This is the second episode of a book club series on Peter Gärdenfors's book Conceptual Spaces. In this episode, we will discuss chapters 3 and 4, in which Gärdenfors explains how properties and concepts fit into his theory of conceptual spaces.
For this series, I'm joined by Koen Frolichcs, who was already my cohost for the books club series on Lee Child's Killing Floor. Koen and I are PhD students in the same lab.
Podcast links
Website: https://geni.us/bjks-pod
Twitter: https://geni.us/bjks-pod-twt
Koen's links
Google Scholar: https://geni.us/frolichs-scholar
Twitter: https://geni.us/frolichs-twt
Ben's links
Website: https://geni.us/bjks-web
Google Scholar: https://geni.us/bjks-scholar
Twitter: https://geni.us/bjks-twt
References
Gärdenfors, P. (2004). Conceptual spaces: The geometry of thought. MIT press.
Kahneman, D, & Tversky, A (1979). Prospect Theory: An Analysis of Decision under Risk. Econometrica. 47 (2): 263–291.
Kriegeskorte, N., Mur, M., & Bandettini, P. A. (2008). Representational similarity analysis-connecting the branches of systems neuroscience. Frontiers in systems neuroscience, 2, 4.
Montague, P. R., Dayan, P., Person, C., & Sejnowski, T. J. (1995). Bee foraging in uncertain environments using predictive hebbian learning. Nature, 377(6551), 725-728.
Murphy, R. O., & Ackermann, K. A. (2014). Social value orientation: Theoretical and measurement issues in the study of social preferences. Personality and Social Psychology Review, 18(1), 13-41.
https://en.wikipedia.org/wiki/Monotonic_function
-
[This is an automated transcript with many errors]
Benjamin James Kuper-Smith: [00:00:00] Yeah. So now finally, after half an hour of us trying to figure out how to record sound,
Koen Frolichs: we did it.
Benjamin James Kuper-Smith: We finally managed to now get there.
Koen Frolichs: Yes.
Benjamin James Kuper-Smith: Um, so this is now the second episode of our book club discussion of Peter again, forces conceptual spaces. Uh, today we'll be discussing chapters three and four, which are about.
I should know this properties. Concepts.
Koen Frolichs: Exactly.
Benjamin James Kuper-Smith: So they're both such generic terms that I just keep forgetting what they are.
Koen Frolichs: Yeah.
Benjamin James Kuper-Smith: Um, it's just like as if they, as if the types are called like stuff and things. That's, I mean, I know like he specifies what it is about, but yeah. I don't know. When I hear those two words, uh,
Koen Frolichs: it's very vague.
Yeah. Yeah.
Benjamin James Kuper-Smith: Yeah. I have to admit, when he also talked about properties, it took me like 20 pages to figure out. He's not talking about the dimensions. Somehow I just made that error when reading the first one, and I was like, this is really [00:01:00] isn't making sense. Yeah. And then, yeah. Yeah.
Koen Frolichs: No, I, I have the same, and actually I have like, sometimes I feel very lost and then.
He brings it back, or for me it gets brought back through a third. I understand. Right. Then it's like, oh, it is like a distance metric in Euclidean space and like, oh, oh, that's what we're talking about. Oh, now it makes sense. Kind of like not, not everywhere, but yeah, a little bit.
Benjamin James Kuper-Smith: I also found those. Yeah, every time you basically gets back to the geometric stuff, I find it way more interesting.
Yeah,
Koen Frolichs: totally.
Benjamin James Kuper-Smith: You know, as we talked about, uh, like before we started recording, I found. A lot of this stuff, especially chapter four, not super interesting and I dunno, I wonder Yeah. But like, I mean, every time when he brings it back to the geometric stuff, I find it much more interesting.
Koen Frolichs: Yeah.
Benjamin James Kuper-Smith: Uh, and I wonder whether that's maybe because I'm reading like, in a way I'm just reading this book, right.
And I, I started this reading with no agenda in that sense, other than Sounds interesting.
Koen Frolichs: Mm-hmm.
Benjamin James Kuper-Smith: But I, I did realize, especially whilst losing interest in the fourth chapter that. [00:02:00] I think the reason why I find some stuff interesting, others not so much, is that I'm not reading, like, so he's writing a book from the perspective of here's my theory, there are other theories that went before it, and I'm going to say how they didn't work and how my theory does work, et cetera.
Koen Frolichs: Yeah.
Benjamin James Kuper-Smith: But I don't care about that. What I'm doing is I'm reading this from the perspective and not like intentionally, but I'm, I realize I'm reading this from the, from the perspective of how can I use this in my own research?
Koen Frolichs: Yeah, yeah. Applied. Yeah. Yeah, sure.
Benjamin James Kuper-Smith: Exactly. For me, it's just like, oh, here's some interesting ideas.
How can I use this? Um, what if this stuff could I, uh, you know, yeah. Apply directly to my own stuff? Yeah. Which not and all that kind of stuff. Of course, I'm in that sense, reading the book incorrectly, you might say, um, I don't think there's a correct or incorrect way, but you know, that's obviously not the book.
He wrote in that sense.
Koen Frolichs: Sure.
Benjamin James Kuper-Smith: But I think because of that, like the entire argument about intentional and ex extensional semantics. Yeah. And like the logical stuff was like, uh, whatever.
Koen Frolichs: Yeah. A little bit. Yeah. But
Benjamin James Kuper-Smith: also for him, that's super important. Yeah, yeah,
Koen Frolichs: yeah, yeah, yeah. [00:03:00] Yeah. And especially to, for him to ground it in his current time.
Right. And like, you know, Ky, I mean, that's the only name I remember of course, like, said this and like some other guys said that, but you know, I think that's wrong because. For me, it's just like, yeah, whatever, you know. Can we euclidean space again, do you, do you remember where you started losing, um, yourself or like your, your interest in chapter four?
Because for me it was very specifically,
Benjamin James Kuper-Smith: um, let's see. Um, I've just flicking through, so chapter three, I think I actually enjoyed almost. I found it really interesting actually.
Koen Frolichs: Yeah, same.
Benjamin James Kuper-Smith: Um, apart from like, the beginning was like super technical. Yeah. And I really had to, I really had to concentrate.
Koen Frolichs: Yeah.
Benjamin James Kuper-Smith: Um, but then also at some point I realized like, wait a minute, I don't need to understand all of this probably. But then in chapter four, let's see. I mean, I have to admit, I think I found almost all of chapter four less interesting. Um, there, I mean, there's not, like, there were, there were obviously some stuff, some things that were interesting in here.
Koen Frolichs: Yeah.
Benjamin James Kuper-Smith: Um, but. I think one problem that I kind of had with chapter four is that it's not entirely clear to me how chapter four is not just a repetition of chapter three. To some [00:04:00] extent. The interesting stuff from chapter four, I feel like was mostly in chapter three already.
Koen Frolichs: Mm-hmm.
Benjamin James Kuper-Smith: Sure. At least from my perspective.
Koen Frolichs: Sure.
