This is the fourth and final episode of a book club series on Peter Gärdenfors's book Conceptual Spaces. In this episode, we will discuss chapters 7 and 8, in which Gärdenfors discusses computational aspects his theory of conceptual spaces, and provides a general discussion of the topics covered in the book.
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
First AI conference: https://en.wikipedia.org/wiki/Dartmouth_workshop
Bauby, J. D. (2008). The diving bell and the butterfly. Vintage.
Bellmund, J. L., Gärdenfors, P., Moser, E. I., & Doeller, C. F. (2018). Navigating cognition: Spatial codes for human thinking. Science, 362(6415).
Churchland, P. S., & Sejnowski, T. J. (1994). The computational brain. MIT press.
Gärdenfors, P. (2004). Conceptual spaces: The geometry of thought. MIT press.
Hafting, T., Fyhn, M., Molden, S., Moser, M. B., & Moser, E. I. (2005). Microstructure of a spatial map in the entorhinal cortex. Nature, 436(7052), 801-806.
Kriegeskorte, N., & Kievit, R. A. (2013). Representational geometry: integrating cognition, computation, and the brain. Trends in cognitive sciences, 17(8), 401-412.
Kriegeskorte, N., Mur, M., & Bandettini, P. A. (2008). Representational similarity analysis-connecting the branches of systems neuroscience. Frontiers in systems neuroscience, 2, 4.
LeCun, Y., Bengio, Y., & Hinton, G. (2015). Deep learning. Nature, 521(7553), 436-444.
O'Keefe, J., & Dostrovsky, J. (1971). The hippocampus as a spatial map: preliminary evidence from unit activity in the freely-moving rat. Brain research.
Quiroga, R. Q. (2012). Concept cells: the building blocks of declarative memory functions. Nature Reviews Neuroscience, 13(8), 587-597.
Rumelhart, D. E., Hinton, G. E., & Williams, R. J. (1986). Learning representations by back-propagating errors. Nature, 323(6088), 533-536.
Silver, D., Schrittwieser, J., Simonyan, K., Antonoglou, I., Huang, A., Guez, A., ... & Hassabis, D.
(2017). Mastering the game of go without human knowledge. Nature, 550(7676), 354-359.
-
[This is an automated transcript with many errors]
Benjamin James Kuper-Smith: [00:00:00] This is the fourth part of our book up discussion on Peter Gaden Force's book, conceptual Spaces, the Geometry of Thought. I realized that it has a subtitle, so I thought I'd add it. Um, very good. Yeah. Proud of myself. Uh, so, uh, again, joined by Kun Foles. Hello. And this is, yeah, I said part four. Today we'll be discussing two very.
Brief chapters, the last two, chapter seven computational aspects and chapter eight in chase of space, which is, I guess a kind of summary or outlook.
Koen Frolichs: Yeah. And,
Benjamin James Kuper-Smith: and then we'll also have kind of, I guess, a general discussion at the end.
Koen Frolichs: Yeah.
Benjamin James Kuper-Smith: Yeah. I guess like last time, this is probably gonna be a shorter one because there's just not that much to discuss.
I feel like the, the two chapters are together 30 pages long, and I think. We've kind of, I guess, had a general [00:01:00] discussion through most of the episodes to some extent. So we'll see. Actually, you know, we had, um, for the, I, I usually do, you know, a hundred pages per week. Right. Which we change for this one, just because it's much denser.
Yeah, yeah. Than other text that would've taken forever to read and. We basically said like, usually I wanted to do a hundred pages per week and talk for an hour. And in this case, I guess we just said, you know, what was like sometimes 90 pages, chapters, 30 pages, whatever. Yeah, it was two chapters. Exactly.
It really varied, but I realized, um, uh, I just edited the, the last one. Um, so I think actually if we talk for about 50 minutes a today, we'll be pretty much exactly an hour per hundred pages again, really. So it seems like we pretty much matched, um, exactly why. Yeah.
Koen Frolichs: Your original predictions were just very correct.
