 Hello and welcome everyone to the Active Inference live stream. It is act imp stream 9.1 and it is November 24, 2020. So welcome everyone to the act imp stream and welcome to team calm today in the act imp stream. We will be talking about a really interesting paper. So I'm quite excited. Team calm is an experiment in online team communication learning and practice related to active inference. You can find us at our website or Twitter at email or YouTube or our public key base team and shared key based username. This is a recorded and an archived live stream. So please provide us with feedback so that we can improve on our work. All backgrounds and perspectives are welcome here and also just as far as video etiquette for live stream. Remember to mute if there's any noise in your background, raise your hand so that we can hear from everybody on the stack and use respectful speech behavior, etc. Just to quickly go through the rest of the meetings that we'll be having for 2020. Today on November 24, we're talking about the project of consciousness model and phenomenal self hood. Next week in 9.2 on December 1, we're going to be talking about the same paper. So anything that gets raised today, whether you're participating live or watching replay, let us know about it so we can cover it in 9.2. On December 8 and 15 in 10.1 and 10.2, we'll be discussing the paper, a variational approach to scripts. And then on December 22 and 29th, we'll be talking about sophisticated effective inference and that will be another more on the modeling side paper to discuss. Cool. Today in Act M stream 9.1, the roadmap will be as follows. We'll start off with some introductions and warm up questions. And then we'll get to 9.1. We're going to be talking about the paper called the project of consciousness model and phenomenal self hood by Williford at all 2018. And punchline of this paper is basically about the combination of a few different fields, geometry, cybernetics, neuro phenomenology, etc. And by combining these fields under the framework of active inference, there might be something there to understand or explain about consciousness. So we'll go through the aims and the claims of the authors, the abstract and the roadmap. And then at multiple points here and throughout, we want to take time for questions. So really anyone who raises their hand will just as soon as we can cut there because it's so interesting what people have come to mind during these provocative topics. Then we'll go to the authors study of consciousness and talk about the two lists that they make and what those lists do for the paper and do for their theory. And those are the functional features and the phenomenological invariance of consciousness. And then at the end we'll go through the figures and with however much time we have, we'll go into thinking about what each figure might represent or invoke. And as stated in 9.2, we're going to be discussing this same paper again. So save and submit any questions you have and get in touch with us if you want to participate. Great. For the introductions and warm up, please just introduce yourself and your location. Say hello and as short of an intro as you'd like and then pass it to somebody else who hasn't yet spoken. So I will go first. I'm Daniel. I'm in California and I'll pass it to Alex. I hope I'm audible. My name is Alex. I'm also in California and I will pass this to Shannon. Hey, I'm Shannon. I would be in California normally, but I'm in South Dakota, so I'll pass it to Lee. Hi, I'm Lee. I'm in London and I'll pass it to Mathis. Right, currently muted. Okay. Hi, I'm Mathis. I'm currently in Germany in Osnabrück and I'll pass it to Ivan. Hello, my name is Ivan. I'm from Moscow, Russia and I'll pass it to Sasha. Hi, my name is Sasha. I'm also in California and I'll pass it to Alex V. Thanks. Hi, I'm Alex. I'm in Moscow, Russia and I pass it to Steven. Hello, I'm Steven. I'm in Toronto in Canada. Cool. That is everyone. So great. Thanks for these introductions, everybody. Let's do some warm up questions and there'll be a few of them and anyone who would like to talk just jump on with raising your hand. Let's try to hear from everyone as well. The first question, and I'll do actually the first and the second question together since I think that they're very related. The first question is what interests you about studying consciousness and the second question is what are some ways that we can study consciousness? Maybe someone's a skeptic. Maybe they don't think that we can study consciousness or maybe they have a really awesome idea about how we could do that. So, Lee, go ahead. What's your first thought? Yeah, so, I mean, I primarily come from the phenomenological angle and particularly kind of neuro phenomenology really. And what I'm really interested in is agency. So can the models that come out of coming out of neuroscience, can they help us to increase our agency? So, for example, if you can become reflexively aware of simulations, modal simulations that are going on in your system, then do those become affordances for actions that we can take? And so you increase the scope of agency. So that's kind of what really interests me about this. Cool. Thanks, Shannon. I love that you bring up agency right off the bat. Something that's really interesting for me about consciousness, which I don't study at all, is when we lose our sense of agency, like when we're in flow states, if we're making music or I guess people get in flow states, if they really like exercising. And we're completely immersed in the experiment experience, but not consciously aware or we don't have the same kind of intentionality that we do. As if maybe we're like writing down a math problem or trying to speak over a live stream and thinking very carefully about each of our words. And I guess just the phenomenal quality of those two types of experience is really interesting to me. I agree. And it's part of this, what is consciousness question and what is the role of awareness. So when we're really paying attention to doing a task with our hands when we're really paying attention to balancing on a beam or something like that. Are we truly at our most self aware at that moment, or are we at our least self aware or in which ways are we self aware we're not thinking about our biographical history, per se. But in another sense, we're the most embodied and an active sort of circuit for action. So where is that going to fit in with active inference and with this idea that we are embodied in an environment. So what is really more experiential of the self doing something that might take up your whole body or bring your full attention to bear or doing something that asks you to write a personal statement, something like an autobiographical or narrative self. Cool. If anyone else wants to go on the first two, those are good questions. Or the last question here is just like, what is another theory or framework for consciousness that you know of or have heard about doesn't need to be your theory to support or deny here. Just like what is something else that is in the space of this paper, and then what's just something that you liked about it or found absent from that model. So all just starts with something that I found missing from a lot of models and I really want to know where agency and where a lot of other topics fit into play here is it's often a little unclear whether consciousness is something that reaches back into the system and intervenes, or whether it's something that's just purely epiphenomenal or kind of just there could be awareness but awareness may not have causal power. So it's consciousness something that's just getting spun off to the side and produced, but not reaching back into the system or is it something that's actually essential potentially even in an evolutionary or adaptive context for maintaining coherent behavior. Steven. I'm interested in this question of self consciousness that seems to be, and I like just the idea of general consciousness being thought about because I think in psychology this idea of the self, and the idea of the individual has become so dominant and the idea of broader consciousness was kind of left 100 years ago for the philosophers to deal with. So this idea of I have a self and I have this idea of a psychology and this perspective on the psychology has become kind of all encompassing but now this ability to bring consciousness in more broadly sort of dissolves a lot of those problems because it kind of stuck psychology this single scale of analysis. So that's one of the things I think that is quite exciting by all of this. Sasha. One thing that I find missing I feel like a broken record sometimes is the developmental side of consciousness. And even if we're thinking about humans, you know, our favorite species. How does consciousness develop across a lifetime, and what are the different stages or versions of consciousness that we pass through. And is there sort of a developmental trajectory, because I tend to think it's not like flipping a light switch. Like it's not that we are born and we're immediately plugged into this same consciousness that we're experiencing today. So, yeah, that's kind of what I would like to see from a consciousness model is different levels or versions of it that we can start to describe, like much like the flow state or sleep. Yeah, I like both of those points that Steven and Sasha brought up. And not just how does our personal consciousness develop over time but how does our consciousness, like our group consciousness develop so like as an infant a lot of your conscious experiences really coupled with the mother or the caregiver. And like slowly you might develop a sense of yourself as an individual separate from that caregiver. But also when we gather in groups, you know, when we gather at concerts or we gather at protests or in events where we're at mix of explicitly or implicitly interacting with each other. So we start to behave like like one group and in some sense we start to really feel like one group rather than like just myself in a room full of other people. Yep. Cool. And something in the ant literature people have wondered about is the nest mate conscious is just the colony conscious but not the nest mate, then what is really happening with this transition from individual to group. And the development of that also interesting the