Hello and welcome to the ninth session of Theory and Object with Reza Negri Stani. I'll pass the mic off to him now. Thank you very much, Leo. Thanks, everyone. I was actually wanted to ask you what sessions we are in. The ninth session and we are still in car nap. I'm not even finishing car nap. don't worry I think this is going to be the record high longest a new center her course so today we are going to you know explore a little bit more of car nap particularly is a statement view and its overall project probably this whole
session will be on Carnap and next session we go to I recognize that I actually confused the order of a couple of sessions so the next session if you want to read you are going to read the face-off the challenge or the controversy between Carnapian inductivism, which is essentially an inductive logic, an observation-based Humean form of induction, and Popperian deductivism.
To my knowledge, I don't have that many texts that can capture this controversy between Karna and Popper. However, if you are interested for the next session to look at it, I would say read the The introduction to the foundational, I think it's foundational problems of theory of probability or something like that, or philosophical problems of the logic of probability.
I will send you the book. It's actually quite a very difficult book. However, the introduction I think can give you a little bit of an indication as what Carnot is trying to achieve. and after that that would be just one session after that we've moved to hemp account of explanation I noticed that we missed a few of our regulars, Peter, Justin, and a few others.
Justin's here. Peter, I haven't heard from. um jeff told me he's going to be a little bit late today okay okay and i i don't know uh where adam is he he's been missing since last class too um poor adam has a back to back classes so i can forgive him for her missing this class then he has to deal with adam after a while. So OK, let's start if you have brief questions, brief questions, like the stuff that you had somehow problems with or difficulty understanding
in the last session. If you don't, I will directly go and start our new session. So please feel free. Maryseus? I finally get my Sundays back from my family. They've released me from all my obligations. I've been catching up, but I'm excited to get back into it. Excellent. Joven, Justin, Laura, Meredith, Sepide, Svetlana, Theo, Christian, of course.
Anyone, anything. Don't be shy. I rarely bite. I fell behind on Carnap a little bit I'm doing some catch up this week okay so if you don't there is no question sure absolutely you asked us not to ask questions ask us not to ask questions I think you you discussed it but maybe I'm confusing You asked to resist. Are you sure you don't have a question or is it some sort of pretext to pose a question? I have lots of questions.
Okay, go on then. Shoot. Oh, so I just, first maybe, first comment, I really enjoy reading. I just went through introduction of Karnapin 20th century book that you recommended and Excellent. It's excellent. It's superb. Literally the last chapter particularly, which is on the notion of explication, is among one of the greatest gems of chapters I have read for a while. It's truly magnificent. I have a silly question because in this introduction, he refers a lot to the concepts or ideas of enlightenment.
Enlightenment, I mean this is every student of philosophy is supposed to know, but I'm not a student of philosophy Yes, whatever else you are I would like maybe to give you like some basic references like especially specifically like the kind of reference to idea that enlightenment is Canotes or implies Yeah, because that supports this idea that he develops in this book. Because my understanding of enlightenment has been quite fuzzy, and I've read this and that, and different texts called What is Enlightenment, and whatnot.
But I'm looking for, I'm looking in my notes, what he talks, how he defined this idea of enlightenment. I can't find it right now, but it's something about universal access to knowledge. yes come universal access to knowledge come diversification of points of entry or points of access which are basically which is basically a thesis of explanation and principle of tolerance for carnapp yeah and unity of unity of knowledge no i would i would rather i'd rather like to
because Karnap, he picks up on others as far as I understand, he picks up on these ideas, but they came from someone else's writings. Who are those people? Like which writings? Like I would like to get those. So you see this whole idea of the war, of course they weren't really in a very implicit way they were addressing in terms of enlightenment but this was a very momentous debate between Mark and sorry Clank Max Planck Ernst Mach and Max Planck now whether the unification of
science is possible or not, precisely because the unification of all sciences literally is what you might call to be the ideal of scientific enlightenment or rationalist enlightenment. And the thing is that of course the debate was far more far from this idea that whether all sciences can simply be bunched together reduced to physics or rather the idea that if we have these partial models of the universe, one is covered by biology, one by chemistry, one by physics, one by engineering, applied mathematics, so on and so forth, then is there a possibility
for such a unification? So as we can say that essentially the idea of philosophy as we have recognized it and so as science, essentially tries to give us a unified picture of the universe or objective reality. So essentially if you want to look at it, look at the debate between Mark and Planck. That's basically the entire roots of the problem comes from for Carnap. Right, right. And references for like social and personal uses of knowledge, something what intent
Sure. I would say that your fellow countryman, I consider him quite very... He has written a few short essays on this topic, on this idea of mother of rationalism or rationalist enlightenment, starting from Euclidean to Medieval to Copernican to Darwinian, Newtonian and then onward. Andrei Rodin. Could you spell it? Sure. Let me type it here.
He's a great philosopher, a Russian philosopher of mathematics. And he's quite very, very generous, so even if you email him, I'm sure that he will give you a bunch of stuff. So Andre, I think, is a good friend. Steve Aoudi. Let me. Of course, Andre Kars, the one that you are reading. These are, I think, give you a kind of indication that what might be the social implications,
social cultural implications of this project of enlightenment, renewed in this Carnapian sense. Of course, Sellars is another one, it doesn't need mention because it's obvious suspect. To be honest with you, I don't have that many more references. Maybe Robert Brandoom in Reason and Philosophy, also but with but the thing that Brandon is a brand is essentially a pragmatist is actually anti-carnapia but but I can see that all of these people in the 20th century particularly in the vein of analytic philosophy even though they
were not purely analytic philosophers like sellers Brandon even Macdowell have been trying to do exactly the same thing but putting different accents on the same topics. Essentially their trajectory is the same, it's that what you might call to be their perspectives are different. Thank you. Absolutely. Has Finley written on Carnap at all? I know Finley was briefly a student of his. No, actually Finley strangely hardly ever mentions Carnap in his books.
By the way, just for the references, when Christian, this is the name, great philosopher. It's actually after N there is a D, sorry, Findlay. Any more? Anything? Justin? You have been very, very silent.
I have some follow-up questions later. I'm not going to shoot right now. Are you sure? Yeah. Okay. What about Joven? I don't have any questions right now. Thank you. Absolutely. So let us start then. And once again, I highly recommend the one that we were just talking about.
It's called, it's of course, you know, it's not part of our course. It is not by any means a work on philosophy of science, but nevertheless it captures a kind of a very sophisticated picture of Carnap without all those caricatures that are usually levied against logical positivism and particularly Carnap. Carnap and 20th century thought explication as enlightenment by Andre W. Karras.
So I mentioned that one of the main ambitions Carnap is to create a constitutional system. I briefly introduced Carnap's constitutional system and what he means by it. So on this account we can say that
we are presented in our power with an epistemically ordered constitutional system that is meant to show that all scientific concepts can be constituted from an experiential basis that no metaphysical concepts can be so constituted this is very important you know obviously Karnak is a chief anti-metaphysician he literally thinks that metaphysics is a pseudo-philosophical problem, and a pseudo-problem for that matter. Little wonder then that there has arisen a received view of Carnap's book of Baal,
the logical structure of the world, that sees it as the high watermark of good old-fashioned empiricism but then again I mentioned that Carnap's and Reichenbach particularly Carnap in that sense are far from a strict empiricists in the Humean sense you know obviously their entire project is predicated on the logical dimension and what logic is. So they are not really empiricists in a Humeian sense.
What do I mean by that? Well, and that comes again in the next session when we talk about his account of induction versus that of Hume. You see, for Hume, empiricism is simply observation. So, if you see a red dot floating in the air, you say that it's like an insect. In that sense, you might say there are basic rudimentary experiences or observations. us something specific about them, about themselves. That it is read, that it is, for example,
a dot, so on and so forth. Obviously Carnap does not believe in such a naive empiricism. In fact, Carnap is far more Szilardian on this front rather than being human in the sense that the red dot floating in the air does not tell anything about itself. You see, for something to be red, it requires to be not blue, not green. requires to be colored for something to be considered as a dot it means that it is not a square it is not a chunk so on so forth so you see this flying red dots
is actually an inferential observation Why I call it inferential? Precisely because when I say that there is a red dot over there, by virtue of inference, it means that for something to be red, it means for that something, x should not be blue, should not be green, but should be colored. And the same thing can be said about the shape of a dot. So in that sense, Carnapian logical empiricism, I would say without any bias with regard to the
project of positivism or logical empiricism can be said to be far more in come in in far more compatible to the projects of rationalism rather than human empiricism the only reason that it's called logical empiricism the empirical bit is that Carnap's believes in Afbau that we can in fact arrive at some rudimentary observational statements or atomic facts. Of course for these atomic facts to have any implication
within the network of other sentences of a logical language, it requires for them to be inferentially connected or caught up in a linguistic network or a linguistic conceptual web. In order for us to say something is red, obviously this needs to be in a network of inferential possible connections. For something to be red, as I mentioned, means that it is
contrasted with the assertion that this is blue or this is green this is white so on and so forth. Is this clear? Yeah, so. So am I right to think that empiricism bit is the fact that the sort of assertions that we can make are sort of bound up with the specific resources. Parological resources, yes. Yes, yes.
