 All right. Hello. Welcome to Acton Flab, live stream number 39.1. It's March 2, 2022. Welcome to the Acton Flab. We're a participatory lab that is communicating, learning, and practicing applied active inference. This is recorded in an archive live stream. Please provide feedback so we can improve our work. All backgrounds and perspectives are welcome and will follow good video etiquette for live streams. To learn more about Acton Flab, go to activeinference.org. We're here in Acton Stream number 39.1, and we are learning and discussing this paper, Morphogenesis as Bayesian Inference, a variational approach to pattern formation and control in complex biological systems by Kuchling, Friston, Georgiv, and Levin from 2020. And we had a fun dot zero and time probably all individually going through the paper. Pretty much in 39.1, we'll just kind of go over introductions and then can go to a blank page and just see where especially Steven and Blue want to raise any questions and also anyone can ask in a live chat. Okay, so we'll just do introduction. So I'm Daniel. I'm a researcher in California and I'll pass it to Blue. I'm Blue. I'm a researcher in New Mexico and I will pass it to Steven. Hello, I'm Steven Sillett. I'm based in Toronto. I'm a sort of action researcher and practitioner in community development and participatory theater. And I will pass it to Dean. Morning. I'm Dean. I'm here in Calgary. And not much to say other than I'm kind of looking forward to hearing what Blue and Steven have to add to this, this rather big math question and how that fits with morphology. So back to you, Dan. Okay. Well, if anyone wants to start with a specific question or figure, or we can review just sort of the main points of the paper, like let's look at the roadmap. So what section would people like to enter in or have a question or idea related to? Or we can just go, yeah, Dean. So you wanted to talk about this and I wanted to talk about this and it was this whole section on the least action principle, which I think we both find quite interesting because, yeah, why? And it led me to the question of, is there, is there something efficient about search over somebody sort of passing along and sharing? So I'm kind of curious what people think of that in terms of the math and the morphology. Least action, math, morphology. Steven? Yeah, I don't know if we can unpack a little bit more what Dean was saying there. I'm interested in some of the implications of that work around looking at the math and morphology as it scales from these, what you might call smaller scales than we can normally perceive. Often what's going on with morphogenesis is not something we can be consciously aware of a lot of the time. However, who we are is built upon morphogenesis. So I'm very curious around how some of these first principles can be thought of as rolling out. So we can talk about least action as being one way of approaching it where people often think about energy and efficiency of energy processes as being the route that everything will take, the more informational Bayesian inference approaches and learning that is implied by this without sort of distracting from the paper. I'm curious about people's thoughts about what sort of implications that has for thinking about these bottom up ways of knowing. Okay, Blue? So I have some questions going back even to before least action in the paper, talking about generalized flow. And it's really interesting just in light of the paper that's coming up in number 40. But here we see that this is like a spatial, like we're talking about spatial motion here or like the foundation for the way that they model morphogenesis is positions of particles through space and time. And I just wonder, we have a time evolution here and we're talking about motion in space. But could we do the reverse and talk about a space evolution in terms of time? Like in terms of like, can we invert space and time here? So that's just it's something that I was curious about if it's possible or mathematically illegal or why am I even thinking like that? I don't know. And another question that I had, you guys were really talking about you know, the in the generalized flow, we talked about the equations of motions, coordinates of motion, and the derivatives. And something that I saw in the paper specifically at equation 15 that I've not seen before is P with a dot. And it was not defined in the paper P with a dot, but I'm assuming it's the first derivative of P. And I just thought that that was interesting. So it's the probability density, but is it the derivative of the probability density with respect to x with the tilde? Or is it just P dot because x has a tilde? Anyway, that's my question there. All right. Wow. Well, we have a lot kind of on the table. I think the thing blue appended with generalized flow is a good entry point because this is sort of what the systems state evolution is. And in live stream number 26 with Bayesian mechanics, we also talked about the generalized coordinates of motion with relationship with control. So it's kind of like a thermometer cybernetic model. If you only have the position of the temperature, there are certain strategies that can be applied, given that modeling just simply the temperature. And then if you are modeling progressively higher and higher levels of how the system changes, it's kind of like a Taylor series, it equates to basically a better depth of control. And then part of the hard part is that because over the time course that action is being planned for, even hypothetically, changes in like the unknown consequences of one's action complicate the estimation of different policies. So this is just not like a total perfect way to roll out the system and understand how it's going to happen. But it gives like approximation terms to higher and higher levels of the system changing. And so that's the end all of the levels of analysis are being linked to each other by being derivatives of each other. So then some stationarity like is capped out at some point. So the derivative becomes zero. And then that's like a heuristically approximatable depth of that model. So that's just like one note of what is being estimated on. So it's like temperature and its higher order components in some either analytical kind of infinite generalized way, or just more practically with some smaller realization. And then the dot is like a derivative pretty sure to blue, like it's like a prime and like prime prime for double derivative two dots for two derivatives. And then morphogenesis is like position of particles. Here it's like the cell body or the cell nuclei, or the center of gravity of something. And then those are spatially moving. So that's the lowest level of this chain of the integrator chain with the generalized coordinates of motion. So that's how it's from a Bayesian mechanics, getting to morphogenesis with just how particles are moving in space. But instead of just modeling like their x and their y, it's like x and y and the higher level derivatives. So that's the developmental motion of particles. Just summarize that part before the space and time. Okay, any thoughts on that? Because the generalized flows definitely where whatever least action is is going to be based in. Okay. So then let's just look at that formalism. Here is the derivatives. Okay, anything on this? Yeah. Yeah. Yeah. So if you're going to look at the at the formalism, did you pull it up here? Yeah, generalized. There you go. So go ahead and then I'll I'll interject. Yeah, here's 2.2.1. X tilde is the total blur of all the dots on top. So yeah, go for blue. So just here, my question more is the noise and fluctuation term. And is that just like Brownian motion or I don't know. So yeah, like what is the where is the omega random fluctuation? Like, is it just, I don't know, random in in terms of Brownian motion? Or is there some other random fluctuation positionally? Like do things wiggle? Definitely someone who knows more about stochastic processes. And I know several people listen who do. Could help with this, like in a math stream on a little bit of this stochastic aspect here. But I think at a first pass, there's situations where the what is modeled as noise versus what is modeled as signal have differential ratios. So there's times where like the regression is very the F, even if the regression and other functions like very tight, and then the noise is modeled as low versus one where just the location of X is just dominated by the noise term. But I'm sure it's way more than that too. Stephen or Dean. Well, I was just going to, you know, often when we see I'm just thinking broadly in terms of the modeling approaches and active inference, often it brings in temperature as a way to sort of bring in the kind of noise fluctuations into models. Whereas here is directly seen as a noise fluctuation term. So I think that's maybe it's again, because we're going down to even more first principles here. It's getting down to stochastic dynamics, which isn't necessarily even thought of as temperature, but as something more foundational. Okay. Yeah, Blue. So temperature is like, yeah, right, like the increasing temperature, like random motion would increase. So that does make sense to me. So another point out of this section, or maybe if anybody else has a point here, going down into the next like set of equations, unless anybody else has anything else. Yeah, just to summarize that it's kind of like the temperature parameter is understood sometimes to play this role of mediating the difference between the particles, motion being driven by some sort of directed motion, like change in position through time, not at stationarity of position, which is like kind of the observable in that chain. That directed motion versus thermal slash Brownian. And then that might be kind of where these statistical assumptions come into play. So it's like variance estimation, but temperature is just one kind of variance with the motion of particles. And it has a physical interpretation and I mean, it's not insignificant piece. I don't know. We could explore more where temperature specifically comes into play, but Stephen, I think you're getting at the right thing that this is not temperature bound, but it does relate to that sort of situation. Okay, blue. So I don't know if you want to go down to equations 15 and 16, probably, where my next question or comment is, do you have them in there now? Oh, copy them right now. Cool. Go for it. So it says equation 15, I think that's x tilde dot, yes, p x tilde, is a probability current, which I thought that that was cool because I've never really heard of that like term, a probability current, like how probability changes over time or how the probability density changes over time, like what is that a probability current, like that was just, it's an interesting, I don't know, like term that I hadn't heard before. And then in the second equation 16, it says that this is a partial differential equation that describes the time evolution of the probability density under dissipative and conservative forces. So where the term on the left is the dissipative forces and the term on the right is the conservative forces. And I thought like that that was interesting, especially