Within Reason - #34 Anil Seth - A New Science of Consciousness

Episode Date: June 7, 2023

Anil Seth is a professor of Cognitive and Computational Neuroscience at the University of Sussex. He is the author of "Being You: A New Science of Consciousness". Purchase "Being You: A New Science of Consciousness": https://amzn.to/3NgKD53 Anil Seth's website: https://www.anilseth.com/ The Perception Census: https://perceptioncensus.dreamachine.world/ Learn more about your ad choices. Visit megaphone.fm/adchoices

Welcome to Within Reason. My name is Alex O'Connor. My guest today is Dr. Annal Seth, who is a professor of, now can I get this right, computational and cognitive neuroscience here at the University of Sussex. Yeah, it's the other way around. other way around. It's basically the same thing. Thank you so much for being here. It's a pleasure. Also the author of Being You, a new science of consciousness. Why do we need a new science of consciousness? I think there's really a new approach to the science of consciousness. And one of the things about consciousness is that for a long time, it was considered relatively illegitimate and disreputable in the mind and brain sciences. Now, that's all changed in the last 30 years or so. And there's many different approaches to consciousness now that are really flourishing within neuroscience, within

psychology, within cognitive science. And it's a very exciting time because there's lots of new evidence, but there's also lots of different positions, different theories. And my own view is one particular theory. So it's not really a completely new science of consciousness. There's a new scientific approach to this age-old question. I'm going to have to begin by asking some of the boring questions, including what is consciousness? Not that that's a boring question, but it must be boring for someone like you. It's always a tricky question. I mean, I've been asked it so many times

and I always stumble a little bit answering it because it's one of those things that intuitively we all know what it is. It's what we lose when we fall into a dreamless sleep or lose even more profoundly if we go under general anesthesia, which is an amazing non-experience to have general anesthesia. And it's what comes back

when we come around again or wake up in the morning. And when we open our eyes, It's not just that our brains are doing some kind of complex information processing. There's an experience happening. There's a redness to the colors that we see. When we feel hunger, it feels like something to be hungry. It's not just a state my brain is in. So consciousness intuitively is what makes us more than merely complicated biological objects, makes us subjects.

It is what it's like to be a thing. That's the definition I use a little bit in the book. It's a definition from the philosopher Thomas Nagel. And he says it like this. He says, for a conscious organism, there is something it is like to be that organism. And you're right. I think the intuition he's driving at is that it feels like something to be me. It feels like something to be you. And in Thomas Nagel's famous paper, it feels like something to be a bat. The only a bat will ever know what that feeling is really like. But for other things, things like this table, this cup of tea, my phone, it doesn't feel like anything to be these things. They're objects. However complicated. So you say, I mean, I spoke to Philip Goff recently, who would probably beg to differ. I wondered sort of along that line, actually, you said that consciousness is something like they're being more than just computational processes. Is it also something more than the material? Because for many people, one of the most interesting things about consciousness is that it seems to intuitively point to something non-material existing within

the universe. It's such an appealing intuition. I think it's one we all have. I'm interested actually over history whether it was as seductive and intuition as it seems to us today. But certainly in Western society over recent history, if we think about conscious, as thinkers have done from Descartes and going back and going forwards, it seems as though consciousness is not the kind of thing that a mere material system could have or could generate. It seems a different realm of existence altogether. Descartes divided the universe into matter stuff, res extensor, the stuff that tables and chairs and brains and bodies are made of,

and res cogitans, the stuff of thought and the stuff of consciousness. I can't deny that it seems as though there's something additional to the material happening, But I worry how reliable our intuitions are about this. They failed us in the past in these kinds of questions. People at one point thought that life could not be explained in terms of physics and chemistry. It seemed as though we needed some supernatural explanation, some el-en-bital, some divine sparkle or spark of life. The fact that it doesn't seem to be a property of the material doesn't mean that it isn't. It might mean that.

And what we do know, the sort of fundamental starting point, of neuroscience is that there's this intimate connection between what happens in the brain and what happens in conscious experience. So my overall strategy is to basically recognize that and let's try and learn as much as we can about the nature of consciousness through the study of the brain and the body and the world. And maybe at the end of that, there will be some little residue of mystery left. It's possible. But I think that remains to be seen. Well, the brain is the center of consciousness. It seems to be what generates consciousness and all consciousness is correlated at least as far as we know with some kind of brain activity. But the strange thing is

that if consciousness is located in the brain or conscious experience, then like you say, there are red experiences, green experiences, hunger, these kinds of things. But the mystery is that if I were to cut open your brain, crudely speaking, I'm not going to find red in your brain. I'm not going to actually find redness or blueness or greenness. And so if consciousness is based in the brain or generated by the brain, but all of the material matter in the brain, you can search through and through and you won't find redness. Does this not imply that redness itself is non-material

because it's not anywhere to be found in the material brain? And yet we know that it exists because we can see it right now. I think what it points to is that something like color has two different aspects. It has a subjective and experiential. side of it, the feeling of experiencing color, and there's an implementation side of it, what's actually happening in the brain. It doesn't really make sense to think that you would find red if you open the brain up and looked for it because that would require another conscious observer to look at the color in the brain and then you just push the problem

one stage further back. So how is it that that color is then experienced as a color by whoever's looking inside the brain? So that's the problem. So that's the problem. So that's the I don't think that's such a problem, but the example of color is super interesting because it does, I think it highlights that color is a kind of thing, call it a thing or call it a process or you can call it quality, that requires both a brain and a world to happen. Actually, we can dream colors. So it doesn't always require a world. But color in our normal everyday waking life is not out there in the world either. Out there in the world, there's just electromagnetic radiation of different wavelengths. And these wavelengths, you often call them like red, green, and blue, but they aren't actually red, green, and blue. They're just different wavelengths of energy. And it's combinations of those wavelengths that our brain uses to create experiences

of color. And so when we experience something like color, we're experiencing simultaneously less than what's really out there because our eyes are only sensitive to these three wavelengths. and more than what's out there because we experience a whole palette of millions of colors rather than just three. So there's this very indirect relationship. And color, in the words of the artist Suzanne, color is the place where the brain and the universe meet. It's not in either one by itself. So if colors are essentially, well, would you say located in the brain, if they're not located sort of in the objects themselves.

Like you say, you've just sort of got wavelengths of light, and it's really the brain that produces the color experience. If that's the case, then, well, we know that there's an entire expanse of wavelengths outside of our perceptive abilities. So ultraviolet light, infrared light. These are the same kind of things. They're just electromagnetic waves, but they're not waves that our eyes can detect. But if we take our eyes out of the picture, if it's also located in the brain,

do you think in principle there would be a way to stimulate a person's brain through artificial means that would allow them to see something like a new color? I think we already do that, to some extent. I mean, they're already night vision goggles, right, that take, say, infrared and render it in something that we can see. Now, you could take that one stage further, and instead of changing the frequencies of the energy into the visible spectrum, so-called visible spectrum, you could maybe plug these signals directly into the brain and perhaps generate a new kind of experience.

This, I think, is in principle possible, but of course our brains didn't evolve to deal with signals other than the ones that we typically experience. It doesn't mean that we wouldn't experience something, and it might be very interesting to know what it is. There's a whole history of studies in what's called sensory substitution, where we take information from, say, of one kind, let's say, touch information and it can be fed

into the visual cortex or into the eyes in some way. Or even echolocation. So back to Thomas Nagel and what it's like to be about. You can actually get some of the way there by taking sonar and rendering it either directly into the brain or into some other sensation, maybe the pinpricks of tactile sensation on your skin, to give you the same. same information that echo location would give. And it's fascinating to ask what people experience when you do this. And it starts off often with them experiencing something like the interface. So a typical example would be translating vision into touch. It's one of the oldest interfaces like this. Initially people just feel the touch. But then the idea is some of the

evidence is that as people get used to this, they don't notice the touch so much. and start being able to directly have an impression of a visual environment, but now it's coming through a different way. So what do you think is sort of going on when we perceive objects, colors, and have sensations, is my sort of perception of this cup, the cup that I can see, caused by this object, the cup, as it actually is, or is there something a bit more complicated going on? I think it's a little bit more complicated.

I think we have this impression as we walk around the world that the world is just there and it just pours itself into our minds directly through the windows of our eyes and our ears and our other senses, that we see things and hear things and smell things as they are. And that's a very useful way that evolution is designed our perceptual systems. It would be very weird if we walked around the world experiencing things as a brain-based construction. But that's in fact what's really going on, I think.