formation of group is it kind of like how many sand grains make a heap. You know, once there's three people it's a collective consciousness or do they have to be doing contact improv or do they have to be rhythmically synchronized. What are those dynamics, what might be metrics, what might be things that we could come to the table and agree on as far as measurements are concerned so that we could kind of move the needle forward and research, or is this a question that because of its maybe fundamental difficulty and measurement, is it a question that's sort of forever going to be in the speculative domain. I think that the authors of this paper would suggest that scientific study of consciousness is meritus, and that there's value in approaching with even a mathematical bent approaching how we could study consciousness. So it's by no means obvious that mathematics is the way that we're going to express the language or the structure of consciousness, especially when we start to think about all these different dimensions like group versus individual and what, you know, who is individual. We were never individuals that type of thinking, what does the brain region know that the neuron doesn't and the brain know that the region doesn't. So all these questions are going to come into play. And a lot of the neuro and social parts are going to get a little bit, not swept under the rug here, but this paper is clearly a little bit more on the cut and dry projective geometry site. And there's a lot of discussion of neuro phenomenology but you don't see any pictures of cells. You don't see any actual details of neuro mechanisms is just assume well you could wrap mechanisms and mechanisms, kind of close to the end. This type of a systems behavior. And so we'll get a little bit more into it. So at any point of course people are welcome to raise their hand we'll just try to immediately whatever springs to mind in a relevant moment so no worries about these prompts. Let's talk about the paper, which is the projective consciousness model and phenomenal selfhood by Willa Ferd at all in December 2018 in frontiers in psychology theoretical and philosophical psychology. The aims and the claims of the paper are basically as such, they ask, can we describe this integrated multimodal and centered spatial structure of conscious experience in a rigorous way. And can we give a good account of why it should be thus organized. Here we answered both questions in the affirmative. Great, big of true. Let's see these be answered. Overall, the PCM delivers an account of the phenomenologically available generic structure of consciousness. And that speaks to Stevens point about the general consciousness and shows how consciousness allows organisms to integrate multimodal sensory information, memory and emotion in order to control behavior enhance resilience, optimize preference satisfaction and minimize predictive error in efficient manner. And this is again to my earlier point about does consciousness get spun off of the system or does it reach back in. So does consciousness control behavior and enhance resilience, or is that systems with controlled behavior and enhanced properties like resilience, do those systems spiral off consciousness as an observer. That's vague here but the way that we phrase these sentences is essential. It's like active versus passive mistakes were made versus I made a mistake versus you made a mistake. Pretty different. And the non jargony way might be to ask, how can we apply active inference as this unifying framework to the study of consciousness in all of its unified coherent mysterious dynamic and profound facets. And we'll see when they get to the functional features and the phenomenological invariance, which aspects of consciousness, they think are sort of the key features to explain and really integrate across into one common framework. So that's what we hope to be getting at today. Even though this paper is not the whole citation network, we're definitely going to be exploring a lot of the key ideas and questions about how and where active inference and the free energy principle could link back to consciousness. So here's the abstract briefly. They write, we summarize our recently introduced projective consciousness model PCM, which is Rudolf at all 2017 and related to outstanding conceptual issues in the theory of consciousness. The projective consciousness model combines a projective geometrical model of the perspectival. That means from a perspective phenomenological. That means what's experienced structure of the field of consciousness with a variational free energy minimization model of active inference comma yielding an account of the cybernetic. That means goal seeking function of consciousness vis the modulation of the fields cognitive thought processes and affective like how it feels emotions dynamics for the effective control of embodied agents. Another plus one for affective and effective in the same sentence. Abstract part to the geometrical and active inference components are linked via the concept of projective transformation. So we'll definitely have a slide where we look at what a projective transformation is, which is crucial to understanding how conscious organisms integrate perception emotion memory reasoning and projective imagination in order to control behavior enhance resilience and optimize preference satisfaction. The PCM makes substantive empirical predictions and fits well into a neuro computational framework. It also helps us to account for aspects of subjective character that are sometimes ignored or conflated pre reflective self consciousness, the first person point of view, the sense of mindness or ownership and of social self conscious so maybe we will get to some of these social topics. And they close by saying that we argue that the PCM though still in development offers the most complete theory to date of what Thomas Metzinger has called phenomenal selfhood. So that's combining this phenomenology component with a selfhood element. Any thoughts on the abstract or we can move forward. Bring them up during the roadmap. So the abstract is just the header of the paper. It's kind of like the metadata for the in group and out group discussion. The roadmap is how they're actually going to implement these kinds of claims that are suggested in the introduction and abstract. So first with a more formal introduction to the topics they discussed in the abstract, and then they went straight to methodology, which I think is very important, especially when discussing something that's abstract like this. Then they talk about the phenomenological invariance and the functional features of consciousness which will go through on separate slides. And specifically they talk about these invariance in terms of a projective geometry feature of the PCM. They talk about the functions and then talk about how the PCM links the functional features of consciousness within an active inference and a cybernetics framework. So they're taking some of the invariance, combining them with projective geometry. They're taking some of the features and combining that with cybernetics in the sense of active inference as a modern phrasing of cybernetics and free energy minimization as an optimization process. And they're kind of connecting experience, cybernetics, projective geometry, all these different things together and asking whether it makes sense. They then have all their four figures, which will go through today that range from sort of sketch like accounts of psychological phenomena to some early solutions to otherwise challenging ideas like illusions, Necker cube. And then they sketch out a few different representations and simulations that they run or ran elsewhere. They then return to this question of the hard problem, representationalism and phenomenal selfhood. So the hard problem is the hard problem of consciousness attributed to Chalmers and others that relates to basically where the rubber hits the road. How does consciousness arise or what is it? I think there's many hard problems and who knows which ones we think are easy but actually are hard. The point is they're going to come at that nexus from a computational and a functional angle. Then they talk about representationalism in the PCM. And I feel like representationalism, we've heard about a bunch on active stream with a Bayesian structural representationalism. And now we're going to come at it from a different angle from a geometric and a cybernetic angle. They then talk about how the PCM is linked to subjective character as well as phenomenal selfhood, which is clearly one of the things that they're out to describe. And then they conclude. So they're going to go from taking a few different threads of phenomenological neuro research and tie that with some more mechanism agnostic control theory approaches like cybernetics, active inference and optimization approaches like free energy minimization. They're going to bring in this projective consciousness, projective geometry bit to tie it all together and then we get to the conclusions and also some of their thoughts on where to go next. So let's just pause there. I've written out a few questions, but does anyone just have a thought or a question or when they read the paper like what was the first thought? Was there something that resonated a lot? Was there something that seemed ill-defined or you would have defined it a different way? Or just something you're curious about, Stephen? Well, one thing I thought was interesting was this idea of perspectival consciousness. So this idea that when you're consciously taking a perspective on your situation, which you normally might think of as seeing it, but it could be some sort of embodied sense. So I think that was kind of important and how that differs from maybe the unconscious dynamics that are just going on that they don't necessarily deal with. So it's somehow sitting, I don't know if it's sitting on top or sitting alongside it, but there's somehow this perspectival piece that somehow kicks in. Yeah, and we often think about vision in talking about how active inference is to be differentiated from, let's just say, a sensory relay model where photons come in, your brain categorizes them and makes an image and then you act. And here it's like it's a generative model of the world that you're experiencing. That's why there's color