Consciousness with Michael Graziano
2026-01-20 13:00:00.000 +00:00

Yeah, we always screw this up, but I think we're gonna nail it this time. I'm David Pinsoff. I'm David Petrozowski. David, what podcast is this? This is Evolutionary Psychology the podcast, and it's about actual evolutionary psychology. Being done by actual evolutionary psychologists. Not exactly the Italian line. It's it's actual evolutionary psychology by actual evolutionary psychologists. Okay, close enough. Tomato tomato. Enjoy. David, we have a wonderful guest today. I'm very excited about our guest. Yes, this is uh our first guest that does not formally identify as an evolutionary psychologist on paper, But when you listen to him speak, you will identify him very easily as a fellow traveler.

Uh the way he thinks about consciousness is very insightful and very friendly to the evolutionary perspective, which is why we wanted to have him on. So what do we get into, Dave? Well, a mechanistic approach of like what are we talking about when we talk about consciousness and how could you put an engineer a system to have this thing grounding this very nebulous, uh floofy idea into very concrete functions, which I just love. Um, and I just love this conversation with Michael Graziano because he talks about um really being a different voice than what is largely being done in the study of consciousness, which on on by my account, and maybe by um uh many accounts is sort of very floofy um and maybe not exactly the right

direction. Yes, I agree. Uh and we get into the the pitfalls of thinking about consciousness and the wrong directions that one might take and how to think about it in a non-floofy way. Uh and also just our intuitions about consciousness itself in addition to consciousness. Um that is itself a really interesting social psychological question. And we get into that. We get into religion and spirituality and ethics and AI. It was a super fun conversation. I think you'll have a great time listening to it. Yeah, I think fluffy is a technical term now. And uh this is a great example of the evolutionary psychology of psychology now uh with respect to consciousness. So yeah, without further ado, here is Michael Graziano and David and I talking about

consciousness. Enjoy Yeah, I have um a terrible to pronounce last name. It's Petri Zuski. Um your last name to me seems straightforward, but I'm gonna take a stab. It's Graziano. That's correct. Okay. So yeah. According to me. Yeah, and that's the only person who matters. It's a convention. Yes, names are conventions. So as long as you go back exactly, if you go back a couple generations, then it's um, you know, Graziano. Yeah, exactly. For me, it's not. Yes, that that takes a little few more seconds, a few more milliseconds. Yeah. In this in this workaday world, uh, we we can't tolerate that. That's mine is Pietroshewski, which, you know, again, it's a lot more intonation and a few more seconds anyway.

Yeah. Um, David, go ahead and you're asking. Oh, I should at some point I'm sure it was Pinsofsky or Pinsov and Levelhof Levelsky, and then it got short into Pinsofsky. What is your David? Yeah, what's the background of Pinsoff? Uh I believe it's Russian, so I have some some Russian ancestry. Yeah. Okay, cool. Thus all the allusions to boring Russian novels. Yeah. Yeah. So let's get started if that's if that's okay with you, gents. Um yeah? Yep. All right, great. Michael, we we always like kind of record an intro, so we'll just get right to it. You know, we're very happy to have you here. Um, this is a podcast about evolutionary approaches to psychology. And um, we've typically had very sort of um evolutionarily focused

folks on so far. But as we've recorded episodes, one of the things that we've gotten to is the idea that, you know, if you take a functional approach to the mind, then oftentimes what you're doing is you're studying important information processing problems and describing it at the level of software, sort of what the function is. And if you take intuitive concepts sometimes, like for example, we've David's tackled humor. I had a recent um episode about free will, really difficult concepts where a lot of times people who might think kind of like an engineer or a scientist, they hear things like free will or humor or consciousness, their eyes roll back in their head and they think, oh my God, this is where good thinking goes to die.

You know, like this is just bullshit, right? Um, one of the reasons why I wanted to have you on is because over the years, as I've encountered occasionally dipping my toe into work on consciousness, every time, and I don't even often know that it's you, I just sort of read something, and then I'm like, oh no, but it's him. Uh, to my mind, you're doing some of the clearest-minded work on consciousness, taking what is otherwise this very nebulous, very ghostly, um, almost magical thing, and just approaching it in the right way. And that's a really big um thing to bite off. And that's a really interesting, I imagine, story to sort of how you got there. And I know that's not the only thing you've studied.

So this is my long-winded way of starting out prefacing saying, How does somebody get to how does a kid growing up uh in the world end up uh trying to tackle this idea. And then why of all people, um, and at least maybe we're biased here, but like, why is it that everybody doesn't think well about this thing? And how did you get to think so well about it? If I can ask you a flattering question. Well, thank you. That was a very nice introduction and a very flattering introduction. And yes, I think this is a a huge topic that everyone thinks about at some point, and most people get wrong. Uh, and that's okay. Um, and one of the reasons why people get it wrong is that they rely on intuitions and they

don't realize the extent to which intuitions are simplifications and inaccuracies, uh, you know, useful simplifications, useful inaccuracies, but they are inaccurate. Um, and I think that's at the heart of the whole problem of consciousness. So, yes, it's a it's a huge field. There are lots of people who traditionally don't like the field of study. I do think that's um I don't know if it's completely reversing, but there are more and more people who find the topic worthy and interesting, um although many of those people still approach it in a way that I think is a little is not gonna get them anywhere. Um so it's still a field that's chock full of fluff and illogic and um kind of um superstitious thinking, I would say.

Uh so it's a very difficult field to study, but very, very fascinating and very, very important, as it turns out. Um, I don't think I understood the importance of that field of study until relatively recently, long after I got into it. But I do think ultimately the reason why it's so important is not is because it's not just philosophy. Right. And it's not just psychology and it's not just neuroscience. It turns out to be the key to our technological future. That's what makes it so important to everyone, because now we're facing artificial minds and questions emerge. Can it ever be conscious? What does that even mean? If it is, what do we do about it? Uh and so answering those questions turn out to be kind of the center of what we have to

figure out as human beings. Uh and it ends up being this incredibly important topic to study. Um so yeah. Yeah. Um so you mentioned that um a lot of people approach this topic uh in the wrong way. And I'm curious to unpack what you mean by that. What are people doing wrong and how can they be corrected? Yeah, I don't know how it could be corrected. I've been trying for 15 years. TMS. TMS, a little bit of TMS. Yeah, no, maybe maybe impair the well, yeah. No, we will let's hear what you if you have a longer answer, or we can offer some half-baked joke answers. We have a 12-step program, actually. TMS, your theory of mind system, maybe first step. Okay. Um so first of all, I think a lot of people come to this

field with hidden motivations. They may not even know. There are people who come to this field because they want humans to be special and they want to be there to be a spiritual mystery. And they kind of revel in that. And so one of the things you see at the start of almost every book or paper or talk about consciousness is it is a mystery. We may never understand it. Um, I actually don't think that's true. I don't think it's a mystery. I think we do understand the essence of it. But that doesn't fly well because literally everyone wants it to be a mystery. They want there to be the mystery essence. Uh, and so there's a sort of celebration of the fact that this is something science hasn't touched yet.

Um, I think there's a lot of people who um approach it from a um a point of fear of technology. And um actually I think a lot of the understanding, a lot of the development of ideas in consciousness scholarship emerged from that. I think you can go back so if you trace it back, the first person to really I mean everyone goes back to Descartes. But Descartes 500 years ago doesn't actually talk about consciousness per se. He talks about the mind and he talks about cognition. And um William James, so end of the 1800s, I think what is it, 1890, writes his giant book on psychology, principles of psychology, and he's the first one talking some kind of scholarly or scientific way about consciousness.

And he's the one, for example, who coins the phrase stream of consciousness. And to him and to people at that time, consciousness was content. It was, I remember this, I perceive that, I have this thought, I have this emotion. It's a whole bunch of whole jumble of things that are in a stream flowing along. Um, and that's the concept of consciousness. And then somewhere around the middle of the 20th century, and people like Turing have a huge role to play where he writes his 1950 paper, Can Machines Think? Although, notably Turing also doesn't use the word consciousness, but can machines think? Suddenly people are realizing that the contents, the things that are in the stream of consciousness, are duplicatable.

Like gradually, computers have memory. Maybe it's not quite the same as brain memory, but they have memory. They can make decisions, they can take in information, compute something, and then choose an action. Um, they can um take in sensory input, uh, they can control action, they can do all these components, and suddenly the philosophers have a problem because if we're special and machines can do all the components. What's left? And it's in that state where science seems to be closing in and stripping the mystery away that a new concept emerges. And this is um what uh so probably best put by Chalmers. He calls it the hard problem, and everyone's heard the phrase the hard problem, and uh and the idea is um I it actually dates back a little earlier, but maybe the per first person

to really articulate his Nagel when he talks about what is it like to be a bat? What is it like? It's not the content, it's not the thoughts, it's not the perceptions in your mind, it's that it feels like something. And so this new idea emerges that there's an essence. It's not the information content, it's nothing physical, it's an essence of experience. And that becomes the big thing. And that's what people mean by consciousness. And it emerges, I think, because people are afraid of the technology. And so that's how the field moves all the way through the 50s, 60s, 70s, 90s is when Chalmers is writing about the hard problem. And now, suddenly, in the um 2020s, maybe 2019s, uh, suddenly technology takes another leap.

And deep learning networks, well, LLMs, suddenly explode, and they're doing stuff and they're thinking apparently, and coming out with cognitively complicated stuff. And now you see the whole field of consciousness studies reacting to that and the prevailing, almost uniformly prevailing view is: well, the machines aren't really thinking, they're just mimicking. They have no nothing like mind or consciousness. There's an extra essence that we have. And it must somehow bubble up from this or that. And some people suggest it's like this the subcellular structures that bubble up into the this and the that, or it's something else, or it's quantum resonance, and then there's electric fields, and people are spinning and spinning, trying to figure out what the how

how to express the special thing that biology has that the machines absolutely can't have. And it's again shaping and driving the whole conceptual land uh landscape of consciousness studies. All of this fear uh of uh of mind being something that in the end is kind of mechanistic and reproducible and people don't want that and so i think that's where that's the the deep motivation uh for a lot of these misconceptions that spread everywhere all not just in the public but across the whole you know scholarly field Yeah, that's fantastic. And I think one of the things, one of the themes of this in uh recent episodes, especially, and I think your work is such a good example of this, is that I think in the long scope of

history, the 20th century and early 21st century psychology will be seen as the time when the perspective dependence of psychology was not made explicit. So the idea that the initial constructs like consciousness or free will, those things are intuitive to people. They're outputs of this complex machinery. And so far in science, we've sort of just assumed at the outset, okay, well, these are real things. I'm going to just consult my intuition about what kind of thing it is and just go looking for it in the mind, as opposed to taking a step back and saying, well, this is an output. I'm a robot, and I'm a squishy robot. And I'm this is the what why would an evolved complex computational system think something like free will or consciousness exists?

