Anil Seth: Consciousness and illusion

Anil Seth: Consciousness and illusion

Anil is a cognitive and computational neuroscientist, TED speaker, and author of the bestselling book, Being You: A New Science of Consciousness.

Anil is a cognitive and computational neuroscientist and author of an excellent and highly successful book called Being You: A New Science of Consciousness.

We discuss:

  • Perceptual illusions, and whether different people perceive the world in the same way

  • The extent to which we perceive reality as it really is, rather than as a useful fiction

  • Exotic experiences like dreams and psychedelic trips

  • The function and origins of consciousness

… and other topics.

Watch on YouTube and listen on Spotify, Apple Podcasts, or any other podcast platform. Read the full transcript here. Follow me on LinkedIn or Twitter/X for episodes and infrequent social commentary.

Paradigm is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber.

Episode links

  • Twitter: @anilkseth

  • Book:

  • Dream Machine:

  • Perception Census:


0:00 Intro

4:10 Blue and black or white and gold?

9:08 Yanny or Laurel?

12:17 Do we see the world in the same way?

18:21 Perceived is a controlled hallucination

24:16 Why d consciousness exist?

31:19 The Hard and Real problems of consciousness

34:15 Integrated information theory

46:17 Beast machines and the emergence of consciousness

1:04:40 Dreams, psychedelics and exotic conscious experiences

1:18:00 Book recommendations

Introduction: Things are not as they seem

This was one of my favourite conversations on this podcast to date. This is a conversation that gets right to the heart of some of the core themes I’m focused on in this podcast and in life more generally.

In Paradigm we look closely not just at what we know, but at how we know - or how we think we know it. And absolutely foundational to this entire endeavour are our perceptions. It is through our perceptions that we receive information from the world. Everything comes in via some combination of sight, hearing, touch and so on. It is using our perceptions that build models of the world, and it is with our perceptions that we test our worldviews against reality. 

If you think about it, everything we think we know about the world is in some way grounded in perception. And you can easily verify this for yourself by picking any beliefs you have and asking yourself why you believe this thing. And you might find that for some beliefs there are layers and layers of concepts and logic chains all on top of one other. But at some point in that line of questioning, perception will always be involved. There will always be some form of seeing, hearing, tasting, smelling, touching - something, in which your belief is grounded. Even in strictly scientific domains like astrophysics, where complicated instruments are involved to collect and interpret data, at the end of the day it is via our perceptions that scientists ingest this data and make sense of it.

Our entire knowledge system is intimately intertwined with out perceptions.

And it’s largely because of this fact that I find this conversation so compelling, and frankly quite terrifying. As we’ll discuss, different people can perceive the same physical reality in very different ways. People might literally see and hear very different things, even when confronted with the exact same physical inputs. We might see different colours in an image, or hear different words in an audio recording. And this is not some abstract philosophical point. Anil and I actually tested this in today’s conversation. We sat down together in the same room and looked at the same image, and listened to the same sounds, and yet saw and heard very different things. And you’ll be able to test this with us when you watch or listen to the conversation. As you’ll see, perception varies between individuals far more than one might think.

Given this fact, it’s extremely important that we seek to understand more about the ways in which our perceptions vary. If we want to understand one another, and coordinate and communicate effectively, then we need to have a better understanding of how people might see the world differently.

Anil is actively working on exactly this topic through an initiative called the Perception Census, which is a ground breaking scientific study exploring the ways in which we each experience the world around us different. And if you’re listening to this before November 2023, then you have a chance to be involved by taking part in it directly. And I strongly encourage you to do so. Check out the links in the show notes for details.

Thank you for reading Paradigm. This post is public so feel free to share it.



This transcript is AI-generated and may contain errors. It will be corrected and annotated with links and citations over time.

[00:00:03] Anil: I see it as a blue and black dress.

[00:00:06] Matt: You see it as a blue and black

[00:00:07] Anil: see it as a blue and black dress.

[00:00:08] Matt: I see it as a gold and white dress. And Tessa, who's my fiance last night, was gold and white. She could not believe that other people see it as blue and black. And it's fascinating. I mean, the actual answer is, is blue and

[00:00:21] Anil: It is blue and black. Yeah. The real dress, if you see it in a, it's like in a photo with less ambiguous lighting, then it's clearly blue and black dress to everybody. So I'm right and you're wrong. Basically I win.

[00:00:34] Matt: It's true, but I mean, what, uh, what could possibly cause this if, um, you know, you look online, I think it's something like a 50, 50 divide or something like that.

[00:00:42] Anil: That's right. And that's what was fascinating about it. I remember when this, this photo first kind of took over the internet, I think it was in 2015. And I'd been just teaching a class on visual perception, getting back to my office. And there was just a ton of voicemails about this dress and this photo of the dress.

And to start with, I was also really perplexed because I'd not seen an example like this, where it was such a sharp divide between one way of experiencing it and another way of experiencing it. Our brains. Do something called color constancy. So, um, I mean, when we, when we experienced things around us, like we're sitting in this room now, it seems as though the colors that objects have, well, they're just out there and we just register them as they really are.

So it seems like this table really is brown. The walls really are kind of an off white, something like that. But that's not the case. And this is not a new insight, right back to Newton, possibly even before, way before, actually, you know, people have recognized that. Colors as they appear in our experience are what the brain makes of how surfaces reflect light.

Out there there's just electromagnetic radiation when it comes to, you know, how we see. And that's not colored. I mean, we might say there's a red wavelength, a green wavelength, and a blue wavelength, but they're just different wavelengths. They're not actually different colors. And our cells in our eyes are sensitive to these different wavelengths.

Out of different combinations, the brain creates this kind of almost infinite palette of color, but it does it in a very, very sophisticated way. The color that we perceive an object to be is in part shaped by what kind of ambient light. The brain thinks is going on. So if you take a white piece of paper from in here to outside It will still look white Even though the light hitting your eyes from the paper has changed quite a lot because outside it's mainly bluish sunlight I mean even when it's cloudy, it's still bluish light in here It's mainly yellowish and this has been well described process, but what had never been Recognized before was that there are individual differences in it, and turns out that that photo of the dress hit this sweet spot totally by accident.

It wasn't like a experiment designed to have this effect, it was just this photo that just appeared organically. And for some people, for some people's brain, The, seemed to be the brain was assuming that there was a sort of yellowish ambient light, like an inside light. And so these people experienced it as blue and black and other people's brains assumed the ambient light was more bluish and they saw it.

As white and gold 'cause different ambient light situation. So there's this incredible world of individual differences, uh, that this photo of the dress suddenly exposed. But the thing that I think is really fascinating about this is that, you know, the dress was a, was a meme. We, you know, it sort of took over for a while, then people forgot about it.

And there's a more fundamental truth, which is your question that you started with this assumption that we all perceive. I think the key word here is sufficiently, because it's not just the dress, like we're all having different experiences all the time, just that most of the time they're not dramatically different.

They're sufficiently similar that we use the same words, like, yeah, the table is brown, it's reasonably warm in here, so on and so on. We get along and we can behave in the world and it works and we function. Okay. But the truth is when probably not having the same, exactly the same experience, even if we share the same, the very same objective reality.