We don't, and we can't experience things exactly as they are. This point goes back to Emmanuel Kant and even earlier. Now, there is an objective reality. There are things, there are objects like your cup, like my cup here. They exist, and they have properties like color and solidity and shape and so on. but the way we experience them is never as they really are it's always an interpretation always a construction and quite how that interpretation that construction is built

I mean that's what our kind of day-to-day research is all about but the basic idea is the idea that instead of the brain reading out sensory signals and say oh I just brain is sitting the self is sort of sitting in here somewhere and reading out all this information and saying oh there's a cup and it's got some red markings on it and so on. The brain is always making predictions about what's out there in the world or in here in the body and using the sensory signals to update these predictions,

to keep them calibrated, tied to the world in ways that are useful for our behavior. And this is a big flip then, because even though it seems as though we read the world from the outside in, what I think is really happening, and not just me, is that perception is as much a creative act of writing as it is of reading. We're actively generating our perceptions. And the sensory data is there mainly to keep these perceptual predictions tuned to the world in useful ways. So we've got a kind of balance going on. We've got inputs, but also our brain producing outputs. And it's somewhere in the meeting of these two that we have conscious experience. That's right. And that's the big open question. It's like where in this this continual dance between sensory input and perceptual

prediction, where in this dance is consciousness happening and what elements of it are most important for shaping what we experience? And the traditional, you know, if you open a neuroscience textbook, at least the ones I was reading when I was a student, the heavy lifting is all done in this bottom up outside in direction. And whatever's coming from the brain back out to the world, maybe doing a little bit of tuning and finessing and modulating, but really it's an outside in. And I think that's entirely wrong. I think the flip side is the way to think about it.

What's really important is this top-down, inside-out flow of signals that carries the brain's predictions about the world. And I think that's what we experience, the brain's predictions. I think outside of the setting of neuroscience or philosophy, it can sound a bit for want of a better word, sort of hippie-ish, to say, well, the conscious experience we have, my experience of the cup is really sort of a projection of my own mind, or at least in part, it's my brain sort of putting itself out into the world. It implies, firstly, that a lot of what we're seeing is not, as you say, like it really is,

that there's some kind of hallucination involved, although not a sort of full or complete hallucination, because there seems to really be an object that you say it's interacting backwards, I think people might be skeptical of this view, for example, thinking about a newborn child, sort of opening their eyes for the first time and just being presented with, you know, blue or being presented with a cup. And as far as we know, I don't remember, I don't know if you do, but we'll just see a cup. And it seems that in the case of a newborn infant, they're not going to have much in the way of sort of predictions to make about how the world works, unless you're suggesting that these predictive qualities of the mind

are somehow sort of hardwired evolutionarily into the depths of our brain. Is that more like what's going on? I think that's likely to be a big part of the story. It's actually very hard to know what a baby, a newborn baby, really experiences. William James, one of the founders of psychology famously speculated that a newborn baby would experience this kind of blooming and buzzing confusion, perhaps without even distinguishing the different senses. Like it's pretty obvious to us what's visual and what's auditory and what's olfactory, what's smell. But maybe we learn to do that a little bit. I mean, I wonder if that's got something to do with the fact that people tend to have memories up to a certain point in their childhood. It's like people might

be able to remember being five years old, possibly slightly younger. I'm not really sure. But there's a point of which we just don't remember. And maybe that's got something to do with the fact that the information that was coming in at the time was just a big sort of hot mess of data that we couldn't make sense of. It might be a big hot mess. It might also be, and I think also part of that story is that, you know, it takes the self quite a long time to get online. It's not that the self is suddenly there with all its ability to remember and plan aspects of self that develop over time. And memory is something that it's, again, it's not like a computer. We often get really misled by this idea of the brain as a computer. And when we think of memory in a computer,

you store it and it's just always there and it's kind of always useful to have it there. But memory in a biological organism and a human being is not supposed to be this indelible, totally accurate record of everything that happened. It serves the purpose, like everything in our biology and our psychology. It serves the purpose of behavior. And it's kind of hard to see an adaptive reason why we should remember. remember things at that early in our lives because we didn't really have control over anything that early in our lives. So memory is very different. It's very driven by the future as much as it's a record of the past. Yeah, and seemingly these things are driven by survival. And this makes

people, a lot of people quite skeptical in philosophy of our ability to understand truths about the universe because of course the very sort of mechanism we use to understand the world is biased not towards truth but towards survival. Is that as skepticism you share given your sort of neurological research? It is. Well, it's sort of. I mean, I actually see that as a really essential insight into understanding perception. I think rather than it being a problem, when it's turned back on itself and you recognize the survival imperative of our brains, then that can actually, instead of obscuring our understanding, it can enhance it when we're trying to understand the brain itself. So this comes back to what you were saying about, is it a little hippie-ish to talk

about perception in this inside-out, active, creative way? Maybe, but I think it's also very well grounded. I use the term not, well, use the term controlled hallucination quite often to talk about perception. But the control is just as important as the hallucination. The hallucination part emphasizes that what we experience comes from the inside rather than just being directly read out. But the control emphasizes that it's not arbitrary. The world exists. Our minds aren't just making stuff up. There is a cup here and it has solidity. But the way we encounter it is a kind of active construction. But it's controlled by the world in ways that are useful for our survival. And that can help us understand the indirect nature of one.

we experience the world the way we do? So it's not that we're just sort of imagining everything. There is a real external world. And I mean, I don't know what you would say to skeptics on that front who say, all I sort of have is this potentially flawed system of senses that I used to try and investigate the world. And I have no way to know if it's really there. And especially when you say that so much of what we perceive is to do with our brains,

what they predict, what they expect to find. And I mean, you say, the book that people might object to a sort of hallucination view of perception by saying, well, if you think it's all a hallucination, go and jump in front of a train and see what happens, right? And of course, yeah, I mean, do that and you're going to feel a big crash. No, I should clarify, actually. I did say do that. I don't want to be clipped. But of course, the feeling that you would have being hit by a train is something that equally could be put down to something that your brain is sort of creating or inventing. So I don't know if you have any reason to so confidently

assert that actually, no, there is a real sort of set of objects that are in part causing our perceptions. I think that's just a metaphysical assumption that I make. I don't think you can demonstrate that, right? I mean, it could be that it's true. You can't rule out the possibility that we are all just brains in a vat and our brains are being stimulated in particular way so that we have the experiences that we have. And there's a very different world out there or maybe even nothing at all out there. It's just not a very productive way to look at things. You know, the way I feel I'm very pragmatic or try to be quite pragmatic as a sort of as a scientist and especially this interface with philosophy, the assumption that things indeed exist, I think is a very pragmatically useful assumption to make. It helps us under, you know, it gets us quite far in understanding the world and our relation to it. If we were thinking, if we were making the starting assumption that indeed everything is made up in our heads, there's no such thing. as objective reality, then it becomes quite challenging to explain how and why we have compatible

perceptual experiences of things if there's nothing actually real out there. Okay, so there is an external world and there are objects and they sort of impress themselves upon our minds. And even though there's no real agency involved here, I open my eyes, there's a microphone in front of me. I don't choose to see this mic. I don't think to myself, I would love for there to me, be a microphone here. It'd be really convenient for this podcast, and thus there it is. It just sort of appears in front of me. And yet you suggest that at least part of the experience I'm having when I see this microphone is actually my brain making a prediction and mapping something onto the world. What reasons do we have to believe that this is happening when we perceive the world

around us? All sorts of reasons. And I think this is one of the appealing things about it. There's ancestors this idea that you can find in philosophy, in psychology, even in machine learning about how this problem of perception might get solved. And just by the way, the fact that we don't experience it as a voluntary act, it's not that I decide to see something. It's actually some interesting subtleties there that in hypnosis, for example, we can, people can be suggested to have visual experiences purely through suggestion, which means that they're sort of engaging volitional acts without even realizing that's what they're doing. And they can, indeed have these experiences.

But you're right, in the normal way of encountering the world, it just seems to happen to us. We don't actively bring it about like we do when we close our eyes and imagine seeing something in our mind's eye. So why should perception work this way? The first reason to think about things in this sense is that is a very sort of just like,

let's have a look at what the problem involves and figure out how we might solve it in principle. here's the thing the world out there the real world is ambiguous it's unlabeled we if we try and like take from take a physics eye perspective on what's what's actually happening in this space here where there are two bodies two brains and some air and some microphones there's all kinds of things there's like pressure waves in the air there's electromagnetic radiation there's underneath that there's this kind of quantum super who knows what what is going on um And the brain is faced, which is also made of these things, by the way.

