Maps are their own territory with Mahault Albarracin

Unpacking Complexities with Mahault Albarracin: on Cognitive Computing, Sexology, and the Patterns that make a difference.

Join us as we try and discuss what cannot be discussed, touching on sexology, cognitive computing, and technology's role in societal change. Albarracin , currently pursuing a PhD in cognitive computing, delves into the definitions and interrelations between disciplines, the patterns that shape cognition and awareness, and the potential of technology to influence social dynamics. We touch on topics from the integration of diverse perspectives in understanding complex phenomena to the role of AI and its ethical implications. This journey is also an exercise in complexity, challenging traditional viewpoints and offering fresh insights into contemporary academic discussions.

Mahault's Google Scholar: https://scholar.google.ca/citations?u...
Connect with Mahault on LinkedIn: https://www.linkedin.com/in/mahault-a...
Research Gate: https://www.researchgate.net/profile/...

00:00 Introduction and Guest Overview
00:41 Defining Sexology and Its Challenges
04:16 From Sexology to Cognitive Computing
10:30 Exploring Constructivism and Predictive Processing
13:22 Phenomenology and Shared Constructs
15:58 The Complexity of Modeling Cognition
22:20 Beyond Dichotomies: Self, Other, and Community
29:12 Exploring the Concept of Representation
29:50 The Self as a Process
30:32 Clarifying the Question

Transcript

A Map is also its Territory: scales of awareness with Mahault Alberaccin

Andrea Hiott: [00:00:00] Hello everyone. Today, we have them Mauhault Alberaccin for a very exciting conversation. We had to keep it in an hour. So we sort of speed through all kinds of things. But. Somehow in a way that's very illuminating. First, she helps me understand what sexology is because I was never sure how to define that. And that was her first field. So to speak before she got into cognitive computing and she's now doing her PhD in cognitive computing. With Karl Friston, which is how we came into one another's orbit, so to speak. This is one of those [00:01:00] conversations where we talk about what cannot be talked about. In a way that actually is

illuminating even if it's not in the ways we were trying to, for it to be illuminating or maybe it is, but. Yeah, you have to do this sometimes. Just jump in and try and understand. These nested fractal

patterns that as soon as you talk about one of them, you've opened up 5,000 more. So that's what we try and do here and thanks to Maul and her articulation and her study and her wide layering of knowledges. . There's a lot in it. We talk about patterns and how they change, where we draw the line between. Inside and outside and Markov blankets. What those mean? Ideas and knowledges and. Where awareness comes into play in terms of changing patterns and how we can understand these recursive patterns that are cognition. We talk about them here. And end up talking about how all of this was it sort of the way that we integrate different perspectives towards changing [00:02:00] society, through changing these patterns that we can't really speak of, but we have to speak of, and we can't really model without reducing, but we have to model without reducing and so on and so forth. Um, actually, none of that, what I just said is even accurate either, but you get the point that I'm trying to understand how something like mind and intelligence and cognition is related to patterns and to the ways that we represent those externally and try to understand them, even as they're always an ongoing dynamical process that can't be represented in full of course ever. That's the challenge, how to link together, all these incredible ideas and theories and all this amazing work that's going on right now. It feels like there's a framework, there's a through line, that's connecting it all. The way making, one thing I was thinking about a lot while we were talking was I was imagining bees or birds or something sort of swarming swallows, maybe swarming [00:03:00] left and right. And slight shifts in the wind changing all of that. We started talking about that and like many different levels here. What changes the direction of the pattern. The pattern slowly shifting from within, and then there's a different pattern. The patterns within patterns. Anyway. It's beautiful. Rich. It's a dancing conversation. About this confusing, mysterious, delightful world. That we are all in. And I just want to say thanks to Maul for all the work she's doing and for her amazing ability to take in so much information and assimilated and articulate it across different disciplines. 

 I hope you enjoy this conversation and that you're having a really good day out there wherever you're making your way. 

Andrea Hiott: Hi Mal, it's so great to see you and to meet you. Thank you for being here today. Well, thank you for having me. So, [00:04:00] I'm very interested in the work you've been doing lately.

There's so many papers you've been writing and to get into that though, I want to think about your personal trajectory, because the papers that I want to talk to you about are a lot about cognitive computing and active inference and modeling and technology and AI and all these sort of subjects.

However, I seem to have discovered or heard that you started in more humanities and sociology and even something called sexology, which I'm not sure how to define exactly. So. Could you help me? So i'll 

Mahault Albarracin: start by defining what sexology is because it is a concern that sexologists themselves coming out of university are trained to answer Oh, wow.

Andrea Hiott: Okay 

Mahault Albarracin: And then i'll tell you about my path. Thank you. Sexology is basically an amalgamation of a lot of fields you can go biology psychology anthropology. We do a lot of sociology as well sometimes linguistics, sometimes media studies, [00:05:00] and effectively it's about the topic itself. So anything that touches on anything related even tangentially to sexuality, like pregnancy or diseases, it can be couples therapy, it can be well being, so these are all the topics that are touched on.

And so we dissect these topics from multiple aspects and try to find the connection points between the fields that find areas of I want to say synergy, but mostly impact, because what we are more and more going towards in the social sciences field is psycho ecological biological approach, where you can't just see one in isolation.

You have to take it all into account, and that creates a problem, because it's a great it's a great desire, but don't have the right bridges, the right paths between the different fields of study such that sometimes the languages don't match. Sometimes we think the languages match, but really we're talking about different things.

And sometimes the methodologies themselves [00:06:00] aren't even aligned. So we couldn't even try to measure the same thing if we wanted. And that goes beyond the problem of scale. Oftentimes when you talk about a field, we're talking about related things, but they're, they're acting at a different scale. So for instance, we're talking about psychology and then sociology.

In general, we think we're talking about phenomena that act on, say, the individual scale. And then we're talking about topics that span multiple people, multiple individuals, and then even institutions and countries that become their own groups.