Benjamin James Kuper-Smith: Yeah. I didn't without, you know, if we completely lost the structure, I guess to some extent, because last time we read three times as much this time.
Koen Frolichs: Mm-hmm.
Benjamin James Kuper-Smith: Um,
Koen Frolichs: yeah.
Benjamin James Kuper-Smith: But yeah. Anyway.
Koen Frolichs: In [00:02:00] terms of also in terms of density and like ease of reading. Like last, last time was, wasn't only like three times more. It was also like three times as hard as it felt like.
Right?
Benjamin James Kuper-Smith: Yeah, definitely. This was pretty easygoing or I mean, uh, yeah, I guess we can do, um, uh, I mentioned some stuff then later in context rather than now. Um, I guess we should probably start with chapter seven, computational aspects.
Koen Frolichs: Yeah.
Benjamin James Kuper-Smith: Maybe I'll just, uh, I have basically only really one point here, or I mean, maybe first, I guess a brief summary is that my computational aspects, and this is something that I, I had a slightly different expectation when I went into it.
He does mean, uh, in the kind of. Fairly literal sense, as in how useful are different kinds of representations for implementations, for computer implementations of cognitive processes. Mm-hmm. Yes, yes. Um, I was kind of thinking that it might be a bit more about like competition neuroscience or something like that, but it's, it's very much actually about the generative part [00:03:00] of cognitive science, as seems to find.
Yeah, the, the minor point that I had, uh, kind of interesting point I had is that he never actually mentions this, actually, probably because this was written in 2000, the book or published. Um, but it occurred to me that kind of the three levels he's talking about are potentially kind of three waves of ai.
Yeah, because AI kind of started off in this kind of symbolic way, right? Yeah. Uh, there's this, this kind of, I think the classic, the, this like somewhat famous description of a, the first conference or something where they're like, yeah, we're gonna like solve these problems. Oh, do this stuff. Yeah, yeah, yeah, yeah, yeah.
Exactly. That. I, we do all this where that's from. Yeah. We do all this thing like this, this with this week or something. Yeah. And we've made lit of progress on most of those things until today.
Koen Frolichs: Yeah, yeah, yeah.
Benjamin James Kuper-Smith: But that was like the very, like how he describes it also. Right? The, the. During machines, this kind of symbolic, logical stuff.
Yeah. Right. Sure. Yeah. And currently, I guess we're in another. Wave of ai. Yeah. Uh, with the deep learning stuff, especially with, [00:04:00] um, the stuff that DeepMind is doing into large extent. Mm-hmm. I guess they're at least maybe the most famous, but I'm sure there are many, many other companies. Yeah. And organizations do similar things, but that's basically the what Gaden, FOS would called the subsequent conceptual level.
Right. That's the artificial neural networks.
Mm-hmm.
Benjamin James Kuper-Smith: And, you know, that started of course, much earlier, I guess, in the. Eighties or whenever. Mm-hmm. Yeah. But I guess it kind of had like a, a, a start and then they realized, I guess they didn't have any data to use it and now they really have data slash that generat it themselves when
Koen Frolichs: Yeah.
Yeah.
Benjamin James Kuper-Smith: Deep mind creates programs that play chess against each other and that kind of stuff. Mm-hmm. Yeah. But it seems to me we're kind of like in that phase right now. That's interesting. Where it's still very much neural networks, it's still very much this kind of associative stuff. Um, there's of course additions, all sorts of stuff that's been added.
Um, but I think that's still kind of generally the scheme and it seems to me then. The next one might be something like the conceptual level.
Koen Frolichs: Very,
Benjamin James Kuper-Smith: I dunno.
Koen Frolichs: Okay.
Benjamin James Kuper-Smith: I don't work in an ai, but yeah's thought I had
Koen Frolichs: a good [00:05:00] prediction. Yeah, I, I mean, I, I, I just, I just didn't think about this. I actually, I have to admit that for me it felt like almost like nothing has changed in the last 20 years.
Right. When you wrote this book, like a lot of the, the quotes and stuff like that are still like, open questions we have today almost. Right?