Yeah. And and doing the science of that before we start just thinking naively it's sort of there. Yeah. find. And otherwise, we sort of just we are dominated by the intuition which might be leading us astray. Yes, exactly. That's exactly right. Sometimes I give the example, since I'm I'm a little bit of an ancient Egypt file, uh I give the example of uh imagine them saying, Well, we know the sun goes down at night into a big hole into the underworld, but we're scientific, we're gonna go find the hole, yeah, And then they expend you know money and resources, and they have journals on the sun hole, and they try and try and they can't find it, and it's a big spooky mystery. And the problem lies right at the root.

Yeah. Just like you said, they accepted as literal truth um uh an assumption or as I would put it um a a model that the brain constructed that's not literally accurate. Yeah yeah and I think I mean I and I mean we'll get to exactly the substantive thing but I think just to preview I suspect that it seems like especially that the self-referential systems, the self-modeling systems are especially where philosophers and people's thinking have really um those are the those are the constructs where there really seems to be a mystery. And and I suspect that's actually a design feature or a consequence of actually the system. And I think we can get to that. Yeah. Um, but I think it's a really interesting thing.

I think it's no accident that the self-modeling representations are the ones that are so philosophically problematic. Yeah. Yep. I think that's right. So if you want to if the um if you want to build a model, the brain builds a model of something out there in the world, those models, of course, are not terribly accurate. But you know they capture some aspect of what's out there in the world. And there are lots of inaccuracies. The most obvious maybe the one that I like to point to is um color or white, in particular, the color white. Um, and Newton figured out that white is actually a mixture of lots of different wavelengths. Um, before him, people assumed that it was pure luminance, stripped clean of contaminating colors, right?

And then the physics question was: how do you take light and clean it to make it pure? And um, and the reason why people thought that is because our visual systems evolved to come up with a really simple, useful, but inaccurate model, bundle of information to describe the mixture of wavelengths. Because it's it's useless to have a really detailed, physically accurate, like the brain doesn't need to know that light is a wave and that a surface reflects some an infinite range of wavelengths within a spectrum. Like you don't need that. You it just a nice simple cartoonish picture of it is sufficient. And that's what we have built into us. We can't help it. We can't think it away. It's there. And so we um we see we see white in that way because that's how

the model depicts it. And um, and one of the most interesting aspects of that, of that kind, which I think is a very close analogy to the consciousness issue, one of the interesting aspects of it is by now people are educated. And so we all walk around with two perspectives at the same time. Because we all look at white things and it still looks white. It's not like we go to school and then things don't look white anymore. Like you can't see all the colors mixed together. The brain, the visual system is what it is that is automatic. We can't help that. So we have to deal with living in a world where to us, from our perspective, white is brightness without color. but we also all know academically what it actually is.

And we're okay with that apparent conflict. And I think the same thing is going to end up have happening or will have to happen with consciousness. It's okay to go through the world and say, well, I have an experience and I have consciousness and I have a mind and I have all these intuitive things. I can't help those models. I can't help my brain building those models. It's not like I'm going to shut them off. It's still going to be part of my life. but I also have an academic knowledge of what they actually are, which is different. Um, and I think people are gonna have to reconcile themselves to that conflict, like one's never gonna wipe out the other one. Yeah, and and that's interesting too, because I think an evolutionary approach, I think all

their it would be interesting to get more evolutionary people like reading your work. uh, not that a lot and many people haven't, and many people have, but I think like this issue comes up in evolutionary approaches too, where it's sort of this idea that your own personal experience, um, it's not a a competition with a scientific approach. It's that the two are just two different things. And I think you can get there lots of different ways. And I think anybody who's taking sort of an evolutionary approach or even just a cognitive science approach, you know, you just your visual system, right? Your visual experience of like, you know, the shepherd Subterra or the Subterra Terra, right? Where like there's two monsters in this tunnel and they're chasing each other.

And actually they're the same size. They, their angle subtended on the retina is the same, but they're um, but you know, with with the visual uh integration of distance cues, like the one just seems bigger than the other. You know, like whenever you uh poke on how something works, then you realize that um I think I think what I'm hearing you say is that you have to go through this one experience, which is you have to realize that what we experience is not a description of how we work, It's a description of what we experience. And if you confuse those two, or if you're not willing to tease those two apart, you're going to be in trouble. But the second you accept that those are just two different things, how we work and then

how it feels, if you if you just keep those clear, you're you're gonna be okay. Yeah. Uh and of course, when you get, yes, exactly. When you get to consciousness and the philosophy, then you start running into terminological problems because then people will say, Well, yeah, but if you think there's how it feels and how it works, consciousness is how it feels. And the fact that it feels like anything at all is the question. Right. And then, you know, then there's a whole set of arguments about that. Well, never mind whether it feels like something. We say it feels like something. We cognitively think it feels like something. And we do that for a very specific reason because there's models in our brain telling us

that there is such a thing as feeling, uh, and and so on. So sometimes you have to drill down into these concepts to really get to them. But I think you're exactly right. It's interesting. I'm my brain as uh an evolutionary psychologist, as someone who has been stuck in the childlike phase of asking why, why, why, why until uh I lose my mind. As soon as you brought up the example of white light, I started thinking about wait, why? Why why do we experience white light as this kind of pure, uncontaminated cleanliness when it's actually from a physics perspective, the opposite? And I started rolling through possible explanations. And it seems to me that once you understand what what color is and what color adaptations

are, it begins to make sense because um certain objects will absorb certain wavelengths of light and reflect certain wavelengths of light. And it's really useful for us to compute statistical differences and the tendency of of particular objects to absorb uh this wavelength and emit this wavelength. And that helps us differentiate objects. Uh and we tend to uh um you know subtract out uh uh whatever the ambient lighting conditions are. So um when we see something as is red, it's actually sending us different wavelengths at different times of day, but we subtract all that out so that we can successfully differentiate objects. And when we experience something as white, I think what we're experiencing is the absence

of stuff that can absorb and reflect specific wavelengths of light. And for an organism that is particularly hygienic, interested in cleaning things, interested in organizing things, you want to remove all the put all the potential things that could absorb specific wavelengths of light and emit specific wavelengths of light. And once you've removed all of those things, you have achieved a kind of cleanliness, and that there's no, there's no adaptively relevant stuff that's in the stimulus. And so it seems to me that once you recognize that the function of color vision is not to accurately represent reality, uh, but to successfully differentiate objects and interact with objects in adaptive ways that the uh the intuitions about white

light start to make sense. It's only when you think about it as um uh uh uh when you think about f uh the function of color as uh faithfully representing what's out there, then you start to get into into problems. So once you you think in terms of how do we uh differentiate stuff in the environment? How do we categorize objects? How do we detect whether something is absent or present? It begins to make sense. Anyway, I'm I'm going off on a tangent. The ultimate question is: is there an explanation along the same lines for why it seems to us that uh computation uh uh does not explain feeling? That when we look at a neural network or an algorithm or uh ChatGPT, that it seems to lack this property that uh is so important and

fundamental not only to our our lives but our ethics. Where is this intuition coming from? So let me try to approach it in several different ways. And first, I'll start with this. If you want to build a self-model, and by model, of course, I mean bundle of information that depicts but also helps predict something. You know, like the visual system builds the model of an object, and it's not just a description of the object, it's also a prediction of what will happen if you grab it or look at the other side of it, and so on, and how so models describe and predict. Um if you want to build a model of yourself, of the stuff in your head, um, you run into a problem right away. Because if you want to build a physically, literally accurate model, it would have to

be information in your head that says there's 86 billion neurons and they're hooked together in this way and that way, and there's synapses between them, and then there's like chemical messengers and this and that, and this pathway from this brain area to that. And it's very, very complicated stuff that no animal needs to know in order to survive. It's like literally useless information. And so the model literally does not need, like, there's no adaptive value to building uh uh evolving um kind of basic inborn models that depict literally accurately what is going on in there. And so right away these models are incredibly simplified. They're detached from physicalism, right? They're not models of how physical computations are made.

They're models of how magic essences kind of inhabit a location without a real physical substrate. That's kind of how these models depict the events in our heads. And that's good enough. And any more than that would be harmful because then it just wastes energy and computing power to try to compute the model. So that's a huge uh um huge part of it, right? Uh maybe that's most of most of the answer to your question. But I I also want to get at from another angle and point out a different kind of model, which is very, very similar. It's also a kind of self-model, and it has much of the same um issues around it. Uh and the example which I often use in in my writing all over the place is the phantom

limb. And I'm sure you know you know all about the phantom limb, but the phenomenon is someday, unfortunately, some people get an amputation, they get an arm cut off. Um, but they feel like the arm is still there, and they can talk about it. And you can say, Close your eyes, tell me about that limb, and they'll say, Oh my god, it's still there. It's like sticking out, and I can feel it's hot, and then it itches, and and the joint feels stiff, and um and then they open their eyes and but of course it's not there. Oh my god, this is so spooky. Um, so phantom limb, what is a phantom limb? Uh it seems like a ghost object, invisible ghost object. So there's two classes of explanation the modern accepted one

and a much earlier one, um, which people actually did speculate about. And the modern accepted one is well, the reason is because your brain built a model of your body. And the model used to be accurate, more or less, but now it's not because the limb is physically gone, but the model still contains a model of a limb. And so when you ask someone about that limb, what's happening is their cognition is accessing deeper models that are automatic. It's not like they can intellectually control those deeper models. They're just automatic and they're constructing this information about the limb and sending it up to higher cognition. And then on that basis, the person can say, Oh, my phantom limb is doing this and this, and

that's the explanation. It's model-based, information based. That explains the whole thing. The second class of explanation is: no, there actually is a ghost limb growing out of the stump. It's an invisible limb. I think Lord Nelson, who lost an arm in battle, famously said, I know there's an afterlife and I know there's ghosts, because if my arm can have a ghost, then the whole rest of my body can too. Right? So the this other explanation is no, let's accept as literally accurate the output of this model. Let's accept as literally accurate that there's a ghost arm coming out of the stump. It's invisible. Objects can go through it, but it's there. Okay, and now we have a scientific problem. Like we all know it's there because we know it, because intuitively it's just there,

because the only information we that the brain gets is the information available to it, and that's what these models are piping up to your cognition. So that's what your cognition knows. So let's accept that there's a ghost limb. Problem one, or scientific question one: how is the ghost limb generated? Like, is it like cells? Is it like uh uh uh uh integration of information at the ectoplasm, actually, if you've watched Ghostbusters, the substance, yeah. Well that's that's question two. What is it made out of? Is it ectoplasm? Is it quantum mist? Is it an electric field? So one, how's it generated? Two, what is it made out of? And three, the question no one ever asks, but is actually more important than the

other ones. Once it's there, how does it impact neurons in your brain in such a specific pattern that you know that it's there? Elizabeth Elizabeth of Bohemia, right? Writing to Descartes, the problem of interactionism. That was that was her. How does the it's the joke is how does a dualist wake up in the morning, right? Like how does the how does the ghost interact with the machine? Yeah, exactly. Yeah, but it has to be such a specific interaction that if prints the like this neuron gets a little more active and that's a little less than just the right pattern, that it represents arms sticking out of stump. Yeah. And then you can uh uh form the thought and articulate it, right? So here's the problem how is it generated?