and good resolution around the edges and no blind spot. And then it's like, well, then why do we see what appears to be coming from our eyes or from a position that's right there? And that's where we're going to come back to with a lot of these questions about the projection itself and why we appear to be situated. Lee? Sorry. Yeah, hi, I don't know if it's relevant, but I was just thinking about the discovery of perspective back in the Enlightenment. And I was thinking, well, obviously, you know, if this is a sort of a more generic theory of consciousness, then I was wondering if it connected perhaps to Metzinger's idea of transparent states so that, you know, the projective geometry would have always been there, but there would have been transparent. And somehow, you know, in the process of the discovery of perspective, it became something that we were able to pay attention to more consciously. Anyway, just a thought. Cool idea. Let me actually unpack that. First, what would you describe as the transparent state? Because I don't know of that terminology specifically. What is a transparent state here? I think in Metzinger's model, so transparent states will be those states that we view the world through. So we're in a stable, we don't know that we're having a particular state. Whereas, he contrasts it to a metaphor. So an opaque state would be a state where we understand that it's a sensory state, if you know what I mean. Cool. The transparent state, it's kind of like, what is water? The water we swim in, it's like the water's transparent. So when you look out in all directions, it's not being seen specifically. Yeah, you're looking through it, you're looking through it, but you're not able to actually observe it as a state that you're having, if you see what I mean. Yeah, just to unpack this one bit for Stephen, it's kind of like, without a theory of air, you could just look forever and you could just have any model could replace this idea about what fills up space. Then there's this chemical theory of air and there have been various chemical theories of air and whether there's four types of things they mix or whether it's protons and neutrons and different ratios and things like that. And then once you have a theory of air, you could ask things like, well, what if I compressed all of the quote, invisible stuff in this room and all of a sudden it looks like blue or something. It's like, whoa, it kind of is there. There's unique experiments. There's unique predictions. I'm taking something that was transparent in the niche that actually was real ontologically real and then through attention and awareness and affordances were able to now like look at it. So we've developed projective geometry and it's like, whoa, what if it really were projective geometry all along, but then now with the projective geometric tools were coming back to experience that. And it's actually something that Stephen and I were talking about just this weekend with a little bit like what people's experiences of perspective have been in the past. Because there is those sometimes funny pictures, at least comical sometimes where it's like a person's face and it looks like it's from almost a Cubist angle, but it's like a Renaissance painting and there's probably so much to say from an art history perspective about the evolution and the development of perspective. But at the very least, it seems like people throughout time have been differentially excited by different types of representations of perspective. People weren't doing photo realistic charcoal drawings in a different era, even if they necessarily could have had the technologies to do so. So something's happening there with perspective and how it's changing the way that we interact with a visual world. Cubism, of course, was shocking and when it came out. And so now where are we with perspective? There's a lot to say about that, Stephen. Yeah, I mean, one of the things that's interesting with like imagined feelings in space is that they can be kind of transparent. So if you like, what's the energy of someone or if I imagine the energies of people around me, it's like a quality. Those qualities, they don't like if you imagine something space, it doesn't become completely opaque. It's not like you can't see that you're in the room anymore. You know, so even if you're on stage as an actor and you imagine that you're in a scene with a king and it's an empty stage and you start to imagine the whole stage. It's still slightly transparent in relation to the physical, perceptive, conscious physical space, if that makes sense. So those two things are kind of interplay. Yep. Yeah. And to bring that to the affordances and the field of affordances, field of consciousness, it's like the generative model isn't just of what the video camera would see if it were next to your head. The generative model is actually generating this field of action that's scaled geometrically so that things that are closer to you and better within reach are seemingly at hand and things that are distant are seemingly distant. And one example that Stephen and I were talking about was like, there might be an object on the other side of my monitor that I'd have to kind of go around to reach. And just right with a quick judgment, it feels like something on the other side of my monitor is actually further away than something that might be linearly closer or further away. Sorry, but like I can reach it directly just by reaching my hand over. I can reach two feet over with my arm, just grab it. But the 18 inches linearly to the monitor, I have to end up doing a 36 inch route so it feels longer away. So even within our at hand concepts and the way that we talk about what is accessible or what is relevant, these all have to do with grasping and the comprehending and the action distortion of a space. So there's really a lot here about the perspectives that we take. And if we say there's one perspective and it's mine, there's one perspective and it's my cultures or is it, hey, we're all at this party and we're all going to have different perspectives of this party. Okay, which perspectives can hang which ones can't. Those are the kinds of discussions that you get to there instead of is it right wrong the way I'm seeing it because you know not everyone's going to see the party like you see it because no one's head is overlapping with your perception because everyone's affordances and location are different. For example, Sasha. Yeah, I really appreciated in this paper, the kind of connection. We're always talking about on this stream between, you know, taking in information and acting. And so it's this feedback loop of two things. And I really like this converting thinking about our space into thinking about the space of affordances, so that it's not just a, you know, a blank slate of information that you could access or not but it's really what you are able to do with that information and how you're going to act on it. So I, yeah, I just thought that was a really nice connection and fits in very well with what we're talking about and kind of builds up this similar theme of how you're going to connect action and perception. Cool. And just one other example that that kind of made me think about is like when I'm dissecting an insect under a microscope. It's very small movements of the fingers but it in the field of vision of a microscope will swing the forceps like, you know, 30 degrees so be a radical motion under the microscope. But it has the exact same embodiment feel as like swinging one's arm, almost as if what matters is how much I'm seeing my body move within a visual field and how radical of an affordance it is like flailing my arm versus a more fine tuned movement. But a fine two movement with the arm is like one inch but that's represented by a certain number of degrees. And that ends up being like visually almost the same number of degrees that I can move the forceps. And that's at a different spatial scale because the attention and the affordances are down at that scale. And I have the visual ability the umbelts to perceive at that smaller scale which couldn't be seen if it were the naked eye. It takes on a feeling like it's in the regular vision, regular body world. And then also the forceps don't have nerves but they're sensitized. So there's a lot to say and think about there. I do questions that I just wrote down here just like these are of course perennial questions. They're not going to be answered in a couple minutes. But what's exciting about studying consciousness? What do we agree on? What do people agree on? What do people disagree on? And then also what do people agree is unknown? What are the known unknowns? Where might there be unknown unknowns? And then also to always return to what is active inference doing here? What could it do here? What could some other theory do here? What do you think, Shannon? Yeah, so your story about the forceps in the microscope reminded me of the first example of like extended consciousness that I learned about in a philosophy of mind class and undergrad. And our professor was talking about divers who are, they're not actually diving, they're in a room and they're controlling a camera like on a little submarine that's underwater. And they start to feel like they are the submarine so much so that their body's actually responding to the pressure differences that they would experience if they were diving. So if the submarine comes up very fast, they'll actually feel like they're experiencing the bends that would happen if they were a diver coming up so fast. I don't know if this is a true story, I've never looked it up, but it's very intriguing that your consciousness can be so extended that you end up having these embodied effects that you perhaps could empirically measure if you had some sensors on these people as they were experiencing these extended consciousness moments. Cool. Lee? I was just going to ask Shannon, do you know whether that depends on whether the people controlling the remote sub were, had previously dived as well so they kind of got the embodied experience as part of their model? I think for specifically the bends it would or maybe they would have to know about the bends because that's a very specific thing that doesn't happen to your body in any other case. Yeah, one thing that makes me think of is like the drop in a techno song