Theories of Everything with Curt Jaimungal - Anil Seth on The Neuroscience of Consciousness, Sapir Whorf, and Daniel Dennett's ideas

Episode Date: November 23, 2020

Patreon for conversations on Theories of Everything, Consciousness, Free Will, and God: https://patreon.com/curtjaimungal Help support conversations like this via PayPal: https://bit.ly/2EOR0M4 Twitter: https://twitter.com/TOEwithCurt iTunes: https://podcasts.apple.com/ca/podcast/better-left-unsaid-with-curt-jaimungal/id1521758802 Pandora: https://pdora.co/33b9lfP Spotify: https://open.spotify.com/show/4gL14b92xAErofYQA7bU4e Google Podcasts: https://play.google.com/music/listen?u=0#/ps/Id3k7k7mfzahfx2fjqmw3vufb44Anil Seth's Links WEBSITE: https://anilseth.com TWITTER: https://twitter.com/anilkseth TED TALK: https://www.youtube.com/watch?v=lyu7v7nWzfo00:00:00    Introduction 00:01:08 What's alluring about "consciousness" research? 01:03:51 Defining "phenomenonology" 02:05:56 What does it mean to "explain" consciousness? 03:14:52 Difference between illusion and hallucination 04:18:16 If perception is tuned for fitness, why care about objective reality? 05:24:44 Applying Donald Hoffman's ideas to Anil Seth's 06:33:39 Does time actually slow down, during high valence emotions 07:36:31 Details on the Bayesian model of consciousness 08:41:50 Can you be wrong about your own perceptions? 09:46:29 Myth of left brain vs. right brain 10:54:31 The role of the "ego" and the loss of it 11:56:45 Consciousness and quantum mechanics 13:00:37 Sapir Whorf hypothesis and Curt's polemics 14:10:24 Language and conscious experience 15:13:33 Can I "see" without noticing? What's the difference? 16:17:27 The Bayesian model of consciousness (and its limitations) 17:25:16 How does the Bayesian model deal with mental disorders 18:36:26 Gödel's incompleteness theorem's implications on the computational nature of the brain 19:42:23 Can we upload our minds? (a neuroscientific perspective) 20:45:48 Integrated Information Theory and measuring Phi 21:50:21 Peter Lush's recent paper and undermining the rubber hand illusion 22:55:09 Placebo effect and hypnosis 24:01:25 Experiments involving psychedelics (LSD and microdosing) 25:04:01 Anil's personal insights from psychedelics 26:06:18 Anil's views on Daniel Dennett's "Explaining Consciousness" 27:12:26 Compatibilism, free will, and Douglas Hofstadter 28:15:21 Global Workspace Theory 29:19:45 Anil Seth and Pragmatism 30:22:39 Curt's list of final questions, repudiated 31:23:35 How do the various AI architectures relate to the brain? (GAN / Convultional / etc.) 32:32:17 How does Anil stay so fit? 33:33:44 Comporting rationalism with irrational behaviors * * *Subscribe if you want more conversations on Theories of Everything, Consciousness, Free Will, God, and the mathematics / physics of each.* * *I'm producing an imminent documentary Better Left Unsaid http://betterleftunsaidfilm.com on the topic of "when does the left go too far?" Visit that site if you'd like to contribute to getting the film distributed (in 2020) and seeing more conversations like this.

Alright, hello to all listeners, Kurt here. That silence is missed sales. Now, why? It's because you haven't met Shopify, at least until now. Now that's success. As sweet as a solved equation. Join me in trading that silence for success with Shopify. It's like some unified field theory of business.

Whether you're a bedroom inventor or a global game changer, Shopify smooths your path. From a garage-based hobby to a bustling e-store, Shopify navigates all sales channels for you. With Shopify powering 10% of all US e-commerce and fueling your ventures in over 170 countries, your business has global potential. And their stellar support is as dependable as a law of physics. So don't wait. Launch your business with Shopify. There's a lot of questions. This is like some sort of super hard interview for some thing. I hope you don't expect me to have answers for all of them. I'm here with Professor Anil Seth of neuroscience at the University of Sussex.

There aren't many people who are brave enough to tackle consciousness scientifically. Maybe there's, maybe there's 1000 people and there's maybe 5% of that are paragons in the field. So that's like 50 people I'd say. And I would include a Neil Seth in that. I'm super excited to speak with him. If you look up his talk on Ted, I believe it's one of the most viewed We Hallucinate Realities. You might want to view that before watching this to get an overview of Seth's point of views. I'm also here with Faraz, who has a background in neuroscience and medical science, a master's in it actually, medical science. And I thought it'd be great to have Faraz tag along. This is going to be somewhat free-flowing because me and Faraz aren't communicating with each other off screen in some way. So at points he might interject and it might overlap with mine,

but that's fine. We'll just take it as it goes. Welcome. Thank you, Professor. Thank you, Kurt and Faraz. It's a pleasure to be here. So Professor, what are you currently working on and how did you get interested in the field of consciousness? So, well, two different things. I mean, consciousness for me has always been interesting. It's one of those questions, I think, that pretty much everyone is interested in at some point in their lives,

certainly when you're a kid. It's just one of these things, these big mysteries. How does it fit into our picture of the universe? How does it depend on the brain? What does it mean to be me? What happens when i die and all these sort of big questions are things that most of us wonder about and then usually grow out of and and do something more normal with with life uh so i've just been lucky enough to

never have been forced to go out of these these questions but of course that doesn't mean that every day is like 12 hours a day just trying to answer this big spooky mystery about consciousness. In fact, day to day working in this area, working in this part of science translates into a whole variety of different things that are related more or less directly to this big question so just for example today we've been talking with my people from my research group and colleagues about various things like how do we mathematically get a grip on an emergent property people often talk about consciousness as being emergent from the brain somehow but what does that mean mathematically that's's one thing. Another thing,

we were looking at developing computer simulations of different kinds of hallucinations to get a better grip on how different kinds of visual hallucinations differ and what the underlying mechanisms in the brain might be. And then, you know, just before this meeting, I was might be. And then just before this meeting, I was with a PhD student at the very early stages, trying to develop new kinds of experiments to understand the difference between what we might call gist perception. If you just see a scene very quickly, you can usually tell that it's a landscape or it's some very, very basic features of perception come out very quickly, but others require a bit longer to figure out, oh, am I looking at this person or that person, or is it a car? What are the details going on? So I'm very interested in thinking about

how to explain these differences in visual perception. So just to say there's just an awful lot of different specific questions that we can ask when we're trying to understand more about consciousness. Is it cool if we get some definitions out of the way? Yeah, I think that's always a good place to start. Right. Phenomenology, the way that you use it, I assume it's different than the philosophical doctrine of phenomenology, the way that you use it, I assume it's different than the philosophical doctrine of phenomenology, which is Husserl's, Heidegger's, and then there's particle phenomenology. And I'm unclear when you say phenomenology exactly, what do you mean? Well, I mean, I must admit, I play a little fast and loose with the definitions. And it's, but the way I mean it is probably quite close to the philosophical phenomenology of Husserl and Melo Ponti and people like that. Really, it's just for me a way of drawing attention

to the properties of experience. So if we talk about, let's say, visual perception. So perception here is how my brain interprets visual sensory data to form representations of the world outside. And you can think of that as having like a functional aspect. So what does perceiving things visually allow me to do? How do I behave with respect to visual information? But there's also a phenomenological aspect. There's how visual experiences seem. The experience of redness is different from the experience of, let's say, an object in vision. An object seems to have a back, even though I can't see the back. There's

a phenomenology of objecthood. There's a phenomenology of spatiality. All my visual experiences have a spatial character. Things are organized in space in a way that, let's say, smell doesn't have so much. Emotions have a very different kind of phenomenology. An emotion feels different from any kind of visual experience. So I use phenomenology just as a way of thinking about what a scientific consciousness should explain. What I think a scientific consciousness should explain are phenomenological properties, properties of experience, why redness is the way it is in our experience and is different from an

emotion or is different from a smell or a memory. That's what I mean by phenomenology. And when you say explain, part of the problem with consciousness, or at least people who are less scientifically minded, who are disinclined to believe that consciousness can be studied materialistically, I think part of the issue is what it means to explain consciousness, what it means to explain any phenomenon. So when you say we're trying to get an account of consciousness, do you just mean correlates? It's a very good question. I think there's a lot of

Ambiguity and confusion about what explanation means mean whole books have been written about what explanation means in science There's a beautiful book by Carl Kraver called explanation in neuroscience, which is anything in neuroscience What does it what does it mean to have an explanation of it? The way I think about it is an explanation in consciousness science does not have to solve the whole problem that we come up with this intuitive eureka moment about now I understand how the electrochemical pate inside my skull gives rise to the redness of red. That would be nice, but we can have explanations that are relevant without that kind of big ticket prize. So by explanation, I mean more than correlation. And that's key.

And that's key. So it's already a relatively standard method in neuroscience to look at the brain and see what changes in the brain as different aspects of consciousness change. What happens in the brain when you fall asleep or when you go under general anesthesia? Or what happens in the brain when you consciously perceive a scene versus you still respond to it but you're not conscious of it. So this is a standard method which is looking for the neural correlates of consciousness. But just finding correlations is not sufficient for explanations. Correlations provide constraints on explanations but they don't provide explanations themselves. To explain something, what you want to have is, in some way, the correlation is no longer arbitrary. You want to have a sense that, ah, yeah, I see why that correlation obtains,

because it explains something about the thing it correlates with. So, which for me means that the right way to think about this in terms of the brain and consciousness is is not going to be so much this part of the brain is correlated with consciousness and this part isn't it's going to be at a sort of intermediate level so consciousness correlates with this kind of process these kinds of dynamics in the brain because they have similar properties that the conscious experience has. So we need a more kind of finessed bridge between what's happening in the brain and what's happening in experience. And if I can come up with something in the brain that tells me that's why consciousness in this case is the way it is and not some

other way, then that's an explanation. In science in general, we judge the success of science, a scientific field, usually if it can explain something, if it can predict when things are going to happen, and if you can control those things. Explanation, prediction, and control are good criteria for the success of science. And I think we should apply those same criteria in this case too. This doesn't mean that it has to be intuitive. Quantum physics is extremely counterintuitive, but in terms of explaining phenomena, predicting and controlling, it's extremely good. So this leaves open that we might come up with a good theory of consciousness

and it might do the job of explanation prediction and control but we might never get the feeling that like oh yeah that's right now it all makes sense it just might not happen don't know yet let's take quantum mechanics as an example. The way that I view it is it's more instrumental. People who are studying quantum physics, they're hesitant to say that it's an explanation. They're more apt to say it's a model and it provides some predictions and it's an operationalist approach. Some people might take a more philosophical bend and say, you know, this is an explanation. But truly, truly, I don't, it's difficult to be a scientist and not be an instrumentalist. And if you're an instrumentalist, you're not, it depends on what you mean by explain. Like,

is the fact that you can write equations that predict a certain number and then you find it in the data, does that mean that you've explained it? Well, I don't know. I don't know if that matches what people consider to be an explanation. But explanation is a tricky, like you said, it's a tricky word to define. Okay, now how's this related to the real problem? You mentioned that there's a hard problem, there's the easy problem, and then there's the real problem. That's your anti. So I'm curious, is what you just described the real problem? And how does that delineate from the hard problem? Yeah, that's basically it. So again, this is a little bit fast and loose with these terms as they're used in philosophy so the hard problem has has been one of the classic go-tos in people thinking about consciousness

from a philosophical or neuroscientific perspective it's due it's uh due primarily to david chalmers and the austral philosopher. And he describes the hard problem of consciousness. It's this metaphysical problem, really. It's the idea that even if we had a complete description of how the brain as a mechanism worked, how it transformed sensory signals into actions, into our behavior, into actions, into our behavior. If we understood everything about the brain as a physical chemical mechanism,

there would still be a mystery about why and how any of that activity should give rise to a conscious experience. Very broadly, why is consciousness part of our universe at all? How does it relate to the operation of physical stuff? That's the hard problem. And it is a very metaphysically loaded problem.

It's really a question about the capabilities and limitations of physical explanations, explanations in terms of mechanisms. and limitations of physical explanations, explanations in terms of mechanisms. And it articulates the intuition that consciousness is somehow beyond the limits of physicalist and mechanistic explanation. Now, there's all this debate

in philosophy and neuroscience about whether the hard problem exists, whether it's an illusion, or whether it can in fact be solved. I put my cards on the table will want to remain agnostic about that. I'm not sure whether the hard problem is a real problem that can or cannot be solved. solved. But what I do think is that we can still ask questions about consciousness without sweeping it under the carpet entirely. So the easy problems are, again, in Chalmers' distinction, these are the problems that don't really appeal to consciousness at all. They're questions about how the brain works as a mechanism, how it can explain perception, sure, but any kind of perception, action, attention, behavior, language.