Yes, yes. And the whole point, I mean, from now on, if someone tries to ridicule or caricature logical empiricism to you, Alakarna, Reichschenbach, or Vienna Circle, just tell them this, that logical empiricism really has nothing to do with human empiricism. The only reason that it is called empiricism, logical empiricism, is because it believes that experiential agents or persons or humans or observers are capable of arriving at some atomic facts.
These atomic facts are sentences, but they are only sentences by virtue of being embedded within the network of other logical sentences. So, as I mentioned that about the good old-fashioned empiricism, this view
however or moreover seems to fit into Carnap's role as the staunch defender of logical empiricism, a role that appears to assume precisely from the Afbal. It is a view Carnap himself endorses in his retrospective piece called Intellectual Autobiography. Those people who want to check it's page 16 to page 19 and that formed the basis of of bow the so-called call to arms for the vienna circle the primary responsibility for this account of the afbow however must rest with quine
in the sense that why is it called the logical empiricism rather than merely empiricism in the human sense well coin has found occasion to discuss the project of Karna with some frequency in his career. Indeed perhaps the most influential documents of the Quinean reorientation of epistemology. His famous paper Two Dogmas of Empiricism and also Epistemology Naturalized and I really do recommend you to read these
to paper, particularly the first one, are motivated in large part by lessons. Quine asked us to take from what he regards as the epistemological failure of Carnap's project in the Aufbau. So the following two statements from these essays by Quine express the essence of the received view of the of Bauer accepted today by most analytic philosophers, which of course does not mean that it is correct. It's simply the received view and nothing more. So the first quote,
I mean the first view is that a radical reductionism conceived now with statements as units set itself the task of a specifying the sense datum language and showing how to translate the rest of significant discourse statement by statement into it. Harnack embarked on this project in Ofpav. So what is exactly sense datum? Can someone please tell me what sense datum is? I mean, those of you who have been in the CAND class surely know by now what sense datum is. Who wants it?
It's the stream of information. It's a stream of data perceived by your senses. Yes. It's basically, you see, in Kant's Transcendental Turn, we already know that sensation is defined as how thus and so we are affected by the items in an external world and this affect is brought on our what can calls outer sense you can think about it as a neurobiological
call interaction with the ball. In the sense that whatever we touch, we smell, we see, is a sense, is a sensation. Why is it called sens datum? You see, Kant does not believe in such a thing as a sense dato, is transcendentalist. Sense dato means that the way how you doth and sew in a de facto manner, by de facto I mean in the sense that you have been physically influenced or affected by something that you
for example a smell. Now Kant believes in this de facto manner of sensation that all sensations are in fact de facto rather than du jour. Simply they are not by right, they are not by judgment, they are simply by physical constitution. So when we say sens datum it means that we somehow believe How we are thus and so being affected by external items in the environment actually in a very rudimentary basic sense constitute our knowledge about such items.
the word data data is this clear or would this so this data would be in some way shape or form sort of like like obviously related and entangled with but extrinsic to the logical resources which we use to understand it yes well the whole point is that uh kent believes that you know he actually he doesn't himself but for example you know more recent commentators like rosenberg's and sellers uh make a distinction between sensation and sensation
understood the difference between the two is that the first one is de facto it's simply physical you see if you are neurologically being for example in your field of vision you sense a red dot this red dot doesn't tell you that it's a red it's just a stuff is literally a differentiation. And this differentiation ought to be determined. By what? Well, Kant says by understanding. By the upper irony. By concept of red. If you don't have the concept of red,
literally that thing cannot be called red. Of course, for a sentience being like a cat or any kind of a robot, for example, who can differentiate red from blue and green. The field of differentiation is there. But the implication of for something to be red is not there. That's the most important thing. That really is the most important thing. and that's the space of implication for something to be read as i said it ought not to be
white black green and blue so on so forth this space of implication can only be achieved by way of the concept not by way of sensation so since you can still differentiate it via sense but this differentiation as i mentioned is de facto it's not what you might call to be in french or dojour so you would say the sense data is more on the side of seeing of not of seeing as
Absolutely, absolutely, fantastic. but the various and entangled mesh of logical resources which we use to understand that sense data and that's where the empiricism is. Yes, yes, yes, yes. And those of you who didn't get the reference that Christian was making, the
difference between seeing of and seeing as, or what sellers call C1 versus C2. could you talk about Carnap's interpretation of the synthetic judgment it seemed like that was really important for Carnap I don't know if that's what you're talking about now but yes different yeah yeah no absolutely no it is it is and the thing is that I think Carnap ultimately makes a confusion enough about with regard to the synthetic statements in the sense that he does do believe that these statements are synthetic in a Kantian sense however he also believes at the same times in their reduce ability on two two fronts one
that such synthetic statements can be analyzed in terms of the logical space of implications between this statement and other statements pronounced in in thus and so language, but also in another front or in another direction, he believes in the reducibility of the synthetic statements to what you might call to be since that. So to continue with regard to question. Like, you go on, but it's synthetic because, you know, when you have sense data, and then you have logical resources, you know, there's a process to be done.
Yes, yes. It's synthetic precisely because such a statement can only be, by itself, it doesn't mean anything. By itself, it doesn't mean anything. It demands explanation. It means being bold in boldface. It does not mean anything. Precisely because for it to mean something, it means that it should retain or imply such and such logical connections with other statements in the set in the in in in a specific language in which such a sense datum is
implicated is embedded so I was trying to say that difference between seeing of and seeing as you can think about that you know when I look at this glass I see other from from a sensory perspective I see of this glass only the facing side of it the color say side of it and nothing more I do not see the back I do not see of the back I do not see of the sides so seeing of is essentially what you might call to be
sensorial perspectivalism whereas seeing as is what you might call to be a form of inferential integration from an uncritical judgment I would say I see this glass as a glass as a cylinder almost a cylinder and then I see the liquid in it being colored almost green through and through so this is the difference between seeing of and seeing as and I forget what the quote was from but I remember that Carnap really stressed that
the importance wasn't of realizing the hypothesis but the totality of the hypotheses is that kind of what you're sort of leaning towards here yes yes the whole idea is that for Carnap you should And we will talk about this. I mean, the reason that he moves from Aufbau, where he lays out the foundations of logical empiricism to the logical syntax of language, his next work, is not that miraculous as some people, commentators, observe. It's essentially the logical consequence of his project beginning in logical positivism in the sense that for Karna,
any statement which is essentially a hypothesis is already caught up in the space of hypotheses, in the space of collective statements, which is that of language or logical connections. I have a question. Absolutely. Am I heard well? Yes. So would you say that this distinction between seen of and seen as parallels, you know, this divide between what I would say Kantian and Goetian argument exactly about color.
It was exactly about color, right? So Goethe writes this book. And what this tradition started with Goethe, but moving on to Goethe and to some other people afterwards, is that they reject exactly this view of seeing a color as not other colors. So what Goethe was trying to show, that we don't see red because it's not green and not blue. We just see red because we saw red in our life a few times we know that it's red and and this is like i think it's a very different approach to how we recognize objects and a different approach to science as well so would you say that it's the same
distinction as between scene of and scene as i think so i think so yes i i and you see essentially what you might call to be the Godelian lemma is that when we are really enter the domain of seeing as then you have metalogics or metamathematics these different hierarchies of different fundamentally different gradients of colors that you have to go and of course for something to be colored it means that it not ought not to be a haptic sensation or not to be as ought not to be a smell so on so forth so then you see that this whole idea of seeing
gas essentially what you might call to be a hierarchy of inferential webs her first you say that this is red precisely because it is not blue green white black so on so forth then you say that this is color precisely because it is not of a smell okay another hierarchy then you say that this is not such and such precisely because it's not haptic and then the proof of your sentence actually requires you to move from your inferential layer to another inferential layer so on so forth to infinity
Can you elaborate a little bit on this? You said that unlike Canada, Carnapis believe that Ukraine and Ukraine be usable too. Can you, OK, I get your question,
but I think I'm still vague. Could you please like somehow encapsulated in one explicit kind of question that exactly what you think is the difference between Carnap and Kant? So I don't go off tangent because you know that these questions can go off tangent. And so as their answers. Chuggies, can you hear me? I can't hear you. Yeah, yeah, can you hear me? Can you hear me? Yes, now I can hear you. Yeah, sorry, sorry. No, I was...
I'm sorry for you, I thought it was important about it. You were like to... that since datum is something actual as opposed in opposition to Kant and I mean I don't know it seems like maybe that would be something like the linguistic dimension of since data he believed like he yes the whole point is that sensation since that one implies linguistic intervention for cans for Kant the name datum already means linguistic intervention the intervention or the ingression of the categories into our judgments about what this dot so-called dot reddish thing
that we are seeing out there. This is a sense datum precisely because it is a datum in so far as it has been brought under certain kinds of categories that qualify and classify what this thing or stuff we see is actually. however for kent sensation is not a datum by itself it is only a datum in a de facto manner analogically with relation to our own categorical or classificatory understanding of a red dot
something that is dots like but also is colored red so the whole point for can't is that we can't even talk about sensation as a datum we can only talk about it as a datum in analogy in analogy and that's really important in in analogy to our own classificatory way of seeing things as we see stuff out there conceptually.