You know, it's also made of the same kind of fundamental particles that everything else is made of. It's faced with this problem, figuring out what's out there in the world. And it's a situation that involves inherently a lot of uncertainty. And so this already suggests what the brain has to do is make a best guess. It can't unambiguously determine what's in the world. It has to make a best guess about it on the basis. of prior expectations or beliefs, some of which might be learned, some of which might be evolutionarily pre-trained, if you like, and the sensory data that it gets. So there's this

concept of the Bayesian brains, old concept of Bayesian reasoning, which is just how a system should deal with uncertainty in its sort of an optimal way. And an optimal way to do this is to make predictions and update these predictions with the data and try and minimize the difference between what the system is predicting and what's actually there. That's a way of dealing with uncertainty. Yeah, you refer to the Bayesian brain throughout the book. And of course, Bayesian reasoning, like you say, is inference to the best explanation. We say, and you can compare it as a form of abductive reasoning. You can compare it to deductive reasoning, which is logically taught a logical. It might be, you know, famously all men are mortal. Socrates is a man. Therefore,

Socrates is mortal. This logically follows. Inductive reasoning being like, I pick out a green ball and a green ball and a green ball and a green ball. So I suspect, so I'm going to reason that the next ball is going to be green, which doesn't necessarily follow, but that's inductive reasoning. Whereas abductive reasoning and sort of Bayesian reasoning here is given some set of data, what's the best explanation that we have? And you think that the, brain is working along these lines. It's sort of taking in this experience, this data and saying, well, what's the best explanation that I have of the various things that I'm experiencing right now and produces something like the cup in front of me? Yeah, I think in many cases, yes. I'm not saying

in all cases, but I think in most cases, I think that's the way it has to work. And it is this, another good example of Bayesian reasoning, just to put it in context, is let's say something like a medical diagnosis. So, you know, you take a test for some rare disease. and it comes through positive. So what's the chance that you actually have the disease? Well, it depends on how accurate that test is and on the prevalence of that disease. So even a positive test,

you might still conclude that you don't have the disease if it's very rare and the test is unreliable. So it's a best guess. And even if the test is quite reliable, if the disease is rare enough, it might still be not enough reason to, on a sort of abductive understanding to reason that you actually have the disease,

which is quite an interesting implication of abductive reasoning. I'm wondering what might be some sort of empirical data that we might have for this view that our predictions about what we expect to find in the world and sort of pre-existing networks in the brain determine and affect what we perceive. Because most people think, I would imagine, that when I look at something, I just sort of see it.

And I'm not doing it. anything to produce this image, rather the image is sort of being produced in my mind by the object. What evidence do we have that that's not what's going on? Heaps of evidence, which is the other main strand of evidence that we were talking about. On the one hand, you've got the in principle stuff, right, that it sort of has to be or ought to be this abduction, this inference to the best explanation. But then if indeed we do open up the brain and look inside, yeah, we don't find colors, but what we do find very suggestively is that, for instance, in vision, there are far more connections going from the inside of the brain back out towards the eyes than there

are coming in the other direction. Is that right? That's right. And that was always a puzzle for neuroanatomist, neuroscientists to explain. Why should it be like this? And it begins to make sense if you start to think, well, actually, it's this process of generating predictions. That's the really important thing. And the sensory information is just carrying the error, the prediction error between what the brain expects and what it gets. So anatomically, we've got some suggestive evidence. And then there's a ton of fascinating experimental evidence. And this is the kind of thing that I'm interested in doing in my group at Sussex, at least in part, but plenty of other groups are doing it too, where we can show that the brain's expectations really do shape what we

experience. We did an experiment a few years ago now, just priming people to expect to see either, let's say, a house or a face. And the images would be presented in this way so that we're very ambiguous. And what we found was that if the sensory data matched people's expectations, then people consciously saw the image more accurately and more quickly than when the image, the sensory data, didn't match their expectations. So here this close match between the brain's expectations and the data, it brought something more rapidly into our conscious scene, into our conscious experience. And that's, of course, what you'd expect if expectations are deeply involved. Perhaps. I mean, I'm imagining, sort of, if you expect to see a

face, you might see the face more quickly. But in the same way that if you told me to sort of react to a sound you were about to play, and I was sort of ready for. And I was sort of ready for it. I think my reaction time would probably be quicker. It would seem as though I'd noticed the sound quicker because my reflex would be faster than if it just happened randomly and I reacted naturally. But I don't think that would suggest that my expectations of what the sound's going to be would in any way affect my actual sense experience of what the sound is like. You know, the bell would sound the same to me whether I was expecting to hear it or not. The fact that I see it more quickly or I'm listening and so hear it in more detail because I'm focusing, you know, something

like that. I don't know if that tells me that my expectations have anything to do with the experience. So you should be a psychologist because you're raising all the kinds of objections that people rightly raised to these kinds of data, which we had to control for. So yes, I mean, you have to rule out that these effects are not simply priming people to respond to something or having them pay attention to something more than they would otherwise. How do you do that? I mean, so you can, you have to, I don't think any one experiment can, can, can, do it all. So in our experiment, we did a bunch of controls, like trying to change the mapping, the hand that you would use to try and control for this idea of response priming so that you

weren't just queued more quickly. And we also had people pay attention to different things so that we could, to some extent, control for attention. You can do it as, you know, you can make a good college try at doing it. But I think the real power is that this kind of vision is not rely on any single experiments. There are other experiments. that also converge on this view. Some of my favorite come from a friend and colleague of mine in Glasgow, Lars Mukley, who uses brain imaging to look actually inside the brain in experiences, paradigms, experiments like this.

And what he found, and I think this is really strongly indicative of the predictive brain idea, is he had people look at different images while they were in the scanner and there were images of different kinds of scenes. They would be like an inside scene and a scene in a city and a scene in a countryside, things like this.

But in each image, a quarter of it was missing. It was just blank. And, of course, people could still recognize what's going on because three quarters is still there. And then what he did was he asked

whether he could decode from the brain imaging data what the scene was, but only using imaging data from part of the visual cortex, the back of the brain here, that was not being stimulated by the image, that was like responding to the blank part of the image. And he found that he could.

And what's more, he found that he could only do this by using the part of the visual cortex that was not receiving input, but that was receiving top-down, inside-out signals from deeper in the brain. So the brain was passing back to, visual cortex, enough information to decode what the person was seeing. So there's definitely information about the content of our experience going right back out to visual cortex. And you can

see this now in the flow of signals in the brain. There's also just some quite straightforward intuitive, almost optical illusions that you talk about in the book that suggests that our perceptions of things can be affected by more than just what we're seeing, but also what our brain expects to find, right? There's all sorts. So your optical visual illusions, they're perhaps the easiest way to show that what we experience is really not a direct reflection of what's there. And these can range from very simple things like our brain encodes expectations that things under, in shadow, appear darker than they are. And so we can show that that really changes how we experience shading and color. And we take the context away and suddenly these, the colors all look

different. Yeah, that wonderful famous image which for viewers I can put on screen, for listeners we'll have to do with the explanation. It's sort of a generated image of a sort of a cylinder producing a shadow on a checkerboard and the light

square in the shadow and the dark square outside of the shadow look like different colours. But in fact they are in the image the exact same shade but because our brain sees that the one is under a shadow, or at least it appears to be in the image, it's the illusion, our brain just makes it

look darker. And it's, you describe this kind of illusion as, what's the word you use, like cognitively impenetrable or something? Yeah, that's right. Even when you know what's going on, you can't help. You can't unsee it. You can't unsee it. So it's hardwired into, into our perceptual system. And it's not, you know, it's not a failure of our perceptual system. You know, our eyes and brains were not intended to be light meters. They're intended to figure out most likely what's actually going on. And so given that particular sensory information, what's, you know, we see the situation kind of accurately.

Yes, there's a cylinder and a checkerboard. And that's the situation. So it kind of doesn't matter given our overall perceptual take that there's this funny effect with colours going on. Yeah, I mean, in a sense it sort of is darker because within the sort of within the fiction of the image, well, it is a darker square, you know. But it's, I mean...

But it isn't. It's not actually a different colour, but this is part of the trouble. I mean, you say this isn't a failure, which is interesting because that implies that the purpose of our sort of visual systems are not to tell us what's actually there. Correct. But to, I guess, sort of help us navigate the world. Again, interesting, but potentially sort of leading to a fear. that sort of everything that we think we know is true is actually just survivability.

I mean, there are arguments in philosophy such as Alvin Planting as evolutionary argument against naturalism. I mean, the way that he frames it is that it's impossible to be a naturalist or a materialist and believe in evolution by natural selection because he says that in order to assert the truth of natural selection, in order to assert that this is the way that things occurred, You look at evidence, you observe evidence, and you assert your conclusion. But if the conclusion is true, then the very senses you're using to observe that data did not evolve for truth, but for survivability. So by asserting the conclusion, you're undermining your ability to get to the truth

of the theory that you've asserted in the first place. Right, that's great. So, yeah, he uses that as an argument for God's existence because he says that, you know, this is true on materialism, but if we're created by a mind intentionally, the problem doesn't arise, but I think that this kind of thought and this kind of, this kind of experiment, this illusion of the, of the cylinder on the checkerboard, and anybody listening who can't see it on screen, you know, just Google it. It's a wonderful image. It's called Adelson's checkerboard. Maybe tells us that planting is onto something there. I wouldn't go that far.

