 Hello and welcome everybody. Welcome to the Active Inference Lab. Today it's October 1st, 5th, sorry, greeting Shannon. And 2021, we're going to be discussing this awesome paper, How to Count Biological Mines, Symbiosis, the Free Energy Principle, and Reciprocal Multiscale Integration. Welcome to the Active Inference Lab everyone. We're a participatory online lab that is communicating, learning, and practicing applied active inference. You can find us at the links on this slide. This is recorded in an archived livestream, so please provide feedback so that we can improve on our work. All backgrounds and perspectives are welcome here. And we'll be following good video etiquette for livestreams, like muting if there's noise in the background and raising our hand so we can all speak. At this short link, you'll find the calendar of past and future livestreams. The first tab is the regular Tuesday livestream with two weeks per paper with the dot one and the dot two. But then we have a few other series of streams related to different topics. So go to that link to check out what's upcoming and see if you want to participate in one of these streams. Today in number 30.1, the goal is to learn and discuss this awesome paper on counting biological minds and we're really honored and appreciative that Matt Sims is here to discuss it with us. So we're just going to introduce ourselves and walk through any of the figures, any of the formalisms. We also have a few other ideas and questions that lab participants have raised. So this should be a great discussion. Thanks everyone for joining. We'll just go around and say hi and introduce ourselves, either just an introduction or feel free to say something that you were excited about with reading the paper and then we'll all non-authors will go and then we'll let Matt just provide a little context here. So I'm Daniel. I'm a postdoc researcher in California and I think I'm looking forward to drawing out some of these threads about collective behavior and the timeless questions about collective behavior with some of the new approaches that are discussed in this paper. So I'll pass it to Shannon. Hi there, nice to see everyone again. I'm Shannon. I'm a PhD candidate at the University of California in Merced. I've blurred my background but there is a dog rolling around so hopefully it's not too distracting. I will pass it to Steven. Hello, I'm Steven Sillett. I'm based in Toronto. I'm doing a practice-based PhD through the UK and around psychological perspectives on practices, particularly around social topographies and spatial meaning making in participatory theatre and the arts. I'm really interested in this paper. I like the folk psychology questions that it brings up and I like this idea of a folk biology, which I found really interesting and the way that we have assumptions around that. It sort of ties in maybe with some of the folk quantum that came up in some of our previous sessions about what people think that is. So I'm really excited to see this folk idea being extended because I think that could be really useful. So thanks. Thanks, Steven. And to Dave. Yeah, Dave Douglas. I'm retired from programming machine translation software. Now I get to get really busy working on history of cybernetic learning theory project for a professor at Carnegie Mellon and all this good stuff here. Thanks, Dave. So Matt, thanks again for joining. We'd be really happy to hear some context and we'll just take it from there and anyone can raise their hand or type in the live chat if they have a question. Fantastic. Thanks so much for having me. So the context in this paper, I suppose, is like I wrote this paper during my PhD over at the University of Edinburgh. I was supervised by Dave Ward, who's an auto-product and activist. Andy Clark, who I'm sure all of you know, and Julian Keverstein, who's up over at the University of Amsterdam. And the focus of this paper was pretty much my obsession on the notion of collective intelligence and what collective intelligence might be and how cognition has developed, say, the evolution of cognition with the background of taking into account the fact that cognition is in its most basic form something that's going to be collective in the first place. So it does pull at there is this kind of immediate tension with how we look at cognizers and how we look at cognition in the kind of folk field with typical cognitive scientific research, I suppose, or traditional, should I say. And the point of this paper was to look at the kind of evolution of cognition, if you will, through a particular lens using some of the conceptual tools that were provided by Aeroset Morathe and John Minnan-Smith and see exactly how this can be formalized and looked at through the perspectives of the Friedinger Principle with the process theory active inference. So that's pretty much it. Very concise summary. Major transitions, which is the Zothmery and Maynard-Smith line. Big qualitative changes over evolutionary history and then just like you kind of brought to at the end, how do we land that plane with maybe quantitative differences or structural differences in a microprocess theory maybe active inference under the Friedinger Principle? So Shannon or Dave or Stephen, feel free to raise your hand. Otherwise, maybe a first question that came to mind for me was like, how do we research or understand these folk beliefs? Scientists are folk too. And is this like sort of the mythical, average citizen? What are these beliefs? And how do they come to be that way? Why are they the way they are? These folk beliefs about biology because that sets up a lot of the discussion and what's relevant of the contributions of the paper. Absolutely. It's a very good question. Folk beliefs we have kind of obviously folk psychological beliefs that take the form of actual folk psychological notions that take the form of say beliefs and desires. And these things I think give us a very good manner in which to actually approach the notion of folk biology also with beliefs and desires being particular type of constructs which play an explanatory role, a psychological explanatory role. And for example, explaining or predicting behavior to us on a certain level. And I think on a kind of first blush level so to speak. And I think folk psychology and folk biology are kind of bedfolk in the sense that folk biology is going to be based upon what we initially see via like in our everyday experience of the world for example is what we initially see. We're going to base the notion of an individual for example on the ideas of things that are going to be easily individuated and like biological systems that are going to be specifically reproducers. And this is where you come into the notion of kind of the kind of Darwinian individual as it was called by Peter Gregory Smith where individuals are going to be demarcated by their ability to be categorized under three different types that combined or joined three different kind of categories which are going to be they have to involve variation heritability and some type of differential fitness which is just natural selection, right? And this type of individual has been kind of we've invested into this notion of individual and the biological sciences up into a certain point up into a certain point because it helps us answer helps us approach certain questions and make certain I suppose make certain inferences about systems over the long run but it falls short and I think this is again one of the grounding points of the paper it falls short in explaining certain types of associations and these associations namely are going to be symbiotic associations and this is a quite important I'd say transition as we spoke about before for the simple fact that without symbiotic associations we don't really have a manner I think of explaining multicellularity the transition from single cellularity to multicellularity now we've had various theories and I think one of the theorists that started this was Lynn Margulis over at the University of Madison, Wisconsin and her research was fundamental in which she suggested that for example you have two two prokaryotes one envelop the other through some type of phagocytosis and they create some type of relationship over time this is thought as ludicrous and it did really push against this notion of Darwinian individual however over time we've come to accept something like this and I think this is like the idea that you have some type of bacteria that was enveloped by an archaea and in order to explain in order to actually think about what we have in terms of this particular type of unit we really do need the notion of moving beyond moving past Darwinian individuals and two ways of doing that obviously I think that the paper tries to point out this and these are not the only two ways but one way of doing that is through the metabolic individual which is something that Peter Gottfried Smith himself tips his head off to which is just to say that metabolic individuals are individuals which they can be delineated by their metabolic machinery and having autonomous metabolic machinery and other individuals can play a part of that machinery and thus you have two systems that make metabolic couples to one another making one system and that's one way to go about it the other way to go about it is was introduced by Thomas Pardoux and this is kind of the immunological version of an individual or a physiological individual and again these are the only two types of individuals or two ways to delineate individuals outside of the kind of more traditional folk biological notion of individual but maybe that's something that we can talk about perhaps there are other ways that might be interesting. Great points and so interesting how the map replaced the territory the map of the sort of Darwinian triad defining an individual in the Darwinian sense and then people project that back out and say reality is explained by this partitioning scheme. Yes. Very cool. Stephen with a raise hand then anyone else in here or in the live chat. Go for it Stephen, thank you. Yeah I'll be interested I think this is really good points about these different ways once you start to have these interactions and these reciprocities across you know then pluralism starts to come into the picture and I'm wondering how you see this moving between fields of practice because I wonder if like people in the biological sciences and maybe those who work with agents would really be getting this right and then people who have a theory about thinking they know what biology is maybe physicists who maybe think they know what everything is then have you know they have a folk biology understanding of biology just like maybe other people have a folk physics understanding