But for Carnap there is... Sensatum is not essentially wrong. However, what you might call to be the difference between someone like Serb, who actually does believe in sensatum, and someone like Celavus, who does not believe in sensatum, is really the idea that for those people who believe in sensatum, as if how we are affected by external items actually tells us something about them. them, that is what you might call to be an over-extended analogy. Whereas for Kant and Sellars and other transcendental philosophers, the red dot, this is soft, which appears as
red and dotted is only dot and so in analogy, in analogy, this is really important, analogy with our conceptual resources about things in the world, which means that analogy with our external knowledge of the world. you're saying that for Carnap that's not the case. For Carnap it's actually a little bit more of a twisted what you might call to be alternative to both Canton sellers and someone like CERN or any other kind of sense that's on
provenance. In the sense that Carnap does believe that, for example, when we say that it's a red dot, obviously for something to be a red dot means that it ought to be in a network of logical connections with other segments so as for it to be identified as red and dotish. However, Stil Karnap, at least in Afbau and not in his later works, believes that we can in fact reduce the connections between logical statements to empirical observations
atomic facts and that's why he proposes an illogical in logicist empirical stance sorry yes but what is the for Carnap empirical observations what you might call to be empirical observation is essentially our raw experiential data but it's linguistic yes now the whole point that why is it called logical empiricism not empiricism
because it believes it is derived by two poles at the same time one empirical and one linguistic one purely raw and one conceptually formatted but that's why i said that uh carna kind of actually kind of position in that regard is far more in human with rationalist position and transcendental turn rather than empiricism however as we will look at it uh shortly still carnapp's beliefs in some sort of basic sense that what you might call to be an observational
statement that is just there, even detached from other sentences. At least, as I mentioned, this is, however, the early view of Carnap. He will abandon this in his later works. So Valentin, you wanted to say something, my apologies, with regard to Godel and seeing as. I think I will maybe next time I will think about it a little bit, prepare a better question.
OK, but I mean, feel free. Don't worry about if the question is not well-crafted. We are just thinking together. So I mean, feel free. Just that this hierarchy is not like the hierarchy you mentioned. It's clearly, I don't know, it's not certain that it actually exists, it's actually used. is not used it's what you might call to be it's essentially a logical hierarchy or an inferential hierarchy you see the whole idea of the godelian lemma is that we cannot prove the meaning of a
sentence within its own hierarchy right we cannot say what and of course i'm valentin the whole point is that i'm literally trying to over extend the godelian analogy and this is an analogy and so forgive me if i am really doing over extending the analogy so you might say that godel if he was kant or brandon or carna he would have told you that in order for us to say this is red we cannot determine the meaning of this is red this assertion for us to determine
this meaning we need to write at a different hierarchy what you might call a web of inferential connections whereby we can say that this is not red precisely because it is not blue it is not green, it is not so on and so forth. Then we say that, okay, this is colored, but then the meaning of this sentence is also underdetermined. Precisely because in order for us to determine the meaning of when we make the assertion that this is red, we should have some sort of access to a different layer or hierarchy of inferential connections.
For something to be colored means that it is not haptic, it is not smelly, it is not you know bulbous, so on so forth. So you see, this web of inferential connections can be said as these kind of Godelian hierarchies that actually do verify from an outside perspective, from a higher position, they verify the meaning of your sentence. I think exactly, like, you're talking about probably what we can call a type system. or something like that. And I think one way to look at what Gödel does
is that he circumvents this type system because he encodes a logical proposition as a type integer, right? Yes. And this is exactly where it all goes to. However, Valentin, now that you brought the type system, I would say that there is actually a more generous way of looking at the Godelian hierarchies. Essentially what generates the hierarchy in the first place is what you might call to be an untyped system because types cannot verify themselves. So you need to have an untyped process undergirding such type hierarchies.
And that's what you might call to be the inferential web. And the contrast which would I give to this way of explaining cognition is the neural network. So if you have this neural network, which can recognize a number of features, so like a typical neural network, you show it to something and it fires up, like this is red, this is a shoe, this is something else. And this is all features which are for this neural network, not, they're not typed, this adjust probabilities of the thing having a particular feature. And this feature works, I don't know, like a tag on Instagram.
It does not, it does not adhere to any logical relationship with other features. Then you can add another layer if you want to. So for example, if you really want to find out the color, you might look at all of the features which were returned by this network and see the most probable color or something like that. But until you actually have this question, you just have these features which are not typed, which do not adhere to any logical schema until it is useful. I absolutely love the way that you expanded this problem and literally I would say that
this to me is too racist for AGI it's essentially the idea of the conceptualism versus what you might call to be a statistical, a statistical paradigm. The way that you should, however, think about this, that the statistical paradigm can in fact model the conceptual paradigm. Basically, after Boltzmann, we know that everything can be in fact modeled on the statistical probabilities. They are not simple possibilities or probabilities in the old probability theory. They are a statistical analysis. Now, what is really happens at this point? You see,
you should understand that we humans are resource sensitive from a computational perspective. it we cannot do a lot of a statistical analysis the way that for example a computer does and even a computer cannot do that in fact for huge amount of data because the computational costs increase and all the you know dangers kick in the whole point is that I am willing to accept your proposition, your alternative, with a caveat. And the caveat
is that conceptualism is far more computationally friendly from a cost perspective than something like a massive neural network in which everything is a statistically analyzed and a statistically indexed you see concepts are compressions from a computational perspective and that's really the most important thing what what do you think reasoning is reasoning or inferences are actually way of compressing selectively compressing data so as we can arrive at the connections or nodes between different possible scenarios so you say for example
learning machine can do that but let me tell you that for such a learning machine let's give it the giant big data. It obviously, even maybe with the progression of computers, it might not find some sort of dead end, but nevertheless it's a theoretical position in in the sense that if you have just too many data at some point the computational cost becomes unwieldy and simply the whole idea of cognition becomes impossible.
It is not that I am saying that conceptualism is different from a statistical point of view. I would say that in fact concepts can be a statistically analyzed and in terms of a statistical learning or formal learning. All I am saying is that concepts give you a far more computationally friendly alternative than simple statistical analysis of the one we use in neural nets. Language. This is really one of the greatest insights of Chomsky in late 60s, that he
he started to think about language in terms of a statistical analysis or statistical learning. Of course Chomsky's later work began to diverge from this paradigm precisely because Chomsky thought that not every level of language or logical connections can be a statistically in fact model. But I would say that putting Chomskyian objection aside, I would say that conceptualism is a and essentially the idea of language or inferential use of concepts
is a far more computational, optimal way of looking at these problems. Of course, we can do every of these things from an statistical perspective. But there are also stuff that we just cannot do at level of neural nets. Of course, you might be able to do as the technology grows and the computational resources grow, but this is, as I mentioned, this is a theoretical problem. Computational cost needs to be taken very, very seriously. Reza, okay. This argument applies to neural networks, but it does not apply to this kind of positive knowledge.
I actually remember the passage from Phenomenology of the Spirit, where Hegel actually says that every effect of every moment of cognition is at the same time a negation of all of the other colors, for example, but at the same time it's something positive. It's not just a negation. Hegel means by negation a determined negation. You see a determined negation is essentially a conceptual negation. no no no in this particular situation he I think he talks exactly not nothing not in the same sense he negates most of the time okay it doesn't really matter I will send you the specimen like I will set on it for two hours anyway to be honest with you I think that this is really a fantastic
idea that you have. And let me tell you that even though I do not disagree with it, I am fine with it, meaning that I don't lose sleep over it. Literally, if it can actually happen, I'm fine with it. Literally, you need to work on this one. This is really an incredible philosophical question. Let me give you another example, which is not statistical, but at the same time positive. So for example, to get back to math, and I think this is, I don't know, so in math we have these structures. So for example, I don't know, groups. If you do any kind of abstract algebra or things like that,
what happens is that we look at some kind of object and we recognize that it is the set of objects form a group, for example. Yes, asymmetric preserving. But it does not always happen so that we look at all sets of structures in abstract algebra and think, is this a field, for example? Is this something else? No, no, yes, yes, and arrive to the conclusion that it is in fact a group. No, we start with an object and we immediately go to the conclusion that this is a group not based on the fact that it is not something else, but based on the fact that it adheres
to some kind of axioms of groups. And one of the things that it is important... This is what Hosterl calls adumbration. One thing why this is, I think, is important is because it is not clear. However, it is not useful for a mathematician to think that all of these structures form some kind of very specific hierarchy. So we can start, for example, asking, is this topological or is this something else, and then go down, go down, go down. I think if this was exactly the process of doing mathematics, mathematics would not go as far as it actually went. And some of the things would be possible, including Godelian theorem, which exactly
works with isomorphism between structures which do not really belong to the same type. Yes. Now, I would say... So, example of... Please go on, please go on. Yes. So here we have this example of recognition, pattern recognition, which gives you some kind of positive idea what thing is without first negating it against every other. Yes, that's important, that's the whole point. You see, I believe in the Hosserlian Adamperation and Kant calls it the difference between an uncritical judgment and a critical judgment. So what is the difference between what you might call to be
recognition in itself in a Hegelian sense and recognition with determinate negation? So you see when I for example there is this glass half of it water I put a pen in it so I look at it From perceptual perspective of what you are talking about, simply recognition in itself... We don't see the glass right now. Oh, well, sorry. The pen looks as if it is bent, okay? So Kant calls it uncritical perceptual judgment.