when everything's building and the tension's building, but that's actually a cultural regime of affordance and different sub genres of electronica or different people who are aren't familiar with electronica. They might get chills going down their spine. It's like, oh, this is the sickest drop ever, but somebody else might just not be tuned into that regime of attention. And therefore it's like, okay, wait, what happened? Why is it just why did the drums stop? Something like that. So similarly it's like the diver, if they're part of this deep generative model and then they're emulating it or they're simulating it somehow, this diving, then they may have a deeply learned reflex about changing their blood pressure. Let's just, you know, go with stock experiment being vaguely, you know, feeling true, just ish that you could, if you're doing a first person shooter immersive video game, your heart is going to start pounding, not because you need increased blood flow. That's a psychological response. So trained psychological responses in increasingly immersive environments like AR and VR will generate physiological responses. And that is actually part of the thrill that was, I mean, there's so much happening here. And it doesn't really go into the AR VR here, but like, what if we un-projected our projective geometry? What would happen? Or what about these sort of Alice in Wonderland type simulators for the AR and the VR, where you reach towards something but your arm moves twice as fast? Or you reach towards something and your arm doesn't move, you crash into stuff and it just feels like you're like super disoriented or the glasses are flipped. All of these transformations are just super deep interventions into the structure of sense and action. So there's a lot that is now possible here with how we can on the fly remap these things. In 1920, the Cubis could put a light on one side of the cube and draw it and then halfway through the drawing move the light and get a different perspective. And that was kind of cool, like a mashup of perspectives in one two dimensional plane. But now you could be experiencing like a lot more stuff happening and the cube could be mutating, it could be a hypercube and stuff like that. It's kind of next level with respect to our geometry and how it relates to our sense and action. Any other thoughts or questions here before we go to the functional features? Just think of them as they come up. So the authors highlight for functional features of consciousness, which are global optimization and resilience, global availability, motivation of action and modulation of attention and simulation enhancement. So their goals are basically these are the things that they say notice that it's very mechanism independent. It doesn't say you got to have neurons you need glia doesn't say it has to be at this time scale. These are all like things that you might say about a cybernetic system a control theory action perception system. So these features are the abstract functional features of consciousness, which through neuro phenomenology will get brought into the brain. Perhaps, but just at the functional level, these are the functions of consciousness, and they write, we claim that the overall function of consciousness. Again, is it that consciousness is getting spun off by systems that have these functions, or is consciousness literally the function that maps back into the system and gives these features presence in the world. We claim that the overall feature function of consciousness is to address a general cybernetic problem or problem of control. Consciousness enables, again ambiguous, a situated organism with multiple sensory channels to navigate its environment and satisfy its biological and derived needs efficiently. This entails minimizing and predictive. This entails minimizing predictively erroneous representations of the world and maximizing preference satisfaction. A good model of consciousness must be able to explain how this is accomplished. The PCM thus emphasizes the following interrelated functional features of consciousness. So ask yourself if there was a robot that could do intermodal sensory things and memory and it was doing counterfactual simulations and it had motivated looking action and modulation of attention and global availability of some variables or some processes and it was resilient within some resilience. Of course, we're not resilient to getting hit with a hammer on the head. We are resilient to, you know, a train coming at us from a mile away, but not an inch away. I mean, it's always going to be about where you draw the bounds. Oh, I mean, that's out of bounds. You don't need to be resilient to that kind of an insult to this drug, but to this molecule you are. So I would just ask if we had a robot that had some of these features, would we include it in the conscious club? Or would Dennett include it in the club? No. Would a panpsychist? Yeah, they might say, well, it's something it's like to be that robot. But where are we really getting if we just say it's something it's like to be this robot and this thermometer and this networking system? Well, if the only way we can ask it is the exact same affordances that we as observers have today, like measuring electromagnetic frequencies or something like that, then aren't we going to be debating these things forever? Because no one really disagrees that these things all appear to happen for systems that we think are complex goal directed systems like humans and cats and things like that. But isn't the question about what else we're going to do beyond this? So hopefully that's where the projective geometry is kind of like that's the hook that they hang their coat on. So to speak, they're going to bring it all to projective geometry. And they're going to say this is what explains all these functional features and the invariance, which we'll get to in a second. And it makes some other predictions that would be unique and testable. So they're going to say, yeah, there's probably a lot of ways that theories of consciousness could explain these features, like attention schema theory or the global workspace theory or integrated information theory, all these different frameworks for consciousness. But we're going to take these things and we're going to go in a new direction by talking about the projective geometry. So what supports in their eyes the need for not just a quantitative or a formalized framework, but also specifically the projective geometry. And for them, it's about these phenomenological invariance. So it's kind of like universal basic income. It's like, it's universal, you know, these are invariant. What part of that don't you get? So these are the things that generalized systems of consciousness, you could debate with them, whether these systems that they're talking about are the right ones or whatever. But these are the invariant features of experience and phenomenology. And they have them as relational phenomenal intentionality, which is that all experience and all consciousness is about relations, especially between one and another system like internal and external relationship, even though as we're going to kind of see it's a little bit trippier than that. But basically it's about the relation of like the eye to not I it's like I thou I you these kinds of things. There's a situated 3d spatiality. And that means that you're starting your in a situated spot. You're not just everywhere experiencing a 3d simulation, but it's always perceived as from a situated spot in spatial dimensions, three of them. And then through time three and four, there's this multimodal synchronic integration, which is in a snapshot, it's multimodal. And then for is through time, everything's happening. So this is kind of capturing through space or through senses and then through time saying, yep, you're at a concert and there's the light and the sound and things bumping into you. All of that is happening in this integrated way. You don't just experience those separately. At the very least, the illusion appears to cover integration across senses, which are qualitatively different. Like a chemical is different than a photon from the sun. And then there's this temporal integration through time, which is kind of related to the perceived event durations. There's messages that can be flashed on a screen or sounds that can be outside of your awareness. But there's also ones that can be within your awareness at a certain duration of time. So if it flashes for a second, it's able to be at a time scale that's integrated into consciousness. There might be time scales that influence you below that, but it doesn't become aware. So like that's kind of conscious and subconscious, but it's dealing with that a little bit blurry about whether consciousness is just the whole neuro system. And anything that gets integrated into the nervous system through time is considered to be temporarily integrated. Or is it that there's experiences that actually play out over certain time scales that we experience. And then lastly kind of related to one is this subjective character, which is that it's about the subject and one and five are very similar. But in fact, they hold that two through five are all implicated in one. OK, so they had a functional features that's pretty abstract and cybernetic. And then they're like, OK, what are the invariance, almost the axioms that we want to be building from that we want our theory to be able to address? And then they laid them all out one through five and then they're like, yep, turns out that two through five all collapsed back onto one. So really one is the part we need to explain in how they've set up the problem. So it's relational. It's about the relationship between the agent in the world. It's phenomenal. It's based upon experience and it deals with intentionality. And that is going to bring in agency affordances, all these kinds of things that have come into play. Any thoughts on the invariance or the functional features of consciousness. Because in a sense, that's what that's what frames the paper. And it's what frames the discussion, the kinds of answers they're going to come to is these are their checklist for how they'll know they'll succeed. Or this is the recipe. These are the ingredients. This is like what they're cooking with for the paper. So if anyone thinks of it, they can raise