But we can answer all the easy questions without referring to consciousness. Now, there's, I think, an enormously rich middle ground, which is to appreciate that conscious experiences exist, let's assume that they do, and we have experiences of colour, of shape, of emotion, all sorts of things. And we can start to try to explain the properties of these conscious experiences in terms of mechanisms. This is exactly what we were talking about. Can I explain why a visual experience is the way it is and is different from, let's say, an emotional experience in terms of

brain mechanisms? And if I can do that, I think I'm making progress on the scientific explanations of consciousness, even if I'm not going directly after the heart problem. Now, other people have expressed pretty similar ideas. Chalmers himself called this approach something like the mapping problem, how to build non-arbitrary correlations between things in the brain and things in the world of our experience. Other people have talked about neurophenomenology. This goes back to Francisco Varela in particular.

It's just pretty much the same thing. Again, how do we link things happening in the brain to things happening in our experience? So there's a tradition of thinking this way. Do you see that the... Okay, Faraz. Yeah. Sorry, I just wanted to talk about the role of perception in all this

because I find the way you have described in the past the perception's relationship with reality quite fascinating. You have previously said before that famously, in a way, you have said that normal perception is controlled hallucinations. Now, this idea of hallucination or illusion, I suppose, first of all, do you distinguish between these two words, hallucination and illusion, or can we use them interchangeably? Maybe you can answer this and then I have a follow-up question to this. Sure. I mean, well, let me first say that I think it's a lovely phrase, perception is controlled hallucination. I certainly didn't come up with it. I heard it from one of my mentors, Chris Frithith who heard it somewhere else and and actually a few of us we've tried to find out the proper origin of it and I've never been able to it sort of there's

this legendary seminar UCSD sometime in 1990 which people refer to but it was never recorded and it's just so who knows anyway the phrase does the rounds but rounds, but I think it's a good phrase. And we'll get on to hallucinations in more detail, I'm sure, in a bit. I'm not sure I would use it interchangeably with illusion. Illusion, I think, is a very problematic word, too, because illusion sort of implies that there is, in the context of perception, that there is a non-illusory perception, that there's such a thing one might call veridical perception, which is the sort of, we're seeing the world as it really is. And when we're hallucinating or experiencing an illusion, then we're not seeing the world as it really is.

then we're not seeing the world as it really is. But my view on this is we never see the world as it really is. Everything that we perceive or consciously experience in our perceptual work is a construction that's only indirectly related to what's actually there. to what's actually there. It's just that the degree to which these perceptions are constrained by what's actually there differs. So I think illusion for me, and I'm thinking about this now really for the first time, how I would distinguish it. Illusion is probably a very systematic relationship between what's there and what we perceive that is not necessarily maladaptive it's it can often be just useful it's sort of a quirk of the visual system where a normal way of constructing our perceptions is revealed to be quirky in some way um and then

hallucination is something a little bit more unusual. Hallucination usually means that I'm having an experience that you're not or that other people aren't. And it is no longer adaptive in that same way, no longer necessarily so useful. But I wouldn't say it's a hard and fast distinction. If you look at, for instance, if we look at fluffy clouds and see faces in clouds, then this is probably something many of us do. And you could equally call that an illusion

or a hallucination. And strictly the word for that is the Greek term pareidolia, seeing patterns in things. So I think there's an interesting sort of landscape of these terms. Right. And then to follow up to this idea, I mean, the general population might think that, for example, illusions are in a way negative limitations that might prevent us from reaching the supposedly noble goal of, for example, searching for the truth, let's say right but um you know you and others such as let's say donald hoffman you've brilliantly argued that illusions are actually could be in a way a gift for example in your in your talk about consciousness you bring up the adelson's checkerboard illusion phenomenon uh and in a way my question with this would be that if we have pragmatically evolved in a way not to experience reality in its true form,

you know, it's about fitness in a lot of ways from an evolutionary perspective, and are better off if truth is, in a way, evolutionarily or perceptively driven toward extinction, then why should we want to get to the root of reality as a whole? Isn't ignorance bliss in this specific case? I mean, there's a lot of loaded phrases there about truth and, of course, truth matters. And there are things that we can't perceive, which make a huge difference to us. If we could perceive coronavirus particles, I'm sure people would be much more willing to wear masks. And if you go and jump in front of a bus, whether or not you're looking at it, it's going to hurt. So I'm certainly not

being anti-realist about this. There's a real world out there. We don't know quite what it is. Best ask a physicist for that, what the ultimate nature of reality is. But it's certainly true that whatever it is, we don't have and can never have direct access to it through perception. I mean, this is a very old point. Immanuel Kant, I think, says it. Plato said it, but Kant especially talked about the noumenon, the sort of the external objective reality that's forever hidden behind a sensory veil. So in a sense, it's an elusive goal to think that we could ever perceive the true world in the sense of having direct perceptual access to physical reality. But what we perceive is constrained by that reality. That's what perception is all about. It's a way of constructing a world for us that is usefully constrained by whatever's really out there. And as you said, and Don Hoffman

has said, a whole bunch of people have said this really, that as soon as you appreciate that our perceptual experience is not a direct reflection of an external objective reality then it's quite easy to to think that well what we perceive is going to be tuned by evolution to what best keeps us alive and i think there's there's a really deep truth to that and you can think about it in one of the very simple examples like color we experience colors as usually we experience them as really existing properties of objects in the world color seems to be a property of an external objective reality but we know that it isn't and we don't even need neuroscience for that i mean newton has shown that color is is not a thing in the physical universe it's there's this whole continuous spectrum

any part of it that we were able to that can shape our experience at all you know we create this universe of colors from the fact that our eyes are responsive to roughly three different wavelengths but we see way more than three colors, which for me is fascinating, because in one sense, we are perceiving much less than what's actually out there, because we don't experience x-rays or infrared or ultraviolet at all. But in another sense, we're experiencing much more than what's out there, because we create a universe of colours just from combinations of three wavelengths. and the brain does this because colors are very useful to guide our interaction with the world they're a way for the brain in this case to keep track of surfaces when lighting conditions change and to distinguish between different kinds of surfaces.

It's a very, very useful thing for the brain to figure out what to do. So I don't think of it as, in this case, it's not like we should, the goal should be to not experience colors anymore and experience just those three wavelengths of light. I mean, no. Even though that's closer to objective reality in some way, that seems like an odd goal to have. I think the interesting thing is just to understand much more about this relation. How do we construct the universe of color from a very limited sampling of physical reality? And why is that a useful thing for the organism to do?

And why is that a useful thing for the organism to do? I mean, one example that comes to my mind regarding this was the idea that, for example, you have an organism that might be tuned to fitness in terms of understanding small and large quantities of some resource as, let's say, one color, let's say red, and then anything that is in the middle as another color. color, let's say red, and then anything that is in the middle as another color. So for that specific organism, it doesn't really matter whether that resource is in a large or small quantity. That's why it sees it as the same color. And what matters is whether it's in the exact amount that the organism needs. So in a way, evolutionarily, the organism uh to not distinguish between small and large amount of a quantity but and and that leads to evolution in a way hiding the the truth that is that is behind

it so that that idea just yeah i think there's something right about that and you know i know various people don often as well use this idea idea of the metaphor of the desktop, right? The desktop on your computer is useful because it hides all the stuff going on in the operating system that is not useful for the user of the computer to see. We just want to drag a file into a folder and be unaware of what's actually happening to make that happen. And so indeed, I think there's every reason to think that perceptual systems are also

built that way. Let's take this Hoffmanian argument a little further. So the argument is that there's no relationship between, or there's little relationship between truth and fitness, and we're tuned for fitness, given that we we're not seeing reality or at least we can't claim to now given that how can we ever make any claim that there is an objective world or what the characteristics of this objective world is like now you made a claim in one of your talks about misperceiving yourself that some mental disorders can be seen as you have the wrong idea about yourself How can you have a wrong idea that to me implies objective, but it's difficult to know the objective without making reference to adaptiveness

Okay, so I think I Don't think anybody's saying there's no relationship between Truth and fitness or perception. It's just not they're not they're not the same thing so fitness or perception it's just not they're not they're not the same thing so uh but there's going to be some overlap like if my if my perceptual system reliably represents things as far away when they're close and i try and cross the road i'm going to get run over and that's that's and so that's that's not a good way for perceptual system to be organized so that's not a good way for a perceptual system to be organized. So there is a systematic constraint between what's out there and what I perceive.

Yeah, I'm not sure if that's true. Let me just rebut, because there are ways that you can form graphs. There's something called quantum graphs. I believe it's used in loop quantum gravity. But regardless, where you can represent what's far as close and what's close as far. And this, actually, Hoffman does make this argument that there is virtually no relationship between truth and fitness. One of the ways he does it, I have my objections, and anyone who's listening can watch the podcast.

One of the ways that he does this is by saying, let's imagine that the universe, that the objective world has some property of structure. Let's say it's a group structure. Let's say it's a partially ordered set, whatever it may be. Then let's imagine the totality of the ways that you can perceive. You find that as the complexity of the system increases, the amount that conforms with reality. Now, in this example I gave,

it's a group or a partially ordered set, but there are various other structures. The amount of ways that you conform with reality divided by the total amount of ways you can experience it goes to zero. So he may argue, now he's not here, and I could also be misrepresenting him, but he may argue that there is virtually zero relationship between what's out there and how we perceive it. And in fact, okay, sorry, please. I mean, again, we probably would need him to see exactly where he's planting his flag. I would I'm also of the view that, and we've said this earlier, that I can never

perceive what is actually there, that the nature of the physical universe, assuming there is one which is still up for grabs, right that is something that is not directly accessible, can never be directly, perceptual mechanisms just don't get that,

that's not what they can do. So in that sense, there will always be a gap between our perception and what is there. However, precisely because our perceptual mechanisms have been shaped by evolution, they will have been shaped so as to be adaptively constrained by the regularities in the physical universe that are important for our survival, which is precisely why, if I'm crossing the road,

for our survival, which is precisely why, if I'm crossing the road, I will see things as far away when they are far away, and as close by when they're close by. And the other thing, useful distinction here, I think, goes back again in philosophy to Locke, one of the British empiricists who distinguished between primary and secondary qualities. And a primary quality for Locke is something that exists independently of a mind or of an observer. So something like having solidity and moving are properties that exist independently of a mind or an observer, which is why if I stand in front of an onrushing bus, it's not going to go well. But then there are other properties like colour, which are for Locke secondary qualities, because colour requires the interaction between an observer or a mind and the physical universe.

Cezanne also talked about this. He said, colour is the place where the brain and the universe meet. And so there's at least two ways of thinking about the kinds of things that are. Two ways of, a way of partitioning our ontology. And they're both our ontology. But in neither case is our experience identical to whatever its hidden causes are in the world, whether it's a primary quality or a secondary quality. We're always constructing on the basis of those qualities or properties we're constructing our experience it's just that for a primary property for a primary thing like solidity

it's going to exist whether we are conscious of it or not and but our perceptual mechanism will still be constrained by it you know we will we A well-designed perceptual system will be constrained in ways that makes us behave appropriately to these things. Right. I don't want to remain stuck on this, but I guess my objection is the notion of farness and nearness is dependent on space, and space-time itself, this is something Donald Hoffman likes to reference. There's a physicist named Nima Armani, Hamed, I believe, but many physicists, almost every single physicist, believes that space-time is not fundamental, which means that location it is you may say that well we're constrained constrained is not the same as and you're definitely admitting that there's not a one-to-one relationship between reality and then how we perceive it but then even in these examples of solidity well what's solid in quantum mechanics where the wave function spreads to the universe. Sure, but there's something... Okay, but firstly I think actually you bring up a very interesting point. We're getting very deep into the weeds here, but for me one of the most interesting

things about this way of thinking about perception is how far you can take it. So for instance, yeah, I can perceive a colour as and I can understand the perception of a coloured surface in terms of an invariance in how a surface reflects light under different lighting conditions and so on. But then what about what you might call the deeper structure of perceptual experience? structure of perceptual experience. We mentioned this right at the beginning as well, the fact that visual experience is spatial, that we overall tend to experience the world in this 3 plus 1 dimensional way, in three spatial dimensions and one time dimension. How is that, can we understand that as constructed in the same way that we would understand the experience of colour

as being constructed? know is our brain generating this this architecture of experience right through the same principles um because yeah from a from a more physics-based perspective you you might say well actually this three plus one dimensional picture is not the most accurate picture of the physical universe. And we could argue and I would not have an expert view about what is. But I know there are many options out there. Even relativity tells you that the three plus one distinction is not as solid as our perceptual world would lead us to imagine. And then, of course, you've got string theory and all some bock universes and all sorts of stuff um but in a sense you can you can be agnostic about what that ultimate reality is if there is such a thing and still ask the question of okay so what's

constructing how do we experience time and space not just objects within time and space and i think for me that's that's a really exciting direction to push this kind of research. Speaking of the perception of time, I remember you said, this is obviously known, that the more adrenaline you have in your system, let's say it peaks, you feel like time has slowed down. I said that I mean that you certainly that there's just this kind of popular assumption about um time slowing down under various sorts of high arousal situations right and actually time here is something that I've been working on just for the last uh four years or so and it's really due to I had this one of my colleagues Warwick Roseboom an Australian cognitive scientist to join join the group in 2015 and brought a deep interest and a wealth of expertise on time perception. So he's been leading our work there.