relating back to the dot zero video, where you guys were talking about a tensor and Dean was talking about tension. And I think that there is like an interesting tension here between like conserving and dissipating. And it just makes me like wonder how, how like, like is death when like the dissipative like term is winning? Like is that like what causes like a system or I mean, I know that the term, you know, winning or term being greater doesn't cause a system to die or when the system is decomposing. If that can be represented here in terms of these dissipative and conservative forces in the process of an organism dying or ceasing to exist. Okay. Sorry, I have way more questions than. Nice. Nice. Let's, let's look at 1516 a little bit. So yes, probability current, it's kind of in this dynamical systems framing. It's being bridged with formalisms that are like more like a current, like a flow on a dynamical landscape. There's that sort of angle, but then they're and maybe there's also more of a fluid flow component blue. Sorry. I was muted. So like it's funny when I look up a probability current like also foreshadowing foreshadowing the dot zero, it says in quantum mechanics, the probability current sometimes called probability flux is a mathematical quantity describing the flow of probability. So it's interesting that it's related to quantum mechanics in terms of these like instead of these classical physics things that we've kind of been looking at or dealing with so far. Yeah. So this 14 is, okay, this might be only a partial element or example, hopefully one that is useful. So this is x tilde with a dot. So that's x tilde, all the coordinates of motion and then x with a dot means derivative of. So this left parameter with a tilde and a dot is how the generalized coordinates are changing through time. So that's kind of what is being modeled, the data that are being, whether they're an observable or an unobserved estimated state in a Bayesian graph. So there's an O or an S that plus it's all of its derivatives. And then that is being partitioned into one big complex function f modeling it flow. And then this noise omega tilde over both are over tilde. So the these are both over like each of these higher derivatives also have an associated noise term. And then it says, yeah, okay, Stephen. In some ways, there could be an analogy. I'm not saying it's the same, but there's a there seems to be a mirroring of the enthalpy turn and then tropic turn should get in Gibbs free energy where one is where there's this the energy is bound within the molecules at play. And then the other term this entropic term where it's kind of dissipating. So it's another because these things are is translatable just like mass and energy are translatable. And they're known to ultimately be conserved. They can only be the overall mass energy of a system is always conserved, but it could be translated between normally, we don't think about that because it only happens in nuclear reactions. But if you're thinking informationally, it's still and this and I see they talk about this idea of probability mass. So that there's almost an ability to think about that conservation of energy in a form which has mass which has form, as opposed to when things have dissipated as light where it doesn't have mass anymore. Okay, blue. So I think where they're talking about the conservation of probability mass, I think that they mean like the conservation of the total probability there, like so the probability mass function, right? Like, so I think it's that which relates to like the it's the conservation of mass, but it's also the conservation of information using that conservation of probability, like the probability of all of the possible things happening has to add up to 100%. Like that probability is never going to change. And so I think that that is there. That it has to add up to one, like a fraction or 100%, has the realist and an instrumentalist take the realist take is something has to happen. And then the instrumentalist take is, if we model it this way, we're going to be able to use statistics. But if it was like, well, the chance of getting heads is 60% and the chance of getting tails is 60% in this next flip. It would open up the space in a different way. It's kind of outside the bounds of the model, you could have a model where both happen, but still there's it's a different thing, you know, can you model things that can't happen in the real world? Or if it's not a good measuring stick for statistics, which is kind of like that one, something must happen, we have to model something happening, we have to model a data set row being added, otherwise what is happening if it were like one, one every one every 1.2 rows. So and then here, yeah, Steven. In relation to the point you just made, I'm just curious whether if these functions can be used in multiple particles, in say, for instance, you have the probe, the idea is the probability of a coin being flipped is 5050, for instance, or a rigged coin could be 6040 in one way. However, the probability if there's noise on how rigged something is, so for instance, you've got a coin and it's flipping between being rigged 6040, 5050, 4030. So there's kind of a random fluctuation in that behavior, which may be contextually contingent. That in itself wouldn't be in the one equation, but if you had multiple particles, they could start to behave not just as one predictable thing, they themselves could have the change in probability approaches. Thanks, Blue. So I think like specifically as related to here, and I think that this is derived like I'm not, I don't like really know, I don't really do these like derivations, it's definitely not my area of expertise. But I think in terms of the conservation of mass, like we're not creating or destroying matter. The conservation of the probability mass, it just says that all the particles, if you're looking at it doesn't matter if it's one particle or a thousand particles, the particles are somewhere, right? Like so the particles are not nowhere, like the particles are somewhere. So it's that kind of concept or that's like the idea that, you know, all the probability has to add up to one, like there is 100% chance that you will find the particle of interest somewhere. I mean, you're right. Okay. Yeah, Blue good section to pull out important some important bridge equations. So we have the signal to noise. Now when the signal that we're modeling the instrumentals take, we're modeling how much the cow weighs defined as the one that comes to walk over to me when I call for it or something like that. That could be dominated by different functions or could be driven by different noise on different time scales. So the signal sort of directed motion component or modelable component of the motion through time is the purple. And here is like some flow with upside down triangle that we talked a little bit about in dot zero, but it would of course be good to hear from someone who knows more about this flow operator object. Signal is like the movement on a probability distribution on those states changing. And then the noise is the dissipative component. Steven. It appears as well that by breaking into the dissipative and conservative components, it's serving a similar role to separating into complexity and accuracy. It's giving a way to break apart for dissipation where the because you're not actually talking about a final target, but more like you're things just doing what they do in some ways and having a very kind of physics fundamental process. I think that's proving to be useful because it's also again something which can be translated into different types of math, different types of physics. Okay, so let's connect it to the morphogenesis. So what do they do in the paper? They're modeling the motion of cells. So why does it matter this whole discussion that we're having about how to get from the motion of generalized coordinates? If you're the modeler, you get to define that and get to set the parameter between this. If you're the experimentalist, then you're inferring this kind of like a regression. Like if this was a regression term and a noise term on the regression that and then there's some function that helps you fit a certain solution to the one linear regression that fits it best. So it's like kind of in that same genre here. So if we take the real estate on that, it's like how the actual system is defined as or inferred to be as from experimental data and then flowing the probability density on that of our inference on that, whether we're getting that from the data or whether we're defining that system formally. So that modeling, we could help from the authors or someone else like it engages some new future. It's like a bridge to some other set of equations. So that's where it would be good to learn more like what is actually enabled here, but it enables a new vista that separates, that enables this separation. So we can talk about instead of this like potentially extremely open ended and nonlinear change on the system can be bounded within a statistical inference zero to one inference framework that is more amenable to certain kinds of analysis. So that would help model cell location and all of that. But I think there's still a lot more to learn there and unpack. Okay, Dean, yes. So I want to just sort of step back for a second here because I want to go, we started out talking about so what are the implications for least action. So the act part and then we sort of stepped into the idea of what came before the action. Well, it was these flow states now kind of want to go back to the action part in this in the under that section 222, the second paragraph says that the least action principle can predict the emergence of form. So what Daniel was mentioning there, that sort of the morphogenesis in terms of the flow or paths of least action in biological systems. For example, in colonies, ants find the paths of least action to harvest food and bring it to the colony. The example considers their paths as flow channels or trajectories finding the least average action for each instance of foraging given available resources. So this gets me thinking about, well, what do we mean by least action as opposed to say most action and that that that general way the general way that things tend to follow. Like if you thought of an analogy of maybe a 19th century war where armies would line up facing one another parallel to one another. This is speaking more like a flow state more like you find in nature where the the molecules we talked about it here are sort of flowing and following down the side of the bowl to the least potential energy. And so now my question is, okay, so why does nature tend to flow with the advantage of least action, whereas people with instruments don't necessarily flow or follow one another unless they're mirroring some other form or some other shape. So it's interesting to me because I think by nature we are quite adaptable. But when we start making non living things as instruments to necessarily take the nature away the life away, we're maybe not as quite as adaptable. So I'm trying to bring it I'm trying to step back and now pull us back into this idea of so what advantage is there in