Benjamin James Kuper-Smith: So in some sense I was. I'd say chapter four wasn't for me.
Koen Frolichs: Okay. Well, that's very fair.
Benjamin James Kuper-Smith: But what was your specific
Koen Frolichs: Well, for me, like when we got to like non monotonicity, I was just like, yeah, I'm sorry. I just like, I really don't care. Like I, I don't even know fully what it is I have to admit. And, you know, I guess, I guess also fair thing to say is that timing wise, like I am always kind of pressed for time.
So reading this book is kind of just an extra, so, you know, I'm not like, I don't have, I don't have like enough leisure time to like. Google non mono and stuff like that.
Benjamin James Kuper-Smith: Oh
Koen Frolichs: yeah. But I really started, because in the be beginning, a
Benjamin James Kuper-Smith: monotonic function is just one that like increases continue. Like it doesn't go down, basically.
Koen Frolichs: Okay. Yeah. But, but what
Benjamin James Kuper-Smith: point, and if you look at just like, I mean I, I, yeah. If you look at, for example, chapter, I dunno whether you mentioned means this specifically, but like page 130 has figure 4.6. You can say, see that the [00:05:00] one dash line goes up and then goes down towards the ends from vase to flowers.
Koen Frolichs: Oh that. Okay.
Benjamin James Kuper-Smith: I think that's, anyway,
Koen Frolichs: but yeah, that's okay. Yeah, I mean after like that's where I kind of was like, you know, because like as you, as you said, I think same for me. Chapter three was very interesting. I think fir the first half of chapter four was still very interesting for me. Like, 'cause that like, 'cause in chapter three we kind of talk about these properties and I feel like that they kind of get combined or like, you know, they become a whole with concepts, right?
A bunch of properties becomes a concept. I thought that was kind of cool. Kind of lost me. And then like, I thought the experiments actually were kind of interesting 'cause it's like, oh, I thought he was a philosopher. But they still do kind of experiments so they, they have like these shell, um.
Benjamin James Kuper-Smith: Shell. Yeah, so just, I didn't read the experiments at, at all.
Okay. I skipped that part entirely. Yeah,
Koen Frolichs: I went to read it, but like a little unfocused. I after a bit. It's kind of fun to see kind of, because they use models kind of to see how they do that. I mean, I can't like give you much detail. Um, before you ask, um,
Benjamin James Kuper-Smith: I [00:06:00] won't 'cause I'm not that interested. I mean, unless like you say there's like something, one specific reason why it is interesting, then I'd love to hear about it.
But if you also didn't find it too interesting then,
Koen Frolichs: I mean, for me, I think as a, from a, I guess the applied perspective, I thought it was fun to see how they create these shells. You know, by like having like a distance, like a, a curvature measure, like, um, a um, spread, like a height measure, stuff like that.
I thought that was kind of interesting. And then they kind of apply that to like, you know, ma like that is a conceptual space in itself, but then they transform it to like, what humans think of it, because humans don't look at these dimensions like mathematically as like. In the same way as mathematically is like they kind of transform it already and then they do that and then they test it for smart, which I thought was kind of cool.
Benjamin James Kuper-Smith: Okay. That sounds kind of interesting. Yeah.
Koen Frolichs: Yeah. That was like,
Benjamin James Kuper-Smith: maybe I should have read it. Yeah, maybe I should have read the text, which I'll think about. Yeah,
Koen Frolichs: you still can. I mean, it was interesting. I, I thought that was kind of cool. And they have like these four different models they use [00:07:00] like nearest neighbor.
And three others like this Vol Roy, how do you call it again? Forno.
Benjamin James Kuper-Smith: Yeah.
Koen Frolichs: Space. Which sounds very cool, right? That's like definite third date, like after Mar and the other one it's OID destinations. Um, you know, so that was kind of cool. I liked the applied like, you know, not just like theorizing about it and like speculating, but also like testing it.
Yeah. I thought that was kind of cool.
Benjamin James Kuper-Smith: Yeah, I mean, one thing that I found kind of interesting about the testing it is that. The, I mean, this is a kind of indirect point, but it's, it's something I've found. I found it slightly difficult about the way he writes this or about some of maybe the way the theory, even like what the theory is.
And that's about, so on page 112 and 13, he has this criticism by, uh, KY and someone else. Um, yeah. Yeah. I love like, as, as soon as your name becomes big enough, it's always that person and Yeah. Who gives a shit, um,
Koen Frolichs: Hutchinson.
Benjamin James Kuper-Smith: Hut. [00:08:00] Um, I mean, I'm sure Hutchinson's is a lovely person and a great scientist. Um.
But uh, by the way, just very briefly, I find it really interesting here to see in this chapter, ma Anderski both cited, but both for things that I don't know about.
Koen Frolichs: Yeah, same. Even with Mon Montague, I think that's the same Montague who did like the bee paper really. Yeah, I think so because it's RM Monga and I think, I think it's, I mean, I didn't check this at all.
Um, I mean, I just checked like in the back the references and I was like, oh, I think that's the same, like Montague.
Benjamin James Kuper-Smith: Montague, okay.
Koen Frolichs: And I was like, he isn't a philosopher. He did like, you know, like cool, like modeling stuff. But yeah, probably he is, no, that's
Benjamin James Kuper-Smith: not him.
Koen Frolichs: No,
Benjamin James Kuper-Smith: I dunno. It says here
Koen Frolichs: it I, it
Benjamin James Kuper-Smith: was Montague are 1974.
Let's put it this way. I saw Reid Montague in like 2015 when I was at UCL. I'd be surprised if he was writing books in 74.
Koen Frolichs: Oh, you right. Yeah. That's not your birthday.
Benjamin James Kuper-Smith: He didn't look that old. He in phenomenal shape if he was writing books in 74 publishing about that.
Koen Frolichs: Yeah. [00:09:00] Okay. So that's so okay, that, that, that, that makes sense.
Benjamin James Kuper-Smith: Yeah, that sense. That's just coincidence then. Yeah. Anyway. But I just found it really fascinating that KY is. So famous for his worth with Kahneman and Prospect Theory and all that kind of stuff. And MA is so famous for, as we talked about last time, his three levels. And so it was just really random seeing like those two names cited both in the same chapter, but about something I'm interested in, but not for the reasons that I know both of them.
Koen Frolichs: Yeah, yeah, sure. Yeah, yeah, yeah, yeah.
Benjamin James Kuper-Smith: And yeah, I mean this also, I once, I can't remember who said this, but I once heard someone say something like. One really interesting thing when you meet someone who's like won a Nobel Prize or something, is that usually apart from their stuff they won the Nobel Prize for, they have this entire other career of stuff that's also really cool.
Koen Frolichs: Yeah, yeah, yeah.
Benjamin James Kuper-Smith: And that seems to be the case here.
Koen Frolichs: Oh, one one. Can I like one quick anecdote? Did you know Einstein never won the Nobel Prize for relativity? You went for something else? Of
Benjamin James Kuper-Smith: course. Good.