So, we are lacking the tools to understand the ways in which these things interact. So that's, I think, gets to how I started my path and where I am today. I can see that. Yeah, right? Like, I started in sexology because really I was interested in psychology but I didn't want to do just what my understanding of psychology was at the time, just like more than 10 years ago.

I really thought that I wanted to get into something practical and [00:07:00] useful and that would get to the root of what is being a human being. And to me, sexuality, there was something Fundamental about it that spoke to the deep connections that we can make with other people and therefore something transcendental about the way that we relate to one another.

Unfortunately, I was a bit let down. The field is not what I thought it was. It, it mostly talks about, you know, communicable diseases and you know, the kinds of things you do when you're pregnant and what happens to your body. And all of this is massively important, but not exactly what I went into it for.

So then I did my master's thinking. Let's push this a little bit further. Let's go into culture. Let's try to understand how culture is this sort of framework that shapes our perception of reality and therefore perception of each other and how we might exchange through these cultural artifacts. And I got a little further to closer to where I wanted to go.

But then again, I was limited by the fact that the fields [00:08:00] are hazy and a little bit blurry. They, they mean to be blurry. It's on purpose because they mean to be inclusive and try to not apply a frame of reference. That is not appropriate to phenomena that is poorly understood, and that may not be where the researcher is coming from.

So they acknowledge that there is a bias from being human and existing in the world, that you can't really get away from. And so, we try to acknowledge and pull the bias away a little bit by keeping things a bit blurry, not defining too much, letting the definition stem from the people, the cultures you're studying. But that doesn't give us a lot of explanatory power. It doesn't give us a lot of predictive power. And so it's, it stays a bit there, you know, it's like, okay, let's make a description. It's beautiful. It's poetic. It's abstract. We can't do anything with it. And I find that that also leads to a perception from the, [00:09:00] let's say hard sciences, the life sciences, that.

It's useless, that it's not even science. Oftentimes, we separate these sciences because they don't even consider social sciences to be a science. I'm saying that it's a broad thing, not everybody thinks that way, but I've really encountered people who wouldn't even qualify the humanities as a science. And so I was a bit flustered by that. I wanted to go further. I wanted to find what constituted building blocks. And anytime I mentioned the word ontology, I was met with heavy criticism because they were like, no, no no. Ontology is an essentialist concept. We are not essentialist here. We're constructivists.

And I believe that. I believe that things are constructed. I believe that culture shapes the way we engage with reality and the culture itself is a, is a construct of interactions and phenomena, which is why I'm very interested in neomaterialism. But I don't believe that this means you can't come up with essential [00:10:00] Processes, essential building blocks that lead people to potentially predictably make certain assumptions or predictably think in a way that will lead them to act in a certain way.

And so that's how I got into trying to think further and, and So I was trying to find paradigms that were heading in that direction, and I found predictive processing. I was engaged a lot with the community at the time that was really involved with AI, even though I wasn't part of that community, but at the same time, I was learning how to program.

I built a company, a sex tech company we were doing at this point, I decided it was web development, but therefore I was getting more interested in it. We were linking patients with sexologists who wanted to give them care across the country, because one of the main issues is access. In rural areas, you don't have access.

And so, but because I was learning how to program, because I was learning about this community and AI and their concerns. I started reading about cognitive theories that had these building blocks, even [00:11:00] though they weren't specifically interested in social sciences for now. And that's how I fell in the work of Maxwell Ramstad and Karl Fristen, who were starting to discuss niche construction, which was very reminiscent of ecological theory, which I was very interested in.

I was very familiar with having studied in my master's so I was seeing that they were making these connections and I brought into the field the notion of social scripts and we formalized this and now we have some degree of the building blocks and the missing piece still, to some extent, is how you map across scales to get an emergent feature, because active inference is scale free, but it does give you emergence per se. You still need to understand the historicity of a system in order to understand how certain things emerge at a given scale through some attractor dynamics. 

Andrea Hiott: Okay, wow, there's so much in there, and now I can sort of understand how you got into this, because it is, there's a kind of strange set of constraints in both these areas that you're [00:12:00] in.

I mean, I've kind of been in both too, and more philosophy, but I think it's similar in a way, we're trying to take account of multiple perspectives, like something like constructivism that you said, where, we kind of understand having read so deeply and like tried to see from so many perspectives.

That everyone's got a little bit of a different trajectory. And even when we're in a shared space and looking at the same object or in the same encounter, even we're, we're perceiving it differently. And so you kind of start with that and then you feel as if you can't really put any anything seemingly objective or ontological onto that.

And then on the opposite side in science, it seems almost. This is all I'm generalizing in a big way, but it seems like we take away all of that stuff and try to get to something that's very distilled that will kind of work no matter what. And and then that also ends up in a strange constraining place, which I think something like active inference is starting to explode a bit.

And it kind of, for me, has to do with the possibility of modeling [00:13:00] and to get into this a bit, I wonder, are you the one who brought in the idea of the phenomenology into all of this? Because I've started seeing that word coming a lot, and I wondered, did you study some of that early on, or was that someone else, or So I'm not the one who brought it.

Mahault Albarracin: That was definitely Anil Seth, Maxwell Ramstad, and a group of other researchers who really started getting familiar with this. They wrote a paper I think on the computational accounts of generative passages. This is not my title. I'll send you the paper later. But they, they started it.

They were very interested in computational phenomenology, but I am also interested in it and I've done some work in that field. And my main interest was how phenomenology becomes shared. So intersubjectivity and how it can become shared is through the common constructs. That we have that allow us to parse reality at all.

And those can be very basic, like time and space. You start with that . Mm-Hmm. . Mm-Hmm mm-Hmm. from those basic constructs. [00:14:00] Then you extend on qualitative expectations. And those qualitative expectations then become sort of, constrained and patterned. And that's how you get to potentially, you know, you know, scripts and, and culture and language and anything like that that has.