this physics view of least action. There must be something advantageous there in nature. So what is it? I'm not the geneticist. So I don't know what what the signal milieu is that makes that so. So I want to tap your guys's and Gals's expertise on this because I don't have it. All right, thanks Dean Stephen. So following on what Dean's saying there, I think with them in the traditional sense we tend to take what we call equilibrium dynamic approaches. So for instance, the arm we the armies are lined up like Dean said, then there's a battle and then there's the final state. So we talk about the initial conditions, which is equilibrium everyone lined up and the final state. And then the bit in the middle, we sort of talking to our shirt or our shoulder or whatever cough a few times and somehow it happened. And in some ways with the biological approach, it's like it's in that because all that bit in the middle is where the flow is happening. So essentially speaking, and biology doesn't have those initial conditions in the sort of traditional sense of pure equilibrium. It's always in this non equilibrium, or at least a large component is in a non equilibrium state. So this is an interesting thing is it's flowing down towards the least energy. It never reaches it. So it reminds you I think that's a way. So for instance in the chemical reactions, you take your pure reactants, you mix them together, you stir it up, you do the reaction. And then once you've got your product, which could be a precipitate, for instance, which could be a ton of the precipitate, it then gets filtered, it gets washed, it gets dried, and it gets measured for how pure it is. And now it's in this kind of stable product form. So I think, yes, this question about what does it mean to flow is really important. Thanks, Stephen. All right, Dean. Well, what this what this flowing and following thing kind of says is that that the desire line or the termite mound, the result is of signaling, which is, I don't know if that's collective signaling. I don't know if it's the stigmergy part of it. Like I don't know why nature sees the advantage in the following. But there has to be something there. Why is that advantageous? Why is that more adaptable versus the alternative? Okay, thanks, Stephen. Well, I suppose one way of looking at that is it's the only rules. I mean, if we're going to take a realist route, recognizing that it maybe it's the only plausible route on the table, they don't get to forge teeth in a kiln or in a way that's partly why all chemicals that are available in nature have to be within certain plausible temperature scales and pressure scales that can yield those react those products at. So, I mean, that kind of doesn't answer the question exactly because I know that's kind of a bit of a but it's part of it's what's available, I think, in terms of biological plausibility. And I suppose in some ways humans we've adapted a cognition to try and move outside that. Okay, yeah, Dean. Now, one last thing on this and that is that following implies that there's also something leading. There always has to be something taking that that initiative for others to fall in line behind. So, shifting that and making leadership something that's more lateral. So, sort of going forward together as opposed to who do we decide to follow behind. That's an interesting thing in terms of sort of assessing out how this least action principles things turns into things that we actually see phenomenal phenomenologically. So, again, we're going to get into the what what the final form takes. I just think that when we were sort of treating this in the point zero. This was kind of pivot a pivotal moment because it went from what are all the things that have to be in place for formation to occur to now let's let's look at what the what the form now results as as I said in here in that one paragraph it said well the previous paragraph says physics offers a useful formalism to understand at a quantitative level the ability of biological systems to work towards. So, now we're we're basically now moving to something that is that is adaptable. So, I think like I said I don't want to over or talk this one point but I think it's an interesting part in terms of especially when we get into the the discussion at the end around why what with some of the assumptions we're in yada yada. Okay, thanks. Yeah, Stephen. I suppose this work towards idea I mean that can be thought of in as a model in and a life realist problem like how how does that work towards and in some ways accuracy and complexity can be used once we if we're at the level of assuming that we are actively inferring to stay alive but prior to that something like conservation and dissipation or conserved are a bit more foundational okay because if I conserving or working within this conserving dissipating noise dynamic at the levels which are below what we are consciously aware of however which all our cognition is effectively built upon somewhere we we're able to potentially give that towards this without necessarily having set a goal and it could be and I think that's that might be an interesting see how that rubs out in the real real world of active influence flabs and papers and stuff. Yeah cool yeah a lot you link there Stephen just now like accuracy complexity and then okay we'll leave it for a future day and work and who knows how many of these are just concordances versus tautological but accuracy within a model is like pragmatic pragmatic aims within a model with the imperative to prefer to fit as much data as possible and then it's like conservative using the reward structure that worked at one time step using that generative model moving forward versus changing it any sort of dispersion around the parameters as evidenced at the lowest level of the chain the position of the particles or the actual parameter that's being modeled like the actual image that's being percepted on versus these higher generative models which don't realize like at the kind of you know tip of a javelin so to speak so then the signal and noise is also like conservation and dissipation with respect to the model and then that is coming all the way back to morphology with like the position in stasis versus vibration of the particle from thermal so it's like a lot that gets linked here but where's least action in all of this like yeah blue so just to read a quote from the paper it says since self-organizing open systems are not conservative their structured flow is quintessentially dissipated dissipative and so and that's like you know down it's in the third paragraph in the least action principle section so it just that goes back to what i was saying um earlier like does that equate to death like when a system totally dissipates like it's ultimately dissipative so you know we stay in this like nest for a little while and then we dissipate right like is that can it can it be modeled like these processes using these equations i find that super interesting yeah um thanks steven this this um this idea of the least action i think timing with what blue was saying there as well is um it's least action maybe we should say dynamical least action as opposed to least action which we normally think of i well like i was saying reactance products what was what's the what's you know what's the route to get from a to b or the army lined up to final space normally we are taking starting conditions ending conditions or you know where i am now what's my goal what's the least to get there but those two things are defined in some ways they're they're they're definable so i think it's a good now all the bit in between is um trying to go to least action um in some way using that kind of in the space of chaos i suppose if we're going to take the kind of complexity approach in human systems um it's in a chaotic state where you're trying to at some time um make sense of how to act but however um you know in some ways there's only chaos moving into complexity which is available at so biological systems theoretically maybe not the idea of complicated and kind of simple sort of reproductive sort of steady like equilibrium state systems like mechanical systems aren't necessarily available although you could argue that maybe certain properties of our morphology like our um our bone structure and that sort of gives something which is approximate to that um once we know how to move our arm we have a relatively simple thing that we can now bring into our regime of attention but uh yeah i wonder what people think of dynamical least action is is somehow what's going on here yeah thanks even just here's one take on that it's a good idea a question i hope this is accurate too because i think we're all learning here but this is definitely a really challenging area in some ways to approach especially given our realistically limited familiarity so again it'd be awesome for people who have more familiarity with these equations to join us like either in preparation or even just joining us on these streams so that we can actually learn and connect to these other areas somebody who sees this formalism like every day just like we might see some other one okay but Stephen brought up is this a dynamical least action okay so first kind of a hopefully non contentious point that we're analyzing what we model we can only ask the computer to calculate the numbers that we ask it to model and add we can't expect something that goes truly beyond that not to say that the models can't have surprising outputs or interactions etc but we can't have it go beyond what we specify and the model is over x tilde which is the generalized coordinates of motion which is the position and all of the higher derivatives the whole integrator chain position in all higher order moments of the statistical distribution they're called moments like the first second third moment of the derivatives of the statistical distribution and these are basically terms of approximation of motion that allow a high order like a Taylor series approximation or a Volterra series approximation to allow snapshot modeling with real-time flow action cognition perception it's the flow over as gets explored the blanket states this is like the flow over s a and i states actions and um internal states so like this is enabling a flow description of Markov blanket states and their perception cognition action their flow over states they're changed their time including higher order moments of time so that it is a dynamical systems model but still there has to be a snapshot like a time series model of a stock price it has a value at whatever time resolution or continuous or discrete it's being time series model that so it's kind of the relationship between snapshot modeling and capturing higher order trends and this is a certain way to think about that in the direction that they're going to take it towards the Markov blanket partitioning and bridging it to everything that that affords okay um Dean and then Stephen and then so this is really important because i think it i want to tease something that maybe we can look at right right here but only in the point two so between the us doing the live stream zero and today there was a report about the rate at which climate i think it was a un