Simulating the World & Remodeling Philosophy (Session 4)

Hello and welcome to the fourth session of Simulating the World and Remodeling Philosophy with Reza Gharsani. I'm going to pass the mic to you. Thank you very much, Theo. Thank you everyone and hello. My apologies. The connection might be a little bit bad because if I go upstairs, which is closer to the router, there will be huge amount of noise coming from our neighbor. So, we'll just take it easy today. I will go over different stages in modeling and also talk more about constructs, the topic that I mentioned last session.

But before that, before we actually hear about the responses, there's people who volunteer to give responses. Let's hear if you have any questions. Anything, anyone? Jovan, Justin I just came back from the bar with Mari and she's skipping this class to go on a date

so I'm like snitching Don't worry, that's good Who wants to listen to a philosophy class on Sunday I just it's just beyond me I mean don't you people have life go do something else but anyway ask ask questions no one Sean so this isn't necessarily a question about the modeling stuff but it's kind about what you were just talking about with rationality and reason. With you, yeah. One question I would have, and it's something I've been

trying to figure out, is how one would define the difference between rationality and reason. Yeah. You see, I think the idea is quite actually confusing, even in the history of rationalism. But I would say that some good attempts have been made. And you can think about it in terms of this unsatisfactory, ultimately, but nevertheless, helpful definition. Essentially, when we are talking about reason as capital R, we are talking about reason first and foremost. And by reasoning, we mean certain kinds of things you ought to do in order to say something,

whatever it might be, right? Whatever it might be. It might be wrong. It might be true, so on and so forth. And then there's certain kinds of doings that you have to accomplish, you have to do, in order to be capable of assessing what you have said. Essentially, this is already in fact in Kant. So there is such a thing is that, you know, when we have a perception of, for example, a red balloon, of course, that perception from a Kantian perspective is not a sensor. It's not just a bunch of a bundle of sensations. It is actually a piece of judgment about the qualities of this balloon.

So, in a sense, you might say that to make a piece of reasoning in a rudimentary perceptual sphere, that piece of reasoning is more like not thinking aloud, but experiencing aloud. In fact, you need to have certain kind of conditions or constraints to be capable of positive. On a model of a public language, your private experience, and that's when it becomes actual perception, a perception as such.

And of course, for it to have a perception, it needs to have an epistemic status, a certain epistemic status. And this is really what you might call to be, I would say, the Kantian definition of reasoning as a faculty. in a sense that for us to be capable of doing something or saying something, such things ought to have a certain publicly modeled on a public language epistemic status. And we need to, even though we implicitly follow them, we need to have some sort of semblance of familiarity with rules of language.

of an influence in order to not only to say what we want to say but also capable of assessing and in fact being assessed by another agent. So this is reasoning as a faculty. Now the thing is that for example with Brandome and Hegel you see that this definition of reasoning already gives a certain kind of status to an aperceptive rational agent, in the sense that the agent becomes the author and responsible for what it says and what it does. And essentially rationality, or reason with capital R, then become the enterprise, the concrete rather

and abstract, the concrete enterprise of making explicit the kind of commitments and the kind of consequences which your thoughts and actions have. This is, you know, a kind of, I would say, a minimalist definition of reasoning as a faculty and reason as a concrete sphere. That's super helpful, thank you. Absolutely. Any more questions?

Artemis, you don't need to come for another session. I was just mentioning that we had Sam's lesson earlier. I need to, I opened the text because I have some notes. I need to scan it again to remember my questions so we can. Sure, sure. But any please, any person who has any kind of Adam. I have one actually. Please. Can you hear me? Yes. Yes. So I'm thinking about what happens with prototypes, because prototypes of objects that does not exist yet, I mean, there is not a target system.

So my question is, does that qualify as a model? Yes, yes, absolutely. In fact, you see, so far we have introduced models as what you might call to be particular representations of a specific phenomenon in the world. Now, this essentially, so I will elaborate this point today, that there are three stages in any kind of modeling, right? And only the last stage is about the relation between a model and a target system. But in fact, there are many, many models in the history of science which don't deal with

this third stage, namely the relation between a model and a real phenomenon in the world. But only halt at the second stage in which it's just we are talking about model, and model can actually be understood as a kind of almost a counterfactual. What would have been the case if this was X? For example, majority of times when modelers, you know, devise models, they don't even actually think about what is going on in the actual world. That is only a specific stage of modeling. Sometimes, for example, for the study of dimorphism or sexual binarity,

people, modelers, need to come up with this idea of a three-sex. Or, for example, Eddington, a British physicist, came up with this idea that uh in order for us in fact to talk about actual phenomenon we should be capable of modeling what might not be actual and this is what you might call to be a kind of model in many relations to itself rather than to the world and okay prototypes in fact are part of these kinds of models yes prototypes in the sense that you don't have any kind of an actual example for it in the real world

yet yes it's something is not just for understanding something but in order to create something no yes yes and not only understanding it's not yes it's not about understanding something but is understanding what thing could have been. Okay. It's about essentially a space of possibilities, both in terms of creation and understanding. Okay. Thanks. Absolutely. Okay. Okay. So let's hear the responses. I can start. I can start. Although I actually didn't prepare a proper response. I mean,

there is something that is not really clear, obviously. So I'll just try to give a sort of brief summary of that chapter. yeah basically yeah the models for Westonburg are a combination of structure so the computational mathematical or concrete structure and interpretation with this interpretation is provided by I believe a human and non-human agents in the sense that even this is actually really a question even like for instance I don't know artificial intelligence can provide an interpretation and therefore

construct and structure with a model but yeah yeah hope this will be yeah discussed later and so yeah The interpretation is basically consisting of modelers' intentions and basically the modelers' intentions is this construal. And again, I didn't actually look up for the word construals because I'm an English native It simply means, you see, it just simply means that, okay, so we have a model structure, which is like the overarching structure of model, which is derived from a theoretical

edifice in which the model has been couched, fabricated. And then, of course, this structure is way too broad for us to model a specific phenomenon. So essentially the modeler tries to interpret certain aspects of these structures and not others. So these are essentially the constructs. According to the modeler's intention, certain aspects of the structures are being interpreted and incorporated within the model and not others. So yeah, I mean basically it is said there are some assumptions embedded in the aspects

and these assumptions are given by the controls, right? Absolutely, yes. So yeah, basically, yeah, I mean every model obviously it has some assumptions and my other question would be, I would actually link to the thing we talked about in the last session about the how scientists think about their work in the sense that how conscious they are about all the assumptions they put in the sense of in the sort of like bias they put in their right but yeah we'll come later and yeah therefore um um so yeah basically the the controls are also tries to set ups of relations of denotation between the real world and the model and there are four main

criteria that tries to evaluate not only the scope the assignment but also the fitness and yeah the fitness and their the fitness of that model and i believe that there are two kinds of fidelity criteria in this sense and one corresponds to the representation of feed representational fidelity criteria which basically consists in uh how how how how much um how much the model can map the phenomena that is trying to represent and there is also the the dynamical fidelity criteria which is um uh yeah uh i i believe yeah it's it's it's like um

i i think but again this is i don't really remember if this is correct it's sort of of how much an output of a model can resemble the real phenomena. And so there is a dynamical relation between the output, the outcomes of the model and the real phenomena. Yes, yes. And of course the whole point of Weisberg is that any model by definition requires degrees of simplification. simplification and or idealization. In the sense that, for example, a good example of this, you see an orary, which is just like a kind of like a concrete model

of the movement of celestial bodies. Doesn't matter whether it is brown or black, made of metal or plastic. You see, some of these features are fundamentally abstracted because they are not relevant. But even some relevant features are being abstracted precisely because they are not part of this model. And these are basically defined under the terms of constructs. As for your second question, you see, it is really hard to say, as I mentioned last time, to talk about whether scientists are aware of the assumptions, of such assumptions, or

not. Well, is it their task? Again, hard to say. It's probably not their task, in fact. But a good scientist always tries to, in fact, bring out the assumptions. And that's how theory change happens, a change in the paradigm of a certain, from one theory to another. But nevertheless, what is really, so where does this task come from? Being conscious of the assumptions and being capable of tracing them back to the origin. Well, this is a task of both historical and the community of scientists, not an individual scientist.