It's something that can be hooked onto that allows us to compute further than just the momentary thing that's happening. 

Andrea Hiott: So I don't know, I just want to say something and see what you think. I feel like in a way we're kind of stuck in like a flat world understanding where let's talk about something like sentient behavior is sort of what we're trying to understand in a way, right?

Or what you got are trying to understand with. Active inference modeling and so on, and this goes into cognition, maybe you can, correct me on that. And, but in a big kind of way, we're trying to understand sentient behavior and cognition and all these bigger issues come out of that, of how communities form and different sort of protensions and so on.

But, It still seems like, to me, when we talk about it, [00:15:00] maybe just in everyday life, when we think about behavior or cognition, we still have a linear sort of model of, you know, we start with a cell and we end up with AI and it should all fit together on a kind of line somehow and, I feel like what you were kind of expressing with the humanities and this constructivist multi dimensional view is very hard to kind of fit into that way of thinking or modeling in science.

But I feel like reading your papers, especially the ones on scripts this, the polysemous idea and the, I mean, there's many, many things where I'm starting to feel like it's opening up and becoming multi dimensional. I don't know, just saying that, does that, what does that correct me? Or how does that strike you?

No, I think it resonates a lot. 

Mahault Albarracin: I mean, the problem, the problem we face with science is that you're trying to establish causality. That's the thing, right? It's easy to establish correlation, but causality is tough. And so what you're trying to do is, like, move away all the possible noise to find [00:16:00] one strand of possible causal relation.

So obviously you want to try to make it as linear as possible, and ideally you want to make it reductionist, because then it's simple, right? If you can start with a thing, you can just kind of, like, start it and it'll become what you need with this one principle you have to, so that would be ideal. But again, this is what I was saying about the, the emergent structures that only occur on a given scale, because we can't yet predict this very well we are, unable to fully grasp how different factors come in.

But we are indeed interested in it, and I think that's what Varela was trying to get to with the ecological sciences. We, we are trying to do the same right now. Active inference is basically also heading towards having a proper understanding of how different kinds of observations lead to some [00:17:00] types of beliefs, and those can be obviously as varied as you'd like.

They can take in different kinds of factors, different kinds of scales. And I feel like because it's being applied in different fields at the same time we are getting some way to sort of hook all these things together, but we haven't quite yet found. the proper way to hook them. We are, we're, we have a language, we have a common language, active inference, we have a common set of processes, which eventually all lead back to cognition and fundamental principles of how information is encoded on boundaries, which itself is pretty fundamental.

So in a sense, it's reductionist in that sense, but it doesn't necessarily entail that there are no emergent features. So the key, I think what someone will have to come in and do. is find properly how these things can be connected. What my work is trying to do is show that the scales, I mean, are an illusion.

That's the work that Maxwell Ramsey have started. I'm trying to show that even the internal external [00:18:00] boundary idea is just a function of measurement. It's not, it's, it's, it's also just an illusion. So if we, if we start to understand this and we can cast away these artificial boundaries. then maybe we can find the true way in which things interact and therefore the ways in which different different modeling ideas that are in active inference can connect to one another.

But we're very far from finding the answer. A lot of work still has to be done.

Andrea Hiott: But I do feel like what's happening is that there's a scaffolding of, of something like mathematics, but also just language and description that's come towards understanding something, some very basic things, like how a body encounters the world.

So, with active inference now, through mathematics, but also through, you know, English or whatever language we speak, we can start to talk about it. It might not be exactly the right way, but we can think of it as, you know, minimizing surprise, or basically a body moving through its encounter. Aligning [00:19:00] to those regularities, and then when something is against those regularities, having a kind of a different response, you could probably explain it better, but and once we can start to understand this a little bit, is it more possible then to begin to understand that multi perspectival or constructivist idea of that every agent would be a different spatio temporal position, of course, and thus would have a different encounter, even as many of these regularities are, are shared.

But precisely. That's exactly the point. That's exactly the point. I 

Mahault Albarracin: think this is why this notion of objectivity is flawed by nature. Objectivity shouldn't mean what we generally think it means. Obviously, any, like you said, any body in the world will necessitate a different perspective. If only because I'm looking at this rock from here and you're looking at it from here.

So our experience of the rock, our regularities related to the rock will be ever so slightly different. The reason we have a [00:20:00] language is because we're constantly trying to collapse these differences by hooking up to common experiences and trying to peel away what constitutes an error of the difference between our perception.

That's what communication is. And we're confronted with this never ending paradox that what do we hook onto? At one point, there's going to be an irreducible difference in our experiences. I think that's what Slavoj Zizek is trying to get to. Whatever you think of him as a scholar, he does sort of point at this idea that at the core, there is nothing.

Because We can't, we can never, it's, it's an, a never ending thing. Dorito is kind of going in the same direction. It's a never ending swapping of different things which, there's nothing behind it. I think the answer lies in Phenomenology and the, the fact that we're not [00:21:00] individuals that have seen only one thing and that we are capable of sort of placing ourselves temporally within the perspective of another agent, and because we're capable of doing this temporal swapping, we're also capable of abstracting away what might constitute another agent.

some degree of difference and only focusing on this statistical regularity that we have both experienced. And it's because we are temporal agents and not simply, you know, agents that only focus on one point in time. That being said math isn't sufficient. That's why phenomenology here is critical.

Qualia in itself and understanding the relationship between matter and information and how matter translates to a representation is not yet a fully solved question. And I think it's going to be the, I mean, it's, it's always been the critical question, right? It's always been a question, but well, we're so close.[00:22:00] 

Andrea Hiott: I think there's two things that come, come to mind there. And it's back to that, in one way, that flatland thing I was trying to say, where I think we're smooshed into this binary way of thinking. This is a lot about beyond dichotomies, and you've already sort of pushed us past them quite a few times in the way that you're trying to talk.

