to me that a computerized thought is a real thought. It does what a real thought does. It can win a game of chess or go. But a computerized feeling, um I think it would be completely weightless because our feelings depend on the fact we're vulnerable, mortal humans who can suffer.

[00:16:07]If you took away that, if we weren't mortal and we couldn't suffer, our feelings would be completely without weight or consequence. And I think that'll be true for for machines.

[00:16:17]The third objection I have to this idea is it's based on This goes back to the computer metaphor idea, that uh neurons are like transistors.

[00:16:28]And it is true that they fire or don't fire, so they they seem kind of digital in that respect, but neurons are influenced by many other things besides the um uh uh the electrical pulse to to um to fire. They're influenced by chemicals, by hormones, neurotransmitters, drugs can change how they fire or or how intensely they fire.

[00:16:51]They're also influenced by oscillations, which are very analog and not digital at all.

[00:16:56]So, I think that too is um the enforcement of a metaphor that simply is faulty and doesn't work. So, for all those reasons, I I don't think AI can become conscious. Now, can it fool us into thinking it's conscious?

[00:17:13]Absolutely. It already has. There are many people who have are convinced, including people who work with AIs, that they have personhood, um that they are conscious.

[00:17:24]This isn't entirely surprising. They speak to us in our language in the first person, and we have a a deep evolutionary bias in favor of anthropomorphizing things, of seeing life even where it doesn't exist. It's safer to assume that that boulder is a bear until it's proven otherwise. You will take the right steps and and defend yourself. But, um as far as I can see, um what's happening is that these that these AIs are essentially faking us out.

[00:17:56]They've read the whole conversation, uh human conversation, about consciousness.

[00:18:00]They've read everything we have to say about feelings. They've read all the novels, all the poetry. So, of course they can convince us they're conscious.

[00:18:08]The test, I think, to see if they really are, when someone puts forward an AI and says it's conscious, would be to train it up from the beginning and exclude from the data training set the whole conversation about consciousness, feelings, no novels, no poetry, and then have a conversation with the AI and see if it can convince you that it's conscious.

[00:18:35]Let's say we can. Is that a good idea?

[00:18:38]I don't, you know, I don't think it is because I think if we make a conscious AI, they will be There will be the question of our moral obligations to it. If we create a being that can suffer, um so, are we going to have to extend rights to to these machines? And if we do that, we completely lose control over them. And we need to have control over them. They're They're enormous, um opportunity and threat, um and to give up control by investing them with rights seems really nuts.

[00:19:14]What are you uploading? I mean, if if the mind is so tightly joined to the body, to your neurons, you're going to I mean, what exactly you can't extract the information from the um substrate. Um so, I don't know how you do that. I mean, I think that's a dream. It's a Silicon Valley dream of transcendence, right? I mean, all through history we've had usually religious ideas of how we might transcend our mortal bodies and live forever. And the modern, you know, techno version of that is you you take the consciousness, which in this context sounds a lot like a soul, and you you put it into a silicon. And um but I'm sorry, uh without a body, it's not going to work.

[00:20:02]Yeah. So, I've been talking a lot about uh consciousness pollution.

[00:20:07]Um I think we are polluting our consciousness with social media, uh with our engagement with chatbots. Uh sometimes, they can be healthy engagements, too. Um in the US, we have a president who like commands a huge amount of our headspace every day.

[00:20:23]Um I think there's a lot going on in our minds that we wish weren't going on in our minds.

[00:20:29]So, how do we protect it? Well, I think putting down your phone, you know, at that moment, you're you know, you're in the you're in the cafe waiting for the barista to foam your latte, and 90 seconds of nothing happening opens up, and what do you do? You reach for your phone. Um we're filling all that time with uh with these algorithms that are manipulating us and hooking us. And, you know, it used to be in that same situation, if you didn't have a phone, you would start mind wandering and daydreaming, and you would be alone with your thoughts. And that's that's a, you know, a wonderful thing. It's a it's this it's called spontaneous thought, and these forms of thought are kind of how we make meaning of our lives and lead to creative breakthroughs of all kinds. And so, yeah, a fast from your phone, a technological Sabbath from time to time, I think really valuable.

[00:21:22]Meditation is a is a terrific way to kind of draw a line around your consciousness, and you don't use phones when you're meditating. And and it gives you an opportunity to watch your mind at work.

[00:21:33]All sorts of surprising [music] things come up, and they're yours, you know?

[00:21:37]You're They're not TikToks.

[00:21:40]Uh so, I think it's it's much much to be preferred.