And by community of scientists, I also mean, definitely mean, not only just the ones who are working in the lab, scientists who make theories, but also those philosophers of science who go over such assumptions and kind of try to triangulate what is actually going in the lab, theory construction, so on and so forth. But of course there are so many scientists in the history of science, individual scientists, who actually started to challenge their own assumptions. Why is it that I am thinking like that instead of the other way? But I don't think that this paradigm of the kind of enlightenment scientist is no longer

out there really. Yeah, the last thing, as you said, I mean, obviously, Weisberg is, I mean, tries to give to give different degrees of representation for for modeling I mean obviously there are several scales of representing a certain phenomena what I was thinking probably I can I can also ask I mean the defect of low approximation and reduction no I believe that one career said something about approximation and I think you with all no new Clidian and what's the word I was

English word mathematician thinkers yeah I mean you think this this I mean do you think that Poincare the non-nuclear is quite like the first that the first theories that like understood that you can like approximate everything in the real world and obviously models can have not only a representational but also production of another. Yes. I don't think that Poincaré is in fact the first one. Yes, he surely systematized this. And let me just get this.

And yes, those of you who want to, these two books are absolutely fantastic. I really do suggest to you reading them. One is called Henri Poincaré, Science and Hypothesis. And there is the second volume, which is called Science and Method. These are really fantastic books, and I highly suggest them. Extremely lucidly written. There is no math to be as scared of. They are quite excellent. But I don't think that Poincaré actually came up with this idea of hypothesization. This is actually goes back to a time of what you might call

to be, I would say, Francis Bacon and the rise of empirical naturalism, a kind of hypothesization. Well, of course, at the beginning of this kind of Renaissance science that took hold in the course of centuries, there was no systematization. And Poincaré not only managed to systematize it, but also invent a certain kind of mathematical systems. Mathematics, during his time, it was called ugly mathematics. because it wasn't really an axiomatized system.

Now it is actually being recognized as the mathematics of chaos, or dynamic systems. So he not only systematized this as part of the idea that we are not simply representing reality when we are doing science, but we actually enrich our concept of reality in a very specific way. Not only did he that, but also he developed a kind of required mathematical or a structural platform for us to be capable of looking at these phenomenon in an entirely new vein,

which is what you might call to be, it's a new vein precisely because it's not built upon classical concepts of temporality, you know, first principle, so on and so forth. yeah and yeah I mean I'm always I've been wondering if this as you said enriching the reality as reaching the reality can be can consist in sort of adding new dimensions to reality obviously like projecting new dimension necessary of a project in the future yeah yes

Yes, yes, absolutely. And you would say that, of course, this kind of way of thinking is quite actually politically relevant as well. Yeah. In the sense that all the stuff about past determination on the future or teleological worldview are all being basically eroded. And that actually, this kind of philosophical scientific vision, I would say, enables us to think about different kinds of worlds, you know, in a very political Marxian way.

And yes, absolutely, I absolutely think that Poincaré, so as Boltzmann, set in motion a kind of new ways of doing science that potentially can erode away and fundamentally eliminate what you might call to be the ideological subsumption of science by political ideology. I mean, Boltzmann is another great example. You know, you have seen this quite often, I mean, among the right-wingers, left-wingers,

so on and so forth. It's a kind of a banal teleological interpretation of the second law of thermodynamics, either in the name of extinction, either in the name of acceleration of irreversible processes, so on and so forth. But literally, at the end of his life, Boltzmann started to question all of these. in fact showed that these are fundamentally human or anthropological biases, the interpretation. From a certain perspective, Boltzmann never actually managed to bridge the gap between

a statistical thermodynamics, which is fundamentally a modern and respectable branch of physics, with the observable thermodynamic principles, which are basically this idea of irreversibility, time arrow, determination of past upon the future, so on and so forth. The kind of soap opera of interpretations that you see in the idea of history as a kind of a thermal thermodynamic unfolding of events. So next one, by the way, Theo, I don't know,

is there any way to put a picture on the sidebar or share it other than... I think you can, you might be able to drop one in. There's like the little triangle. Oh, yeah. She's links. Oh, okay. If that doesn't work, maybe if you wanted to screen share. Okay, I did it. Okay. Okay, so the next volunteer. That would be me, but I pretty much just sort of summarized the entire chapter,

and Andrea kind of also went through that, so I don't know if you wanted me to sort of like ask questions about it. Sure, ask questions, yeah, sure. I was thinking a lot about this sort of like and I think in the sidebar Adam wrote about cybernetic effects and how the certain levels of tolerance you know plus minus like when you're looking at like say an analogy that comes to my mind would be something like sort of like a closed circuit electronic kind of device that has like very like tolerances like when you when you insert a resistor into into an into what do you what do you even

call it I work with electronics and I don't even know what this is what this is called but uh-huh so so into so so say you have like a feedback loop and you're inserting like resistance, a resistor into that, you have tolerances plus minus 10% either way. And you can use sort of go down the line of phenomenal, of a phenomena in terms of, you know, I know that one of the examples they used is like of the hyena is so, I don't know. that would cause a predator to either initiate or terminate predation.

And so one of the things that I'm interested in wondering is, like, these tolerances, how do you accurately measure them when there are so many factors that go into sort of like evaluating the criteria of like how a model reaches its own target, you know? Yes. Well, the question to that would be exactly, you see, there is this fantastic book by Mark Wilson. It's called Physics Avoidance. Okay. What does the word physics avoidance mean? Physics avoidance is really the fun topic of all engineers. When we are dealing with any kind of mundane object, we just don't want to know much.

It's literally, we don't want to know much. It's a computational hazard. It's bad for brain. Yeah, no idea. So the whole point is that, in fact, all models try to abstract certain features that are only relevant for the kind of aspects that they are trying to deal with. Absolutely, you do not want to have too much complexity. Yeah. Too much complexity increases the size of your model and becomes a computational hazard. Yeah. So, yeah, you pretty much sort of answered my question there, which was kind of actually implied in the chapter itself when it goes through, you know, the actual model.

They use it as an example. And so to just assume, you know, that there are all of these things within two of those criteria, is like the target and scope that weren't necessarily represented like predation or um that in on other things within this within the scope of uh dynamical fidelity that were not talked about they're like yeah we just don't need to know that stuff it's not it's not actually important if you need to know if you needed to know you can yes you can you can you can supplement them but you don't need in the first thing which is which i kind of thought was uh interesting right yes yeah that that basically kind of covers it it was pretty much

self-explanatory of course when we come to the so from hopefully from next session we start to talk about toy models yeah our toy models are a class of formal models uh and they come in a small and big varieties a small model is essentially what you might call to be you know getting rid of again details simplification yeah big toy models however you can plug as you go different supplementary details that you didn't have your initial models and we will talk about what kind of models these big toy models are and how they function yes there are different kinds of models but a model in a kind of a is what you mean by model as a as a kind of a indirect abstraction

of world phenomena absolutely always requires as i mentioned simplification and or idealization there is no way to make any model without that it would be it's essentially you see this comes back to the idea that when even when we are talking about you know a red dot flying in my visual field right you know a common sense assertion well of course the red dot can be fuzzy at this point time t1 the fuzzy texture can change at time t2, so on and so forth.

