Podcast: Play in new window | Download (Duration: 44:42 — 41.0MB)
On W.V.O. Quine’s “Epistemology Naturalized” (1969), featuring Mark, Wes, and Seth.
What justifies scientific theory? The classical epistemological project found in figures like Descartes and Locke seeks to find basic, indubitable premises that serve to ground the rest of our theorizing. Quine begins by considering Hume’s attempt to do this by claiming that all we ever experience are sense impressions, so anything else we know must be based on that. Rudolf Carnap took Hume very seriously and tried to actually reduce the meanings of complex propositions to their base sensory components, and Quine’s goal in this essay is to show why this does not in any way work. In fact, Quine wants to reverse the order by which philosophers base science on epistemology to instead propose a less problematic way of pursuing epistemology: through psychology, i.e. science. Making epistemology part of natural science is what “naturalizing” it amounts to.
Quine relates the question of theory vs. observation that we’ve considered in looking at philosophers of science such as Popper, Kuhn, and Lakatos to the pre-scientific context. According to a verificationist like Carnap, the meaning of at least certain basic sentences, which philosophers call observation sentences, is cashed out in terms of what sensory situations they relate to. So we tend to think that “it is raining” is certainly related to a person being in a situation where they’re perceiving rain happening right now in their vicinity. If the sentence is being used literally and observationally, i.e. in its “normal” usage (as opposed to its use in a play, or as metaphor, etc.), then the sentence literally means that perception… or we should better say “perceptual situation,” because we don’t want to have to take into account the minor differences that different perceivers might have. The point is that anyone with the relevant senses in that situation, on hearing that sentence and understanding it to refer to that situation would both then know the meaning of “it is raining” and (importantly distinct for Quine) be able to confirm the truth of that sentence.
All other sentences/bits of knowledge according to an empiricist ultimately rely on these basic sentences. But as we’ve discussed in other episodes (e.g. Sellars, Hegel), the idea of a “theory-free” perception like this that everything else could be based on simply doesn’t work. There are several basic assumptions about physical objects, time, and space that are already assumed by talk of rain happening now. Kant’s solution to this was to say that those basic categories are part of the human makeup and are necessary to think at all. Quine doesn’t want to go so far as that; instead, he says that knowing a language is knowing how to use words in their appropriate context, and this involves a mass of theory. To say what an individual observation sentence means, we’d have to also characterize the rest of the theory. Two speakers of different languages could have observation sentences of the same event (the occurrence of rain, for instance), and these sentences would actually mean different things, because of their different theoretical contexts.
To explain this, he uses the image of two speakers who don’t understand each others’ languages trying to come up with a translation. If I see the person uttering some sentence “bango!” in the presence of rain, I might assume that word means rain (or rather, the one-word sentence means “it’s raining.”). I could then test it by pointing at other phenomena (e.g. snow, sunlight) and saying “bango?” and seeing if the person assents. So it seems like this is a route for translation, that “it is raining” and “bango” both mean (in the verificationist’s sense) the phenomena of it raining now. But not so fast, Quine says. What if my foreigner thinks of rain as a persisting entity, as a rain god showing his liquid essence, whereas I have a meteorological view of the matter. According to Quine, our three sentences don’t actually mean the same thing due to their different theoretical assumptions. Well, perhaps this difference could be tested for, if I were able to ask “is this the same rain as before?” For me, the answer would be no, but for the foreigner, yes (it’s the same liquid essence). Ah, but what if there’s a additional difference in linguistic convention, such that while I say “it’s raining,” the foreigner’s remark would better be translated as “he appears as rain”? So if I ask, “is that the same rain?” the foreigner understands this as “is this the same appearance as rain?” Both of us answer no, because for me, it’s a different cloudburst, but for the foreigner, it’s a different occasion on which the singular wet god appeared. We’re using the sentences in the same way, in the same perceptual circumstances, yet because of the interlocking nature of our various beliefs and linguistic meanings, we don’t actually mean the same thing, and we can’t even figure this out. Quine calls this indeterminacy of translation: an experience underdetermines its observational description. Quine even conjectures that two such descriptions might even contradict each other, yet each accurately describes the perceptual situation. (Though this claim is actually pretty wild, and doesn’t seem possible to sketch an example of; in my example, the two speakers are just talking past each other by employing different conceptual schemes, not actually in contradiction.)
To get clearer on this, you may want to review our past episodes on Carnap and Quine, which provide more information on Carnap’s “Aufbau” project that Quine is reacting to here and also gives Quine’s argument specifically against the idea of synonymy, i.e. two expressions’ having the same meaning, which is obviously an essential component for Carnap who’s trying to reduce complex expressions to simpler ones that are in some sense synonymous with them. We also refer to Russell’s reduction of mathematics to logic and set theory and Gödel ‘s demonstration of the incompleteness of any axiomatic system (like Russell’s attempt).