[00:04:42] Matt: Yeah. I mean, in the, in the color space, you could at least still argue that we could just be substituting words and we're still somehow qualitatively experiencing. Something the same thing, you know, you could argue that all our colors are inverted for example, but I think there are other Illusions just very similar to this that are not Color based so there's an audio one as well I feel like we should we should do it because a lot of people will be listening to this I'm not sure if you've come across this one, but I have one prepped here and this is Somebody's name being read out, and I would love to hear what you think.

[00:05:19] Anil: Oh yeah. Yeah. That's Yanni. Yanni. I've heard this one too.

[00:05:25] Matt: You're hearing Yanny?

[00:05:26] Anil: Yanni.

[00:05:27] Matt: I hear Laurel. And I cannot for the life of me hear the word Yanny.

[00:05:31] Anil: exactly. So I think it's another great example of, of, um, a bi stable perception. Right. Um, but instead of something that flips back and forth. Within a person like there are these, um, optical effects. There's one called the necker cube, which is like a wireframe cube.

And if you stare at it long enough, it sort of flips. Like sometimes you see it from below. Sometimes you see it from above the other things like this. The famous duck rabbit picture as well. Sometimes it looks like a duck. Sometimes it looks like a rabbit. These. Other illusions like this, this Yanni Laurel or the dresser.

There's various other ones as well. Green, there's something that to me sounds like green needle. Other people here is brainstorm.

[00:06:12] Matt: Yeah, I have again my fiancé heard the opposite to me and we could not reconcile.

[00:06:17] Anil: thing is they seem very different, right? It's not just like green needle versus grown nodal or something, you know, green needle versus brainstorm.

What's going on. And I think that's, that to me is a real insight into the fact that. Even though it seems as though we experience things as they are, whether it's visually or auditorily, or even olfactorily, or through whichever sense, that's not the case. All of our experiences, and this is, you know, something that's driven a lot of my work for years now, are active constructions, and the brain is It's interpreting sensory information, which is always fundamentally ambiguous.

And the role of these expectations is so important that it can make all the difference between hearing the same sound as brainstorm versus green needle, which seem to be very, very different kinds of things. So it really opens. the mind to the possibility that we can't take for granted how Veridically how realistically our perceptual experience reflects not only what's actually there What other people are experiencing too?

[00:07:32] Matt: Yeah, but I mean it also, in some sense it feels like as soon as you recognize that one of these things can be true, the whole bottom drops out from beneath you because, uh, there is, I still, I still think, you know, it's very commonly asked this question, philosophy 101 question, do we perceive the world in the same way?

And as soon as you encounter just one of these illusions and you realize that the answer is not, it does open up the whole spectrum of the question just how differently. Do we perceive the world? And I guess part of that is an empirical question, and I put it to you. How differently do people perceive the world?

How wide is the gap between, you know, people walking around in the street?

[00:08:10] Anil: It is an empirical question. It's one we're actually trying to to answer right now There is a long literature on aspects of this. So there's a whole subfield of psychology that studies individual differences in perception and typically it may look at individual differences in maybe one aspect of perception like vividness of mental imagery, you know, if people bring something into their mind visually for some people it's very For other people, it's much more abstract or there's really nothing going on.

Uh, we may differ also in, for instance, how we perceive three seconds and how we perceive duration. Um, but not a lot is known about how people differ in lots of different ways, all at the same time. Most of the work that's out there is also another, well, another aspect of this literature of this field has focused more on the extremes too.

So there's lots of work on so called neurodiversity. I mean, it's a very, I think, useful concept, um, that first originated. As a more of an activist term, um, in the context of autism. So people with autism, it was used to underline the fact that people with autism really experience things very differently.

And maybe sensory information seems much more overwhelming often than for people with autism. But the term neurodiversity, partly because it was very successful in, um, helping us understand. Some of these conditions like autism, ironically, I think reinforced the idea that if you weren't a neurodivergent, then you were neurotypical and experience things as they are.

And the dress was a kind of nice wedge into that because it affected everybody. And so this is the real question for me. It's like, not just at the extremes where the differences are large enough that they end up. Manifest in differences in behavior and we might slap a label on them like autism or synesthesia or something else What about the differences?

Are still there, but that we don't recognize because we use the same words and we basically, we can get away with it. So it's an empirical question for the last year and a half, we've been running this experiment called the perception census, which is a large scale attempt to characterize exactly this.

What, what is the diversity out there? I mean, we all differ in height and skin color and so on, and we can see that, but what is the, you know, the wild diversity out there really like? So this experiment has been running for just over a year and it consists of lots of little, like these examples that you were just showing, really fun little tests and illusions and games, but covering vision, covering sound, covering time, emotion, uh, all sorts of things.

And people can. Basically do engage with the census, all you need is your own computer and you just, you can do bits at a time, keep coming back and we're hoping the results from this census, this perception census will shed light on this diversity in a way that's not been done before. We've already had 30, 000 people take part from a hundred countries from age of 18 to 80, which is quite a step up from an average psychology experiment and we are.

I'm still collecting data. I know it's going to be down to the wire probably by the time people hear our conversation, we're closing data collection at the end of October, 2023. So in about two weeks from now, um, so if it's still open when you're listening to this, then please do give it a go because everyone who takes part really does contribute to our understanding of this phenomenon.

We haven't looked at the data yet because we're being good scientists. We want to, you know, not, not peak before we've got some idea what we, what we're looking for, but I'm very excited about what we'll find.

[00:12:18] Matt: Yeah, I mean, what, what, what do you, what do you expect to find? What is the hypothesis?

[00:12:24] Anil: Well, the overall hypothesis is that there will be quite a lot of differences. Um, I, to be honest, there's so much data we're collecting, it's going to be more of an exploratory. I feel like it's mapping the territory rather than going in to test any specific hypothesis. There are going to be some things that we'll test.

I mean, there are things that we already have hints about from previous experiments, like people differ in their degree of. Hypnotic suggestibility, like we're all suggestible to some extent. It's not a bad thing or a good thing. It's just a thing. And we know that differences in suggestibility predict some differences in perception.

It's as if you're more suggestible, your, your perception is more influenced by context. So we can make a lot of specific predictions there. But really what I'm interested in is just what is this distribution like and how does it compare? Previous studies, which have generally been more limited in their coverage, you know, in terms of countries and age and, and so on.

So we can ask questions like, does our perception change with age? Does it depend on, I don't know, which hemisphere you, you live in, the patterns of daylight that you get, but also what goes along with what. So the idea here is that we might have something like. Perceptual personalities or perceptual phenotypes that explain how different aspects of our perceptual experience go together.

Maybe if you're really sensitive to sounds, maybe you're sensitive to time in a particular way. I would be, we don't know yet, but we'll be able to. to look for these kinds of cross modal relationships and correlations.

[00:14:08] Matt: Yeah, I think a lot of people will find this a little bit sort of disquieting or uncomfortable to realize that you're, you know, I think, I think there is still this belief that people have that perception is somewhat of an, a window into reality and. As you said earlier, this is, this is definitely not your view, and I don't think it's the mainstream neuroscientific view.

Now, I think you, you've used the word in the past of a controlled hallucination, and maybe let's, maybe, let's get to that. Could you walk me through this idea then of the world as we perceive it being a controlled hallucination rather than a window into reality as it is?