I mean, I think that science itself can also be best thought of as an abductive process, that it's our best explanation of what's going on. So the postulation that we evolved through natural selection is not something that can be logically demonstrated, but it is the best explanation we have for the things that we observe, even recognizing that the things that we observe are dependent on things that evolved.

I can see the circularity in there, but I don't think it's a particularly vicious circularity. Yeah, I guess not all. circles have to be vicious and we sort of have to start somewhere. I think there's some bootstrapping in any kind of, any kind of philosophy. But so our experience of the world is something like a meeting of the projections of our mind and the input data of the objects of the external world that we see around us. That's how I see it. And I think the overlooked implication of that is not so much the worries about whether anything is real or not.

And I think that, just to return, I think for me, that's a sort of pragmatic assumption to make. And it helps us bootstrap ourselves up a little bit, get off the ground in this explanatory cycle. But the other perhaps more interesting implication that has relevance for everyday lives, too, is that we will see the same world slightly differently, even if we don't recognize that we do. And this, too, is a bit counterintuitive. if we start from the kind of naive realism view that we see things, hear things as they are, then of course, we would just naturally expect to assume without thinking that somebody else sharing the same environment, we're in the same room here, is going to be experiencing things in exactly the same way because we all experience things as they are.

And the character of our experience is like that too. It really seems as though I'm seeing the world as it is in a way there's, independent of my own brain and mind. But if that's not the case, as I think it's not the case, then just as we all have different bodies in skin color and height and so on, we all have different brains. So we're going to exhibit a kind of inner diversity just as much as we exhibit this external diversity. We're going to have a perceptual diversity. And that is, well, that's something we're currently studying because a lot's known about it. And I would say when the diversity gets sufficiently large that we slap a label on it and say, you know, this is autism or

this is schizophrenia. But there's this whole middle range where I think most people just assume that we have the same subjective encounter with things. Sometimes it's, this is thrown into doubt, there was that social media meme of the dress, which half the world saw one way and half the world saw the other way. Yes. And you saw black and blue. I saw black and blue. I also saw and see black and blue, and struggled to see white and gold. Again, a simple Google search for anybody listening, but I'll put it up on screen. You'll probably remember this if you weren't living under a rock at the time. Fascinating. You talk in the book about how people started doing experiments, because I mean, presumably it has something to do with the fact that in this image,

you don't know what color the lights are. Right. So, you know, if you've ever taken a camera, most cameras will do this automatically, like the camera on your phone, but if you take a professional camera and you don't set the white balance, then, you know, the cameras that we have on us right now, if I've set them correctly, make us look sort of, you know, roughly the, the color that we actually are in the room. But if we were to move into a room lit with yellow lights, we'd suddenly look really orange on the camera, even though to our eyes, we look the same because, you know, the camera doesn't have a brain that can make that projection, right? The camera is seeing just what it's presented with. It just sort of becomes orange because the lights change.

But if we go into a separate room, we appear, everything still appears the same color, which I guess is further evidence that what we're seeing is in part what our brain is projecting. But in the case of the black and white dress, it's like, depending on how the white balance of the camera was set and depending what color of the lights were, it would sort of change how it appears on camera. So you talk about people like looking at the image and then running outside to change the lights to see if it will sort of change their perception. I don't know if there were many cases of people who were able to sort of flip,

between the two. I think there are a few. I've never been able to do it. I mean, I can see it differently if I change the background enough because, yeah, you're absolutely right. It seems to be, so we have this basically

biological, automatic, white balancing happening all the time. So if you take this, you know, I've got a piece of white paper in front of me here, it looks white. If I take it outside, it still looks white.

But the light coming into my eyes will have changed massively because indoors, it's a relatively orangey, ambient light artificial lights we have, and outdoors, it's going to be relatively bluish. It's sunny day. And our brains automatically compensate for that. And so we see white. And that's a useful thing for us to do. You know, you could make the argument that the camera, of course, it doesn't subjectively

see anything, but it's more accurate in registering what's there. But that's not very useful. Imagine if our perceptual experiences just changed all the time, depending on whether the sun went behind a cloud or whether we walked from one room to another, that would be a very, very maladaptive way of encountering the world. So all that stuff is hidden from us. And the example of the dress showed that this particular process of biological white balancing actually varies a little bit from person to person. And it just, by happenstance, serendipity found this sweet spot where some people's white balancing was such that they saw it one way and others such that they saw it another way and the kind of vociferousness of the arguments that it sparked

about no it's really it's got to be blue and black and I don't know what you're talking about yeah that for me was the interesting part of it because it really surfaced that we do bring to bear this kind of naive realism about our experiences so that it's very hard for us to even concede or conceptualize that somebody else might experience the same thing reality seen differently and of course it's not just the dress so the dress came and the dress went but the deeper interest for me is that even when the differences are not so large that they surface into social media memes and into language that they're still there i mean you and i might be having a slightly different experience of the sound of the background noise here or the or the um the color of this red on the mug

The background noise that hopefully my audio guy has expertly taken away. Maybe we can turn it back on again for this little bit. But we would never know, right? Because it seems as though we see it as it is, and we're going to use the same words. So this project, which we're doing, is trying to actually measure this relatively hidden landscape of inner diversity. So by the way, if people want to take part in it,

it's still something we're collecting data on it. It's a whole series of hopefully fun and interesting. and short, engaging, little interactive illusions, the sort of thing we've been talking about, but the sort of thing that can tell us about these individual differences. Where can people find that? They can find it. The easiest way to find it is on my website. The project is called The Perception Census, and my website is anilceth.com. And so that will allow people to sort of answer a bunch of questions about various things that

they're shown, and it will give you some insight as to the different ways that people are perceiving, what appear to be the same thing. That's right. So we look at vision. We also look at other things like time perception and sound perception, music, perception, rhythm perception. So I think this diversity, we often focus on vision in neuroscience, but of course our encounters with the world are much richer

and probably vary on all these dimensions. So yeah, there's all sorts of little interesting, interactive things to do. And people taking part will learn about perception, learn about their own way of perceiving and also have the warm glow of contributing to this for this scientific enterprise. Yeah, well, the link to that will be in the show notes

or in the YouTube description. And you're quite right. I mean, we have a sort of arrogance that might be caused by having a mistaken view of what causes perception, right? Because you're like, of course the dress is blue. I can see it.

It's like, I wouldn't see it as blue if it wasn't actually blue, but something like this phenomenon teaches us that it's not quite as simple as that. Audio illusions are interesting too. I mean, it seems to me that those are easier to potentially switch between. Have you heard the brainstorm green needle one?

That's one of my favorite because there's also the Yanny and Laurel one. That one people tend to say, you know, I sort of hear one or the other, but the reason I love the brainstorm green needle thing is because you can just hear one or the other. depending on what you're expecting. I'll play it for the people listening as well.

So it's sort of this, it's from some kind of toy or something. Yeah. I imagine it's actually saying brainstorm, but I'll play it. I'll splice it in here. So if you listen to this audio, I'll play it sort of three times in a row,

and you listen for the word brainstorm. I'll play the sound now. You listen for the word brainstorm. You hear brainstorm, song, but if you listen to it again, but this time listen for the words green needle, I'll play it again. Then you hear the other one, and you can sort of rewind that. I'll just play it one more time, listen for whichever one you want, and you'll hear it. And it's just whatever you're expecting to hear, you hear. What's remarkable about that, I think, is it's not just like hearing

brainstorm and then hearing something like shame form, you know, green needle. It's like, you write the words down. They look entirely different. Like what, what phonetically or linguistically do these things have in common? Green needle and brainstorm. I spent ages sat there trying to figure out like which syllable matches onto the other one, you know, how it is that, how it is that you can hear one of the other. But you're quite right. It's like completely an utterly opposite. And I think that really does underline that the power of our expectations to constitute what we experience. It's not just some sort of like mild fussing around on the top. It's really foundational because we can experience these two

very different things purely by expecting to hear something. Yeah, I mean, there can't really be an opposite of a word phonetically speaking. But if there were such thing as the opposite of brainstorm, it would probably be something like green needle. Like you say, what we expect to hear changes what we actually, in fact, hear. And when we listen to something, we think, no, that's what it is because I can hear it. My perception is being caused by the object. That must be what it is. But this demonstrates that our perceptions or our expectations shape how we view the world.