Quine’s account does not so much debate with positions like Lakatos’ so much as complement them. Quine agrees like these modern philosophers of science that scientific advancement is not a matter of deductive certainty, because there’s no theory-free way to interpret an observation made on an experiment. Like Lakatos and Kuhn, Quine would agree that in the face of an experimental result, there are choices to make about which parts of one’s web of theory to change. However, he described Kuhn’s position as an “over-reaction” for reasons that he doesn’t go into. My sense is that Quine is interested here in the technicalities of meaning, not in actual scientific practice. The fact that in theory two contradictory observation sentences could both be true of the same observation doesn’t mean that this actually happens in scientific practice, because the various competing scientific theories are not so profoundly different as the competing metaphysics included in my rain example. But one can also see how thinkers like Kuhn and Feyerabend might appropriate Quine’s argument to support radical differences in conceptual schemes, and this is exactly why Donald Davidson responded to this by arguing that any two schemes have to be actually translatable into each other for them to count as language at all.
Buy the book containing this essay, Ontological Relativity and Other Essays, or read the essay online. Two secondary sources we recommend for this are Gary Kemp’s Quine: A Guide to the Perplexed and Peter Hylton’s Quine. Both of these authors worked on the Stanford Encyclopedia entry on Quine.
Image by Genevieve Arnold. Audio editing by Tyler Hislop.
Continue listening on part two.
Transcript
0:07 Mark: You are listening to the Partially Examined Life, a podcast by some guys who are at one point set on doing philosophy for a living, but then thought better of it. Our question for episode 294 is something like how do we ground scientific theories? And we read Willard Van Orman Quine's essay epistemology naturalized from 1969. More information please visit partiallyexaminedlife.com this is Mark Linsenmayer in Madison, Wisconsin. In all my philosophical theories came built right into my language, y' all jive turkeys.
0:36 Seth: This is Seth Paskin with only a limited Alphabet of perceptual norms in Austin, Texas.
0:42 Wes: This is Wes Alwan under Determined by the evidence in Cambridge, Massachusetts. Hey, I want to just start by plugging. I'm doing seminars. So for each episode we do, we'll almost always do a follow up seminar. You can find out about those at partiallyexaminedlife.com seminars or go to wesalwin.com if you want to sign up to be alerted when I do them. I will be doing one on Quine.
1:06 Mark: So this was something that was on our list for a long time. We did some Quine a long time ago. The two dogmas of empiricism. And this is sort of the next thing that's been on the list is supposed to explain why Quine is sort of the analytic philosophy take on philosophy of science. Not exactly the same line of debate as Lakatos and Kuhn and those guys. Though he mentions Kuhn at one point, thinks it's a silly overreaction, but he definitely has that idea of you got a theory and there's an observation, and what is the observation change about the theory?
So that thing that we've talked about in several episodes is in here. It's just he's also rooting this straight to historical empiricism to Hume. And then Carnap is his immediate predecessor. So we did an episode a long time ago on Carnap's Aufbau on his project of trying to reduce all the complicated things that you'd want to say about science maybe to statements about immediate perception. Right? That's kind of what Hume wanted to do in the first place, is all we have knowledge of is our ideas, our impressions, and then when the impressions stick around a while and become stale, they're ideas.
So they're all basically things from perception. And Carnap tried to do that for real. And Quine, in this essay and other things he was writing at the time, is explaining why that was kind of a silly idea. And how he wants to naturalize epistemology, make it no longer the foundation for science, but actually part of science. What's your initial impression, Seth? You're on the spot.
2:33 Seth: This is a very strange text. My perspective. I was familiar with Quine from all the analytic tradition, and I think he fits in the canon there with, you know, A.J. ayer and of course, Carnap and Frege, kind of like these early 20th century, early to mid 20th century Analytic philosophers who were focused on language. And I expected to come to this text and be bored and disappointed. And I was surprised. First off, he's a very good writer. But the other thing is I found the project once we got to kind of what his turn was. Typically, when we're talking about skepticism about the external world in epistemology, my eyes just glaze over, right?
Because deep at heart, I'm just a. Even though I don't want to admit it, I think I'm a Heideggerian, where I think that question just doesn't even make sense. And after reading the Quine and then reading that secondary article that you recommended, Wes, which was also quite good, I think his project is actually quite interesting. It's not a question of skepticism so much as I was just. You're not going to believe this. I was listening to our old Quantum Mechanics episode in the shower today, and there's a point during that episode at which Wes says something to the effect of Heraclitus was Socrates's and Plato's skeptic in the same way that Hume was for Kant.
So Kant was responding to Hume's skepticism, and in a certain way, Socrates and Plato were responding to Heraclitean skepticism about the reality of the reality of the world or what we can know because of the changeability. And in a certain way. That's funny, because it's an odd framing for this discussion, because Quine is taking on skepticism and taking it seriously, but at the same time, he's making a move that to me is kind of oddly Heideggerian, where he's saying, look, I'm not going to deny or quibble about the existence of things in themselves or epistemic access to real things.
What I really want to know is, how does knowledge happen? How do we even build knowledge from what seems to be a relatively limited set of perceptual experiences? And what he means by that is something to the effect of how is it that we build concepts from this limited set of perceptual experiences, which in turn allow us to characterize them in A way that becomes scientific and then how can we trace that back so that we can have some validity about our scientific characterization of the world? And that's a surprisingly interesting question if you take the position, as the commentators said about Quine, that we don't have any kind of a priori concepts that we can bring to bear in advance.