of quantum physics so I'll be interested in that and this might even bring up some work of Chris Fields that we had on here as well in terms of other ways of approaching this multicellularity but I'll be interested in your thoughts first about you know that tension even within the sciences that's a really very good point I do think this is before I hit that point directly I do think this is a call for interdisciplinary work it's an absolute call for interdisciplinary work this is something that's I think trying to actually use folk psychological notions or folk biological notions or folk physics notions we will have some success doing these things however once we run up and there are going to be certain constraints with doing that and we'll run up to problems run into problems and these problems I think can only be can only be avoided or actually addressed when using various disciplines together and I think this is one of the things that kind of the free energy principle actually offers as a kind of it's an opportunity to work interdisciplinary on various various phenomena various processes and you're absolutely right in saying that okay well how can for example in biology you might have a typical assumption about what cognition is so we'll have a lot of biologists for example there's a there's a a type of we were speaking about before this there was a notion of fossil cognition that has been pushed within the last 15 years and a lot of various research programs scientific research programs whether it's going to be plant biology or microbiology has kind of fed into this which is it's a lovely program and since it's questioning how we actually look at cognition and these biologists that are kind of taking these cognitive terms and using these cognitive terms one of the things that I think that pushes against this is going to be kind of a traditional psychology that refuses to with lack of lack of better words I suppose refuses to actually engage in some type of interdisciplinary discussion so it's happy a psychology that's happy using for example more folk traditional and folk psychological notions and with these kind of limitations I think we're going to actually be stuck we're really going to be stuck so I guess to summarize the point is that it is a call for disciplinarity it is a call for kind of questioning or putting aside the fact that we can give satisfactory explanations but whether or not our satisfactory explanations are actually the explanations or if there is just one explanation that's the question that's the question and I think the whole notion of the whole idea about biological individuality is that you can be a once you take into account various angles from various disciplines it's fine to be a pluralist it's fine to be a pluralist thanks I was thinking of just a pluralism of what the word folk even means some people might think oh folk chemistry that's being wrong that's the middle school level of understanding chemistry and then the experts they know about chemistry that's folk versus expert but also it seems like folk is being used in a way appeal to first principles like appealing to directly intuitive observable processes and what is active inference if not a first principles approach so there's a way to bring in some of the technical expertise the computational backbone but also fully retain the conversational and the casual approaches that appeal and spread to people in their understandings of different sciences yes so Dave with a raised hand anyone else with a then yes go for it Dave there's been so much good work done by people outside their specialties I guess Erwin Schroedinger may be the most well known of in that but Ben Mecca the Swedish neurophysiologist wrote a lovely little paper from about 1999 I don't think it was ever published about the developments of the mobile eyes in one extremity of the early vertebrates has forced neurological developments that have resulted in consciousness simply because having mobile eyes creates such an enormous burden on sensation that you can't just use a simple nervous system otherwise it's just overwhelming have you looked at any of Mecca Mecca's work okay I'm going to have to send you that please do please do and also Mark Solms who is a direct collaborator with Professor Friston just wrote a popular account of active inference from the viewpoint of his own work with Mecca and Yock Punkssepp and Antonio Demasio on the effective foundations of consciousness and especially his own thesis that in Freudian terms it is the id that's conscious and the ego the cortical and the neocortical processes only gain to become conscious when it really has to because it's so expensive metabolically and in terms of distraction and he's one of the few people who was actually aggressive enough in trying to go out and market his dissident from the received wisdom and he's also did an amazing update of the original project for scientific psychology in terms of Fristonian active inference so first word to last word of that 1895 Freud essay is in 2017 terminology but still in exactly Freud's order of presentation wow yes please do send me that paper thank you awesome so Stephen with a raised hand then Shannon or blue welcome and anyone else in the live chat of course yeah just also this idea of the folk psychology and I agree with your point about interdisciplinarity and I think it's really interesting is how the expert's perceptions of the world then gets translated into a folk publication which pretty much everyone then starts to incorporate and one that I found really interesting with active inferences I was looking back at older ideas of vision before they discovered the retina and the image on the retina and they actually thought of the eye as projecting out and sort of feeling the world if you go back like 500 years and then we were told by the experts how it is with the retina and the image projected like a camera so then folk psychology moved to that or folk perceptions moved to that because the experts said that but actually it turns out that the original folk psychology or the original perceptions is actually more accurate now that we're looking at saccades and all of that so this is an interesting what is folk and what's expert and where does it come from absolutely it's a it's a both and because it's not like the retina emits the photons that it then senses back it's not doing echolocation but there is a tale of two densities there's a generative model and a recognition model and there's a process that's a conversation which we're going to get to with the users and resources distinction so it's like there is that bidirectional conversation and then if this direction is folk and then that direction is expert in certain social contexts which direction is going to be prioritized as an explanation and the other direction is going to be seen as a countercurrent literally going against the stream mainstream so pluralism allow us to recognize both directions in that case yes even that's a good point Daniel bidirectional and one question I wonder sometimes I often work I'm sort of connected to I'm seeing in the world of coaching and psychology as well in a ways and the arts where these things like Bayesian brains sort of did come through there about ten years ago in a very but it's it's normally come from research which has been interpreted into a book by someone from that field which then gets picked which is still quite technical which interpreted by someone into something else and by the time it's rolled down into some other areas these notions can be quite quite watered down you know so yeah this question about how to you know have that conversation when some of the stuff is quite cutting edge from even people in the fields themselves and how to get it used in the real world is a big one absolutely very true so Dave with the raised hand and then anyone else I think it's in the pandas thumb where there's a rather exasperated essay about how you'll have an innovator and if he's really lucky he'll have a few people that work with him that fill in the gap but typically and he doesn't even have one level of competence it just immediately goes into the folk press and becomes just a complete mush with all the almost all the original valid insights just buried under folk foolishness possible so I guess just we paused on this slide because this concept of individuality is a central one here and we mentioned a few of these different ways of looking at individuality and kind of contrasted pluralism the idea that multiple perspectives can coexist and that that's okay versus I guess absolutism or monism that one of these or some other unlisted criterion of individuality is like the broader category or the right one what is on the table with definitions of individuality like why would it matter in a technical capacity or for just all people what is being discussed or what matters depending on how we approach this question of individuality to Matt first and then to I think there are a few things that we can say why it matters one thing is that before we engage in some type of investigation about a particular type of phenomenon process it's quite good to know how to individuate what those process what are involved in those processes or what that process belongs to the relations between elements in a network pretty much and in order to understand those elements or to locate those elements we have to have some type of criterion in order to actually even think about the kind of larger structure of a network and the relationships between them and I think this is biological individuals they actually play this role and I'm reminded by something that Gregory Bateson said right that in order to actually understand what the mind is we have to actually understand how to delineate things with minds but those delineations are going to be completely dependent upon the processes that we're studying those delineations are going to actually involve the processes that we're like so it's going to be a process dependent or a phenomenon of dependent individuation that doesn't mean that these kind of individuals are going to actually they might contrast or conflict with one another in any type of logical manner but it's just putting in place a type of pragmatism in terms of where we delineate things and I think biological individuals provide us with a manner of talking about for example specific phenomena now if we're talking about replication or we're talking about for example competition with respect to something like replication then we might want to actually focus on Darwinian individuals however if we're talking about the evolution of multi-sailarity or these major transitions in evolution we might want to think about individuals in a