[00:14:43] Anil: Right. And the control is just as important as the hallucination in this sort of metaphor, which I think is often overlooked because, yeah, we, we do all experience things differently, but in ways that are still for most of us, most of the time tied to the world in ways that make perception useful. So yes, you're right.

I'm not. Suggesting that our perception is random and sort of completely made up disconnected from objective reality, that's clearly not the case. Otherwise we wouldn't survive very long. But it's also not the case that our experience is this direct window onto how things are. So I think the best way to, to get into this is, is to start with.

Just underlining that, that for most people is how things seem, like, for me, even right now, I don't know about you, but it seems to me that the world is just there with colors and shapes and space and objects in different positions, and, and I just open my eyes, and the world just pours itself directly into my mind.

And maybe, you know, if I, you know, if my contact lenses are... Season up or something, then I won't see so well, but it's still, I'm looking through a blurry window onto to reality, but still a window. But what's actually happening is not like that at all. We already talked about color. I mean, color is not a mind independent property of the world.

It's something that the brain creates out of colorless wavelengths of light. And I think the same thing goes for everything else, too. So, if you imagine being a brain, I mean, you're locked inside this dark, silent vault of a skull. There is no direct access. All the brain gets are these electrical signals.

Which are only indirectly related to things out there, or indeed in the body too. And to make sense of these ambiguous, unlabeled, noisy signals, the brain has to, or the idea is the brain has to make use of its prior expectations or beliefs about the way the world is and combine those with the sensory signals to reach a kind of best guess.

The best guess of what's actually out there. Formally, this is a process of Bayesian inference. Like I have some prior expectation of what's going on, some data, and I kind of combine the two, or my brain is combining the two in a way that's an optimal guess of what's going on, given the prior and the data.

And this really changes this intuitive. Not only an intuitive idea, but the sort of classic textbook picture of how perception works is as this bottom up outside in readout of sensory information flows into the brain. But in this alternative perspective, which has long history, you know, Plato, Kant, Helmholtz, um, the brain is constantly generating predictions and the sensory signals.

Predictions primarily calibrate these predictions. They keep the brain's predictions tied to the world. And perception is the continuous process of the brain trying to minimize prediction errors. Minimize the discrepancy between what the brain is predicting and what it's getting. Or everywhere and all the time.

And this really does change things, because now, instead of the brain reading out sensory signals from the bottom up, what we experience is conveyed by the brain's inside out, top down predictions that are reined in by these sensory signals. So this is why I think the term... Controlled hallucination is, is useful because it underlines that all of our perception is sort of internally generated.

It comes from the inside out. So when we hallucinate, we typically experience things that are not there or others don't. And we can think of hallucination as a kind of uncontrolled form of perception when the brain's best guesses go off the rails. But normal perception is just really fundamentally the same process.

But in this case, the brain's predictions are tied to reality in ways that are not determined by accuracy necessarily, but by utility. You know, we see the world in ways that evolution has decided is. Maximally beneficial for us. Colors, again, being a good example, the brain creates colors because they're useful to guide our behavior.

I think this is this, it's still to me one of the more surprising revelations of recent cognitive neuroscience because it just doesn't seem like that. I mean, even though I believe that's the case, it still seems to me that the world exists out there independent.

Ah, it's being actively generated everywhere and all the time.

[00:19:42] Matt: Yeah, there is an open question, like, you know, looking at the neuroscience, one can intellectually understand that it can make sense. And I think reflecting on it, maybe one can get somewhat closer towards appreciating it. Uh, you know, in a more subjective, you know, feeling sense. But then there is the question of, but why does it have to feel like anything?

Why does it have to be an experience? Why can't all of that be happening, um, in the background, in the dark? And this leads, uh, very nicely to a lot of the core focus of your work, which has been the consciousness aspect, and the consciousness, because everything we've been talking about, underlying all of it, is the fact that there is.

Subjective experience. This perception isn't an in the dark control mechanism, I'm not a thermometer, um, I experience it. And so maybe let's turn then to that question, you know, if, if this is what perception is, as you explained, a control mechanism, um, a, a, um, something that compares what we're getting from the world from our sort of like internal prediction of what is there, why should that feel like something?

[00:20:49] Anil: Um, yeah, so this, this is the big question. I mean, already the other questions are quite large as well, but this is probably the, the one that underlies all the others, because this is where the other questions flow from, like if we didn't have experiences, we wouldn't kind of wonder about their relation to the world as it might actually be.

And perception can also happen unconsciously. There's plenty of research that shows that at least some aspects of perceptual inference of this kind can go on totally under the hood. And so consciousness is not inevitable for at least some forms of perception yet. Seems to be the case that We have conscious experiences, some philosophers will deny that and say, actually, we're just mistaken about the way we think about consciousness.

Um, but I, I think I, I take a fairly common sense view that we have conscious experiences and they depend in some way on what happens in the brain and the body. And there's two ways to address this question of how and why, and I think that is the how and the why. So the why question, why should they, why should we experience anything?

Well, it may be that conceptually you could say, well, I can imagine a machine that does all this, but there's no experiencing happening. That could be the case, but in fact, it might be the case that it, the only way to build a sufficiently flexible, powerful perception. machine, um, given the kind of tools, the material that's out there in the world is to make one that has conscious experiences.

Now it could be that, that consciousness in this sense is, is not optional for the kind of flexible perception that, that we have. It certainly seems that our experiences are very functional, right? It's not that they're arbitrary with respect to what's going on. A typical conscious experience If we think about it in its totality, we have perceptual experiences of the world that reflect the world in a way that, again, is useful for our behavior, and we see things relative to how we can interact with them, um, and our experience of the self is also.

Relative to the survival prospects of the organism, things feel good or bad in our emotional lives. And we feel fear, hunger, excitement, and in ways that, that are highly adaptive and it's all bound together. So we have a single experience at a time, sort of integrates. It's an interpretation of, uh, opportunities for action with a sort of evaluation of their consequences for our future survival.

It's incredibly functionally useful. You can see why evolution would have selected for a system that represents the world and the self in this kind of way. Now, why should that be conscious? Well, consciousness is a particularly interesting format for this, isn't it? Because conscious experiences are unified and are integrated, that sort of defines what they are.

The. So, I think there's a very compelling story to tell about why, but the question still remains about like, what, what is it about neurons and brains that creates consciousness? Why is it still not just all happening, integrating all this information, but maybe in the complete subjective dark? This is David Chalmers hard problem of consciousness.

Why and how should any physical processing at all give rise to a rich inner life? It seems objectively unreasonable that it should, and yet it does, is how he puts it. And there is a deep sense of mystery about that. And I don't pretend to have an answer, I don't think anybody has an answer. But I also worry that it's the wrong question to ask.

Because we can make a lot of progress addressing what, with tongue slightly in cheek, I call the real problem of consciousness, which is to accept that it exists, but then try to explain its various properties. What kinds of conscious experiences do we have? What are their functional uses for the organism?

And as we do that. We chip away at the hard problem. So instead of thinking it as one big scary mystery, it becomes a series of interconnected related, but more tractable problems. And as you do this, and what I think is happening, for me anyway it's happening, is that this, this total sense of confusion and mystery that seems to be there if you face the hard problem head on.

begins to change and it begins to soften and it begins to dissolve a bit. And you realize that once you've explained different aspects of consciousness, the sort of force of the heart problem, the intuitive, emotional force of it is weakened. And the question is, is it, would it dissolve? Entirely in a puff of metaphysical smoke, or will there be a little residue of mystery that remains?