But why is it I come with such an assertion? Precisely because it compresses, qualitatively compresses, and I allow it to be assessed at a certain scale. Of course, if I want to get more involved with the details, I don't change my assertion. I don't change my model. I just simply add different contexts to what I have just said, to my piece of judgment. So in a sense, model also like this is what you might call to be made in congruence with the simple judgments that we make about the world.

Precisely because our judgments about the world, both from an evolutionary perspective, a computational perspective, and the perspective of logical judgments, these are optimal in their structure. They are quite sufficiently simplified to allow for tracking and further complexification of the model. it would have a start from a complex viewpoint then it would be extremely extremely difficult to revise the model yeah that makes sense I had I had another

question but I'm sort of kind of jogging my memory for it at the moment I didn't write that down. I'm trying to remember. I just had it in my head. You might have to come back to me on that one. Sure. Don't worry. Don't worry. Don't worry. Take your time. Okay. Any more questions? Lenka, Mikey? Oh, yeah, yeah, yeah. Okay. Anyone?

Well, one of the – I do actually have a question. It's like – so when we talk about this sort of like simplification of these sort of unnecessarily details, it's like the ways in which humanity we're constantly sort of negotiating our position within subjectivity like wouldn't that be dynamical in and of itself you know so like that's sort of one of the subjectivity within the world humanity yeah yeah yeah gains its subjectivity by renegotiating its position in the world. That's an individuating process. Yes, absolutely, and it is quite a dynamic phenomenon in the sense that there is no

specific end in sight and you might have anticipations of how things will move forward but anticipations if you want to think about them more close to a scientific way of understanding any dynamic phenomenon, anticipations are only relevant within a certain abstract trajectory, within a threshold. Only within the threshold that you can have anticipation. Otherwise, all anticipations ultimately will be doomed. They will not eventuate. Yes. Yeah, absolutely. This is essentially a Hegelic perspective. Hegel is in fact one of the first people, one of the first philosophers, who sees this kind of historical unfolding

of the idea of the human, and by idea Hegel means a very different kind of thing than the common sense word idea. It's essentially what you might call to be the actualization of the concept of human. He sees it in dynamic terms, precisely in a very, you know, way that we were talking about Juan Carrezz analysis of dynamic systems ugly mathematics mathematics of chaos so on so forth can I ask a follow-up question um is it a skeptical question it is kind of skeptical but I I'm actually just kind of curious how this is technically done,

how you would go about determining which information can be excluded from your model, because it seems like it's only something that's easy to do if the model has a very specific task, right? Yes, and of course, all models have specific tasks. Essentially, that's what is the difference between a model and a theory, as we talked about. Right, I see. So, essentially, there are what you might call to be falling under what Karna would have classified as descriptive pragmatics.

You are describing a phenomenon in pragmatic terms, in terms of use. And the whole point of models is for engineers, for example, they need to get the job done. Right. We don't need to get into the nitty-gritty theoretical details, because the intention of an engineer, modeler, is simply to get the job done. That's it. And of course, you see, it's not that always a modeler tries to sacrifice details for the

sake of modeling idealization. And there are in fact, as I mentioned previously in last sessions, there are great amounts of tools in the arsenal of a modeler and an engineer which allow them to approximate great amount of details at a lower scale, translate them to less amount of details on a top-race scale or level. These are called normalization techniques or approximation techniques. There there are quite a lot of them they do work

okay yeah i think that kind of answers my question for the meantime okay excellent okay shall i start then OK. And by the way, look at that diagram that I shared on the sidebar. So we are going over this. So I said in the previous session that one of the most important insights behind the semantic view and other attempts to reconstruct theories as sets of models

is that the theory should not depend on a particular linguistic formulation. More importantly, for understanding the practice of modeling, a modeler often conceives of a model in a vague way, writes down some equations to describe the model she thought she had in mind, studies the model actually specified by the equations and determines whether or not they pick out the right model. Situations of course can arise where the modeler's imagination picks out some set of models and her model description picks out a different set of models, necessitating a refinement either to her imagination or to her model description.

Modelers often use models in order to learn about real-world phenomena, but as we will see, the model discipline is not exclusive or limited to simply learning about real-world phenomena. When it is actually about real-world phenomena, the model must be similar to a real-world phenomenon in certain appropriate aspects. As Quine, as I mentioned, as Quine has pointed out, similarity is a vague notion and we therefore should not be content with such a simple formulation of the model world relationship. One of the most active contentious areas of the structure of

theories literature concerns the question of how to give a more precise and detailed formulation of the model-world relation. Some philosophers such as Geyer argue that appropriate relationships between models and the world is one of a structural similarity. Your similarity to real-world phenomena lies in some parts of the imaginary structure literally having similar properties to parts of the real world phenomena. Other theories like Van Frossen and Suppi conceive of similarity more abstractly,

describing it as a relationship between mathematical properties of the model and the real world phenomena described mathematically. And the third view, as I mentioned, holds that model have partial isomorphism to their intended target systems via a series of models that are ultimately tied back to data. For example, D'Acosta and French espousing this kind of view. This means that some substructure of the model, for example, the relations between its properties stand in a one-to-one correspondence to properties of another model, which can be a model of the

real-world data. None of these views has clearly emerged as dominant in the structure of theories literature. Each of them has its critics and supporters. Sorry. Nevertheless, we are going to abstract from some of these details with regard to the relation between the model and the world. Naturally, a complete and final account of modeling will require this issue to be settled. Some accounts of models treat this relationship to the world as determinable simply by nature.

knowing the structure of the model and of the real-world phenomena being represented. For the purpose of understanding the practice of modeling, this view, however, is too restrictive. Models do not have a single automatically determinable relationship to the world. Different modelers employing the same model may intend different parts of it to correspond with different parts of a real-world phenomenon. They might abstract some details and not others. Some modelers may require the model to faithfully represent the causal structure of the relevant phenomenon, as well as make quantitatively accurate predictions.

Other may only require that the model make accurate predictions. For example, Volterra's model of prey and predator is a good example of these more nuanced properties of the modeler's intention about the model world relationship or isomorphism. Volterra believed that his model captured essential causal relationships that gave rise to the unusual fishery data following the First World War. ecologists think of volterra's model as a minimal model a template for building models of greater complexity by you know supplementing and plugging in more details precisely because the the platform

which is the car which which is the interpretation of the causal pattern in uh you know uh among the population of prey and predator is fundamentally sound. Therefore, if a modern ecologist deploys Volterra's model to study a real ecosystem, she does so with much lower standards of fidelity than Volterra did. Her use of the model is only intended to give a first approximation to the most important dynamics of the system. The relevant intentions of the modelers with regard to an application to a real-world phenomenon

are what Michael Weisberg calls the construals of the model. They are essentially interpretive factors with regard to the structure of model. What features of this structure should be included and which ones should be excluded. The control of a model is composed of four parts in the Weisberg paradigm. assignment, which is the modeler's intended scope of the model, and two kinds of fidelity

criteria. And of course, I will supplement the two fidelity criteria of Weisberg with one additional one, which I will elaborate later on. So the assignment and scope determine and help us evaluate the relationship between parts of the model and parts of the real world phenomena. Whereas the fidelity criteria are the standards theorists use to evaluate a model's ability to represent real phenomena and at a certain degree of accuracy. Now, of course, this certain degree of accuracy is where I would say that Weisberg's notion

of fidelity criteria becomes a little bit vague. That's why we need to supplement it with the third fidelity criteria to make it more determinate. So the first aspect of a model's construal is its assignment, which is the specification of the phenomenon in the world to be studied and the explicit coordination of parts of the model with parts of the real world phenomenon, itself described mathematically according to some accounts. Now this explicit coordination is important for two reasons. First, although the parts of some models seem naturally to coordinate with parts of real-world phenomena, this is often not the case.