report the rate at which climate change is happening so rapidly that as as the people who are sort of in that highly changing flow state aren't able to necessarily going to be able to adapt to the rate of change and so i wonder if some of what this paper speaks to might help us understand why the authors of that report are indicating to us that the the environment in which we we are existing in is going to change so quickly that as cells within that larger structure our form is not going to be able to adapt quickly enough to the the way that the external system is is changing so i want to say that for the point too but i think this least action principle part would be a good entry point into some of these bigger questions going forward so i just want to park that but i think it matters yeah just to give one note on that the rate of change and how it changes those are kind of natural language descriptions of derivatives of how things are changing might be a simple claim to some but that's really important to keep in mind so how things are changing and then that's always going to be unknown to some degree so how things change and so on this is the generalized coordinates of motion and so it is about modeling rates of change and predictability of systems that have rates of change and okay yes steven with this we mentioned there about your action perception cognition and in some ways the cognition is cognition on the generative model and the generative model is an action policy model okay so we've got this recognition coming in but it is derived by this action and of course one challenge you've got in the climate change scenario is really what you think is one thing but being able to act and being i mean the danger is we hit a point where no matter how much we know we it's beyond the ability to act in our capabilities and capacities as a species so this while i think the word cognition is useful maybe useful to be careful just that the inference process is more general because it can be a process which is like say happening beyond thought so thought is a is another action almost giving a higher order understanding on the generative motion product generative model for an action inactive inference and of course a lot of what's going on underneath is is beyond our perception literally way beyond so that that that can be interesting or useful to strip out nice yeah blue thanks so just to kind of go back to dynamical least action i do think that that's what they're referring to at the very end of the section the authors say from our perspective the key observation here is that any dissipative random dynamical system can be formulated as gradient flow on the log likelihood of its states this is reflected in our solution to the Fokker Planck equation in 17 which means the action is the time or path integral of the marginal likelihood or self-information for any system or model m so this is really the key thing this means the least action integral over the Lagrangian turns into an integration over the self-information of states which is known as entropy in information theory in short the principle of least action manifests as a principle of least entropy for systems that possess a random dynamical attractor and thereby obtain non-equilibrium steady state thanks a lot blue great um point good day france hi hi sorry that i'm late and uh i only have a half an hour too i'm in the middle of finding preschool so daycare is a whole mess right now i do apologize i do want to take this time it's a little bit of mess next week should be better i think we found a school so hopefully next year we'll be more easy cool unexpected but preferred you know how will we model that uh but thanks a lot for joining this is really cool so we were just describing some of the formalisms but where would you like to begin it'd be awesome to hear like any just introduction and context on the paper and then we kind of jumped in at least action and would love to hear your take on that sure um all right full introduction so i guess um i would have thought a little of myself so um i i work in my clever slap and and taffs and doing a phd in biology and this is the whole kind of active inference um part of my work is came actually before my entire phd so i joined my slap so i wanted to really learn what kind of information and physics do in a biology context so i didn't i didn't want to study the whole level of you know i mean i'm still doing it as a technique but i didn't want to have to focus on protein interaction and you know genetic information um where you measure all these things so i was based on one something a bit more broad than that and so when i reach out to mike and once i got semi accepted he i was like you know have some months to kill can you send me somewhere or do something cool and then he sent me to call first and that's when that's the first time i heard the term active inference so i spent about four months in carl's lab um and then basically started working on on on exactly i think however half the work in the paper that's actually done in those those couple months and when i was in carl's lab so the the overall goal of the paper and of that part of my work is to see if we look at morphogenesis and similar biological processes can we look at something where you have a very kind of um kind of baseline stem cell like behavior where you have cells that cannot do like morphogenesis basically i'm much you're familiar with the whole biology aspect of this paper but essentially it's the ideas you model cells that are uninformed they have some kind of you know genetic code something that encodes their their their structure that basically you have the genitive process already ingrained in the genitive model into the into those cells but they're completely right there stem cell like they don't have actually yet achieved their final form um and they're just like in the simulation and biology right they're starting off a very few asymmetries i think that's a famous who said that but uh some physicians that once said most of physics is basically just uh symmetries that's symmetries everywhere and that's exactly kind of um how most biology things biologists think of of morphogenesis there's some asymmetry and in the beginning in the egg where some there's some some different information some gradient of something some localization of certain agents and then from that on basically everything else just follows through but it's it's it's hard to believe that's the whole side of the story because there's just so much complexity and so many cells doing this at the same time so if there isn't any capability for them to adapt to signals in their environment and to learn from each other it's hard to believe that the time frame they're given they can really um you know achieve the complex morphogenesis outcomes that we know um do exist so coming to this paper basically this model is actually already um like the the father or mother of this paper was already done before i even heard of this topic that that was the knowing one's place so the model structure itself was already done by i think how that most if not all of it and Mike just basically was on the paper and got Giovanni as well they basically gave him some input on on on what what the background of biology would look like and so they already figured out basically let's do a simplistic model eight cells there's more as far as i know from carlo and as far as myself and my simulations many practical concerns don't want to complex if you have too many cells at once um then then you don't mind the problems in the beginning something i i actually add on to the model um even the baseline model was that damn thing parameter um which i talk later on in the results which might often like it's not a key part but i find it very interesting because it's one of the requirements for active infants um for for phenogenization variation of phenogenization there's a smooth landscape and when you have all these cells initially clustered together and they're trying to infer their place and they all kind of have some random randomly initialized prior beliefs they don't know where to go from there and if you right if you have already high precision in there and their sensory apparatus initially they have a really hard time for they still end up mostly fine getting later but they jump all over the place and you can imagine that'd be very bad for an in a biological scenario if every cell immediately jumps to the first cue they have we actually think now in the mike levin labs that love cancer initiation happens a little bit like that where repositions are as i said too high there's paper coming out soon by a colleague of mine that talks a bit more about the argument but so basically from um we had the target morphology already um carlo's encoded that and you saw what you see in the in the baseline control experiments was already done well i was interested in is basically again coming back to my my question and my approach to this from the biology side of you is how can we use the aspect of information flow and specifically active inference to manipulate and better control morphogenesis outcomes and control biology so the two main results in that in that paper are looking at if you instead of you know instead of like a normal biology experiment where you basically interrupt one part of the machinery um and then see how everything reacts you actually control the process information processing itself so one was basically if you put in a an asymmetry and the response of the cells to the to the signaling um ligands to get the signaling concentrations in the environment how can you basically completely remodel the entire morphogenesis outcome even though you actually left the entire coding itself that's always one thing i tried to stress in that paper the target morphology in all those relations is exactly the same none of them they basically start exact coding um of what red is supposed to be but how the process information has been changed um so the first figure is that where you have those two head and two tails um i don't know if i did the two tails and that or if i just did it myself and then publish it i think it's in there as well um then that's basically something that we also see in the lab i think that was mentioned as well that that's something that that my 11 lab has done with an area where they basically induce those two head types right that's something that's really weird to a virologist to and again in the type of um manipulations that we do in a lab that have been done in this lab they didn't manipulate the genetic and the genetic code again the the gene a from those scenario that was exactly the same but some of they perturb the actual biometric network um itself so basically this on the state space essentially so that kind of like where this inspiration came from and then the second part where we the the malformation or i also talk about cancer information and of course it's it's completely called cancer because we didn't put proliferation in this in this simulation so that's it's the initiation stage but it's the idea is that again under something that we talk about also in the lab and we've called many emails afterwards and it kind of also goes back