Benjamin James Kuper-Smith: Yeah,
Koen Frolichs: definitely. People. I guess it's like, you know, there's this, this optimism you said, but also people just saying like, oh, we have to kind of figure this out. And I feel like a lot of this stuff is, we're still figuring it out kind of.
Right. Um, but I think
Benjamin James Kuper-Smith: that might also be because we're still in the second kind of
Koen Frolichs: Yeah, that could be. Yeah. I mean we might've way
Benjamin James Kuper-Smith: to some extent. I mean, I dunno. Yeah, you could obviously go completely different ways. I have no idea.
Koen Frolichs: You know, but
Benjamin James Kuper-Smith: yeah, it could be.
Koen Frolichs: That's very interesting. I mean, it's the, like, I, I kind of got confirmed in my s especially when he had this quote by John Locke from 1604 or something like this.
Oh,
Benjamin James Kuper-Smith: right,
Koen Frolichs: right. See, see, like, I always like, nothing has changed since. Yeah. John Locke, when was it? 1604, I think. Yeah. Can't find it. Um, but I thought it was like damn, like a
Benjamin James Kuper-Smith: [00:06:00] while ago.
Koen Frolichs: A while ago. Yeah. Like at least. 1690. Oh, sorry. I, I lied. 84 years
Benjamin James Kuper-Smith: corn you. Yeah.
Koen Frolichs: That's not too long ago. I'm sorry.
Benjamin James Kuper-Smith: You almost skipped a century.
Koen Frolichs: Yeah,
Benjamin James Kuper-Smith: but it's still 300 years ago. Yeah,
Koen Frolichs: yeah. Yeah. I mean, yeah, I mean it's, it's a, it's a different topic to think about, but it, but like, you know, these, these, these, a lot of this stuff in chapter seven from like Patricia Churchland and, and Sinofsky, you know, that's still very relevant today. Um, which I think is interesting.
Yeah.
Benjamin James Kuper-Smith: Yeah, definitely. I mean, I guess, I mean like, I guess something we already discussed like in the early stuff, is that, uh, like in the early chapters Yeah. Or discussions, is that, I guess to some extent this is a weirdly, I. Not famous book, at least in the circles we are in. Yeah, to a large extent. Yeah.
Yeah. And also Peter Gfo, I don't think many people know necessarily. Um, but [00:07:00] then, actually let me just check. I think. Just so I don't talk nonsense here. I think the book is cited quite a lot.
Koen Frolichs: Okay.
Benjamin James Kuper-Smith: Um,
so Google says it's from 2004.
Koen Frolichs: Yeah, I think there's some.
Benjamin James Kuper-Smith: Oh, it is. Uh. The paperback edition from 2004. Okay. Um, copyright is 2000, but I guess they, that was probably a hard cover. Oh
Koen Frolichs: yeah.
Benjamin James Kuper-Smith: Um, anyway, this book has been cited 3,258 times according to We've Got it. Yes. Not this is a very influential book and it's also been cited.
It's been cited pretty much since like 2005. It's been cited like 150 to 200 times a year pretty consistently. So this is not a, this is not a niche thing, but somehow it's, um, so I, I wanted to, like, the reason I mentioned this is because I wanted to, um, address the point you made about not much has changed.
Yeah. And it seems to [00:08:00] me that to some extent, I think it was just in a field where people haven't really. Looked at this or, I don't know. It's, it's really weird because I, whilst reading like chapter seven and eight, I did, I did feel sometimes like, oh yeah, this is exactly what I want to do and these are some people I've heard of before and then often just, I have no idea who these people are and
Koen Frolichs: Sure.
Benjamin James Kuper-Smith: Yeah. Yeah. So it's this really weird thing where to some extent, like, I guess the, the progress might not have been made because we. Uh, because the people re we read and the stuff that we do somehow hasn't quite picked it up as much as other fields maybe.
Koen Frolichs: Yeah. Could be. Could be. Um,
Benjamin James Kuper-Smith: but yeah, it's,
Koen Frolichs: yeah.