For example, harmonic oscillator models were first developed to make predictions about the periodic motion of physical systems. But as mathematical models, they remain abstract objects without obvious analogs to the properties of the springs, molecules, or even a pendulum. them. Chemists use harmonic oscillators to model vibration in bonds. Therefore, they need to represent atomic positions as points in a coordinate system and treat the periodic offset of these points which corresponds to molecular vibration as the behavior described by the dynamics of the harmonic oscillator model.

Now, models typically have a structure not present in the real-world phenomena they are being used to study. This brings us to a second role of the assignment. And what is that? It is to specify which parts of the model are to be ignored, essentially the criteria of abstraction. This aspect of the assignment is well illustrated by a practical, everyday model. For example, I'm sure that you have seen those egg timers, you know, the ones, I'm not talking about the ones that are cloth based.

I'm talking about those ones you can buy from a $1 store. It's made of a specific kind of plastic. You put it actually in the hot water at the same, right at the moment that you put other eggs in the water, and then it has like three basically color variation, you know, for hard, you know, runny and, you know, in between. the color of the egg starts to change according to what kind of you know preference you have for running or hard egg or in between so this is a let's call it a a model egg this is a model egg

so the model is usually comes in red plastic egg shaped object that has the words hard medium and soft printed on it. Now if you want to make a medium boiled egg, you drop the plastic egg in simmering water along with your real egg. As the plastic egg heats up, it gradually changes from red to black, starting from the outside and working inwards. This mirrors what is happening inside your actual boiling egg. Heat is slowly being transferred from the outside of the egg the inside. So when the egg timer has been sufficiently heated, the black color reaches the word medium and then you can remove the real egg from the heat. Now

apparently the egg timer works because the plastic out of which it is made has similar heat diffusion or transference properties similar to a real egg. So the plastic egg provides a high fidelity model of heat diffusion, the most important property associated with getting a well-cooked egg. Any question before I continue? Okay, so let me finish this next section and then we'll have a rest.

Okay. So, although this is not a mathematical model and not a scientifically interesting case, the plastic egg model demonstrates the second important role of the model's construal. Many aspects of the model egg are irrelevant for the purpose to which it is being used. We actually don't care that the model egg is red, made of plastic, a bit lighter than the real egg, completely homogeneous, and printed with words. However, we do care that the model egg has approximately the same heat diffusion properties as the real egg.

And the color change on the model egg represents heat-induced coagulation on the real egg. The construal tells us which parts of the model correspond to parts of the real phenomenon and which parts can in fact be ignored. No one would ever assume that the real egg is made of plastic or is red. However, this issue often arises in a more subtle way when one considers mathematical models. For example, Volterra's predatory prey model is described by two coupled differential equations. These equations, one for the prey population, one for the predator population.

Volterra's predator, so these equations and the mathematical model they describe are defined for any real valued number of predators and prey. However, Volterra certainly did not intend the fact that this model could describe dynamics of non-integer numbers of predator and prey to correspond to any real or possible population of fish. Therefore, in his construal of the model, Volterra only assigned the integer values of the model and probably only certain ranges of these integer values to the population in the Adriatic and other possible populations.

So, before I move to a scope, let's have a break and then coming back. Sounds good. About five minutes. Great. Cool. Thank you.

Thank you. Thank you.

Thank you. Thank you.

Thank you. Thank you.

Thank you. Thank you.

Thank you. Questions?

Yes, yes, models are reliant on theories. However, yes, there are, in terms of their structure, they are reliant on their theories. But when a modeler makes a model, there are certain times that in fact a modeler do these weaseliest stuff which are not in fact permitted inside that theory or simply they are not existent inside that theory. For example, a good example of this is an approximation technique.

You see sometimes modelers do use approximation techniques in order to map the details of a particular level to a more accessible level in which they can do the actual modeling. But of course, the assumptions of this approximation, the methods of making this approximation, this bridge between different levels, might not be even existent in the theory in which the model is fabricated. I'm kind of wondering, like, I'm kind of wondering how you can have a theory without

a model now, right? because it seems to me that you might have some sort of not well stated model or very implicit model or something that's not very well separated out from the theoretical sort of aspects of it but um I I kind of if it's a scientific thing how can you not have a model this becomes basically the question that we have been addressing is essentially that very fine distinction between direct abstract representation and indirect abstract representation. Essentially, yes, all models in order to frame data, empirical data, require some sort of modeling. But of course we

we can call them models as like Van Fressen does, but then that I would say, it becomes just simply a taxonomic issue then, in the sense that, okay, if we overextend the definition of model to all theories and the paradigm, the paradigmatic of theory construction, then we actually rub ourselves of, you know, more juicy details about what actual models are. Yes, sure. Okay. Yeah, I mean, I would say it's more of a taxonomic issue, which is useful to keep in mind. So there's always modeling going on.

Yes, absolutely. Yeah, I mean, you know it very well. The whole idea that there is such a thing as empirical big data where basically theories feed off of. No, that's just complete travesty. Every theory models data. Essentially, models are and framing of the data. Right, right, right. So there's modeling going on, but to say that the modeling gets to a certain level of quality, essentially. Yes, yes. And we can talk about there being a model and a theory and a meta-theoretical embedding. And, you know, we can start using this taxonomy only when the quality of the modeling gets to a certain level, in a way.

Right. And, of course, Theo is saying that, but isn't the problem also about the scope and capability of theories too? Yes, of course. Isn't it really what we were talking, and we are getting back to this issue in our later sessions, how we were defining theory constructions in the vein of Joseph Vesnid and his secular. So you can think about models as scopes in terms of not just concrete, like these kind of models that we have been talking about, you know, what we call to be, you know, clear-cut modeling discipline, but in terms of possible models, partial potential models, so on and so forth. Yes, absolutely.

But as I mentioned to Adam, I think it's actually useful to retain some sort of taxonomic distinctions between these. So as not to confuse every theory with a model, and every model as actual theory. Because if we do that, essentially, we are making a very, very kind of dangerous mistake of thinking that, yes, theories are actually capable of being conscious of certain assumptions that goes into the construction of the said theory,

but models are not. This was one of our first things that, yes, they are made inside a specific theory. For example, Volterra, you know, any like Schelling within the game, theoretical choice theory. But the thing is that models are not really capable of actually telling what are the theoretical assumptions behind them in a coherent way. not to mention the meta-theoretical assumptions. I think it's best to think of this as a subset taxonomy

in the sense that, so you have theory, and then within the theory, you have various classes of models in the vein of Joseph S. Snead, you know, partial potential model, partial model, than the actual model. And this is how we can actually think about the structure of these specific models much more coherently without eliding distinction between what they mean by a structure at that level. don't worry we i mean those of you who are not familiar with the work of joseph snit we have

covered this a little bit in our previous class which was on philosophy of science we will we'll bring this back later on in our class and you know we talk about this more coherently Questions? Yeah, I was wondering, aren't the model descriptions also models? Because there are mentions some diagrams. Models as such, they are what you might call to be the encapsulations of the structures. Now, what does this mean, really?