And I don't know the answer, but I'm optimistic that it might dissolve entirely and either way, I think this approach of like, you know, divide and conquer is in any case, the most productive approach to follow, because even if we don't resolve the mystery entirely. We'll learn a lot and maybe the residual sense of mystery will be explained more as a property of human psychology rather than as the insufficiency of science because It's just the curious thing, or one of the curious things about consciousness is we're trying to explain what it's like to be us.

And so we tend to be less, I think we set a higher bar for a theory of consciousness than we might do for, let's say, I don't know, a new interpretation of quantum mechanics or of the origin of the universe or something like that, because we want it to make intuitive sense. Like, Oh, that's why it's like this to be me.

And it may not be like that because it's not that way in other areas of science. We don't ask that.

[00:27:07] Matt: Yeah, I mean that's, that's a perfect stepping point to that question as to, as to why we do have, I mean, I think there are even, there are people who hold the intuition so strongly that consciousness is something that could never be intuitively grasped. I think people, it's almost like seen as a thing that's separate from other questions we ask.

So you mentioned quantum mechanics, I think it's a great example, or gravity or something. We're quite comfortable with the idea that we can explain more, but there will always be unexplained whys at the bottom. You can always ask why again. And I think people don't see this as necessarily a problem to understanding gravity.

We'll assume one day we'll feel confident enough that we understand what's going on there. That, yes, you can always ask more whys, but we would consider gravity understood. But I think with consciousness, there are people who hold the view that that is, that just cannot happen. That, you know, people who really believe in the hard problem so much, um, they think it's, it's of a different type of stuff.

Um, what is your view? Is your, is your hope that the real problem approach, um, will, uh, you know, even for those people, they will eventually see the problem of consciousness in the same way as we see any scientific question, you know, what is gravity?

[00:28:16] Anil: I think it will always be different, but I, I, I hope it will become more similar actually to these sorts of things. I think it's a really good example that we're more willing to accept sort of intermediate explanations, um, for other things. And, and why is that? Well, part of it is because I think these other things aren't us.

So we worry less about any explanatory gap that might remain. Uh, the other thing is these. theories we might have of say gravity still have an extraordinary amount of explanatory and predictive power. You know, Newton did us very well for centuries, relativity, even better now. Uh, and even though we know neither of these theories is complete, um, but they're still extremely useful.

I think one of the issues with the science of consciousness is that it's utility. It's there. So there's already interesting applications. And I think a lot of just applications, but it's maybe not as evidently useful as something like a theory of gravity was within physics and its utility hasn't been maybe as widely appreciated.

Um, as might be, I think that is because the insights that we've got into consciousness so far, and from one perspective, just don't even touch the surface, you know, for a lot of people, well, that's not about consciousness. Like the only thing that's about consciousness is if you come up with like, here's the solution to the hard problem, everything else is just, just not in the, in the game.

And I think that is a mistake. I think if we can explain. properties of consciousness, functional and experiential, then we're in the game of scientific theories of consciousness that may only be partial, but that can be very useful.

[00:30:03] Matt: There was a little bit of a controversy recently in the consciousness space, um, on a question sort of like this. There are many scientists, some of whom have actually been on this, on this podcast, um, Signed an open letter, um, basically likening one of the. leading theories of, of consciousness, integrated information theory to a pseudoscience.

And, um, you know, there are various reasons for that, but, you know, I think, um, a lot of people do think anything that is at its bottom inherently subjective, like consciousness, uh, this is somehow not really addressable in the same way as, um, as things that we typically address in science because of, you know, science seeks objectivity.

Unconsciousness seems to be the only thing that is always going to have this subjectivity, um, baked into it. Um, and then in the integrated information theory, the context is slightly different. But again, um, it's framed as a pseudoscience because it commits you to something that is not amenable to be tested, uh, with the standard tools of conventional science as we know it.

Um, so what do you, what do you make of that, uh, of this recent, maybe you can talk to what happened and, and what you make of it.

[00:31:17] Anil: Sure. Um, actually before doing that, I think it's worth just addressing your first point, which was this aspect of consciousness science, which again is challenging. And one of the things that makes it distinctive, which it's a science of something that is. It's generally agreed, not directly objectively observable.

I cannot put a conscious experience on the table. And we, you know, we make measurements directly of that conscious experience. Its nature is that it's private and it's subjective, that may not even be subjective. It's subjectivity might be a optional property of conscious experiences, but certainly it's a distinctive methodological challenge.

And one of the main reasons why. My consciousness research was really off the table for most of the 20th century, even up until when I was starting my undergrad degree in the 90s, it was like, no, no, no, you don't go there. And behaviorism was really dominant in psychology precisely because of the skepticism that you could study objectively something in mental, unobservable mental processes.

But we can make indirect measurements. I can tell you what I experience. Maybe I can do that only with a certain degree of fidelity and a certain degree of, of, um, maybe not completely exhaustively, but still in ways that are useful, you know, that still constrain explanations. So I don't think it's a fundamental limitation to...

Doing a science of consciousness, but it's true that it may impose some limits on exactly how far you can get, but I think that's a methodological limit. It's not the same as this issue of, of like, is there a hard problem of how consciousness can relate to to matter? I think those things are, are distinct.

So, you know, I think we just need to accept that there are difficulties, but nonetheless it's not completely.

[00:33:15] Matt: I mean, maybe just even following that point, there is actually a sense in which even the standard objective science has that problem, right? Because, uh, you know, what, even, even mathematics, something as pure as mathematics, what convinces someone at the end of the day that a theorem is true? Um, you know, you kind of go through line by line and eventually, You have a feeling that it, uh, it makes enough sense.

And actually if, if one applies, that, that, to me, that seems to apply to all of science. At the end of the day. It always comes down to , you know, subjectively, as a scientist, um, I have to eventually buy into the evidence. I have to believe the results, the, even the data that I'm interpreting that is a subjective experience of interpreting data.

Is is that, is that the same as, as what you were saying, or is,

[00:34:00] Anil: think it's similar, sorry, I think, I think you're right that, that, I mean, science in a sense can be thought of as the progressive, um, distancing of what is subjective from what isn't. You know, we, we, instead of just feeling what's hot and warm, we develop thermometers, but the development of a thermometer was bootstrapped by what felt hot and warm.

And we still look at, you know, the results of these instruments. So you never take the subject completely out, but you try to sit, you try to sort of. make systematic the involvement of the, of the scientific observer. So I think that's what one thing the scientific method does. And it is a bit more complicated in consciousness because what you're trying to observe is, is in a sense the, the subject itself.

Um, but you're right that it's, it's not as if the rest of science is. Just transparently objective, if it was, there wouldn't be really any need for a philosophy of science and, and we would all, you know, proceed uncontroversially, just grind the wheel and facts come out and it's, it's just not like that, of course.

Um, so there are, yeah, I think the methodological challenge for, for consciousness. It's a little bit special, but it's not, um, insurmountable. Now, what about the pseudoscience thing? So you could say, you know, okay, because of the nature of consciousness, basically all of consciousness science is pseudoscience because you, you know, you can never make, you can never put a conscious experience on the table.