to this whole idea with um how how different psychiatric orders work in the brain where you have some of these inference processes being disrupted and then lead to large-scale outcomes but they kind of start somewhere so the idea was here is that if we disrupt only the um the information flow from one cell to the other cells and the other cells to that one cell then what will that what would that cell do so basically just completely reduce its um sensitivity to the environment um and then saw it happen and what happened was that the this one cell basically kind of like it didn't it didn't move a whole lot it got it inferred the wrong kind of cell type um and it was completely out of place right so that's always batten biology if a cell is doing something that's not supposed to in a place where it's not supposed to be um and what i thought was interesting afterwards well okay so how can we rescue that and again like right i'm not trying to rescue it in a traditional cancer therapy idea where you bombard the cell or just kill it how can you actually have the system remodel itself so that the idea was to increase the um so the sensitivity was still disrupted from that one cell but then the flow like how much concentrate how much the other cells reacted to it so basically we manipulated how the information flow from at the at the other cells to that cells was flowing and increased that so and then what was interesting is that the kind of they were just kind of like going more closely into an interaction which is seen this this time lapse figure um and then eventually they reshaped so even though the cells like normally if you run this it's it's a deterministic simulation right so like um it's run it's always the same random seed so you run the simulation and you run it again it will be some outcome so you can kind of know which cells go where even though you know no cell starts off with like knowing i'm supposed to be in that cell that's not how it happens but because we run the simulation with a fixed randomized seed we always know which cell ends up going where and then in that cancer simulation that was not the same anymore so actually even though at the end after that rescue experiment they ended up with a perfectly normal shape in the mythology at the end it wasn't what normally was supposed to happen right the cells had normally were gone to their fixed positions didn't all do that so there was some reconfiguration which is something that that's that's an ideal strategy that you would like right something that you exploit the systems interaction itself to then rescue it and overall talk about you want to work on the the fume type but you want to work on what's actually wrong you don't care exactly what cell does and and is implemented there's a lot of cool signs little tangential but that's really cool work by eve martyr that works on um neural networks and and lobsters where they see that this works last because it's a simple enough system to work normally but what they're seeing is that there's actually a multitude of implement implementations that the neural network can can use and do use individuals to use entirely different parameters with like two to four four differences in ion channel concentrations to achieve the same end result and it's actually the the response to resilience to stress to perturbations in the environment isn't in the details but in the end the actual homostatic aspect of it is kept the same right the goal of having something um end up the same steam would like to speak uh go ahead oh hi thanks yeah i was just gonna one question is the idea of flow in a way more flow is better even though that um so unlike i say a genitive model where you maybe don't want too much complexity you can have a lot of flow but as long as that flows not noise um it just is able to inform so that's the first part and the second part is are you saying it's it's with this flow it's about finding where the choice point is or where the threshold is to say okay that's now going to decide that that's going to create a morphological target okay as opposed to it just being a gradual gradient descent on energy it is more the gradient aspect so the it is i would say it's it's not ideal to think about decision points even though in the end it doesn't happen it's something that we're just we're used to in our level of speech but on the level of of outside underlying cells it's very much a gradual process right there's basically you have all these different priorities which are probabilities of being one cell nearer in the simulation and the idea of increased information flow or increased sensitivity with an aspect you're modeling and manipulating or precision as well is that you are trying to get the vibration of energy to be minimized but also specifically to converge on the system where those probabilities actually end up somewhere meaningful right where if you don't want to be in a state where they're constantly kind of fluctuating constantly randomized either you mentioned noise that's one aspect you want to avoid you also want to avoid that they basically do this and that that that's where you write the decision point right like that the cell is basically going towards one state and then there was a wrong one and then you increase the information flow of the other cells and the cancer rescue simulation to bring it away from that but you're not you first of all not actually working on like you know identifying that point and then you're fixing that you're basically just realizing that the the simulation overall basically you basically look at the the probability changes over time so how are they updating and how fast are they updating and then of course in the end what are they updating towards and that gives you a cue if you think from an experimental point of view or even like an experimental or a simulational point of view then that tells okay so um if if that happened too fast if there was misfilm malformation having too fast then you probably want to act and you want to increase sensitivity or something um early on but you don't do it at one time we just set the parameters for the whole simulations but it tells you basically the strength that the sensitivity is with respect to how early and how strongly that misinformation is happening does that answer your question somewhat yeah I think that does one I suppose one question just on that is is it is there like an optimum level in terms of this flow um i it peaks or is it like the more flow you can get the better and it's just literally a limit on how much information flow is possible hmm I think I don't know for a second um so the short answer is that the boring answer is that always will depend on the context I would say um there is no I don't think you can say you know more like the broadsaber like more information flows always better doesn't make sense and I think you already gave the answer to that why that is right because in the end we'll end up too much noise based on the system so that answer has to be with respect of how much information actually is being can be processed what is the timescale of the sensory apparatus um my my experimental work works a lot on um um basically having different uh variation of input signals to my model system which is actually an algae I won't get to that but long story short I actually had once got the question that where I was had these randomized signals um that was defeating it um and then someone asked like well you actually this is actually taking in more information because you're having and I was like well and that like white noise technically carries more right more variation but it's not as formative right so that's more information flow by itself is not better more informative information flow the informative part being is actually does it how quick does it change over time so there's a lot of the important aspects that are there to register and with respect to the also the sensory precision and you know the timings there's certain timescales involved in like how sensory states how fast sensory states lead to updates of internal states all these things together make up basically how fast an agent can react and how much information can process and that's the the optimal level will be dependent on that essentially okay awesome dean or blue before we have anyone go again so I have a question um can you flip to figure four Daniel please yep um just so you had mentioned earlier front hi by the way it's nice to see you I'm really glad we I'm glad you could make it I it's a pleasant surprise um but here in figure four you mentioned that you didn't model proliferation in this model but it seems to me that there's like more like it looks proliferative here so is it just like a very dense cluster at the first time step or do you duplicate it every time step or like how how did you end up with so many more um like beads or whatever you know agents felt at the end then at the beginning that might be an optical um so or communication problem in my part so the the the actual cells are eight in each of those images all of them but what you're seeing is the trace of it right so only the ones that have like the the the the strongly colored dot and then like the little star around it at the end but if you look at the last frames on there those are actually the the cells and what you see in the back of basically like time lapse kind of like snip snip snip shots they're not actually cells in there it would be actually interesting I know you thought of that that would actually be a fairly simple uh introduction that each of those point steps you just make that a new cell that would be pretty cool actually um I I know I think basically Carl and one of his when he initially published that I think he had way at the end of the simulations he basically just added some new cells but he never went further with that and I think it's the same problem that I mentioned that when you have it initially and like in the first time frame at t equals one that if you have to make cells close together it's um you have to kind of set the sense the precision lower because otherwise they're too close together I mean it works somewhat but like it's it's very susceptible to to perturbation in that that stage so in order if you basically want to reintroduce proliferation either like you said or like like you gave me the idea to do in between or you do at the end you would have to have more that which of course what cells do right if you have new cells formation that are not going to be the same kind of sense of rest that then a mature or form cell is um but that was basically I think just for the simulation what was making a little bit complicated but um it is I'm not right now I could be working on that simulation anymore if I did um that would be great yeah maybe I'll come back to that no that would be way cool and if it needs to be like less sensitive at the beginning like you just double the earth or yeah you just double the sensitivity at each um time step and then that way like you know you would get that enhanced sensitivity over time and it's also very interesting from the point of view of fact also and there are some really cool right right now also done by I recently read that Wolfram