Koen Frolichs: Oh, I never knew that. I learned that this,
Benjamin James Kuper-Smith: well, that was, that was kind of the, I think the interesting thing there was that he, [00:10:00] I think it was still.
At least in some circles con, controversial when they gave him the royal prize. And I think some people thought like, ah, they're maybe giving it him for the photoelectric effect, right?
Koen Frolichs: Yeah.
Benjamin James Kuper-Smith: And, um, so that they, they're keeping the door open for relativity. Ah, but then they never gave it to him for that, so I dunno.
Koen Frolichs: Fuck. Okay.
Benjamin James Kuper-Smith: Yeah. But, uh, yeah, so that was just like a random aside that I found really interesting. But the point I wanted to make is that, um, so he has this, uh, so Ky I guess, I dunno where the direct. I guess it's not really criticizing Gaden for, but more like the general idea or approach. And the one thing that he criticized it for is this symmetry part where he says, um, sure.
Koen Frolichs: Yeah.
Benjamin James Kuper-Smith: Tel Aviv, New York. Like New York and Tel Aviv, or Tel Aviv in New York. Like, um, depending on how you phrase it, one will seem more similar than the other.
Koen Frolichs: Yeah.
Benjamin James Kuper-Smith: You know, that kinda thing. And that basically, if you are, if similarities are determined by the distance, then it shouldn't matter which are around you phrase this.
Koen Frolichs: Mm-hmm.
Benjamin James Kuper-Smith: So to which then again, [00:11:00] fast, kind of in general. Or like generally argues that, well, if you are starting with one thing, then that kind of becomes the, uh, not, not the norm, but like that becomes kind of the default and thereby the salience of the different dimensions changes. Yeah. And therefore, like, you know, you, you attend to different things.
Yes. Which makes complete sense. Right. If you think of New York, you have a very specific image in mind. Yes. When you think of Tel Aviv, I mean, I don't have a specific image really of Tel Aviv, but. Other associations being it just like a Mediterranean city or,
Koen Frolichs: yeah, yeah, yeah. I've never there,
Benjamin James Kuper-Smith: I'm assuming it's, dunno whether that actually is a port city, like that's a coast.
I think so, yeah. Anyway, like that makes complete sense to me. But I also find it slightly difficult because in a way it feels to me whenever it, it seems to me it's almost the kind of thing that's. In principle, falsifiable not, but not practically. Because you're right, you can kind of always come up with something like this.
You can always say, well, in this thing the salience changed. Well, maybe we're using a different [00:12:00]metric for calculating the, I like the distances. Maybe, you know, like there, there's, I mean, of course some of these can then be refuted, but it seems like there's always this wiggle of room whereby the general thing.
Is kind of immune to any criticism because you can always come up with some sort of ad hoc explanation of why it's still true, and I found that a bit difficult.
Koen Frolichs: I had this a little bit when he was talking like the, I think in the beginning of chapter four, they talk about. It can't be like the number of, I think, properties or dimensions you share because as like an infinite amount of dimensions you always share, right?
Like the, the piano and a goat or something. Like, they both have four legs. They're they both, yeah. Both above the
Benjamin James Kuper-Smith: ground.
Koen Frolichs: Yeah. Like you can come up with an, an infinite amount of things that they, right. So I feel like you are right, and I was thinking here as well, okay. If you wanna have a property, you need to have boundaries basically.
Right? Within these boundaries it will, it will work. And if not, if we can find something that kind of. Functions or are not functions outside, then [00:13:00] like we have an issue. But you're, yeah, you're right. Um, it does seem like it's a vague or broad enough theory that it's hard to falsify. Um,
Benjamin James Kuper-Smith: I mean, again, in some sense I don't actually mind that much because I'm not reading this book for the theory itself.
It's more like for the general ideas in there, but. Yeah, it's weird because I, I think a lot of this, yeah, I mean, yeah, it's difficult for me to judge in some sense because, you know, the parts, especially where he compared it to these other theories, other parts that I, some of it, I didn't read that carefully.
Koen Frolichs: Mm-hmm.
Benjamin James Kuper-Smith: And a large part of that was because the stuff that I did read carefully, I didn't understand that much because it feels like there was a lot of stuff, background knowledge that I just lacked because I don't know much about first older logic and semantics and all this kind of stuff. So. Um, in some sense it's, I guess, difficult for me to really evaluate the, the value of the theory that he proposes in here.
Koen Frolichs: Okay. Yeah.
Benjamin James Kuper-Smith: But yeah, this was thing, one thing that kind of jumped out to me a [00:14:00]bit where, yeah,
Koen Frolichs: I mean, I, I get it. Like it's easy and like, I think one thing I realized while reading it is like. Because you know, when you read it, it's like his arguments make sense. Right. But that's like you have to remind, so you're reading his arguments.
Right. Because he even says at some point, like, I don't have enough like space to like fully argue, which is very fair. Right.
Benjamin James Kuper-Smith: Sometimes he even said like, this isn't even developed properly at some point in chapter three or four. Oh,
Koen Frolichs: okay. I can't
Benjamin James Kuper-Smith: remember. Yeah. Said at some point, like, in this aspect, this theory isn't developed properly so we can't really say too much about this or something.
So it is just
Koen Frolichs: okay. Yeah, that's very
Benjamin James Kuper-Smith: Includes stuff like that. But. Yeah.
Koen Frolichs: Shall we maybe like do a little dive into chapter three into what properties are? 'cause I thought that was kind of like, as it was kind of developing in my head right? I kind of liked it. Sure. Um, and maybe one thing is, 'cause he always seems to use the same examples.
I, I sometimes wonder. Because he has the color spectrum, I guess. Sorry, I'm doing two things now, but that the examples came up. He, [00:15:00] he either uses the color spectrum as like, you know, a very solid example of, you know, his conceptual spaces. Yeah. Or he uses birds and emus. So I was kind of, I didn't have enough time, but I was like kind of on, on the bike, like trying to like think of other things that would fit his theory and I couldn't come up with any, but I, I guess that's more, um, my fault.
But
Benjamin James Kuper-Smith: doesn't a principle anything fit the theory yet?
Koen Frolichs: I guess, I mean, that's his idea, of course,
Benjamin James Kuper-Smith: or to which, I mean to what. Example, do you mean, because sometimes you could think of anything, right? Like you could think of some two personality traits and you can, uh, see how similar two people are by calculating the distance between them and that stuff works.
Or do you mean something more specific or,
Koen Frolichs: yeah, no, I mean like, 'cause like the, the, um, the visual or like the, the color dimens, like the, you know, this, this, this chromatic thing. The color
Benjamin James Kuper-Smith: wheel. Yeah,
Koen Frolichs: color wheel plus. Then the, the other axis that like, that is so. Perfect for his. It's such a perfect example, but I was, I was wondering are there any, like, this is like the stereotypical, [00:16:00] the prototypical example of his theory kind of.
Right.