You know, it's always going to have this relation of indirectness, but you know, I think that's really unfair because a science to be a science. There's very general definitions of science, but a very, very general definition is sort of, um, systematic investigation of some natural phenomenon through theory, experiments, some combination of theory and experiments.

Scientific explanations should, um, generate testable predictions and shed light on a phenomenon. That's very, very general. And consciousness science can do this, you know, we can make observations of what happens when people fall asleep or go under anesthesia or see, see the dresses of blue and black or white and gold.

And we can build theories around these things, but there has always been this, this slight suspicion of consciousness research, um, that goes back to some of the other concerns about the methodologies, the limitations of the methodologies. So over the last 30 years. People within consciousness science, I think, have fought hard to give it the kind of legitimacy that it enjoys.

You know, now you, you know, we attract, you can get research funding, smart students come into it, and there are applications coming out of it, and it's all, I think, a very healthy and exciting field. But... It is still a field where there is no single dominant core theory, you know, in physics, you've got the standard model.

In consciousness research, there isn't. There are theories, and this is great, you know, initially there weren't really even theories. Now there are theories, there are a few leading candidates, um, and one of them is called integrated information theory. And it is a particularly counterintuitive and challenging theory about the nature of consciousness.

And it tries to directly tackle the hard problem that we've talking, we're talking about, and it makes the claim that consciousness is identical to, in the theory, irreducible maxima of integrated information. And quite what that means is very difficult to summarize, but it's very precisely mathematically stated.

But the key thing is it's. It's stated in a way that makes that core claim very difficult, if not perhaps impossible, to test, at least as things stand now. And it seems that that fact about IIT, together with some other things that have been happening, like the feeling that this theory was getting perhaps more media coverage, then it's...

Deserved compared to other theories. Um, and that it has implications like a certain form of panpsychism, that if the theory is true, then consciousness is gonna be much more widespread. Then we might intuitively think, uh, this led to a bunch of. scientists and philosophers, including many of my friends and colleagues and people whose work I admire, writing a letter, uh, concluding that integrated information theory should be considered as pseudoscience.

I strongly disagree with this

letter. Um, with respect to all the people writing it who are trying to, I think, ensure the health of the field by, you know, by their lights. Drawing attention to a theory that they think may undermine the legitimacy of the field. I rather think it's a valuable part of the landscape. I personally, I think integrated information theory is very likely wrong.

It's certainly very. Crazy and counterintuitive, but it's still science. It tries to, it's in fact, a very, perhaps the most precise statement of a possible solution to the hard problem of consciousness that has been developed yet, and it does generate testable predictions. Maybe not of the core claims of the theory, but it still generates testable predictions.

And that, for me, is enough. And philosophy of science perspective from Imre Lakatos, the Hungarian, talks about science in this, in this regard, that there are often theories where the core claims can be in principle untestable or in practice untestable. But nonetheless, if over time they generate testable predictions that have explanatory and predictive value, then the whole research program is productive.

And if they don't, then the research program becomes degenerate and sort of fades away. And I don't know what will happen with IIT, with integrated information theory, but I think it deserves. The chance to be tested. I think it, I wrote in a recent piece about it. I think there's theories, so long as they play by the rules of science, have the right to be wrong.

And I think IIT has every right to be wrong. And because we don't have a standard theory, you know, that really does already explain most of what we want to explain, I think if we, if we dismiss ideas in part, because they're counterintuitive and difficult to test, then we risk stifling the kind of creative thinking that we still might need in order to make progress.

On the science of consciousness. So I, I, yeah, I've, I've sort of, I think it's an unfortunate episode for. For the field, because it's creating a polarization where I really think there's not so much of a need for it. I think, you know, there's many different approaches that are being explored by, by different groups of people.

You know, I have my own approach. We've been talking about controlled hallucinations. It's very different from IIT. It's also probably wrong, almost certainly wrong. Um, but it's still, I think, uh, a useful perspective and so long as theories satisfy these basic constraints of. generating testable predictions and explaining phenomena in ways that are not totally inconsistent with the rest of our scientific understanding, then I think we should embrace a diversity of theoretical approaches.

[00:42:05] Matt: Yeah, well, let's, let's maybe then get onto your, your theory of how this all comes about and. You know, it shares some aspects with integrated information theory. It, it draws on work from Carl Friston, um, and the free energy folk. And so maybe let's, let's get to that. The phrase that you, that you use is beast machine.

Um, and so maybe let's anchor on the beast machine. Could you explain to me this, the use of this phrase and then, um, let's expand into, into your... Sort of theory or model about how this all comes to be.

[00:42:41] Anil: The phrase came from, um, from Descartes, Rene Descartes always gets a bit of a bashing when people talk about consciousness these

[00:42:50] Matt: have unfortunately bashed him on this podcast.

[00:42:52] Anil: Uh, it's a bit unfair really, isn't it? Because, you know, we, we bash him now, um, from our perspective of knowing what we, what we know now. And of course, he was incredibly, um, important thinker in his time and made a lot of really important contributions.

But he also said some things which now look a bit strange. And one of the things he said. I mean, normally he gets bashed for the whole mind body problem and the pineal gland and all this stuff. But another thing that he said when considering consciousness in non human animals was that non human animals should be treated as beast machines, bet machine in his French.

And the point he was making, you know, whether it was what he really believed or had concluded from his philosophy, or whether it was what he was saying to sort of play nice with the dominant religious attitudes of the day, was that the fact that other animals were living creatures did not endow them with any Claim to conscious status or all the kind of conscious states that had some moral ethical implications.

So beast machines just was to connote that they're flesh and blood, but merely machines. They don't enjoy ethically meaningful inner lives. That was his use of the term. And I've, it struck me as an interesting term because I ended up basically concluding exactly the opposite. That in human beings and very likely in other animals, and quite possibly in general, conscious experiences happen in virtue of our nature as living creatures, not in spite of it, so that we experience the world And the self with, through, and because of our living bodies.

So I'm, in a way I'm trying to rehabilitate or invert or recapture that term, but change its meaning entirely.

[00:45:04] Matt: Yeah, um, it's, I think it's not the first time somebody in this space has, uh, has taken one of the, one of the old greats and, and used, um, one of their terms, uh, in an inverted way. I'm, I'm thinking of, and I, I don't think this person has a, has a theory that I particularly buy into, but, uh, at least they're, they're doing interesting things, which is Philip Goff, um, and Galileo.

Um, but I digress. Um, on, on to the beast machine. You said that we have, um, these conscious experiences, you know, not in spite of, but because of our status as living beings. And without getting to the weeds of what it means to be a living being, uh, let's maybe stick to, to that point. Walk me through the, walk me through the thinking and the steps there.

Um, why, why, why do we have experience because of being living beings?

[00:45:52] Anil: There's two ways to take this argument and you sort of, they meet in the middle, which is I think a useful way to think about it. So the first way starts from what we were already talking about in terms of the nature of perception as a kind of controlled hallucination. So starting there, we've already got the idea that our experiences of the world Uh, perceptual best guesses, that's shaped by evolution to be useful rather than, than accurate, but we've got this machinery of prediction and prediction error in our brains, that's underlying experiences of the world around us.

The next step. is to realize that it's the very same kind of process that underpins the experience of being a self. Now the self is not this mini me inside my head that does the perceiving. It's, it's another collection of perceptual experiences. I experience my body as a particular object in the world.