their hand. I just want to clarify that because it's like you could think about this stuff all day, but at least for a journal club, it's like what did the authors actually specify to do? Did they answer something that, yeah. So Stephen. Yeah, I suppose also coming back to what Lee was saying about perceiving things in terms of our cognition in relation to our history, when we started getting into perspective painting and thinking about things that way, you get this sense of how much is it about us constructing cognition as a cultural way of being? How much are we actually constructing this relational phenomenal intentionality culturally? And that then, because in indigenous cultures, you see, they think a lot more about shape shifting and the land being much more fluid and enfolded, whereas we tend to have it much more linear and structured and straight edged. So there's some question there about how much this consciousness becomes like a construct that we actually put on the world. Yeah, one researcher who's done a lot there is the professor Tanya Lerman and collaborators at Stanford. And she studies like hearing voices and various cultures. And so what is the difference between a society that says, hey, the voices that you hear are your ancestors and they're telling you cool stuff, or it's a molecular thing and it's all random or it's bad voices that are actually getting into your head. That is going to be shading the experience of the people experiencing voices. And so when we're talking about consciousness, what if one society, they tell everyone and multi generations, it's completely baked into their culture. It's like, yeah, it's an illusion, you know, denitopia. Everybody, it's an illusion and just the ideas that come to your head. They're literally defined by being illusory. And you're basically just in this perpetual auto gas lighting scenario. Just don't worry about the policies we're pushing on you, just go with it. And then another culture says actually the spark of awareness is what should be followed even if it appears to disassociate from social norms. That's going to be a different outcome. And so all of those different cultures could contain individuals that are 100% convinced that they're right. So being right isn't the same thing or thinking you're right isn't the same thing as having an answer to the consciousness question. And yeah, there's a lot of implications with how we actually are constructing our world and our perspective. So that is very related to Lee's point. I think he's frozen for a second, but yeah, I agree. That was a really good point about how it's like we're in feedback. We're doing niche modification for how we perceive and what affordances we have in the consciousness space. If you train up people in a insight meditation tradition, since they're children or something, they're going to have affordances in their consciousness. It's another framework, good or bad layer on top of that with the judgment. But they're depending on what people are taught about their consciousness growing up. That is going to color their experience and what affordances and what trajectories they take later on. So very interesting stuff. And I wonder like what the authors think about that or how they conceptualize these issues. So just to cover this projective geometry bit because they write the concept of 4D four dimensional projective transformation is central to the model. As it yields an account of the link between perception, imagination and multi point of view action planning. So they're going to try to link some of the cool stuff like we've talked about in this action perception loop. We've asked where's imagination, where is imagination and speculation and all these different things that we appear or experience engaging in. And then also how do we square the circle with the fact that we're so situated in our affordances. Yet we can also imagine things that are seemingly related to our four inches like a city's grid. It doesn't seem like it's tied directly to my affordances of course in a traveling sense it is but not like it doesn't depend on where I'm sitting in my building in a different city. So geometry is about the shape and the size of things whereas topology is about how they're connected. So in topology we talk a lot about nodes and edges and about connectedness and structure. Whereas in geometry structures related to angles and sides and edges and things like that. And Euclidean geometry is kind of like linear algebra is a specific case of geometry where certain prerequisites are true like parallel lines don't intersect and the angles of a triangle are 180 degrees. There are non Euclidean geometries just like there are non linear models and non Euclidean geometries which is kind of like the out group to the exclusion of Euclidean. So it includes a vast number of esoteric geometries and the way that you can move from some geometries to other geometries is through a projection and that projection is called a map. So it's kind of like seeing the map that's in front of you it's like it's a map it maps your representation onto the territory. So this is like a map that maps one territory onto a different territory or one geometric projection of a territory onto another. So you could map between two different kinds of maps of the world. You could map between a flat one and a spherical one with certain distortions taken along and things like that. But the key mathematical features are that in the Euclidean paradigm there aren't special points. There aren't points at infinity. It just is like a vast plane that doesn't end. There's no horizon. There might be an angle of vision that you can't resolve below, but there isn't a horizon and a flat plane. And then the origin is arbitrary in that every point you choose you could just say, oh, well this is, you know, zero zero zero and the local neighborhood is exactly the same here as it is everywhere else. It's all square. Contrast it with non Euclidean geometries which are like you can imagine though there are many non Euclidean types have features that include things like special points or points at infinity. Like there is a ring of points at infinity and they're all equally infinite far away. There can be horizons. There's non arbitrary origins, all this kind of stuff. And they're basically going to write the perspectival phenomenal space. So our perspective of what we're experiencing in a geometric understanding allows us to locate ourselves and navigate in an ambient space that is nevertheless essentially Euclidean. So you can be in a 90 degree room, but you can be having this like wacky distorted hyperbolic experience of this 90 degree room and kind of seamlessly integrating across that. Even if it was a bunch of cubes or connected, you can climb through it. You can navigate in this Euclidean world and measure the angles. They're 90 degrees. It's all Euclidean. But your perception is still going to have this distorted effect where things closer to you are seemingly different than things further from you. And that means that the angles as you look at them, unless you're in a very specific position will not look like 90 degrees. Look around the room you're in, even the corners don't look like 90 degrees. And they basically say because the Euclidean space has no privilege points, no points at infinity, no horizon, no non arbitrary origin. It very strongly suggests that a non Euclidean frame must be operating with the help of sensory motor calibrations to preserve the sense of invariance of objects across the multiple perspectives adopted and relate the situated organism to the ambient Euclidean space. So like put your hand on a 90 degree angle. You can pretty much feel from proprioception. It's 90 degrees. You look down. It's not 90 degrees feels normal. Everything's chill because we're experiencing this distortion of the Euclidean space. Okay. Any thoughts on the projective geometry or about geometry or where would geometry fit into a consciousness model? I just concur with them that there's some functional attributes that would point towards non Euclidean spaces. But then every time like, wait, what in the how with geometry? Just how is it giving us experience again? I guess that's sort of the question. But yeah, Sasha. Yeah, I also thought this was really trippy. I really liked the 9.0 kind of pre-digestion of this topic. So I found that really helpful. Yeah, I think this is kind of trying to walk us through the possible, you know, points of view on consciousness that not only does each individual have a different perspective necessarily in Euclidean space. Like we're all seeing it from different locations, let's say, but that even the mapping is different between individuals onto our own non Euclidean, you know, projection of experience. And so I think that's kind of many steps to walk through to recognize how different we see the world and how we can define this word perspective. So I think that is really an important aspect of understanding consciousness is where individual perspective is baked into the whole process. Yep. And it makes me think about like, let me see it through your perspective or do you see it how I see it. So we have visual metaphors for alignment and then we have put, you know, walk a mile in someone else's shoes or put yourself in their shoes. Those are embodied and active metaphors of understanding what else someone's experience is like. And then what we're saying here is somehow we should integrate that how you see it and how you walk it together so that your perspective is based upon both how you see it and the shoes that you're walking in. So your actual embedded extended body and how you walk in those shoes. And there's people for whom they can't walk a mile in any shoes. So what are other metaphors that help us go beyond just simple walking to really understand that more generally we're talking about like perception and action. And that because perception and action are linked in us, in fact, perspectively linked in us, it's also that way in someone else's. So it's not just about coming to agreement with what they think or believe or state they believe, nor is it about specifically reenacting their action sequences, which we wouldn't either be able to do or wouldn't even have the same perspective on them as they do. How are we going to collaborate? And so there's like a lot of