And one of the, I think, really fascinating lines of argument that he's making, and I agree with him, it seems right, is to push back against this idea that we experience time in virtue of having a little clock inside the head. A lot of work in this area boils down to an assumption like that, that time unfolds in the world. And certainly when we're making judgments about time, we often get it wrong. If we measure an objective duration and then we ask people how long that duration was, people can be wrong about it. And they're systematically wrong in the same way they're systematically wrong about many other things in perception. For instance, there's a very general law in perceptual psychophysics where people are biased towards the mean of the things they they're judging perceptually if you you know if you like see a

whole load of circles of different size and then you're asked to judge the size of the next circle you'll be biased towards the mean of the circles that you've already seen it's called regression to the mean and in time perception you do exactly the same if you if you experience a load of different durations let's say just beeps of different length and you do exactly the same. If you experience a load of different durations, let's say just beeps of different lengths, and you're judging the duration of each beep, you'll be biased towards the mean. And this is a very useful signature that there's something like Bayesian inference going on here, that your perceptual judgment about a situation is based not only on the sensory data, but on your prior expectations about what's going on. Bayesian

inference is simply this way of combining what you already know with what new information you have to come up with the best guess. When you say that we're biased toward the mean, do you mean that of what we experience first? So it's like we take a sample of our first experience and take the mean of that and then our later experiences are then biased toward that mean? Or do you mean we're biased toward the totality of the experience's means? Well, this is a good empirical question, right? It might depend. So it probably depends on the details of the experiment over what time scale the brain is taking the mean. And it's going to vary by task, by context, and so on. In this case, I was really talking about fairly short time scales where you can show that you're just on the last one or two minutes of what's going on, then you'll be biased

in your next decision towards the mean of that. But then you could make the argument, although it's much harder to demonstrate empirically, that, for instance, maybe my experience of duration in general is biased by the mean of every experience that I've had and my ancestors have had and so on. But of course, I can't manipulate that in the lab. So in the lab, I can only manipulate context over relatively short time scales and see what the effect is. But this is a very circuitous route to this idea of time slowing down. Firstly, for me, there's a very interesting distinction here between perception of duration

and the experience of the flow of time. That's interesting. So there's a difference between those two. Yeah, I think there is. We can perceive something as lasting a certain duration, but the experience of just how fast time is going is different. It's different in this, again, in perception in general, we can separate the change of perception from the perception of change.

So I can experience my perception as having changed, but then the experience of the change is itself a kind of perception. There are various phenomena, things like change blindness, which show this very nicely, that if something, if I'm watching an image which I'm assuming is not changing but actually maybe the background colour changes very slowly, you'll find that people won't notice that change. And sometimes people say there's a paradox here because they didn't notice that change but they're now looking at a different

thing so how does that work and i think that the resolution is to say well no their perception has changed but they didn't perceive the change because change of perception is not the same as perception of change so i think the similar thing plays out in time as well that perception of duration is not the same thing as the perceiving time flowing um and it's much harder to to get an empirical handle on the flow of time my favorite experiment here i don't know if you've come across this by uh a guy called david eagleman and and ches stetson i think was his researcher at the time And so they had this idea that if indeed my perception of time is slowing down at moments of high arousal, so let's go back to that. How can I test that? then maybe if I do have some sort of internal clock that's the ticking of which is underwriting this perception then maybe there are more ticks of the clock going on in a particular objective

duration so then I would experience time as dilated and so that was the reasoning and so to test it he had this beautiful experiment where he gave people watches that were specially designed so they would flash numbers. But so quickly that if you just looked at them now, you'd just see a blur. You wouldn't be able to decipher what the numbers or the letters were. And so he then had people bungee jump while looking at this watch several times. It's a great experiment. Look at this watch several times. Go to Crane, go to Crane, look at the watch. And the hypothesis was, well, if subjective time is slowing down because an inner clock is increasing its rate, its frame rate, let's say,

then maybe these people should be able to read what's being flashed. And then report back. Yeah. What do you think happened? I would imagine they did. Right. No, they didn't. No. So it was a really interesting failed experiment. I mean, it wasn't failed because it came, the evidence was pretty clear. But it was, it sort of ruled out that simple explanation that yeah, this clock is speeding up. Because people subjectively still had experienced dilation of time,

but it didn't seem to be due to this acceleration of an internal clock. And of course, I think the reason for that, following my colleague Warwick, is that because there isn't an internal clock. So that's super interesting to me. Because to me, it sounds like what you're saying is we can be wrong about our own perceptions For example, okay, or I'm posing this as a question. Can I say I see this cup is red I see sorry, let's look at this. I see this as blue So you're wrong, I see it as blue. I thought it was red

There are two cups here. Okay, I see this cup as red. I'm sorry. I see this cup as blue. Could I be wrong about my own perceptions? I think I see it as blue. But is it possible, like in the example of these people who say that time has slowed down, but they're wrong about that. Can I be wrong about seeing this as blue? Now, in fact, you know, with our shared reality, you may be able to objectively measure it as blue.

Can I be wrong about my perceptions? I think there's at least two very interesting responses to that. Philosophers often talk about the ineffability of conscious experiences, this idea that we can't be wrong. If I'm in sharp pain now now that's just a fact and i can't be wrong about it there's when i think about claims like that the first thing is one problem is is a sort of epistemological problem like how would you know when the claim is about a subjective experience what's the right way to validate that against so does it even make sense as a as a question being wrong about your your own experience

um but what might it mean it might mean that you are wrong in your interpretation about it and i think that's usually what people mean when they say that they're wrong about their perception. They might say that, for instance, I experienced time as slowing down, but then if you actually look at how they estimated the duration of something it may be basically the same as in another case so they are you know right in the sense that they're giving a sort of face validity you know what they're experiencing but all the ways one might interpret that

might be wrong. Take another very illustrative example. People for thousands of years have reported out-of-body experiences. They'll have experiences where they're floating above their physical body or floating around a roof or going out the window, whatever. Are they wrong about their experiences? Well, in one sense, no. If that's the experience they're having

and they're not just sort of confabulating it later on, then that's the experience they're having. But if they interpret that experience as evidence that their self has in fact left their skull and gone flying around, then they're wrong. That's not the right way to interpret those kinds of experiences. Or rather, there are plenty of other options by which you might interpret those kinds of experiences. And you can also be wrong just in the small space that opens up between the moment of your experience and the moment of your report. And if I tell you what I'm experiencing, I have to, I have to remember it, even if it's only for a very short time. And you could be misremembering. You can systematically misremember stuff.

Um, there are, yeah, there are many ways where we can introduce introduce indirectness between your experience and your report of that experience so you can be wrong in that sense too i think but there is some there is something for me that that it almost doesn't make sense to ask the question can i be wrong about my experience in the here and now bracketing all these things about interpretation and report, because there's no criteria to judge it against apart from the experience itself. Yeah, Faraz. If I may, because this is one of the questions I had in mind, which ties really nicely into this.

You most likely know about Jill Bolte bolte taylor's famous ted talk the stroke of insight the one where she goes into detail about her experience of having a hemorrhagic stroke on the left side of her brain and then she experiences these feelings of euphoria and she goes into the state of nirvana because the way she's describing it, she thinks the right side of her brain is sort of taking over. And then once her left brain comes back again, then suddenly she comes back to her version of reality that we experience. And she realizes, okay, something's wrong. I have to go get some help. And that talk generally was really fascinating to me.

But I just wanted to know your thoughts on that kind of experience and maybe we can follow up with the idea of from a neuroscientific perspective the left and right brain dichotomy that is created and the way she describes it now her talk was 12 years ago

I think it was around 2008 and maybe since then have we had any advancements any any new discoveries that sort of shed light on the on the differences between these two well i suppose my question is twofold one is what are your thoughts on on the on the way she was describing that out-of-body experience in a way you could call it an out-of-body experience and and secondly the distinction between left and right brain. Yeah, I mean, I don't have too much to say about the first one, about the sort of experience of bliss or nirvana.

So it's, you know, people, again, people do report these kinds of experiences in various conditions. There are examples, what comes to my mind, are these examples of so-called ecstatic seizures so some people with with epilepsy uh experience reliably these moments of bliss and awe and presence of the numinous and all sorts of you know wonderful experiences um either in the in the penumbra or the periphery of their seizures or maybe even during the seizures, I'm not entirely sure, to the extent that sometimes they

don't want to be medicated, they don't want this to be relieved from their seizures because they get this from them. Quite what's going on for these sorts of experiences is extremely hard to say, I think. And part of the problem is it's been extremely difficult to investigate them in controlled conditions in the lab where you can actually see, okay, this is what's going on. Is it a wash of a particular kind of neurochemical

or activity in this or that part of the brain it's unclear so some of the stuff from epilepsy is it shows that these some brain regions seem to be more implicated than others but but i wouldn't i wouldn't want to put too much stock in those uh but it's a fascinating class of experience sure and i think again there's a lot of you know certainly a lot of interest um again, there's a lot of, certainly a lot of interest, and there has been a lot of activity for ages, about the neuroscience of meditative experiences. It doesn't have quite the same end point,

but certainly there are examples of people in meditative states who will experience the dissolution of the self and the ego and so it's kind of fascinating to see back to where we started so what in the brain can explain that kind of phenomenology, that kind of dissolution. An open question. The second thing about laterality in the brain left versus right hemisphere and this is it's a huge it's a as other than most people will know we hear these tropes and these cultural tropes about the left brain being analytical and the right brain being

holistic and artistic and left-braining and as with all these things there's a grain of truth underlying this but it's only a grain and um pretty much for everything you do the both hemispheres of the brain are working together some things are more lateralized than others and so this is a few of these things a couple of these things when you get more towards output and input, things tend to get more lateralized. So if you have a stroke in one half of your visual cortex, that will affect one half of your visual field. If you have a stroke in your right hemisphere, your motor cortex will be paralyzed on your left hand side so these things are very lateralized um and that's not particularly mysterious thing to say then when you get into sort of more uh less obviously

lateralizable things and language seems to be basically the only thing that's reliably lateralized in terms of brain function in in humans which is usually on the left, but not always. So most of the time it's working together. I think the super interesting studies here... So I don't really like very much this habit of describing people as left-brained or right-brained, and I think that's some kind of neo-phrenology stuff um but what's still really interesting are these these old case series these neurosurgical case series of people with what we call calisotomies these are people who've had split brain operations um this used to be done uh it's pretty rare now because other medications become more effective, but

for people with really, really severe and intractable epilepsy, one of the options was to cut all the majority of the fibers connecting the two hemispheres, hence the term split brain. And there's some absolutely fascinating observations. This goes back to people like Mike Gazzaniga and Joe Bogan, who would give the two hemispheres different input to try and ask the question, are there two separate conscious subjects now, now that you've split the brain like this? And there are very suggestive hints

that there's something weird going on like you can you can show uh you have to always get my left right mixed up because so many things cross over here it's very hard to keep keep straight about it but if if i show my hold on this has to be my right visual right visual cortex anyway if information gets into my right visual cortex that doesn't get into my left then typically the person is not able to

verbally say what that is because the left language side doesn't have access but will maybe try to confabulate what's going on so the right hand side might still be able to draw what they see and then the left hand side will make up some story to justify why the right hand has drawn it and so these are these are really really super suggestive but it's still never been entirely clear this basic question of have you divided one person into two or not? Because

a lot of the early studies turned out to have been methodologically limited in various ways, so now with more modern methods in the few patients that are still around, perception does seem to be more integrated across the visual field than we would have thought. But to me, it's still a fascinating case about could we divide a single conscious person into two conscious people? And what would it be like to be one or other half? Just to clarify one point then about it the idea that for example the right brain