Benjamin James Kuper-Smith: Although I think not actually. I think what the reason he uses it is because. So the way I understood that is that he uses the color spectrum as an example of something that's difficult to explain with other parts or to, to show some difficulties, maybe even when using his theory, because you have, you know, because as we discussed last time, the interesting thing about the color spectrum is that we, or the circle is that we perceive it as a circle rather than a linear
Koen Frolichs: Sure.
Benjamin James Kuper-Smith: Uh, increase in color or whatever it might be, right? Mm-hmm. The way we would with heat, well, I guess heat is a bad example, uh, because. I think he, there's some receptors that respond to cold and heat or something like that. Yeah. It's
Koen Frolichs: probably not linear or like, what
Benjamin James Kuper-Smith: Yeah. Anyway, but, but yeah, it's not like a continuous, uh, dimension, but it's, it's one that's circular, so.
Sure. Therefore, these di these, um, distance calculations become more complicated. You know, he used this example where you go, uh, wait, lemme see which figure that is. So I'm not too vague. On chapter, on page 72, figure [00:17:00] 3.5, uh, he uses this as like a possible counter example. The title is possible. Counter example to the convex of color.
Yes. Where he goes through red, but you go kind of through the gray part in the middle. Mm-hmm. And he, he, I think he uses that to kind of say like, that color might seem like an example that doesn't fit his theory. But then he finds a handy way around that problem.
Koen Frolichs: Yeah, sure. No, I mean, I get,
Benjamin James Kuper-Smith: which I agree with though, if your, if your thing is circular, then yeah, don't use a, I dunno, direct,
Koen Frolichs: you can't use like, or like non this way.
Yeah, you're right. Totally. I mean, but still, I still feel like. It's, it's one of the, it's like a prototypical example of where theory works out. Like, I mean, and then it's not me like shitting on it, not at all. But I was wondering is there anything that fits as well, kind of, that has, you know, these multiple dimensions kind of, and like, I, I, I don't think, not like in such a solid example that disen, like, you know, that has these kind.
Components that all belong to the same thing? Kind of, I don't know. It doesn't matter. It doesn't matter.
Benjamin James Kuper-Smith: Yeah. I mean, I think there's, [00:18:00] I mean, I think one reason why he uses Kala is because it's so, one thing I noticed once, while reading something, someone used all sorts of examples. At some point you get disoriented because it's just so many different things.
You just want kind of one example to go through. And I think, you know the, the one thing that I did quite like in chapter four is this example of skin color. Of how
Koen Frolichs: true,
Benjamin James Kuper-Smith: you know, if you say someone has white skin or whatever, it doesn't mean you don't mean white. Mean it's literally white. Yeah. What, and instead, what he kind of has, he has this, uh, what is it called?
The spindle of color. Yeah. And then basically when you talk about certain thing, you just take. Uh, basically you copy that version in, uh, centered around, let's say the, the average of that thing or something like that, right? Mm-hmm. So that then you, it is this reference dependance that he talks about a lot that I'm also interested in, that I found kind of cool.
It's kind of that you. These concepts are not evaluated like absolutely. But relative to the context in which you're talking about them.
Koen Frolichs: Mm-hmm.
Benjamin James Kuper-Smith: In this case,
Koen Frolichs: yeah, sure. Yeah, yeah, yeah.
Benjamin James Kuper-Smith: Skin color. Yeah. [00:19:00] Or hair color or whatever it might be.
Koen Frolichs: Yeah.
Benjamin James Kuper-Smith: And yeah, I guess color is also cool because it applies to so many things.
Koen Frolichs: And I guess the one good thing is that it's very visual. Like literally.
Benjamin James Kuper-Smith: Yeah. And that,
Koen Frolichs: yeah. You know, as an example like that. 'cause that's, for me, that's kind of what, what grounds me? I said this in the beginning, I think the visual examples to thinking about, you know, Euclidean space, which is, you know, three dimensional.
It kind of like brings me back, you know, when we get to these small drawings, it's like, oh, oh, oh, okay. Yeah. This is what we're talking, you know, kind of like, so I guess it makes sense.
Benjamin James Kuper-Smith: Yeah, exactly.
Koen Frolichs: Yeah.
Benjamin James Kuper-Smith: Yeah. No, and I also, I actually quite like the most of the figures actually. I think they,
Koen Frolichs: yeah, they, they're simple,
Benjamin James Kuper-Smith: actually.
Eight, understanding.
Koen Frolichs: Yeah. They served their purpose really well. You're right. Yeah.
Benjamin James Kuper-Smith: Um. I mean, if you have something, you can just start.
Koen Frolichs: So what I really liked about chapter three is kind of how he starts building up the two, like these properties. And in the beginning it seems kind of vague. I mean, I think you said as well, like the first couple pages, you're really like, it was like kind of like a struggle to get through a little bit like Right.
'cause like where we,
Benjamin James Kuper-Smith: it's just the background on the [00:20:00] other theories.
Koen Frolichs: Yeah,
Benjamin James Kuper-Smith: yeah,
Koen Frolichs: yeah. Sure. Yeah. What's the point we're getting to kind of, and then all of a sudden, like it starts to make sense. You get to these examples where you talk about like, you know, this convexity. You know, connecting and disconnected, blah, blah, blah.
And then like it starts to make sense and then he makes it like a little bit less fake by talking about these natural regions of conceptual spaces. I guess that is kind of, that is, yeah, I guess that's maybe is like the non falsifiability, right. Okay. Sorry, I'm kind of vague now. Maybe we should just start, start talking about properties and we get to everything I, I'll, um, I just got to,
Benjamin James Kuper-Smith: okay.
Yeah. The natural properties somehow didn't quite get the point of that or why that was so important. Okay.
Koen Frolichs: I, I wrote, okay. I have to like check my own writing.
Benjamin James Kuper-Smith: Yeah, maybe it was, uh, actually, here's a, here's a related point that's again, a criticism of mine, which is weird because I'm, I, I find this book so interesting and I'm being quite critical now.
But one thing I also didn't like, and this actually, I'm just mentioning this because this appears in the natural properties, uh, sub-chapter.
Koen Frolichs: Yeah.
Benjamin James Kuper-Smith: [00:21:00] Is this whole like evolutionary. Argument he brings. And I mean, to be fair, I think I have in general a kind of skepticism about a lot of evolutionary psychology.
'cause a lot of it I think is just making up stories.
Koen Frolichs: Yeah, sure.
Benjamin James Kuper-Smith: Some of this goes a bit into this, um, region or whatever you might wanna call it. So, and the, I mean, it's, it's in, in page 70 with network properties. It's kind of there, but the strong example for me was on page 82. The question is, why does our innate subjective space of qualities accord so well with the functionally relevant groupings in nature?
Kind of like, why are humans so good at this? Mm-hmm. And, and, you know, why can we. Conceive of everything as a space and that kinda stuff. And then he says, the answer for me comes from evolutionary theory. Natural selection has made us all develop a conceptual space that results in inductions that are valid most of the time and does promote survival.