different way so that's that's what I think Thank you Matt, Lou and then Shannon So hi Matt it's nice to meet you I love this paper found it super inspiring and you know I think that the concept of an individual is closely related to maybe the concept of autonomy and also the concept of emergence and so as and especially I mean it's also of course related to the concept of a cognizer but the cognizer aside so I've heard a lot of definitions thrown around and an individual I liked in the paper how you described emergence with the circular causation because that like echoes a lot of I mean it's the information theory of individuality paper by Krakauer and that group and it just echoes a lot of what's required for an individual so I just am thinking about autonomy and it's one of the things that I mean it depends how you define it and in the artificial intelligence world there's no such thing as an autonomous system I mean really we program in whether we want it to learn whatever it wants to do but I mean they just don't little baby robots aren't just set free with no goal there's always a self directed or a human directed goal kind of programmed into this system or maybe I'm wrong maybe there's examples where it's we just create robots for fun or artificial intelligence just to create it but like an artificial intelligence system doesn't really have that same autonomy or at least I don't think about it and so I just am kind of curious as to you know in the paper you kind of hinted at or I don't know if you said explicitly I was kind of flipping through looking for a quote but you know you hinted at like maybe a cell doesn't also doesn't have that same kind of autonomy and like so I I want to challenge that so in terms of active inference and the FEP like I think about it always as a scale free architecture and a continuum and so why is the cell not autonomous but an organism is or maybe an organ is is an organ autonomous or where do you draw the line and why are you making a line in the sand around a system like that. Fantastic question and I agree with you I agree with you that's one of the things in the paper that I think that I pushed I did draw a line I didn't try to draw a line that was too deep in the sand and I think in one of the footnotes I do say something like okay well I'm not suggesting that a mitochondria is going to be something that's non-autonomous however autonomy is something that also varies in degree and I think a certain level there's a certain level of autonomy that it's going to be provide a different degree of say degrees of freedom with which some system can interact with its environment in ways and that environment provides perturbations enough perturbations enough kind of enough surprise such that whatever that system is has various strategies and has to have various strategies behavioral strategies or we can talk about this in terms of active inference in order to actually adapt to that particular environment now I'm not this gets us into this notion of adaptive active inference versus kind of like the non-adaptive active inference and I think the way that Michael and Julian have thought about if I'm not mistaken I have thought about adaptive active inference is that it is based on a certain level a certain degree of autonomy and with respect to cells I do agree with you cells do have a certain level of autonomy however cells are again dependent upon other cells in a way that other systems other more autonomous systems aren't dependent now you'll say of course we're dependent upon our environments we have to have oxygen there's certain constraints in our environments that we have to we really have to kind of like our environments have to be a certain way so we are dependent upon our environments but it's distinction I think between being dependent upon other systems and cells for example bodily cells are dependent upon other cells not only plasma but other cells in order to actually maintain themselves and I think there's a difference in degree and I shouldn't again I should impress upon this idea that I'm not saying that cells are not autonomous but most certainly I think there's a difference in degree and that difference in degree does matter cells might even be cognitive to a certain degree and this is something that someone like Michael Levin suggests and I agree with him but I suppose the notion of adaptivity there's a degree of adaptivity that comes along with the emergence of a certain amount of degrees of freedom so cells are dependent on other cells but aren't humans dependent on other humans? I mean we're born as an infant and completely rely upon our parents to provide everything for us for many many many years and even as adults I mean still we don't exist in a bubble I mean perhaps we can go off to a cave and exist in isolation but I think humans are dependent on other humans as well so maybe it's a degree of dependence or not or I mean I really feel like there's a continuum here even in terms of adaptive active inference and mere active inference I mean and I fall in with Mike Levin in a lot of the ways that I think I think that's right I do think that's right I think there's a distinction to be made also between kind of developmental dependence and homeostatic dependence and I think homeostatic developmental dependence can have an effect on homeostasis most certainly but those are two things that might not actually always overlap and for example sure we as humans absolutely we have a long developmental period in which we're absolutely dependent upon our parents or being reared by some other human or some other animal absolutely and that happens that when we actually when we get older we probably are dependent also absolutely agree with that but there isn't any kind of development with a cell that can be mapped on to dependence or non-dependence it's always dependent thanks Matt, Shannon that actually flows really well into my question blue that I know your paper you specifically say we might want to talk about humans or groups of humans but we're not going to in this paper but I'm going to ask about humans and so when you're talking about this squid in its bioluminescence and they form a temporary unit in which they are together an individual and then it sheds the the bioluminescent cells during the daytime and then they breed again afterwards I'm wondering if there is a time like a critical time period at which units would have to be have this symbiotic relationship with each other or if there's a necessary strength of the coupling or a repetition of how many times the organisms need to meet up and if that would change for this animal compared to systems of humans because there's a certain way in which the humans that you interact with in everyday life or if you're co-living with someone even your metabolic states become co-regulated by your sleeping patterns with each other or stressed and hormones between different humans are affecting each other when you're together at a fairly regular rate and so I wonder if we might define individuality or collectives slightly differently depending on the system we're looking at and the type of coupling that's occurring. I think that's a great question, Chad. I would say yes. I think there has to be some that you know about the empirical literature in terms of entrainment and I think that's a really good point to make in terms of having those types of interactions with humans whether or not for the Bob-tail and Vibrio there's a chance that after the Vibrio actually colonize the light organ they're typically not taken in again once they're expelled they're out and they might be taken in by another squid or juvenile squid but the Vibrio that continue to colonize are the ones that are left over and it might be the case that the ones that continue to colonize for that shorter period of time might become the relationship with the squid might change to a certain degree but I don't know any empirical evidence about this point that suggests that's the case but I think that's a really good question to ask a very good question to ask So Blue and then I'll make a good question I think Dave was first Oh yeah, go for it Dave and then Blue Yeah there's Dr. Alan Shore has done some I think controversial but really evocative work on the formation of the not just the personality but actually the neural functioning of the newborn in the dyadic relation the intimate dialogue with the mother face play and the rhythm of moving in and moving back and regulating each of them regulating each other's expression of interest and each other's degree of involvement and at a really different scale Gerald Adelman's topobiology talks about the cell and the tissue as being just so dependent on these points out we have almost no chromosomes we have almost no genes compared to how complex we are where does the regulation come from it comes from adjacent cells adjacent tissues Thanks Dave, Blue So Shannon kind of she was saying about becoming co-regulated say if you live in an environment with someone like it's a super nice transition to something that I have really been wanting to discuss in the context of this paper which is like okay you become co-regulated with someone that you're roommates with and co-regulated perhaps by the environment but also perhaps by the microbes that live in our environment so there's like that entire microbiome concept that is fundamental and key to humans and I don't know I mean I'm not like a huge studier of symbiosis but I'm not sure if the microbiome like I know we depend on a healthy microbiome I mean for healthy brain function and you know just regulation of our own internal like metabolism and so forth and so is that a symbiosis or not of the Holobiant which is like introduced by Fields and Levin if you've read that paper the skill free biology paper I mean it is like to what and that they talk about physiological individuality there as well so is that symbiosis is it not is symbiosis also a spectrum or is it like a you know a discrete category of things and I also think about like symbiogenic viruses which like you know the very early viruses that it's like you know effectively like instantiating speciation right like the viruses insert into the genome and you know developing species and so forth and and it's beneficial to the virus and the host right so so anyway I just wonder like what are your thoughts on that or or if you have any insight there absolutely that's again a wonderful question but with respect to kind of symbiosis I think there's been some work on that and some very I think convincing work done by Queller and Strassman who suggest a particular type of XY locating symbiosis on kind of an axis pretty much so you have cooperation I believe on one part of the axis I have