But I do think like the two things are, one is this self other thing, right? I can see what you're saying about Slavoj Žižek and all this. There's nothing there. We could see it that way, but we could also see it that we have a kind of a strange idea of what we are. That we've, we've taken the Markov blanket a bit too seriously for good reason, to survive.

I mean, we couldn't have even gotten here if we hadn't. So it's not to say that we did something wrong, but we're in a different place now, right? That maybe we can reimagine what, what's happened, what is the self actually and what is the body actually. And we can start to understand that there's a power in that community, you know, you, you write a lot about community intelligence is almost the only kind of intelligence and in a way, if we could kind of have a different realization as the spatiotemporal agent that we [00:23:00] are it might shift a bit what's possible in, in those realms.

And I think that's also connected to that flat world sort of understanding that we have in the modeling too, that because you, you could, if we could understand that everyone has this, unique spatiotemporal trajectory from birth to death, and if we could model it somehow, really, like, literally, we would see that there's so much overlap, and that there isn't a gap, actually, between all of this stuff.

It's just, how in the world do we visualize that, and model that, and put language on that? Yeah, I mean, I think we're getting to the point of like, 

Mahault Albarracin: do we need to recreate the universe to understand the universe, right? Well, we're sort 

Andrea Hiott: of doing it, right, with digital twins of everything and spatial web. 

Mahault Albarracin: Like, it's, a lot of people are like, well, you need to make a map, not a territory.

And it's like, well, maybe, what constitutes, the map that allows you to understand what is [00:24:00] fundamentally just a map, right? Like, literally everything you perceive is just a map. So what we're trying to get to is as close to a map as 

Andrea Hiott: a map is. It's, it's, so eventually Exactly, that's what I mean by if we can understand it's both.

That's the dichotomy. It's either a map or it's a territory. That's the way we think right now. It's either I or other. I mean, but it's so hard for us to, Open, like that's the cognitive shift. Where's the third space or, or even the multiple space where both of those are okay, you know, map and territory are not a choice.

They're different ways of understanding a process that's kind of, inseparable. No, absolutely. And I mean, 

Mahault Albarracin: again, it's like this map territory idea gets back to again, and information and, and matter. Okay. And effectively, there may not even be a real difference between information or matter. It, it, now we're getting into, like, like, physics, and I'm not gonna go there too much.

I will refer people to Chris Fields if they're interested, and he will have much [00:25:00] better things to say than me. But effectively, I think what we can, what we consider to be a territory at a given scale is just a map at another, or, and, and so, and so, that distinction is really just a function of your perspective and the way that you measure reality at this, at this point in space and 

Andrea Hiott: time.

Mahault Albarracin: But it's still 

Andrea Hiott: important. That doesn't mean it's not, like, it suddenly becomes meaningless, right? You know. 

Mahault Albarracin: That's, that's kind of what I mean as well. Like, I feel like people put valence upon certain terms. Yeah. Like, you, you had to sort of, shield yourself from criticism by saying, it's okay that we've made this blanket important.

It was, yeah, of course it is. Every, every process. Is effectively natural just by virtue of existing, so we shouldn't put a label on it, but we should Every time we qualify something as bad or good, it's relative to an objective that we had and that is being hindered by this thing and so you could say that relative to this thing, this is not helpful in this local scale, but if [00:26:00] you blew it up to maybe 10, 000 years, maybe it would be Because maybe this is a calling mechanism or something of some kind.

So, like, we don't, we can never say something as blanket as that. We can identify tractors relative to a state we have identified as preferable, right now, from our perspective. But, obviously These things change, so if you're talking about the distinction between, like, or, or the ways in which we are focusing on what constitutes our Markov blanket, I agree.

And I don't think this is a new idea, or that, that we need to focus on the community. I think religions, at least a lot of them, have been saying this for a long time. I'm not going to talk about Judeo Christianism, although, you know, there's interpretations of it in that sense. I'm really talking about, more about Hinduism and Buddhism, and the religions that preceded them.

It was very much about this idea that this, this Markov Blanket is an illusion. You are hyper focusing on what constitutes the ego and you're, you're kind of missing the, the [00:27:00] forest for the tree. And if you understood that this boundary can be relaxed, that you can sort of shed that even in your own mind, which is something that Sandved writes a lot about, about meditative practices.

Yeah. Jonas Magoo as well. You can relax these and effectively your, your feeling of boundary goes away. Maybe effectively your boundary does go away because boundary is a function of measurement to some extent. So, yeah, I think this isn't really a new idea. I think a lot of our scientific. Concepts have been, like, largely influenced by these ideas.

I think it would be I think a little, a lot of hubris of us to say that we've come up with this. In, in effect, like Foucault kind of nailed it. Althusser as well. We never really get away from prior philosophies or ideologies, kind of build on top of them, even if you push them down. So, it's the same thing here.

I think the one is all or all is [00:28:00] one is definitely central to this idea. It's just that now we have very formal math to put on top of it. 

Andrea Hiott: Sure. I mean, even kind of something like what Slavoj Žižek is saying, and the Buddhism could be understood as different ways to a similar place, in a sense. I mean, all of us are, in a way, trying grasp it and also at the same time changing it.

But as you were saying that it, it does make me think that there is a importance too, in understanding that the representation, which I think is what it usually meant by map, isn't the process. I don't know. I think you might push back on that or maybe, maybe you agree, but I do think sometimes we confuse, The representation of all of this, like, the actual maths, or the language, or the image or the marker itself, with the action that that marker is creating in the world.

Do you see any distinction there, or do you still just kind of say it would all be, you know,

I'm not sure I 

Mahault Albarracin: understand the question. I feel like I might have a ninkling, [00:29:00] but I can answer one part and then you'll have to maybe clarify. Yeah, I can explain 

Andrea Hiott: if you want. 

Mahault Albarracin: So, to me, a representation can be a process, the process of representing, and representation is simply a momentary byproduct of this process.

this time of representing, and you could qualify a representation as, let's say, a manifold or a path but it's still a function of this action of representing, and therefore, I don't think they're differentiable. Like, for instance, the self, I know that there's been a lot of theory that, that push in that direction, suggesting that, The self is selfing.