And this is, you know, one of the things where I go like, sure, I guess. [00:22:00] But it could also not be the case. It's this kind of, I feel like it's, a lot of these evolutionary arguments to me are, they're not even arguments. They're kind of statements that may or may not be true, and it kind of makes sense that it would be, but I'm not really, it's just that like he, I've only mentioned it because you mentioned it twice and in each case it seemed to me a bit like a kind of.
A way of getting rid of the argument almost by just saying, go, that's evolution. We're made for this.
Koen Frolichs: Yeah. Why be because of evolution. Yeah. Yeah. No, I, I, I guess it makes sense. I mean, it is always a compelling, and I think it's a compelling argument and that like, you know, we're still alive, right? So what we, the reason we must be doing a lot of stuff must be because we like, are kind of, you know, evolved to survive in this world.
This world, right? Yeah. But not though.
Benjamin James Kuper-Smith: But not necessarily. I mean, again, I dunno too much about this, but the, I guess the idea of, of evolutionary selection is just that you [00:23:00] can, that traits that are useful
Koen Frolichs: Hmm.
Benjamin James Kuper-Smith: Or that no traits that increase your, uh, fitness,
Koen Frolichs: yeah.
Benjamin James Kuper-Smith: IE your ability to reproduce and for your children to reproduce are more likely to be selected.
But like, it doesn't mean that you need to, that you need this. Like, I dunno if, if moving around randomly. It's an effective strategy for you increasing your fitness, then that's that. And you'll have that.
Koen Frolichs: Yeah,
Benjamin James Kuper-Smith: but it doesn't mean then like that randomness is, per se, a good representation of nature or something.
Right? Like,
Koen Frolichs: but he
Benjamin James Kuper-Smith: does talk, I mean, he doesn't go into this that much. I think he just mentioned it as like a, as an aside, twice. But I think it's just in general, I have a problem when people use kind of evolution as a reason for something.
Koen Frolichs: Sure.
Benjamin James Kuper-Smith: Um, because it seems to me that, yeah, it just, it, it rarely convinces me those kind of stories because Yeah.
Yeah. I guess I haven't thought about this too much either, so. No, no, but it's fine. It can't be too specific with my criticism, um, of a lot of evolutionary psychology, but. Yeah, [00:24:00] just inherently seems, yeah, kind of like a lazy answer that gets you around actually having to think about it.
Koen Frolichs: Yeah. I mean, what I heard now is that you don't believe in evolution.
That's, I heard that correctly.
Benjamin James Kuper-Smith: Look, God, uh, look, I mean,
Koen Frolichs: yes. Yes.
Benjamin James Kuper-Smith: God wrote the Bible.
Koen Frolichs: Yeah.
Benjamin James Kuper-Smith: After creating, uh, Adam and Eve, et cetera. You know, the story
Koen Frolichs: like the rabbi
Benjamin James Kuper-Smith: said. Yeah. And, uh, here we are now and 4,000 years later.
Koen Frolichs: Yeah.
Benjamin James Kuper-Smith: No, no, it's more than four. Oh, okay. Um. Uh, what I'm saying is it shouldn't be in biology books in school.
They shouldn't teach this. Yeah. Really. They should teach creationism. That's a theory.
Koen Frolichs: It's just not enough evidence. It's just not enough evidence. Okay.
Benjamin James Kuper-Smith: Yeah. Anyway, now that we've cleared that,
Koen Frolichs: yes, yes. I mean, it didn't think about it that much. I mean, I think like the whole idea of these natural properties, kind like we work towards it, right.
What is a property it can have, like, you know. I have to like, look, look up a little bit [00:25:00] like it's properties that stuff are, are is something that objects can have in common. Right? So it's that kind of, yeah. Things you can share like, I guess like a color or, or like, you know, I guess that, I guess I go to color myself.
Um, '
Benjamin James Kuper-Smith: cause that's the only example.
Koen Frolichs: Yes. Yes. And then kind of, we, we talk about these properties a lot and then like, he kind of goes to like this, this idea of like, you know, properties can be anything at this point, but then we go towards natural properties that are properties that kind of exist in. Our world, our reality kind of.
Right. But yeah, this is a bit vague, I guess. Yeah. So I guess I'm, I'm also stunting stunted myself here a little bit. Like, but what exactly is that? Like, can I get it in con concrete terms? Not really, but like anything in our world that's not like in abstracting Oh, no, no. Like, like symbolic, like logic and stuff like that.
I think that's something we can't, like anything physical, anything that's really in our world that's natural and like stuff like, I think even like language. Um, it is not natural. Then maybe language, [00:26:00]
Benjamin James Kuper-Smith: I'm not sure. I mean, the, the one thing it just says is I, so I'm not entirely sure what a criterion is actually.
Um, but it says Criterion P and natural property is a convex region of a domain in a conceptual space.
Koen Frolichs: Yeah.
Benjamin James Kuper-Smith: So I guess you can have different kinds of regions, but if it's convex or it doesn't say, if it's convex, then it's natural. It just says natural ones are convex.
Koen Frolichs: Yeah.
Benjamin James Kuper-Smith: Um. Yeah, it's, it's, I mean, just like whilst we're kind of, uh, fumbling over this, I find it kind of weird how at some point I thought like, yeah, in a way, like there's so much in here that, I mean, I guess we both also just lack the background for this, right?
Totally.
Koen Frolichs: Yeah.
Benjamin James Kuper-Smith: Neither of us is, uh, knows much about philosophy or logic or semantics or linguistics or whatever.
Koen Frolichs: Yes.
Benjamin James Kuper-Smith: Um, or anything, but, um, uh, I did feel at some points like, you know what, we kind of should, like, we should almost do this kind of podcast by doing like reading five pages, then like doing research for, for like a few months.
Koen Frolichs: Yeah.
Benjamin James Kuper-Smith: And then coming back to it, doing the next five [00:27:00] pages and then.
Koen Frolichs: I was thinking the
Benjamin James Kuper-Smith: same because there's, there's so much of this where, um, there's just a lot to unpack and to really kind of put into context of actually doing science and thinking about your own stuff and
Koen Frolichs: Totally. I, the one thing that always like discouraged me so bad, it's like, this is 2000, right?
When I was written like 2000, I think.
Benjamin James Kuper-Smith: Yeah.
Koen Frolichs: Like that's 20 years of research on top of this. That's like, I don't know how many books and I don't even get like the two thousands. Right. Um, so that's nice.
Benjamin James Kuper-Smith: Yeah. I, I mean, all you have to do co is read all the references in this one. And all the reference in those ones, et cetera.
And all the people that reference
Koen Frolichs: these people.
Benjamin James Kuper-Smith: Yeah, exactly.
Koen Frolichs: Okay.
Benjamin James Kuper-Smith: Then you'll get it. It's just like a few million pages of thought.
Koen Frolichs: That'll be my Christmas.
Benjamin James Kuper-Smith: Yeah. Okay, so here's a, here's maybe a more specific question that I had, which is, so he mentions on page six or seven that each dimension is embodied with a certain geometrical or topological structure, you know, which makes sense, right?