So anyway, I found it kind of invigorating and interesting and it was chunk size, just about the right bite to take for me, like not too much along with the secondary literature, it was just about enough reading.
5:34 Wes: So we did Two Dogmas of Empiricism. Right.
5:37 Mark: And on what There Is is the name of the other essay. Yep.
5:41 Wes: Yeah. It's funny, I had written a kind of angry essay about on what There Is in grad school at ut, because all the analytic philosophy was just so bizarre and horrifying to me when I first encountered it. I just couldn't figure out what the hell was going on beyond the sense that it was just dry and soulless and all that. Just, it seemed confusing. And not just confusing because I didn't know enough about it, which I didn't, but confusing because it just seemed to be missing the point, which, you know, one would see if one just knew more about Kant.
Now, reading Quine, I think I don't remember what my experience was with our last recording on Quine. I think I was more positive, but I found very interesting and it prompted me to do a lot of extra reading, including the one I'll recommend listeners is Peter Hylton's Quine. That's part of the Arguments of the Philosophers series by Routledge. There's another one by Gary Kemp. Quine, a Guide for the Perplexed, also very good. And I think they wrote the Stanford Encyclopedia article on Quine together. But my take on this is that Quine still takes skepticism seriously.
So this is what I think we need some element of skepticism in order to take epistemology seriously as a project, even if we naturalize it. So to take skepticism seriously, the way Quine puts it at one point is just that we want to figure out the ways in which theory is under determined by evidence. And theory is a big word, but it. It means everything from more robust abstract scientific theories, but it also just means any kind of talk about physical objects in the world, physicalist discourse. So what Hume is doing epistemologically, when he tries to say, hey, look, we have all this physicalist discourse about there being these objects out there and that they're causally related to, well, what we get in terms of immediate sensory evidence, all these different sense impressions.
It really underdetermines everything that we think about those objects. The causality is unsupported and things per se are unsupported except as bundles of successive sense impressions. Those are theoretical constructs. So they're under determined by the data. Which is why rationalist wants to come along and say, actually the things we know best are intellectual intuitions of these very clear ideas. After all, this is what science is about. What do we know better than causality? What do we know better than extension? Says Malebranche. We have immediate intellectual access to those things.
So our epistemology is based not on sense data, but on those rationalist categories. That's what makes epistemology and skepticism of some form interesting. It's the sense that what we think we know is under determined by the inputs. Let's say where Quine is going to end up is he's just going to say, look what this turns into. What Hume's project turns into with Carnap and the logical positivist is this attempt to reduce physicalist discourse to sense data or to show how you can construct that discourse and therefore all of science and scientific theory out of sense data.
And for various reasons we have to give that up. That's not going to work. But once we give that up, we're kind of free to, because we're no longer trying to do something deductive and we can't be accused of circularity. We can actually use science and psychology in particular, in order to investigate the foundations of science. So we start looking at language acquisition in infants, for instance, because, right, what is theory all about in physicalist discourse? It's all embedded in language. And so we look at, instead of saying, hey, how do we construct physicalist discourse out of sense data?
We say, hey, let's observe infants being bombarded with having their sensory inputs irradiated, right, with light and other inputs. And let's see psycholog how the construction actually happens. This is like an empirical thing. We don't have to do rational reconstruction. We can actually look and say, do some cognitive psychology and say, hey, how does the mind construct all of this stuff from the inputs? How is language acquired as opposed to how is it built up?
9:49 Mark: The picture of science that includes philosophy that he comes up with is some kind. You were saying, cognitive psychology. Cognitive science was supposedly my specialization in grad school, though I drifted away from that pretty quickly. Ends up being something that I would be less interested in. I think there's a Reason that we haven't done texts that are trying to get. Let's use neurological data to see exactly how beliefs are built out of sense data. That's the kind of thing that we would want here. And I'm not sure, you know, if it's just.
Well, historically we hadn't gotten to there yet, and now we're approaching that. But that seems to fall short of, you know, Seth, you'd said you found doing science stuff a little less interesting because it seems to go out of date so quickly. And also, it's just as philosophy people, we're sort of interested in the bigger picture. As long as we have some sort of sketchbook of like, schematically how these things get put together. You know, is it gestalt, is it synthesized from little atomic bits? Those kind of larger categories of theory are more interesting to me than the specifics, you know, And I don't see any evidence that Quine, in any part of his career gets into the specifics.
Like, he is a philosopher. He's laying groundwork of here's how philosophy and science should fit together. Not that he actually wants to do cognitive science himself.
11:04 Wes: Yeah, we don't have to do cognitive science. He probably doesn't really want to do it either. He just wants to point to it and say, hey, people can do that. That's what's so great about being a philosopher. You just sort of assign things. Yeah, I do the big picture stuff.
11:19 Seth: I'm an idea guy.