That's interesting. I mean, 'cause I like in the beginning, I think in the first episode we talked about that. You know, 'cause I kind of, I looked up some papers like the RSA paper, stuff like that. Right? Like, like the, the, the, the first R RSS a paper. Um, like if this book was, um, cited, but it wasn't, right.
Which I think like, 'cause the RSA, that's, that is like conceptual space in a [00:09:00] nutshell, right? I maybe not. I mean, but, but like it's, it definitely uses the same ideas, right. But yeah, for some reason it's not. So you're right. Maybe it just didn't make a big, in fact, impact and, uh, cognitive neuroscience.
Benjamin James Kuper-Smith: And I'm just looking right now at kind of who is citing this.
Koen Frolichs: Yeah.
Benjamin James Kuper-Smith: Okay. Interesting. The second most cited one that's citing this one is a book on, it's called Party Policy in Modern Democracies. No idea why they're citing. This sounds interesting
Koen Frolichs: though.
Benjamin James Kuper-Smith: So there is one. Sorry.
Koen Frolichs: Can we do
Benjamin James Kuper-Smith: that next time? Exactly. Um, yeah, it seems to be more the philosophy side. That's such actually goun.
Koen Frolichs: Mm-hmm.
Benjamin James Kuper-Smith: You should know. How have you not seen? So representational geometry, integrating cognition. Confrontation, and the brain by kri Scott in. Ki, is that how you said it
Koen Frolichs: could be? Oh, okay.
Benjamin James Kuper-Smith: Uh, that's the, the trends in cognitive science 2013 paper. Apparently they're citing this.
Koen Frolichs: Okay. So he does know it.
Okay. So I guess it's, there's not enough overlap then [00:10:00] for the RSA paper, you know? 'cause if he's aware of it, and it would've been. Relevant for the SA paper. I guess he would've cited it properly. Right. But, but well,
Benjamin James Kuper-Smith: they did, right? Or is this a different paper?
Koen Frolichs: It's a different paper. I think it's kli and is there,
Benjamin James Kuper-Smith: yeah.
Yeah. But I guess it's still, yeah. Anyway. Um, I mean
Koen Frolichs: it's still, yeah. Okay. It is relevant. It,
Benjamin James Kuper-Smith: it's, they are kind of citing it, but yeah. It seems to me that a lot of, if I just look through the citations here. Um, language of color. I'm not surprised. Meaning changing grammatical? Yeah. I mean, I guess the stuff, yeah, it's kind of, I guess you have these, like, these, these small world networks kind of mm-hmm.
Citation patterns. Yeah. And that's the same here. So the stuff that they're, that he's citing in this book, they seem to cite this side impact. Yeah. Um, and a bit that, a bit of stuff we're interested in also, but most of the stuff here. Yeah, there's, there's language stuff, there's, I mean, the whole semantics, like to say, right?
An AI kind of stuff. Um, yeah, there's a lot of language stuff here. Actually. Most of this seems to be language. I mean, it's also, yeah. Sorry
Koen Frolichs: to interrupt.
Benjamin James Kuper-Smith: No, [00:11:00] so just to finish that, maybe that thought is just, it seems that maybe, you know, I'm, I'm now at the, like, 70th, uh, page of this.
Koen Frolichs: Yeah.
Benjamin James Kuper-Smith: And, um, I've recognized very few names so far.
Koen Frolichs: Yeah. Okay.
Benjamin James Kuper-Smith: Um, yeah, there's a. It, it seems to be like language philosophy kind of people. And now, okay, now on page eight, uh, we get, uh, Koon's paper navigating cognition.
Koen Frolichs: Got it.
Benjamin James Kuper-Smith: Um, but that's the second basically that I cognize. Yeah.
Koen Frolichs: Yeah. Okay. Yeah. So, but I, I, I think it might also be because it's not as applicable, right?