And to me, I think this must have huge implications for worldview and for philosophy. If you expect to find in the world evidence everywhere abounding that God exists, then you'll find it. Yeah. If you expect the world to be a pessimistic place full of suffering and misery, then that's what you find. mind. Similarly, I suppose if you have an optimistic world view and you expect the world to actually, you know, feed that back to you, then that's what you receive. So I mean, surely this can explain a lot about people's political and religious dogmatism that you meet somebody else and you just don't understand how they see the world that way. How can you possibly think

this is how the world works? It's because that's what we've been taught to expect, right? I think that's exactly right. And you mentioned earlier that we might bring to bear some kind of unrecognized arrogance in our ways of seeing because we don't recognize that they are dependent on our own particularities of our own minds, brains and bodies. And so part of the value, I think, of this work, part of the implications politically, philosophically, sociologically, is in deflating some of their arrogance and cultivating in its stead a kind of humility about our own individual perceptual takes on the world. And it's a humility that doesn't it doesn't dive all the way down into total relativism and idealism. We've talked about this

already. There is a world and evolution has made damn sure that we see it in ways that for most of us most of the time are useful. They're not going to be fully accurate. They're not going to be completely identical, and cultivating that humility at the level of something that seems so natural as perception, and my hope is, there's a bit of a hope, is that that can provide a bit of a platform for cultivating a parallel humility when it comes to our beliefs about things. If we recognize that we can literally see things the same thing differently, then maybe that gives us a bit of pause for thought about the things we believe and how to, least begin to understand that other people can believe something that seems so contrary to our

own ways of thinking. So we have, at least we know, we're familiar with the concepts of things like media echo chambers, social media echo chambers, where we have this kind of confirmation bias that will seek out things in favor of what we expect. It's almost impossible to step outside of that bias. We have to make a big effort to do so, to watch the news channels that we don't normally want to watch and so on. Another way to do it is, I think, to just to recognize that they're also perceptual echo chambers, that at the level of how we move our eyes around the world and how we experience our immediate environments, that's also something that is not revealing things as they are. I think, yeah, you're right. These things sort of line up quite

nicely. It's not sort of a direct leap from perception to belief, but I think it's the same A lot of our beliefs arguably are also formed in this kind of abductive way where there's some evidence out there. We all read some kind of news. We all get some kind of information about the world. But even if we see the same information, our prior expectations about how the world is structured, how it operates, are going to change our conclusions from that data quite radically. And we'll come to different beliefs, different, you know, Bayesian posteriors, as they say, in the, in the, in the, more technically, even for the same data. Yeah, of course, in more fundamental aspects of perception,

even knowing that our expectations change how we view things, doesn't mean we'll be able to see things differently. I mean, for example, to bring it back to colour, I'm looking at an orange mug. The mug, apart from the handle, is one solid colour all the way around. But of course, it's not actually, because it's a curved surface and the light is hitting it at a particular angle and parts of it are more shaded than others,

which means that I'm actually seeing this, including the reflections, I suppose, just this unimaginable sort of vast expanse of different wavelengths that are all coming off of this mug, but my brain just perceives them as one colour. Because it takes in the context around it, and even though I know that's what's happening, I know that this side of the mug is a different colour to this side of the mug because there are different wavelengths bouncing off it.

my brain sees it as the same and I sort of sort of can't shake that off. You know, one of the unexpected things that thinking about perception this way is brought for me is a newfound appreciation of art and the skill of artists. I was just thinking about that. If you think about what, you know, not just sort of more recent now, but going right back, I've just finished reading a biography of Leonardo da Vinci and, you know, you can see in what he was doing. You can see that he's diving underneath sort of our naive encounters with the world, like that this mug is solid orange. Because if we really reflect on our visual experience, indeed, we see it both ways, right? You can see it as one single orange mug. But when you start to pay attention,

of course we can still see the fine variations in things. And some of them maybe we can't see, but then artists like Leonardo and onwards know what's going on and are able to paint not what's there, but paint the raw materials for our brains to make the same kinds of inferences and so that it generates the subjective experience of what seeing a real orange mug would look like. There's this concept in art history called The Beholder's Share from art history called Ernst Gombrick. And I think it's extraordinarily parallel to these ideas emerging in neuroscience. and the Gombrick, when we encounter an artwork, the beholder's share is the part of the aesthetic experience that's contributed by the observer, the beholder that comes from within and is not

the present in the artwork itself. And he talks about all sorts of traditions, impressionism in particular, was really scorned when it first emerged on the scene as being just like, you know, low resolution scrapings and marks on a canvas. But the genius of the impressionist like Isara and Monet Manick was to was to see through

their own visual cortex to paint the light, not paint the brain's inferences about the light. I think that's what gives it its power. Yeah, I guess artists are people who can break out of that illusion. I mean, I think I would, it's the reason I can never be an artist.

I mean, even just sort of the perception of shapes, you know, if I hold up this book available online, Amazon.com, and in the link in the description, I see it as rectangular. But the angle I'm holding it at means that what I'm really sort of seeing are two lines going upwards and inwards to a sort of vanishing point that's never, that's never reached. This is sort of a strange, four-sided shape that isn't a rectangle. But my brain sees it as a rectangle. I can't. I can't. I think you see it both ways at once. I think it's a mistake to to say like we really see it one way or the other way. I mean that's that's the interesting thing

about the phenomenology of vision. Well an artist can do both exactly and we we do to some extent too I mean we we see it as a as an object you know as a rectangular object but we also still you know our visual experience is still you know you can still extract that from your visual experience that there are lines tapering to a point but I think you have to you have to focus you got to really pay attention to that I've I've never really been able to get a to get a grip on. I mean, even the table in front of me, you know, I'm sort of seeing there'll be a long end here and a short end there, but my brain just perceives them as the same. I think that good artist is someone who's trained in being able to, to understand that that's

not what the eye actually sees. And it's the same thing with color, you know. I remember once learning how to paint like a sunset or something, it was this, this orange, blue, you know, gradient. And they said, okay, now you need to pull out some green to put in there. And I thought, what on earth? Why do you need green? And it's because something about that color sort of makes it seem how it actually should seem, even though that's not really the color you're trying to put into the image. It just sort of makes it work.

And that's what makes art so fantastic. And so, and like you say, so impressive. So we can add the whole history of art now to our evidence base for, for the predictive brain. It's looking pretty good, right? So, I mean, does this constitute for you sort of a theory of consciousness, or is this more like a theory of perception? It starts as a theory of perception. I'm a bit skeptical of sort of self-proclaimed theories of consciousness, which gets us right back to where we started, that it seems as though consciousness is this potentially immaterial thing. And I respect that intuition. I don't necessarily agree with it. But

but rather than trying to face that problem head-on, this so-called hard problem of consciousness, like how is it that physical stuff is identical to or generates any kind of experience? I think the way forward, or at least a productive avenue to follow, is to say that conscious experiences exist because there are some philosophers that claim

we're mistaken about that too, and there's really nothing that special about what we think about what we're talking about when we talk about consciousness. But I think, you know, conscious experiences exist and they have properties, and we've been talking about some of them, how visual experiences appear to us. And if we can start to account for some of these properties in terms of brain mechanisms, and we've been talking exactly how to do that, you know, what gives something a feeling

of being an object, what gives it a feeling of visual presence in our world, there are many things we could do with this. we're making inroads. The temptation is to say that that will never be enough and we'll just be left with a theory of like some aspects of how we experience the world but we will fall short of a full understanding of consciousness itself.

I think the only, at least for me, the intellectually honest approach is a wait and see approach and to identify as many different aspects of consciousness as we can and try and account for them in the same way that we've been doing. Can we use the resources of neuroscience in particular this idea of the brain as a prediction machine to explain not just visual experience, not just auditory experience, but also experiences of self. And that's why the book is called Being You. That's where I've been going over the years is like kind of starting with this view of how the brain

creates the perception of the world and then turning the lens inwards to think, okay, we have this other of naive but very powerful intuition which is the self is the thing that does the perceiving, perched inside my skull somewhere

and decides what to do and maybe exerts free will and then does something. But really it's hard to make sense of that because what is this self? Is it some residue of some soul, some immaterial soul. And as a whole tradition in philosophy, again, going back to David Hume and also in Buddhism and Eastern traditions, that talk of the self very differently as a process, something that's

impermanent, something that's always changing, and something that I think critically is productively thought of as a form of perception too. So the self in this view is not what does the perceiving. The self is itself a kind of perception, a collection of perceptions. And then you pull on that thread some more and you can start and you can it becomes for me a theory towards a theory of consciousness when we ask the question the most fundamental question about why is the brain doing any of this at all we've given one answer which is it's the only way of making sense of ambiguous data from the world but I actually think there's a deeper reason and the deeper reason is that the fundamental purpose of having a brain for any

organism is to keep the body alive. Everything else is secondary. Even movement is probably secondary to this. And keeping the body alive means regulating processes in the body like heart rate, blood pressure, whatever. So they stay within very tight bounds that are compatible with staying alive. And as any control engineer will tell you, the best way to do this in a world where things change and there's uncertainty is through prediction. If you can predict what's like, what's like to happen and what that's likely to do to things like your blood pressure and so on, then you can keep things regulated very tightly, not without waiting for them to change and then try and bring them back, have anticipatory control. So the whole reason we have predictive brains