It is a judgment, but it is uncritical. Why is it uncritical? Precisely because it is not determined in a Hegelian sense, versus other kinds of elements that might actually be responsible for giving rise to it. A critical judgment is when I say that this is an optical illusion, that it appears to be bent. This optical illusion requires a fundamentally new class of type systems and computational untyped processes to generate that. a neural network cannot learn for example we can we can train it we can train it but what about
other scenarios you see this is the difference between our science of the world and a neural machine and let me tell you that I do not believe in such a distinction between human machines machines and human dichotomy I would say is essentially a false dichotomy so that's not I am talking about I'm simply talking about that neural model is that it can only gives us analogically speaking the uncritical perceptual judgment of Kant in the sense that it learns that there is a pen in the water and this is a glass and so on and so forth, however the pen appears to be
bent. But the critical judgment, which is essentially an infringement and logical, requires a far more kind of training i am not saying that it cannot be a statistically decomposed or model i am simply saying that it requires a new form of a statistical training okay um so the impression i get is that uh this um this logical language which optimizes calculations twice like first it optimizes as a language and then as a set of logical links between sentences it's in my opinion it is the most
a posterior thing which can ever be like it's not even it's not even acquired with language it's acquired with like training it's acquired with um well you see when i'm talking about language and logic i'm a pragmatist by training in the sense that what is language there is no such a thing as language out there language is simply in a persian or brandomian sense is a conversation i make an assertion this is does and so and then you query like a computer You say that why is this that it is thus and so? So assertion and queries is what might call to be the core of the pragmatic view of language.
Language is not just something out there like this kind of naive Platonism. It's essentially a form of interaction among a multi-agent system where we can triangulate our logical connections plus those kinds of pattern recognition together and then the matching of this in a multi-agent system what you might call to be giving us an optimal an optimally invariant picture of what this whole thing is about this is what is actually carnav calls objectivity
It seems to me that in practice, this kind of perfect hierarchical language has never existed, and not a single science has moved beyond this constant triangulation of hierarchies, even within the same problem when you look at it, like in physics, when you do Lagrangian or do something else. You see, Valentin, it's not perfect. Obviously, it is a messy problem. The whole idea of objectivity, even for Karna and Kant, is essentially a messy problem. When we are dealing with different experiential agents who are interaction, who are in interaction
with one another via a kind of language, this problem is far from perfect. It's messy. However, just because practice, what do you mean by practice? You mean computer science? computer science i i can tell you the computer science if it could solve these problems it has already solved the problem of agi the very fact that it has not solved the problem of agi it means that it is not having the enough resources to tackle this problem um could you say a little bit more about yes well the whole idea is that at least since the syntax of
language which is the work after of bow connaps believes in the idea that language he actually tries to somehow retain some with can a spine in pragmatics in the sense that for Wittgenstein language was nothing other than its use. It's essentially what you might call in a system engineering perspective, language can only happen within a multi-agent system. Language is essentially multi-agentia. So Carnap's believes that the only way that we can arrive at something called
objectivity with the understanding that we do in fact have different individual streams of experience about the world. I see perspectively this, you perspectively see that, you might actually see a different thing. The only reason that we can arrive at objectivity, and this objectivity is what you might call to be a judgmental or justified invariance, an inferential invariance, something that does not change according to our perspectives, but is preserved as we change our perspectives, regardless of who we are.
He believes, Carnap believes that such thing can only be possible by what he calls language as the organon of the structure. language simply stabilizes and retains a structure of our streams of individual experiences as projected onto the external world this i mean was kind of like kind of what i wanted to comment on with like your and valentine's conversation. Because this, you know, I do really tend to be kind of, you know, sympathetic to Valentin, but
you know, I think it seems like, you know, your point is kind of along the lines. You know, where this is like the other side of logical empiricism, where, you know, these linguistic resources for creating things are contingent upon how they interact within the mesh of languages. Yes, yes. Or what you might call to be a medium, which is called language. And by that, Karnap absolutely does not mean natural language. Right, right. Language in general. And then these non-perspectival critical judgments would be arrived at through some sort of process of explication. But I mean, I feel like, you know, in tandem with what Valentina is saying, this process of explications is necessarily a posterior, right?
Because, you know, it doesn't happen in some ethereal realm, but has to do with actual concrete linguistic resources, which when combined with sense data is able to give a non-perspectival judgment upon that data, correct? Yes, yes, yes. Can I give another story? Absolutely, you can, you can. And then we will have a rest and then we will come back. Yeah, this is not to interrupt the conversation. No, no, no, absolutely. I really do appreciate your insights. So what I really like when we're reading, I think it was Kuhn about paradigms. So he has a story about how he would read a lot of Aristotle and his physics would make zero sense to him.
He was like, what is this nonsense? How Aristotle could be so stupid? And then he tells... Well, Aristotle was stupid. Not as a philosopher, but as a scientist. So this is not what Kunt says. Kunt says that for him it made zero sense until he looked out the window and suddenly everything made sense to him. So, you know, there was one language and there was another language and he explains Aristoteles' language beautifully, fascinatingly. And I don't know, I think it's, I don't know, I started to respect. Well, Valentin, I think you are overextending the metaphor of language here. I think this
language is a stylistic language rather than the language that Aristotle was using. No, Aristotle, I mean very specifically that Aristotle has this language which is based on the fact that each thing has its place and things like that, like whatever. Well, that's a philosophical language. It is not, however, a natural language which Aristotle used. You see, the reason that Aristotle was able, in fact, to arrive the objectivity, a semblance of objectivity in the first place, was because he was simply, what you might call to be a computational agent, interacting with other agents in his time via something called
language and this language is not a philosophical language is simply language okay yeah so the way Kuhn arrived at understanding Aristoteles language in whatever sense language we can say it was not by directly translating the concepts from one to another because it would not work but it was exactly by looking out of the window of just like looking at the object again. It's not even like objects, it's looking at something again. And like so I think this look out of the window is this moment when you
So in this point of view that you're explaining, it's like first you choose a language, and then with this language you start seeing the world. Yes. But like... You don't choose a language. Obviously for us, language comes naturally. However, with science, we do in fact choose the formal language. I think this has a lot to do with that other side of logical impure, not just sense-data and logical resources, but then the posterior nature of the logical resources themselves. I mean, it's a very interesting story. I mean, I don't know the story, but just from
what I'm hearing, it seems like what you're talking about is through some type of know, being spawned from this looking out the window. These different languages were able to interact. And then those resources, which are conceptual, interacted, and thereby new insights were reached. And I think, you know, correct me if I'm wrong, I think this is why Reza brought up neural networks, you know, is because, you know, yes, you know, we are able to draw statistical inferences from various sense datum,
you know, but then the question of conceptuality is introduced, which means that, you know, the explicated axioms which we have to be able to arrive at various conclusions from sense data are necessarily entangled in this a posteriori mesh of other logical resources. And from there then, you know, conceptually, we can reinvigorate the axioms which we're able to employ to draw various statistical conclusions. Okay, with that, sure Meredith, you say and then we will go.
I'm still trying to grok all the implications. My apologies to intervene. With regard to your last question on the sidebar, I said the language enables objectivity, it's not objectivity as such. So I'm just talking, so I'm thinking about this notion of a posteriori and although I I'm really interested sort of in the emergent, you know, language emerges in, I mean, a recursive, it's not a posteriori, especially in the case of like a neural network where the refinement of what the object is comes as there's more sense datum apprehended by the system.
So I'm sort of interested. It's only sense datum. You see, first of all, we do not know what the machine actually does. The only way that we do are capable of looking at something like neural net and saying what it actually does is in analogy to our own theoretical and practical reason. But the language, forget neural net, but the language that we use to describe red, is constantly being refined. There is no static language. It is constantly becoming as more information enters the system. So I don't know if it would be a posteriori really.
It doesn't exist. Well, I do not think it is a posteriori. I would call it an implicit a priori, in fact. uh however let's valentin uh was supposed was going to say something when i said something nasty probably all neural networks it's not actually as a you know i i know where is this uh myth about neural networks not making really sense it's just that you know you can you can can look at specific, it's not neurons, but specific points in the matrix multiplication that is going on there.
And you can actually recognize the things that these specific points recognize. But the thing is, so for example, if we would say, like, this person brunette, and he has brown eyes, and he has come to a conclusion that it's someone. But each of us think, I think what the concept does for us is that if you look at someone and we say, he is this probability he is like brunette. But then, because we have this concept, it's no longer a probability for us. He is exactly brunette. and then it's kind of easier for us to go on.
Yes, yeah, I mean, this is essentially what you might call to be one of the greatest insights of Ludwig Boltzmann, the person who put a statistical probability, what you might call to be modern notion of probability, on the table. According to Boltzmann, you know, a statistical probability cannot, you cannot talk about And a statistical system or a system understood or analyzed in terms of statistics, in terms of what it actually does, it's just simply a statistics. When you say that the system is doing this, then you are resorting to what you might call to be macroscopic concepts.
And the macroscopic concepts are not statistical. They can be statistically modeled. But we are entirely in a different domain. Can I maybe ask one last stupid question? Let's have a break right now. And then we have brief questions. And then I have to literally start. Yeah. No, put a chain on all of us.