It's, it's, it's, it's not a thing which exists. It's a, it's a process of engaging with reality, and through the boundary, therefore having a sense of what constitutes what you need, and how you can act relative to what you need, and what are the constant [00:30:00] elements, or, or more consistent elements of what you need over time.

And that's why we have a sense that something core to us remains, but over time we change. We can see that nonetheless. You are not the person you were 10 years ago, hopefully. So can you 

Andrea Hiott: clarify your question? Like the ship of Theseus, I bet, you know, all the parts can change, but there's a pattern that remains.

I think I really mean it in a literal sense. I mean, of course, this word representation and neuroscience, we use it one way and philosophy, we fight about it in another way. It's like, but I really mean it as re presenting So let's try to go back into the personal a bit, like, something, you know, about with, with social media or something where there's an image that, that is itself not ever static, but that we might confuse that image with the actual trajectory to which that image is connected, I guess, you know?

Mahault Albarracin: Yeah. Okay. So that's, okay. So there's multiple ways to. tackle this question. In neomaterialism, the idea [00:31:00] would be that everything is about how things interact. So effectively, you should always separate the interacting elements, but at the same time, there is nothing that exists prior to that interaction.

So in that sense, the image is its trajectory. there is nothing that precedes its trajectory because that it's it's not it's not a thing in itself it's a thing which has effects in the world there is another way that you can consider it which is there are multiple paths that can be taken and will sort of find a connection point even if they are not aiming from the same direction and they're not going in the same direction.

In that case, you could say, well, yes, okay, so that point here is essentially meaningless because you could have taken it from that vector or that vector and therefore the image in itself doesn't necessarily contain anything which dictates [00:32:00] the vector you're going to take. So I'm going to, I'm sort of going

caveat that in two ways. The first way is that, yes, you could have taken multiple vectors, but there's a finite number of vectors that go on. It's not exactly true, but let's assume that given some regularities, there's a finite number of vectors relative to all the other vectors that go through that point.

So it is constraining. In any case, it is constraining. For instance, will be perceived, and I know this sounds silly, by people who can perceive it. Which means anybody else who can't probably will not be affected by the image in the same way. So there is a 

Andrea Hiott: regularity here, there is a constant. Sure, I think there's an affordance, an insects not gonna, you know, 

Mahault Albarracin: encounter that the same way.

But the second thing is, there's other regularities. Like, for instance, you could say that This is a pink shirt. This is just a pink shirt. There's nothing in the shirt that says anything more than it's a pink shirt. It's not even really a shirt, [00:33:00] actually. It's just cloth. But I know that it's pink. You know that it's pink.

And most men in North America know that it's pink. They therefore, most likely, will not wear this sort of cut with the shoulders and the pink. Some might, but in regularity will associate this very traditionally with femininity. And you could say, well it's not contained in the shirt. It's like, well it's true.

But the historicity can't really be abstracted from everyone's cognition because they exist in a world where that historicity shaped their cognition.

it's not quite true that the images should be separate from their trajectory because part of their trajectory was already defined by the, by the, by the patterns that we gave them. So I, 

Andrea Hiott: I don't know, I hope that answered the question to some extent. I mean, it's, it's more just a topic I try to explore, but yes, I mean, I love what you're saying.

It's just, I guess my, I think there's something different about something like a [00:34:00] language or image because. of the intersection of all the different trajectories in this kind of way that isn't connected to a body one body. It's, so like, as you described your shirt, that's, that's a very interesting, and of course it depends on all of our, as humans, all of our trajectories of how we've been taught what pink is as kids and like, you can't separate it from every moment of our life, exactly where we live and so on.

And of course in different geographical or emotional or conceptual spaces we'll have different understandings or experiences of, of that shirt or a doll or whatever we want to talk about. But I think we, something happens in society all the time, I don't know if you agree that we assume. That we're all experiencing that shirt.

Your shirt is now the object of, of this conversation. But we all experienced that shirt and we think everyone's experienced in the same way we are. And so what I'm trying to do is like, I don't feel like we pull it apart enough to understand that [00:35:00] yes, it's connected to trajectories, but not one trajectory and that all the time, everything that we're encountering is being processed differently according to each spatiotermal point.

So there's a sort of system three, I like to think of it, you know, instead of fast and slow within the body, it's more of a community, you know, Daniel Kahneman kind of processing that's going on. And I just, I feel like if we could put some focus on it like that, It could be helpful, 

Mahault Albarracin: absolutely, and I think, okay, so, let's talk about maybe a topic that's a little bit, it shouldn't be, but it is, controversial, and explain how that factors into the question. What has the queer community tried to do by calling themselves queer? They've tried to take a statistical pattern and sort of nudge it away from where people want it.

So, okay. This isn't, saying that there are statistical patterns doesn't mean they're set in stone. It just [00:36:00] means, when you take a population, they're likely to be on some distribution to think in a certain way. But you can push the distribution somewhere else. And that's why I like this statistical manifold, because it doesn't really first of all, it doesn't say that you and I will be on the same spot.

It says we are likely to be within a range, most likely, but we might be in different points on the range. And then it says, well, okay, if this constitutes an attractor, and I know how much What factors contributes to it being an attractor here. I can use these same factors and push them over here. And I think that's, that's kind of brilliant and beautiful and empowering because now we know, well we can just re appropriate a bunch of things.

I can decide that, Pink is not a feminine thing, in fact, it's not, it doesn't represent that at all. Now the problem is I don't control every factor, like, I'm still gonna get perceived as a [00:37:00] woman, and so, well, by virtue of association with a woman, okay, but I can make or ask people who engage with me if they are not in considering themselves women to wear pink and I enjoy that and it makes me happy and if they want to contribute to that they can and lo and behold most of my boyfriends have worn pink like it's it's no longer associated with femininity to them it's just associated with happy and joyful and cool I 

Andrea Hiott: was about to say usually pretty cool guys wear pink these days right exactly yeah but it's that actually speaks to something really fascinating too is that how these prior uh statistical regularities Can shift according to something like awareness.