What is it made out of, and then how does it curl back and impact your brain in a specific way, right? So the intuition is there. The very powerful intuition is there because the deeper models are telling you it's uh it's there. And the deeper models are there because it's adaptive. Like we evolved to have models of the parts of our body. So um this is very adaptive. Maybe there's circumstances in which the model becomes so inaccurate that it's actually maladaptive in that particular circumstance. These models are not perfect. But then the scientific error, the heart of the scientific error is always at the very outset saying, let's assume that the model the brain generated is literally accurate.

And now let's try to find the scientific explanation for that. And that's that's where people right at the outset, 90%, 95% of the scholarship on consciousness gets stuck. Yeah. First step is the wrong step. I think if it's okay with you, like I think a lot of people listening, uh well, here's what I like to do. Um there's two questions. One is sort of what is the system that's producing these representations? So let's talk about like what is the system? Then there's a second part, which is why does that system produce bad scientific intuition, which is a separate and additional thing? And I think what you're studying is this unique thing of any kind of representational system is a hard nut to crack.

And it's in you also have a very interesting history of having done earlier work on other kinds of representational systems that are in this category of self-representation. And I suspect that's no accident either here. And this is very cool. So um, but let's get, if we can, let's get to let's sort of for now ignore even that people have been thinking wrong, that there has been this issue. but what is the computational system that we're talking about here? What does it do? And what is its job? And then I think that might be a good lead into once we've sort of talked about well, what is the thing we're talking about? Then we can start unpacking given how it works, given what it's for, when people are using that system to then do science, why does that then lead science

astray? Yeah, sure. So um part I think part of what's going on is as I said self models. So a system needs a good model of the thing that it's controlling or trying to influence. So you need a good model of the stuff in your world in order to interact with it. You need a good model of yourself to predict your behavior and understand, expect what's going to happen next. You need a good model of yourself. And there's lots of different aspects of yourself to model. And people probably have, you know, if you if you have a sense of your personality and who you are as a human being, that's in a sense a self-model. Um, and um but what we're talking about here is something really specific. It's essentially how the brain models, how you understand what it means for

your brain to deeply process something. Right. Can I can I maybe do an example or a question just to make it concrete for some? Yeah. So for example, uh to and to put it in for our listeners who might have listened to other episodes, this is sort of putting out a functional or software level description, which is that there are systems in the mind. So we're breaking, we're at the subpersonal level. There's no agent anymore. So one of the jobs of one of there's a system that's representing at a gross level, sort of what is the state of the system. So for example, right, um, that there is a decision to be made. Let's say, am I gonna go to this place or this place? Yes. And that decision being faced, obviously, there's all kinds of things

entering into whether that decision is made. And there's computations going on. But there's another system that at a or set of systems that at a very high level is saying the state of these other systems is that options are being considered, and the output of which option will be selected hasn't been chosen yet. Yes. So that will be sort of, and and part of the function of doing that would be, for example, not starting to walk yet if the decision is, um, if the decision is which way am I going to walk. Or if if we're talking about where are we going to go to dinner, it would be something of like, no, I'm thinking about it. Like that's what would give meaning to I'm thinking about it. It's that there's a system, it's not doing the job of making the decision, but it's

representing and talking to the system that sort of detects the output has been arrived at. And it's just saying, nope, we're not there yet. There's still processing going on. Yeah. So that's like a very mundane, concrete example that demystifies it. That one system has the job of representing states of other systems, not their full computations, because that wouldn't do anything. And but it has this uh affordance function of, for example, saying, look, we haven't made a decision yet, therefore we shouldn't proceed yet, for example. Yeah, yeah. So that's right. So uh if there's a a decider system, which is probably very complicated and spans much of the brain, if there's a decider, there's also a thing that

builds a description and a set of predictions about the decider. So you have, you might say, a schematic description of what it means for your brain to make decisions. Um, that's exactly right. And that somewhat cartoonish or at least detail poor description is I have an essence inside of me that magically makes decisions. And people call that will. So you were talking about free will, one of the topics you talk about. So these intuitions about free will are not really uh pre-scientific mistakes. What they really are is deeply ingrained, um, simplified but useful models of yourself. Uh, and so, yes, so decision making has a kind of schematic model of it that we build and that we can then articulate and talk about, and

we describe it as like a will inside of us. But there's another type of process in the brain, um, also very fundamental decision making, one of the most fundamental processes, and another process that uh cognitive psychologists typically call attention. And um attention, you know, it goes is so many different versions of what attention is and colloquial versions and this and that. But what I really mean is uh focusing, processing power on something, processing it deeply. That's what attention really is. Um, so you have tons of stuff in the room around you, and some of it you're focused on and you're processing it deeply, and some of it is being processed so shallowly it never gets beyond a few low-level brain systems.

So uh there are systems, so there's a way of directing attention. And directing attention is something that you need to control and move around so that you can do anything in the world. If you can't control your attention, you can't control your behavior. So there's attention and how it's being controlled. And then there's the brain's description to itself of what it's doing when it's controlling attention. So how does a brain describe attention and attention states and attention control. How does it describe that? And I would say it describes it as there's a weird, non-physical magic essence inside of me that can focus on things and experience them and have and grab them. Like my mind can possess this thing right now and then switch to that thing right there

and and it and when I do that it allows me to know about the thing and to respond to it and to experience it and feel it and so on. That's the brain's way of describing what it means when it pays attention to things. And so That's yes. Yeah. And if I can make another just concrete example for listeners, then you can tell me if this is so. So, like a good example would be something like um, let's say this actually happened to me. I was walking home from school one day talking to somebody and walked into a tree, like when I was in high like elementary school or something. And right, the intuitive idea is like I wasn't paying good enough attention to that, right? Like, so I need, and so the idea would be that again, we're in translating this intuitive

idea now into a software functional level description, is that um that I walked into a tree. So that is an event that's important. Their mechanisms representing I just walked into a tree, right? And again, this is like no free lunch principle is you we can't take for granted that there's a system or systems that have been modeling what has been the primary or what have been the important kind of real-time um contents out in the world that were that that many systems were were handling in the triage, right? So I was to bring it back intuitively, it was like I was having a social interaction and I therefore missed a physical interaction, right? Yeah. So this system has detected an error of the prioritization of a physical danger

got missed by and the social systems were were getting more of the um prioritization in this sort of finite, you know, uh uh, you know, muscles can only do so many things, right? So in action planning and et cetera, all these so the the the influence of the physical hazard system on motor planning and modeling was just wrong. And so the system just detecting, okay, we we made a mistake here. And this and then consequently there's changes to that. And then the high level description of that is I wasn't paying attention to the right thing. But in fact, the translation is what I just said. Yeah. Yeah. And of course, the high-level description, people typically conflate attention and consciousness.

And so people will say, well, I wasn't aware of the tree, or I wasn't conscious of the tree. I wasn't paying attention to the tree. And actually, those tend to go together: attention and consciousness. So there's a wonderful phenomenon that, of course, you know about called inattentional blindness, where attention, focusing your neural processing on something and being conscious of it, usually go together. I mean, you can separate them as it turns out, but they usually go together. And um, and of course, the famous example of inattentional blindness is the old gorilla in the basketball game, where just for the listeners, where you tell people here, look at this video and count how many times a basketball switches hands in the game.

And they're so focused on that that they don't see the video properly. And then you say, okay, I'm going to show you the same video again, but just take in the whole thing. Don't look at the ball so much. And they look at it and they're like, This can't be the same video. There's a guy in a gorilla suit running through the game. I didn't see it before. And the reason why they didn't see it, even though it was getting in their eyes, it was getting in their visual system, is because they weren't paying attention to it. And so when you're not paying attention, you are not conscious of it. If you're paying a little bit of attention, you might be a little bit conscious of it, and so on. So inattentional blindness.

Um, and what I think is going on is that attention is a mechanism, it's a process, it's the brain focusing more on this than on that. Consciousness is the brain telling you from the inside about what its attention is doing and what it's paying attention to. And those are usually, they usually go together, but not always. The brain's concept or model of what its attention is doing can sometimes make mistakes. Right. Yeah. So I want to get into the functional specifics of why we need a model of our own attentional state. So it makes sense to me why I would need a model of physical objects in my environment and their orientation to me in space, how far apart they are, even what color they are, and

how I can differentiate them such that you know red things pop out from a green backdrop. That seems like an adaptive model if I'm foraging for food, say. Um uh it makes sense to me why I would even need a model of, say, uh, your mind or Dave's mind, such that it allows me to anticipate if you're going to take a punch at me, I can duck. Uh, if you're going to say something uh that uh upsets me, I can come up with a comeback in my head before you finish saying it, so that I have a witty retort. So it it makes sense why I would need objects, uh models of objects in my environment and even models of uh other people and their uh uh mental states and the likely consequences that would flow from them.

But it's not as obvious to me why I need a model of my own mind or my own uh attentional state. Uh because that's just gonna go and affect my behavior regardless. Why why do I need to uh uh uh a model of my own uh uh attentional state um uh as opposed to say someone else's uh mental state. Yeah, excellent question. I'll give you several specific examples. I was gonna say three examples, but then I'll probably forget the third one before I get to it. So I won't be quite so specific. Several examples. One, um, you're at a party and there's a big plate of donuts next to you, and you don't want to pig out and eat them. And you know, because you have a predictive model of your own attention, you know that

if they're next to you, they'll snag your attention. And if they snag your attention, then you're gonna have a hard time um resisting eating them. And so you make the very simple choice of either turning your back or going into the other room, because now they're not going to be foremost in your attention. That requires that's a simple control of yourself on the basis of predictions. And the predictions require a predictive model of how your attention interacts with your environment and your behavior. So you got to know something about how your attention works to be able to do that. Here's another example. You're trying to study, and you have your textbook, whatever it is, I don't know, quantum

mechanics, whatever it is you're studying, uh, and you're trying really hard. Um, but there's uh an annoying conversation in the room over there. And um, and you want your attention on your book and not over there. Um you don't want to totally shut off the conversation from your attention because what if they start saying things about you? And so you want to keep like a quarter of an ear out, but you most of your attention is on here. How do you do that? Well, as you're working away, you realize, ooh, wait, too much of my attention went over there. Control it back. Focus, focus, focus. Oh, wait. I got so into this paragraph I totally lost track over there, and I still want to keep a little bit of tab.