11:20 Wes: I communicate the plans, the client's plans to the engineers, and then they do it. Yes, to quote office space. But I, you know, I wanted to say that in a way, I think Quine is actually getting back to the origins of epistemology with this idea, because it does all start with cognitive science. Modern science is sort of what puts skepticism on steroids, because people observe, you know, they're coming up with theories of light in the early modern period, and they're saying, hey, wait a minute, we see different colors because there are different wavelengths of waves of particles going into or frequencies of particles going into our eyes.
And Galileo says, look, there are primary qualities and secondary qualities, and secondary qualities reduce to the primary qualities, and you get all that stuff. That is where epistemology really takes off because of that early cognitive science stuff. And how do we start to explain our perceptions and much less our more abstract ideas, in terms of actual inputs, in terms of what actually happens between us as physical objects and then things that affect our senses?
12:24 Mark: So an interesting mechanism that he uses in this essay, which we haven't really touched on yet is the idea of translation. So in trying to figure out what the empirical content of a sentence is, right, If I make a sentence. And as we were saying, Carnap wants to reduce that to, I have perception, you know, sensation of red in front of me at time t. What? You know, some kind of basic sentences like that, and build theories out of those. Well, the way that Quine wants to say, well, what would you actually do if you met somebody that didn't speak your language at all and you were trying to figure out by just, like, pointing at stuff and what his words meant?
Like, let's go up to a tree and point it. And I say tree, and he says, goomba. And I'm like, okay, I guess goomba means tree. Like, and how you would actually build things up out of that. And that is a way of saying, there's an assumption there. I hadn't really thought about this based on. We were just talking about, in our symbolism episodes, this very kind of thing. But how that makes that human empirical assumption that whether or not the belief or the trees have bark or what, you know, whether or not a sentence can be somehow reduced in this carnapian way to observation, sentences.
Let's say still that is the only way that we can communicate with each other. Or the way we would start, at least the way we would get started. Like, we couldn't necessarily use this ostension to teach each other logic. That's a level of abstraction higher. But we sort of are assuming that there are at least some sentences where I can tell what you mean without knowing your language. Because you and I are presumably having the same perceptual experience that I think I'm looking at the tree. You're looking at the tree. Therefore whatever's coming out of your mouth probably is referring to that thing.
I didn't realize that we are all, he's saying empiricists in the way that we learn language in the first place.
14:15 Wes: So I think we are. Although, you know, according to Hylton, he's actually a little bit friendly to some of the a priori stuff as well, because there's a question of salience, right? Are you pointing to the tree? Are you pointing to the bark? Are you pointing just to oblong standing things? Are you pointing. So we have these things called similarity spaces. And some of that stuff is just defined by how the brain is structured based on evolution and all that stuff. But putting that aside for Quine, ultimately this comes down to basic sentences spoken in the context to Us and in front of us and ultimately by us, you know, as we acquire language when we're infants in the context of sensory stimulation, basically. So it's not necessarily ostension that does the work early on for language acquisition. Right. It's something not unrelated to that, but I think something a bit more complicated.
15:07 Seth: We know the general problems with the ostension. Right. This is the Augustine. Right. Who proposed that. This is the theory of language acquisition. Quine's not that naive. He's very clear in the essay that there's talking about the social component. It's not as fully explicit as some of the things we've looked at where we talk about the social ele. Trying to think of some examples like
15:29 Mark: the private language argument. He mentions that, or Captain Middleings mentions that.
15:34 Seth: Well, just talking about Langer and talking about this social nature of language acquisition and how if you're not socialized around language speakers, you can't pick up a language, in fact. And when you miss that window when you're young. And since he's going to be talking about a genetic or a developmental aspect of it. So I think Quine recognizes that it's not just about pointing. You know, ostension is not sufficient. He's not that nice, naive. And that's why he's trying to define the satisfaction conditions for language usage. It's not that you point at a tree and you say tree, and then the child or the foreign language speaker says, huh, okay, I wonder if he's pointing at the bark or whatever.
It's that through the process of using the word tree in a propositional statement. In other words, there's a context that requires more than just I've forgotten all my analytic language. It's the indexical, the nominative. I've forgotten all those terms they use, but it's like the name is not sufficient. Learning language is not about pointing and naming.
16:36 Mark: Demonstrative.
16:37 Seth: The demonstrative predication or.
16:40 Wes: Yeah.
16:40 Seth: So he's going to get to the point where he says something to the effect of you can say somebody knows how to use a word or knows the language when they appropriately respond to the stimulus or stimuli in a way that most normal people would or in the same way that most normal people would, something like that. And coming to understand and use language, he's much more use focused in the sense that you have to be able to in the appropriate context where it makes sense to utter the word tree. You would utter the word tree. And how you come about getting to that point is the question. But the One thing, you know, you don't get there is just by pointing and saying the name over and over again.