As, as we talked about, like, some [00:28:00] dimensions might be, uh, continuous variable like height or sure height of, of whatever, uh, or width. Um, others might be some sort of network structure if you're talking about family re like a family tree or something like that.
Koen Frolichs: Yeah.
Benjamin James Kuper-Smith: And this is something where maybe we just actually need to understand some sort of math to see whether this makes sense.
But I was just curious. Can you just combine these like completely freely that you have, like, I don't know, you have a three-dimensional space, but like one dimension is a graph, another is like a circle or something, and the third is continuous or. I mean, I guess maybe to, um, so corn is looking like a d in headlights right now.
Um, so yes. Um, I mean, I guess maybe to, to some extent, I can answer that a little bit myself by saying that I guess with the color spindle, as he calls it, you do have a circle and a di and a continuous dimension.
Koen Frolichs: Yeah.
Benjamin James Kuper-Smith: Uh, with luminance, but yeah, I don't know, [00:29:00] like does, it makes. Because I, I guess when I always imagine these, uh, dimensions, I always imagine them as continuous because it's just easiest to imagine, right?
Koen Frolichs: Yeah,
Benjamin James Kuper-Smith: totally. Yeah, sure. Very continuous dimensions, and then it's really easy to imagine how you move around in that. But I found it a bit harder to do if one is like a graph or something, or.
Koen Frolichs: I haven't thought about it. I mean, what I was really surprised by in the beginning, 'cause he talks like, you know, in, I think in chapter one or two, I, I can't remember, he talks about like graphs as being like, I guess like, kind of like a, um, lower dimension of continuous stuff and stuff like that.
Right.
Benjamin James Kuper-Smith: Uh, can't remember exactly. Okay.
Koen Frolichs: I mean, like, it made sense. It made sense that, I mean, yeah, I guess what I was like, how do you represent gra You can't do like, you know, linear algebra. Right. You can do like graph connectedness, graphs
Benjamin James Kuper-Smith: are done with linear.
Koen Frolichs: It's basically, yeah, it is basically, yeah, you're right.
Like so can't you just have like, kind of like what would be a continuous field with like holes in it or like kind of like connect like that?
Benjamin James Kuper-Smith: Yeah. I mean, that's what I'm wondering, right?
Koen Frolichs: Yeah.
Benjamin James Kuper-Smith: Because you [00:30:00] can imagine like a graph, like where you have like the, you know, have a two dimensional space. Yeah.
And then on top of that you have like circles for graphs. But I don't know. Yeah, and like I have no idea what you do with that mathematically if you want to calculate like distance, 'cause then you have two different kind of metrics in there. And what do you do if one dimension is best represented by one to one by the other?
How do you combine those kind of calculations to get the distance between two things?
Koen Frolichs: I have no idea.
Benjamin James Kuper-Smith: I mean, maybe it doesn't work. Maybe that's where then people just kind of ignore one of it, or I dunno. I mean, yeah,
Koen Frolichs: I dunno either. Yeah.
Benjamin James Kuper-Smith: Our confusion is actually some, a problem inherent to it. I mean, who's more similar?
Your tall sister or your short uncle. Right. In a way that is a kind of really weird comparison, right? Where you go like across kind of different Yeah. Dimensions completely.
Koen Frolichs: I guess that's what he then kind of like this salience thing, right? Like what's, you know,
Benjamin James Kuper-Smith: exactly.
Koen Frolichs: Is it like genes relatedness or like gender or, you know?
Benjamin James Kuper-Smith: Yeah. So maybe like depending on context, you, you, I think so you focus on the height or how close [00:31:00] someone is related to you or something like that.
Koen Frolichs: Yeah. But still, yeah, it's a good one. We should invite Daniel Bassett.
Benjamin James Kuper-Smith: I want, I want to answer,
Koen Frolichs: well, let's, let's email her own and she might know she does graphs, right?
Benjamin James Kuper-Smith: I mean, yeah. I mean, yeah. She's one of the people who does graphs. Yes. I don't think she invented them, but yeah. Did any Bassett, we heard you do graphs.
Koen Frolichs: Yeah. Can, can you do graphs on multiple dimensions with like continuous variables as well?
Benjamin James Kuper-Smith: And she probably won't respond. Um,
Koen Frolichs: can't blame her.
Benjamin James Kuper-Smith: She'll just respond with, yeah.
I can't really block you in my email, but I've, I automatically send you now to the trash, to the waste basket, whatever it's called.
Koen Frolichs: Yeah. Every time I read your emails, it feels like I'm having a stroke, but I'm not sure. Okay.
Benjamin James Kuper-Smith: Actually, uh, one random, I mean, I guess a lot of this discussion is kind of random because, uh, it's more like random thoughts we had while reading it.
Koen Frolichs: Yes.
Benjamin James Kuper-Smith: And another thing that, that I thought that was actually really cool in [00:32:00] that he doesn't really talk about too much, but that to me seemed really interesting is the idea that, um, I mean he does mention that properties are useful for reducing complexity. Right. Mm-hmm. Instead of like having to take in everything, you just extract kind of the relevant dimensions, and then based on that you have a end dimensional vector.
Yeah.
Koen Frolichs: Which
Benjamin James Kuper-Smith: you can use to represent someone.
Koen Frolichs: Mm-hmm.
Benjamin James Kuper-Smith: I don't know, and just as, as he kind of wrote something like that, I was like, yeah, that is a really efficient way. Of storing like information, like even if you, if you've never met someone, you just go by like the most, you know, depending on context again, you just choose whatever dimension is most relevant and then you can kind of, uh, summarize that entire person across a few dimensions.
Koen Frolichs: Yeah.
Benjamin James Kuper-Smith: And yeah, I mean it's, I mean actually many of these things are like super relevant once you think about them, um, and have been said a lot of other times, but it's, I just, yeah. I found it really interesting when you said that. I was like, yeah, that is. Suddenly you can represent someone with, with like three numbers or whatever it might [00:33:00] be.
Koen Frolichs: Yeah, yeah, sure.
Benjamin James Kuper-Smith: Um,
Koen Frolichs: no, that, that is, that is definitely true. I mean, I like to kind of like relate it. I feel like a lot of this stuff, a lot of humans or science have the same ideas and we kind of rehash them right in like a slightly different context, you know, like RSA, stuff like that kind of. I guess RSA is almost like a, a follow up of this and like, uh, maybe not, but like, you know, it's also like you treat neurons or whatever, or like anything basically as like an plain like Euclidean space.
Um, and like do an analysis on that, kind of like on relatedness or whatever similarity. It's like all like very similar things. You just like put in kind of different context and like make it work like that, but like this overarching interest in a lot of this stuff. Comes from the same kind of pool or, or, or, you know, anything like that.
Yeah.
Benjamin James Kuper-Smith: Yeah. And in a way that's, I guess what he's suggesting people do in general, right? If you have conceptual spaces, you're doing is you, you know, how to measure distance, so now you just put everything as distance.