So redirect. What I'm describing here is a control system that targets an ideal or desired state. And to get to that state and maintain that state, it has to monitor it. So it has to have a representation telling it this is my state. This is the state that I want. Or this is my state. And now there's a gap or an error between what my state is and what I wanted it to be. And now I have to readjust it. So all of that requires a model. In fact, probably the most important reason for a system having a model of itself is to control itself. All control systems need a model of the thing they're controlling so they can monitor it, so they can error correct and so on. So that's a second example. Um a third example comes down to what you were saying about interaction with people.

It's very important when you talk to another person to understand their state of attention. Because let's say I'm interacting with you. Right now it's simple, we're just paying attention to each other, I hope. But in a in a not in a different setting, I don't know, we're at a picnic and there's all kinds of things around us, and kids running around, and food, and this and that. your behavior, as far as I'm concerned, your behavior is 90% determined by your attention. Because what you're paying attention to, you're more likely to choose to act toward. And what you're not paying attention to, you won't react toward. And if I see some kid flinging a football at you and you're not paying attention, I might

warn you. I'm like, ah, watch out. Because I think you're not going to react, because I think you're not paying attention. And therefore, I predicted how you'll behave on the basis of your attention. And if I see your attention slyly going to that last cookie that I kind of want, that tells me I better run in and grab the cookie first for myself, right? Because I'm predicting your behavior on the basis of your attention. So it's very, very important for me to model and predict and understand your attention. But think about how the interaction works. It can go layer after layer. I predict that you'll behave in a certain way. I predict if you behave in that way, I might react in this way. I predict if I react in that way, then you might react in this way.

And each one of those steps requires understanding first what your attention is likely to do and how it will affect your behavior, and then what my attention is likely to do, and how it might affect my behavior, and so on. So it's not enough for me just to have attention that controls my behavior. I need models, predictive models of it, so I can engage in futuring, so I can predict what I'll do next, what you'll do next, how I'll react, how you'll react. So having a good model of mind is for myself and you is crucial. And maybe the most important component of mind is knowing what is emphasized, what is uppermost in your mind, i.e., what are you paying attention to, and what's the consequences of that for your behavior.

So those are some examples, three really specific examples of why we need to have a really good, rich predictive model of of attention. And it was three. It was three. You did it. Um and what I'm hearing you say too to to for listeners is it's a it's an information processing version of we can make this concrete with motor control, right? Yeah. Where sort of you you don't have to um trial and error do physical things you know so if you just if I put my arm behind me and kept it behind me and then I sort of would say well what's the way my right arm you know what's the way I'm gonna pick up this coffee cup um I don't have to like try to sort of pick it up I know I'm not gonna be able to right so all

of the sort of stumbling around trial and error things that like a roomba still does you know like we look at a roomba and it's like it's so stupid because it's bumping into the wall tries again bumps into the wall bumps again It doesn't have a sufficient self-model to run through. If I did this, would I still bump? If I did this, would I still bump? If I did this, and then selecting among all of those, the one that doesn't lead to the bump, it doesn't do. And that's why it looks stupid to us because we do that, right? So the ability to represent model if my limb, if I did this with my limb, would it work? If I did this with my limb, if it worked, if you don't have a self-model of the limb, then

you're not gonna be able to produce that. You're gonna have to do trial and error Roomba stuff, which would look really weird. Yeah. And what I'm hearing you say is that's a concrete, well, it's actually not so concrete, but at least it's more intuitive, I think, for people that oh yeah, you need a model of your arm such that when it's broken, or when we're babies and they're only this long, and now we're adults and they're this long, you know, or something is different, or my hand is tied, or I'm handcuffed, or whatever, right? Or there's something in my way, I can model then what other behaviors I should do if I want to catch something that's falling or if I want to avoid a ball that's thrown at

my face. And your argument is just like that for the motor system that the motor system needs a self model of where are my limbs, how long are they, are they encumbered or not? The same thing is true for classes of information processing. Yes. And even though it feels more abstract and we don't really have a vocabulary for it that is anything other than these intuitive categories, it's the same principle that in order to effectively motor around with this control system, interfacing with the world, having goals, realizing that my goals might not be met, or the system might have the metaphorical hand time behind its back. Like, I don't want to eat the donut, right? Like one of my goal systems is I don't want to eat the donut, but I know some part of or

the system knows that certain systems are saying eat the donut or will produce that output. So the system can anticipate that. So it's a similar self-model. Yep, yep, exactly. So yeah, and also I would say yes, there are multiple models because there are multiple kinds of information processing. And so, like I alluded to before, a model of decision making might give us our intuitions about free will, and a model of attentive processing might give us our intuitions about um Experience the intensity of a person, you know, prospective experience. We probably have models of our memory. I mean, we have a good concept of what what how our memory works and what we're likely to remember or how we're likely to forget.

So we probably have good models of our the processes involved in long-term memory and short-term memory and so on and so on. So we have all these all these interesting models of processes in the brain. And the models are really necessary. They're really useful. They're not silly mistakes. They're really useful. But they're not scientifically accurate depictions of the processes in the brain. And that's the thing that I think um Scientists have to get their minds wrapped around. Yeah. And how far? So I mean the motor stuff, I mean, there's a lot of, you know, that's one area where there's lots of good work. People understand this is an engineering problem. They understand the problem and the function.

So I mean, Michael, what you've already done is already, you know, you're solving one of the biggest, you know, Gordian knots of philosophy in a way. So that's plenty to do. But I imagine it's frustrating. But I'm just like one question is ignoring all of that. is there how how far have you or other people gotten into what are these models? And you can imagine doing it in a problem-based approach, which is sort of, you know, like, well, what are the information processing problems a self-referential organism like humans would need to solve? Um, what mistakes could it make, or what conflicts could there be in systems? Another one would just be like starting with introspection and or meeting the two in the

middle. Are the are there have we are there starting to be models of what the models are? Um I think it's first, it has been tricky. Yeah. Uh even in the motor system, as you say, it's much more concrete, it's much more engineering and easy to wrap your mind around. Uh and I spent years and years studying that system, which is of course how I got to the other system. Yeah, exactly. Yeah, yeah, yeah. So um, but even there, even though there's lots of work and lots of acceptance of the idea that the brain builds models of the body, I don't think anyone can point to a spot in the brain and say the model's there. Right. Uh or say this is how the neurons compute the models specifically. Like that gets very hard to do.

And um, in the case of models of self or models of my of information processing in the brain. Where do the where are those what are those models? Um I think the best that we've been able to work out is that they're they massively overlap the social cognition networks that compute models of other people's minds. It all seems to be the same kind of stuff. We're building models of minds, attributing them to other people to ourselves. And there's a certain vagueness in trying to pin down exactly what the models are. I think where the um progress is being made right now, uh, in my opinion, is in artificial systems. Right. Uh, because now, finally, for the first time, you can have systems that literally build

models, really rich models of other people's minds, maybe in some way of themselves, although this is still debated. But you can go into those systems now and you can see exactly what they're doing. You can look at the representations in their hidden layers and say, ah, look, this is this pattern of activation that represents this, and this pattern represents that. And now we can see how it's doing this and that. And so you can begin to understand the principles by studying these artificial systems in ways that I don't think people could studying the biological system. So I see that as, you know, consciousness research is becoming more and more a matter of AI, which is interesting. Yeah, and just and just to maybe a slightly funny thing.

So on this account, and you can tell me if this is wrong, The thing that people are worried about, which is like a machine becoming conscious, what that actually will look like, it's actually going to be quite boring because what it will actually be is it'll be systems that have very cursory representations of the states of the system. And it will say things like, I'm currently paying attention to this and not to this. It's going to be very mundane. It's not going to be this all of a sudden glowing light, um, you know, and all of the world's knowledge flowing into this Pandora's box, like a Ghostbuster scene. Yeah. Um, so that this is a very deflationary view of consciousness. Again, probably for the reasons that um you alluded to the beginning maybe people might

not like because there there's another agenda um and we can get to that in a minute but but is that right that basically the idea would be that really what consciousness in an AI would be is a very boring prosaic thing in a way. Yeah I think you're right. I think that um the first so f uh part of the confusion comes from differing definitions of consciousness. And some people think oh consciousness is when you wake up and have the ability to choose to do anything you want, but you have personal desires and you care about yourself and distinguish getting all kinds like piles and piles of things get added to this concept. So they think if you make AI conscious, that means it will wake up and start killing us all

because it'll want to protect itself against us and it'll get mad and um and it'll have all these feelings and um it'll murder us all. Uh and it'll be able to make decisions because when you're conscious, you can make decisions, and right now it doesn't make decisions, it's just some mimicky thing, right? And and so that's a very strong concept people have. And so you ask them about making AI conscious, they look horrified. Um, But on the other hand, I would say AI already makes decisions, very sophisticated ones. It can kill you just as dead, whether it has a subjective experience of doing it or not. The consciousness question is actually orthogonal to these worries. And it's also orthogonal to emotion, like whether or not it gets

upset about how it's being treated is irrelevance to the question, the larger question of what is this consciousness thing. Um, so I think you're right when I talk about AI having self-models and therefore having the thing we call consciousness, I'm talking about it having this very limited kind of model that tells it about its own um state of attention. It's what what is it focusing its resources on more now versus uh before, here versus there? And exactly how does it model that? Because you can model that in different ways. Um well, I think the heart of the problem, the conceptual problem that is so confusing, and I admit this is very confusing. People ask, okay, so if you give it information about what it's paying attention to,

how does that make it have an internal feeling? How does the internal feeling emerge? What is that like? Why? Um, and the answer is it doesn't give it that, nor does it give us that. But if it has a particular kind of model of itself, that particular kind of model will describe, it will have information in it that describes the machine as a thing that has subjective experience, and therefore it will think it has subjective experience, and therefore it'll tell you it has subjective experience. That's what it could have. That's what we have. Yeah. But it's a special kind of model that it would have to have to give it that information about itself. Yeah, I love I love that. I can and maybe I can um again translate or maybe to talk a little bit about an