These are ideas that he tries to put in a framework, but like, there's not that much, I mean, there's some examples, but not for, you know, as cognitive scientists. Like how do I apply this? Right. And I think that's might be, it's like interesting to think about, but there's no real, real life examples, right? I think there's like one study that he cites from himself [00:12:00] with the sea seashells, right?
That's applicable, directly applicable, like, but very little. That's that you can just take and put into an experiment, right? So I guess, yeah. It's a, it's a level of extraction away from what, what scientists
Benjamin James Kuper-Smith: think. Yeah. I mean, like one thing I've said early on, and I'll probably say again later, is that like I've, like, I find this really interesting and I think this is gonna be at least.
At least a minor part in my thesis. Sure. Potentially more. Uh, 'cause I do find this stuff really interesting, but I don't exactly have a specific idea right now of what I do or how I do this.
Koen Frolichs: Mm-hmm.
Benjamin James Kuper-Smith: Um, it is still kind of a vague idea of like, okay, seems like it's really cool and I want to spend some time integrating this, but there's no, as you said, it's not like obvious immediately what you would do with this necessarily.
Koen Frolichs: Yeah. Yeah. No, it makes sense. Yeah. Yeah. No, exactly. I mean that was kind of what was fun about chapter seven 'cause it kind of turned around to stuff. You know, I can kind of understand, again, I can relate with, again, especially the first half the chapter, right, the, the churchland synsky kind of [00:13:00]stuff.
Yeah.
Koen Frolichs: That I've kind of read before or looked at before. That kind of made, made it like, you know, tangible again, which makes read reading it so much easier and you know, also a little bit more fun. But still, yeah, I mean it's still like a level of, of, of abstraction away from what the way I like to think about, which is like, you know, can I put this in a matrix and can I do like, you know, an analysis on it kind of.
Right. Um, is there a MATLAB
Benjamin James Kuper-Smith: function that will calculate this?
Koen Frolichs: Exactly. Yeah. Yeah. Is it significant?
Benjamin James Kuper-Smith: Yeah, exactly.
Koen Frolichs: I mean, there was this one quote by. Patricia Churchland. I'm not sure if it's uh, you mean
Benjamin James Kuper-Smith: 1, 2, 3, 6?
Koen Frolichs: Yeah, exactly.
Benjamin James Kuper-Smith: Yeah, I had it at that one.
Koen Frolichs: Yeah, me too. But I'm thinking, does it fit into what I'm saying now?
Let me have a quick read.
Benjamin James Kuper-Smith: Oh, no, we can just talk about it anyway, I guess.
Koen Frolichs: Yeah. Should, should I read the quote or do you wanna do the quote?
Benjamin James Kuper-Smith: Do it.
Koen Frolichs: I do it in my own voice. I will not, um, I don't know what Patricia,
Benjamin James Kuper-Smith: exactly, because you have no idea what she sounds like. You can just randomly, I guess, go, [00:14:00] uh, you can just randomly take a stereotype of an.
Presumably American woman and pretend, but yeah, just do your own voice. It's probably
better
Koen Frolichs: no voice. I therefore propose that the hypothesis that the scattered maps within the cerebral cortex and many sub cerebral laminate structures as well are all engaged in the coordinate transformation of points in one neural state space into points in another.
By the direct interaction of metrically deformed, vertically connected topographic maps. Their mode of representation is states state space position. Their mode of computation is coordinate transformation. Reading a quote on a podcast feels like. Sitting in school and having to read. Um,
Benjamin James Kuper-Smith: oh really? I actually, I always no called, reading in school was always the best thing because you could kind of pretend to be engaged without having any idea what it, what was going on in class.
Koen Frolichs: Yeah.
Benjamin James Kuper-Smith: So I was like, who wants to read it? So I always, I always read everything I could in class because it clever boy, I didn't have to know what any, what we were actually talking about. I just had to be able to read a [00:15:00] section that the teacher told us I should read. So.
Koen Frolichs: Well, if there's any young people listening
Benjamin James Kuper-Smith: advice from that.
10 year olds. 10 year,
Koen Frolichs: exactly.
Benjamin James Kuper-Smith: Number one. Good to bad. Uh, it's probably too late.