uh as i mean she was describing in this way where she sort of when the right brain was coming into into effect she was she felt like she was connected to everything around her she no longer felt like an individual she felt like she she was everything and she describes it as the we uh we sort of experience and then when the left side comes back in that's where the individual self comes back on now this definitely sounds like a really interesting analogy dichotomy that sort of thing but i just was wondering from a neuroscientific perspective how much truth um exists here and and whether we have any evidence of this I versus we sort of distinction. Well, I think there's certainly, there is a distinction between, or rather the experience of being an individual

separate from others and from the world is something that varies. And it varies for each of us at times if you're fully immersed in something you tend to experience yourself as less separate from it and people call this sometimes by the by the term flow experience um if you're in a crowd dancing or a football game the crane you feel part of the crowd and that's not just a metaphor, it picks out I think a valid aspect of your experience of reduced individuality at that time. People on LSD, that's also one of the classic

reports from psychedelic experiences, is a dissolution of the ego and a feeling of connectedness now not just with other people but with nature, with the world and so that and all these things together suggest that well there's a brain basis for the strength of our experience of individuality but I don't think it comes down to simply right right brain versus brain that's all i see and just just so that i can go back to one of your other statements that i found really interesting you you mentioned if i'm please

correct me if i'm wrong you mentioned that uh an object would exist regardless of whether we perceive it or not and i believe i believe that's what you said i'm wondering how this ties to the i to john wheeler's ideas regarding uh for example he makes he makes an assertion in the sense that um a conscious observer is a necessity in order for us to declare that something exists without the conscious observer it's like part of the formula that without it if we remove it we cannot make any claims whatsoever about whether something exists or not so how does how does the idea that you brought up tie into that do you agree or disagree with that with that well you said two separate things though which were different So one is what are the conditions

for something to exist and the other was what are our conditions for us making claims that something exists. And I think if it's the second, then observer is probably needed. You can't make an inference about an existence without an inferrer. But whether there are objects, whether objecthood intrinsically requires observer, that's again, I think that's a deeply physical, metaphysical question. So John Wheeler, who you mentioned, I always associate him with this idea of giving information some real status, some ontological status.

He's famous for the slogan, it from bit, that everything that we measure, whether it's using the tools of science and physics or experience, fundamentally our universe is written in the language of information. our universe is written in the language of information. And then from that, then there's all sorts of ways of understanding that kind of claim because some sorts of informational relationships are fundamentally relative. Shannon information, you can think of that way, a sort of depends on a sender and a receiver other ways of thinking don't but to me that that's that's now that's a sort of i'm not trying to dismiss the interest of these i think that they're

fascinating and actually they do play into how we might read certain contemporary theories in consciousness science, in particular things like integrated information theory from Giulio Tononi. Then it makes a big difference what you think information is, what you think it's, if you're with Wheeler or not. But for the most part, I don't think what I was saying about objects, I want to bracket those discussions away from that. And it may be true that in some physics-based way that we can't say anything exists unless there's an observer, but for the story I'm trying to tell, which is just this idea that what we perceive is always a construction,

but underlying that construction, there are hidden causes that we can never directly have access to. I think that stands whatever you think about the ontological status of information you're just looking at it at a different level in a way these are distinct levels that might make sense on one aspect but not on the other that makes sense what are your views on the sapir sapir wharf hypothesis now now it really does feel like an interview like bang bang one hard question another um oh and just okay you can gather your thoughts on that just as an aside because this bothers me i'm not a theoretical physicist but my background is in theoretical physics

when people make claims of consciousness being associated with the observer and quantum mechanics much of that is over exaggerated and taken up by the new age community because they tend to like to use the word consciousness and tend to want it to have some ontological foundational status and it might be that might actually be the case but it's over exaggerated and it's extremely contentious and i believe something like 90 of physicists don't believe that to be the case the way wave function can collapse when there's another electron or another set of electrons. Let's say objective quality, like atoms observing another photon, then you can collapse the photon.

But even that's a bit, that has to be stated more technically because there's something called coherence and decoherence. Either way, take those claims that the conscious observer is necessary for the collapse of a wave function with a grain of salt absolutely and then you guys get to know much more about physics i mean i started off in physics but i i switched to psychology when physics became too hard i'm so i'm super interested in your background man because you also went back

to do a computer science master's and an artificial intelligence PhD, if I'm correct? Yeah, that's right. Okay, that's super cool. Well, we'll talk about that after. So let's talk about Sapir-Whorf. Okay, so Sapir-Whorf hypothesis is this idea that language either determines or shapes perception. that the linguistic environment we live in influences the perceptual environment that we live in, I think is probably one way to say it.

Technically, people will talk about cognitive penetrability here, that is the way we see the world or hear it or smell it, is that invaded by higher level beliefs or cognitive processes of which language is exemplary? So the Sapir-Whorf hypothesis is that yes, that the language that we use shapes the way we perceive it. And this can be in kind of sort of subtle ways, like how do I perceive time? Is that dependent on the way I use language? There are all sorts of lots of interesting anthropological experiments

looking for languages where, for instance, in Western languages, we tend to think of the future as ahead of us and the past as behind us. But I think it's the Aymara, this is something I'm very secondhand on, but there's reports of some indigenous South American cultures for which it's the other way around, languages for which the other way around, that the past is ahead of you and the future is behind you. The reason being kind of makes sense to me is that, well, you know, you can see the past. I know what happened in the past. It's happened. So therefore it's within my line of sight. But the future is unknown. I don't know what's going to happen. So I can't see it. So it's behind me.

But would that linguistic difference, that sort of metaphorically shapes how we talk about time, but might it actually influence the way we perceive time? So there are all sorts of interesting questions about how deep these influences might go. To me, it's sort of a priori that I'd be surprised if the Sapir-Whorf hypothesis is not true. The reason people seem to find it appealing, I think, is because of this idea that there's this modularity in the brain, that we have perception, which is some sort of encapsulated at some lower level. We form this perceptual representation of the world and then we think about it and we talk about it and we make decisions about it.

And then we go and we do stuff. We make a plan for what I'm going to do next. We do stuff and then we go and we do stuff we make a plan for what i'm going to do next we do stuff and then we perceive the world again but the brain doesn't work like that i don't think and and what you were describing was a bottom-up approach well it's sort of more just an approach of of extreme modularity so people like jerry fodor is known for for this view that you know one of the longest running arguments in neuroscience has always been about functional localization to what extent are different things the brain does localize to different brain regions and the pendulum has swung back and forth on this obviously both extremes can be ruled out to start with you can't have a brain where bits different bits do different things and

don't talk to any other bits um and you can there would be no point to that because you'd have no cross communication anyway right exactly and you can almost equally rule out the other extreme although you can't quite rule out conceptually but certainly um empirically it's ruled out very quickly you have damaged different bits of the brain, different things go wrong. But you can still, at a more subtle level, there's still this argument about the degree to which functions, let's say a function like language, is just served by a particular

network or brain region. And so those bits of the brain do language. Yes, they talk to other bits of the brain, but they don't sort of bleed over into those other bits. So computers like this laptop computer is kind of modular in this way. There's a chip that does memory, a chip that's the CPU, a chip that's the IO ports and chip that's the I.O. ports and so on, you can replace these somewhat interchangeably. So it's pretty modular.

So this kind of design is a reasonable way to design systems. Many of our technology is designed this way. But typically biological systems don't end up being designed this way. And if we look, now that we know a bit more about how things like visual perception unfold in the brain, they're not limited to just the visual cortex. So the way I and many others think about perception is that it's this hierarchical set of predictions about the causes of sensory inputs. Many of these happen in visual cortex, but some of them happen outside of visual cortex. So things like language will interact with visual cortex in interpreting,

coming to the best guess of the causes of sensory inputs. And there's a lot of experiments now. There's a great paper by Gary Lupkin, I think he's at University of Madison, Wisconsin, who sort of took this debate more into the realm of experiment and just summarized a load of evidence showing that words influence perception in in pretty clear ways so this to me is not surprising but it's also there's just there's lots of interesting questions about

how far this goes again this is always a question how deep do these influences go what kinds of things are are amenable or influenceable by things like language. It seems like there's a controversy around it when you even just look at the Wikipedia article about the Sapir-Whorf hypothesis. Pinker and Chomsky are both against it being true. And I haven't had a chance to look up their arguments. I'm curious if you could explain to me,

as well as the audience, what are Chomsky's and, to a lesser extent, Pinker's, but I'm more interested in Chomsky's, arguments against it? Yeah, to be honest, it's not something I've studied much either. My feeling is that, I mean, Chomsky's known for his ideas about universal grammar. I mean, he's such an interesting character because he, in many ways, put the nail in the coffin of behaviorism. Sort of saying that, OK, we need to take internal mental processes seriously, and we can't just infer them always just on the basis of behaviors.

We need to assume some complex internal mechanism. Of course, he was doing this in the domain of language, and so you can't explain linguistic behavior purely from a behaviorist perspective on what people say, just how quickly human beings learn language. He would say that there's not enough information in the stimuli that babies are exposed to for them to infer grammar and people still will argue about that. They'll say well actually there is and it just depends on because they're interacting with it and so i'm not a linguist i don't want to come down on one side or other there but chomsky's view became this idea of this sort of language

acquisition device this very very circumscribed system in the brain that was for language and so i think if you take that perspective you are you're going to be disinclined to the sapir-wolf hypothesis which is really motivated by an idea that there isn't this kind of modularity in the brain that things there is to some extent there's specialization, but there's not a complete kind of encapsulation, firewalling of one kind of process from another. Is there a difference? Sorry, go on. reminded me of was the i believe there was a study that they conducted with regards to color color detection uh being tied to uh if if if in the language that that individual spoke there was there were there was act there were actual words that would define a certain color then then that individual was capable of identifying that specific color but if a language was was not um it did not have that specific color um like just it hadn't been developed in that language and

that person would not be able to to detect it so i'm just wondering what this is sort of a piece of evidence right that yeah i mean that's sort of connected yeah that's a of a piece of evidence, right? Yeah, that's another piece of evidence in favor of Sapir-Whorf, right? Exactly. And I think the example that comes to mind when you talk about that, I think it's Russian, right? And I don't speak Russian, but I believe Russian has more words for blue than English. Maybe, maybe.

And colors, you can test this really nicely with colour because colour is this beautiful example of categorical perception. You have a continuous wavelength that we typically experience as a number of discrete colours with bits of blurry boundary. What happens when you look at a rainbow? I mean, that's what you see, even though the colour is, the spectrum is continuously changing. So to assess categorical perception, you sort of ask, you present people with differences on this continuous scale and ask them to judge how far apart they are. And if their judgement changes quite a lot at some times, but not other times, then you infer, oh, there's a category boundary there um and so yeah i think this is again for me it's a bit second hand but i'm prepared to accept that people who

who learn russian from birth i don't know about if you learn it as a second language may divide their experiences of blue into more categories than people who have a different linguistic environment. Against me, it's like, okay, this is kind of fine. I don't understand. For me, it shouldn't be a controversial perspective to adopt because I think this is how perception works in general. We also notice for things like, let's say, wine tasting. notice for things like let's say wine tasting you initially you all wines all red wines basically taste the same and then as you learn you sort of learn to distinguish them but then you learn to

apply different labels to different kinds of wines and then those help scaffold your ability to distinguish them still further what the problem is you've got a sort of chicken and egg thing here you know what comes first is it that you experience things differently and then label them? Or can it be the other way around? And with many of these things, I think it's probably not a good idea to seek an answer to that. It's a circular kind of causality that bootstraps these differences into existence. Do you mind repeating that last part why is it that you can't start with the concept and then be able to perceive that concept afterward so for example i don't know the difference between cyan and turquoise and magenta and purple and and their

cognates do you know it just it just looks like blue and some blue and i know green and red and so on and i could take 10 seconds to look it up and then probably solve it for the rest of my life but either way i see that as bluish green i see that as purple am i seeing the same as someone else but just not noticing and is that between seeing and noticing yeah yeah okay that's good i mean in that particular case i don't know we'd have to do experiments to know that right we'd have to we What we'd have to do is give these different colours to you and present them in sort of pairs so you make judgements about how similar they are to each other. And then we'd have to see, are you making reliable distinctions

between these colours that you linguistically describe as the same compared to somebody who has different words for these different colors. So I think these things all boil down to empirical questions. Now, here's one test. This might not be doable, but imagine you look at a scene

and let's say it has some discriminable phenomenon that are indistinct at first because you don't have the language label for it so cyan and turquoise for me for example and you show that to me and i just oh that's bluish or this bluish green then a day later you teach me the phenomenon of cyan and turquoise and i'm able to to discern then you put me under hypnosis to remember to remember yesterday and you say was cyan on the left or the right in other words i'm wondering because i don't notice it initially do i also not see it does it not get recorded or does it get recorded so that i can go back later now with my proper disambiguations and disambiguate okay that's interesting

I mean I like the experiment design part and hypnosis bit I think there's it knows whichever method choose whichever method allows you to get a better nose is an accurate emery it's hypnosis is real but it's not what most people think it is it's not sort of it's not this kind of stage I wave a watch in front of you and you can remember your child anyway, but I think there's, so I can indeed imagine this experiment where you make a perceptual judgment I then give you some additional knowledge you make the judgment again and then you just try and remember

what you did the first time and then you try and you just try and remember what what you did the first time and can you be more accurate uh when you're retrospectively making a judgment given now some additional knowledge than you were at the time i i there's almost certainly stuff like that has been done i i don't have it to the tip of my tongue. But yeah, that's an interesting design that could certainly address the question of... well, I don't know quite what question it would be addressing, because there's problems, there's many things being conflated here, that sort of memory along with perception at the time. There's all these things, it really depends on what the specific question is you're trying to ask.