evolutionary approach to that, which is I think what you're describing is so what we're talking about is what consciousness is a self-model. And that's the scientific sort of that's what it's doing. The thing about, well, if if machines become conscious, then it's gonna uh you know care about itself and try to take over. That is the concern of an evolved organism that has evolved to have representations of what do conspecifics do, right? It's populated with I'm dealing with a complicated, evolved human conspecific who is self-preserving, who wants status and resources, and under the right circumstances will try to kill me or not, you know, depending. And so I think that it's no accident that we probably have

very complicated systems that are assessing degrees of consciousness because that's a system for representing, you know, is David asleep? Well, he's not going to try to go after me and my family, but if he's awake, he will, right? And the fact that it's sort of pre-populated with this default of it's a it's no accident that what we mean by consciousness is the full range of human capacities. The argument would be that that's an evolved representation because what you're doing is you're tracking: is David asleep? Is he drugged? Is he not paying attention? And the full thing is not a representation of how he works. It's just the default assumption of like, what's the worst case scenario? He might be a powerful, malicious, intentional human actor

who has power, who wants to go after me. And that's exactly what people are concerned about with consciousness. So I think it's a nice example of like the scientific model of the self-model combining with then you just have this evolved organism who what it's trying to do is assess is David asleep or you know what's his intentions? But it's no accident. The default is it's all the human stuff, right? Yeah. Yeah. Yeah. So I think there's a huge part of the way we think about consciousness that is richly intertwined with ethics. So if humans are conscious and animals aren't, then it is ethical to just treat them as objects, to torture them on factory farms, whatever, it doesn't matter. They're just unthinking automata.

You know, Descartes famously thought that uh animals were just automata. Um and uh it seems to me that th this question of how far in the animal kingdom consciousness goes is um inextricably intertwined with how we should ethically treat animals. If bugs are unconscious, it doesn't matter if I swap them or or whatever. If they're conscious, oh boy, what a nightmare. We're surrounded by these sentient beings flying around us, and we have to consider their interests. And so I want to ask you about like how your model of consciousness fits in with the ethical nature of it. And um it it seems to me that if AI is conscious, the the aspect of it that we really care about is like, does it have pain and pleasure?

Does it have feelings? Uh, does it have affect? If I am mean to ChatGPT, will its feelings be hurt? Right? It's not so much does ChatGPT have a model of its own attentional state. I don't really care about that much. What I really care about is does it have feelings? And so, how does the the affective and ethical dimension of consciousness fit into your your way of thinking about it? Yeah. Yeah. So, oh boy, there's lots of things I could say about that. Um, well, I'll give one one point. Um, theory of evolution by natural selection has been used to argue for and against a vast range of ethical propositions. Some very problematical and some very lovely and nice. But the theory itself is actually ethically blind.

It has nothing to do with ethics. It's just this is how it works. Um, and to some extent, the theory that we've developed this attention schema theory or this way of thinking about consciousness works the same way. It says this is how it is. um, ethics aside. And from that, I think people could come up with all kinds of different ethical assertions. Well, we have to do this, we have to do that, this is conscious, we have to do this, this is not, so we have to do that. Um, so that's the first thing I would say is I don't think the theory prescribes what we should do or not do ethically. It just gives information. Ethics is something else. We're all ethical beings. We all have, and ethics are fundamentally irrational.

And they're rooted in these kind of um built-in uh you know um emotional states. and that's great, and we all have them, but I don't necessarily know that the theory dictates the ethics. Uh, what kind of animals are likely to have this kind of stuff in them? Um well, that's not that hard to trace out. So along the vertebrate line, at least. You're looking at animals that have the kind of attention that needs this kind of control system that needs really sophisticated models. That's the first thing you would look at. And so you're probably looking at all mammals and all birds, and maybe many reptiles, if not most of them, and somewhere in that range, maybe some amphibians. I don't know if amphibians um four brains and

really sophisticated covert attention is probably non-existent. So maybe they don't have these mechanisms. But it's a huge part of the animal kingdom that has some version of that, undoubtedly. The second question you're asking, though, is about emotion. So we can be conscious of all kinds of things. I mean, classically, people like to study visual consciousness because it's easier. Consciousness, the conscious experience of red, or uh the conscious experience of getting poked with a pin. But um uh emotion is a thing of which we can be conscious. Uh, It is not itself, it doesn't define our consciousness. It's just one of the things we can be conscious of, but we can feel sadness, we can be upset, we can be angry.

And so you're asking about AI. You're saying, well, what you really care about is does it have emotional, is it does it have emotions and is it conscious of those emotions? Right. And so it's an interesting fact that people have um emotions and are often not conscious of them, as you find out if you ever go to a therapist. Um they're always trying to get you to be aware and conscious of emotions you didn't know you have, and then you're like, oh yeah, I guess I was angry all day. How weird is that? Uh but um uh so there is a difference between the emotional states um and being conscious of them. And now you get to the question of AI. What we worry about is can it have emotional states and would it be conscious of them?

I think at some point the answer will be yes. I think probably right now the answer is no. Because emotional states are are complicated. Right now AI is really good at what we will call cognition and visual perception it's getting really good at. But and language processing. But emotion emotion has so many complicated layers in biology in human systems and other animals that are not that well understood. And so those layers start with the physical body and the autonomic nervous system and the flight or fright systems, or the resting and digesting systems that change the physiology of your body, and then sensory neurons that detect those changes and carry that information up into deep systems in your brain, and those builds

uh all kinds of information structures, and then eventually that gets up into your cortical systems and more into your cognitive networks, and you start interpreting those and saying, Oh my god, I feel this emotion, I feel that emotion. I think AI currently lacks most of those layers. And so, at least right now, it's probably not an emotional thing, an emotional agent. Um, but will people eventually fare all that stuff out? I mean, aside from the question of should we, yes, people will probably figure it out and probably build artificial stuff, because that's what people do. They're gonna build it, then we're gonna have AI with very human-like emotions and very human-like consciousness, and then we're gonna have to deal with the ethics.

Uh, and I don't know what the solution to the ethics is, but it's a it's definitely a um a quagmire, I guess is how I would describe it. Oh give it a go. So I do you see um if a separation between consciousness and feelings? So could you be conscious without having feelings? Could you have feelings without being conscious? Yes. Um uh some well, some people distinguish feelings from emotion. And uh and the terminology is a little subtle, but some people have proposed that the word feeling be limited to emotions of which you have become conscious. Okay. And um emotions can be processed at unconscious levels, and there's lots of brain systems and the amygdala and the hypothalamus and the this and the that that are computing these

states that um affect your body chemistry and your body physiology and affect all aspects of your brain, but it's going on unconsciously. So some people would make that distinction. Yes, of course, I think you can be conscious of, like I said, you can be conscious of colors. You can be conscious of a poke on your arm. You can be conscious of a sound, you can be conscious of a thought. You can think two plus two equals three and have a conscious experience of thinking that thought. And all of those could, in principle, be totally stripped of emotion. So the emotion doesn't need to be there to be conscious of something. You would still, if asked, you would still be able to say, there is a me. I have a subjective experience of those things.

And if someone asked you, yeah, but what is the emotional component to it? It would be possible to say there is none. And I think there probably are people who are like that, because there's a huge range of people. You get people who are so disconnected from their emotions, um, through whatever, you know, some subvariants of autism, for example, or other um conditions where people are are not conscious of their emotions. I think you can have uh uh people, minds, life experiences that are really shockingly devoid of emotion. They're really different from the way most of us operate. So I do not think consciousness and emotion have to go together. Um, but for most of us they do, of course. Right.

And when they do go together, you get feelings. So an emotion um is uh doesn't need to be conscious, but when it becomes conscious, it turns into what we would call a feeling. Yes. If you, you know, according to that terminology, yes, by definition, a feeling is a conscious emotion. Okay. All right. Cool. Um Yeah, it's it's super interesting to me because I I feel like what we ultimately care about is um is feelings in terms of you know whether AI is quote unquote conscious. But do I care if like an AI has an emotion but it's not conscious? That's that I don't know. It's kind of a weird question. Like if it's mad but it doesn't know it's mad, do I do I care? I don't know. Yeah. Yeah. So um if you poke yourself with a pin, yeah, um, but

something has disconnected your cortex from that experience. Yeah. Uh you know, all the pain processing is going on, let's say all the way up to the subcortical level, but you're not aware of it. Is it pain? I don't know what to say to that. I mean, that's a philosophical question with no real answer. Yeah. Yeah. And I I think it's a pro I mean, m it seems like it's a proxy for is this a severe event? So in other words, like I think the the intuition and the philosophy is not considering all possible worlds. It's it's uh assuming the the evolved psychology we have, which is specific specifically um their emotions are emotional systems or situation detectors, right? So, like there's a certain amount of pressure or incursion into my skin at all times.

But then when a pin pokes me, I react, but I'm also aware that I'm experiencing pain and I'm reacting, right? And so the idea is that the self-modeling system is detecting such an important or severe event that it even is registering in the self-model system that this is this is the state we're in, and that probably has other consequences too. That's why it is activated. Yeah. So when we say something like, Will AI be conscious? What we're really saying is, Will it have such severe feelings that it would it would pass the level of the it would engage the self-monitoring system in us. Because, you know, the things that are important, like pain and sorrow, and disappointment and terror and anxiety that are important to me become self modeled because

they're so severe. So I actually think that it's a very human-centric thing to say consciousness is what matters. Yeah. Because it's a readout of the fact that they're conscious because they're a severe level of that. And that's why they're conscious, probably because it just it adds to the inferential machinery that, like, oh, we better stop all the other stuff we're doing, call all hands on deck to deal with this pin in our foot, or the wife who left us, or, you know, the child who died, or whatever. Um, you know, it's not just the mundane scrapes of of everyday life. So so I actually think that it's a readout. People saying consciousness, what they're actually saying is the readout of it's the

severe events of people. But the consciousness is sort of the confabulation. I don't think that's actually what matters, if that makes sense. So I if I could uh translate what you're saying, it seems like um because intense emotional experiences like intense pain, intense sorrow, um by design activate consciousness, activate full attention and a representation of that attention. Yeah. That we conflate um consciousness with feeling because feeling uh automatically by design activates consciousness. And whenever you have intense feeling, you automatically get consciousness of it. And that's that's why it's so intuitive that those things are conflated, but they don't actually have to coexist. Yeah.