I experience my identity over time. I experience some actions as being voluntary and others not. All of these things together constitute what it means to be. be me subjectively. David Hume talked about this is a bundle theory of self and this is just putting a kind of predictive processing gloss on that old philosophical idea of the self as a form of perception.

So that's the second step. The next step is to realize that the experience of self is probably grounded, arguably grounded, Um, not in some abstract notion of personal identity, you know, me being Anil Seth, but in the brain's perception and regulation of the interior of the body, the physiological condition of the body.

We already talked about the fact that the brain has no direct access to the outside world. It has to infer the relevant states. The same is true for the body. All right, there's sloshing around in chemistry and so on, but it's still, you know, has to infer what's happening within the body, heart rate, blood pressure, and so on and so on.

And that's arguably the fundamental role of the brain that everything else is built on. You know, if brains can't keep the body alive, then, you know, that's game over already. Everything else is secondary to that. And so the experiences of Just being a body, things like emotions and moods, and perhaps at the very bottom of it all, a lot of interesting phenomenology here about what is the simplest kind of conscious experience that's imaginable, that might be reached in sort of deep states of meditation and so on.

Often described as just the, just the feeling of being alive, you know, without shape, form. Without emotional contour, it's just the feeling of being alive. Now that could be the root of all experience. That is really the consequence of the brain's predictions that, in this case, are not geared to finding out what's out there, but geared to control and regulation.

As soon as you can predict something, you can control it. And there's a whole way of motivating predictive models that start from their role in control, rather than in... Um, so that, that's one way through it. And then we sort of realized, well, and that of course is just intimately related to our nature as living creatures.

This imperative for control goes all the way down, you know, even individual cells have have their imperative to sustain themselves over time. So, you know, the weakest claim, and I think the most defensible claim is that we will never understand consciousness except in light of our nature as living creatures, because.

All the machinery that underpins our conscious experiences evolved and develops and operates. From moment to moment in light of this drive to stay alive, that animates our body, the stronger claim is that this is actually a necessary condition for consciousness, that consciousness is not something that can be manifest on a, you know, on a.

Let's say inside a computer, but it is a property of living systems only, you know, this is a position sometimes called biological chauvinism or, or, um, uh, biological naturalism, carbon chauvinism sometimes. And it's often pooh pooh, because why, you know, why, why are you just being closed minded about not imagining alternative manifestations?

I don't think it's an... I think it's actually an instance of, of open mindedness because we become just more suspicious of some of the common assumptions that are out there that consciousness is a form of information processing, which a lot of people say, but very few people actually articulate what that really means.

And old, you know, analogies, the brain is a computer, which is still hanging around even though the brain is not a computer. So that's sort of one way to it. The other way to it very, very briefly starts from Karl Friston's free energy principle and related ideas like autopoiesis, um, which is a sort of theory of how, um, cells self organize and maintain themselves and produce themselves over time.

So the material aspects of. Life becomes very important. Life isn't just a process running on some sort of, um, wetware of the body. You know, life is fundamentally a metabolic process where the components matter, you know, energy is consumed and generated. And from the free energy principle, which is basically this, this idea that things that exist, basically, firstly, they, they define themselves from other things.

There's a boundary and this boundary can be statistical. There's a sort of statistical independence of things. Within the system from outside the system. And so things that exist induce a boundary, a separation. And then in virtue of, of this, you sort of end up deriving the rest of this predictive idea of the brain.

So in order to keep doing this, according to the free energy principle, systems have to minimize. It's the surprisingness of their environment, like a fish in water will stay alive, but a fish out of water will rapidly die. So Things that persist over time tend to minimize the statistical surprise of their situations.

But in order to do this, then systems need, they can't do it directly, they can only do it approximately. In just the same way that our brains can't do full Bayesian inference, they have to approximate it. And this is what's called free energy. This is this term free energy. It's not like electricity that costs nothing.

It's a term from thermodynamics and statistical mechanics that That kind of systems can minimize as a proxy for minimizing the surprisingness of their surroundings. And so from here, you begin to get something that looks very like predictive processing systems that do what they do in virtue of minimizing something like prediction error.

And so the rest of it then follows. So what's nice here is you go from something that's Either you start from like how we perceive the world around us and you, you pull on that thread until you end up right down into the depths of our internal biochemistry, or you start the other way around or ask what it means for a system to persist over time, and especially a living system, and then you go outwards and you end up in exactly the same place.

So for me, the combination of these ideas motivates this strong connection between life. And consciousness, that's the pull factor and the push factor is to just recognize that this assumption that consciousness is substrate independent, could be implemented in anything, um, and might is a form of information processing.

That's just a really. Strong assumption to make, and I think we don't often realize how much of an assumption that really is.

[00:54:17] Matt: Yeah, I spoke um, a couple months ago on this podcast, I spoke to Mark Soames, who has a, a view that seems very, very consistent with, with what you've just said about, I guess the function of, I think he, he focuses more on affect, but you know, we are drawn or, um, repelled from certain feelings and again, as you just articulated, his view is very much that that happens in line with How, um, how divergent our predictions are or expectations are from the signals we get externally or internally from within the body, but sort of external in a way.

Um, but I think his view focuses much more on, um, it does seem to be more substrate independent in the sense that if you could replicate that functionality. You know, a system, whatever it may be, having those functions, he, in his view, that is, that is what is required. And so, um, when you talk about sort of being a bit skeptical of the substrate independence view, is, is your view more that the typical view of substrate independence is, you know, running something on a standard computer, or is it, you know, even if you could build a system that could genuinely replicate that functionality?

Um, would, would, would that be, in your view, a likely candidate for this emergence of, of consciousness? Maybe it would be , maybe it would be life, maybe that would be life in a different substrate, but what is your view there?

[00:55:40] Anil: That's a, it's a good question. And it's, it's really something I'm still, you know, thinking a lot about. Um, and I think my views will probably evolve. So there's a lot of shared ground with Mark. This is, this is true though. There are a lot of other differences as well. He's a strong focus on the subcortical mechanisms more so than me.

Has a sort of injection of Freud, which, which I don't have. Um, but you're right. There's this difference that he, in this sense is a more orthodox position that it's the function that matters. Now it might not be the function of a standard digital computer. But it's still the, the functional organization of the system.

And if you get that right, then you get consciousness. So that's, that's a sort of more general functionalism, the sort of functionalism that, you know, many, some people in AI might run with that. Um, yeah, if we get the next generation of language model or whatever will be conscious, that's more of a computational functionalism that it's not any kind of function, but it's the kind of.

Functions of the sort that a computer can execute and get those right. Then you get consciousness. That's a really strong form of functionalism. Marx and more kind of liberal and a bit more plausible. Um, but I still remain to be convinced. And I, in fact, go through a slightly similar sort of experiment in my book.

Like if you had a synthetic robot that had internal states like battery and so on, and it was monitoring them and regulating them and so on, you know, would it be, would that be enough? Would it be enough? I think to claim to be sure about the answer is to claim too much. I think we need to recognize humility here.

I don't really know. But my, my intuition and my credence's shift on this is that no, it wouldn't, it wouldn't be enough. It wouldn't be enough. You'd still have, um, a kind of model, a simulation or be a kind of hardware simulation rather than the thing itself. You know, there's this difference between, you know, some things are like, um, playing chess or like mousetraps to classic examples that are often used.