But I think it's pretty clear. There are lots of experiments that show indeed that expectations, what you expect to happen, can directly influence perception. These expectations can be set up by linguistic phrases. So I think there's lots of good evidence that what we think, what we say, the structure of our thoughts and our language can indeed influence how we see the world. Is there a limitation to the Bayesian model of the brain?

I know that it can't be a holistic, it can't explain everything because there's no there's no model that explains everything you'd have the holy grail right there and I know that you are a large proponent of the Bayesian model of how the brain works I'm curious where are its limitations I'll give you one example from my life I was playing a video game and someone in the I was expecting the person I was expecting the person in the video game to say the word extinction. And I was watching and she said something and I thought it was, she said entity instead. But I was expecting extinction.

So then I rewound it and she in fact did say extinction. So I thought that was strange because I expected extinction. She said extinction, but I heard entity. Now these are rare, but when they come up, I do write them down. Another one was I was looking for some object on my desk. Let's say it was a pen. I don't remember what it was. Let's say it was a pen.

It was right here. And I didn't do a cursory glance. I looked for a second, not half a second, but a glance. And I didn't see it. And I was expecting to see it because I'm looking for it. And I looked back, and it was right where I initially looked. So there's some... Okay, you understand.

In other words, what are the limits of the Bayesian model? Am I just nuts? Let's just take a step back and say very briefly what the Bayesian model is. And so Bayes' theory is just this way of reasoning with probabilities. It's this idea of inference to the best explanation.

And so in terms of neuroscience, the idea is that that's what the brain is basically doing all the time, that perception is a best guess about the causes of sensory data. It's not a direct readout. There is sensory data. It's trying to combine prior expectations about the causes of sensory data with the sensory data to come up with the best guess. And all of these things are sort of probabilities. It provides a standard of optimality. Bayes' theorem in mathematics and probability theory is really telling you what the optimal way is to combine information under conditions of uncertainty. under conditions of uncertainty.

And so in one sense, it's not really something that can be right or wrong. It's a standard of optimality. How you use it to think, well, okay, that's what the brain, that's the problem the brain should be solving if we've got the cost function right. If we're assuming the brain is solving the problem we think it is, it tells you the optimal way to do it. That's one way to interpret it. Another way is more literal to say, no, this is

actually what the brain is doing. The pattern, the neural firings, the circuits in the brain, what they're doing are actually the computations needed to implement this kind of probabilistic reasoning. That's a much stronger claim. And... Razor blades are like diving boards. The longer the board, the more the wobble, the more the wobble, the more nicks, cuts, scrapes. A bad shave isn't a blade problem. It's an extension problem. Henson is a family-owned aerospace parts manufacturer that's made parts for the International Space Station and the Mars rover.

Now they're bringing that precision engineering to your shaving experience. By using aerospace-grade CNC machines, Henson makes razors that extend less than the thickness of a human hair. The razor also has built-in channels that evacuates hair and cream, which make clogging virtually impossible. Henson Shaving wants to produce the best razors, not the best razor business. So that means no plastics, no subscriptions, no proprietary blades, and no planned obsolescence. It's also extremely affordable. The Henson razor works with the standard dual edge blades that give you that old school shave with the benefits of this

new school tech. It's time to say no to subscriptions and yes to a razor that'll last you a lifetime. Visit hensonshaving.com slash everything. If you use that code, you'll get two years worth of blades for free. Just make sure to add them to the cart. Plus 100 free blades when you head to h-e-n-s-o-n-s-h-a-v-i-n-g.com slash everything and use the code everything. I think that second claim is probably right, but probably not universally. Universally means across the whole brain yeah yeah i mean that oh across well both i mean i think i think in any given brain it's probably not sensible to describe all the things going on in the brain as involved in

implementing bayesian inference for the simple reason that some things can be pretty hardwired. I mean that's one reason, there'll be other reasons too. But you still can probably interpret it as if it were doing Bayesian inference because you can always tell a story, you can always say oh yeah I can describe these dynamics as they're compatible with the brain doing Bayesian inference but it's a different question from whether it's actually doing it and again this is actually a pretty

controversial thing too that you can find I think there are lots of cases where things will look like exceptions to this rule. And they may well be accepted. A very kind of basic example of this is the phenomenon of adaptation and perception. That, for instance, if I show you a load of lines of that orientation, let's say 45 degrees sloping up left, let's say, and then I show you a line of a different orientation, you will perceive it as further away from what I was showing you before.

This is the opposite of the regression to the mean effect that I was describing earlier. So I wasn't being entirely comprehensive when we were talking about that. Adaptation goes away from the mean. If you go towards the mean, that's like an example of what you would expect under Bayesian inference because you're expecting that, so what you perceive is drawn towards that. So examples of perceptual adaptation seem to be anti-Bayesian, if you like. And so there's then been a lot of discussion about what that

means. I mean, it's just a fact that that's true. I mean, that's what adaptation does and you can sort of understand it in terms of different neural populations getting fatigued or worn out and so across the population that's that's what you see you can still tell a bayesian story about it i forget what it is now but um but there are certainly things that it's not i don't think the bayesian brain hypothesis is this sort of magic solution to to every everything we see but i think it's a very useful perspective at that first level about it's a standard of optimality and i also think that a lot of perception in the brain and action an action can, I think probably is,

the brain implementing and approximating Bayesian inference. The other, just last thing I want to say about that was when you were giving your counter examples of where you expected to see the pen and then it was where you were looking all along or you expected to hear a word. There's a confusion that's often made here because when people use Bayes' theory to talk about

things, they will use words like expectation and belief and prediction and so on. These are not necessarily personal level things. It's not something you as a person need to believe or expect or predict. So if I talk about a Bayesian belief, it makes as much sense to say that I as a person believe that Neil Armstrong landed on the moon as it would make sense to say my visual cortex believes that light always comes from above. A Bayesian belief is a probability distribution. So it's not really...

Which means that it could very well be that your brain is doing Bayesian inference even though you as a person had an expectation that turned out to be incompatible with what you see. Your visual cortex might be believing something different. I see, I see. Can I just interject here a little bit? Is the Bayesian brain model in any way,

can it be tied to mental disorders, like mood disorders, let's say severe depression? Because when we discuss ideas such as optimal, optimal conditions, well, someone who's severely depressed, clearly they're not in an optimal condition. Can the Bayesian model sort of identify what is going wrong here and perhaps alter, even alter that suboptimal conscious experience?

Any studies on that? Well, let's first be careful because i don't want to to imply that uh sort of i mean clearly being depressed for instance is is not a great state to be in so it's suboptimal for the person in terms of their lived experience but i would worry a little bit about about sort of saying that there is a standard of optimal performance of the brain and then deviations from that are wrong and to be avoided. Optimality here simply means a sort of mathematical optimality about

the most rational way of combining sources of information that are uncertain. It can very well be that being perfectly optimal in that way is psychologically suboptimal. In fact, many people would argue that we have better psychological lives and perform better if we are slightly irrationally optimistic about things. So these two things can come apart and there's this, I think there is this general worry, I have this worry about this sort of reification of things like optimality from one language to another and we suddenly start making claims about people's psychological lives and ethics and morals and things like that. There's a big gap.

Having said that, you're absolutely right that this is a good framework for thinking about phenomena in psychiatric conditions, mental illness, other things too. Perhaps the best example, one of the oldest, is in schizophrenia and psychosis and we go back to hallucinations what is a hallucination in the language we've been

talking about now a hallucination can be usefully thought of as a a bayesian inference where your prior belief or your visual cortex's prior belief is unusually strong. So it's overwhelming the sensory data. So the conclusion the brain comes to is guided more by what the brain expects is there than what the physical signals from the world are saying. If you think that every perception is this balance between what we expect and what we get. Again, we had this example earlier of looking at faces in clouds. Everybody sort of has a minor hallucination going on there, partly because our brain comes preloaded with expectations to see faces everywhere. Faces are very important stimuli for us as human beings. So we see faces in clouds quite easily. When people are having hallucinations in, well, perhaps in psychedelia,

but also in psychosis, you can think of that then as, again, their perceptual predictions are overwhelming the sensory data. And I think this is a really powerful approach. Paul Fletcher in Cambridge and Chris Frith in London pioneered this way of thinking many years ago and now people like Phil Corlett at Yale are taking it further because it makes very specific predictions. at Yale are taking it further because it makes very specific predictions. If somebody with psychosis is, if their symptoms are explained by this imbalance, then you could presumably put them into a very limited psychophysical situation where we can test in a very, if you like, boring way what the balance is between their expectations and their perception.

We can basically just, and this experiment Phil Corlett did, was to just associate a dim flash of light with a noise, but then very occasionally leave the flash of light out and then ask people what they perceived and people with psychosis would perceive the flash of light report the flash of light more often when it wasn't there than people without psychosis because you know the interpretation being they're expecting the flash of light and that expectation stronger in these people so it actually generates the experience more often and i really like this because it if you think about how medicine works

you want to get away from defining conditions purely based on the symptoms that people report. And you want to get towards understanding what the underlying mechanisms are. If it used to be people would have fever and that would be their disease. They have fever and maybe a yellow fever or another fever. But that was it. And you treat it by bloodletting or something. But once you understood that fever is a symptom, but there's a mechanism that's giving rise to that fever, and that's an infection.

You treat the infection, the fever goes away. Medicine has always progressed like that, from getting underneath the surface-level symptom description to what the underlying pathology is. And in psychiatry, this is still early days. If you look at the Diagnostic Statistical Manual in psychiatry, it's basically a big list of symptoms. And depending on what cluster of symptoms you have,

that's what you get. But we're beginning to move beyond that now. And this is one example. Okay. Something like psychosis, one of the underlying abnormalities or aberrant processes is Bayesian belief updating in the brain. And that explains the symptoms. But it also means, well, maybe there are more targeted ways to develop prognosis or therapy even. And so I think this is the whole area which is now called by computational psychiatry, coming up with computational models of the symptoms in psychiatric conditions. So psychosis, schizophrenia is one one but then people are also thinking about this in terms

of things like depression as well and also things which which i would rather label as neurodiversity rather than psychiatric disorders but things like autism so conditions that are not psychiatric problems but certainly involve people having different kinds of experiences. Again, what's actually underlying the surface level phenomena here? I can see how depression and psychosis might be related to a miscalibration of the Bayesian priors. What about other psychiatric conditions, like you mentioned autism that one i don't see how it's a it's a mistake in in how they're viewing the world and it seems to me to be more hardwired well this is gonna be clear that i i'm not i don't think it should be described as a psychiatric

condition autism i think it's you know it's it's a it's an example of neurodiversity and um so it's again i think it is a very interesting example there because typically when people are given the diagnosis or label of autism or being autism spectrum it's often when some social difficulties become apparent and so people typically have thought of it in terms of an issue with theory of mind, that there's a difference in autistic people about the way in which they perceive the states of other people's minds. And that's you know that's been for for a long time sort of the general understanding of what autism is uh but then if you look more carefully closely and people have always done this because

about what what um what people with autism report about their lived experience. It's not just about theory of mind or anything like that. There are some very basic perceptual phenomena happening too. There's often this sensation of being overwhelmed by sensory data. Everything seems overwhelming in the sensory world, loud noises are very disturbing, there's a sort of need to control to make inputs more predictable. Things seem more unpredictable. And to me this, and to others, this begins to speak to something going on that can be more usefully articulated in a sort of Bayesian language here

are brains of autistic people they're dealing with uncertainty in a different way they're not adapting to changing probabilities quite so well there's sort of there's a lot of hypotheses out there the basic point is that we can again go much deeper than how these conditions superficially present themselves to understand well there's something more basic going on about how these brains are processing sensory information that then these high-level phenomena are built upon and Back to schizophrenia, actually a lot of the symptoms of schizophrenia become clinically relevant when they become delusions, when people start having false beliefs about paranoia and being controlled by the CIA or something like that. But these false

beliefs can be understood as higher order inferences to explain away unusual perceptions. So it's the perception that comes first, not the belief. You begin to understand that the primary pathology in something like schizophrenia is an alteration in perceptual ability. Only then only then, later on, as the condition progresses, does that become manifest in beliefs, in different kinds of beliefs. Speaking of, well, not speaking of, let's take a tangent, Gödel's incompleteness theorem. I'm curious to know what you see as the applicability

or non-applicability of it with regard to how the brain operates. So someone like Penrose or Lucas would say that it implies that the brain is not purely computational in the way that we imagine computation, let's say in a Turing sense, because of so-and-so. I'm sure you know the argument.