So, you see, what really makes a model a model, like Volterra's model, is a structural core. The structural core pertains to the kind of phenomenon that you are in fact trying to study in accordance to a theory. So here it's about the competition or the dynamics between a population of prey and a population of predator. Here the structural theoretical core of the model, which makes the model the model, pertains to

you know, competition in a dynamic system terms. Whereas the equations simply describe this structure. You see, they don't actually say anything about what the structure is. It just simply is like pictured, you know, like, you know, And that's why a model description can be equation, can be a graph, can be a picture, so on and so forth. It's an encapsulation. And this encapsulation already sacrifices so much of information about what kind of phenomenon you are in fact trying to deal with.

According to what theory, so on and so forth. Yes, you can, of course, garner some insights, for example, even from the differential equations going into Wultero's model, as what would be the structure of this model. But these are quite trivial. So, and that's why I think Weisberg is right. A model structure is very different from model description. These two are not equal. It's like an allegory for it, a metaphor for it. So you have a map and all the navigation points, the system of map making, you know, according

to what kind of system of map making you have made this map. would be your structure. Now of course the kind of pictorial representation of the map would be your description. The description doesn't really tell you that much about the kind of decisions and determinations have gone through the process of map making, according to what system of map making, so on and so forth. Yes, yes. Adam, can you detail this last point of view? I guess I was just thinking, like,

y equals mx plus b, right? That's such a simple bit of math. It's simpler than the Poltero examples, right? But that's applicable, like, for instance, in Newtonian physics, where you're describing displacement on the constant velocity, for instance, but it's also, you use linear equations in so many different places, right? But the model, the structural feature of the model is just that equation, right? And so it's just almost like the machinery and it's without the description, added to it, you can't do much with it in practice, right?

It's sort of like a hammer without... Yes, absolutely, yes. Without that kind of interpretive core, which is that of the structure, I mean, you might as well apply it to human behavior, rather than the movement of celestial bodies. Right. More, any more, anything? So could you say that? I don't think it's a metaphoric, Sean.

These are, it's just, you know, I need, okay, let me, actually I don't think that I'm in the position to answer these questions coherently. But I don't see this model description to be a metaphoric relationship and the other one a metonymic relationship. I just need to think a little bit about what exactly the kind of relationship between the description and the real world phenomenon is, what would be the relation between the structure and the real world phenomenon,

and more importantly, the relationship between the description and the structure in a kind of more of a common sense vocabulary. Let me think about this. Let me think about this. You see, when you guys are saying metaphor, what do you mean by metaphor? metaphor. I think that the word metaphor is a little bit vague here, but any of you, can

you actually kind of elaborate and expand on how you see this description as a kind of a metaphoric component? So I'm thinking of metaphor as the description of something that I suppose gives a certain kind of space or a gap between two things. Like in this case, it would be the description of a model gives a certain gap between that description and the real world phenomena. It's as though the model description doesn't touch or something like the real-world phenomena

actually like grasp it in a certain way. Whereas like metonymy, I understand is kind of sharing like a more continuous relationship that then could be mediated by something like theory or various meta-theoretical positions. But I think that at a higher level, perhaps that metaphor kind of relationship between Okay, in that sense, yes, there is a gap and there is a continuity correspondingly with regard to description and the structure. Yes, but the use of the word metaphor always assumes a kind of what you might call to be,

you know, kind of an inferior abstraction, a kind of, you know, something that is simply is going to give us a semblance of impressions about the kind of stuff that we are trying to actually relate. But the description, I don't think that completely falls under this category, if that, you know, what we mean by metaphor.

because a description, yes, there is a gap between the description and the real full phenomenon, but the thing is that you should, we should also think that the description of the model is not opposed, it's not something opposed to the structure it is actually a derivation from this structure according to some mathematical computational paradigm and so on and so forth that allows us or symbolic that allows us to kind of encapsulate this is structure within the domain of the symbolic pictorial computational mathematical so on and so

forth and in that sense what counts as a model description is not really a metaphor by any means but it's just simply what you might call to be an informational package which which tries to which is kind of okay let's think about this as a metaphor an informational package which is opaque you don't see what is actually going inside it, okay?

What would be really the act which you don't know what the kind of theoretical structure is inside this box. It just that this box allows you to mobilize a structure in a very effective manner with regard to a real world phenomenon. Now, well, if that's a metaphor, that would be a different question. But as I mentioned, I think that, I mean, at least the common sense word metaphor can be quite misleading as a kind of a qualifier for the word model description. But yeah, I'm just thinking about this.

Yeah, don't take me seriously at this point. Yes, it is a specification. A specification, however, not of the structure, specification of the structure according to a certain symbolic regime picture, graph, equation, so on and so forth essentially description doesn't give you more information than what already the structure

the structural core gives you it in fact limits this information I think I'm not really clear as the difference between the model description and the model instantiation. Could you elaborate a bit? What do you mean by model instantiation? Now I'm even more confused than you. Okay, sorry. Yeah, in the chapter, Weisberg says about instantiation, what's his term?

Instantiated Modules. Yeah, it's in page 37. Steven Oasek and Nadia Tsober introduced a helpful concept of model description instantiation. An uninstantiated model description is an equation in which values are not assigned to the parameters. Can you repeat that sentence one more time? An uninstantiated model description, that is the description that is instantiated, so my bad. An uninstantiated model description is an equation in which values are not assigned to the parameters. Instantiating model description means adding in values for the parameters. So, the instantiation, it seems that an instantiation model description is one wherein there are specific models assigned to the variables, it seems to be.

Yeah, yeah, yes, yes, no, I remember this. It's essentially, from what I gathered, and this actually comes a lot in, you know, more, you know, kind of scientific works and philosophical works on model development. is essentially, it's not really a kind of, what you might call to be an epistemological distinction or a kind of a grand distinction. It's just that, so instantiation of a model requires for you to have certain correspondence

with the real world data, empirical data. And these empirical data, these values that you plug into your equations can actually you know, constraint your variables. They can change them in one pattern as opposed to another. So that's what you might call to be instantiation. instantiation essentially in this sense simply means a deployed model, a deployed model. Whereas description is just free floating equations.

It still does not have, you know, you haven't yet plugged or punched in the data and hence it still needs to be yet to be determined what would happen once it actually comes into contact with a real world phenomenon. Okay. It seems then that instantiation comes at a later level than model description. Yes. You have first model description and then you have instantiation or un-instantiation of this model description. Yes, absolutely.

And in fact there are quite massive amounts of examples in the history of modeling in which at the time of application the modeler notices that oh my equations look like crap and then you have to change either the model structure or the model equations the description okay and i i'm wondering if thinking about the instantiation of a model description could shed light on the metaphoric or not nature of the relation between the model description and the real world phenomenon in the sense that if the model description is instantiated it seems there

is a i mean it's purporting to be a somewhat referential relation not not a metaphorical relation, but it's a referential relation according to a determined set of parameters. Something like that. Right. Yes. It's not really a metaphor. It's not really a metaphor because a metaphor, I mean, there's a whole lot of things about metaphor around, but I like the Goodman the Goodman definition it is a predicate with a past

which it's an affair between a predicate with a past and an object which yields to it protesting right yes yes it's even worse okay yeah no yeah absolutely it's more of a referential yes absolutely it's a referential And precisely because of that, that's why I was, you know, making that claim that I just, yeah, sure, I mean, to the extent that the word metaphor can mean so many different things according to so many different theories of metaphorization. but still I don't think that it's of a metaphoric relationship.