Let's just skip the questions for a while and so I can, because we are a little bit behind, let's just let me give you a very brief account of the Carnapian view and then questions maybe if you can take them, write them down, and ask them next session. Maybe if there's time just at the lingering end of class, I kind of wanted to just bring something up about the Kuhn analogy that Valentin used. Sure, sure.
right now in Michigan. Yeah, I am. I'm adjusting. Where are you exactly? I'm in Kentwood, which is just outside of Grand Rapids. It's on the... I see. I think we can start. Maybe go ahead. Sure. So as I mentioned, the language which Carnap adopted as a starting point was not in fact
a senstatum language in the narrowest conceivable sense, for it included also the notations of logic up through higher theory. Carnap's starting point is a very parsimonious, however, in its extra logical or sensory part. In a series of constructions in which he exploit the resource of modern logic with which ingenuity, with much ingenuity, Carnap succeeds in defining the wide array of important additional sensory concepts which bought for his constructions one would not have dreamed were definable on so a slender base.
He was the first empiricist who not content with exerting the reducibility of science to terms of immediate experience took serious steps toward carrying out the logical reduction. To account for the external world as a logical construct of sense data, as for example in Russell's external world program. It was Carnap in his, you know, logical structure of the world who came nearest to executing it. What then could have motivated Carnap's heroic efforts
on the conceptual side of epistemology when hope of certainty on the doctrinal side was abundant? There were two good reasons still. One was that such constructions could be expected to elicit and clarify the sensory evidence for science. The other reason was that such constructions would deepen our understanding of our discourse about the world, even apart from questions of evidence. It would make all cognitive discourse as clear as observation terms and logics. Thus, on the received view, Carnap is seen as taking his epistemological inspiration from a strict empiricism.
We begin with an epistemologically privileged language of sensation and seek to exploit the resources of modern logic to make good an empiricism longstanding promissory note that this language is conceptually rich enough to capture all of scientific knowledge. The project extends our understanding of the world by simultaneously clarifying what we are saying when we engage in theoretical discourse and indicating what the empirical evidence is or must be for such claims to be true. In objection to Carnap, Quine argues, however, that this phenomenalist and constructivist
program in the Appao does not, however, succeed. Not only is it impossible to derive the totality of science from sensory experience, scientific discourse cannot be couched in exclusively observational terms even with the aid of the logical system of Whitehead and Russell for us quine argues in two dogmas of empiricism highly theoretical sentences of
of empirical science do not have sensory import individually. Underlined rather only large portions of theory have empirical consequences. And hence only large segments of theory can be translated into empirical observational terms. The objection Quine raises is not a bolt from the balloon. However, Quine sees the objection as implicit in a change of method of definition found in Carnap's book itself. The problem comes right at the crucial moment for the phenomenalist project.
When Carnap seeks to move beyond the solipsistic world of phenomena, or in Carnap's terms, of the psychological realm, to the physical world, at this crucial stage, Quine notes, Carnap drops the physical world, sorry, at this crucial stage, Carnap notes, Carnap drops the project of giving explicit definitions. Rather, he provides methodological principles for the mapping of qualities from the private realm of individual experiences onto physical space time points, external world.
Thus, at the crucial moment Carnap turns his back on the empiricist only allowable methods. as Koine objects in page 40 of the book 2 Dogmas. Citation from Carnap. A statement of the form quality Q is at point instant XYZT where according to Carnap's canon to be apportioned truth values in such a way as to maximize and minimize certain overall features. And with the growth of experience,
the truth values were to be progressively revised in the same spirit. I think this is a good eschemization, deliberately oversimplified to be sure, of what science really does. but it provides no identification, not even the sketchiest, of how a statement of the form quality Q is at point XYZT could ever be translated into Carnap's initial language of sense data and logic. The connective is at remains an added undefined connective.
The canon's counsel us ensues, but not its elimination." End quote. This feature of Afbaho is unavoidable and finds and indicates that there is, in fact, no way Carnap could have succeeded in the task he set himself in the book. Rather, we must abide by the lessons inherent in Carnap's use of these astrological canons and give up the projects of sensory reductionism, or sensedatum reductionism. This, then, is the received view of the logical structure of the world.
It was the most ambitious and successful attempt to use the resources of modern logic to carry out the reduction of all scientific discourse into the terms of immediate individual experiences. The principal legacy of the book is that it failed in this reduction, and that it failed not merely in fact, but in principle. That is the important lesson of the Akhbaw, that Carnap so rigorously formulated the empiricist thesis of reducibility and used the logical resources available to the empiricists with such precision and ingenuity that it became clear that the failure of Carnap's attempt at providing a constructional system is symptomatic
of the impossibility of the program as a whole. This view is not only Quine's but also Putnam's. In Reason, Truth and History, page 181, Putnam says with regard to Carnap's of Bao, he says this in quote, the claim that statements about material objects are translatable into statements about actual and possible sensations seems as a matter of fact to be false. Careful logical investigation of this claim starting with the work of Karp and the Vienna Circle
in 1930s convinced the phenomenalists themselves that the claim was unfounded my apologies so then if the overall project of carna does not indeed follow russell's lead of the program
of external world where we can say that there is an isomorphism between logical statements and those of atomic facts of experience or observational statements, then how could we possibly make out Carnap's rejection of metaphysics? And you know that Carnap was anti-metaphysics. after all rejection of metaphysics simply follows from the doctrine of verification that is to say our inability to consider the concept of say reality in the metaphysical
sense from the experiential basis of the epistemological system is how we show that that those concepts are in fact illegitimate if Carnap is to have an argument against metaphysics he must be verification is and hence a Russellian empiricist after all this argument is not correct however and it is instructive to see why Carnap does not follow Russell's lead on questions of the relation of ontological issues to epistemology at all. This is indeed the central issue that divides the two within the text of the Avdow itself. In the very section, section 176, in which Carnap makes the point that metaphysical concepts
are not constructible within the epistemological constitutional system that he sketched out at the beginning of his work. He notes that this seems to agree with Russell's view, but he goes on to say quote, this does not seem to be consistent with the fact that in Russell's questions of the following sort are frequently posed which, in parenthesis, regardless of how they are answered, parenthesis close, implicitly manifests a realistic conception, whether physical objects exist when they are not observed, whether other people exist, whether classes exist, etc.
End quote. Precisely this sort of question, which is given an answer, rather than rejected as meaningless in Russell, is what Carnap seeks simply to set aside. For Carnap, unlike Brasso, the rejection of metaphysics is not governed by the acceptance of an ontology of objects, of acquaintance, and a method that shows how to do without anything else. Carnap seeks to reject all questions of ontology. Epistemology has nothing to say about such questions. In fact, within Carnap's philosophical framework, in the logical structure of the world, their role played by the argument of section 176 is minimal.
Carnap takes no great pains in showing that metaphysical concepts are not constitutable in the epistemically ordered system. This is because if they were, they would be scientific concepts. But Carnap takes it as a presupposition of his dispute with metaphysicians that both sides agree that metaphysical concepts are not scientific. The question at the stake is what sort of concepts are employed in philosophy itself. Carnap metaphysicians are in the business of employing metaphysical concepts such as reality or essence,
essence, in giving a philosophical interpretation of science. Carnap is dissenting himself from this point of view in his anti-metaphysics, and for this task his epistemology plays only a limited one. We can see this, for example, from the fact that the physicalist constitutional system that Carnap takes as an alternative is also claimed to be metaphysically neutral. Nothing internal to the business of any constitutional system for empirical science is playing the major role in the rejection of metaphysics. The important role is played, rather, by the logical language in which any and all constitutional systems are cast.
The language of logic provides the metaphysically neutral vocabulary within which any and all philosophical discourse can be represented or presented. The importance of the neutrality of logic is already given voice in Carnot's claim of neutrality in section 5 of Alfao. of Baal. We have already noted that Dirk Karnab claims that the very language of constitution as made out in the of Baal is neutral within the Nookantian metaphysical thesis of objects created in thought and the realist thesis that objects are merely grasped in thought
but exists outside of thought. Hocanap, this dispute is misguided because when we get clear on the language available to the philosopher, engage in constitution, neither of these theses of a constitutional system. Sorry, neither of these theses is formulation at all. All we are left with the sequence of definitions of a constitutional system. This is the whole content of the epistemological perspective. No further external or interpretive questions are raised, and none need be answered.
Logic thus provides the philosopher with two things necessary for the rejection of what we call metaphysics. It provides a general vocabulary applicable everywhere, and that conditions the very possibility of judgment on any topic. This being the case, logic neither requires nor allows a deeper explanation in terms of metaphysics. The role of logic in conditioning the possibility of objective judgment cannot be explained through appeal to some particular doctrine, whether metaphysical or empirical, logic is universal and hence fundamental and unavoidable for any and all theories.