What do you see? What does Gödel's incomcompleteness Theorem have? I don't think it has that much relevance to be honest. I remember reading Penrose's book The Emperor's New Mind about 30 years ago now, I think shortly after it came out, just at the beginning of when I was just about to go to college, when I was just about to go to college, waiting for this enlightenment about why it was going to be relevant. And it's clearly a very important piece of mathematical philosophy, right? It's telling us that there are axioms with any given formal system that cannot be proven to be true within that formal system.

And as I understand it was a sort of big blow against the Hilbert program of trying to bring closure to the mathematical explanations. But quite what that has to say about the brain I've never been convinced that it has that much to say. And this is partly because, you know, I don't think the brain is a computer. And so the fact that... Okay, what do you mean when you say that? Because clearly... Well, again...

Oh, not clearly. It's sort of, yeah, I mean, it does get? Because clearly... Well, again... Oh, not clearly. It's sort of... I mean, it does get a bit complicated because, you know, firstly, quite what exactly is Gödel's theorem saying? It seems to be... Because, of course, the theorem class of computation

is something where there is closure. And... There is closure. So I guess for me, Gödel's theorem, if you apply it to the brain, it may tell you that the brain is not a computer. On the reasoning that there are things that we can understand to be true within a formal system where a proof of that doesn't necessarily exist. This is going way back

because it's been a long time since I've thought about this. But firstly, that doesn't seem that surprising because even if the brain were a computer of this sort, we don't know quite exactly what kind of computer it is. So we don't quite know what sorts of true statements should be not establishable by the kind of computer that the brain might happen to be. But basically I'm kind of happy with the idea that the brain cannot be, does not operate according to these principles of logic, that it doesn't reach true conclusions about truth by combining these logical operations

in a particular way that Gödel's theorem rests on Gödel's theorem is a sort of deduction I think it's a real strong form of logic that's fine does that put you at odds with some of your colleagues? Because as far as I understand, most of them believe that the brain is some computational device that could be wrong, could just have a bias set. No, I think you're right. I think that, I don't know if it sets at

odds, but well, I mean, firstly, I don't think it particularly, the whole point about the Goodell's theorem for pen raisers is then made a claim about consciousness. And that was a connection that was, to be honest, never very clear to me. It's sort of the idea that we consciously experience things that as being true or false, that we know cannot be proven that way by a system of logic through Gödel's theorem, so therefore consciousness is not a property of these kinds of formal systems. Again, fine. It still might be the case that the brain does do some information processing. It still might be the case that it is just of a constrained or restricted kind. So I think you can have that. You can accept that the brain is an information processing device,

but maybe it doesn't have the kind of class of universality for which Gödel's theorem becomes relevant. Not all computers have to be universal. and becomes relevant. Not all computers have to be universal. But I do also worry that people use that language too loosely when they talk about the brain. It's a very easy thing to say that the brain processes information and that consciousness is a form of information processing

of some sort that we just don't quite understand yet. Maybe, but... It speaks to mind uploading. Well, yeah. Yeah, I mean, to be honest, I think there's this... It's quite a big assumption that consciousness is to do with information processing and that the brain is actually an organ of information processing. Yes, there is an indirectness between what sensory inputs do

when they impact our retinas and our things and what the organism as a whole does. You can always apply information processing metaphorically. You can describe a system from a third-person perspective as processing information. It's transforming inputs into outputs. We can describe inputs, we can describe outputs. And there's an indirect relation. But is that what the brain is actually doing?

Is it exchanging symbols in some way? How is its mechanism actually working? Or is it just a very complicated bunch of levers and pulleys that look as if it's processing information? Does that difference make a difference? I think it does when you draw these kinds of implications from it. So, as you just mentioned, if you believe that consciousness is a property of a particular form of information processing, then you might believe that it doesn't matter what the substrate of that is.

What is the physical mechanism that's underpinning this information processing? And you're not convinced that it's substrate independent? No, I'm not convinced. But again, I'm disappointingly agnostic about it. I think that the only systems we know of, confident in, that are conscious, are biological systems. It's difficult to come up with a good reason why that should be so. Why should it be that consciousness is restricted to things made out of neurons? But there's also many other levels of substrate independence. There are things like,

well, maybe it doesn't matter quite what the molecules are, whether they're carbon or silicon, but maybe the structure of the mechanism still matters. So theories like integrated information theory pick out this middle ground. For integrated information theory, the substrate doesn't matter in the sense that silicon is just as good as carbon, but the mechanism does matter. You can't have a feed-forward neural network without any recurrency. Be conscious. It doesn't. On that theory, that's ruled out because it's a property of the mechanism. So there are many different ways in which things other than input output mappings might turn out to be relevant. And the assumption that none of these things matter, I think, is a dangerous and weak assumption. And of course, that's the assumption on which

these things like mind uploading are based, that The idea that it really doesn't matter. That if I get the input-output relations right, if the system is functionally identical, then I'm going to be conscious in my iPhone. I just don't think that's a safe way to think. Speaking of Tononi, I understand how psi can be measured in, let's say, toy examples in somewhat trivial idealized states.

I think you mean phi, by the way. Psi is very, phi is his measure. Psi is like some sort of, you know, hokey stuff. Okay, that's great. Although there may be a relationship, who knows, right? Okay, so let's talk about phi. Is there a measure of phi that you have found in living humans? So phi in, do you guys know? Well that you have found in living humans? So phi in...

They have to be living. Yeah, they have to have dynamics. So... Is there a way of using neuroimaging to determine the phi in someone, let's say? The straight answer is yes, but it's not the same phi that Julio would want to measure in his theory. There are approximations which I think he can measure. So phi, mathematically, is this measure of how much information the whole of a system has about its evolution over time than the parts do

considered independently. It's sort of a measure of emergence in this sense. You can also think of it a bit more loosely as a measure of combination of integration and information. The system is going to have higher phi if it can enter more different states that make a difference for itself. So the information has to be for the system itself and not for an external observer. So it has to be integrated in that way. Now for Giulio Tinoni to really measure phi, there's a problem in that to quantify the amount of information a system is generating by being in a particular state,

technically you need to know all the possible states the system could have been in, even if it's never been in any of those states. In practice this is nearly always impossible to know. And it's impossible because we don't have a high enough resolution or for some other reason? Well for some other reason, I mean even if you could measure everything in precise detail, there are only certain systems, I think there are gothic systems, I forget my physics terminology now, but most systems will not, unless you push them in some way, they may not explore all the possible states they could be in. They'll only explore a sub-state of their state space. And so you've got to push a system in all possible ways to figure out what all possible states it could be and are.

And you can't do that just by having high resolution brain imaging. What you can do, let's assume you've got this fancy new technology you can measure with arbitrary high, arbitrarily high spatial and temporal resolution. You can measure now an approximation to find. There's still massive difficulties, computational difficulties, in considering how you cut all these, you know, what the right granularity of time and space is that gives a maximum value of it, and how you cut the system up to find where, what partition makes the biggest difference between the whole and the parts.

There are massive problems, but you can do it. And with my colleagues here, with Adam Barrett in particular, and Pedro Mediano in London, we've been developing for nearly 10 years now some approximations to phi that capture some of the spirit of it, that it's a measure of combined integration information, but which you can practically apply to brain imaging data. So that's certainly possible, but it has to be recognized that they are approximations. and as of now they don't work very well when we apply them to actual data in the sense that they they're pretty insensitive to or rather that they're very noisy you can change subtle things in a system and then the measure changes quite a lot. So I wouldn't have very much confidence applying these measures to empirical data, although it's certainly something we're still working on.

I was reading a paper by Peter Lush a few months ago, before I had researched you at all. Turns out you're a co-author on it, so that was pretty cool. It was about the rubber hand illusion and imaginative suggestion, demand characteristics and phenomenological control right exactly now i don't recall it precisely and i wasn't sure about the relationship between the three as far as i understand demand characteristics serve as an implicit imaginative suggestion and as far as i understand phenomenological control is the degree to which you can act on that suggestion or the degree to which you can act on your expectation.

Is this something like that? Do you mind just, if you know which paper I'm talking about, outline it for the audience because I found it extremely interesting. Yeah, this is a switching of gears again, but I'm glad you brought it up. There's a lot of terminology there. In one sense, they're all vaguely unsuccessful attempts

to avoid saying hypnosis because hypnosis has all this unfortunate historical baggage. I think there was the word hypnotizability in the paper it'll be in there but but you have these these so let's take them one by one so demand characteristics is a phenomenon that psychologists have known about forever since the 60s at least and was formalized then but i'm sure people knew about it before this is the idea that if you're doing a psychological experiment oftentimes your subjects might implicitly or even explicitly know what you're trying to you know what the right answer is so to speak and either just deliberately tell you what you were expecting to hear because they don't want to disappoint the experimenter or unconsciously

actually have that sort of experience if you if you expect if you if the context of an experiment is strongly leading you to have a particular kind of experience in one condition rather than another then you may well have that experience and so psychology, it's always been important to rule out demand characteristics if you want to say that something else is responsible, which basically means trying to say that people don't have different expectations for different experimental conditions. And then if you still see a difference, then you can say, well, it's not because of their implicit expectations about what should happen. And then we get to suggestibility, imaginative suggestibility, hypnotic suggestibility. This speaks to there's quite reliable individual differences in how hypnotically suggestible people are.

And by suggestible here, I mean something like you can be asked to experience something like a fly crawling up your arm or something like that. And you might have you might actually start to experience that your hands are being drawn together. you might actually start to experience that. Your hands are being drawn together. There are ways in which people vary in the extent to which a suggestion from a third person can induce an experience. That's what we mean by hypnotizability here. So it's not the kind of Darren Brown,

big magic, stage magic thing. Well, that's fascinating too. It's more like just how much can your experience be influenced by what i say and people vary a lot in this you're in the news there's a huge celebrity right he's brilliant actually i love listening to his interviews as well i think it's underrated here in at least can. Okay, continue. So now you can start to see that if you've got an experiment where

there might be demand characteristics going on, such that the setup of the experiment leads you to expect some sort of experience in one condition but not another, if you're highly hypnotizable, highly suggestible, the demand characteristics might have a stronger effect. I see. Because they basically act as a suggestion, and if you're more susceptible to suggestions, they're going to have a stronger effect. Now the third piece is this term phenomenological

control which Peter Lush and Zoltan Dienes, my other colleague at Sussex, they sort of introduced this term basically to try and emphasize firstly that these suggestions can affect your experience hence phenomenological phenomenology and the idea of control is that it's the subjects, you yourself, I give you the suggestion but then you're controlling you're generating your experience consciously even though you're not aware of doing it yeah that that's the point so i can induce you to experience something if you're highly suggestible you're generating yourself but you're not aware that

you're doing it this sounds like the placebo it's related to the placebo effect it's very related right again people who respond to the placebo effect arguably and there's some evidence that they're more highly hypnotizable um they're more suggestible they're changing their experience in response to a suggestion so there's this very big, typically uncontrolled variable in a lot of experiments of hypnotizability, traits for non-logical control, suggestibility, take your pick. And this matters if your experiment involves demand characteristics, because now the results you're going to get will depend on who you test. And so we've thought, and this is work led by Peter and Zoltan and Warwick, who I mentioned earlier, Warwick Roseboom,

who's worked on the time perception stuff too, thought that this is what's going on in the rubber hand illusion. The rubber hand illusion is this very famous thing. I mention it. I demonstrate it in my TED talk, and I tell the old story. So this is something where I've had to change my mind about it a little bit. You tell the old story? Yeah, the pre-phenomen the old story. So this is something where I've had to change my mind about it a little bit. You tell the old story? Yeah, the pre-phenomenological control story.