Yeah, I think that's right. And again, I would say there are probably people who aren't wired that way. There are people who just are not emotional people, or they are, but it doesn't, you know, there's a disconnect somewhere in there. There's a neurological uh abnormality. So um, but here's an interesting point. Almost all the discussion so far, not quite all of it, between us on consciousness has been this personal thing. I am conscious, or that machine is conscious, or but but but consider this. Um maybe of equal importance is whether when you interact with that AI, whether you project consciousness onto it, whether you see it as a conscious being. And that's that's the special thing. I'm not talking about intellectually interacting with it and then deciding, ah,

I think it's conscious, or no, I think it probably isn't. I'm talking about that intuitive feel that you get when you interact with what you think is a conscious being. Um, and one of the examples I often give is ventilquism, where we all know it's not conscious, but we still get that sense anyway. We it it pokes our buttons, it makes our social machinery go into overdrive, and we generate a a model of a mind and project it onto it, and now that to us that becomes a conscious being. Um, and that's a very important threshold, um, because that's how we interact with each other, first of all. Uh, as many philosophers have pointed out, I actually don't know whether you're conscious in some absolute sense.

I can't know that. But I build a model of a mind and project it onto you. I, in effect, perceive consciousness emanating out of you because that's how my brain has built the model of the world, and that helps me interact with you. And we do that to other people. We do it to our pets. Some people do it to their houseplants. We do it all across the board. We live in this sea of perceived consciousness, conscious minds all around us. And when AI, well, not when, it already is entering that community. That is more and more people interact with AI and at a gut level perceive it to be a conscious thing. They interact with it that way. And they do that every time you talk to another person because you're exercising and

using your social machinery, you're also shaping it. and very slightly tweaking it and retraining it. Um and so when you interact with a a chat bot or one of these companion bots and it's changing who you are as a social being because it's an a social experience you're having with it and how that affects how you interact with other people and how you think about other people and how you feel about the social world around you. Now that's a really interesting threshold that we reach. So to just sort of say what you're describing is that we have models of others. Yes. And one important problem is that if you think about it, you know, animals are complex agents. And we have agency detectors and inference systems, but there's different kinds of animals,

right? There's like complicated animals and simple animals. There's also like humans. There's babies and adolescents and adults, and then like, you know, people, you know, you can be sleepy. Somebody might not be able to hear. So we have these agency detectors, but to some degree, they are also designed to be able to learn whether it's a specific person, like, oh, Sam can't hear, or this kind of species. If I'm a hunter, you know, they detect this part of the world, but not this other one. And I can take strategic advantage of that. So there's this agency detecting system. Maybe the full-blown version is human level, but then there's always these sort of modifications. But what you're saying is as we interact with technology, that's spawning a new, like this

is another kind of agent to interact with. And then when I interact with you, Michael, and you don't just give me per, Then I might evaluate you more negatively because I'm comparing you to this artificial thing that I've now spawned sort of automatically as this is an optional kind of agent I could have. Yeah. Yeah. So uh or to give another example, So social contagion is a very well understood uh phenomenon, and you can have a company where there's one person who's just so cheerful that everyone around them gets cheerful, and then they make other people cheerful, and pretty soon the whole company has this uh measurable uptick in mood because of that social contagion, Um, and it can go the other way.

You get the really sort of cynical jerk, and it brings everyone down in this waves of spreading um contagion. Uh but we have now these agents that are like one type of agent trained in one way, now has the ability, well, has access to has uh fairly extensive interactions with maybe a billion people around the world. Now, what kind of social contagion can happen as a result of that in terms of just the subtle habits of how we interact with each other? Uh I mean, we know that's happening. We know that has to be something that's completely rewiring our our um social interaction patterns. Yeah. So what you're saying is it's not we should be less worried about AI consciousness and more worried about the consequences of people automatically and implicitly sort of

ascribing consciousness to AI. Consequently, it's actually contaminating our representations of what agents are, what to expect from them, or what what what we should all do now that there's this agent who's acting in this way. Right. Yeah. Yeah. I wanna I wanna get your thoughts on um uh religion and spirituality and how it ties in to uh your thinking about consciousness 'cause it it gave me an idea while while you were talking about modeling other people's minds. Yeah. Um and it occurred to me that you know we live in an extraordinarily social species where every tiny little thing we do is scrutinized by other minds. And even if we think nobody is watching, there's a good chance that someone might be

watching or might overhear us and then gossip about us. And that could totally destroy our reputation or totally, you know, uh enhance our social status. And because we have this just ambient uh uh surrounding environment of consciousness that is scrutinizing us at all times. And it is useful to sometimes overperceive that ambient level of consciousness or ambient level of attention on everything we do. Um, it becomes very intuitive to just personify that, to say that we're just we're just surrounded by consciousness, and there's just an inherent consciousness that's that goes with me wherever I am. Uh, and I'm just gonna call that God, right? Or I'm going to call that the spirit world or something.

And because we we we live in in an environment that is so hypersocial and is so awash with um extreme attention on everything we do, it can be it becomes very easy to overperceive uh consciousness and to think that we just live in a universe that's just made of consciousness. So what what do you think what do you think about that idea and that that way of thinking about spirituality? Well A, I think it's essentially correct. Okay. And B, I have written pretty much those very thoughts many times. Uh, yes, I think I sometimes think we should have been called uh homo socialists instead of Homo sapiens, because we're not very wise, but we're sure as heck social. Uh but yes, I think our success as a species, the way that we're so cooperative,

is because we build models of minds at the drop of a hat and then project them onto everybody. And I think basically this is a case where you know, like evolution made the giant giraffe and the and the big horns and the big thing here and the giant peacock tail. I think evolution got hold of that uh that volume knob on the social machinery, on the machinery that makes us project conscious minds onto everything and it just cranked it up to 11. And so, yeah, there's always a bit of a downside when you crank um a feature up that high. But the point of evolution, of course, is it finds the balance point where the um you you get um benefits, and the benefits increase and increase, and the and the downside

increases a little bit less, and then you crank it all the way up to the point where if you cranked it any further, then the downsides would start to have a delederous effect. I think that's where we are. I think we attribute mind to things so readily because it is valuable to us, because it made us the cooperative, incredibly capable species we are. And one consequence of that is um we see mind in empty spaces. We see mind in um you know a lightning strike. Oh my god, uh a malevolent spirit did it. Um a tree falls, oh my god, that tree had a spirit and wanted to crush my house. You know, the river spirit that carries people away. Uh every event we see, we have a tendency to attribute a motivation and a mind process and a consciousness as the cause of it.

We're very, very prone to that. And yeah, I think God, the concept of God is basically the extension of that to the universe. So that's to say, your behavior is up to you. This other person's behavior is up to them. But then there's all this stuff that happens that people didn't cause. But somebody must have decided, somebody must have had the motivation and the consciousness to make it happen. So maybe that was God or spirits. I think that's that, yeah, religiosity is not um a mistake. It's not um some, you know, as many hardcore atheists would put it, uh, would describe it as some kind of pre-scientific stupidity or something. I don't know. There's lots of mockery of it. And I do not believe that.

I think that religiosity is something really deeply ingrained in us that we cannot get rid of it. And um uh there's no reason to, in a sense, no reason to try to get rid of our our um evolutionarily built-in models. Uh I think some of us have intellectual knowledge so that we don't cognitively believe this or that. But probably every one of us, if you put us in a dark, spooky room, start imagining uh creepy things behind us. Like we all we we we all get mad at our computer when it doesn't work right. Um we all fall for the ventriloquism illusion. Uh when you read a story, the characters in the story have personalities because you're projecting it onto them even though they're not real things.

Right? All of these are acts of effectively spiritual belief. They're all acts of a social machinery building models of minds and projecting it onto things that don't actually have minds. Yeah. No, I love that way of thinking. I mean, it it it makes me think that even the psychology of storytelling, that I have to model a sentient being uh telling me about other sentient beings and their motivations and their betrayals and whatnot. Um you could easily just extend that onto the universe itself, and that everything is just a story being told in the mind of some sentient being or being imagined in the mind of a sentient being. And it becomes very easy to just put everything into that frame. Everything everything happens for a reason.

It's all part of the grand narrative that is being told uh by this mind. Um it it th this hyper sociality and this um constant need to model the mental states of others, I think makes religious ideas very sticky and intuitive. It makes them cognitive attractors, um, as you know, Sperber w would put it. Um so I I I love this way of thinking and how it sheds light on some of these uh sticky religious intuitions. Yep. I can just poke on that too. I mean, I think this also means that this psychology of agency and inferences of what would an agent do or intention, it also means that getting back to the science thing, I think it's people's default way of understanding large-scale social dynamics, right?

That in I'm I'm for whatever reason, I'm not trying to humble brag, but I'm listening to Karl Popper on audio tape, like or like on CD while I'm driving around my car. And his book, Conjectures and Refutations, he talks about people's models of social sciences. The wrong version is the conspiracy, which is exactly what we're talking about, which is outcomes, social dynamic outcomes, are because there's like a cabal of people who intend for something to happen. And it's this idea that there's sort of this conscious desire to make a outcome happen. And Popper's argument is that most large-scale social dynamic outcomes are unintended consequences. I mean, not always. Sometimes people succeed. And so social science, a large part of like modern social science is agent-based

modeling or formal models of complex dynamics that nobody sees coming. And so I think like getting back to the science thing, we use this evolved psychology of consciousness or modeling an agent to then apply it to somewhere it doesn't really belong, which is like macroeconomic, you know, macroeconomics or institutional dynamics. And modern science is sort of saying, no, that's the wrong tool to use. But then it suggests that what Michael's studying and what we're talking about is the psychology of thinking wrong about large-scale social dynamics, too. Yeah. Yeah. Yep. I mean, the only caveat I would say is um in talking about it as thinking wrong, sometimes it is right in the sense that it is

useful. Yeah. I mean, that's why it evolved. Sure. Um, even though it can be wrong on the details, the factual details. And of course it can be overextended. But that's one point that I often like to make is that these models they're there for a reason. They evolved because they're useful. Yeah. I mean, it it gave me another idea about um the intertwining of of religion and morality. Um, if you think about like you know, we have these intuitions that moral uh norms and moral rules are ha are like somehow built into the fabric of the universe, that they're objectively correct, that it's wrong to steal or to kill or to lie or something like that. And the wrongness doesn't depend on anyone judging it as wrong.