17:25 Mark: He goes through this exercise of trying to learn, you know, imagine how you would figure out a language from somebody that doesn't speak yours at all, and you don't speak theirs. And it sounds right that what you referred to, Augustine's position, is the starting point. But as Wes was saying, that there are complexities and that's actually his point. And apparently this book, Word and Object, which I thought was just another essay, I'd never actually looked at it. But no, it's a 300 page book is all. I guess a large thing that it does is build up this whole idea of translation and how it is problematic such that we actually then get something that speaks very directly to what we were talking about with Kuhn and Lakatos and stuff like that.
That, yes, you have these, what seem like observation sentences. And it seems like this is a tree and they can say in their language, and we think like we've started, but once you get into the center and you start trying to make like, you know, trees are material objects or something like that, then like, how in the hell would you. You couldn't use ostension to determine that. So he comes up with this idea of indeterminacy of translation. So it actually becomes that it's not that these individual sentences, in fact, even the individual sentence, this is a tree is part of a theory.
Right. The whole language is kind of like a theory. Like Lakatos called it a research program, or Kuhn called it a paradigm that it's a web of beliefs. And this was in our past Quine episode, I recall, you know, so whenever you're trying to figure out what somebody means, you can't just try to figure out the individual sentences. That's not the unit of meaning. The unit of meaning is in the theory as a whole. Because it could be just that. Maybe they don't recognize that that's a tree. Maybe, you know, you think that's a tree, but they think that that is, you know, what you would call a large piece of grass or something like that.
19:06 Wes: Yeah. A tree is already a theory. And any physical object is already a theory. Like this comes down to like Husserlian phenomenology. Right. What do I see of the chair in front of me? You know, I see one angle, one perspective. And my theory includes the idea that there's something in the back of it. If I walk around it, I'm going to see the back of it. That It's a three dimensional object that it's resistant to touch, all sorts of stuff that go well beyond the immediate sensory experience. And this is what Carnap tries to do, is he tries to say, hey, I'm going to show you how to translate all that physical object, theoretical physical object language, into these fundamental statements about sense data. Like.
19:49 Mark: Yeah, conjunctions of them.
19:50 Wes: Yeah. So very mathematical and abstract, ultimately, but, you know, it involves. And it's not carried out. He points to the whole project, but he doesn't carry it out. But the idea is that you talk about placing specific colors at points in space. Right. And that everything could ultimately be reduced to that sort of talk. So with Quine, like an observation sentence, I don't think this is a tree. Would be an observation sentence. It's too theoretical. I might be wrong about that. I think something like, hey, this is sweet, would be closer. Right.
And maybe we can only talk about how close and how far away things are to being observation sentences. Right. Ultimately, his definition is that it's just going to be whatever is in closest causal proximity to the stimulation of our sensory receptors. He's going to give a kind of more externalist account of what it means to be an observation sentence. You know, something that leans much more on concurrent sensory stimulation than on what he calls collateral information, which is all that theoretical stuff. Right. All that additional stuff that might make me want to say, hey, that's a tree, as opposed to, hey, that's a rough brown blob or rough, you know, round cylindrical thing or whatever.
21:02 Mark: And just like you could have a different Kunian paradigm or a different Lakatosian research program that, you know, very different theoretical apparatus that is supposed to explain the same observations. So for Quine, this idea of indeterminacy of translation means that the theory as a whole, the language as a whole, you could have multiple languages which mean very different things in the sort of colloquial sense of meaning. It's a very problematic word for him, but you can have different theories that account for the same observation. So I think ultimately, which is the proper kind of observation sentence, he kind of raises that, as this is something that the people in the Vienna circle in the 30s argued over.
And it's kind of a moot question. He says, like, if you have to insist on something, fine, let's talk about whatever is closest to the sensory organs. But ultimately there are no observation sentences because there are no theory free sentences. They're all part of a theory. And that means that you can't actually do this translation project that it is, or rather there are multiple ways you could do it. And it's unclear. Carnap sort of agreed with that. Like, well, maybe there are multiple ways you could cash out. Like we could explain things in terms of basic physical, atomic facts, like Wittgenstein, early Wittgenstein said, as he seems to try to do in the Aufbau, just in terms of basic phenomenal things, which I just listened the beginning of our Carnap.
At that point, it was like my total experience at Time T. It was not red spot in this. That was something we were talking about with Wittgenstein. But these are all different ways that you could try to axiomatize, that you could try to break down into the basic principles reality. And the fact that you can do that in multiple ways, I think is a way of prefiguring Quine's idea that there's an indeterminacy of translation. It's just that I think for Carnap, these all have to be ultimately consistent and translatable into one another. Whereas the point of the indeterminacy translation is you could have two sentences that actually contradict each other, both of which are correct translations of the original sentence.
And this, I think, is sort of the most controversial, you know, hit you in the face point of this, because if you think about it in. I approach an actual French speaker and I'm trying to figure out what they mean, and you're trying to do the same thing. And I decide that what they're trying to say is this thing is black. And you'd say that the person is saying, this thing is not black. And somehow, according to this thesis of indeterminacy of translation, it could be that we're both right. And that just seems completely counterintuitive. So that's at least something we need to explain before we're done with here.
23:28 Wes: Yeah, unfortunately, he doesn't get much into that. Right. I mean, he mentions it, but it's unclear to me how that works. I'm not sure I fully understand the indeterminacy of translation.