Yes, I mean, in a sense, the model is absolutely referential. It is about a certain aspect of the world which is tractable. Yeah. Okay. not only that aspect of the world is tractable but also the relation with that aspect of the world is tractable yeah in the sense that you you get to ideally control the metric between the the model and the world I mean and

Right, right, yes, yes. Okay. I mean, the application of the model, many people think that it's just like plopping up some sort of haphazard system on some haphazard system of reality. No, the whole idea of application is controlled metrics. Yes, that was a fantastic term that you just invented, by the way. It's a triangulation process. Yes, I see. Anyone, anything, complaint?

OK. Don't ask me. Ask Jean-Pierre about this. Jean-Pierre, expand on your notion of controlled metrics. Yeah, I'm not sure. You sure about the systematic aspect of this. I'm just thinking that I'm just freestyling on the metaphor reference problem. And it seems that how does a metaphor function? You transfer some predicates from some domain to another, and the fact that you are applying these predicates to another domain seems to bring forth some characteristics of the new domain.

So in a literary metaphor, this seems what's going on. I mean, in an idealized sense. Trojan horse, Trojan horse deploys the resources of one realm into another. Yeah, yeah. Greek into the... Greek into the Troy. Yeah. Kingdom. Yeah. But this... I think maybe it's an issue just of the... relating to the objectives of the metaphorical deployments, but regarding the model, I think it's not as free-floating, I mean the effect shouldn't be as free-floating as the metaphorical

effect in the sense that you have to be able to understand the degrees of approximation between what is from the point of view of the model purports to be a reference, a reference to something in the world, but it is a bounded reference, a reference bounded to the model itself to the parameters set forth by the model. Bracketed by the structure. Yes, something like that. So the controlled metric would be some way to decide of the degree of approximation between the values that the variables are taking in an instantiated model description, something

like that, and the real world phenomenon. So I'm not sure what the question is in the sense that how to do, how to control this metric. I don't have this answer. Well, there are actually, in response to Theo, I would say that this is something that I have only recently come across. And, you know, in old stuff about modeling, this is mostly talked about various, you know, technological instruments. allow us to make this approximation, what you might call, to retain the fidelity of this approximation.

Whereas now it is actually not about the way that I'm seeing it in the new stuff about modeling. It is not about the technological instrument. In fact, it is a purely theoretical discourse. In a sense, however, it is a theoretical discourse that is usually this kind of controlled metrics is usually posed in terms of statistical probabilities, statistical probabilities for approximation. There is a whole new genre of this stuff, some of which are philosophical and some of which are absolutely beyond my pay grade, unfortunately.

But I can, in fact, give you a few references. I mentioned James Crutchfield, who is actually working on some of this stuff. some of his more new essays, but a few others. I can, if you guys are interested, I can give you some references. But yes, yeah, sure. I mean, really, I don't know exactly what would be the nature of this control matrix. But yes, as a buzzword, let's just think about this as a kind of a triangulation by different kinds of methods,

some of which, as I mentioned, exactly like approximation techniques, are in fact not informed by the theory in which the structure of the model has been couched. They are coming from auxiliary theoretical assumptions, statistical physics, Bayesian inferences, so on and so forth. However, now that it just struck me with regard to what I'm going to read just now, that what

What Sean was talking about metaphoric and with the comments that Jean-Pierre gave us. So I think that yes, the idea of the metaphoric relation between the model description doesn't hold when the model description is about a real world phenomenon, namely referentiality. However, what if the model is about another model rather than a real world phenomenon?

And we have a lot of this in the history of science, where models are not actually about a real target system, it's about another model. And that model can be about another model and down the line the model of the empirical data. something to think about. Or maybe we are just simply overextending our common sense concepts, like metaphor, to something that is highly,

highly specialized and scientific and is out of the purview of our common sense concepts. But nevertheless, it would be actually, I would say, with Quine and Karna, it's actually sometimes important to talk about what is actually happening inside the scientific domain with some sort of help of some sort of common sense concepts, such as metaphors, such as metonymy, and so on and so forth. Okay, let me instruct.

So we talked about the first one, the first construal, which was the assignment. Now the second component of a model construal is the model's intended scope. So what is a scope? It tells us the aspects of phenomena intended to be represented by the model. This is the definition that Supi gives in his, one of his earlier essays in 1977. This is related but somewhat different from the assignment criteria, which tells us about

how to coordinate particular parts of the model to particular parts of the target phenomenon, and which parts of the model should not be taken to represent anything, like abstracting away from them. So a scope is best illustrated by example. So let us turn once again to Volterra's predatory prey model. The model itself only describes the size of the predator and prey population. the natural birth and death rates for these species, the prey capture rate, the number of prey captures required to produce the birth of a predator, so on and so forth. It contains, however, no information about spatial relations, density dependence, climate

and microclimate criteria, or interactions with other species, neutral species that are not prey can be categorized as prey or predator. So if the scope is such that we intended to represent those features, Volterra's model does a poor job precisely because it would indicate that there is no density dependence, no relevant spatial structure, etc. By choosing a very restrictive scope, we indicate that Volterra's model is not intended to represent such features. In Weisberg's account of the construals,

the third and fourth aspects of a model's control are its so-called fidelity criteria or constraints. So while the assignment and scope describe how the real-world phenomenon is intended to be represented with the model, fidelity criteria, on the other hand, describe how similar the model must be to the world in order to be considered an adequate representation, adequate in bold. Now, there are two types of fidelity criteria in Weisberg's account, dynamical fidelity criteria and representational fidelity criteria.

Now, dynamical fidelity criteria tell us how close the output of the model must be to the output of the real world phenomenon. It is often specified as an error tolerance. For example, a dynamical fidelity criterion for a predator-prey model might state that the population size of the predators and prey in the model must be minus or plus 10% of the actual values before we will accept the model. So it's like a threshold tolerance. Dynamical fidelity only deals with the output of the model, its predictions about how a

real-world phenomenon will behave. Representation fidelity criteria, however, are more complex and give us standards for evaluating whether the model makes the right predictions for the right reasons or not. these criteria usually specify how closely the models internal structure must match the causal structure of the real-world phenomenon to be considered as an adequate representation now I mentioned about the precision of defining these two fidelity criteria.

And I mentioned that I think that Weisberg missing something here. And I would like to add it by saying that there would be a third fidelity criteria and it is a resolution fidelity or a scale sensitivity, level fidelity. In the sense that these representational and dynamic criteria only make sense in the procedure of modeling, if and only if we are capable of distinguishing the very specific scales to which such criteria

are being applied or certain scales or levels at which we are basically pertained to our model making procedure coming back to that old tiresome example that I always make the metal beam you know so when an engineer wants to deal with a metal beam of course you know we are dealing with different scales our models of a metal beam should be a scale sensitive otherwise we are not going to make it

bridge we are not going to make some sort of bunker that can withstand that can withstand nuclear bunker buster technologies it's just going not going So what are we going to think about as engineers? So we have different concepts of hardness, models of hardness, hardness 1, hardness 2, hardness 3, hardness 4, ad infinitum. Each of these concepts of hardness correspond to a different scale of the metal beam, from the microscopic level of elasticity down to atomic and quantum level. You see, at these levels, the concept of hardness fundamentally changes.