It is therefore universally available for exploitation by the philosopher, in this case the philosopher of science. also provides a new means of capturing the intent of traditional philosophical debates. Thus, as we have seen, old philosophical terms such as constitution itself, which is a very Kantian term, can be given precise meaning within the language of logic. The universal applicability an expressive power of the new logic does all the serious work in the rejection of metaphysics carnapp's most perspicuous a statement of the swill occurs not in the about but rather
in the pseudo problems in philosophy section one quote from carnapp it has frequently been expressed that the epistemological question grounding or reduction of one cognition to others must be differentiated from the psychological question of the origin of the content of a cognition. But this is merely a negative determination for those who do not want to rest content with the expression given reducible foundational or the like or those who do not want to use these concepts in their philosophy the task of epistemology has
not been formulated at all the goal of the following considerations is to formulate this task the definition of epistemological analysis without having to use these expressions of traditional philosophy. We only have to go back to the concept of logical implication, the conditioned relation, as it is expressed in if-then sentences. This is however a fundamental concept of logic that cannot be rejected or even avoided by anyone. It is indispensable in any philosophy in every
branch of science questions Theo, you are twitching. You better spill out your question or you will die. I'm just trying to think of a way to phrase what I was going to ask what you just said.
And I don't have a very clear way of doing that right now. I'll let other people ask questions if they do have some. Sure. If no question, I will continue. So what you just said, it all refers to the first version of Karnaf.
This is... Yes, but you already see that the seeds of the later Karnaf is already in the Akhbal. In the sense that, insofar that he's not a strict empiricist, insofar as... He rejects the question of metaphysics in favor of the question that can only be posed in logical terms broadly understood. And of course that in Aufbau we should understand that Carnap believes that mathematics is reducible to logic.
In later works, however, Karnam no longer believes in this and that's why I said logic in the broadest possible way understood. In the sense that you don't even need to have such logical reductionist view of mathematics. All you need to know is that essentially all relations within mathematical, computational
linguistic terms can be captured logically or proto-logically if you want to use a Girardian more contemporary way of looking at logic Doesn't Carnap have a specific definition of logic? He sort of challenged this fundamental notion of the truth of logic, so he develops his own sort of conception of what logic is? You see, Chagis, in earlier works, at least even before Carnap, he's fundamentally free
Okay? But with the logical syntax of language, he actually thinks of logic in a plural sense, as an unbound, what he calls an unbound ocean. The language is essentially what logic is. And language, by that he does not mean natural language, he means a formal language in which the relation between symbols the transformation rules between them is explicit and we can have many languages in the same vein that we can have so many logics hence i would say that
a later Carnapian view of logic is essentially plural. It's not freaky in any sense. I have a soft spot I want to push on, but I'm not... I'm so afraid of your soft spot, because usually they are the most dangerous ones. Go on. Well, I just, I don't exactly understand how language is supposed to be plural. Or, or logic. You believe that, okay, let's, let's look at the constitution of our own natural languages.
Look at, for example, Arabic, Farsi, English, Spanish, French. Or Korean for that matter, which is fundamentally different from these in the European languages. Then you notice that some concepts actually function very differently within the web network of logical connections among other statements or assertions that can be made about other
words. For example, even think about a very basic perceptual word like green and blue. Well, I think that you know that already Japanese and Korean and some other languages do not have in fact green and blue. In fact, in Farsi also we do have green and blue, but we have words that somehow bunch together the green and blue, the word green and blue, like Nili. My force is really rusty,
so please, Safideh, if you find this atrocious, please remind me. So the whole point is that obviously words are not the names of things, okay? They are not labels. they are assertions when we we only say this is green by virtue of an assertion that this is green it's a piece of judgment the percentual judgment so obviously when we do not have the word blue and green in our language then we have different kind of logical connections
even words or assertions you can even go beyond this and think about languages which fundamentally have different kind of computational or proteological processes that can be plugged together into a whole language this is why I would called logics as plural versus logic because you see I feel the word logic it's not that logics plurally understood are not logical in the singular
universal sense it's a rather this is a dilemma inherited from Frigge in the sense that they at least since the time of Frigge and Boone they understood logic as that for example the connections the logical connectives and the logical behaviors that are responsible for the diversification of the logical notations are exactly the same and can be applied across the board. But of course we can make new computational processes. We can even see the logical connectives that are shared among different frameworks of logic
generated by different computational processes by different kind of logical behaviors and that's what I mean by logics it's not that logics essentially gives you a different senses of logic no it's the same reference obviously all of these refer to logic as such or what we mean philosophically by logic but also diversified in terms of what you might call to be the methods of generating and explicating logical connectives and behaviors so like Claude van de lois sort
of talks about this in spatial prepositions where if you say like something is behind something, it means different things in different languages. Is that kind of like in one language, it's like you wouldn't be able to see it because it's too far away, but in another language, it means it's behind something. Yes, yes. Is that a different logical behavior? Yeah, absolutely. I mean, great work has been written on this front, special prepositions by Claude Van de Loew. I think I have mentioned it in our previous classes. where Bandoloid talks about for example a spatial you know connectives in at on so on and so forth
and of course the meaning of sociological connectives are far from obvious across different languages You know, the way that we use the word in in English and the way that we use the word in in French is actually quite different. It's precisely because we are using different kinds of mechanisms which make explicit the meaning of in to some extent. But you can even go further. render the natural language to a formal language to a bare minimum formal
infrastructure and then you see that such formal elements inside your language can be actually generated or even defined differently Paracassism logic versus consistent logic Girard You know some structural logic versus free or boo Georgie Japanese computational logic versus you are so on so forth Um, I noticed also that when Carnap Carnap there's an equivocation incarnate between material and formal implications um do you think that's
significant in any way or interesting yes that is superbly interesting but let's not go that way yet because that that requires a lot of uh extra i i i try to cover this but did you read uh the paper that I suggested by Andre Karras, his paper, his essay on Sellars, idea of rationality, where he actually talks about this, the idea of material versus formal inference, precisely because many people in philosophy, in analogy, philosophy thinks that Karnab really did not
understand what material inference is then you know what is exactly material yes the asphalt is wet the sky is cloudy my material inference is that it has been rained You see, material inference is a kind of inference that what you might call to be falls in the realm of Kant's transcendental logic in the sense that what you might call to be the combination
or synthesis or integration within formal logical resources and my perceptual judgments. All logic however does not require perceptual judgments or perceptual inputs. So one of the, for example, one of the greatest I would say and this has brought up by Robert random and different works one of the greatest what you might call flaws in the program of artificial general intelligence is that they still have not figured out what actually material influences so you see it by polarity
between neural learning that Valentin was talking about or simply formal syntax manipulation, formal languages. Material inference is when these two are being combined. I have a question. But I wrote it down so it's surely quite nasty. Go on. Okay. So it seems like the empiricism and logical empiricism
make sense into these three domains. Maybe not exclusively, but just things that can be done. One has one foot in intuition and understanding. That is, it has axiomotized forms of sense data which is isomorphic to observational experience. It means that different axiomatizations of base observational data give different inputs for further conclusions to be drawn from the other logical resources at stake. Two, the conclusions which can be drawn from the axiomatization of the form of base observational data is contingent upon the logical resources which we have to draw from the axiomatized form of sense data.
constitutional system yes right allowing discrediting revising and testing testing of resources within the mesh of logical resources which we have available to us three reducibility to experience means that both of these you know both you know the axiomatization axiomatized forms of observational data and logical resources from which we can draw a conclusion from those are ultimately both isomorphic to observational data. Both the axiomatization of the form of observational data and the logical resources from which we can draw conclusions are a part of what is observable.
Yes. And so in the second case, what would you call it? I'm actually... I'm almost done, I'm almost done. No, no, no, I was going to actually, I'm going to answer you this by way of how I will develop this whole idea of a Satan view of Karna. This is called Erleb. Erleb. Erleb is a short German word for elementary experience. And literally what you are talking about are airlines. Axiomatically indexed, but also observational. Can I get to my questions?
Yes, sure. Go on. So in the second case, what would we call the interaction of different logical resources which allows their testing, discarding, revision, etc. Is it right to call this interaction empirical or a posteriori? If not, what is it? And there's a second question. It seems to me that especially from the third case, that is the reducibility to experience of both the axiomotization of the observational form and the logical resources which we have to draw conclusions from those, there's really not much purpose of talking about sense data itself, namely because we simply do not have the resources, be it logical resources or the cost of computational resources,
to explicate the content of the totality of the streams of sense data, but only from the axiomatized forms of the sense data that we have available to us. Both the observation of base observational data and also the observation of logical resources themselves are both afforded by the axiomatized forms of observation itself. Putting again emphasis on the resources for logical implication and the conclusions reach from the interaction afforded by them. Is both sense data and logical resources subject to not observation, but the axiomatization of base observational form and subsequent logical resources which allows to draw conclusions from observables? then is the innovation of the resources of logic proceeds solely from the
consequences of a plural logical process itself yes I would say that this is much more in tandem with the later work of Karna you see at least in our foul if not enough power in earlier works of Karnap he does in fact believe in some sort of isomorphic between the axiomatized elements and the observational statements namely elementary experiences the air legs but yes yeah yeah absolutely no so
this is what you are actually we're talking about is more of a later Carnapian development but in off all you do not fundamentally get this precisely because of power is made or built upon the reducibility thesis in the sense that we can do in fact reduce our logical statements to some axiomatic statements which are isomorphic to elementary observational experiences and hence the thesis of logical empiricism
let me let me let me start this section before we go I hope that we don't run out of time okay Okay. Okay. So, let me give you, let me tell you that there are at least three interpretations of Carnaps of Baal, logical structure of the world.