It's another topic I wanted to ask you about a little bit is like, do you see a difference between this cognitive and the kind of conscious. Or being, or awareness, or attention. Because, you were talking about Buddhism and all of this, and I think, in some way, if you start in that tradition, you're aware of everything as one, and the whole, and so on.[00:38:00] 

And there's some kind of way in which the dichotomy side of things, the, the reductionist side of things is kind of, I don't know, not as familiar. And then, you know, vice versa if we want to give these big kind of awful generalizations of different religions or parts of the country where you start with a more reductionist and the idea of thinking of everything as whole is different because you're, it's about the individual and and all of this.

And it seems like the thing that changes that is what you're just describing with in terms of The shift, like we want to shift, pink is gonna be cool, and, you know, who knows in 200 years maybe only guys wear pink because it's shifted all the way to there, but what does that? Is it something like awareness?

And I want to try to link this, you know, to the active inference work, but I wonder how you see that, like, consciousness, cognition, intelligence, are those different? Where's the attention in that for you? 

Mahault Albarracin: Oof. 

Andrea Hiott: That's a rough one I know. 

Mahault Albarracin: No, I mean, I, in general, consciousness is broken down [00:39:00] awareness is just one part of it, and I think that's probably correct in the sense that sentience doesn't necessitate awareness per se, I don't think.

think, I think you could have, well, I mean, awareness in the sense of an integrated perception, an integrated qualitative perception is more what I'm going to say. Yeah, 

Andrea Hiott: and I should have, I'm sorry, I don't mean consciousness the way it's always talked about, where it's like you're aware of yourself. I really just mean the feeling of being alive or being alive, being itself.

And that can of course, you know, become Being aware of your own self as a self and all that. 

Mahault Albarracin: Yeah, I mean, see this is an interesting question because it gets to the core of what constitutes qualia, right? Like, is, is qualia only This, right now, what we think we're experiencing together, or does it go much deeper and are yourselves, to some extent, conscious and have some degree of [00:40:00] qualia by virtue of being able to react to information in a way that Like pushes entropy away, because, you know, lots of things react, like, potentially if you kick a rock, it moves, like, it has, you, it has had a reaction to what you did, it's just not a reaction that effectively felt like an action.

It didn't feel like it had control over where it intended to go next. Relative to where it wants to go. And there's a lot of words in there, right? Like, want to go. Like, the notion of wanting seems to be central to what we consider to be life or consciousness. So if, if these are more what we consider to be consciousness along Mark Solms, maybe, then a lot of things might be more conscious than we think, and they may not require what we understand to be awareness, but if we break down awareness to its more, most central component, i.

e. There is some information that is encoded and acted upon, [00:41:00] or at least gained more precision. I feel like that's, that's closer to a panpsychist even conception of what constitutes consciousness. It feels like something's missing though. Like, if, if awareness is all that's required, then what happens when you meditate and you become sort of, Flatly aware.

Are you not conscious anymore? Is it, is there no precision here? What, what constitutes like the, the very nature of a color or something? Is it, is it that you break reality apart somehow and you encode it such that it has differentiability? Is does it, is it, is that required? I don't really have the answers to this question.

No. Of 

Andrea Hiott: course, 

Mahault Albarracin: we're aiming in a direction. Mm-Hmm. . But obviously this isn't solved. 

Andrea Hiott: Well, so, yeah, I guess, like, you said something earlier about it not, nothing being set in stone, and I kind of think sometimes we do think of representations as set in stone, [00:42:00] like, the color, you know, is, is what it is, or the math, even, when we, when we represent something, we just assume it's static, and in a way it is static in this sense.

weird universe that we've built of language, right? It does move, of course, language changes, words change, but in that kind of, if you just take a very tiny slice of it, it seems static relative to everything else. And I guess, you know, when you talk about qualia or something that's digging into that a bit of I don't know that I really think there is qualia, actually, like, I think, what if we just go back down to the very basics and think of something, we're always measuring or assessing, so this is a measurement or an assessment already, but we do that, the body does that with itself, right, and that could be something like qualia, or that could even be something like thinking and that's kind of different than me verbalizing that in language, but the action itself, it feels to me, action of an agent moving through its environment in an active inference sort of way, [00:43:00] adjusting You know, all the stuff that Active Inputs describes, I don't see how that doesn't also apply to something like me as an agent who now has language, considering myself as an agent, and thinking of my own thoughts.

To me, it seems like the same process. 

Mahault Albarracin: I mean, I agree. Effectively, everything is recursive, right? Like, I know that Maxwell really likes this idea of Oh, crap. You know when you have these figures that are recursive? Fractals? Fractals, thank you. I love fractals. Yeah, so he's passionate about them, and to him that's how he can perceive time and reality and space.

And I think I agree with this. It's all quite recursive, so the notion of, Saying words and hearing them back if only for yourself, but also from others and getting pinked back. This is very fractal pattern Where we're gonna get more more precise So I think there's something to that nature and if we go back to your initial question Which was how does awareness and consciousness factor in being able to shift societal patterns?[00:44:00] 

I think it gets to an interesting question of whether or not A group can be conscious, as opposed to just individual units within the group are conscious, and then they produce patterns, emergent patterns. And I think Mike Levin also gets to these questions as to whether intelligence, and therefore to some extent potentially sentience, can happen In the relationship between nodes, and I agree with that, that's what you said earlier, right, that intelligence happens within communities, even if it's communities of cell that exchange information here I think that there's a possibility that what we're experiencing is Just not fully being at the level of the sentient layer that is above us, potentially, but we are getting some messages from it, some signals from it, which constitutes culture.