Koen Frolichs: Yeah.
Benjamin James Kuper-Smith: But yeah, I mean, [00:34:00] like one thing I, I, I had whilst I think actually my first note I had, um, or the one thing I wanted to mention at the beginning, but forgot to mention, is that most of the stuff he writes about.
At least the stuff that is, I find interesting I've heard of before. Mm-hmm. Um, sure. I hadn't heard about like the other theories and blah, blah, blah, but most of the stuff, like when he talks about space and how you can like, calculate the distances between these kind of stuff and all that kinda stuff, like, I kind of heard a lot about this.
I kind of know a lot of this already.
Koen Frolichs: Yeah.
Benjamin James Kuper-Smith: But I just realized like he just provides like a, not even completely, but quite different angle
Koen Frolichs: Sure.
Benjamin James Kuper-Smith: Of looking at all of this.
Koen Frolichs: Yeah.
Benjamin James Kuper-Smith: And this kind of, you know, yeah. It felt like this was really like adding, um, another dimension, uh, to my thinking about this.
Koen Frolichs: Yeah.
Benjamin James Kuper-Smith: And um, I also find it then slightly disorienting because I. It's like you feel like you know something already, but then suddenly it like, you know, it kind of slightly shakes the boundary or the, the foundation of some stuff you think you kind of know already.
Koen Frolichs: Yeah, yeah.
Benjamin James Kuper-Smith: By showing like, yeah, you can look at it [00:35:00] from this entirely other way, but because it's not really, because it does that like repeatedly in like smaller ways here and there.
I I never really think too much about each of them.
Koen Frolichs: Sure.
Benjamin James Kuper-Smith: But yeah, I found that kind of interesting that he, yeah. As you mentioned, like a lot of the stuff is already. I mean, in a way, a lot of this stuff, you know, when I'm especially trying to think of how I would use this in my, my own stuff, and I've thought of some very rough outlines of experiments where I thought like, oh, this, I could do this and this and this with, you know, with some of these ideas.
You know, there's super basic stuff in science, right? You just say like, Hey, we're just gonna represent these things as two dimensions and Yeah. Calculate the distance from this or whatever. Yeah. I realized at some point you said, oh yeah, if you have two dimensions, you can actually combine them to create a, a, a higher kind of dimension from this.
Koen Frolichs: Mm-hmm.
Benjamin James Kuper-Smith: And uh, on page 92, he has this, uh, two thing where he says like, one is more than the other. Then that means it's everything below the diagonal, for example.
Koen Frolichs: Yeah, yeah, yeah, yeah,
Benjamin James Kuper-Smith: yeah. Then you can kind of get this concept of that and like, [00:36:00] you know, that's exactly what we already do when we study, like social preferences.
Mm-hmm. If we talk, if we talk about someone as pro-social. What this means is that this person prefers something when it's close to the, the diagonal, for example, the identity line, this kinda stuff, right? So like we're already doing all of this stuff that he's kind of talking totally and we're, yeah, yeah, we're already thinking about all this stuff.
But somehow, like the way he just puts it just makes me think again like, oh yeah, now I somehow understand this more, even though I don't really know how.
Koen Frolichs: Yeah, I, yeah, I, I, and I totally agree. I think that's also how I've been treating the book. You know, I, I guess I'm kind of like, I'm trying to understand philosoph philosophical parts.
I don't fully, because, you know, I lack background. I lack knowledge. Just sheer intelligence probably. Um, I'm just trying to, Hey, um, I'm just trying, I'm just trying to like, I just
Benjamin James Kuper-Smith: wanna emphasize Yeah,
Koen Frolichs: yeah.
Benjamin James Kuper-Smith: Yes.
Koen Frolichs: Yeah, I, you know, it's like Dinner for idiots or what's, what's the movie called? That's why I'm here.
Right. I did that. Yeah. I'm
Benjamin James Kuper-Smith: just,
Koen Frolichs: just for entertainment. Um, [00:37:00] anyway, Blake, I'm really just trying to like, get a different viewpoint and it's actually working. Yeah,
Benjamin James Kuper-Smith: yeah, exactly.
Koen Frolichs: It's hard for me to kind of vocalize or verbalize exactly what I'm getting, but I am kind of thinking about stuff while reading.
Um,
Benjamin James Kuper-Smith: yeah.
Koen Frolichs: And you know, of course, like, you know, it's, it's, I I always tell myself this stuff, like I do wanna reread it in a bit. And like, see if I can get more out of it, which is of course, like, you know, I, I know it's gonna be very difficult to do this.
Benjamin James Kuper-Smith: Oh, I have like three books. I'm really good now where I think like, I should definitely reread this.
Koen Frolichs: Yeah,
Benjamin James Kuper-Smith: yeah, yeah,
Koen Frolichs: yeah.
Benjamin James Kuper-Smith: Let's see where that will,
Koen Frolichs: I mean, you know, after a PG we'll take a year off and I will read, right?
Benjamin James Kuper-Smith: Yeah. I mean, actually, I mean to, I have to say like one thing that I'm, you know, I mean like I have stuff I have to do for my PhD, right? Like some projects that I go that I need to finish and that kind of stuff.
Yeah. And so it's not as if I have like any. Specific time where I can say like, oh yeah, I can definitely like do an entire experiment with this stuff or something, you know? Sure. Like, I mean, I can, but this would be like in a year earliest or something like that. Mm-hmm. [00:38:00] But one thing I, I do wonder whether I, I feel like this has, like reading this book has in some sense like renewed my interest or a kind of faint commitment I've had to actually f.
Finishing the linear algebra course and to also doing more geometry.
Koen Frolichs: Okay. Very good.
Benjamin James Kuper-Smith: Because there's a lot of stuff a well, and like this is really interesting stuff. It also, a lot of it relates a lot to what I'm interested in.
Koen Frolichs: Mm-hmm.
Benjamin James Kuper-Smith: And I should probably learn more about this because this, it seems like there's some, you know, like when I'm reading something, some of the stuff I feel like Yeah.
Similar to what you said, like I'm not entirely sure what I'll get from this.
Koen Frolichs: Mm-hmm.
Benjamin James Kuper-Smith: But it seems like this is an interesting direction.
Koen Frolichs: Yeah, totally. Same. Yeah, no, totally. I mean, I have a unfinished geometry book lying right behind me. Um, I, I blame COVID or something. No. Yeah. But so yeah, I have, I have the same, yeah.
I actually do think that chapter five and six are gonna be a little bit more difficult for us because they're named semantics and induction. [00:39:00]
Benjamin James Kuper-Smith: I have to admit, I'm not looking forward to chapter five.
Koen Frolichs: Yeah.
Benjamin James Kuper-Smith: As I said, as, as say many times, uh, not necessarily on the podcast, but in real life, uh, linguistics is the most boring thing on the planet.
Koen Frolichs: Yeah. Probably. I have to admit. Yeah.