23:38 Mark: Seth, do you have any more here, or should we just get into the text here and start moving? Because this is what he ends up with. This is kind of the end of the essay.
23:46 Wes: Yeah, I think we should start from the beginning.
23:48 Mark: This is from a collection called Ontological Relativity and Other Essays released in 1969. So it's based on a lecture, I guess, from earlier, but it is not. Even though in this book he makes reference to the other chapters it was meant to be a standalone essay, a standalone lecture.
24:03 Seth: The problem statement that he starts with is epistemology is concerned with the foundations of science. Conceived thus broadly, epistemology includes the study of the foundations of mathematics. As one of its departments, specialists at the turn of the century thought that their efforts in this particular department were achieving notable success. Mathematics seemed reducible altogether to logic. In a more recent perspective, this reduction is seen to be better describable as a reduction to logic and set theory. This correction is a disappointment epistemologically since the firmness and obviousness that we associate with logic cannot be claimed for set theory. So I'm not going to lie. I think I have a vague notion of what that means.
24:45 Mark: Folks could listen to our Russell on math episode.
24:47 Wes: Yeah, I mean, beyond the idea that, you know, seemed like math could be reduced to logic, that you could take one plus one equals two for instance, and prove that that's true from basic logical axioms, which I think they. Russell and Whitehead did. Right. And there could be a mathematic, or at least they thought they did.
25:05 Mark: But you had to use set theory, I think, even for them.
25:07 Wes: Did they use set theory?
25:08 Mark: Okay, yeah. It's just that that was before we had our Godel episode. So even that was problematic. That with Godel's incompleteness theorem, which I don't really want to sum up, but the idea is like, you can't. For any consistent theory, you can't prove all of its true sentences based on a finite set of axioms. There are always going to be some sentences that will be true according to the theory, but are not provable within it. So that by itself just busted up and we have an episode on that. I will refer folks too.
25:37 Wes: And as far as why set theory is not obvious, I have a vague sense of that. It just seems kind of crazy, right? You're treating sets as if they're entities. And, you know, you could be very Platonic about it and it aid for. I don't know if you guys and at ut, but it aid for the logic sets and theory class with Causey, who kind of wrote the textbook on it. So I taught this stuff and I just still found the whole. All the set theory stuff quite bafflingly abstract, you know?
26:04 Seth: Yeah. I mean, I think at root. And Quine's going to tease when we talk about set theory, talking about consistency and completeness. And then Quine makes a distinction that he calls conceptual and doctrinal. So conceptual studies are concerned with meaning, doctrinal with truth. And it comes down to derivations.
26:25 Mark: That was a pretty new distinction to me, to actually have those separate words for that. But of course, if you're doing, there's. What we were describing Carnap as doing seems to be about meaning. But I guess it's supposed to be both. Right. We prove a scientific theory by breaking it down into. Well, what it really means is this set of observations, and those set of observations are what proves the theory. So it has two relations. It's a conceptual relation, meaning, and a relation of derivation.
26:52 Seth: You're talking about the truth part of it.
26:54 Mark: Yeah, the truth.
26:55 Wes: You know, by the time we get to physical objects, what Carnap was doing was the conceptual part, like conceptual analysis, defining bodies in terms of sense data.
27:04 Seth: Yeah.
27:05 Wes: And then the doctrinal part when it comes to physical objects is just how do we justify our knowledge of the truths of nature in sensory terms? So justifying the truth of our statements via sense data is the doctrinal part. The conceptual part is analyzing our concepts into sense data. Right now he's at the mathematics version of this.
27:28 Seth: The important thing is this structure of derivation that Mark mentioned. The way I conceived it when I read it is something like, okay, imagine you have. The ideal state is you have this very complex mathematical formula that represents some wackadoodle thing. And what you want to be able to do is prove that that thing can be constructed out of much simpler, less abstruse, less objectionable, if you will, mathematical formulas that are simple and clear. This is all the derivation aspect of it. So ideally, what you want is some subset of your notations from which you can derive all the rest of them, the mathematical language.
And then ideally, what he says is that small subset of foundational terms can then be translated or represented in logic. And logic is where meaning comes in, because logical structures, the meaning and the intuitiveness, the obviousness of it should be there. And then the idea would be, okay, if you build this set of terms scientifically that are all reducible to the subset, and the subset is then can be mapped into logic, where you can get meaning and some form of validation corresponding to our sense data, then you would have a foundation for science.
But the problem is that that breaks
28:51 Wes: down like in logic, for instance, right? You can prove, like modus ponens, right? You can prove, if it's the case that A then B and A then B, you can do a logical proof of that in terms of more fundamental concepts. So there's a theorem, which seems obvious to us, but there's even more obvious stuff that we can reduce it down to.
29:11 Mark: Well, I mean, the proof is a truth table.
29:13 Wes: At a certain point you're just defining essentially.
29:15 Mark: Yeah, I'm defining the condition if then.
29:18 Wes: Right.