And so what we're doing at this level is telling it, whoa, this doesn't work for me. Oh my God, move, move, move. And, and when this pattern agglomerates enough, when there's enough of a [00:45:00] signal, it gets passed upward and potentially. That's when it starts shifting, like an Ising model type thing. So, I could totally see how by being the only way for you to move things somewhere else is by being able to measure what the thing is now, right?

So yeah, I do believe that in this sense, there's a degree to which awareness factors in our capacity to move things, because they think awareness or.

Focus gives you the capacity to add precision and determine how far away from your desired or preferred state. You might be. 

Andrea Hiott: Yeah. You mentioned Michael Levin and we could think of it as kind of as, as ways where we're navigating like intelligence, we could think of it as the way that life or being or whatever we want to call it is making its way or navigating whatever the encounter is that, and it could be a different landscapes.

It could be in the geographical world, you know, as an insect doing whatever, or a [00:46:00] human, it could be in conceptual space. It could be in linguistic space. Morphogenic, all these spaces, right? That, that we are setting as we're measuring. And I guess something like that, what you just described, which is a very important issue it's kind of in the human social space and that space perhaps changes through communication or, you know, you talk about potentiations and it's a change in potentiations, I guess, or potentions.

Am I saying it wrong? The Herschel, the Herschel phenomenological idea of like, Opportunity. Yeah, right. Thanks. So, in a way, when something changes in society that's changed in the laws or something, such as the way women are treated or what our rights are, it, what has changed there? Is it that, In this navigating space, certain spatiotemporal agents have expressed through the system three languages or whatever, what's happening, and that it's wrong, and somehow spread that [00:47:00] through, through the group, so that the group itself understands that it's wrong or I mean, this is all very hard and awkward, but I just like, you know, how do we get an idea of what's, what's happening there in terms of something like changing, minimizing surprise or changing the priors or. So to me, it's akin 

Mahault Albarracin: to structure learning where you're essentially carving different ways that you can have observations mapped to states or different ways that just states in general are carved. And then you do Bayesian model selection, where you determine that a model actually works better for you to minimize free energy.

So I think that's, that's what it is when you add a new state, for instance, like when you're when you're saying there's a new concept, here you go, you're now proposing that there may be a better way to measure reality, and therefore It's coming out of 

Andrea Hiott: something, right? Like, it's a script to You're changing the script.

So it's not that you just proposed a new concept. It's that you've [00:48:00] together shifted the script's regularities, right? 

Mahault Albarracin: It's embodied, exactly. It's embodied. None of this is fully one or the other. It's completely connected. And this gets back to the historicity, right? Like, I There has been historically groups that feel a lot of error, And eventually, it gets to the point where they, they have to act on it, they have to do something about it.

And because some slight, subtle shifts happen in other places, they get more and more and more of an opportunity to have a policy that is going to actually phase shift them away from the system. The, the prior situation they were in. Unfortunately, those phase shifts are never as clear cut as we would like them to be.

It seems like anytime you really integrate a new concept, a new a new script, let's say it does constitute a phase shift, but there are still things left over from the past. So we can see it in the historicity of the U S for instance, right now. A lot of people saying, Oh, no, no, this thing is over. It has been over for 200 years.

And the rest of the people are like. [00:49:00] Maybe some concept seems to have changed, but there's still a lot of artifacts from the past that have followed us, regardless of this ability to break the representation in different ways. So it's not, I don't think it's as simple as we change the representation and the embodiment will change.

It's more like, there's errors, you change your model of the world, and thus you will now have also probably to change your actions because, With a new model, it may be possible that the prior policies you were on no longer fit. And so now you have an uphill battle to get the rest of the system to follow you to actually changing this policy.

So, I think, I think this is what we can observe. And I think it gives a lot of credence to this idea that social groups in general act like their own mark of blanket, like their own agents to some extent. 

Andrea Hiott: Yeah, sure, it's again where you're setting the scale of the Markov blanket, but that's very fascinating too.

Think about in terms of the role that [00:50:00] technology plays in that change so something like one of these movements where people are treated wrong and, and clearly, you know, seen in a way that's not accurate. And they have to change that by coming together and you know, shifting the regularities more or less often with, you know, Something like, you know, force whether through language or image or whatever.

Often technology plays a really big role in that. I mean, because as you were talking, we have to remember all that's developmental and there's no gap. It's all coming from many trajectories joined together. And technology has been a way that people, like the first images of, you know, when people saw it and they knew it was real or something, or radio, TV.

Now, you know, we can go all the way through the internet into something like AI. So, two things as we kind of wind down is I wonder how you see the role of Of technology in this and you know, it does relate back to your own work and where you started and where you are now you're now using technology in a way to maybe help [00:51:00] open up some of this community change space, or I don't know, maybe that's too idealistic, but I would just wonder what you think about that.

And yeah, I'll just leave it there for now, but 

Mahault Albarracin: so this is a sort of grandiose idea. But if we understand consciousness as the capacity to integrate information such that it is encoded on a form of boundary and that boundary is sort of aimed at potentially maintaining the agent that has the boundary.

And you consider AI to be anchored within a system. I think what it says is that there's a possibility to make more manifests, this integration of information through what AI can do, which is basically integrate information and synthesizes to do something useful, which is what cognition does, right?

That's what we do. We're trying to Unfortunately, currently, we're getting to a point where We can only integrate certain kinds of information, [00:52:00] and when we do integrate it, it can only do a certain kind of thing. Like, it doesn't do everything. It does one thing very well, or it does another thing very well, but not both at once, and it can't generalize from one to the other.

We're trying to get towards that. That was what people thought they saw in Chat GPT. They thought they saw a generalist. Turns out it's not exactly a generalist. It's more like a really, really, really, really good language machine that, because language is structured and generally what you feed into it is Structured patterns of language.