interesting stuff and I would hope that this provides a little bit more structure to sometimes discussions about everything from issues people disagree on to accessibility. How could we, whether we bring in projective geometry or not, how could we just start with the idea that everyone is going to have a unique perspective. And then work for something collective rather than get to this point where it's like, well, we disagree. Okay, well, now what? Yeah, Alex, go ahead. Yeah, I just wanted to say this, this brings up it. It reminds me of general relativity in a way like consciousness, some sort of fundamental. It's a, if two folks are in different perspectives, and they can observe the order of events occurring in different ways, or if they can observe events occurring at a border based on their perspective and third remote thing that they're looking at. Maybe consciousness is the way of observant of a larger system when the system is essentially a manifold. The consciousness is the recognition of relativity in actual space. Right. And this projective geometry seems to kind of be getting there. If we are unable to observe the entire world and we're only able to observe from our perspective, consciousness is the recognition that that isn't a variant across space and other agents should have the same. I don't know phenomena. I apologize for terminology. No, that's that's awesome. I really like this. You said the recognition of general relativity is consciousness. And so in Euclidean space, it's like the invariant is angles going to sum to 360 and this is going to work this way and parallel. Okay, so those are geometrical invariance. But because the world is so quintessentially perspectival and embedded in an active, actually the invariant is this relationship of action and perception. And that you move in a apparently non Euclidean space because the invariant is actually like the locus of what is generating awareness is consciousness. And then the two examples that came to mind are first off, you hear a noise with each ear at a different time and you hear, you know, in photons probably, you know, hit the lights at your eyes at different times, but also they see different things. But you get a monocular perception from two different eyes. And, you know, but you can push one eye, you can violate an expectation and you can see two images, but basically when it works, it's monocular. And then I was thinking of a concert where like the bass is on one side and the treble speakers are on the other. And then it's like the group, they're actually hearing the events in a different order. One person hears one note before the other one and vice versa, but especially locally, that doesn't mean it's incoherent. So what is happening there? And that's kind of like this group consciousness question. And just thanks for bringing it to the general relativity level, because also with the general relativity innovation, it was a movement from a Cartesian Newtonian Euclidean. All these things movement towards like, yeah, maybe gravity isn't like a non linear force operating on a Euclidean field, but rather it is a non Euclidean distortion of another thing. And so there's really a lot to say there. And then also there's like a lot of inroads to quantum and consciousness, whether it's Hemeroff and Penrose and about the work on microtubules or whether it's just other more broader physics grounded theories. There's so much there about where quantum itself might fit in. So a lot of points of contact there with non Euclidean geometry perspective. And then the fact that we don't overlap with any other system in terms of our perspective. And so then how do we take that? And it's like, well, just like you can build a computer in a world of general relativity. Can't we design information and sense making and attention systems that work in a world where people have different perspectives? I'd hope so. Steven. Yeah, I like this, like really broadening the idea of relationships, because I think what's really helpful with this diagram is it and talking about the map as a way of projecting between types of spaces, rather than this idea of projections, representations in the mind being projected on the world, which it's kind of the slight danger with that language, you know, because it gets used in that in that representational way. So if it's somehow the way that we engage a type of spatial arrangement, almost bringing in spatial cognition work, but bringing it much more into the body. And I think this relates maybe some of the work with Harry Heft. He does some really interesting work in the world of spatial cognition. And it'd be interesting to see his perspective on this. Yeah, what you said about projection. So that's like in the psychotherapy. You're seeing how it is you're projecting, you're, you know, you're, you're, you're envisionate your bias, you're projecting onto the world, you're projecting that emotions onto them. But it's like, what if you start from, hey, it is a projection, you are in a projection, you are the projection, your experience is the projection of all these things that we're talking about. So it's like, let's start from that point of view that you're not treating this person like a robot, you're assuming they're having emotions. That's what's in this relational projection. That's intentional about your relationship. So how could we connect, for example, optimism about the relationship to how you are going to map their facial expression onto your behavior. Not onto what do they think, but on how you're going to act in a certain situation. That's a pretty different context. So I think that's like very interesting to bring the psychological projection that almost the psychotherapeutic use of the term to geometric progression. And I'm sorry, geometric projection, and then connecting non Euclidean geometric projections to general relativity and to control theory and about systems design, all these different things. So just a few quotes that we can kind of go through here. Basically, this is just a summary of what they believe the PCM's strengths are and how they summarize it, which is that the structure geometrically of multimodal lived conscious experience closely approximates a projective three space. So that's like a three dimensional space with kind of this geometers lingo on which a projective transformation. I believe PGL four is saying it's a four dimensional projection projected down, but we could hear from a geometer with a little bit more information here or what would the physics community notate this as. Our model fits intuitive data, but it also helps account for other phenomena that might seem at first unconnected, including working perspectival imagination inferences about the points of view and aims of others, like the intentional stance modulo active inference, and many peculiar phenomena associated with consciousness. So OBE out of body experience, he autoscopy, derealization, oceanic and mystical experiences. It also sheds light on some visual phenomena, which will definitely get to and it sounds like there's a bunch of papes and they're probably writing more. So if any of the authors want to, you know, share what they're thinking or doing now that's cool. Here is an example we talked about a little earlier, which is that when we consider like think about the Empire State Building. We think about it in perspective, like looking up at the at the Empire State Building or just floating away far away so that it basically looks Euclidean because it's far away. But it's going to be both conceptualized independently of any particular point of view like you could zoom in in your little brain AutoCAD simulator and imagine yourself on the 40th floor looking at a 90 degree angle in an I beam, and you wouldn't have any deception that it's actually Euclidean 90 degrees. But then at the same time, we can naturally co experience almost this like looking up at it and understanding that no, it's still not going to look 90 degrees. So that's what optimally allows for this projection, the perspectival accessing of Euclidean models, the world in memory for appraisal and updating of such models or active sampling. So we get to benefit from the sort of statistical reality, realism, let's say at least not reality of having a model that's actually rectilinear, for example, but then also have this total fluid projection that allows us to act and perspective a lot understand ourselves in that space. So there's the relative perception and then there's that perspective independent objective knowledge base, which doesn't mean it's correct. It just means that it's about an object. It's of an object. It's not truth, objectivity. It's objective like it's about this box on my desk and productive geometry helps functionally connect subjectivity. So being about the subject that's relational and an epistemic objectivity, again, not truth, just knowledge about objects, allowing us to see consciousness as a kind of mediator between the situated organism and the objective world. It inhabits very cool stuff. They also write so active vision thus becomes like a palpation of a 3D spatial user surface to feel the epistemic affordances that are quantified by expected free energy. This anticipatory palpation with temporal depth requires a generative model that encompasses the consequences of actions and their effective values. So in some other work by first and colleagues where they've looked at, for example, reading and informational foraging with the ocular motor with the eye muscles. And so the way that the eye scans around the page, it's not based upon maximal reward. It's based upon the maximally informative observation that could be made. And what is that observation maximally informative in the framework of, well, top down priors about semantics. So when somebody is coming towards you, there's like subconscious icicle movements that first assess, for example, the broad category of person that it belongs to, and then it will zoom in on different components of the face and culturally informed areas of the face that might signal different types of things like whether there's going to be a good relationship in the future or not. And so it is so interesting what the authors did, which is to tie this information foraging, optimal foraging, info foraging, info thermodynamics perspective. Take it from the cybernetics where it