This is why I wrote down, because I remember you referencing the rubber hand illusion plenty. And I thought, okay, are you using this as a basis for your model? And I know that there's some holes that are poked in the rubber hand theory, at least by Peter Lush. So I was like, okay, that sounds interesting. But then I found out, wait, you co-authored that paper, so you should know about this. Am I incorrect in my interpretation of it? No, I mean, in fact, we should change our minds, right? I mean, if I didn't change my mind about something,

I wouldn't be a good scientist, right? So yeah, this is work. So that talk was from 2017 and this work in the rubber hand illusion started maybe in 2018 or 2019. But actually they both fit with this general story so this rubber hand illusion is this idea that my experience of you know i can i can change my experience of body ownership of what is my body in some fairly simple lab-based way if if i put a rubber hand in front of you and your real hand is hidden from

view and i stroke the rubber hand simultaneously with the real hand so you're seeing stroking on the hand you're looking at a hand it's a rubber hand but you're seeing stroke and you're feeling it too because i'm stroking your real hand then for some people you start to have this weird experience that this hand is somehow part of your body, even though you know it's not. And so the simple story explanation for that is that it's just Bayesian multisensory integration, that the brain sees touch and it feels touch and it's putting these things together and decides that's my hand. Yeah.

But there's this other factor factor which is that's a situation which is going to very strongly induce demand characteristics you're seeing something looks like a hand and you're being stroked and then the experimenter asks you does that feel like your hand you can see it's going to set up that expectation and so what peter's work beautifully did was to show that in fact individual differences in imaginative suggestion and suggestibility correlate with how strong the rubber hand illusion is i see and he also showed that typically people use as a control they stroke the rubber hand out of time with a real hand yeah so now this is going to break this multi-sensory integration because now you feel touched but it's no longer aligned with the touch that you're seeing.

And what happens? Well people experience it much less. But the point is people also expect to experience it much less. If you ask just people what would they expect to experience when it's out of time, they say well I wouldn't expect to experience it much less if you ask just people what would they expect to experience when it's out of time they say well i wouldn't expect to experience it then so it's not a good control for demand characteristics because what you expect to experience differs in both cases i just okay go ahead i thought that what you did was great with the pulsing of the heart because i don't imagine that someone would think or would even be conscious of how their heart is pulsing to be able to associate the one that is in line with. OK, does that not show that it's not purely the demand characteristics or the imaginative suggestion? Because you don't know, you don't have access to how your heart is beating. You do it some way, but you're not paying attention.

So I think I think unfortunately not in two different ways one way is that in fact uh suggestion effects can influence basic physiology as well it's not just limited to what you might subjectively report about your experience you can induce physiological responses through imagine interesting interesting okay so that's one thing. And the other thing is, get that experiment again. So I was really, and still am, I think it's a great experiment

that we did at the time. But we did not look for individual differences in people's suggestibility. This was like before we did this new work. So we did it a small fat... I mean, the problem is this makes experiments in this area really hard

because to Peter's experiment where we measured hypnotizability and then looked at the rubber hand illusion. We did the experiment, or he did the experiment, well, a bunch of people did, on, I forget, was it 430 or 340? Hundreds of people. on, I forget, was it 430 or 340? Hundreds of people. Normally, these experiments are done on groups of like 20 or something like that. And so I don't know. In this experiment that you're referring to, where what we did was instead of stroking the hand, we gave people a virtual hand and it flashed either in time or out of time

with their heartbeat and again we found that people experienced it more as their own hand when the flashing was in time with their heartbeat so sort of what we called a cardio cardio something version of the rubber hand illusion intercepted version um i don't know to what extent that's driven by suggestibility i just don't know because we didn't measure it and so we need to do it again and we haven't yet for us i'm also wondering and i'm just thinking out loud here if there's any utility to test test this experiment when the participants are on psychedelics, in a sense that...

There's always a psychedelic question on this podcast. There's always... I was asked that. Because, you know, you discussed the level of complexity, how that's... In one of your talks, you were discussing how it gets... It's more complex when these individuals are on psychedelics versus the wakeful resting state.

Yeah, so we've done some work in collaboration now with the group at Imperial College London led by Robin Cahart-Harris on what happens in the brain during psychedelics. And there are lots of interesting differences. One of the challenges is that when someone's on a large dose of psychedelics they can't really do anything. They can lie in a brain scanner but you can't get them to do an experiment and press buttons and report what's going on. So one of the things I'm excited about is the potential for actually research using microdosing here. So if you give people very small doses, sure, they don't

have the big, vivid hallucinations. Any differences in the brain are going to be correspondingly more subtle, but you have the big advantage that people can still do experiments so you know i'd certainly like to i think that's that's a very rich vein of work to to follow by the way just to connect those two points there was there was a lovely paper i think by olson and colleagues from a couple of last year where they they claim to have induced a psychedelic experience through hypnotic suggestion. Interesting. It's called like tripping on nothing. And they sort of set up the whole context,

have a very nice, relaxing lobby area, and people come in fully believing they're going to be taking part in some experiment with psilocybin, and then they don't get any, but they're still led to believe that they have got it and then they often report like all these hallucinations and things going on so so potentially you can even induce psychedelic effects given strong enough expectations about what's going on now of course i would say there's probably still going to be a difference between a placebo psychedelic experience and a real one.

But quite what the difference is, is an interesting open question. What insights have you found from any of your psychedelic experiences, if you can talk about it, that you feel like were, let's say, worldview shattering or worldview forming? shattering or worldview forming? If anything, it was more reassuring because the thing is, a lot of my work over the years has been about trying to understand perception as this kind of construction, this perceptual best guess. And we've been talking about examples of seeing faces in clouds and all these effects that fit very comfortably with this view of how the brain works and how it does perception. So you think when you're on psychedelics, you're temporarily shifting with the Bayesian priors?

Yeah, I think so. I think it sort of makes sense that way. Firstly, it gives you very, very immediate insight that your perceptual world is a construction of some sort. It starts departing from what it used to be quite quite quickly uh and you know it's interesting people can have very different i think take different lessons from experiences like this depending on their prior beliefs about them again so if people believe that there's some sort of believe that there's some sort of it some people might believe that psychedelics are evidence for a more immaterial form of consciousness because you suddenly depart from your in yourself you don't experience yourself so much and and so you might think well this just this is good evidence that that consciousness is is not this thing created by my brain somewhere behind my eyes.

But for me, you also might reinforce beliefs in things like panpsychism, that consciousness is everywhere because the whole world seems to be alive and vivid with meaning. But the other way to think of it is that if you come, say, somebody like me with a prior belief that the brain is the organ of experience, it's just more evidence for that because you mess with the brain in a very straightforward way.

You just activate these serotonin receptors and bang, your experience of the world and the self massively changes. So I sort of interpret it as very strong evidence for the material basis of our conscious lives. What do you make of Daniel Dennett's claims of, well, let's say Daniel Dennett's view of consciousness? Have you read his Explaining Consciousness book?

I always find it insightful to know how, let's say, luminaries like yourself and others disagree. Well, Dennett is the luminary in this group. I started reading his stuff again at a reasonably early stage. So in the early 90s, I was lucky to read it, Consciousness Explained, and it's a remarkable book. And I have been a lot inspired by Dannett over the years and

I agree I think to a large extent but maybe not the whole way but that might partly be because I found it quite difficult to know what the whole way

actually means for Dan. Meaning you're unsure of what his viewpoints are? Well, I mean, I've had many conversations with him, and I know there's a lot of... So one thing that I very much like about his ideas is this sort of idea about multiple drafts and the centre of narrative gravity and this sort of idea that the self, the narrative self, this identity that we have, the thoughts that we think, that it's not some sort of internal essence of self. There's just many things going on and we kind of construct our experience of being a continuous self. He's also, for me anyway,

bang on about some of the mistakes that people make when they think about perceptual experience. So, for instance, he's very much... Actually, I'll tell you one story, because when I gave my TED Talk in 2017, he was in the audience. And I was massively terrified giving this talk.

It's a terrifying experience. Oh, you looked calm on stage. Yeah. It's a totally terrifying experience but i was most terrified of the fact that danny was in the audience and wondered what he was going to think of it and he was very kind and he said it was almost perfect and okay what's the problem what's the problem he said well you said and i have like a father figure he said it's not good enough there's this no he was it was it was well he he picked he

picked out precisely the mistake yeah he wasn't to know about the rubber hand illusion stuff right that wasn't a mistake nobody knew about that we hadn't done the work i'd said as part of the talk this idea of conscious perception as like an inner movie that and i can't remember the precise words i use now i used to know them totally by heart but um this idea that i described this multimodal fully immersive inner movie as a description of what perceptual experience is like. And that's what he said. He doesn't like that because it implies there's a little person in your head viewing it? Exactly, exactly.

It implies, even though I didn't mean it and I don't believe it, just that linguistic slip, it sort of, it gives that impression. It's easy to interpret that way. And it might speak that i still fully hadn't grasped that point if you speak of an inner movie you imply somebody watching the movie no there is no inner movie right and that just begs the question but i do recall um i do recall dennis saying that it's okay to have a little person inside your brain watching the movie as long as that little person has another one and it doesn't regress infinitely because each little person can be different than the last do you remember that still doesn't know

this argument not not really i don't i just think it's easier to do away with that argument at all yeah there's no need for it it's a it's a solution for there's no problem that that it's a solution to a problem that doesn't exist. Why would you need an inner homunculus anyway? Yes, we do have what we can call metacognitive processes. So I can make judgments about my conscious experiences. I can tell you whether I'm confident or not, or I can tell you whether I can reflect on and evaluate things that happen. But that doesn't mean there's a sort of inner observer.

It just means there are hierarchical organizations. There's a hierarchical organization of my cognitive architecture. So, yeah. So he's certainly, he's always, I've always found him very on point about picking up on errors like that. Where I've never been entirely clear is these claims about the extent to which we're mistaken about what it is we're trying to explain when we try to explain consciousness. He's often criticized or understood as saying that qualia don't exist. I'm not sure that that's what he means. I think it's entirely

right to say that qualia, the redness of red, the painfulness of pain are not what we think they are but that doesn't necessarily mean they don't exist but yeah, for me he's definitely one of the people I still always learn from every time I speak to and it's been an interesting seeing, talking to him and seeing how he's talking more also in terms of the

brain these days and trying to put his ideas within that framework which I think is a really interesting project Do you agree with his conception of free will? The compatibilist approach? I'm

a compatibilist yeah I think free will is another kind of perceptual experience and I think this whole approach? I'm a compatibilist, yeah. I think free will is another kind of perceptual experience and I think this whole debate about determinism is totally irrelevant to understanding free will. I'm trying to get Dennett on the podcast. He said that he's

busy writing a book so he can't come on. I'm also trying to get Douglas Hofstetter. What are your views on Douglas Hofstetter's model of i don't know i don't know him personally and i've read only really the amazing um it's one of the best books everybody should read that it's phenomenal um it's just playful what i take from that is just this incredible playfulness, creativity. I have it right here. I keep only three books beside me, four books. It's one of them.

Yeah, that's one of them. I don't know if you know this show, the Desert Island Discs. We have it on the radio here in the UK. And the idea, it's been going on for like decades. And the idea is you, what are the eight songs that you would take with you if you were getting banished to a desert island, never to return? You can only take eight tracks.

And then you're also allowed, so most people in England spend half their lives figuring out what these eight tracks are going to be just in case they get invited onto the show at some point. You want to be ready. But you're also allowed to take a book. Cool. For you.