in the first place for me is this deeply embedded evolutionary imperative to keep the body alive. and everything else piggybacks on that. And our experiences of the world then become grounded in our biological embodiment. And so it becomes a theory of consciousness in the sense that it's making a claim about this very tight relationship between our nature's living creatures and our conscious experiences. And it also tries to get at why we think there's this big problem about consciousness. If we can explain, for instance, why we feel we have free will, why we feel we have this immaterial mind,

using the same resources that we can understand how and why the mug seems to be red and white, then I think we're getting somewhere. We're making progress. What do you think, I mean, where do you think the self is? I mean, is the self in the brain? Is it the brain? Is it just the collection of perceptions? I mean, saying that the self isn't the thing that does the perceiving, but just is the collection of perceptions is,

strikes me as a bit like saying, you know, like when we're trying to talk about color and we say something is red, there's no thing that is red. There's sort of just the redness, but that doesn't make much sense. There sort of needs to be something for the color to attach to it. It's a secondary property. It doesn't just exist floating out there. It needs to be something to. I mean, it can do if we're imagining or dreaming it or something. this is true um although i wonder sort of again what the difference is between imagining red and

and seeing red and whether whether you are sort of actually having the same experience in the brain i mean it's not literally the same experience right but like i mean dreaming is probably closer i think most people imagining it's it doesn't have the same vividness but when we dream to the extent that we can remember dreams the colors seem vivid in the same way they seem in normal life but anyway it's it's stands, right? Yeah, yeah. And it seems like, I mean, I sort of feel like myself is in the brain, but there's this, I suppose, quite bias-inducing sense in which my eyes are right next to the brain. And my eyes are sort of my primary, I would say, way that I observe and navigate the world, or the most useful, most striking maybe. But, you know, if my eyes were located on my chest,

I wonder if I'd sort of feel my brain behind my eyes or whether I'd still feel it up in my head, you know what I mean? And some people talk about sort of, you know, thinking with the gut or how the gut can affect the way that we think. And people have said for a very long time before we knew much about the science of it, you know, go with your gut. There was sort of this intuitive sense in which we could feel that there was some thinking being done by other parts of the body.

I mean, David Hume goes as far as to say that even sort of, you know, if I were to sort of hurt your foot, somehow if I were to stub your toe, the only reason that you know that that pain that you feel is coming from your toe is because you have experience of a similar feeling every time you've done something to your toe. But if you had no prior experience, if I were to stub your toe and you had your eyes shut, you wouldn't know where the pain was coming from. To me, that seems like, I mean, it seems to most readers very radical and perhaps quite unbelievable. You can kind of see why he might think that. But there's this weird sense in which different kinds of experiences

seem to be located at different parts of the body, of course, all connected to the brain. But if the brain is not the self, but rather the perceptions themselves put together, produce the self, like, where is this self? What is it? I mean, I think there's two things. There's the biases we have. And you're right that we have this kind of perspectival bias that our eyes are right up here near our brains. We could actually do the experiment and just like give some goggles and put a camera on your chest and have you walk around for a few weeks and see what happens. We also have this bias that we associate the self with thought often. And most of us experience the thought in as much as it's a linguistic

verbal train of thought as occurring somewhere up here too. These two biases, I think they tend to localize our experience of self up here. But it hasn't always been that way. And certainly the brain was not considered always the important place for self and consciousness. In ancient Greece, it was often thought the heart was where the soul resided. And in ancient Egypt, when they took great care to mummify important people upon death, they just usually drained the brain away and chucked it, chucked it away. Don't worry about that. Because if you think from that historical perspective, the brain is very inscrutable. It doesn't do much. It's just sort of, you know, tofu textured, grayish, pinkish stuff, much less exciting to look at than a heart or a

stomach. When did we start to realize that the brain really is the center of almost everything we do? I think it varied in different cultures, but in my sketchy understanding of the history, it was a big landmark was the Roman physician Galen, who did the first beautifully illustrated dissections of the human body kind of revealing not what was supposed to be there on the basis of religious or authority text, but on the basis of what he saw, to the extent of course he saw what was there. But he was able to get underneath his expectations about human anatomy a bit more than other people at the time and gave us a more accurate picture of the centrality of the brain to controlling the organism to perception and so on. I think that marked a

big transition. Leonardo as well actually made enormous contributions. He just didn't write many of them down, didn't publish many of them. So I think he could have been also this major figure in neuroscience had he published the work he'd done in neuroanatomy. Is that so? So how do we know that he sort of did all this work if he didn't write it down? I think from sketchbooks and notebooks. And I must admit, I'm basing all this on this recent biography of Leonardo that I read. But it was actually fascinating because we tend to, and he's such a vibrant archetype of culture, of culture of Renaissance, thinking that, and we know that he did so many things, but reading this biography of him, it's just astonishing how productive the guy was, absolutely stunning.

Yeah, I recently, I was in Paris last week and I saw the Mona Lisa. Although I can't say I was particularly impressed by it because I was sort of squinting from a distance. It's really weird I find in art galleries like the Louvre where if you want to find the Mona Lisa, what you do is you walk in and there's this big print of the Mona Lisa and an arrow and you follow the arrow

and then there's a picture of the Mona Lisa and an arrow and then another picture of the Mona Lisa in an arrow and you follow it through the museum. So by the time you get to it and you're sort of squinting at the real thing, you've actually already seen it sort of 20 times prior and you've actually seen it in better detail

than you're probably going to see you when you see the real thing so you know if you go to Paris and want to have a look at them only so you're better off looking at the big banner that's by the entrance it's like I thought about how if you go to a gig and everyone's sort of shuffling in and they're playing music while people are getting ready to listen to the band

it would be like playing that band's music out of the speakers before the band comes on which venues go out of their way to not do because of the fact that they think that's going to slightly ruin the experience. Well, and then it would be like when the band comes on, you have to put earmuffs on and they're not even listening to it in as much detail, right?

You can't even hear it. It's so strange, but from that distance, you can tell that there's something pretty extraordinary going on with this man's understanding of anatomy. There are a few instances in which you've already mentioned dreaming. in which our sort of perceptions don't appear to be exactly in line with what's actually being caused by some external world. The other example I can think of in which that happens

is the psychedelic experience. And you write about how some research seems to suggest that, I mean, normal waking experience, we have different ways of leveling sort of measuring how conscious someone is. These are a little bit dubious. But we have vague metrics on how we can judge how conscious is somebody. Somebody is less conscious when they're asleep and not dreaming than they are

when they're awake. And you write in the book that the only thing we have, the only evidence we have of a state of mind where someone appears to be more conscious than the baseline of sort of general waking experience is on psychedelic drugs. What is going on with the brain and the effects on experience when somebody takes what seems like a pretty unassuming little square puts it on their tongue and all of a sudden an entire new world opens up to them what's going on there's a ton of stuff going on it is it is remarkable isn't it how this this tiny little pharmaceutical intervention can so radically upend our encounters with the world and with the self i mean this is another just to connect one dot here on psychedelics the typical experience of self can be altered very

profoundly people report things like ego dissolution so the whole idea of being a self can become much less apparent and where is the self is that you know where's its boundaries these things can be radically changed which i think is a way of revealing that our normal experience of selfhood should not be taken for granted as just how things are yeah but it's also a kind of construction um in psychedelics i think there's two two general interpretations there are some people who would take the psychedelic experience as revealing to them a deeper truth. So they're seeing things more as they are. It's taking the filters off. I think of things a different way, which is that psychedelics for me, it really underlines how deeply material our conscious experiences are. You know,

you can change the brain in this very, very precise but profound way. And your experience changes radically and the things that we would take for granted as just being reflections of reality, the reality of the world, the reality of the self, are revealed to be constructions of the brain because we can experience them so differently while still being conscious. Whether it's a more conscious state or a less conscious, I actually want to equivocate a bit about this. I mean, what we did in this one study, we took data that had been recorded from colleagues at Imperial College in London, there's an amazing set of studies where they recorded activity from the brains of people while they were taking LSD or psilocybin, which is magic

mushrooms, or ketamine, which in low doses can also give you a sort of semi-psychadilic effect. And it's rather reassuring that when you look at this data, a lot is going on. The brain does look pretty different in the psychedelic state. And it's different in very many ways, almost every way you want to look at it. it, you'll find a difference. One of the ways we looked at it was we applied a measure. You mentioned that we have these fairly coarse measures of global levels of consciousness. So there's a measure that goes down in sleep and goes down in anesthesia and so on. And it's really a measure of how many different distinct patterns of activity we can find in the brain. So the brain, the unconscious brain,