already had been very well entrenched and now say like, that's our experience of vision. That's why even when your eyes are circling at the level that you're actually perceiving, it's actually a lot smoother because our experience is sort of smoothed out over those underlying saccades. That being said, those saccades aren't driven in a top down way as a compromise with top down forces, let's say, by culturally informed priors. Your eyes won't know how to scan if you don't know how to read the language. You won't know which parts of the letter informative, which parts of the word. But when you're fluent in the language, then we know that like all the letters in the middle can be mixed up because you're really just doing a generative model of the text. And that turns out that it includes information about font and includes sentence structure, all these other things. You can't understand the sentence with the words mixed up in a different order. If the letters are in the right order, but the words are mixed, the semantics don't allow the top down to inform the bottom up. So your visual foraging is bunk because you're going to be looking at the difference between a B and a D. But you're not going to understand what the sentence is meaning is because that's the level we're actually at. This framing of expected free energy through projective transformation in imagination provides a method for envisioning how things would look feel or be from another hopefully better location and thus guide action. It's like, where would I be looking at if I were figuring out what this word were under my generative assumptions about language? It is important to emphasize that according to the projective consciousness model free energy minimization is always operating as the fundamental algorithm governing the field of consciousness. And thus there's never an end to the process as there's never in practice a solution that reduces free energy to zero, which because it's defined as this informational distance against the non stationary realism world. It means that basically it's always going to be bounded by zero. And I thought was very funny. Hegel might be happy to know that on the projective consciousness model consciousness is thus always somehow unhappy even when it reaches the best possible states available to it. So this is the well known Hegelian dialectic of thesis antithesis synthesis often shown as a loop as if it were just this run around rock paper scissor. Here it's actually like a virtuous or a vicious spiral and it just showing how this semantic ordering of events with thesis antithesis and synthesis. It's like it never ends free energy minimization optimization cybernetic control processes multi scale evolutionary adaptive systems. They don't end. They can't take a break. There can be relaxing times, but they can't rest in the actual like stasis. There can be stasis with respect to a certain dimension. But that's reflected by this seething activity that actually structures the strange attractor so that the system returns to these homeostatic preferred expected and optimal states. And if you prefer over evolutionary time something that isn't optimal or isn't functional, you're not going to exist. That model will not exist of the statistical regularities in their niche because their preferences, the observations they seek out will include things like toxins or jumping off of cliffs. These are not adaptive observations interceptively or from a visual perspective. And so you end up not seeing models that prefer those things. And so it's like this endless preference hunt policy adaptive selection stuff happening pretty cool stuff by the authors. Any thoughts before we just look through the figures before we close out. See you Sasha. Thank you. So just to spend a couple minutes on each figure and just see what they kind of do. And then maybe we could either reach out to the author or just everybody who's listening to this treat it as an open call to come participate if you're curious about this or something does or doesn't add up. Let's hear about it. Cool. So figure one is this account of psychological phenomena. And I think it's almost like Minecraft helps us understand their simulation. So let's just imagine that they're doing this in Minecraft. So in the A is like the builders mode of Minecraft, which ironically is still from a perspective, even though they're trying their hardest to represent the Euclidean world. It turns out that can only be done from like the position of a camera, wherever that camera is. There's no location independent representation of this Euclidean space unless they did like just a matrix and they said look just render it yourself. But when you rendered it, it would have to be from a perspective. You can't have all the perspectives at once or flattening an invention that doesn't work like that. But then the lived space S2, which is they're using geometric notation. I'm not totally sure about it. Maybe again, a geometry could come on. But like basically what they're saying is even if you have this block of interest, when you put a person's head there or their gender, the locus of their generative model, they're going to see things in this distorted way. B is the anti-space beyond the plane at infinity. So I don't know exactly what it is, but it was interesting. They said with a transparent plane at infinity. So it's actually bringing this like transparent notion back. But this is like, it's the opposite of what you see in some way. I guess it's like it's the yeah, it's everything what you don't see. It's like when you're watching the sunset, what is not being seen and maybe is the anti-space. And then this God's eye projection are like somehow a transformation of the space that includes a lot of the planes of symmetry. They write almost all the space is visible but completely warped and the world model manifests as structures with complex symmetries. So that's quite interesting because whether you agree with it being God's eye or not. I would say it's just another projection of the system or just another, at least as we're seeing it. But maybe it's in like a ton of dimensions or it has like other trippy features. But this is kind of like highlighting just the the the invariances of the visual field. But that doesn't guide action. Like if you're a baseball player and you're running towards the ball, maybe there's like, you know, the one symmetry would be all the grass when you look down and all the blue sky when you look up. And then there's a ball and that's like breaking the symmetry in some very nuanced way in a dynamic way. But what do you want to actually see as the ball comes towards you? Well, it's actually what we do see, which is that the ball gets bigger and apparently more manipulative. And we can do increased resolution on things like its speed when it's further away. That helps us estimate where to go. But when it's close, we can find tune. So it's like exactly what you would want as an embedded agent is what we're experiencing. And so we're not experiencing this like, whoa, it's just like hidden symmetries. Nor are we experiencing like the Minecraft or the Grand Theft Auto perspective. We're in our own embedded situated experience, relationally engaging with affordances and the field of consciousness in the field of affordances are very related. Interesting stuff. Great representations in very least just stuff to think about and learn. This one I thought was very informative. So this is a Q. We can think of this visual illusion as a Q that's ambiguous. So it's called the Necker cube. And basically, depending on how you see it, cough, cough, you can either see the bigger, wider side as like pointing up into the left, which is left and right already are relational spatial and down into the right. So it's like, you know, which let's just let's just have a quick show of hands. Raise your hand if in looking at either of both of these. Option one is going to be I see it basically going up to the left. Two is down to the right and three is it switches for me. Okay. Option one, who sees it go up into the left primarily. Okay. Option two, who sees it go down into the right primarily. And then who sees it just like rapidly fluctuating. Okay. Yeah. I'd say it seems like people raise their hand in every category, small sample size, but yes, it's an illusion. It's something that is ambiguous. The reason why is it's a three dimensional cube. At least it suggests it hints at being a three dimensional cube. But of course it just lines on a page. Okay. We unpack that in a very generous, optimistic way as being a two dimensional drawing of a cube. The child drew that you'd say the child is trying to draw a cube. They're not drawing like a logo with like, you know, a double helix and two, you know, it suggests a cube. It hints at a cube, but it's a two dimensional projection onto a piece of paper. Well, duh. But it turns out that without shading one of the sides or adding a light, which implies a perspective, it's like you can flip between these two solutions. So to speak of how to represent it. And so you in relation to this cube is actually this ambiguity that's being flipped back and forth between and then specific situations might cause it to prefer one side or the other. Yeah. Lee, what do you think about that? I think this, I mean, I'm really interested in this because it kind of relates to the point about unconscious simulations. So if you're looking at it and it flips, which you can do and it does do, if you start to pay more attention to that, you realize that it doesn't just flip. There's a sense in which there's a kind of an unconscious doing, which is probably I'm assuming to do with the predictions that you're making. So, you know, if I pay really close attention, what I notice is that I'm isolating or tracing maybe different portions of the cube in different ways and then labeling them in different ways. So I kind of go, oh, that's, you know, if this bit is a corner or that's the front or that's the back. So and it's that that actually makes it kind of shift. And I think it points to this idea, you know, like of the, you know, the perceptual world is kind of a co-construction in a way, but it's the, it's the, it's that I suppose what I'm interested in is that it's a kind of a doing, you know, there's a way in which you're active in it, which is more often than