29:18 Mark: That little arrow or whatever sign you use for conditional is defined by. If they're both true, then the conclusion is true. If the first premise is true and the second one is false, then the
29:29 Wes: conclusion, well, you need, actually, you need to do some work. You have a lot of other assumptions to prove, modus implication, distributivity, law of excluded middle, a bunch of other stuff.
29:39 Seth: So all of which from Quine's perspective, we can claim a certain amount of certainty about.
29:45 Mark: Yeah, they're clear and distinct in Descartes.
29:48 Seth: Right. And the problem is that the axioms of set theory are not as obvious and as certain as these axioms of logic. And that's where the problem comes in.
29:58 Mark: But mathematics is still better than what we've been able to do for any other area of knowledge as far as this kind of derivation is concerned.
30:06 Wes: The other thing to make note of here is just that the conceptual and the doctrinal kind of interlink with each other. So the conceptual clarification serves the purpose of derivation, because the clearer our terms, the easier it is to establish inferential connections between them. And ultimately, if we were really clarifying everything and we get down to the base level, self evident truths, then we have that possible scenario where we could do something like reduce math to the obvious, obvious, the self evident, clear and distinct ideas of logic, to use a Cartesian way of putting it. But to the extent that we don't do that conceptual stuff and we haven't clarified our concepts, then our derivations, our doctrinal stuff, not as easy or not as clear, not as forceful.
30:54 Seth: So having introduced this on page 71, he says, I refer to the bifurcation into a theory of concepts or meaning and a theory of doctrine or truth. For this applies to the epistemology of natural knowledge no less than to the foundations of mathematics. The parallel is as just as mathematics is to be reduced to logic or logic and set theory, so natural language is to be based somehow on sense experience. This means explaining the notion of body in sensory terms. Here is the conceptual side, and it means justifying our knowledge of truths of nature in sensory terms, here's the doctrinal side of the bifurcation.
There's an analogous structure here. If you think of the body of natural knowledge, the subset of terms, or the subset of sentences, or whatever it is from which all the natural knowledge can be derived is the observation statements, but the certainty of those observation statements is to be derived from our sense experience or from sensory experience. So the parallel is that there's the certainty that comes from the axioms of logic because of their self intuitiveness or whatever self evidence. Parallel in natural knowledge is our sense experience.
And then presumably there's some set of statements from which we could theoretically derive all of the rest of the natural knowledge that would be built upon that.
32:25 Wes: If we are empiricists, we think that sense data, sensations, or what does Locke call them, just ideas, that those are where our knowledge comes from and what it's based on. And in a way those are what's clearest and most distinct. Whereas the rationalist, right, the sense stuff is what's least clear and it's our a priori ideas that are clearest and most distinct. But you know, this is kind of the two big epistemological alternatives, right? And Quine just assumes we're all empiricists and we understand that as people who embrace natural science that if we can reduce everything to, you know, if it were just sensory terms and logic and hey, even set theory, fine, we live with that.
Live with set theory. If we really could fulfill that project, that would be amazing to construct all of our scientific discourse and talk about physical bodies and all the rest of it out of logic plus sense data. But again, it's impossible.
33:21 Mark: So yes, on this page he is talking about Hume, who made this distinction between the conceptual and the doctrinal. More important, because all this hopefulness that we would have is really on the conceptual side. It's like we could under, we could better understand what we're saying by any given theory by translating it into terms of sense data and. Right, this is what the verificationist did. If your philosophical theory does not cash out in experiential terms, if it's a theological theory, right, there's no thing that could happen in the world that could convince you that your religious views are wrong.
Then that means that your religious views, according to these guys, these Vienna Circle logical positivists, have no cognitive content at all. They are not philosophy. They don't even make sense. So throw all that stuff away. This is the tradition that Quine is coming out of. But that's just meaning he looks at Hume. Induction is something we absolutely need to do science. We need to be able to have individual observations and make generalizations based on them and think about especially causal generalizations. And so this is the whole reason that Kant had to come up with the idea that these things are a priori, the idea that everything has a cause is an a priori principle that we just come to the table with because Hume says we only know individual impressions and how these build up over time, what these reveal to us, and they do not reveal to us that everything has a cause.
It's just. That's just something we expect. It is a sort of an irrational thing, necessary. You know, why the stupid joke about my cousin thinks he's a chicken? You know? Well, why did you get him fixed? Well, because we need the eggs or whatever. Like, so this is what induction is for the empiricist. It's like there's no rational justification for it, but we need it.
35:04 Wes: So, yeah, this is what I was getting at in the beginning with the fundamental motivation for skepticism and the idea that the evidence under determines our theories, because our theories include induction and they include causality and they include even write spatiotemporality, which according to someone like Malebranche, is also not something we get from the sense data, right? In fact, we get nothing formal. What the early epistemologists were onto was the worry that data is just data. The fact that we get it through our sensory receptors doesn't account for the fact that it's structured in some way, but it's formally interrelated in our system of knowledge, in our experience of the world.