to look it up again actually I should look that up but it's a you have varying degrees of symbiosis and again symbiosis in the particular paper that we're discussing now I've concentrated on mutualisms but you have various forms of symbiosis that I think you've actually pointed out previously that can be kind of modeled quite well with kind of game theoretic and this is going to be it's going to rain from parasitism to communalism to kind of mutualisms and I think like these are all again degrees and these degrees can change there have been quite a many theories I suppose and I think Queller and Strassman have also suggested this that depending upon the condition of a system with respect to another system say they're symbiots a parasite can turn into a mutual parasitic relationship can turn into a mutualism and vice versa so these things move these types of things move and if you think about this is a dramatic thing that we can even apply to kind of cellular biology when you think about something like the origin of cancer when you have cancerous cells they've actually grown within a host they're part of a body but end up being some type of competitor and I think this is quite interesting to think about this through the lens of symbiosis obviously that's not a symbiosis but we can use symbiosis to kind of think about that but with respect to kind of thinking about host cells and parasites host cells and mutualisms there's like there's many many different areas of the spectrum that one can actually analyze our relationship with our own cells and our relationships with microbiota like Scott Gilbert obviously you know and Scott's work is quite incredible he's the one I believe who introduced the notion of the hollow vion hollow genome excuse me and hollow vion and I think this notion of hollow vion is specifically it includes symbiosis but it includes actually a kind of community of various systems that themselves qualify as one system can qualify as one system if you put it in a different way they create a stable state a steady state themselves and one that's kind of resistant to perturbation and that's the manner in which I think it's a very easy manner in which I can think of the hollow vion as an individual whether or not one wants to call two symbiotic systems that are in a mutualistic relationship a hollow vion that's another question but I think typically it's been applied to more than two systems thank you Matt so for Shannon's questions about the co-regulation and humans and trust it made me think about how one-off games lead to kind of community or population-like game theory whereas repeated games over evolutionary time like in the case of the squid and the vibrio or in the case of the eusocial insects that leads to organismal-like dynamics and then what you just described and what Levin also got at in the paper is the breakdown of for example, organismality into population of cell versus population of cell so it's absolutely a cool framework and then also a really important part on figure 4 is you have the peak of integration and then greeting survival the peak of integration and then that's the y-axis so it's like instead of saying we're going to say which organisms are tall or short out there in the world it's like you do the population level survey and they range from you know 2 to 4 feet and you say okay it's in the 50th percentile it's tall like we're renormalizing our comparison stick to the maximum integration of the squid bacteria so we're not comparing it to some other system we're renormalizing and we're able to have a system-specific understanding here it's relatively more integrated here it's relatively less it's a second question whether the peak of integration of the squid and the vibrio is like the integration of some computer program with another computer program maybe it's apples and oranges maybe we'll have the right framework that allows us to transmute apples and oranges but here you can just relativize to the system itself and that's like very specific and powerful so Stephen with a raised hand and then anyone else yeah I'll give you a chance to get a breath because we're throwing a lot at you you mentioned allostatic and homeostatic forces if I'm right so you've got that difference there between something trying to maintain or trying to move forward and change and I'm interested in how that reflects so for instance I suppose you could have within a set of cells you could have some cells or some organs trying to maintain homeostasis and others say even maybe a parasite trying to allostatically enter that world and create a niche for itself and I think that that question about that balance between the two is a useful divide or a useful way to think about the same cell and of course in cells they change to become different types of cells so that's something that whole organisms have less ability to be completely completely different types of organisms you know they can't you are what you are I suppose that's one defining thing about but you can still change your focus from homeostasis to allostasis I think that is quite common and I wonder what your thoughts are around that and how that itself could be a useful way to frame things that's really interesting the matter in which I am looking at homeostasis and coupling of homeostatic regulatory systems to allostatic regulatory systems is based upon a more traditional idea that homeostasis might be considered more reactive where allostasis is anticipatory and allostasis takes a particular form it's quite interesting with respect to in the context of active inference because we talk about expected re-energy minimization and this is something that one can tie together with an ocean of allostasis when you start to think about the homeostatically coupled systems one system with respect to symbiotic systems might be playing a larger role in being able to make inferences further and further into the future and I think this is something that you've looked at previously on a slide Mike Levin's Cone slide that Daniel actually had up previously and I think this is a quite interesting manner of thinking about kind of a transition perhaps we've always been allostatic systems living systems have always had capacity to anticipate however when we start putting systems together we have a larger capacity or more capacity to anticipate further and further in the future and I think one of the dimensions on this particular Cone is this dimension of allostatic control this is my interpretation of Michael's work and I think this is very important and again you'll see that there's a balancing that if one is given a type of approximate Bayesian inference that's going to be what we suggest is going on you have to have to balance this with priors which are going to be respective to the past so you have another axis on the Cone and as these things expand sometimes the past is going to be a bit deeper on the Cone than the future but allostasis is on the top there and I think that like with multicellularity with the interactions more interactions perhaps we'll see a capacity a larger capacity to go upwards in the kind of future oriented direction and this is something again and I appreciate blues bringing this out this is again a matter of degree it's not a line it's a matter of degree and so just simply we know that there's certain types of micro organisms that do anticipate that do and this is kind of outward anticipatory behavior whether I think it's going to be Phaserum polycephalum which is kind of a plasimodial slime mold various experiments in 2011 have suggested that Phaserum can actually anticipate various periodic bouts of dryness this type of thing this is an outward periodic behavior an outward anticipatory behavior that's not to say that anticipation is not happening occurring prior to that in the internal dynamics of systems that's just an outward manner of kind of measuring it again I do think that this it's a quite important notion allostasis and perhaps all systems again I would be safe at least my position now is that all systems are pretty much allostatic systems that kind of enslave homeostatic regulatory systems to maneuver and invest their metabolic machinery in metabolic capacities to a particular type of state that perhaps they haven't reached yet considering you know a manner of trying to avoid perturbation thank you Matt very similar to mere adaptive inference being more homeostatic more like the ball just in the bottom of the bowl and then increasingly anticipatory rosin's ecology adaptive inference Steven go for it yeah this ties in actually a lot of work around coaching funny enough the world of coaching and team coaching is like and team coaching and I think this also brings in these other types of clusters groups symbiotic relationships beyond maybe what we might just think of as a system you know so the idea of a team in a in a systemic context is really interesting because and this might be how allostasis in a social context starts to enable that larger horizon so each even allostasis is sort of stasis and then working together you may start to break your own need for stasis and it just becomes some other organization and I think it would be really interesting looking at how the way that team coaching either for performance because there's performance orientated team coaching or there's developmental orientated team coaching but that idea and then the enabling context becomes what we would call the ecological niche I suppose becomes as much the focus of attention as the system if that makes sense and I think this might tie into what you're saying Thanks Stephen so Dave I know that you are a little bit to share but I'll just bring up one point related to this definitions of individuality and then through all anything you want to ask feel free to raise your hand because here a lot of terms like autonomous and self-organizing which sort of highlight the ability to cleanly partition a system like people will say the embryo it's a classic self-organizing system right except for the context of the womb and so there's sort of on one hand the desire to individuate and partition and look at a genetic system agent-based modeling, autonomous and self-organizing and then that is critiqued by the ecological perspective or the relational perspective so there's sort of a dialectic between like it's all individuals all the way down and it's all relationships all the way down we see that debate in particle physics we understand it less than we might some of the biological systems or in a different way but it's like two pillars that we're navigating between and it is a continuum in between so that's why it's such a nice discussion so Dave if you're around feel free to anytime you're ready you can go for