If they do. If they play the functional roles right, that's enough, doesn't matter how the mouse is caught, if it's caught, there's a mouse trap. Other things are more like the weather or like diamonds. They have to be made of a certain stuff in order to warrant the label. Diamonds have to be made of hard, you know, particular kind of carbon.

Otherwise it's not a diamond. Rain has to be made of water. Otherwise it's just not wet. So just putting it that way to me makes it a very It's very clear that it's still an open question, you know, is consciousness more like the weather or more like chess? Is it more like the mousetrap or more like diamonds?

If you implement all the functions, maybe you simulate the brain, you model the brain. Maybe you don't even just model it on a digital computer. Maybe you actually instantiate the, the causal structure of the brain in a particular way. Would that, would that be enough? I, I'm still not convinced because of the reasons that we were just talking about, because there is, there is this.

Sort of idea that fundamentally consciousness is so intimately tied up with, with our biochemistry, with our metabolism that it, that might be the unnecessary substrate. Now you cannot prove that, and in this way it sort of suffers from some of the same criticisms as IIT. You know, if this is a core claim, this is an untestable part of this theory, currently anyway.

There's no way, at least I can think of, to test this difference because it remains a bit of a philosophical difference. Um, but I think it's, Yeah, my credence is still that, that life really matters because of these very, very deep, intimate relationships. And because of the, the thing that in a living system, there is no way of saying like, where does, what is the substrate?

You know, substrate independence is always. Suggested when it is assumed that we know what is the substrate and what's running on the substrate. This is how computers are built by, by definition, but it's not how brains are. Every time a neuron fires, the structure changes. Every time it uses a bit of energy, you know, something happens to ion channels and things like that.

The brain is continuously changing. So it's really challenging to say, okay, that's the substrate. If you can't say where the substrate is, then the question of whether consciousness is substrate independence or not, doesn't even, doesn't even make sense.

[01:00:35] Matt: Yeah, and there is a question, almost whatever the answer is, so substrate independent or not, or maybe substrate dependent but certain subclass of substrates, whatever the answer is, if you do buy this idea that the, the function Um, of consciousness is, as you said, to sort of keep us in line with the expectation.

Um, there is a sort of question that opens up on the other end of things because, you know, we all have experience on a day to day basis of having sort of small divergence between expectation and, and the signals we're getting in and, you know, that might draw us towards or push us away from things. But then there are very extreme conscious experiences, things that are very exotic.

And, uh, you'd almost be surprised that they can happen. And I'm thinking of things that you've explored, for example, in, uh, research with psychedelics, in the Dream Machine project. These experiences are so far removed from anything that you could imagine could be serving that sort of, uh, purpose. Um, and so maybe let's, let's move to those sorts of more exotic experiences and your more recent work, um, and maybe the Dream Machine.

[01:01:46] Anil: before we do though, there's just this one tiny thing that I think is really worth underlining here. It just came from what you last said because we've talked a lot about, um, the function of consciousness. And we've also talked about functionalism as a sort of view of what might be, you know, the necessary conditions for consciousness.

The two are very different things, right? It's totally fine. And I remember, I remember sort of having to clarify this for myself, not that long ago, that it's totally reasonable to talk about consciousness, having a function as, as we've done in terms of bringing together lots of information in an organism relevant way, um, that doesn't mean that you've got to be a function list about the material basis of consciousness.

So, yeah, I just want to keep those things separate.

[01:02:30] Matt: No, thank you. Yeah, that's, uh, that's, that's very true. Um, that being said, um, assuming that the function is, is, uh, as we say, you know, uh, managing expectation versus what's coming in, it does seem quite kind of odd. Then, you know, and take this psychedelic example. This is a very small change to what's happening in someone's brain in some sense.

It's a very, you know, little bit of chemical. And, um, but the experience is, is vast, um, and, and same with dream. So, um, like let's explore that. Is, is it, um, is this just a spandrel? Is this just a misfiring of, of the hardware? Uh, how do you think about this issue?

[01:03:11] Anil: mean, so there's, do you remember this? There's a book by, I can't remember who wrote it. Was it Terence McKenna? There's a book called A Stoned Ape,

[01:03:18] Matt: I think it was Terence McKinnon.

[01:03:19] Anil: was sort of this idea that pharmacologically induced states of mind, you know, they played an adaptive role in, in the evolution of mind as we know it now, I tend to think that's probably kind of unlikely really.

And. Um, you, you're right that psychedelic states, on the one hand, they're very notable by, it is, it is quite a subtle perturbation to the, you know, the biochemistry, you take a very small amount of LSD and, you know, you get on a roller coaster and stuff happens. Um, but, you know, I think if, if our brains were to react to, I don't know, let's say, Um, pollen in the same way they react to LSD, then evolution would have rapidly selected out that sensitivity to pollen, right?

Because it's so prevalent. Um, so I think it's probably a bit of a spandrel and the fact is, even though it's a small amount of chemical, it's, it's, it's not something. That has exerted much selection pressure on our, on our evolution. Of course, it's, there are psychedelic compounds in nature. I mean, some of the original ones were, but they're not everywhere.

And so while human culture may well have accepted the brain sensitivity to, to psychedelics. Um, I think it's unlikely that the core psychedelic phenomenon. played a significant role or has a significant fundamental biological function. Dreams are different, of course. You know, dreams happen most, if not every night, many times.

It seems very unlikely that dreams have no function. So I think we, we can, um, we can take a A functional perspective, functional analysis perspective on dreams in a very different way. So psychedelics give us some functional insight because just like any system, if you, if you push it out of its normal operating regime, you can learn a lot about how it operates.

Um, but dreams are part of the normal operating regime of the human brain and of the human mind. So that's a different window into, into consciousness. We have things, dreams probably do have a purpose.

[01:05:34] Matt: Well, so tell me then about the dream machine and the dream machine project. What, what is the, what is the dream machine?

[01:05:39] Anil: It's, it's kind of the, one of the craziest projects I've, I've ever been involved with. It's been a lot of fun. It was a bit nuts. Um, about 10 years ago in my group with my postdoc, David Schwartzman, we started working. On this phenomenon called stroboscopically induced visual hallucination, it turns out that if you shine a fast flickering light, a strobe light at somebody and their eyes with their eyes closed, if you sit in front of a strobe light with your eyes closed and it's flashing at the right frequency, which is roughly the alpha range around 10 hertz, there's a lot of slack around that you will have vivid visual hallucinations.

Colors and shapes and patterns and movements, sometimes complex hallucinations, people in places, scenes. Um, even though it's just unstructured white light that's hitting your closed eyes, bright enough to get through into the retina and sun, obviously, if you have got no eyes, it wouldn't work. So this is a really fascinating phenomenon just because dissociation between the stimulus, which is just white light and the experience.

So it immediately raises questions about. What's happening in the brain in response to this stimulus that is creating these experiences which are very detached from, from the world. So we've been working on this for a while and honestly is a very much a back burner project. We never really published anything on it.