Benjamin James Kuper-Smith: And I know it's very interesting, but not for me. And, uh, semantics seems like the kind of. I know it's like a really interesting question in principle. Yes. But to me it just seems like pedantic
Koen Frolichs: Yeah.
Benjamin James Kuper-Smith: And annoying. I'm not looking forward to like 40 pages of that.
Koen Frolichs: Yeah, no, same.
Yeah, no, same. Um, but I mean, we'll get through it. Who knows? We might like be, we, we might be converted into like fullon linguists. Worship Tomsky, um, within like, you know, next week. Um, I doubt it. Yeah. But I mean, we'll see. I mean, I'll try to be open. I think, you know, as I just said, I'll, I'll try to just try to get out of it what I can and, um.
Try to be open, so,
Benjamin James Kuper-Smith: yeah. Uh, yeah, actually, I just realized the, the semantics is actually [00:40:00] 52 pages.
Koen Frolichs: Yeah, I know. It's, it's
Benjamin James Kuper-Smith: not an induction is only 30 pages. I wish it was the other area.
Koen Frolichs: Oh God. Yeah. We'll see. I mean, like, yeah. Well, I'll, I'll, I'll, I'll rest my way through it. Who knows? It might be very interesting.
Right. I, I just don't know. I mean, just from the title and the pad. Oh boy. Um, yeah.
Benjamin James Kuper-Smith: Yeah. This is gonna be. Interesting.
Koen Frolichs: And not a lot of pictures.
Benjamin James Kuper-Smith: Oh no.
Koen Frolichs: Come on.
Benjamin James Kuper-Smith: I have to admit that there was some parts where, where the pictures got me through it.
Koen Frolichs: Yeah. Yes.
Benjamin James Kuper-Smith: It's a slightly boring now, but there will be some graphs.
Koen Frolichs: Yeah, yeah. Like, oh, I can see like the pictures shining through like the back. Yeah. It's like, okay, yeah, I can do this. Um, and I love the she the seashell, um, pictures. You even, you didn't even get that. You didn't get that reward?
Benjamin James Kuper-Smith: No, I saw the pictures though. Oh, okay. I saw the pictures and my first thought was.
Uh, the resolution on some of these is not that great.
Koen Frolichs: I was, I,
Benjamin James Kuper-Smith: it looks like it was done. I mean, I guess it was, it was from 2000. So my idea that it looks [00:41:00] like some of these graphics were generated in like mid nineties is probably actually accurate.
Koen Frolichs: It's probably because, yeah, he even talks about it, they're experiments in 1994, I think.
Benjamin James Kuper-Smith: Oh, well, there you go. Yeah. Because it does, yeah.
Koen Frolichs: I mean, you
Benjamin James Kuper-Smith: didn't criticize a study. For the graphics when I haven't read the study. Even,
Koen Frolichs: God, it looks like it was made in the nineties. Like yeah, it was actually, um, so yeah, I can't, I can't find it now, but I think it was like he said something, 1994. Um, he even says like, you know, the great things about these shells is that they can be like.
Um, displayed on a visual computer, something like that. So that's like, yeah. Makes sense.
Benjamin James Kuper-Smith: Um, I mean, I guess we are nearing the end here. Yes. But I just have one point that I found was interesting and which maybe goes slightly counter to what's something I said earlier. So one of my earlier criticisms was that.
It's kind of difficult to falsify or that it's sometimes a bit vague and that kinda stuff. You know, what the dimensions are when, which one is most salient and relevant and all this kinda stuff. Right? But one thing I thought was kind of quite cool is, so I kind of had this question. It's like, okay, so how do [00:42:00] I know which metric to you measure something in?
When should I use? Uh, not which metric, which, uh, like measure should I use to calculate? Should I use like a metric? Should I use the city block, um, distances? Mm-hmm. You know, like, which one should I use? Right. Yeah. Because he doesn't really go into it. But then at some point he does mention, he does actually kind of go into this a bit and he says that this Mormon, whoever that was basically suggests that.
You use your prior knowledge as a scientist, you use your prior knowledge
Koen Frolichs: Yeah.
Benjamin James Kuper-Smith: To, yeah. Sorry, this is like a scientific question, not necessarily how people do it, but it's like as a scientist. Yeah. How do you know which metric is most relevant? And this moment suggested basically that you kind of use your existing knowledge to limit what you have, which I guess is a common thing.
Koen Frolichs: Yeah,
Benjamin James Kuper-Smith: I thought, I thought it was kinda interesting you said like, you know what, with color, we know that there are three receptors in the human eye. Sure. So use that to start off with and then you can kind of. That actually, uh, in, in principle you have this probably very vast space of possibilities.
Koen Frolichs: Yeah.
Benjamin James Kuper-Smith: But through [00:43:00] this kind of, I guess what you would, I guess what you could call this is just a biologically realistic model. Then by using the restrictions from a low level. To inform higher level processes.
Koen Frolichs: Yeah.
Benjamin James Kuper-Smith: So I guess as I'm saying this, I'm realizing this is a really basic point they're making, but I did, I did have like, ah, yeah, that's a good point whilst reading it.
Koen Frolichs: No, that's good. I mean, I mean, even, I think the most basic points are usually the best ones to have grounded, right? Yeah. So, but it didn't fully get what you were getting at.
Benjamin James Kuper-Smith: You didn't,
Koen Frolichs: no, I'm sorry.
Benjamin James Kuper-Smith: Uh, like even the problem with like, which metric to use?
Koen Frolichs: No. Yeah, I got that. Yeah. I guess that it's just like that you, yeah.
Okay. No, I get, I guess I get it. It's just
Benjamin James Kuper-Smith: because I guess before I felt like it was slightly, um, it's just at first it was like, okay, here's all these different ways in which you cal can calculate distance.
Koen Frolichs: Yeah, sure. Yeah.
Benjamin James Kuper-Smith: Like why this one or not that one. You know why? Like for example, the city block metric is something I've not really heard [00:44:00] of.
I mean, I had kind of heard of it, but not really. Um, so I'm like. I'm not entirely sure where this is used or why you would use this, or like when would you use this versus another measure of all this kind of stuff like
Koen Frolichs: that. Okay. Yeah.
Benjamin James Kuper-Smith: And yeah, it just seemed to me that he did kind of make it, it did make me kind of believe that this wasn't just some sort of arbitrary thing where you go, well let's, let's just calculate it this way and see what happens.
Koen Frolichs: Yeah, sure.
Benjamin James Kuper-Smith: But that there, yeah, that it does kind of suggest some sort of path for doing this in a sensible way.
Koen Frolichs: Sure. Okay. Yeah. Makes sense.
Benjamin James Kuper-Smith: Yeah. Um, enough rambling for today.
Koen Frolichs: Enough rambling. Yeah.
Benjamin James Kuper-Smith: Next episode will be about good are my favorite topics, semantics and induction. And we are very much looking forward to reading the next 80 pages about this.
So we
Koen Frolichs: might
Benjamin James Kuper-Smith: see you again next time. Gonna be amazing.