the conversation theory otherwise blue go for it okay sure now slide yeah slides 18 and 19 we inserted that into the slide show that we prepared for Dr. Sim's presentation and these slides bring in some matters that he doesn't address in his paper so to justify that but not to paraphrase the insertions the free energy principle and active inference jointly constitute a relatively mature macro level framework for describing living psychological and other complex self stabilizing systems as of the moment no comparably mature micro level framework has been both proposed for and coordinated with this macro level of analysis I suggest that the basis for such a needed micro level framework was proposed by my teacher the late cybernetist professor Gordon Pask of Brunel University in the form of conversation there Matt's essay how to count biological minds considers the puzzle of whether an individual can come into existence and pass out of existence repeatedly Matt's account of this puzzle and please excuse me if I stray from over simplification into parody his solution in terms of degrees of integration is a perfectly acceptable insightful account macro or molar level the metaphors of quantity and elsewhere probability nicely express falsifiable common sense attributes of real world processes but if one states this paradox the paradox of the merchant the transient individuals in terms of conversation then one speaks of a conversation recurring over time between the same conversions this terminology makes the proposal more intuitive less shocking and more readily assimilable for readers better yet familiar ideas attention rhythm of discourse punctuation of a conversation through understandings or through agreements to disagree they all play their roles now what place would be occupied by the special disciplines that have enjoyed detailed treatment by conversation theory as a micro theory during his career professor pass carried out an array of experiments studying human learning of complex subject matters configurations of using configurations of software and hardware to administer and collect data from those experiments and developed a vocabulary allowing elegant description and restatement of the results of those experiments moreover he developed a theory of micro level learning teaching as a single literally irreducible complex of activities regardless of whether they were realized by living organisms by machinery or by complexes of those two fabrics I see many deep and broad analogies between conversation theory and active inference each addresses important aspects of real world phenomena that in a systematic way such that each treats of matters that are not so well handled by the other I believe that active inference and conversation theory can enrich and streamline each other by carving natural science scholarship closer to its joints very nice thank you Dave so maybe Matt any thoughts on that and then we'll go to blue I have to say that I'm absolutely interested in the cybernetic theory that I've concentrated upon and most influenced by it's very limited and conversation theory I'm absolutely interested in seeing more of this I suppose analysis and understanding more of this analysis so I'm quite ignorant about conversation theory and psyche most certainly piqued my interest let me think about this for a few minutes I'm still internalizing what you just said one note on the conversation a lot of times cybernetics falls into that individualism mode like it's the one self-guided missile it's not the conversation between the flock of birds so cybernetics isn't just about individuals it can be about collectives and it was our colleague Alex who told us via one of his teachers Chuck just yesterday that cybernetics died when it could explain homeostasis but it couldn't go beyond so we're talking about the transition from individuals to collectives and from mere homeostasis to adaptive allostasis over developmental and evolutionary time scales and so it's sort of like you can have homeostatic individuals and then they can have emergence towards allostatic collectives and then what is that process of emergence it's a conversation or potentially something like that using our folk understanding of interacting pieces which are us blue so my question is maybe technical but also maybe Cerval has some insight to provide as well so I wanted to ask here in the beginning of your paper you talk about free energy being an additive quantity so the free energy that arises at the macro scale is only the free energy that there is for the ensemble and maybe quantified by summing the free energy associated with each constituent micro scale Markov-flanketed system so is that true and like does that account for emergence because I feel like there has to be some like greater free energy minimization and so I'm not well versed in the technical mathematics of the FEP because I think about just in terms of the bioluminescence of Vibrio fissurei there's like a criticality that happens almost like a phase transition that the system undergoes to become bioluminescent after there's a certain concentration in the light organ and so does this criticality, this phase transition, does this not to minimize free energy to a greater degree or is it really just the sum of its parts? Wonderful question really wonderful question I do think that okay there might be a distinction in terms, here a distinction in terms of what the free energy of the actual constituents as a group have as an ensemble have and a distinction between that for example and bioluminescence as a particular type of free energy minimizing strategy I think it's a fantastic question I'm getting the notion of the summation of free energy from Bios and Ramsted and I think that pretty much the idea is that I'm not sure if there's going to be any, well I'd like that maybe I can ask you Blu, do you see that there is some type of inconsistency between saying that for example free energy is going to be an additive quantity and the fact that there can be some type of emergent property that is not going to be reducible to say the constituent relations and the elements do you think that there's some type of inconsistency there? I definitely do think that free energy across scales is additive like I agree with that wholeheartedly but I don't think that it reduces to the mirror some of the parts and I mean this also it's the work, the recent work of Rosas and also Eric Hall looking at the partial information decomposition over the phymetric that's used in integrated information theory of consciousness and so this kind of speaks to that there has to be some mathematical I just feel like I don't know that there has to be but I feel like there should be some mathematical quantification of more so it should be a greater degree of comparison when the system is more than a sum of its parts and so I think that that's an argument that you make in the paper as well that it can't be, it's not reducible to just the component parts so is the free energy then just reducible to the component parts there? That to me seems inconsistent. Okay, the manner of what I think I see absolutely see your point, the manner of what I'm thinking about this though is that okay well there's a very informative paper by Santos, he's a philosopher he actually deals with emergence and it's a really nice way I suppose I really like this, the paper that you suggested about kind of using the phymetric and various manners in which to do information decomposition over various time steps in order to think about some way to some metric for emergence I think there's been various metrics for emergence, I haven't picked my favorite metric that's something that I'm still getting into but I think it's quite interesting to think about okay well what makes an emergent system, is an emergent system going to be more than the kind of relationships and elements of a network and what makes it so? Is it going to be the relationships themselves or is it going to be something the combination of the relationships or is it going to be the interaction between the system the emergent system and the environment and I think when you start thinking about the interactions between the emergent system and the environment there's a distinction between what free energy is going to be minimized and the free energy of the system itself does that make sense? Is it there's a distinction being made and perhaps I'm wrong about this but emergence is not something one of the reasons why you have an emergent system is going to be for the simple fact okay well you have it's not only the interactions of the parts and the parts or the elements, it's going to be the interaction of a particular type of persistent and I think this is the idea of persistence with a particular environment or with an environment and it's the perturbations within that environment that occur along with that system in which kind of free energy is going to be free energy that has to be minimized is going to arise and I think it's like so you might say of course that the system itself when it has component parts the relationship between those component parts is going to be that system can be thought about in terms of an additive or summation of the free energy of the various parts and relations in that system but that's distinct from that system's engagement or interactions with this environment then you'll have extra free energy you have an additional amount of free energy and maybe that's another way a manner of thinking about okay well an emergence as measured I don't know if you can measure emergence in terms of free energy but of course you're going to be able to measure it in terms of some type of mutual information very nice, thank you I think that was a question there I don't know if you can measure emergence in terms of free energy like you said and you might have to look at something like mutual information and I was thinking that perhaps the free energy that's minimized by each individual system then when they become a symbiotic unit or something might end up being identical and again I'm thinking of humans again but if humans are walking and they happen to start walking in step that's going to look like an identical oscillation as they're coupling with each other so there might not be more free energy that's minimized by this coupled system walking in step together but it might look like the same free energy once they're at a coupled system and one thing that we talked about I think in the dot zero or around there is there situations where the cooperation the coordination increases synchrony so just increases the mutual information of the humans walking in step but also you could imagine anti-phase relationships which is kind of like division of