There is no global concept of hardness that can be applied all the way down, okay? It is local concepts of hardness specific to different scales of the metal beam. And to that extent, when an engineer tries to make a model and work out the so-called fidelity and representational criteria, she should be capable of exactly determining the relations between such criteria and the very scale at which the model is being construed

questions by scale at which the model is being construed you mean the type of application that the model will serve or? Yes, but also at which structural scale of the real physical phenomenon, the model we are imagining or envisioning the model? Is it nanometric? Is it atomic? Is it crystallographic? Or is it microscopic elasticity? Now, this I would say that not all modelers do actually follow these last criteria.

But in fact, looking into the new literature about the history of science and modeling, we see that the question of a scale is becoming ever more relevant with regard to modeling, with regard to theorization, with regard to parametrization and bracketing of our real-world phenomena there is in fact a direct relationship between the idea of ontological scales and the epistemological requirements for singling them out and the idea of complexity in physical sciences

Now here you should ask some question because that's too weight and if you don't ask question I would say that you just didn't get what I just said. Do you mean like, tell me if I'm wrong, but um, so it's not simply like at like a solid level you have one skill and then at a liquid level you have another skill, a gaseous level you have another scale, but rather, are you saying that the fact that these three scales have a kind of like infinite idealization itself has a kind of like epistemic criteria? Right. Yes. Yes. That in fact, how do I put this? It's in fact like you're putting a scale on top of another

different kind of scale or something. Right. You know, Jovan, think about Darwinian evolution, okay? like the evolution of the nervous system. So, I mean, evolutionary biologists believe that the first instances of the nervous system emerged among the invertebrates, possibly a lamprey. Does any of you know what a lamprey is? It's a delicacy, it tastes yummy. Theo, I'm sure that he knows, terrifying. So essentially a lamprey is kind of like a sea worm but with massive amount of teeth. It represents one of a concrete model of evolution of the nervous system among the invertebrates.

Now the thing is that this is really a kind of thing that happens in complexification And evolution is really about complexification also. Complexification, however, needs to be here defined correctly. Complexity doesn't mean increase in variations. Complexity actually is disproportional to variation. It means reduction of variations. So let's think of like this. So first we have this kind of invertebrate nervous system. essentially it can be thought about, you can define in terms of sensitivity to

light, sensitivity to some sort of acoustic or haptic touch, such that the organism is capable of differentiating itself from its environment, right? That's the first trigger for the evolution of the nervous system, telling the difference between yourself and your food. Then as the complexification of the nervous system emerges with new relation processes right at kind of at the end of the Cambrian period we see that there is a kind of like

a systematic, from an epistemological scientific point of view, a systematic way of everything that now adds to the previous features should correspond the constraints which were already in place. So it was sensitive to light and this sensitivity to light becomes more complex, more complex, And essentially leads to, for example, something like the morphogenesis of I in developed organisms. Now the thing is that here something happens precisely because evolution, the complexity

of evolution means that each later stage of evolution should in fact respond to the constraints which are already instantiated, the tolerance or the threshold of its variation decreases. You can't do, like you can't just simply grow an arm out of your head at this point, like humans. But nevertheless, this is, so variation is no longer an element of complexity. Complexity means reduction in structural variations, but increase in the functionality, what they

can do. And this is what is usually called genetic entrenchment or generative entrenchment. Generative entrenchment means that complexity increases as more limitations are being introduced for higher level evolutions, entrenchment. And generative precisely because addition of these constraints does not in fact reduce complexity, increases complexity, it reduces variation,

but variation is not complexity. Now, this is in fact the very biological counterpart biological counterpart of today's physical theory of complexity, which is called a structural stability or a statistical stability. From a statistical point of view of physics, we can show that the more decisions you have made in the past the more complex your decisions become in the future not variation though they don't become varied from the past they become more complex yes history yeah

yeah, Hegel I thought that like biological complexity would have a very different it's a different kind of scale or complexity even from physical complexity that it's not necessarily commensurable You see, of course, many biologists would say that it would be a mistake to compare, for example, biological complexity with physical complexity proper. But the thing is that two things.

Of course, this comes back to this idea that, yes, of course, we are working at different scales. But nevertheless, we can have a specific statistical criteria for a biological system as in contrast to a proper physical system. And in fact, we can show, as it has been shown recently in the past few decades, that such a specific statistical system applied to the biological system and a specific to biological systems show the same kind of complexity, vacation processes and criteria, but of course bracketed, qualified, similar to physics proper.

Yes, you know, I understand that we just cannot simply overextend biology to physical laws or talk about them as if they were just physics proper. But nevertheless, even if we reinvent a statistical modeling, statistical theorization for each of these scales of physical system, physics proper and biological system, or chemical, in fact, processes, we can show that, yes, this happens. And there are many, many theories pertaining to why that is happening. Well, basically, it's actually a computational constraint

that holds across different scales with regard to complexification. So it's statistics that is a nexus, they're like? Yes, yes, yes. So Theo, would you be able to, yeah, I can speak out loud. I was just, I'm just wondering, I was just wondering about the differences between the necessity to think about scale when you're modeling against, say, a theory's scale specificity, or do theories have scale specificity? I mean, it seems like in some ways theories don't have scale specificity.

Right, they don't. Traditionally, they don't. So then, however, this is why I've always tried to propagandize the idea of complexity science precisely because complexity science is not just about complex phenomena. In the question of the scaling physics, the classical text by Romeo Baadi and his colleague, which is the canon of the complexes of science today, they actually advance a scale sensitivity

not just as an ontological problem with regard to target system, but essentially as a main feature of how theories should be constructed from now on. Kind of like a Carnapian explication in fact. Just as we need to have refinement of concepts from big ideas to more fine-grained explicator, We should have fine-grained theories which are a scale sensitive. Yes, just because it hasn't been done in theory for the past hundred years doesn't mean that it shouldn't be. Precisely because this is part of the theory construction of modeling data, understanding

what a law is. Right. I mean, in some ways, I think the I hadn't really thought about this until this class, but it seems like. the project of theory making I mean it has a huge limiting I guess it limits what models are capable of doing yes yes it does but it's not essentially a negative limitation it's a limitation that refines right right but then i mean the i guess the upshot of thinking about it from the modeling

end is that in order to have a uh purposeful or uh constructive productive kind of theory you have to have i i guess just a um my apologies my cat is well it's okay i i guess i'm just noticing that like theory generating practices or theory making practices are have to find themselves being limited by model making practices reciprocally so right yes yes absolutely precisely because i would say that the canonical paradigmatic of theory construction at this point is beholden

to centuries-old paradigm of theory-making, whereas model-making is more in the side of complexities of the world and complexities of conceptual systematization. And I think Yes, absolutely. Yes, that was a fantastic insight. Yes, absolutely. This should be now, they should inform one another. Yes. It shouldn't be a unilateral from theory to model. It should be also model theory. It seems like in some way models could potentially challenge theory. Yes, yes, precisely because they can actually, in the vein of Boltzmann, they can actually either point to new facts of experience or in the Kuhnian

sense they can point to anomalies which are out of the purview of a theory. Right, I mean I think the example of the social Darwinian theory comes up a lot but I think maybe it's not the best one to use but I think social Darwinism seems like it's embedded into the theory of Darwinism because of the metatheoretic assumption. Yes, yes. You brought up a fantastic point. Unfortunately, if I say something like this on Facebook, I would be astounded to death. Well, my...