But then just around 2020, pandemic times, I got a call from a woman called Jennifer Crook, who's a Produces big art events. And she told me about an artist called Brian Geisen, who in the 19, I think in the 19, late 1950s had discovered this phenomenon while sitting on a bus, as it went through some trees with the light shining through the trees, and he turned it into an art object called the dream machine.

And it was a very low tech light suspended in a spinning cardboard cylinder, but it was, it became this kind of cult object. I'd never heard of it. And what Jennifer suggested was the idea that we should. Reinvent the dream machine for now, um, using modern technology, but also make it a collective experience.

So make situations where 20 or 30 people at a time would have both an individual experience. And right back to the start of our conversation, a unique experience. Everyone has a different experience, even though it's the same light. And so we wanted to, we wanted to basically bring this to a wide audience and use it, not as just some sort of gee whiz, look at this.

It's fun, but as a way to. Ignite people's curiosity about the mind and the brain, because the experience within the dream machine, really one reason it's called the dream machine is because it's not like exactly like having a dream, but the experiences seem to be coming from within. It's not like you're looking at something.

You're really immersed in a way that makes it quite clear from a first person perspective. That these experiences are coming from your own brain, from your, from your own mind. So in the course of, in 2022, we, we had some crazy government funding to do this through some bizarre mechanism. Um, we built Dream Machines in four cities in the UK, in Belfast, Edinburgh, London, and Cardiff.

And all together, we had 40, 000 people come through the Dream Machine, each for about an hour and a half. Each having a fairly transformational experience. Some people would describe it as life changing. Some, you know, it's really, for them, many, it was emotionally very potent and, and usually, almost uniformly very positive.

Um, and then they would come out of the actual experience and We had areas where they could reflect, draw and talk and write about what they experienced. And we have 15, 000 drawings that people have made of their Dream Machine experiences, which, which showcase again, an incredible. Uh, diversity. So this, this is the dream machine.

It was, it was enormously exciting to be part of. And I think, you know, I hope very much that it's ignited the, an interest in the mind and the brain and thousands of people who would otherwise not really think about it because it's very easy to go through life and not give it a second thought, right?

You know, there's the world. I see it, I fall asleep. What's to worry about, but once you start to, you know. Encounter some of these illusions or just have some of these experiences. I think you plant a seed in many people that keeps growing and it may grow in different ways for different people. And. And that's, you know, what, what really excites me for the longterm is, you know, what will happen years out from people who've been through this kind of experience.

We're also, by the way, we're trying to now figure out really what happens in the brain and the dream machine. And we're also very interested in the therapeutic potential of this. Now, you know, there's a lot of excitement around psychedelics for treatment of depression and other conditions, but psychedelics.

You know, there's still a lot of controversy and access issues and legal issues. But here we have a situation where like psychedelics, we give people a very. Unusual kind of habit breaking experience, um, but it's not pharmacological, um, it's very controllable when the light stops, the, the experience stops.

And so we're very interested in the potential for this to have some systematic therapeutic benefit. And of course we had all this data from 40, 000 people. So we know anecdotally that it can have a very beneficial effect.

[01:11:41] Matt: Is this still something people can, can try? Is it, is it still running somewhere?

[01:11:44] Anil: It's running in a very limited way in, in London still at the moment. So we have a very reduced setup and what we're, what we're doing is we, we're, well, we're currently working on taking it on tour outside the UK as well. So we'd love to bring it to Australia. We'd love to bring it to, to other countries.

We would really like to bring it to unusual places. Um, so even in London, we were in a. Sort of artistically unfashionable part of London somewhere. I won't, you know, actually that'll probably offend people who

[01:12:15] Matt: I know where, I know where.

[01:12:17] Anil: it was not, it was not like in the Tate, it wasn't in the center, in the center of London, um, because we want these experiences to reach people who might not choose to go to like the latest avant garde art experience.

[01:12:31] Matt: Yeah. Um, if people want to follow up, find out more, um, where Where would you send them?

[01:12:38] Anil: For the Dream Machine, the, um, the thing to do is look up dream machine dot world, and there's a mailing list. And if you sign up to the mailing list, you'll be the first to know. about future plans and where we'll be and how to get tickets. And actually the perception census that we started with as well is also part of the dream machine project.

So that came out, I think almost, I mean, there must be not many projects like this, where we have this quite ambitious science project that really started and was done in total collaboration. With this art project, you know, science and art usually have some sort of, I don't know, the relationship is not often as satisfying as one might like, but in this case it was really intimately, deeply intertwined, which has been very, very satisfying.

[01:13:28] Matt: Well, people should definitely check it out. I'm sure they'll be very interested and, uh, really they'd be interested, I'd imagine in, you know, what brought you to this. And I'm sure there've been books involved. I'm sure you've read widely on this and I'm sure some of them have resonated. And I would love to, as we, as we bring this to a wrap, turn to the topic of books.

One of the questions I like to ask my guests towards the end of conversations is which book have you most gifted to other people and why?

[01:13:56] Anil: The book that in practice I've gifted to most people, it's quite a recent book actually, but I think it's phenomenal. It's Clara and the Sun by Katsuo Ishiguro. Katsuo Ishiguro is a British, Japanese novelist, a Nobel laureate now in literature, who's written some just amazingly delicate and insightful.

Portraits of Society, Remains of the Day was one of his famous novels that became a film with Anthony Hopkins. Um, Clara and the Sun is the most recent of his novels. And it's a kind of science fiction that is not set way in the future in some sort of space opera style. What Ishiguro does is he. He sort of gently introduces some slight conceit, some slight change, might be quite implausible, and then uses that to understand aspects of society, aspects of human psychology.

And the reason I've been gifting Clara and the Sun a lot is that it's, it's about consciousness and it's about the possibility of machine consciousness. Clara is a, is a robot, um, and we're encouraged to consider the world from her perspective. So the conceit, of course, is that she has a perspective.

Everything else about the book is sort of kind of familiar. So what it does is it casts the familiar from a perspective that's deeply unfamiliar and then explores the consequences. And, um, and there's, as with most, probably all Ishiguro books, there's some, some real twists in it as well, but it's. It's, it's almost a work of philosophy as much as literature and it's just beautifully, beautifully written.

[01:15:49] Matt: Amazing. Yeah, it's a, it's a, that's a great recommendation. I've not read that one yet, but it does lead very, very nicely to, um, the last question, uh, which is maybe a bit of a science there, but it also deals with the topic of artificial intelligence and the prospects of artificial superintelligence. Uh, and my question is, you know, suppose we were to be visited one day by an AI superintelligence.

Who should, uh, and we had to pick someone to represent us, who should we pick?

[01:16:16] Anil: I, I, I, yeah, I'm gonna have to think of this now. I, I, I, um, yeah, I, I should have given this some thought before.

[01:16:23] Matt: Past or present. Past

[01:16:25] Anil: past or present? the, my answer to that question is going to be carefully embedded in the next book that I write, whenever that might be. So I'm going to leave people hanging on that one.

[01:16:37] Matt: Yeah. So I'm going to leave people hanging on

[01:16:41] Anil: It was, yeah.

[01:16:41] Matt: was, that was my suggestion. So anyway, it's been an absolute pleasure speaking with you today. Thank you so much for joining me.

[01:16:46] Anil: Thank you, Matt. I've really enjoyed it.


Conversations with the world's deepest thinkers in philosophy, science, and technology. A global top 10% podcast by Matt Geleta.