labor or task allocation where the emergence is not due to copy and conform but the emergence is actually you know you go left I go right and so in both of those cases there would be a change in the non-equilibrium in the steady state which is what is introduced throughout the paper and so that starts to build the road towards quantitative discussions and you could take a system that you suspect already has emergence in nature and remove a part that's kind of like the gene knockout approach so gains of functions and loss of functions and all the other ways that we study function can now have something like a common grounding potentially using these metrics here so Dave and then maybe Matt if there's a response otherwise Steven thank you now blue asks is free energy additive does it exercise that I'd suggest it should bring that question to a more bound earth and more poignant focus take all the discussions you've heard of free energy and substitute for the phrase free energy the word anxiety it already is that way Dave what do you want me to do what would that show though do you think it's a natural fit could we put any word there what do you think it means well so just using anxiety it totally leads to the mob mentality and like ultra paranoia and conspiracy theory so like definitely additive possibly multiplicative though could be exponential there actually have been clinicians that have really used free energy although they didn't use the term as probably before the usage was coined William Glasser in control theory and his earlier work really has this quantum of anxiety that his patients had to manage these tended to be out of control criminal adolescents so they were pretty screwed up and he found ways to get them to reign themselves in and control themselves but think about the Blitz effect London Blitz you had objectively tremendous stress on everyone all the time but there was almost no hospitalization for shell shock among civilians including air rig wardens that were in digging a building beside them get bombed and they rush in and try to find people and drag them out extraordinary low rates of psychosomatic disorder there's a certain there's multiple paths people's brains are really complex there are certain ways that people actually do get shocked into rationality I've been in situations where the right thing had to be done right now with no hesitation no anxiety, no blame just do the right thing and it was not a question just do the right thing whether it puts me in danger or not a child's in danger it's not predetermined but typically yes what does happen is a thing called reorganization it's a technical term that Glasser uses it essentially means going crazy and in the really worst case the reorganizations get faster and faster and faster and more and more drastic as the personality becomes more and more simplified and that's something that Kurt Levin or Lewin worked through in his work on driving children crazy by giving them more and more complicated tasks and under more and more anxiety and he was able to make the kids personalities and thinking degenerate and simplify more and more and become more and more infantile and you're right Blue that is one of the things that happens in a mob it's terrifying to be in an even unified angry mob where everybody's mad and they're probably not going to attack each other it's terrifying and people that keep going into that situation things happen remember the presentation two weeks ago on the mutual suppression of the limbic system the reticular activating system on the one hand and the prefrontal cortex on the other you get these things that look like beliefs they sound like beliefs they act like beliefs but they're not beliefs because reason has never been invoked in the biological decision whether to incorporate this magician's spell into my feeling and my actions do you have to have reason to have a belief Dave? oh yeah in the sense that Phillip jaren's uses it yeah that's kind of the way he defines it a belief is something where the prefrontal cortex has been involved it's been in the loop and there's ways to bypass that I asked him whether he's worked on the social psychology of this or the ego psychologist he says no no no I have enough trouble worrying about people with damaged brains and how that shows the bypassing of rationality in the formation of delusions but I think it works the other way you may have seen my shotgunning of the apathemic wall by Nathan Rosenberg or rather Gabriel Nathan Rosenberg he's an academic in I think Kentucky he says look this is what's happening with this person with a non-existent personality who's doing what the mob tells him to do through the mob's megaphone called Foxman why do we like him so much because he's less than any of us he will always obey us so we'll give him power and then he'll give us what we feel like oh yeah and there's no reason it's very much a bypass yeah Rita LeBone's study and Mackie's study on the madness of mobs and the mass production of delusions and crazes and McCarthyism and all this unhealthy stuff thanks Dave and that was the guest stream 10.1 with Phillip Garans and so just to highlight one point and then Steven we're really talking about control theory maybe the first pass the first term in the Taylor series is about control narrow the error and that's kind of definition one control as homeostasis even if there's a shifting set point but then again it's the ecological and the relational insight and just the emergence of interactions that control takes on this broader usage to control your body temperature is to sometimes do a lot of movements that from that first pass look like a non sequitur like sprint so that you can get into the air conditioned room that's the expected free energy through time so now we have a common grounding for talking about control when it looks homeostatic as well as control when it looks allostatic or even innovative so it's very important to pull out some of these threads and also recognize their cybernetic origins and other areas too so Steven and then anyone else yeah these are all really good points I'm curious when we talk about free energy minimization I think it's interesting to talk about that from a systems perspective however if we want to get back to the kind of organism so the key thing with free energy minimization that occurs at different scales physical scales number of unit scale, time scale teleological scales we think about these different scales but that reduction there has to be a way for that reduction to be meaningfully engaged through actions or interactions like the metric ultimately what's going to be filtering through probably in some noisy maybe backdoor way is that if it's going to be sustained over time there needs to be some meaningful engagement over time so for instance the idea of the path that we walk starting to wear away the grass in our niche it's because we walk on the path more than once that we have an ability to observe that if we only ever walk that path once it wouldn't have a way for any variation to be noticed but we notice the path being worn by agents over time so when we're talking about these different scales these variational dynamics in free energy it's not just a reduction because we don't get a number each time saying this was lower than last we just get some new state in the world and we have to infer if it's changed from some other state in the world and we it may go up it may go down so we may infer at times the free energies has not been minimized and we might take over allostatic processes for instance if we have a big room that's not tidy and we have a group of people and they want to organize it into a studio space you know they may work together initially putting in a lot of extra free energy time and then eventually they know that they will be able to minimize their sort of systemic engagement as a group while it's not necessarily that they're each minimizing free energy as a system so this is almost where maybe challenging to get out of the biological sense and maybe come into some other disciplines where it's not all a system it might be more of a systemic context which is then more teleological so I'd be interested that question like where does this crop through this idea of being able to have a pragmatic metric that links the non-equilibrium steady state of the organisms phenotype or individual so they're actually able to then infer something from whatever that's happening to the free energy out there or in there or around them and how that changes things if we frame it away from systems per se it's something more systemic different uses of the highly polysemus word systems systems of systems for some people are systems so that's why we increasingly will try to be exploring the active inference ontology so that we know which pieces of the puzzle we're talking about and then that will unpack so interestingly into different disciplinary contexts Steven so nice points so in the last half hour I think there's a few different things let's maybe get to one of the key pieces and specific contributions of the paper which I would love to hear Matt your thoughts on which is just first what's going on with users and resources and second what is the difference between unidirectional multi-scale integration and reciprocal multi-scale integration right those two notions I suppose of unidirectional and multidirectional reciprocal and users and resources they're very much so connected the idea of a user we can think about this and I think Steven's to use Steven's example of an elephant path I believe that's what these things are called sometimes so you see these paths that are treaded and are typically used or referred to when we're thinking about kind of niche construction like paths for niche construction this is something that is in the environment that is being used say is a resource that's being used and it's the effect of previous action on the environment returning to the notion of so that that would be a resource used by a particular cognitive agent or an agent that's engaging in adaptive active inference so number one it has to be recognizable and it has to actually in order for the system actually to engage with that particular resource it has to be recognizable as such when that recognition can come through for example epistemic foraging or actually just apply the model to a particular type of resource that has been applied previously this is also the case with something like stitch-mergy where you have something it's quite lovely what you have for example ants leaving pheromones within the environment and that pheromone acts as a particular type of mediator directing further behavior by agents of a similar type and that's again environmentally