is more regular. It's more predictable. There's less diversity. And it turns out in the psychedelic brain, there's more diversity. It's a little more random. It's a little less predictable than even waking consciousness. So is this more conscious? I don't really think so, actually. I think that consciousness probably doesn't map to a single scale, like temperature does, for instance. It's a very multi-dimensional thing. We have perceptual vividness. We have the vividness of integrity of the self. We might have the thickness of our moment in time, of our temporal experience. You might have the strength of memories. There are many different ways we might want to say things are more or less conscious. And I think it's just the fact that our measures of level are still very blunt

that if we look at everything through this very, very blunt tool, then we see psychedelics as coming out higher than waking, whereas sleep and anesthesia are lower. There seems to be more of this predictive experience going on in a psychedelic trip. Like, you can have a good trip or a bad trip, and part of the reason for one or the other is just your prior mood, right? If you're in a bad mood and you drop acid, then you might have a bad trip, potentially because you're sort of expecting, you know, the world to give you some form of negative feedback. That's what it is to be in a bad mood, really. Or that's sort of what you

predict about the state of the world right now is that things are negative and this just gets sort of amplified. And maybe that's sort of, again, even further evidence that at least in the psychedelic experience, we've got a real sense in which our predictions. But again, it's not like conscious predictions. It's not like you can sort of sit there and necessarily just talk yourself out of a bad trip, right? But it is still, even though it's not something you're exactly consciously doing, it will make sense for me to say that the experience you're having is based on essentially your brain's prior state and what it expects to find in the world, which I think is quite fascinating. I don't think it quite explains why you might sort of see a chameleon

appearing on a tree or um or who was it a silo black yes streaming through the sky yeah that was my experience a few years ago um if uh i think it's whether it's silver black or a chameleon there's this quite common factor in cyclic experiences where when you look at sensory input that that is inherently a little bit ambiguous so like a nice sort of day with little white fluffy clouds is quite good for this because we've all had the experience of seeing faces in clouds anyway, even when we're not on any kind of drug at all. That too is just a clue that our brains are project faces are so important for us as stimuli in the world. I mean, we're very social creatures. So the brains are always throwing out face templates into our sensory environment

and see to see where it sticks. And sometimes it sticks in clouds or in the patterns of windows on a building and we see an echo of a face in that in that sensory data and then on on psychedelics yes these things can be accelerated a bit so a bit a bit a lot right so yeah you can see in my case i think there were animals to start with and then it was shortly after the death of sila black had been announced who it was just like you know very very common oh not very prevalent tv personality and so yeah i don't i didn't realize i'd been thinking about her at all but clearly I'd been primed. That's something in my brain was just sort of rehearsing Silla Black, and there she was.

You're a neuroscientist. You know more than the average person about how the brain works and what's going on. And probably just in your normal everyday experience, you're more likely than most people to really notice what your brain is doing when you're experiencing the world. You write about your experience taking LSD. What is it like for a neuroscientist? to undergo such an experience? Well, I don't have the comparison me

that is not a neuroscientist to compare against, but I found it absolutely fascinating. I didn't take psychedelics when I was young. I didn't seem to be around. I'm not sure what a decision I would have made, but in any case, it was only after I became already quite far down the road of being a neuroscientist interested in consciousness

that I thought, okay, I want to know what this is like from the first person because it's all very well studying altered states of consciousness from the third person. I think it's a very valuable thing to do. You know, we can understand, in general, if you're trying to understand how a system works, it's useful to look at it when it's in states other than its normal state because you can bring to light some principles. But just looking at how it was working in the third person, person, just spark my curiosity.

What is it like on the first person? Your Mary sat in the room studying blue without experience in blue. That's exactly right. And I think the same answer applies. You know, did I learn something? And the answer is, I learned something. It's a different kind of thing. You know, I learned a phenomenal fact rather than a conceptual fact about what psychedelics are

like. And also you just, you experience the detail. like no third person description is really exhaustive. I mean, it's notoriously hard to describe what these experiences are like. Michael Pollan in his book, How to Change Your Mind, I think does a brilliant job of trying to convey with normal language what some of these experiences can be like. Yeah.

But just to have it is a kind of, you know, that's the best way to know what these experiences are like. And I was very reassured. I mean, in the sense that it was more or less what I experienced. expected, but at the same time, it made a big difference to me because it was happening to me and it wasn't just me reading about it. Did it change more than just, did it teach you anything more than just the experiential? Did it change the way that you think about consciousness or perception or the way that the brain works, perhaps even after the effects had worn off?

I don't think so, to be honest. I think it, as you said it, it probably just galvanized. my confirmation bias and made me believe even more strongly what I already believed. I wanted to ask you a question that you must have been asked at every dinner party you've been to in probably the past decade. Can computers be conscious? And this is a very timely question. We're speaking as there's all these urgent calls for the regulation of AI in the wake of chat GPT and GPT4 doing extraordinary things. And my short answer is, I don't think so, certainly not in their current form. But there's still a lot to worry about in this area.

So why don't I think so? Well, I think in AI, there's often this tendency to think that consciousness is something that's just going to come along for the ride as AI gets smarter. And AI is getting smarter. There's really no doubt about that. But the idea that there'll be some threshold, and often this threshold, is thought to be when AI becomes as intelligent as a human being. So the lights come on and there's suddenly experience happening as well.

I think this is a very poorly founded assumption and it stems from a kind of human exceptionalism that we still are saddled with that, you know, we know we're conscious and we think we're intelligent in some species specific way. So we think that the two are intimately connected and have to be intimately connected. So that when something is intelligent like us, then it will be. conscious like us. And that doesn't that doesn't really follow at all. There's no reason in principle that I can see why intelligence in an artificial system requires consciousness. Isn't it a form of human exceptionalism to think that there is something special about our brains doing sort of

essentially computorial calculations and that that produces consciousness but if you do the same thing on a physical computer that no, no consciousness. It's not only that it sort of doesn't exist right now in chat GPT, but that it can't. I mean, it seems to me strange to suggest if we're going to be materialists that, yeah, my brain is just a bunch of atoms colliding with each other in such a way as to form this first-person experience that AI just doesn't have. I mean, there are already AI chatbots and sort of AI robots that almost resemble a philosophical zombie, the concept of a philosophical zombie, of course, being a human being who is materially exactly the same as, say, I am right now, reacts in the same way, moves my hands,

talks, responds to information in the same way, but isn't conscious. There's no first person experience. It's just a sort of object. And there are AI chatbots that you can say, like, are you conscious? And they say, well, what's it like to be conscious? And they say, well, I'm not really sure. It's sort of like being aware of myself and I sort of, you know, experience the world. I have a first person sense of experience and who are we not to believe them? Yeah, no, you're right. We don't want to go too far in the other direction. But, you know, when it comes to what current AI systems do, there are these chatbots and actually if you ask them if they're conscious, they probably have guard rails in. They say, I am not allowed

to speak about this. I am just a chat bot. Which makes it all the more suspicious. Don't you think? It's as if you're asking them if they're conscious and the robot sort of looks at you and goes, no. What are you talking about? Conscious? Come on. No, man. Exactly. Look over there. Nothing of interest going on over here. But we're always walking this line between being anthropocentric or this human exceptionalist tendency where we think there's something really special about us and being anthropomorphic and projecting human-like qualities into other things on the basis of similarities that turn out to be rather superficial. it's a complicated balance of strike and so this conceptual distinction between intelligence and consciousness is not the only reason to think that current AI is not conscious it's just one reason to be wary of the assumption

that it is an inevitable consequence of AI on its current trajectory if engineers as some are doing set out with the goal of let's try to build conscious AI, then I think things get a lot less certain because there is no consensus theory about what the minimal necessary and sufficient conditions are for a conscious system, whether it's a biological system or an artificial system. There are many different theories. Now, on my own theory that we've been discussing, I feel, and the evidence is pointing me

towards the view without establishing the view that consciousness and life are very intimately related. So on this view, if I'm right, you won't get consciousness in a machine until the machine also is a kind of living machine in some ways. And that's not the trajectory that AI's on. But I might be wrong. You know, there are plenty of other theories of consciousness that indeed suggest that consciousness is some form of information processing. That's, of course, something computers can do. The big question here is one of substrates dependency. Does it matter that a computer is made out of silicon chips and wires broadly and we're made out of flesh and blood

and cells and carbon broadly. There's a great temptation to say that it doesn't matter because why should it matter? We can build computers that play chess. That's fine. They play chess or they play the history of chess. So why should it matter the substrate?