the guy behind Wolfram Alpha he said sign this whole new physics approach where they basically just have a simple set of rules and then they just use fractal kind of multiplications and on each each time that that goes on and they create all this kind of physics laws um which I haven't yet gotten deep into that can't really say anything about how how good and useful it is but it was fact that we all know is is a very kind of informative some mechanism that's used in biology extensively and that's something you could use here as well where basically at each step where you introduce a new cell you kind of just have the same rule set with the initial and you run like a small simulation in that simulation essentially and with the same simple set of rules you could probably make up a whole much more complex so it probably looked nothing like this I wouldn't expect this to then afterwards still have like the same shape but cool stuff cool ideas too bad not to have templates right now but one day well just kind of unpack that because it's a great suggestion and thanks for kind of giving your take on two friends it's really cool to hear like when the cells divide you may need a first symmetry break or to introduce a symmetry break you get some gradient it can be gravity it could be a nutrient gradient it can be the entry point of the sperm that triggers like a calcium wave but it's like asymmetries can give rise to asymmetries and then interacting you get two morphogens and then now it's high high and low low and then the alternations influence gene regulation and that's like very complex but um adjacency effects are really important and that's how we get the self-organization like of the insect eye or of tissues because they don't have to do what this challenge is which is sort of like getting formations but spaced out this is like the bird flock morphology which is also so cool because it does apply to other systems too which I'm sure will be exploring more but like bodies fill out into a morphology not just dissolve this is yeah cool though I mean of course we really learned a lot let's return to least action just and feel for just to leave anytime you'd like but on least action where does it come into play or what is it doing here yes so it's it's there was more of a background section so Giorgi Gorgiev was also on that on that paper was also my um um excellent Bertie supervisor we kind of talked to him because this is not nothing that I did in there was like novel um I didn't like it was not with the way we framed it but I didn't make it I didn't make new math for that but the idea was that is that what Giorgi was doing all these action principles you know the classical action principles apply to other questions and he was like well you know it's really um I think you're missing like a communication um aspect of this where you're not coming coming clearly how this fits into these action principles um and you're not coming clearly of of what like where diverge essentially and um Ka actually did write a paper on to where the he spingles it in there but I wanted to make it more explicit essentially that that the variation variation principle is at least action principles by by definition essentially but what I was trying to show here is um it's simply kind of like the whole idea of this paper from from the background was that we could anyone could start from this that has more of a physics background and then can work themselves to through active interference to biology and it's something that's very familiar with white and and the idea is actually where does this um what is the interesting take on it from the from the variational free energy and that's what I think in the so you start with a general definition what is an extrema and then you kind of see later on that the the whole definition of the um of the club of clever diverging this innovation of energy very much corresponds to that and it's also um just interesting to see basically where where it takes from so that that's how that answers your question and that's basically the point of that was more not not that least action that I used that particular equation I didn't use the classical the typical equations that Carl uses for his variation for energy there's more motivation and I hope that goes to show essentially where this all flows from like how how do you get from from you know from from from from the opponent function in the beginning possibility how all the function function how do you get from that to a variation of the energy to really make that um kind of integration to physics title which I know I failed at because I didn't that there were the comments to that paper that were published with it and if you read that one or two of them were like we still don't see how the integration is doing and then my comment to that was then I think I called that that mini paper response paper like a deeper integration to physics so I know I didn't succeed in that I apologize but that was the goal okay awesome Dean would you like to go anywhere ask anything I don't even know I know I'm just going to sit back and then because well well for me you've taken a whole bunch of stuff around what a form might turn into based on what signals it's capable of acting on is that that is that a gross oak besides being a gross over simplification can we start there does that make sense yeah so like yes yes that makes sense basically what what I always like like talking about like what you put in what you get out right so what we put into the model is it's an encoding of signaling responses it's like a classical like if you see signaling a b but not much of c and d then you think you sell type a whatever that means head type and then you adjust your own signaling expression so each cell basically has four signaling molecules that they can exchange they can express they can sense and based on different combinations of those they think they're one or the other and and that is semi place encoded so they have basically from that map that's what the target morphology is is meaning so what we put in basically is is that encoding the generative model um general process how they update that the the take on diffusion constants and all that stuff that basically holds information and spreads to each other and then what we get out essentially is is what kind of shapes like you said can can emerge from that um of course initially right the initial model was basically you don't interested in how many shapes because you're interested can't do the shape that you put into it so the point of this paper was is keeping that same target the same kind of like I um each cell has these options that it can be based on what it's sensing and then it's on secreting but keeping with that can you get different shapes just by messing up the way that the information flows in outside of the cell and and it's being processed around the things that's again this that the inspiration from that comes from the lab that I'm working in where we do a lot of this where we don't don't actually mess up the genetic coding so we don't mess up the genetic model in inactive inference jumps but we mess up how how basically the the state space looks like how the information at each point in time is updated and that's that's the idea of this what does it look like to engineer or design a different target morphology how did you fit the Bayesian model or fit the flow descriptor whatever it was in its representation towards the represent towards the um anatomy that's describing figure three yes so um the the the outcomes that we get like the experimental outcomes here we didn't like design the type of logic like I said we keep those the same we just basically had some intuition that some of them try not to be true of how we could change it based on just the asymmetry that we put in an information flow um how would we design this one um mainly so this is on the absolute level and again this model itself is not that's already called this before I joined it so that's not I didn't this didn't do as new but how do you approach this so basically the idea is um this the idea of is fairly biological the idea is that you have joint encodings where certain transcription factors basically right will based on whatever you sense will then basically activate gene expression in the cells that will then allow to differentiate so cells most of morphogenesis essentially starts off from um from something I mentioned initially you have some asymmetry in in signaling that there's this thought that basically even when you already have the the the unfair less egg in an animal system there's some asymmetry where that's being stored actively by by the whole organism of the mother essentially and based from that asymmetry then you have a gradient of already of difference and then so basically some cells will sense more of signaling type A and less of B and then from that they will make they will be um so certain things will be expressed that will then um change its cell fate to a sub to sub cell fate and that then and then you have more cell fates and the same thing happens there you have now more symmetry but starting off in the same one so you again will have sent cells that will sense more and less of some signaling signaling signaling ligands that will basically then turn out to be incorporated through transcription factors to um differential genetic expression um to differentiation so what this is essentially what this is doing right you you're you're coding with what the codes here are essentially those different uh relationships between transcription factor between signaling receptors transcription factors downstream to a genetic code where then basically we don't do perversion again but this is the the first step there's basically one snippet of top mythology one aspect where you have at that point in in this hypothetical morphogenesis experiments you have four different signaling cell types four different signaling ligands um that that are being that being expressed that can be expressed by the cells and based on where you are in the cluster you will see more or less of it and based on that information you will then yourself as a cell start to express more or less of course you basically start differentiating um so then you will start expressing more or less of that and at some point of that you will reach a certain threshold where they actually will fully have gone undergone the differentiation process and have a new cell fate and that's what these different um head tail and stuff so of course like you know there's no animal system I know after it has a you know eight cells and four cell types it makes a whole body that would be pretty crazy although there are some interesting um there's some interesting model owners and systems where they are um the these um the celigans these worms are one example where they have a complete map from from from cell to cell from basically a gene they know exactly which cell basically turns out um to be what cell at the end so they can go from the embryo and they can tell exactly this cell is going to be at that at the end of time um this is the minority of examples most example most model orders I know of are a lot more messier than that and that's