mediated communication or what they might call environmentally mediated conversation between various elements of a system and the pheromones themselves of course are resources in the environment that's going to be perhaps contrast with the idea that not all resources need themselves are themselves they have to be limited to being resources also to make a strange case we can think about stitch-mergy actually what if it wasn't actually the pheromone that was being used but the other ant the presence of the other ant that was being used the interaction with the ant itself so here the two using perhaps you have to tell me if I'm using this wrong that's terminology but you have two discussants right but both act as senators and receivers and actually they're communicating something they're having a conversation but there isn't any mediation in between the two that can be offloaded on to in this particular situation offloaded on into the environment so the signal itself is going to be the direct kind of interaction between the two rather than a pheromone now that in that case we can map that on to something that would be a kind of hypothetical case of reciprocal multi-integration right and again stitch-mergy just happens to be typically in each cases in the way that Pierre-Paul de Grasse suggested it when he was studying termites this is something that's actually involving a particular type of signal in the environment that's offloaded to the environment so it's going to be naturally something that's going to be a unidirectional multi-scale integration between a system between various systems whereas for example when you have direct interaction between two cognizers or two systems two adaptive systems they can also of course use some type of resource they can be signaling and this is something that of course happens with the degree of fissure eye and the bobtail but there is also going to be some kind of other direct interaction between two systems and that's kind of the reciprocal multi-scale integration awesome summary I believe the technical term would be converse ants for that kind of a conversation but it's a great point how when you have an adaptive active inference agent and then a niche that's being modified that does mere active inference like the spiders web that's unidirectional but still multi-scale whereas when you have adaptive and adaptive in discussion in conversation even again if only in a folk sense you have adaptive on either end and it can allow for a new non-equilibrium study state to arise so not that there isn't a non-equilibrium study state of spider plus web rather than them separate in fact there will be but it still is a different type it's like a double directional or reciprocal stigma g so just such a cool way to connect that blue and then Steven so like again this like line between mere active inference and adaptive active inference like is it like makes me want to attack it a little bit I think about the niche modification that for example humans do is extensive and transcends like generations and I mean work it like contributes a lot to the collective intelligence of the human species and so forth and I also think about in terms of like the Gaia hypothesis right like so if if the world is one giant organism and this is something that we talked about with Mike Levin when he came to the live stream so if the world is one giant organism is the niche constantly adapting to the species that are in it like is this a continuous adaptive active inference and just in terms of transcending time scales you think about like these symbiosis I guess they're symbiosis but like relationships that develop between you know like orchids and their pollinators for example so I mean these are intense co-evolution things that happen over many many many time scales time scales and I mean we can think about the plant is the plant a cognizer I mean that's debatable I would say to a degree like I'm not binary here like yes no is not really a good answer but the niche may be constantly adapting to systems to a degree I think that's a great point really really great point I do think the niche is adapting to systems absolutely but I would say that the niche is adapting in a different way and we can use of course we're going to have to be specifics because this adaptation takes as a policy in a sense adaptation in terms of what the niche is doing if the niche is not itself okay perhaps you want to I can be accused of being biocentrist in one sense if the niche itself isn't a biological system then we can kind of perhaps push against this idea that it's adapting in the same way that a biological system would adapt to maybe one way of phrasing this and I'm not sure if this is going to hopefully make it clear my point is that biological systems one way of understanding them is that they have set points moving set points throughout a particular trajectory this particular trajectory might be a trajectory across a phase space throughout their lives so you have a moving set and this moving set is something that's organized around a particular type of phase space organization and it's recapitulated within a particular type of phenotype defines a phenotype it characterizes a phenotype or a generative model with a particular say a field of dirt we might be able to actually walk across a field of dirt or a set of cement stairs it might actually over time for a few hundred years you have these interesting stairs at the castle of Edinburgh that actually dip and these dips in the cement are evidence that people have walked there before but we tend to walk where other people have walked and this perhaps one might call that an adaptation to a certain degree but I think that might be pushing the limit on the word adaptation given that that set of stairs does not have a set point it doesn't have the same type of it's not the same type I suppose of self-organizing system in that sense and as such I would think that okay well maybe the niche does adapt to a certain degree but there's an adaptation with biological systems that are self-organizing and have organizational closure and I think this is quite important that it's different than say abiotic systems and this brings that I suppose I would love to talk about this in a bit it's a question that Blue brought up in the last in the last kind of a live stream was the manner in which we can actually understand systems or identify particular types of individuals that are involving the environment so it's not actually saying that the environment is separate from the system otherwise we wouldn't actually be able to actually talk about the emergence of actually living systems we can get into that later though so I have to just push back on you so think about the earth and the atmosphere of the earth okay now this is not necessarily a biological system this is nitrogen, carbon dioxide, water, clouds, etc but here we are humans releasing tons and tons of carbon into the atmosphere totally changing the set point as you described it like the homeostatic set point or whatever the non-equilibrium steady state density that the atmosphere of the earth currently resides in and has resided in for x million years right so but we are changing that and we are seeing the earth adapt to us and the changes that we're making in just in terms of hurricanes storms I mean the weather patterns are completely and they're destroying large amounts of humans right and drought, famine all of the things that are going to cascade through climate change as we see it progress so is this not adaptive active inference on the part of the earth or what do you think of that really interesting I think in order to make that particular type of assumption one would actually have to ascribe to something like a guy hypothesis like Lovelock's guy hypothesis with the same type of thing that Lynn Margulis was into later in her life and I think that's an assumption that has to be made given the fact that okay well just because a response that happens chemically doesn't necessarily mean that it's not actually an adaptive response but you know gene expression is biochemical adaptation right so it's not about just being chemical using chemical or something like this changing molecular structure as some type of criterion for adaptation I think there's an assumption that has to be made about what the system is there and if you're thinking about the earth as a system maybe the earth is adapting to its constituent parts right but that would involve intrinsically the earth being a particular type of living system given the fact that it has living parts one other quick thought on that is as a model comparison framework so go back to SPM DCM all of our categorical model comparison techniques in the toolkit we can actually evaluate should we have cells as the units and then organisms and then the system and then the world or should we use a different partitioning which one of these models is explaining predicting designing controlling better or which one is more parsimonious I prefer simplicity over utility or I have a function that balances simplicity and utility there's all kinds of frameworks that's kind of part of pluralism is starting with and being with that multiplicity of answers and then knowing who we are and what we use to decide on which models to use so Stephen yeah I'm very much like to keep that ecological niche in the picture so if it's organism in niche and groups of organism in niche and then watershed and micro environment or micro climate as a niche I think that kind of becomes like a defining organism or unit in a way is like watch the watershed what's the kind of climate as experienced by that ecosystem and then of course you've got many ecosystems around the world from the Arctic to the Antarctic and it will be in between now they might all be collectively affecting the broader carbon dioxide levels of the broader ecosystem but how that then jumps to make the whole world an organism becomes for me slightly challenging right because it's always a niche up to then and then suddenly it jumps to be an organism because we look at it from space and we see a nice round ball and it sort of looks alive I wonder whether there's a one of the things that I'm really interested in is is like you're saying how things get ascribed to the niche and we talked about the path walking on the path and in a way that dispersive thing animals can sort of also intuit some of that as well from the affordances and the directions and again bringing stuff in from systemic design type approaches is one thing that's interesting is if you then say humans like humans what we do is as a child and maybe some apes do this but not many is we point at something as a child and then we point to the parent and that's