Two of them are considered to be old interpretation or old theses and one of them is more a recent one, a recent thesis. So according to the traditional interpretation exemplified for example by Quine, good man in the structure of appearance, required in two dogmas of empiricism, and retrospectively by Carnap himself, the aim of the Akhbao, these are the old theses, the aim of the Akhbao is to support the following theses. So these are both what you might call to be the old theses.
of our bow. Every scientific sentence can be translated via explicit definitions into another sentence that consists solely of logical signs and terms that refer to the given, the sense that, such that in each of the underlying definitions the defined expression and the defining expression necessarily have the same extension. The second thesis is that, is an old thesis, every scientific sentence can be translated via explicit definitions into another sentence that is purely a structural, that is to say
It consists solely of logical signs such that in each of the underlying definitions, the defined expression and the defining expression necessarily have the same extension. So what really sets apart these two different old theses is that the first one is about the reducibility to the sense that the given the second one is about reducibility but the emphasis is put on the idea of logical structure and how it can be understood by extension as isomorphic to our experiential encounters with items in an
external world the new thesis is consists of two parts and is rather different the The new thesis says that every scientific sentence can be translated to an epistemically, sorry, to an empirically important, to an empirically equivalent one which consists solely of logical mathematical science and terms that refer to experience such that, part of it the translation image expresses a subject invariant constraint
on experience the so-called objectivity pieces now let me go over the old thesis for you the one that has been criticized by the likes of hume freedman good man and so on so forth let me share my screen
Well, let me first read a few things and then... So, as I mentioned, that the old interpretation of Baal is essentially a Russellian interpretation. It's precisely because, why is it Russellian?
Because Russell believed that the laws of thought correspond to the laws of the observable universe. such that, according to Karna, we can in fact obtain an isomorphism between logical statements, axiomatically defined, and our elementary experiences. The kernel of logical empiricism. So, while the old interpretation considered the Alf Bauer as the result of applying the
then new logical means of Whitehead and Russell's Principa Mathematica, the traditional empiricist phenomenalist program, the second one, the one that I mentioned that every scientific sentence can be translated via explicit definitions into another sentence that is purely structural that is to say which consists solely of logical science such that in each of the underlying definitions the defined expression and defining expression necessarily have the same extension so this definition essentially understand the of pao as being influenced by the new chanting tradition of Kazirer, Hermann Cohen, and Paul Nator, and emphasizes its neutrality with respect to all traditional epistemological positions.
While the intention of the Aftbau, according to its traditional interpretation, is to show how scientific claims may ultimately be reduced to claims about the contents of our immediate subjective experience, the more recent interpretation has it that science is ultimately about the structure of experience the structure underlined and bolded the structure of experience where a structure is supposed to denote something that is inter subjective rather than subjective that is to say it is logical rather than
experiential. Because experience in a Kantian sense or no Kantian sense always applies to individual agents. Whereas when we talk about intersubjectivity, we essentially arrive at a new realm. And this realm, sure, has experiential content, but also has something more. The very reason that we do agree on us seeing the same thing in an external world is because we share the same logical connections and behaviors so let us consider the two theses from above in more
details in the first thesis that i mentioned earlier on the two thesis from above in more in the first thesis given in the first thesis i mentioned given which i'm all means also since datum um denotes what is given by experience in particular are by sense experience indeed for the first for the for the rest of this argument you can consider sense experience in lieu of the word data so when for now we are talking about
data unless otherwise specified to the contrary you should understand that data means sense experience in the same vein scientific sentence in the first old thesis of our bow refers to any sentence in a language of any scientific discipline that uses its terms in clear and non-ambiguous way essentially which means that it ought to be formal because formal language make explicit the relations which are implicit within already a language hence science cannot be even imagined
without a formal language or a formal medium in the same vein the translation is to be regarded as mapping from the set of scientific sentences to itself the two theses the old thesis claim that there are translation mappings of a particular and distinguished kind one dark induced by a system of definitions in the way that the scientific sentence a
is translated to another scientific sentence tr parenthesis a parenthesis close Translation of A. If and only if the stepwise replacement of the defined terms in A by their defining and ultimately primitive terms yields TR of A. Translation of A. Two, the corresponding primitive vocabulary conforms to the syntactic restrictions that are explained in the theses only logical terms in the second one three finally the transition from
a defined expression to its defining one is to preserve extension necessarily and i do hope that you know what we mean by extension here does anyone know his reference to free gives idea of extension can give me some for others. Like definition by like mentioning like the different things which belong to such and such predicate. Yes. Like if men all, like if you were to give a definition of what a bachelor is, you would give a list of all the bachelors in the world.
a equals a is a definition by extension you say Socrates equals Socrates Socrates is a historical figure and Socrates as a philosopher This is by extension. Obviously, extension is always implicitly implies a reference. We are implying the same Socrates in our formula.
I don't want to go to details, but you can think about reference in terms of invariance of senses of who Socrates was. Socrates was the gadfly of Athens. Socrates was a pimp, according to Xenophones. Socrates was the first martyr of philosophy, according to Plato. Socrates was a man who used to get married to many wives because he thought that they shouldn't be left alone to their own device. So all these senses, the invariance of such senses,
what you might call to be a reference, essentially what allows you to make such an equation Socrates equals to Socrates So, as I mentioned, so these are the three main points that you should understand with regard to those two definitions of the old theses that I mentioned earlier on.
Now, as Carnap explains in Section 50 of the Afbaho, the translations of sentences and terms are claimed to preserve what Carnap then called logical value. You see, for Carnap logical value, something like meaning exactly, but syntactically defined in the sense that as long as some transformation rules are always preserved or invariant between your symbols, then you say this transformation rule or rule as such, has a logical value if we were in the semantic domain or pragmatic domain a
la brandon we could say the logical value is nothing but semantic value in disguise so logical value for Carnap is in the sense that all kind of really cares about is about the kind of relations that the sand within such and such symbols and the kind of transformation rules have been obtained between them you can say that logical value for Karnap is essentially logical extension in a free sense in the preface of the second edition of the
Well, Carnap-Clap refines his view by pointing out that what he actually demands is a necessary preservation of extension, namely invariance, i.e. the translation image of an expression ought to have the same extension as a translated one by logical rules or by laws of nature. Now you see here this is already implies that we are in the domain of logical empiricists a la early Carnap in the sense that Carnap like early Russell believes that logical extensions
be in fact thought about as natural extensions to the extent that logical laws of thought are isomorphic to natural laws of the universe is this clear we are we are talking simply about that a statement view that experiential atomic facts where we can finally say in a lot early corner that the logical extension how a equals a star
can be mapped onto red of this object equals the red of that empirical object or experience. We are simply talking about Carnap's view. I'm not defending this. So don't come up after me with some sort of super Kantian. I'm just thinking that you probably need some sort of like intentional component here, you know, because if you're trying to, you know, preserve extension through some sort of invariant. You see, according to Carnap, Carnap actually never took the idea of intention versus
Sellars and Brandon very seriously and I think that was a that both gave us some actually new vistas about what intentionality may mean but also ignore some of the more subtle problems of intentionality or aboutness in the sense that first of all Karnap is what you might call to be the hero of Valentin. Valentin's idea of neural net is essentially pronounced in detail in the last chapter of Carnap's foundational problems of a theory of probability. He does
not think that intentionality should be understood in terms of conceptual terms, but you can in fact talk about intentionality in logical empiricist terms in the sense that you can couch them in terms of probabilities or statistical analysis formal learning so Carnap yes Carnap is quite very you might say subtle but also problematic with regard to the question of intentionality.
Yeah, yeah. I mean, it just seems to me that if you're trying to preserve invariance over translation and invariance with extension... Well, the whole thing that Kant would say... You don't want to... Kant would tell Karna, you can go to hell for all I care. precisely because intentionality you can't even you you can't even understand you can't even you see the whole idea is not about the perceptual invariance and Carnap really knows this he tries to understand perceptual invariances in terms of logical invariances chance would tell you that well we should be
able to tell me first of all that how is that the logical invariance correspond to a succession of your experiential data if you do not have something like a perceptive subject a singular object that is purely intentional would really get angry at this stop see I wouldn't necessarily insist on like that per se but like I would insist you know and I do say like okay sure like you know logical invariance you know but then my question is is it kind of is similar to that after class with Theo that you were talking about with like about psychologism but it's like you know how do you ensure that the
Let me finish. The range of essentially defined things actually abide by such a logical invariance. Well, let me tell you this. What do you need from... You see, the idea of a logical reference is simply invariance of logical senses. Socrates 1, Socrates 2, Socrates 3, Socrates 4. refer to the same Socrates by virtue of us understanding in a Kantian sense their relation between the senses of Socrates I, Socrates II, Socrates III, Socrates IV, then we can say that this is a logical reference.