what this um trying to basically answer given that you have a lot of messier that but cells can be different things and then if perturbations happen like they happen all the time in environment and often we do see then you know misinformed embryos now that we start in the lab but other labs do it too we see it in nature all the time where you get completely different um uh outcomes in the morphology even though not all of these occur because of some genetic mutation there's not always a case sometimes it's just basically some perturbation in the environment and the same genetic code with some little differences that which here is basically the the difference is here between one simulation and different if you did it that way we didn't but if you did at least not in this paper if you had different minimization initially of the prior beliefs you can also get different outcomes but even for one minimization of of the initial beliefs you will get certain outcomes for those cell types based on um on the coding but also on what the environment does that's the kind of the the goal is do you have cell types you have you want to see what how can you change the the line of cell type differentiation based on the interplay not just like saying you do that and you tell the cell to do this but based on how they interact with each other i'll answer your question it makes me think of waddington's landscape the epigenetic landscape which i'll copy and how would you say it relates to epigenetics or the epigenetic authority if i have my shirt here i have a shirt with it but i don't see it right now yeah i love the what it's not in it yeah it's exactly um right so that that was a the big i love you big enough i'll put it on to a slide yeah but basically the right the whole the whole revolution was initially when we started understanding genetic information that it was thought oh well that's that's the end of it now we know we're done here like we know basically that there's some certain signals that we saw us and then you will get certain genes but the thing rolls down um but basically just from that it's all all straightforward what was new then later on was the epigenetic part of it that was basically actually modifications happening not on the like you did not with other mutations of the genome but which genes were more or less accessible for expression which you know again so basically you kept the coding the same but based on modifications it's a lot of system modifications right the the way that the chromosomes by the gene is wrapped around itself in chromosomes the more or less tightly they are called to each other the more or less accessible two transcription factors um and that started basically happening later on and now we think a lot more about this and we even see there's even that's even heritable um and it's it's it's also interactive with a lot more the physical forces of it so you know perturbations in the in the mechanical or bilateral environment will lead to to a lot faster epigenetic transformations of course a lot faster it's also important to mention that there's a lot faster process it's faster to basically modify some of the proteins that are make up the histones and then make up the accessibility of the genetic code and to start off going some evolution experiment where you start to mutating and cross over until you get the right combination so it's a different layer of information flow that is interactive which negative level so what this paper kind of interacts a little bit or you know intersects with it is the aspect of of health in that initial kind of formation where you have the same genetic code within basically one lifetime how do certain modifications and and sensory precision which you can think of right I didn't make that explicit claim I hate to make claims that I did not actually then afterwards answering on some relevant biological mechanism but I could think of easily that way that used this these modifications of the the the response function that the sense the sensitivity of the of the double tail experiments and the miss the reduced sensitivity and the cancer initiation those can easily be thought of as as you know short-term epigenetic modifications and we do see that a lot this happens especially in stress environments and embryos and stressful environments and embryo and genesis where where there's very quick changes that can lead to drastic outcomes and it's thought of as a as a quick response mechanism to two basic perturbations and and it's involved of course in cancer as well that's that's well known so that's kind of how I would think about that just the landscape here is can be thought of as the variation of energy landscape in the sense of like what different outcomes you can get how how quickly the update you're probably based on the changes in the environment awesome thank you Dean when I was reading this I was thinking of a company that I'm invested in that has taken stem cell derived islets because people have diabetes and they have created an environment to protect against the perturbation so instead of injecting those pancreatic cells directly into the portal vein they try to they have or working on trying to sort of create the environment in which the islets survive so so essentially it's again rather than looking from the bottom up it's it's sort of creating that safe environment in which these islets can then carry out their function and so that's what I was kind of thinking about throughout when I was reading your paper now I'm not sure if that again whether I was thinking of the correct analogy but but yeah in terms of what the pancreatic islets can do depends on their survivability so if you don't have the right kind of environment for that they tend to die or worse they don't die they become cancerous so it's kind of interesting that you're you're looking at it here and then I know of some sort of practical application but the research is really trying to build these environments so that they can actually protect against the perturbations yeah no that's exactly the long line so I'm thinking as well I do but I had a quick two more minutes um if I didn't know how to push and go um really appreciated you joining if it works for you to drop in or out at any point in the dot two please feel welcome and any other time you or anyone else are welcome to participate but thanks a lot for joining you're welcome my pleasure see you later okay bye bye thank you excellent that was great fantastic great yeah fun times um okay so here's here's one thought on the relationship with foraging kind of picking up on what we were just looking at so he described some of the manipulations as kind of like interventions that were epigenetic so it wasn't the addition of like another changing the model structure like changing there from four to five transcription factors it was just changing their state it's like picking one cell up and moving it would be changing its position but then there's other levers of intervention like changing its sensitivity or other aspects of its cognitive or its generative model and the whole point with science and active inference though see anticipating brain is not a scientist by brunberg paper so it's not exactly the same thing but it's about informed or directed experimentation and that's true from the epistemic foraging with the isocating to the scientific experiment um when we make targeted interventions into systems based upon a good understanding of them we can learn about the cognitive model that's underlying it or the as if cognitive model so there's a lot of interventions that would not be informative about morphogenesis like jokingly you know hitting the petri dish with a hammer it's it's an intervention that does kill cells it kills cancer but of course isn't that the joke right like there needs to be more to it in the experimental design and in the the measurement of the outcomes so that we're actually learning something and so to make increasingly nuanced designs and learn more and more and apply better and better that's where having like this kind of formal modeling really matters so this is kind of taking just saying like well changes in receptors might be a precursor or a biomarker or an early mechanistic warning of x disease you'll read that a hundred thousand times in the molecular biology literature and this is picking up there and putting it into a modeling architecture where we actually can talk about changes in signaling and position in an integrated way so that it might be able to actually model real settings just like people use the epigenetic landscape heristically but also increasingly quantitatively to model selfie decisions but just to kind of close that again it's the interventions that we as the investigator know based upon our understanding of like the natural history of the system that give us updates on our cognitive model of the system so like an ant foraging example is they'll take a desert ant that forages alone the decimate goes out without following like just simply a pheromone trail and pick the ant up when it's like let's just say 30 meters away from the nest or 10 20 meters away and then move it to like 90 degrees rotated the same distance so one extreme case would be it continues from that location 100 as if it were walking back from the north let's just say that would be a pure direction step integrator model there might be an angle it takes between that and for example heading towards the nest that might reflect the usage of other sensory or cognitive features like detecting the polarization of light or scene memory or other ways that it might be able to adjust and turn its vector to some other direction so that's the kind of experiment where we can start to learn like how much are the the nest mates in their foraging adjusting between their own onboard memory versus these external cues that might be picked up like in one instance by any nest mate in that location so there's some cells we're moving it into a different tissue the first thing it would do is die because it's just like too acidic or just like not a good environment for it but there are certain interventions that do seem to happen that lead to at the very least we can say states we don't prefer or just more directly like states that are abnormal or unhealthy though it really does come back all to blue's question about death in aging and how is that related and what is this living phase between the developmental part and the death part even though there's such a complex continuum between them Stephen can I just ask a question on what you're just saying that Daniel um do you feel that or do you think that the the level of what's happening with epigenetics is fairly flat and once these other types of changes these types of changes happen that the adjacencies permeate out and can the cells are kind of finding their place um and is that a bit different to the nestedness you know that we're talking about the nestedness is because um like for instance a tissue or is that also does it then fall into some other type of um behavior where it's like into this idea of the nested the nested kind of Markov blankets and is it then a case of what was the types of information flows which were conservation at one