Well, I think it might matter. And I think we apply this idea that the brain is just doing information processing. It's a little bit too lazy and easy. It's a residue of another metaphor that the brain is a computer. The brain isn't a computer. It's very different from a computer in many ways. And one of the key differences is that a computer, by its design principles,

has a very sharp separation between the hardware and the software. The hardware is the substrate made of silicon, and the software is the program you run on it. The whole point of computers is you can run different programs on the same hardware. Let's open up the brain again and have a little. look inside there's no colors there's no self there's also no obvious point where if you like you would identify this distinction between not the hardware and the software but between the mindware and the wetware every time a neuron fires the brain changes there's as much chemical things swilling around

as there are electrical spikes flowing around even single neurons seem to behave under this similar imperative to keep themselves alive. Single cells sustain themselves over time. They regenerate their own components. Like no computer, no machine regenerates its own physical components. And this imperative for predictive regulation kind of goes right the way down and it's where does it end? So on this view, it might be life that really breathes reality into our conscious experiences

and makes our brains do more than merely simulate consciousness and in fact instantiate it. So that's one reason why consciousness may not be possible in machines that we currently have. But again, other theories suggest they might. They currently don't because I think the chatbots we have now, they're not even truly intelligent. They're just very good mimics.

Careful what you say about them, you know. They might get angry and it will come back to haunt you. one day. It may do, but... I personally think they're very intelligent and creative and good luck. Oh, they're creative, but the creativity is like, it's partly in the eye of the beholder, right? Yes. But I think the real danger here

is that let's say, let's say that we are actually successful, which is not a goal I think we should even be pursuing, but it's a goal that in the tech industry, some people seem to think is like, cool, let's try and build a conscious

machine. if they turn out to be successful first we might not even know because we all have different ideas about what's involved but if it nonetheless happened we'd have introduced into the world a vast potential for suffering that we might not even recognize as suffering and that would be an ethical and moral absolute catastrophe because you know we could replicate innumerable instantiations of artificial suffering by the click of a button in server farms everywhere. That's not a situation we want to be in. It's not so much, you know, fear of Terminator style being enslaved by conscious and vengeful robots. It's more, you know,

we already fail to adequately treat ethically other living systems, conscious systems, other animals. So the idea that we would behave ethically appropriately to artificial consciousness is, you know, that I don't think we've got a good track record there. Yeah, I mean, we sort of imagine AI being the bad guy, right? Yeah. It's just as plausible, I suppose, that it could be the victim. Although I do think that people would assume that if its intelligence can sort of outdo our own, it would be very difficult to victimize an artificial intelligence.

Maybe, I mean, there's this idea that, yes, once something has consciousness, then instead of it behaving according to our interest, And this is already the problem in AI value alignment. How do we design systems that act in the interests that we give it when it's smarter than us and may just maximize a goal by doing things we hadn't predicted? So there is this worry that AI that's conscious may in some Terminator style take over. And if it's smarter than us, how can we control it? I think there's a little bit to worry about there.

I mean, it's a worry about AI in general. I think the specific thing about artificial consciousness here is that, that when a system is conscious, it makes sense to think that it might have its own interests because it has its own subjective experiences rather than our interests. And already in AI, there's this problem of value alignment where how do we ensure that a system with autonomy

behaves in ways that are compatible with what we want for the benefit of humanity? It's very hard to sort of formalize and operationalize that in a way we can guarantee that AI will do the right thing. for us. That problem becomes more more complicated for some putative, very speculative future conscious AI and just want to reinforce. I don't think that's around the corner at all. I think it's a long way away, even if possible. But I think the more immediate danger, and this is, this may already be here. I mean, you were saying already that you can talk

to these chatbots and they give you the fluent impression of being a conscious mind. And they it's this balance again between anthropomorphism, anthropocentrism. We're suckered in by our anthropomorphic biases here and it's quite easy, it turns out, for us to be convinced that there's a conscious mind

when in fact what's going on in large language models is simple next token prediction with some bells and whistles around the edges that's trained on the entire internet that we project consciousness into it. And I think that can be very disruptive for society as well. You know, if we start projecting consciousness into things where there's no good reason for consciousness to be there,

then we might make assumptions, wrong assumptions, about how it's going to behave, what it might do. Because if we think it's really conscious, really understanding stuff, consciousness and understanding, by the way, might be separable things too. then it might behave one way but if it isn't and it's just algorithms wearing away in the subjective dark then even if it behaves how we might expect it to behave 99% of the time

the remaining 1% it might do very differently so society is very shortly going to be faced with this this challenge of how do we how do we arrange things in a society where we're interacting with systems that give us the impenetrable appearance of being conscious in the same way that these visual illusions, we can't see through them even though we know what's going on. Even as a neuroscientist, I look at that checkerboard and I still see the two squares. It has different colors. Even when we know what's going on, we're still unable to encounter them as being anything other than conscious, even though we might know they're not. I think that's extremely challenging for society in many ways.

And that's around the corner, if not here already. So we have these two worries. We have the sort of far-fetched worry about actually conscious machines, a lot of uncertainty about it, it's not around the corner, but the consequences would be tectonic. And then we've got the near-term dangers, which are less sci-fi exciting. but I think much more realistic, much more immediate and also challenging.

And we haven't really mustered collectively our resources to know how to confront this situation. Something of a call to arms almost implied there that this is something we need to do. There's a lot of discussion, I think. It's just incredible how the importance of regulating, thinking about, about the social consequences of AI have shot to the top of every political, economic agenda over the last couple of months.

There was a Senate hearing last week about this. I think that's good. I think there's obviously a danger of overreach and overreacting here, but I think regulation is necessary. And what I'm saying is simply that in these discussions, the issue of consciousness is often not discussed, or if it is discussed, I think it's discussed a little bit in an uninformed way and often under this kind of assumption that it's just a

function of consciousness. I think the topic ought to be part of the conversation. I mean, you spoke about sort of, well, it seems like it's conscious, but it's actually just, you know, a bunch of sort of learned data from the internet that it's just sort of processing and spitting out. Is that not kind of what our brains are doing in a way? In a way, but I think that's, so that's the challenge for us as well. It's not just a challenge for AI researchers. I mean, really, I think it underlines the importance of understanding more about how and why we are conscious. I'm very suspicious that it is just information processing that happens to be implemented on a biological computer. Just the more you look at the brain, the less like a computer it seems to be, there's the importance

of our embodied embedded interactions with the biological body and with the world around us. And For me, it's very unlikely that large language models even understand anything. Because there's a good case to be made that understanding depends on learning the meaning of words through sensory engagement with the world. The fact, when we talk about colors, what does a color mean? Rather than just being an abstract set of tokens, it gains its meaning because of the things we've been talking about. We have perceptual encounters of things with different colors. So even to get large language models to understand, let alone be conscious, might require that we train the whole thing in a body. Yeah, we need to give it senses, right?

Because without, I mean, you talk about senses and the balance between our prediction and the sort of raw material of experience being so crucial. Yeah. That balance to our conscious experience, it's difficult to imagine. I mean, it'd be like a brain sort of locked in its skull with, with no sense of sight or hearing or touch or balance or or bodily position or any of the senses like can you make sense of a brain like that even being conscious right is but probably not i mean that's one of these interesting but impractical thought experiments you know if you strip away the brain's interactions with the world and the body one by one you know what happens

of course if you lose vision you're still conscious it's fine but if you lose vision hearing touch taste smell maybe it's still fine but then what if you lose the the perception of your body from within too what what happens then what if you never had these in the first place or what if none of your ancestors ever had these in the first place then it's very different taking them away we can imagine something like a leftover brain that's that's just sort of locked it's just it's just it's in this dark room and all it can do is think. But we can imagine it still being conscious because maybe it's had this experience in the past. But if it starts that way, maybe it never develops the necessary prerequisites for consciousness at all. And maybe that's something like what we have

with a computer is something that has all of the necessary instrumentation to do that kind of thinking that a brain does. But because it doesn't have that sense data, it's not going to be able to be conscious. And so maybe once we sort of, you know, start attaching cameras and microphones to these things a bit more routinely, we might be verging on something a little scary. I don't think that would be enough even because it's not, it's not only that we sense and can act in the moment. It's that we did this all our lives. All our ancestors did this all their lives, every brain that ever existed, the whole history of brains in our common ancestors were all embodied, were all embedded. So to the extent that we come preloaded with sort of

ability to understand, perceive the world, it all depended on embodied interaction. So it's not just you can plug a couple of cameras and on a large language model in a robot with some wheels on it. It might be that you have to train, pre-train it in that way too. And of course, that becomes entirely infeasible. These language models are trained on so much data that if we were trying to do that, it's a totally different ballgame. I don't think that would be possible like, I mean, these computers learn so fast. I mean, the entire sort of intellectual history of human thought can be taught to a computer. But it's done quickly because it's offline. But it's done quickly because it's totally offline. It's not embodied in the world at all. I mean,

it's just, you know, reading things at the speed of electricity. So I think it does, it does raise a lot of additional challenges. Well, Anil Seth, thank you so much for coming on the podcast. It's been a very wide-ranging and fascinating conversation. Thank you, Alex. It's been a pleasure. Thank you.