You don't just feed it garbage. It tends to have encapsulated some of these regularities, but it hasn't encapsulated the the understanding, the causal, the causal layer that allows you to understand that because thing is this way, another thing is this way. That's not quite what Chat has or GPT in general. None of the others do either LLMs in general.

We are getting towards a point where we might, but the point is, it may become this sort of, you could say conscious, you may not want to say [00:53:00] conscious, you could say overarching layer that integrates information. And it, that's kind of beautiful, but at the same time, a little scary. Because I think I know who I am.

I think I know what groups generally think I belong to them. I don't know what boundaries. are encapsulated within the system that is bounded by a given AI system. So when you, when you, when you're sick, when something goes wrong with your body, your body fights it. It tries to get rid of it. Because obviously right, you wanna maintain irregularities, , I think everybody is sort of seeing AI in this way and see, and that's why you saw the emergence of grok, which was a non woke ai because people were like, well, I want to be the thing which is represented in that boundary.

I don't want something else to be represented. People are very afraid of biases in AI because they're like, well, if AI is going to be this [00:54:00] integration layer and it thinks something is good and I don't constitute what is good, it's going to discriminate against me. And we've seen that. We've seen AI discriminating against sub communities because the historical data chosen had this pattern of discrimination that identified these groups as groups that were not part of what is represented in the AI. So 

Andrea Hiott: that's I think that's clear sometimes that the LLMs at least are trained on. And it's just like we were talking about with a life, you know, as, as we develop, we're sort of, everyone says stuff to us and they tell us pink is for girls. So that's what we spit out unless we see another option or someone, you know, breaks that open and, but with the LLM, it can't learn anymore.

So it's just whatever you've put into it. It's just going to stick on that. So, 

Mahault Albarracin: well, I mean, there are our LHF patterns. So, you know, we, we, we could, the problem is we haven't figured out how to do it well. So the continual learning is not exactly continual. And even kind of what active inference AI is sort of [00:55:00] trying to do, right.

Right. So that's the thing. So active inference is sort of, It's supposed to be non static by, by very nature. And so you, you would constantly have a system that adapts to its surroundings. But I think this notion of a biased AI needs to be a little bit flipped on its head. Maybe what we don't need is an unbiased AI.

Maybe what we need is a lot of very locally biased AIs that just focus on this perspective that understands it has a perspective and it's going to adapt to it. Crouched in that such that we would know what to use at a given time. 

Andrea Hiott: That's a wonderful idea and so dangerous and so wonderful because it really matters then that everyone understands that they're biased, that the AI is biased.

So that's, again, that awareness thing of how do we, because that, that's actually really smart what you said. I mean, that, that could work if everyone understood that it's biased. It's kind of like with the social media where we, people don't understand that they're in a feed that's giving them. [00:56:00] An orientation of the world.

But if we understood that we could even play with it and look at all the orientation. So there's always a scaling too of awareness, I guess, to go back to that. 

Mahault Albarracin: No, absolutely. I think it is. I think the key question is who gets to define the bias that we're going to adopt for a given situation. And that's not a solved question.

That's where we get into ethics. And again, I think technology is starting to wake up to the fact that they need ethics in order to do anything because their power can no longer go unchecked. Like lots of companies have ethics boards now. So these tools are coming. I hope they come fast enough. 

Andrea Hiott: Yeah.

And there's a transparency in active inference AI, at least that's trying, they're trying to build that in your, you know, so it's, people are aware of it, it reminds me of like the priests who used to only be able to once to read so they could sort of dictate what the Bible said and then everyone learned to read and it sort of changed.

I mean, maybe we're there with AI in a weird way and who knows, but you know, just to, to kind of end it and come back to your own. You know, we started with, with your, your [00:57:00] story. And I wonder what those problems that you started with and ideas and questions, have you found, you know, in learning to code and going through all this really hard work?

I mean, this is not easy to digest all the information you digest and like what, how has that changed your own trajectories and paths kind of in the scope of what we've been talking about or has it? I don't know. 

Mahault Albarracin: So. Initially, I started with a very deep love for postmodernism, and then I had to exist in different circles that were very much not postmodernists.

And it, it changes your perspective on things, it makes you qualify things differently, and now I'm sort of going back over the pendulum where I'm like, hmm, this qualification of these things I used to like, had a purpose, they were trying to do something, which was postmodernism. to push back on the possibility of critique and push back on the possibility of control of what they're doing because they want to maintain a certain order of things.

And so now I'm [00:58:00] sort of going back over these things and having perhaps more nuanced perspective over my previous leanings but I'm still finding that these tools are useful like Foucault constructivism. These are very very useful frameworks to understand what's going on and how we might be able to counter it such as Educate everybody.

Just give everybody the tools to understand what's going on, and use them as well. Don't just let it be in the hands of a very select few, who then get to decide without your input, and therefore, maybe you'll be on the lucky side, but maybe you won't be. 

Andrea Hiott: Does that have to do with self esteem too, another topic that you've written about?

Does that, does, does all that stuff you just said, having that and bringing that back in, do something for the phenomenology of the spatio temporal sort of awareness within the Markov blanket in terms of empowering or? Giving some Yeah, I think, 

Mahault Albarracin: I think self esteem is, is just a tool, a tool that enables you to either try to fit within the [00:59:00] group if you find benefit to it, or flip against the group if you find that this isn't working for you.

And there are obviously psychological factors that either keep you within a low self esteem or allow you to try to flip such that you can now grow your self esteem relative to another group. I think That's why it's important for certain movements to say X is beautiful, because it doesn't matter whether they believe it or not, it doesn't matter whether others believe it or not, what it is is a social tool to help them grow their self esteem, to create enough momentum that they can now be represented in something where their interests and beliefs are forefront.

Andrea Hiott: All right, that's good. There'd be more questions, but it's over the hour is over for now. And that was that was a fun ride. So thanks so much. I really appreciate your perspective and the work you're doing. 

Mahault Albarracin: Well, thank you for reading it and chatting with me. It was great. All right. Be well. [01:00:00]