becomes it's in the niche but it becomes what they would call a boundary object in systemic designs and you're trying to bring together a team around something which is beyond one discipline so what you need to do is you say okay we can't just use the variational free energy landscape of this domain ecosystem which we have all attuned to in our sort of own way because now people are running by different rules right we're running by the physicists and the biologists right they're just going to be so you get a boundary object which you point to so it's something in the niche and then you sort of work to that so anyway I just think there's something quite interesting then about like Daniel was saying there about you know what sort of language in do we use but this scale problem comes back every time but I wondered like the difference between a boundary object or something which can be defined which I'm not sure if other animals can really do in the same way maybe a dog when it urinates up a tree I was just thinking about that does that become a boundary object in their world I don't know but anyway it's something that comes to mind Thanks Stephen and with a transdisciplinarity in cancer biology they've recognized the niche the tumor and the stem cell niche in the micro environment and there's so much more to say on these very interesting topics just in the last few minutes we can go to a few of these final points in the dot one it's always awesome to have the author and unpack some of the ideas so this is kind of our little fractal dot two setting us up for our discussion next week and thinking about implications and how does it influence how we act moving forward at what kinds of questions we're looking at you wrote about some of the implications for thinking about individuality from a cognitive perspective to say it broadly and it'd be awesome to hear your thoughts on the folk psychological notion of the cognizer and where does the FEP come into play right Thanks for the question so the folk psychological notion of cognizer again my argument is that piggybacks upon this idea of the folk psychological biological notion of the individual and once we kind of free ourselves from thinking about individuals specifically in terms of replication dynamics that opens up a space for us to think about cognizers that are discontinuous with replication or replicators and I think this is like that's going to be very important again when taking the this particular root of thinking about the evolution of cognition when thinking about for example okay if cognition goes all the way down to a certain degree then what part does it actually play in the construction of larger systems and why would the construction of larger systems actually be some type of adaptation for each of the systems involved so when we're thinking about for example I think one particular thing that it's quite important will be quite important to think about this for is the manner in which we think about artificial intelligence for example one way of course two ways of course of doing this is we go for various kind of specific tasks whether it's going to be kind of computer vision or GPT-3 type tasks where we can actually analyze a large corpus of text and kind of recapitulate that that's one thing the other thing of course is the task general or kind of domain general notion of AI and I think in order to perhaps tread into the department or tread into that field of more kind of task generality we're going to have to think about even simpler questions as to how can a system actually start maintaining itself according to its own tasks and again we come back to this notion of how can we actually set a system up to have its own intrinsic motivation and one way to do this is perhaps to piggyback on living system so you introduce a living system to an artificial system and have them communicate in such a way that's analogous to some type of symbiotic system this is something again perhaps that Michael Levin has been thinking about doing actually in his lab and I think this is quite an interesting kind of method to think about the importance of symbiosis in developing artificially intelligent systems you know a system that can identify go to a room an embodied intelligent agent that can traverse a room and identify all the dogs that's great with an accuracy, a high accuracy level that's fantastic but if it can't respond to the dog knowing that a dog is going to actually dismantle it I'm not sure if I would want to call how intelligent that system would even be in terms of having basic intelligence and I think that FUP and active immigrants allows a particular type of perspective on the notion of basic intelligence or intelligence that's surrounded or that's grounded in the notion of a nest density or grounded in the notion of a type of moving set point that's equivalent to a homeostatic moving homeostatic states and out of stasis supporting states thank you Matt, Shannon thanks for that I wonder how important drawing on what Lou said and David and Steven I think also earlier how important some sort of effective or emotional state or awareness in knowing and cognizing and how important that is for a symbiotic relationship to be called a cognizer and whether a neural network that is working with the human may be to compose a piece of music how important it is for that neural network to have cognizing capabilities of its own or maybe some sort of effective awareness or some sort of effective processing that it is able to develop in some way on its own or maybe only in symbiosis with a person and I don't know if the Spoptile Squid and Vibrio like do they have sort of effective awareness together or does the Squid seem to experience something like pain I suppose would be one of the most basic effective states we might have so how important would basically is effective processing in delineating individuals or cognizers great question Shannon I'll take a go at it but I do think that at a certain level of complexity effective states play a huge role but because I suppose I'm making an assumption that cognition goes on way down and not every system is going to be an affect had that capacity to have some type of effective awareness one might actually think about some other type of the subcomponent plays a similar role as affect and perhaps that just might be something like you know something that grounds that grounds a particular type of motivation to stay in homostatic states in particular homostatic states and this could be Pamela Lyon has written about this just recently actually and it's not affect necessarily but it's valence negative and positive valence that might not and it has less to do with awareness or sentience in that sense but more to do with actually a particular type of organism or particular type of system responding in ways that again responding in ways that keep it centered and whether or not whether or not that kind of valence a bacterium might climb a gradient of glucose for example but bacterium might not have any type of experience of that glucose gradient but somehow we might say that that bacterium is actually it's positively valence it's something that it's attracted to right and maybe in simpler systems you have something more like the notion of valence rather than affect but a certain complexity after a certain perhaps degree of complexity arises affect is very important and I think particularly when it comes into multi-cellular systems that perhaps that have to actually compute larger like much more information has to compute much more information and has to interact in their environments their environmental niches are much more using a term by much more opaque so affective information affective the matter in which we can use affect and how that affects cognition I think it becomes quite important but I'm not exactly sure number one is probably not a line it probably develops but just what that function is a different story I don't know do you have any ideas about that I don't have an idea about where that would start but I feel like it would fall off so we might have the small scale like our bobtail squid and Vibrio who might have something like valence that they're responding to and then as we get slightly larger organisms you know dogs, people then affect will play a lot more of a role but then as we have bigger systems we have multiple humans what is the affective state of a crowd of people like we can ascribe like they look angry or they sound angry but they as a crowd are they feeling something as a crowd kind of like group intentionality is there a group affective state it might kind of start falling off again so you might have this sort of U shaped curve of where valence matters and then somewhere here in the sweet spot affective states or emotions really matter that's super definitely something absolutely something I want to think about and also reminiscent of the discussion of affect as well as higher order parameters in deep parametric active inference where we even talked about which states were opaque or not whereas opaque usually meaning like you can't see it you can't see in but actually the opaque mental states are the ones that we do experience so that was in the sand bed Smith paper and whether other systems feel things what is it like to be a squid what is it like to be a symbiosis that's the phenomenology and the metaphysics but what is it like for us to model that system as if there's a higher level control parameter we can know what that's like it's a lot like this conversation so just everybody can provide a last comment on this awesome discussion I think the takeaway for me is that in a post Darwin biology it's easy to use the simplified trifecta of evolution by natural selection phenotypic variability heritability of the phenotype fitness correlation and variation and fitness and then project that onto all the kinds of individuality cognitive the rights and responsibilities the economic unit the sentient unit all of those as just simply under the coattails of the Darwinian replicator and I think pluralism and a bevy of other perspectives suggest that that's just not the case blue and then anyone else with a last comment I just wanted to say thanks for coming I really enjoyed this discussion and I just started reading your paper your 2021 paper with Pizzolo the modeling ourselves and I hope you'd consider coming back to talk about it with us thanks so much it's been a really incredible conversation discussion so thank you all for inviting me Shannon or Stephen or Dave any last comments awesome well all participants thank you so much and Matt especially for writing the paper and coming to talk about it seeing the dot zero before so peace out everybody see you next week bye bye