not transparent, you know, we don't necessarily, so for example, a more real world example might be, you know, if we look at a tree, we might label it an object, you know, typically we do. But, but actually it's kind of more of a more of a process, but we don't, we don't see it that way because we just kind of, you know, we look at it and, you know, for all intents and purposes, if it's not salient, because we wouldn't notice that. But anyway. Yeah, well, it's super interesting. And the semantics and the illusion, it's like there's the old woman, young woman illusion. And then there's like the bunny and the duck. And those, the semantic level can be like, I'm looking at a bunny. And then you'll basically not see the duck as much. Okay. So the semantics can inform us, especially when you label for action, like this is the side with a handle. All of a sudden be like, no, I can feel it dragging. I'm only seeing now one of those two sides fixed. I'm keeping my eyes right where I want the handle to be. And all of a sudden it feels easier to keep it in one zone. And then also it's like, there's illusions that are super ambiguous for some people and other people like, I'm sorry, I just don't see the young woman in that illusion. I just don't see the bunny. I don't know what I don't see that dress as that color. Remember that whole thing? So that's like because of the subconscious, you know, the iceberg of the actual neuro phenomenology, the underpinnings and the mechanisms. Some people are near a critical transition point in their perception where just by scanning their eyes, the ballerina will flip one way or the other. Other people, because of how their deep priors guide their visual foraging and their information foraging and their by generation of model, they just will never be able to have the ballerina switch directions, never be able to have the duck go one way. If you haven't seen a duck, you're not going to see it in the back of the head of a bunny. Yeah, Stephen. Yeah, I saw an interesting presentation by Lotto. He does some work down in neuroscience with Cirque du Soleil, actually. And he showed this picture. I think it was very similar picture. And he said that we mostly see it from above because we expect to see things from above first, which I thought was kind of interesting. So there's kind of like you do end up with this kind of inactive situated piece because it's like, well, I'm sucking something here. I'm probably looking from above. I'm looking from above. These things pop out. And then, well, maybe it's. That reminds me of children's cereal being put lower on the aisle to get them in their zone of affordances. It's in their proximal zone. So you can do niche based partitioning of surveillance capitalism. Great. I love it. And then also there's a lot of other natural situations where we can track what an animal looks at. And yeah, they're looking at things that are actionable. And that's like us grabbing something from the top. There's not maybe foraging for branches. You don't visualize it from the top, but that's something that we can do. But really we specialize on picking up things that were above. Okay, let's look at just this last two figures. So figure three, I think is related to this paper 2018 paper that they mentioned in the very end for more details on this family of simulations. But essentially it's showing like multiple views of the expectation map and the prior belief map and basically an agent has to move from the left side to the right side. So like it moves from here's its left here starting and has to move through his pit. I don't know if that's a bad state or what about it. But basically here it is blue. It believes there's a goal over here. It's going to pierce into this darker blue with its zone of proximal awareness. Essentially move through the pit resolve uncertainty about what actions to take here. The action path is simple to keep walking in a straight line, even though it looks like narrow. And then finally it gets in contact with the goal state purple. And so from its local geometrically distorted point of view, projected point of view, there's a free energy thermometer. And it's like first it scans around and then it's like, okay, I'm going to go that way towards the goal and reduce my uncertainty and then it's just getting lit up. And that's interestingly where dopamine comes into play. It's something I would love to talk about soon because I've worked on dopamine, the free energy principle perspective. Dopamine is usually thought about as a reward molecule. Oh, it makes you happy or it's the reward, you know, slot machines and all that. But if we think about it in this context, dopamine firing patterns in this phase anticipatory of the reward actually have to do with free energy. And when you get to the reward, your precision is confirmed because you already believe there was a reward there. So actually, and this is shown in figure four. As you approach the max zone of maximal information uncertainty reduction, that's when the free energy is getting out of control, so to speak. And then when the goal is achieved, precision is achieved again. So when we think about what the neurophysiology actually is and how different interventions in our neurophysiology might map on to differences in what about somebody who doesn't have motivation. Even though they know the goal is right here, but they're just not even if they can see there. And other people say it's well lit and you can just do it and they just don't have the motivation. Maybe someone else is here and they go, actually, I like the other side more or all these different things without passing judgment on what these states are, what level of analysis or what culture says they're rewarded or not. Just how can we map some of these cybernetic parameters onto the neuro and eventually the phenomenological parameters. And I think figure four gets out a lot of similar ideas because it's talking about this relationship between the lived space and the peri-personal space, one might almost be able to say, and also events that are perceived or imagined. And that surprise all and then reduction of uncertainty, the sort of realignment, reordering of either action or the inference side to reduce surprise about the world given this very deep generative model. That's kind of where it's at. And those were some really cool ideas brought up in the paper. And just if anyone else wants to raise their hand, just a few more interesting notes on this figure I thought were cool. They show a loop from prediction to the perception to prediction of the change in free energy, imagination and then action. And so we've been talking a lot about cycles like a like perception, action world, right, perception, agent, action world like that. But here this is a loop that happens in an organism or in a model of analysis. Still it's a nested Markov situation and it doesn't replace the other one. It's not the other one. The other one is about agents in the world. And this is actually like saying in these reflexive conscious systems, what is actually the loop that allows imagination, imagination events to influence action. So I don't know if this is and it shows which directions different things can influence each other. Here it has perception action influencing each other. So maybe that's the other world in the agent loop. But then here's these other processes playing out just very cool. And then here's like the vector of time on a free energy landscape through time. It's projected and it's going downhill, so to speak, through time in this like vector weird way. And as I understand it, and I barely understand it, that's where the math and the research is, is it's like, OK, chemical reactions always go downhill. Well, a candle burning is still downhill. OK, what about a cell? Well, a cell is also downhill in a really specific way. Well, all of our policy actions are downhill in what landscape? That's the free energy optimization. How is the landscape changing? That's a relationship of how our meta optimization proceeds amidst different kinds of environmental regularities. So yeah, that's all I would say about these figures before we close. Anyone else have any last thoughts? What would be something to go into more in 9.2? Yep, let's do Alex or then Stephen. I'll try to be brief. It's interesting to me, like relating this to the energy principle that you model these dynamic systems as agents, which are trying to export energy from their identity. By performing action in greater systems that are part of their action actually creates new unknowns that they have to perceive and solve. Maybe cognition is just the boundary between direct feedback and the larger environment they are part of. Maybe it is the self-referential structure that necessarily creates itself. Cool. Awesome thoughts. Stephen? Yeah, just thanks for unpacking these diagrams because I had looked at this paper a fair bit once in the past, but some of this stuff was really hard to, so this is really helpful. I think this idea that there is this imagination that's going on is really useful. I don't know how many other papers really deal with this, so I think that's quite interesting. One bit in the paper that we might want to think about that I did notice is this. It does say in the paper something about how this doesn't try and deal with all the subconscious dynamics underneath, but that it's about trying to find this kind of extra layer of conscious imagination or projection. It might be interesting just to explore that next time. Cool. Thanks for the suggestion and everyone. Thanks a lot for participating. We have follow-up forms for the participants. It's in your calendar invite. And via that mechanism for the live participants or just a Twitter or YouTube comment for anyone else, please provide us with feedback, suggestions, questions. We'll be able to go into them for 9.2. Good groundwork today. And then, yeah, 9.2. We'll be able to talk a lot more, hopefully hear from a bunch of different perspectives. And just staying in communication with team column. It was a great discussion. Great. Thanks everyone for participating. It's just like super fascinating stuff. So thanks again for the conversation. And we will talk next week. Everybody who's hearing it, you're welcome to join next week or the following week. But thanks again, everybody, and we'll see you later.