Um, even or it doesn't depend on um you know whether someone's watching me when I when I steal. Or even if you know my culture says that stealing is okay or killing is okay, I have the intuition that it's still it's still somehow wrong, that there's still some consciousness somewhere that's saying bad, bad, bad, that thing is bad. Um and uh it's really hard to reconcile that way of thinking with um the sort of physicalist scientific way of thinking. You can think of the pro you know the the evolution of science as the as the constant batting down of our of our mental modeling machinery, saying, no, that's not, that's not the place for that stuff. You know, stop thinking about consciousness, stop thinking about minds and intentions and

conspiracies and spirits. We just need to think about physical stuff and their interactions. We need to turn off your folk psychology and turn on your um your folk physics and your folk tool use and and and that is is how scientific progress has occurred and and consciousness is sort of the last bastion of the of folk psychology. You know it's it's like the last thing that we're still clinging on to that um that that science has not yet defeated. You know it's because it's it's at the heart of of uh of spirituality of storytelling of our social lives of everything and to um let that fall to folk physics and folk tool use is is just deeply threatening. Yeah. Well, of course, the trick is um not to let folk

psychology fall and disappear, but rather to explain folk psychology, to acknowledge its usefulness and then explain it through these other means. Um, so that's where things get difficult because it's really easy in science to say, oh, I know how to solve the consciousness problem, ignore it. There's no such thing, it's fluff. We don't even think about it. We're we're only work working with neurons and how they interact. That's that's easy, and of course, doesn't solve anything. The tricky part is to say, no, actually, there is folk psychology, but now let's try to understand how it got to be that way and um what it means and and so on. Right, So um yeah. I I think though it's going to be a huge challenge to us to think that way.

Because I think the um folk consciousness system, for lack of a better term, and the folk physics and tool use system, th those two systems are are deeply antithetical in the brain, such that when one is active, the other tends to be suppressed. And for good Darwinian reasons, you wouldn't want to empathize with your hammer every time you hit the nail, right? You wouldn't want to empathize with a physical object, with the ground you're walking on, with the animal that you're hunting. Uh, you wouldn't want to empathize with your enemy that you're trying to exploit or out fox. Uh and on the other hand, you wouldn't want to treat your loved ones as physical objects to be manipulated or rearranged.

And so because of of this conflicting way of thinking, you don't want to be treated about that way. Yeah. You don't want to be treated that way. Exactly. We have this metaphor of manipulation, the way you would manipulate an object. We don't want to be manipulated. We want to be treated as an end and not as a means, you know, to put it in in Kant's terminology. So I think these two ways of thinking the manipulative mindset and the uh empathizing consciousness mindset are um fundamentally intention. And you can even see this, there's a lot of neuroscience data that's just showing that the so called default mode network, the brain network that we use for empathizing and imagining and remembering past episodes, that that network is uh deactivates the task positive

network, the network that we used for accomplishing tasks, for understanding physical objects and how they interact, for manipulating models. And so, because those two systems are antithetical and intention, it makes it really hard for us to think of a system as both conscious and a bunch of neural mechanisms at the same time. That as soon as we start empathizing, the neural mechanism part of our brain shuts down and we forget that it's there. And as soon as we start thinking about it as a machine, the the empathizing network shuts down. And so I think it's going to be a really big challenge for us to understand the social world in in physical terms. And I think it's going to be threatening in the same way that it's threatening uh, if you

were to view me as a physical object to be manipulated. If you were to try to take advantage of me or exploit me, that's threatening. And to view um the entire world or the entire social world in that mindset is kind of it feels like an icky sociopathic mindset. And I think that's a big part of the reason why people are so threatened uh by consciousness being scientifically explained. Yeah. It sort of activates our fear of being manipulated. Yeah, I mean that could well be. I've also described I I I've put it this way. The the theory that we're working with, this theory of consciousness, explains why nobody believes the theory. Yeah, exactly. So that makes it a little tricky. Yeah. It's it's if we're as if we evolved not to do science.

Yeah, exactly. It's like uh it's it's we've evolved to not understand the theory. Yeah. Yeah. Yeah. Yeah. So I it's funny. There's a paper I'm working on right now, and the the running title is called Why All Theories of Consciousness Are Unsatisfying. But you need a subtitle, which is uh Graziano accepted. Yeah. Well, I do have a line in one of my books that says this theory is not satisfying, but a theory doesn't have to be satisfying to be correct. That's right. Yes. Yeah. We talked about, we I talked about this, uh, the thing about free will uh earlier episode. We talked about how it's kind of like uh Douglas Adams answer 42. You know, it's sort of, you know, there's just sometimes uh you get an un the correct but

unsatisfying action answer. Yeah. Yeah. Um I want to ask you about um bringing it back to the science. So what do people what are people doing right now in science that's wrong because of the wrong way to think about consciousness. And if you could wave a magic wand and change the scientific study of consciousness, how would that look different than what's happening? Yeah. Good question. It's very tricky. Um because I think the concept is wrong at the outset, Um, it's very hard to proceed from there. So if you start with the concept, I think most people who are consciousness scholars would say there's an essence essence a thing. And call it magic if you want. Some people like that word, some people don't.

It's the feeling, it's not itself information, it's not a physical thing. You can't poke it and measure a reaction force. It's the feeling itself. How do we explain that? How does it emerge? That is framed as the question of consciousness. Right. And so once you've framed that, then you're out of business. Like you can't go anywhere with that. Because as you said at the very beginning of the conversation, then you've just accepted an intuition, a faulty intuition. You've accepted it, and now you're saying, let's go try to explain how it got to be that way. Uh, so um, you can collect data, and people collect tons and tons and tons of data, just massive amounts of data on um, well, here in this circumstance, people say they're

conscious, and this they don't. And the data might actually be useful and interesting, but it's motivated and predicated on these um incorrect, this foundationally incorrect idea. And so now you have, you know, for example, the the great um adversarial collaboration, as they called it, between two major theories of consciousness, the integrated information theory and the uh global workspace theory. And tons of like literally millions of tens of millions, hundred million dollars or something like that got spent around the world to try to conflict these two views, which in my eyes are the same view, because they both start with the same premise and then they come up with slightly different answers of what the you know alchemical

mixture is that makes the magic emerge. But they're basically alchemy. Um and so and then it turns out that none of the experiments quite distinguish one from the other. And then people can say, well, I respect your opinion, but I can interpret your data this way. And then the other side says the same thing, and then everyone shakes hands and that was it, a hundred million dollars. Um and the and it's not down the drain, it's not useless because the use of it was to advertise a fundamentally magicalist perspective on consciousness and make sure that dominates. Right. And that worked. Yeah. Right. So how we go from there, I don't know. But I do think, I hope, that the advent of deep learning systems of AI

will give us something to sink our fingers into. As engineers, one of the things I've discovered is that when you talk to engineers or um data scientists or um you know uh machine learning experts, they're very warm to my work, to the attention schema theory. Because if you try to talk to them about consciousness, the first question they'll ask is why do I care about a magic essence? Like, what does that have anything to do with this machine? And then they'll ask, Well, supposing it did have a magic essence emerging out of it, why would that affect the behavior of the machine? What is the use of that? Um, but then when you get to a theory that says, Well, okay, never mind the magic essence, your machine might at some point think that it has a magic essence.

Yeah. Because it might compute that self-description. Yeah. And that will surely affect how it behaves. Yeah. And it might actually even make it a better social interface with people. Yeah. Because now it's similar enough to people that people can understand, quote, unquote, understand it or socially interact with it better. And now suddenly the machine learning experts say, Oh, wait, I get it. You're not talking about magic. You're talking about information processing that makes the machine generate these representations and therefore put this output. Once you start thinking in that way and doing experiments and studies on that end, what does it mean for the thing to have a self-model? How many kinds of self-models are there?

What does a self-model have to have before the machine starts talking like a person and saying, Yeah, I have a subjective experience. I know what that is. Like, what goes into all that? Um, those are the the I think that's the only way forward. That's the only way we'll begin to understand this stuff. Yeah. And I I would add to the people who are positive towards your stuff would be evolutionary psychologists too, because I think this is a group of people who are taking what usually are sort of fluffy, abstract, ghost-like things and trying to think like an engineer about it and say, look, reality imposes a problem, or there's just dynamics in the world, and there's like evolved, biologically engineered mechanisms.

And so I think like a great area, you know, if there's graduate students in this area listening, like I think doing an evolutionary task analysis of what would a self-model system need to have. And there might be a hundred thousand of them. Like, you know, you don't have to explain everything, but what would a good self-model need for st social tr strategy, for um for self mod correcting errors like me walking into a tree, you know, like that's still low hanging fruit um from a from a this software level. Like what would we expect the self models to look like? Yeah. Yeah. I would echo that. Yeah. It it does seem to me the most friendly to um an uh a Darwinian model of mental evolution. Because if you think about, you know, integrated information theory, what's the adaptive

function of integrated information? What are the selection pressures on integrated information? I have no idea. You know, what's what's what's the function of a global workspace? I don't know. But like, why is it useful for an organism to have a model of its intentional state and the intentional states of others? That seems really uh clear uh and and it's it's something that you can sink your teeth into uh from a reverse engineering perspective. And I think that's that's to your credit that I think it's a it's a uh a useful way forward in in in this morass of of consciousness scholarship. Yes, yes, exactly. Yeah. Yeah. And it it provokes the next best question, which is like, well, how would that work?

Right? Like it says, okay, look, there's a self-model. Well, what would the problems be? What what would the system need to have and why would it have those? What what situations is it detecting? And therefore, what representations would it have? And there's a reason why it's shallow in this case, you know, because it's just, you know, it's like I I the thing I always think about, you know, is I'll I use this example on a paper, is like a gazelle doesn't need a neurohave cognitive account of a lion. It just needs to detect, does the lion see me or not? You know, or is it is it gonna start coming towards me or my kid or my, you know, um I don't know if gazelles have husbands and wives, but my extended interests, right?

Um and so yeah, so exactly as you said earlier, you know, it doesn't need to have this deep detailed neurocognitive account. In fact, it wouldn't to the extent that having that detailed neurocognitive account doesn't afford, you know, m affecting behavior, like what are the affordances, right? And so, really, like thinking about these, these are states of the world that then are hooked up to detectors that if there is this state of the world, do this one thing, and if not, do another. Yes. And it's at that level of resolution. And I think that exactly what you're describing is very applicable to then young students think, okay, what states of the world would there need to be in these systems?