scale have become noise that makes sense so maybe I'm just curious about if that flies in here um it very well might I mean what it just makes me think about is the eye specification in root flies and and similar in other insects so this is like an example of a developmental genetics pathway at a relatively downstream level being worked out through a lot of work but these cells are pre-committed to being eye precursor cells but then they undergo a second or a future stage of differentiation that results in a compound eye and then it's the modularity of this developmental framing of first eye precursor cell and then to a specified eye compound cell that lets like the insect eye scale in different ways for example in the mammal eye developmentally so over evolutionary time there's different affordances for insect eye evolution versus mammalian cell eye evolution um because they're on different pathways so it's kind of fractal because this eye field precursor it's not the second one that comes out of the embryo so there's multiple like fractal layers and pre-committals and it's not possible to jump from a canyon way at the bottom directly over there's a few cancer cell papers that do a little bit more of like a reversion and around or a perturbation that pushes and then changes to like susceptibility are kind of like changes in the elevation of these hills between so it's a landscape which isn't just simply flat um it has some scale and then the relationship between how jagged it is and how much noise or flow there is is about how navigable that landscape is in that one model blue so I have a question for Daniel and Dean specifically um so when Franz was here he talked about the block matrix operator that he used in equation 39 and you guys well maybe it was Daniel specifically we're talking about some kind of housekeeping term in the dot zero and I couldn't remember or place the housekeeping term that you were referring to and I wondered if it was like the same or similar in a similar or used in a similar way I think it's Connor I'll bring in a housekeeping term slide it was from 32 livestream 32 stochastic chaos Markov blankets and um here is the slide in slide 45 here it was um basically the Helmholtz decomposition is usually discussed in terms of just the gradient the solenoid so it's sort of like directed term and then the isocontour circulation change in elevation ruler on the hill and then around the hill on the isocontour and then this paper 32 um brought in the housekeeping term and then there was like a supplement so we didn't really totally go into it but together um they represent the total flow so it's a slight kind of um frist and variation on the Helmholtz decomposition in this context but as Connor mentioned in 32.3 it also has like sort of antecedents and maybe even similarities with other areas so totally remains to be a little bit unpacked but it's the influence of how movement changes the landscape at least that's a little bit how we were framing it like change on a trampoline where the movement is going to influence it you can't just snapshot the landscape and then calculate what would happen and what the gradient and the solenoid will flow would be at every point because it's going to be the future. That might be related to what Franz was saying because he said he used it to smooth out the landscape because otherwise you get stuck in like a little local pocket when you're talking about like you know something that that's bumpy so it sounds like it could be a similar type of um like housekeeping thing or just like a smoothing smoothingness when you're when you're going into the bowl you don't want to get stuck in a bump down along the way right if you're a marble traveling down like I thought it was a cleanse there's just too many too many jokes to be made um but um wow I think in our last you know 10 ish minutes we should just chill land the plane prepare for dot two it was pretty unexpected and awesome to have the um time that we got to speak and I think that's like the housekeeping term in action which is a process and a protocol and then people who have different perspectives on a situation but the group can always re-adapt to a changing landscape and so even if that's interpreted as structure learning at a higher level from the sort of agent view looking up it's equivalent to adapting to a different situation in some fractal way slightly different or very different using the same affordances and perceptive and cognitive features as it had in the time step before and that's why um this model that looks at the flow of the autonomous states so um the states that can actually be controlled in the Markov blanket partition in 30 by defining a specific function that we want to like be fitting well by really focusing on the flow over the blanket and internal states it allows inference on that component perception cognition and action of the given modeled entity to be uncoupled in a certain way from inference on the hidden external state evolution which is fundamentally unknowable but in a slightly different way if that makes any sense and a classic moment blue to suggest the the mitotic elements and like just to bring in the continuity and then the way that that gives rise to symmetry breaking sort of for free um and that the sort and so this cool conversation that we had today and so so I did up as a dot one might I didn't get to mention and I was going to mention it but like you know there's that fractal fmri brain pattern paper that I'm like dying to read and would love to discuss with the authors also but it's interesting just in terms of the the fractal dimension also that Franz was talking about is super cool um um is there anything else that we want to like um write down to think about in the dot two or questions that we'd like to ask Franz if he returns or just things that we want to take moving forward in the dot two already I'm still in dot one yeah I agree it's like it's like you're eating dinner it's like what do you want dinner tomorrow um inverting space and time we didn't really get to um and some of these initial points it was right least action math and morphology and then generalized flow and taking us all the way back Dean so I would like to maybe ask ask Franz about quantum mechanics and what the role if any that that played in um like what is there a quantum perspective that that he was perhaps viewing in from when he was writing this paper I would I would like to ask just that one question because that will encompass inverting space and time and it will encompass what we're talking about with the probability current and the the probability mass conservation so like that all is like bridging into the quantum where we'll go in the 40 so I think I would like to ask about that one thing Dean what are you like looking forward to what would you want to add here at the end no I'm just going to go back to the drawing board and basically look look a little more closely at some of those math operations because I don't tend to give those as much attention as I need to I think it's really interesting now that we've had a chance to sort of bring more people into the conversation I think think there's three parts to I think we can agree there are three things that we can really center on now there's the the formally form forming which is what figure four and five do but as soon as you go to figure five what's interesting is is that the the attention seems to be on the shape that the cells eventually take and yet in the in the sort of in the blackness around each one of those moment captures there's there's signaling and there's changing right there's so there's all the part we've talked about today which is the whole flowing piece and what where is it going and how is it channeling and how is it narrowing or how is it spreading and how is it dissipating and all that stuff but in those even in those diagrams it's hard to figure out where the signal is even when they introduce arrows they introduce the arrow pointing at the form not the necessarily the signaling and the changing and that's where the math really comes in I think because without the math you you lose sight of those other two present factors that are you know all moving and mixing and and turning into something so I hope France comes back because man he's got a heck of a lot more you've got a lot more to tell us about really because this is such a dense paper like I think I mentioned to you Daniel before before the point zero this could be a 13 week course this one paper if you really wanted to I don't know how many hours he spent on this but he already admitted that going to Carl's lab and and and from the lab he's at there's way more multiples of 10,000 hours of work that were done before he started typing things out so yeah awesome with that made me like think about is this idea like multiple kinds of invisibility or overlay or unpacking like projection up into a bigger dimension of interpretation like you pointed out how the like the focus was on morphology which is to say the final realize like this picture is of the morphology of the animal it's not of the signaling density but this could be like presented as just a gradient of transcription factor a or it could be presented as a gradient of vitamin a or like it's kind of like carillion photography like looking at the morphology with the energy field I don't know if that's exactly what it is but that sort of idea with the field based perspective but then it gets hard to show many overlaying fields because like what do you do 20 overlaying colors but we we can't actually see that and so that's just kind of interesting like a question about visualization of higher dimensional models with a lot of overlays and then that's like one kind of invisibility which is something that's modeled but just not graphically shown easily like the density of 50 pheromones or 50 transcription factors or vitamins but then also but but it does exist and it's like modeled as an actual chemical component of the biological system like the generative process and then there's this generative model with the mathematical derivatives the generalized coordinates of motion of cellular position and those are also like invisible in a different way because they're a modeling tool and the derivatives are not anywhere in reality like what is where's the seventh derivative of the baseball's movement just relative to what where is it hiding in the current moment this is just a purely tool driven way of thinking about the current moment not just in terms of its composition and like anatomy and position and the lowest level of the observable model but like these higher levels which are real yet not existing in the moment structurally real in the structural realism sense i'll take your word for it i've been wanting to say that for a long time i have to now go back and figure it out so um but what a discussion i hope like for those who stick around to the end here that they enjoyed this this wasn't just um total uh total morpho-pastrophe all right well fun times and awesome to have all the good discussions and appreciate everybody who was here blue and steven and dean so see you all next week or any other time peace thank you thank you bye thank you