That would be definitely, well, I wouldn't say it would be the one, but it would certainly be up there. I haven't made a decision about the book yet. I haven't made a decision. So I love the way it just playfully explores our intuitions about what cognition is, what mind is, what explanations in biology and physics consist in consistent I think it's I think it's a synopsis he's a genius full of insights yeah have you read his analogy book I

haven't I haven't it's a great one it's all this but it's a far too long sometimes it goes through lists and lists what I find it to be somewhat tedious okay so what are your views on his views of consciousness where do you agree disagree well it's a tricky question because to be honest i i it's been i i would be hard pressed to articulate what they are i mean to me it got he talks about strange loops and things like that I don't have a particular strong view because to me I've always just associated him with these things about language, recursion, all these playful insights so I don't think he's, as far as I know, he's not

come into my radar specifically about consciousness more about what self consists in and what we think of as a self and that's when you're saying that the homunculus it's a it's a foolish model in some ways what about global workspace theory as far as i understand i don't know much about it i just had a cursory glance at it sounds like there's a homunculus because there's something that's viewing or there's a... Well, you can explain it much better than I, but what are your views on the global workspace theory? So the global workspace theory is one of the sort of most popular theories in consciousness science, neuroscientific theories these days. And it's first developed by Bernie Baars and taken on mainly by stander hayne in paris

neuroscientist and it makes a very simple sort of uh proposal that what we are conscious of the contents of our consciousness they become conscious in virtue of being broadcast in this workspace in the brain that gives them access to many different cognitive processes, which is why I'm conscious of this glass of water. It means I can do many things now in virtue of being conscious of it. I can drink it, I can tip it over the computer, I can take it to the kitchen, I can put it down again, drink from it I can talk about it and so that whatever in my brain is representing the perception of that glass

has access to all these different flexible cognitive processes and for global workspace theory it's that process of being available broadcasted around this workspace that endows that perception with the property of being conscious um the analogy i think it's done it again uses the analogy fame in the brain for that fame in the brain fame in the brain so something becomes famous famous within the brain all other parts of the brain know about it suddenly my like my auditory cortex knows about it my language

centers know about it my motor cortex knows about it it's famous there's a cup there's a there's a glass of water hooray everybody can see it it's famous yeah and that fame in the brain underwrites conscious perception. So that's the claim. The nice thing about it is it explains a lot of the functional properties we associate with consciousness, because indeed there are things I can do when I'm conscious of something that I can't when I'm not. And this flexible action is central to that. It's also testable because you can put people in brain scanners and change,

present them with stimuli and vary whether they're consciously perceived or not. And you can look for is there this ignition of this global workspace in the brain when they're conscious? And many experiments would say there is. The issue with it is that it's not for me explaining a great deal about and back where we and as we sort of begin to wrap up with back where we started which is with phenomenology so the fact that something is famous in the global workspace explains a lot about what we can do behaviorally and functionally, but it doesn't really explain to me, at least, well, why is this visual experience the way it is? The phenomenology of this visual experience has a character.

Just being famous in the global workspace doesn't really explain that. It also is not going to address the hard problem of consciousness. Yeah, that's what I'm thinking. No one would claim that it is. but there's this often inverse relationship it has a benefit of being eminently testable um and although you know you can finesse and say well the global workspace is wherever you find it in the brain and there are problems because how do i separate the signature of a workspace being active from simply my consciously reporting something has happened? Maybe the difference I'm seeing in the brain is just the difference in

my verbal report rather than in the mechanisms that give rise to the experience. So there are some tough questions there too, but it's an appealing theory i you know i don't focus on it because for me i want to focus on what kinds of mechanisms explain phenomenology and for me the bayesian perspective is much is much more interesting for us why don't you ask a question to wrap up and then i'll ask my last question as well first of all it has been a fascinating talk thank you so much for this conversation been Thank you, it's been great. I thought it was going to be half an hour. It's certainly not half an hour.

I'm sure our audience... Our perception of time has changed. Yes, it has. Just to wrap things up, maybe very briefly, if you could just let us know, do you consider yourself as someone who looks at the world through a pragmatic lens as opposed to let's say maybe an objectivist lens because you know i i always look at the world i think there are there are different frameworks from which we can we can choose to

to look at the world and to me the pragmatic approach seems to be a the one that is that matches really nicely with the evolutionary um evolutionary perspective so i'm just wondering what your thoughts are on this very briefly i know it's been a long time do you mean as a as a scientist or as a sort of person going about my daily life here do you think there's a i suppose i suppose both would be interesting as a scientist it makes sense that that you want you might want to take a more objectivist approach but maybe as a person could be uninteresting well i think there's always a blend right i mean even as a scientist there's a sort of objectivity is is an ideal right you you don't get there and much philosophy of

science emphasizes the how in practice pragmatic it is you've got to do experiments you come up with testable hypotheses and you don't have a kind of complete it's not hypothesis test ruled out either it's it's sort of what ideas are more uh and lakatos talks about experimental fecundity what frameworks give rise to experimental predictions um and so yeah i think scientifically i'm quite pragmatic as well experimental fecundity, what frameworks give rise to experimental predictions. And so, yeah, I think scientifically I'm quite pragmatic as well. I want to focus on areas of the problem where progress is conceivable, which is why I don't focus on coming up with a, trying to think up some eureka solution to the heart problem of consciousness.

I prefer to see if I can get there by doing stuff that can be done. And yeah, I guess that answers the other question as well. Is that part of the reason why you dislike the heart problem, just because of its ostensible intractability? I don't dislike it at all. I think it crystallizes in a very, very important way an intuition about consciousness and a metaphysical challenge for explanations. It's just that I don't think addressing it head-on is the most pragmatic, productive path for consciousness science to follow.

Okay, so I have quite a few questions. I'm going to list them and then you can just choose one. Whichever one feels like it would give you the most bang for the buck. I was curious, you mentioned Granger causality and you said that it wasn't used often in neuroscience and I want to know why is that the case then okay another question was the relationship between consciousness and intelligence not the way that intelligence is used when people say artificial intelligence but more like iq the neuroscientific iq and i was curious if there's been any tests where you give someone anesthesia how much anesthesia is required is proportional

to the iq of the person so that is one way of testing is someone who has a higher intelligence more conscious in some sense okay so i was gonna ask you okay the one i want i wanted to know about your opinion on box theories there's a slide in your presentation i believe it was about the free energy principle and how it relates to hold on you just pick the one you want me to answer okay let's get back to artificial intelligence because you have a phd in it there's different architectures some of the famous ones like gan convolutional now yosha bach said that one of the ways that the gan that the generative adversarial network relates to the brain is in dreams. He thinks that dreams are like your brain testing you at your weakest point, almost like the adversary.

Okay, so you get the idea. Now I'm curious. There's GANs, there's convolutional recurrent LSTMs, there's transformer and Boltzmann, let's say. That's right. Are these different architectures, do you see them as related to different modules in the brain? Much like Yosha said, hey, well the GAN might be

somewhat related to dreams. Is it like the convolutional is a visual perception and it's pretty much that. Transformer is your ability to generate dialogue. Okay, can we pick a different question? Okay.

I'm kidding, I'm kidding. Let's go with that. We'll have that one. We'll take it a little way. Why, is that a difficult question to answer? It's kind of difficult. I mean, so yeah, my PhD in AI was quite some time ago, but we're actually still working on it.

In fact, working on it more and more these days, again, with some super smart colleagues and grad students. And I think for the reason you just mentioned, because we have this, I think... So it's easy to say that the brain is making predictions about the causes of sensory data, and that's what we perceive, and it's doing some sort of Bayesian inference. But exactly how it's doing it and exactly how you can collect evidence to test these ideas of exactly what's going on

depends on having a much more sophisticated computational model of this process. having much more sophisticated computational model of this process. Which is why I'm very interested in these different classes of network. What can they explain about perception and behaviour? How well do they map onto what we know, what we can measure in the brain? And actually the other way around, can we actually develop better machine learning architectures through understanding more about how the brain does these sorts of things. And there's good example, one of the things we're working on there is for instance most convolutional neural networks which are extremely good at classification convolutional neural networks, which are extremely good at classification, require bucket loads of data.

One of the reasons they started to work is that they can be trained on the Internet. There's huge amounts of data and compute computation power and lots of interesting little tweaks. But the principles of deep networks were there long before the Renaissance. long before the Renaissance. But we learn things quite quickly. We don't need exposure to everything in order to learn new concepts and things like that. So we're doing it slightly differently. One of the things that humans and animals do when they learn is they select what data to learn from. You pay attention to different things, you forage for data. And so this opens a possibility, can we define, can we design more efficient machine learning

algorithms that gain efficiency through selecting the data that they learn from? And then how do they know what data to look for? Well, that's a problem we can investigate in terms of what humans do. Is that related to the frame problem? No, not particularly. It's more, I mean, that's really a problem about how things come to have meaning or a symbol grounding frame. Well, yeah, I guess they're sort of related.

In the sense that... This is really a much more a much more kind of yeah they're related in the sense that that basically how much do you need to know about context and things to learn stuff but just very pragmatically what's the size of data set that you need in order to achieve certain performance and generalization capacity in a network so that's one way but the other way I think is more interesting from a perspective of what we've been talking about, which is do any of these classes of architecture in AI,

do they serve as useful ways to understand what the brain is actually doing, how it might be actually implementing or approximating something like Bayesian inference. And we'll finish just with a sort of a simple line of thinking here which is the still currently off-the-shelf networks that people are still excited about are basically feed-forward networks. These deep convolutional neural networks like AlexNet, they're feedforward. There are many layers and they're trained on a lot of data,

but they process it in this kind of feedforward direction. Seems the brain doesn't work like that at all. In fact, the whole frameworks that we've been discussing are more about top-down predictions being reined in by sensory data. They put a lot of emphasis on the recurrent top-down, inside-out connectivity. And so to model that kind of activity, a feed-forward deep convolutional neural net is not sufficient. So we need to look at architectures that have top-down connectivity, that have generative models. And there are architectures like this? There are architectures, there are. And there are generative models and there are architectures like this have model there are architectures there are and and they you know there are generative networks now i mean

there's a long history of generative networks they go back to variational auto encoders and things like that um helmholtz machines helmholtz being the same guy that is historically associated with predictive perception and the bayesian brain um but they became a little bit less worked on because deep networks became so powerful so quickly. But now there's a lot of movement back towards the power of unsupervised learning, of generative models in networks, generative adversarial networks of one example of that, where one network will try and produce examples to fool

another network's classification. You said that Joshua talks about that as a good model for dreams. Indeed, it might be. There are many other ways to think about what dreams might be doing. It's just sort of pruning the complexity of a generative model. You let it free run and then you kind of prune its parameter space down a bit. And so there are lots of, I think, lots of interesting things to be explained about perception, cognition and consciousness through coming up with the right kind of machine learning architecture, computational model and things that involve explicitly generative components for me that's where the action is at the moment professor thank you so much i'm definitely going

to be following your work and many of the people who are listening are going to follow your work from now where can people find out more about you what are you working on next well thank thank you firstly thank you for us that was a very stimulating conversation um we could go on for longer but I'm about to die so I don't think we can um but thank you for a wonderful conversation um thank you if people are not already completely sick of stuff that I have to say then um you can find out more easiest thing is on my web page which is www.annielseth.com i'll leave a link in the description and i think as you mentioned at the the top there's you know there's a 17 minute version of all this is the ted talk which is out there and uh i am putting the finishing touches to uh a book um aimed at the

informed general audience so hopefully the kind of people that would be watching this listening to this um the book is called being you a new science of consciousness and it's all being well will be out uh sometime between may and september next year we haven't got a precise publication date but it's the plan is to publish it in the middle of 2021 so in less than a year and you can sign up on my website for updates about that book if that's something you're interested in okay everyone go to his website sign up get notified when he comes out with this book follow him on do you have twitter i do oh yeah yeah that's so you can follow me on twitter uh you'll put the link in but it's just at annual k seth. So you can follow me on Twitter. You'll put the link in, but it's just at Anil K. Seth. But yeah, you can follow me on Twitter too.

Thank you again. By the way, how is it that you stay so fit? Do you have a regimen? Do you meditate? Do you fast? I don't know why you think I'm that. well no i mean actually this you look healthy one of the could be wrong for me one of the weird things about the the pandemic this year um is i've not been well not women but none of us have been traveling but actually i've been lucky enough not to not to come down with it so far and slightly paradoxically i feel fitter than i have normally done because I've been able to keep a better routine. Just don't have jet lag all the time. And I can build going for a run two or three times a week properly into the routine. So actually looking after myself has become a little bit easier in some ways during this period.

ways during this during this period i find that academics and people who are more interested in intellectual domains tend to do well during this lockdown because they like to sit and write and not be disturbed oh yeah well if only it was like that i mean the the the flip side of that is i mean this has been a lovely conversation on zoom but the number of hours that i spent on zoom time zones also seem to mean less because i can now just get wrapped into a meeting in california and the right night in australia the next morning so yeah 7 p.m there right now or 8 p.m it's just coming up just uh yes okay and lastly you went like this when you said that you don't have covet do you feel strange as someone who at least i would imagine you think of yourself as a rational person, and this obviously doesn't do anything, but it feels right and it feels like if you don't do this, something bad may happen. So how does that comport with how you see yourself?

That's a really interesting question to finish on. You know, yeah, I mean, it's still true. We still have these, I mean, partly it's a social signal, right? I just wanted to do that to avoid having to say that fortunately so far I have been able to avoid. It's kind of just a gestural shortcut. But still, academic understanding does not trump our lived experience all the time. I still see colors as really existing in the world, even though I know they are constructions of the brain. So the way the study of consciousness changes me as a person

is that everything changes, but everything stays the same too. It's great. It was a great talk. Thank you so much. Thank you. I look forward to reading your book, professor. Yes. I'm going to do it hopefully on vacation if it's next summer after a long time, that should be great. Get some rest, man. Okay. Thank you.