And because what would come follow from them? What would an organism that's well designed need to do in response to these? Yes. Yeah. And so just to point out, your gazelle example is literally an example of an attention schema. Yeah, exactly. Because literally the gazelle needs to know how the lion's attention operates. Right. Yeah, exactly. I love there's a jumping spider. Um, and then it actually maul it it uh it pretends to be a um uh uh prey on another spider's web because it eats other spiders. So it has at some level this representation of the representations of the other spiders' attention. Yeah, yeah. So it's a really interesting question When I was just talking about the um evolutionary path or where what does and doesn't have this system, I

talked about vertebrates in specific. It's a really interesting question what other lines of or branchy bushes of evolution contain similar kinds of solutions, and that nobody knows. Uh it would not surprise me if insects have well, we know they have something kind of like attention, attention mechanisms. It wouldn't surprise me if they have some kind of um models of their own attention at some level. And then you have animals famously like the octopus that people love to study because it's so complex. Uh, does it have these systems? Um, and then of course you run into the other problem, which is uh people's social machinery being at what it is, as powerful as it is, we can't help looking at an octopus,

seeing its big old eyes, and saying, Oh my God, it's conscious. Uh and so it gets really hard to separate out one thing from another. Yeah. Yeah. And then there's there's the question of whether groups are conscious. So um the philosopher Eric Schwitzgable says, you know, if if materialism is true, then uh the United States is probably conscious. Yeah. I would disagree with that. I do too. But I think it's an interesting possibility that um it it makes sense to think of groups as having um attention. You know, a group can pay attention to some things and not other things. Uh, you know, a political coalition will pay attention to the things that are relevant to its political interests. Um and whether it needs a model of its own intentional state, that also seems like it's

plausible that a group of people would want to know what they are attending to as a group. Uh, and maybe that's facilitated by leadership and by organ, you know, meetings and and so forth. So um I don't know. Maybe maybe maybe groups can achieve a certain kind of consciousness. It's certainly intuitive to us. We certainly talk about, you know, groups as having a kind of mind. We say, you know, Microsoft decided to do this, or the United States did that. You know, we talk about world history and you know, you know, uh countries at war. We talk about France did this, France believed that, France wanted this, uh, as if France was a sentient being. Yeah. Right. So what do you think about it? I don't think they have the self-models in the same way.

I don't just like people talk about plant consciousness, because they say, look, they're sophisticated, they're complicated, they actually process information. Yeah, that's all true, but I don't think they have self-models built into them in the same way. I don't think groups do. Here's how you could get a bunch of people and turn them into a conscious entity. Take 86 billion of them, have them all hold hands in complicated patterns, and then uh devise a system where if they feel a squeeze and there's enough of a squeeze, then they're supposed to squeeze the next person's hand, and all you've done is turned people into simulated neurons, and then you build a giant brain, and that giant brain would be a brain, and it would too have a consciousness.

That's a very, I mean, that's a thought experiment that can never happen in practical life, I think. But you could in principle do that. But I don't think that's what's happening. I don't think larger groups uh have that. But of course, other theories have very serious problems. Like famously the integrated information theory uh has as many people pointed out well, if that's true, then a lot of things in this universe are conscious. and um, and that has actually forced theoreticians who are very committed to what they believe, to say, okay, then panpsychism, then I believe it. People will say yes, because I believe in the integrated information theory, therefore, potatoes are conscious and electrons are conscious, and planet Earth is, and the solar

system is, And um you get into real serious problems there where then suddenly everything is conscious and then it's not clear what that word means anymore. So these the this world can get very loopy and very fluffy really quick. Yeah. It is quite bizarre that our minds evolved to find the possibility of electrons being conscious more intuitive than the the uh the the possibility of neural mechanisms being conscious yeah that's actually really interesting yeah yes yeah exactly yeah it's it's a good it's again it's a lovely example of um sort of the psychology of psychology right and that uh we did not evolve to do to do this science um Michael, what do you very briefly, um how did you get into this?

Or or another version of the question is you are sort of uh not alone in the wilderness here, but there's as you said, there's this enormous, powerful, well-funded ideas and you uh are are sort of saying, hey guys, look, it's this other thing, by the way. Um is that the way that you're um uh is that something that you sort of enjoy being in that role or is that something you found yourself in that role um in your life or how do you get to that point where you're um you don't want to be in the the uh evil empire you're you're the um scrappy uh rebel uh yeah I I am often in that role and have been so multiple times in my career in different topics within neuroscience. I don't particularly enjoy it.

Why would you? I do not. No, I find myself the same way, so I'm I mean it's a yeah, it's a but yeah, I don't either, but yeah, exactly. Yeah. I do think and now we really get to very deep philosophy of science. I think the purpose of science is contribution. I do not think the purpose of science is self-aggrandizement or finding a brand name or being admired by colleagues, um, building an empire, and I think probably many people in science. And this is more exaggerated as you get to the more successful people. Those kinds of motivations tend to drive people's behavior in science. They may not even know it, but they're evaluating science, whether it's good or bad, whether they should do it or not, on a kind of gut impression of whether it will make a big

um positive impression or not, whether it'll give them the accolades they want or not. And I really don't want to do that. Never have. And when I try a new field, I think I see insight that should be useful. And I want to put it out there. And I'm always hoping that the colleagues will look at it and say, Wow, cool, that is useful. And it never works that way. It's always a fight. It's always a struggle. It's always look, guys, you're not thinking about this logically, and it's gotta be this, and the data shows that, and it's such a struggle. And then I get to a point where I run out of ideas, and it's not new insight anymore. And I've reached exactly that point that apparently everyone wants to be because you know

I'm established. An extinct volcano. Right. People know me and I'm established, and I have my brand name, and that's where I now should sit for the rest of my life. And and and that's the moment that I decide I'm not interested. I gotta change the field that I'm studying because I don't want all that junk. Yeah. So I think co consciousness I have not reached the the thing that keeps me in consciousness right now is this sense that it's hugely important for all of us. Because it's not just an intellectual question. It's about our future and it's about technology and it's about literally the most transformative event in our species, which is the development of this new thinking machine. Um, and yet people don't understand the concepts yet.

So I still I feel like I have this responsibility to just keep saying to keep preaching, so to speak. That that I think is what keeps keeps me in that that field, as frustrating as that is. Yeah, yeah. No, it's uh um, well, I mean, uh, you know, for for what it's worth, when I read your stuff, I thought, oh, this is cool. So um it happens sometimes. Yeah. Um, likewise. Yeah. Um, is there any question we should have asked you? Oh, not that I can think of now. I wanna oh well I have one question if if uh or if David has got one, but uh you have a keyboard behind you. Yes. Um, and you also have books behind you, and you have you have several hats, including I was reading you, also write kids' books, and you you have a pen name.

Um, and one one reason is so that kids don't read the wrong book. But I love the idea of a precocious kid reading about uh self-model uh neurocognition. Um so I I love that idea, but um, can you talk a little bit about those other hats or one of them? Um, you know, whatever you want to say about it, but that's an interesting thing to to also have the hat of a a kids book author or a musician. Um, and maybe it's not entirely afield from what we've been talking about. Yeah, I so I definitely have a lot of interests. Um, I've been working on these things forever and ever. Um, you know, it was always a question do you become a writer or a composer on your own time while being a professional neuroscientist?

Or do you do neuroscience in your garage on your own time and be a professional writer? And it turns out you can't do the neuroscience in your garage on your own time. So it has to go one way and not the other. So that's a very career-defining requirement. So um, but I've always been, you know, since I was very little, I've been involved in in music. Uh I have a um a YouTube channel. You should check it out. It's called Gras Gorilla, all one word. We'll put a link, we'll put a link in the show notes as they excellent podcast world. Yeah. It's um basically a set of tutorial videos on music composition. Oh nice. So um that's something that I learned a lot about. And then I felt like I might as well pass on some knowledge.

And uh yes, I write I write a uh the kids' books are probably mostly like the seven to ten year range. They're they're fun. Um I've written some novels more for the adult age range as well. And um yeah, it's all it's all good. When you get stuck on one thing, it's good to have something to switch to. Yeah. Well, in in in in light of the previous comments about science, I imagine the kids' books are pr there's probably a very cool um carving up of the world in the kids' books too. I guess so. I mean I think because they're written by me and I have a scientific perspective. Yes. Uh but I don't think they're meant to be science education books for kids. They're they're just adventures. Yeah, yeah.

Well, maybe what I would put by that is uh I think sometimes a good scientist is just sort of like an honest kid who never grew up, which is just sort of your like it's the David Pinsoff thing is like, is this bullshit or not? Yeah. You know, and a good scientist, a real scientist, not somebody who just sort of wants to be famous, fundamentally is sort of like their bullshit detectors still working, you know, and they're just like, this is bullshit, this isn't. And even if the bullshit has lots of money and it's a you know, sunny grass-filled pastures, and the real the the thing that looks right, looks more difficult, a little more uncomfortable, not as many people like it or care about it. That's what I'm gonna do.

Yeah, yeah. Yeah. So I have one book. Um it's called the The Last Notebook of Leonardo. And it's about a father son adventure. And the father is a scientist and he um figures out how to transform himself physically into a gigantic orangutan. And unfortunately, as a gigantic orangutan, he's no longer allowed in his uh apartment complex. And so they get kicked out, and then the adventures start. And so here's this genius scientist trying to interact with his son, uh, and they have their adventures through the world. So I think the science infiltrates into the books in in strange ways. Fantastic. Well, um, David, do you have any final No, this was a really delightful and and uh wonderful conversation.

I really appreciate you uh giving us your time and uh and your wisdom. Uh this was a blast. Thank you. Yes, I enjoyed it. David, I never know what to say in the outro. Well, usually a good place to start is the topic of conversation uh that you just listen to as the listener and you were just listening about consciousness, which means that you're still conscious if you're hearing these words. And if you continue to be conscious after hearing these words, then you can use that consciousness to support Evolutionary Psychology the podcast. And how can uh those sentient beings do that, Dave? Well, they can um click buttons on a computer which mean um specifically support us on Patreon, comment on social media, uh, do a review,

ask us questions, send us an email or over our social media. Um, we'll have a ask me anything episode um in the near future. Yeah, though I think you nailed it. Those are all the things. And even if you're not conscious, just do those things too. Just like a Yeah, we we don't really care if you're conscious or not, as long as you're you know executing the behaviors, we're gonna be happy. That should put that on a t shirt. We don't get unconscious or not. As long as you're executing the behaviors, we'll be happy. And send somebody, David, send uh somebody needs to send you a fanny pack. Yes, please send me a fanny pack. Um, if you're conscious or unconscious, uh click that uh buy button on the fanny pack

and uh let me know when it's arriving. What what's your favorite color, David? Uh I like I like green, reminds me of nature. Okay. Green, how about black? Uh I could do black. Black goes with everything fashion-wise, so that might be a better choice. I can, you know, append it to any outfit uh without it ruining the outfit. Yeah, I think a fashion oriented fanny pack purchase is a bit of an oxymoron, but you know it's 2026. Let the uh well, I'm I'm trying to bring fanny packs back. I think I'm gonna make them fashionable again. That's my master plan. Okay. Well you heard it here first. Yep. All right. The finger on the pulse. See you next week. See you next week.