And that it's got to be isomorphic with. It's got to be a picture in a way of the formal structure of the external world. And then the question is, well, how does that structure get inside us? We don't have sensory receptors for forms, for structures. We only have sensory receptors for light and for particular sense data. So that's the fundamental skeptical insight. And the problem of induction is just one part of it. So general statements, and even Quine says singular statements about the future, right? You don't get any increment of certainty by the fact that they're construed as being about impressions.
So as Mark said, even though it helps us conceptually, it doesn't help us on the doctrinal side. It doesn't help us gain any certainty in our inferences. In other words, science is never just going to be deduction, it's never just going to be math. It's induction, it's generalization, and therefore it's never, can never aspire to mathematical certainty. Then he gets next into this. Conceptual advancements, right? Contextual definitions.
36:44 Mark: Yeah. Jeez, do we need to go through Russell indefinite descriptions again?
36:48 Wes: I can skip us Past this with a very quick summary if you want.
36:52 Mark: Go ahead.
36:53 Wes: He's just described how there's hope on the conceptual side. We're kind of screwed on the doctrinal side. And then we get historically, these conceptual advancements. So one of them are contextual definitions from Bentham, I think, where we. That allows us to make our translations more advanced. So we translate thing talk into sense data talk without doing the Humean thing of identifying things with sense impressions or even anything else. Basically, the contextual definition is the idea. So, for instance, to give the meaning of a term is not to give a synonym.
It's not to say what its reference is. It's to show how we can translate all the whole sentences in which the term is to be used. So translate talk of bodies into whole sentences about impressions without equating the bodies themselves to anything at all. So sentences become the primary vehicles of meaning, which is something very important to Quine. The other innovation beyond the contextual definition is the fact that we've kind of brushed over this. But the idea that you could use set theory to kind of enhance logic in support of arithmetic.
You can use that for your physicalist stuff too. You can use set theory as auxiliary concepts in trying to given ontology of sense impressions. So you can turn bodies and their properties into sets upon sets. Quine says of sense impressions, a quote unquote drastic ontological move.
38:15 Mark: Right. I think we need to say a little more if we're going to use it this much. Just about what set theory is. I'm not well prepared to do that. But I mean, I think the example that you just gave is probably good enough that you're talking about. When you're grouping things together, you're going beyond the observation. Even in Carnap's thing, there had to be some, I think, fundamental, undefined notion of similarity. Right. Because there's this impression and.
38:42 Wes: Yeah, that's his basic relation.
38:44 Mark: Yeah, this impression and then this later impression. They have to be similar.
38:48 Seth: That.
38:48 Mark: That's even in Hume, I believe.
38:50 Wes: Yeah, for Carnap, right. It's gestalt. It's like your whole. The entire sensory experience that's basic. And then the other basic thing is a relation that there is some similarity within that gestalt hole. You know, it could just be a pixel, but there's some similarity between one of them and the other one.
39:06 Mark: So if there's similarity, that means you can put them together in a classification, in a set, and you can say, this particular horse is not just the impression I have right now. It would be bizarre to say this impression of horse I'm having right now, that is the same as horse. No, clearly, a horse is an animal. It exists over time. That impression that you just had for a second there's did not exist over time. It only existed in that moment. So, all right, well, let's make a set out of these things. And then where a theory comes in, I think, is when you get to that bullshit about Russell's paradox, you know, you could use a description to create a set. And this creates difficulties because you could have, like, the set of people who
39:47 Wes: cut everybody's hair do not belong to sets.
39:51 Mark: Yes, that everybody cuts their own hair. So the question is, does the barber save himself? So whenever you start talking about sets, you can run into paradoxes like that. And that's why you need set theory, because you end up talking about higher levels of sets or something like that. That this is how you solve these paradoxes is to say, well, there's a second level language that I can talk about the first level of. So this is what I think makes the sets into a set theory, is that you just end up having more complicated shit than if you're just talking about the set of horses or this bunch of perceptions being grouped into horse. Right? That's just sets. There's no theory involved there.
40:28 Wes: The theory comes in with what you're allowed to do with sets and all the operations, right? So the union of two sets. And there's a whole theoretical apparatus associated with sets in the same way as with math. And I think that plays some role in. With carnap, it's very complicated. And we read it and discussed it, and I still don't understand it. So. So there were two conceptual innovations that Quine was telling us about. One was contextual definition, and the other, we can even do away with contextual definitions if we can add set theory concepts and then talk of the external world as a logical construction of sense data.
So you get sense data, logic, set theory. That should allow us to construct our science and just the way we talk about physical bodies and all the rest of it. And that would be great, he says, if we could do that. But again, so even with set theory, even though that's kind of a flaw, be great if we could do that. But we can't because of the problem of induction. Our generalizations always cover more than we can observe. So we just come up against the same sort of problem that he's described before.
41:35 Mark: For the thrilling conclusion of this essay, you need to come back for Part two, though actually there's also going to be a part Part three supporters. Part two should be the next thing. Your feed and part three, which will be recorded on a separate day, will come out next week for supporters through Apple. All of that is delayed by a week. So this week part one, next week part two, the following week right before the new episode, Part three. Thanks.
Leave a Reply