So wouldn't you need an intentional definition to say that like all of the Socrates I, Socrates II, Well, yes, but basically intentionality is about conceptuality, it's about aboutness. Aboutness is the ultimate, what you might call to be the conceptual form of judgment. It's an inference in the sense that if this is not an empirical association, no neural net can ever by itself if it is purely neural net a statistical analysis can never actually derive the logical reference of socrates socrates one socrates two socrates three socrates four
quote senses of socrates in order to arrive at what we call a reference called socrates with capital s that requires inference and that inference requires exactly what cance calls and a perceptive subject or a perceptive intelligence you remember this is really the attack of cans against hume hume believes that simply if we have the association of empirical datum we can actually say that the person who ate the hot ate the dog who ate ketchup who ate the bun who ate the mustard is exactly the same person that
we should refer to as the person who ate the hot dog can shows that is absolutely impossible this is actually a fallacy So it seems like the difference between Kant and Carnac here is Kant puts an emphasis on the aperceptive subject as a precondition, whereas what Carnac does is the intentional aspect is a logical consequence. Absolutely, absolutely, superbly. Superbly explicated, yes. However, you should understand that Carnap obviously is a student of Kazirer and Kazirer
is a known Kantian obviously, he's a student of Herman Cohen. So he is by definition a Kantian. It's just that he does not fully appreciate the idea of transcendental logic or transcendental philosophy, the way that Kant's understood. And believe me, the thing is that in the logical syntax of language, he becomes even more Kantian than Kant. And I will talk about it. But I think, unfortunately, it's 2.54 and I have my gardening duties. I have to do my duties. So let us stop this session, the next session for 30 minutes and I will assassinate every
single one of you if you make a question at the beginning. Let us just really go through the material and then you ask the question. So first the thing that I want to do is to finish this Carnapian vision of the statement view as put in the of-bottle and then make Quinean and Gutmanian objections to such Carnapian view of correspondence between logical statements and experiential statements or experiential facts. And then we can go on with the induction versus deduction. Yes, Jeff. I was saying that unfortunately I do not know any good work on the challenge
or the controversy between inductivism and deductivism. However, you can read the first few the first chapter actually the first introduction of i think it's called philosophical problems of a theory of probability if i'm not mistaken i will i will put the exact title on the on the Google Hangar, sorry, Google Classroom. So, I have a silly question, but maybe like what we were just saying,
like the difference between Kant and Karnapp is that Kant looks for conditions for the possibility of experience, whereas Karnapp looks for conditions for the possibility of drawing conclusions from experience. Yes, but the thing is that Kant would say that the same conditions, the same a priori conditions, and this is why it's called transcendental term, the same a priori, qua transcendental conditions, which constitute experience are exactly the very same conditions
from which you can derive conclusions from what you have just experienced. Which isn't really a problem for Karnap insofar as he tries to axiomatize the form. Yes, yes, yes, yes, absolutely. It's Carnap, yes. Okay, let's hear your last heckles, questions, swearing, so on and so forth. I'd like to, I'd like actually Chris to repeat what he said when he made this very good conclusion about difference between Kant and Karnap. So Kant assumed the uperceptive subject and Karnap what? Could we just repeat this?
So, I mean, I guess from sort of my understanding, and I think I think roughly along the same lines as Reza here, but is that Kant looks at like the transcendental upper subject as like sort of like an absolute precondition. Whereas with Carnap, and then that's how the problem of intentionality is solved insofar as the apperceptive subject is the conditions for experience. So therefore, it is a condition for satisfying what it is to be experienced in an intentional definition. Whereas with Carnap, it's like, oh, all sense data is extensional.
But we are able to draw conclusions from that sense data insofar as the logical resources we have to explicate such sense data have intentional components. It can thus satisfy, oh, you know, we have Socrates 1, Socrates 2, Socrates 3 as extensional associations within sense data. But then we have logical resources which has an intentional condition for satisfying what it means to be Socrates. And then we can all tie that up. Christian, let me interrupt you. Let me make the ultimate intuitive example.
that I made in the Kant class. You see, Svidlana, you see the import of synthetic or analytic perception, sorry, synthetic or analytic forms of a perception, unity of a perception, which is basically means that an intentional subject in today's world, is that you see the the earliest challenge with regard to such a stuff came with between cans and hume so you see that you let me tell you this that
you would say that there is someone who ate the dog someone who ate the bun someone who ate the ketchup, someone who ate the mustard. Now, Hume in this kind of scenario would say that simply by empirical associations, I would be able to say that the person who ate such ingredients is exactly the same person or the subject a perceptive subject who ate the hatha. Now this becomes a very big challenge in the critique of pure reason and the ultimate Kant's take on
humans empiricism that he shows can't show that simply by empirical associations you cannot direct derive the statement that the person one who ate the hot dog the reference is the same as the person as the persons person one who ate the dog the person two who ate the bun the person three who ate the ketchup and the person four who ate the mustard cant believes that we need something more than empirical associations this is what you might call to be the unity of our perception
Essentially, what is really unity of perception? It's a concept using subject. A concept using subject that by virtue of conceptualization is capable of not only unifying its experiences or encounter experiential encounters with items in the world, but also is capable of uniting them in one and one only experience. So I would say I ate the hot dog and not any other person precisely because I have received
such experiences conceptually in the sense of now I can derive finally an invariant reference between all such episodes of eating the dog, eating the bun, eating the mustard, eating the ketchup and refer it to as me, my own experience. This is what Kant calls unity of our perception. But Carnap wouldn't need such an expanded intentional condition as an apperceptive subject. In Afau, Carnap actually is, to be honest with you, with all due respect to Carnap and
you know that I'm now Carnapian, I don't think that he can actually manage to do that project. Oh, okay. Okay. It's essentially precanted. Yeah. I mean, my thought here was that then, you know, like the difference is between, like, you need the apperceptive subject versus with Carnap. Maybe not really Carnap. He might not be able to answer the question. But I don't know Carnap so well. but that you could have logical, intentional conditions for what it means to be someone who ate a whole hot dog. and then that can be something that can obtain a conclusion that such and such person ate a hot dog
from such extensional data of, you know, maybe these three people eating bun, hot dog, whatever, but then this person eating everything together. Yes, the whole point is this conceptualization, really. The whole point is an intentionality aboutness. Brandon really clarifies this quite fantastically. It says that intentionality is not as, it's not about consciousness. Let me tell you that Scott Baker, those of you who are familiar with his work, and I love, I do love Scott. I think Scott is a fantastic cynic, and we actually do need more cynics. because a philosophy without cynics is just idealization so I even though I
have huge amounts of respect for Scott what Scott thinks that when we are talking about intentionality of the external furniture inside the world it's It's simply about the problem of consciousness. Consciousness of course can be naturalized, can be reduced to basic neural processes. But it is not about consciousness. It is what Kant and Hegel would call you is about self-consciousness. The idea that we think of ourselves as a person by virtue of the concepts that we have.
virtue of such concepts we are capable of seeing a uniformity a logical uniformity between de facto items that we have observed in the universe so then i would be able to tell you that i am in fact the very person who ate the hot dog and i am in fact as a matter of fact the person who ate the dog the bun the mustard the ketchup so on and so forth it's actually a really a really i would say a thorny problem and majority of neuroscientists majority of the so-called you know ai people do not really understand the scope of this problem
the scope of this problem is huge is massive the problem of intentionality or what can calls unity of our perception without it literally you cannot move to the next level maybe it's my stupid logic brain thinking you know but i mean i feel like the connection to the unity of the apperceptive subject is just it's it's really not so necessary i mean to me it's like we're talking extensionality versus intentionality it's are we listing things or does this thing satisfy the logical definition and is able to obtain this predicate let me
interrupt you christine let me interrupt you the reason that the unity of a perception is necessary at least in kent why do you think that he actually comes up with this term unity of a perception he calls every perception every every empirical perception is ultimately a perception namely a a conceptually reformatted perception. Why do you think that Kant says something like that? Precisely because how can we put together in the most optimal way,
not only computationally optimal but also from an intersubjective point of view different senses of the same thing of a hot dog for that matter you see the problem of unity of our perception for Kant is essentially the element that undergirds the sense of what Kant calls objectivity. If you do not have unity of a perception, objectivity is in the trash bin. The very reason that we do in fact have objectivity
is precisely because we are treating ourselves as persons responsible for what we say and what we do in the sense that if I have it might of course illusory I might say that I saw something like that you know jumping in the backyard just precisely because I've already have a kind of unity of different senses logically bound together and apply to my empirical rudimentary associations so another person wouldn't
tell you is that of course what you saw was just an illusion another person will confirm it you see unity of a perception is what you might call to be the most basic element of what it means for an agent to be an agent and for an agent to be an agent it can only be what living or inhabiting a multi-agent system where you not only get your own experiential view of the world no matter how conceptually conceived
but also by virtue of your subjective unity you are capable of conversing and relaying to another person who might actually refute you and object to you namely another subject in this environment And this is really what unity of our perception is. Unity of our perception is really what you might call to be a computational element of what it means for us to have a multi-agent system where agent is ultimately defined by its interactions broadly understood with other agents within the same system.
I think that makes a lot of sense. I feel like it's a bit extrinsic from the intentional and extensional problem, namely because I do think, yes, that these issues... You see, the problem of extensionality is essentially the problem of judgment. Kant shows that you cannot have a judgment without the unity of a perception, namely subjectivity. So, you can't even have a reference or different senses of the same thing if you are not a subject in a Kantian sense.
So you're just like, just systematically and brutally saying that like the moment that logic, like before the moment that logic takes a single step, you're already in the domain of the transcendental aperceptive subject. Absolutely. Absolutely. Absolutely. That makes me the worst kind of transcendental cultist. I'm always asking that we wrap up so that we can go to the other class thank you everyone Theo I'll stay home bye my friends