Yes, but it's absolutely true. Yeah, precisely because social Darwinism is not some sort of fascist addendum to a respectable theory. It's just a natural consequence of the Darwinian theory not being a scale sensitive, namely not being actually aware of its metatheoretical assumptions. right yeah I mean my purpose is not to bring it up in obviously I hope but in sort of advocating a type of social Darwinism but merely to show that the theory itself contains social Darwinism and that right yeah absolutely because

it's on the scale sensitive yes the same thing about Hamiltonian laws of conservation in physics and and we see that it creates massive amount of problem in the history of physics down the road yes absolutely yes definitely the scale sense and that's why I brought this scale sensitivity and of course it is you know I am just a philosopher I'm just I don't have a good grasp of some of the more detailed elaboration with regard to the question on the scale precisely because the question of a scale can be posed also in different contexts what do we mean by a scale or level do we mean a scale with reference to how we predict a regularity do we mean by a scale and you know the size do

we mean by the size of the aggregates covered by that scale so on so forth There are so many contexts. And of course, these are all extremely, extremely detailed questions which require a lifetime to be answered. Questions. Adam, would you be able to elaborate your question in the sidebar. Yes. So I was just back and forth with my kids a little bit. So I missed some key parts of the good discussion

that was happening now. I guess it seemed to me that at least when models are well constructed, the models themselves sort of have correspondence to these ontological levels right and what you're talking about is sort of when models and theories are well constructed they are well associated with scopes and ontological levels in the sort of well yes but however that brings back to the question to the to what i was just telling to theo that is one way to interpret the idea of a scale correspondence or scale sensitivity, namely ontological scales. But of course, we can have different context for

how we can distinguish one scale or level from another. Prediction, regularity, assemblage, size, you name it. Yeah, okay. Okay, Lenka some questions, sorry. Someone, Theo, would you be able, I can't read this. Sure, yeah. So Lenka says, I think she has two questions. One, does this world of multiplicity or plurality of scales and theories slash models proper to them imply that only universal

epistemological criteria is uh has local fidelity i guess uh and two if knowledge is created enriching reality rather than acquired does the modeling go both ways from the model structure to the real phenomena slash target structure and the other way around simultaneously okay the first question let me type this I think I have mentioned this already this is I absolutely suggest that every one of you should read this it's online on academia

EDU engineers and drifters. So you see at the turn of the century, at the turn of the 20th century, a great debate emerged between those who believed in universal epistemological procedures and hence they were in one way or another like novales like romanticists believed in the unification of all sciences at the end of the road max plan was a you know a

proponent of such view in opposition to this camp there was another camp represented by ernest mach Marx had this idea that there would only be ever more fragmentations, precisely because neither our epistemological tools are universal, nor can we in fact do proper science by some sort of recourse to universal concepts. We can only go with local concepts, local models, and when we are in the business of local concepts, in the absence of a possible unification,

there will be ever more fragmentations of local theories, local concepts, local models, so on and so forth. Now, of course, this is a very, very, you know, a strong debate that has continued among the children of Planck and Mosh for the past 100 years, if not more. I actually take side with something like Karros, who represents a very Carnapian vision, a dialectical vision, between local fidelities and universal or unificatory principles.

A good example of this that Karras also makes, so yes, I mean when we look at the kind of procedure of science that is going on today, we only see ever more fragmentation. It's just that the how science evolves at this point with our models, we see that how can ever be a possible unification of all sciences, the kind of universalism in the realm of science. But the thing is that, as Karras points out, universalism actually shouldn't be always thought as an ideal at the end of the road, which might not actually happen, but actually as the first

principle in which local theory and local fidelities take root example of this after Kepler but even more strongly Newton we see that every other theory no matter how diverging these local theories are about gravitation they have to abide by a specific set of constraints or astringencies with regard to the equations of motion and gravitation. Now, this is what shows that if this is in fact the case,

this whole point of local fragmentation or local fidelity and universal epistemological procedures is not really a kind of one way or another. It should be actually thought about dialectically. And in fact, the question of unification of science, even when all we have at this point is local fragmentation, specialization, modeling at ever more divergent scales, this does not rule out the idea of unification of all possible sciences.

And in fact, if we look into the history of science, we see that there are so many examples of that already happening. So why shouldn't it happen now? Now, well, to say that it's not happening precisely because of this or that would be an apparel of limitations. But we can look at Copernican revolution, Keplerian revolution, Darwinian revolution. Each of these are in fact unificatory moments for the kind of fragmentations that have gone through modeling of different aspects of reality, bring them into a coherent whole. So, yes, I think this question is fundamentally open, and I wouldn't take side with either

of these two, precisely because, like, pure universalism or pure local fidelities, precisely because I think that there are just these kinds of bipolar reactions to science would, in fact prevent us to imagine new forms of sciences, new form of epistemic procedures and unificatory processes. With regard to the second question, I think that I need to think about this. Let me copy paste this.

Give me one week, if you are kind enough, to think about this coherently, because I don't want to just come up with some sort of, you know, rudimentary answer. I just want to take my time. I think it's a really important question and it's something that I haven't really decided no so far will be answered. So let me think about the second question. Okay, Barrett, would you be able to elaborate your point? I guess that was just a general statement of pessimism, I guess. I just, you kind of answered what I was actually, yeah.

You cannot call yourself a pessimist. Pessimism is a psychological state. We are in the business of science. Science doesn't care about your psyche. No, no, no, no folk psychology. I'll stay away from that. But no, I was just sort of, you kind of answered that by saying, you know, I don't believe in this sort of like pure universalism or pure local fidelity. So that was kind of actually answered what I was trying to get at because I kind of agree there. So I don't really have anything further to say. Yes, no, I mean it's essentially dialectical.

In fact, to think about universality without fragmentation would be a pure caprice. To think about fragmentation without a universalizing concept would be pro-humanism. is that a lot of people tend to, and again, this is a generalization, but one thing I see is people think of universalization without the possibility of fragment. Right, right. And that's like the one thing about today that just kind of drives me crazy. Yes, but also as an opposite, you can see, for example, unconditional acceleration. is the majority of them think that you can have fragmentation all the way through with no possibility of universalization yeah but i don't agree with that either yes and i and

i flirted with with the sort of unconditional acceleration of some stuff but i was like but wait isn't this just yielding because i think uh what's isn't it the whole methodological individualism all the way yeah but then it just revert it just reverts itself to this other sort of theocratic dimension, which is why a lot of these sort of like unconditional acceleration people have kind of like embraced this like neo-baroque Catholicism kind of thing. Yes, absolutely. I would say that both of these are not philosophical in nature. Any search for a fundament, a solid fundament, would be a theological discourse, not scientific or philosophical. Well, it's aesthetic. Yes, aesthetic. Yes, absolutely.

Yes, but yes, I really, really suggest, read this, please do read it. Okay. Andre Carus, Engineers and Drifters. So, drifters are the ones who are going for the local fidelity all the way down, and hence, they only get fragmentations. Yeah. Whereas engineers have a more dialectical side to them, in the sense that they need to to have a universalization process, consolidation process, and also fragmentation. Okay, if there is no question,

let us end this session and we will meet. uh next session i haven't determined for what we should read for the next session if you are uh generous enough allow me to think a couple of days and then i will post it on the google classroom thank you very much everyone fantastic class as always great participation thanks rosa all right I'm going to end the broadcast. Okay. Take care of my...