Koen Frolichs: Yeah.
Benjamin James Kuper-Smith: Uh, and, uh, welcome I guess to whatever this world is
Koen Frolichs: Exactly. Yeah. Um,
Benjamin James Kuper-Smith: yeah,
Koen Frolichs: so, so what I like about the quote, let, let's get back to the quote. I hope I read it well enough for people who, who didn't, um, um, we'll understand.
It's, it's this idea of that neurons when they communicate with each other, right? Like, so let's say you have, like I mentioned, like, you know, a bunch of neurons that are connected that represent something and by way of communicating to other neurons that, you know, like a step away, the communication is already the transformation and like the, this representation changes from one set of neurons to the other others.
And I find that a very intriguing. Yeah. Way of thinking about the brain. Right? So, so like visual cortex, we kind of know, I, I am not sure if that, if [00:16:00] we know, but like it kind of happens there, right? From layer one or from V one, I mean, sorry, to V two V three. There is these transformations that are happening, you know, and it must be from the neurons in layer one, communicating it to layer two.
You know, they, they do these transformations and I think that's
Benjamin James Kuper-Smith: layer or.
Koen Frolichs: Um, um, sorry, not layer. Yeah. Um, phase V one to V two. Um, not, not like cortical, uh, layers, of course. I'm sorry. Cor,
Benjamin James Kuper-Smith: same.
Koen Frolichs: Yeah. And for some I find that a very interesting idea. I think it's really cool. But then also, like this is already like slightly abstract of like.
You know how, but how, how do I test it? Right. That that's, that's me coming in. Yeah. Okay. Okay. That's kind of the
Benjamin James Kuper-Smith: point you made. Yeah. I guess that is, yeah, that is a good point because I, I mean, I also highlighted that quote in particular because, you know, I think what I want to do in the next few years is kind of see how people in these social interactions make decisions that have outcomes for them and for others, and kind of how you integrate all that different information and kind of how, how you just use that to make a decision.
And so when I highlighted that, it's like. So I'm, [00:17:00] this is what I'm, I think I will be working on for the next few years, but I'm aware that this is a more general question than just about social interactions. Right?
Koen Frolichs: Sure.
Benjamin James Kuper-Smith: It's, it's just, it's a kind of more general question about how the brain uses and transforms information to some extent, you
Koen Frolichs: Yeah, yeah.
Totally,
Benjamin James Kuper-Smith: totally. Put it in a really broad way. And so when I read this, I was like, oh, yeah, here we go. But you're right, there's no, it doesn't lend itself immediately to some sort of experiment or something. It's just a general way of thinking about it or something like that.
Koen Frolichs: Yeah, yeah, yeah. Exactly. Yeah.
And, and even like, and, and I'm, I can't capture this, right. Um, probably, I mean, who knows? That's, you can write next thing.
Benjamin James Kuper-Smith: Yeah. But Well, maybe, yeah, who knows?
Koen Frolichs: Yeah.
Benjamin James Kuper-Smith: I
Koen Frolichs: mean, they
Benjamin James Kuper-Smith: can do interesting stuff. I guess wasn't made for fmri because I guess the stuff that the, the kind of fields where you look at, I think, I think the fields that have done more when it comes to.
This kind of stuff are the fields where you can measure single neurons and that kind of stuff?
Koen Frolichs: Yeah, yeah. Like animal literature. You mean like, like
Benjamin James Kuper-Smith: the grid
Koen Frolichs: cells
Benjamin James Kuper-Smith: came from
Koen Frolichs: rats. Yeah, yeah, yeah, yeah,
Benjamin James Kuper-Smith: yeah, exactly. [00:18:00] And that, but like even lower level stuff, right? Where they, I think like a lot of this computational, like, I think that a really precise computational neuroscience, uh, where you can really measure stuff precisely.
Comes from when you can.
Koen Frolichs: Yeah. Yeah.
Benjamin James Kuper-Smith: Um, yeah.
Koen Frolichs: Yeah, you have like a very accurate representation. This is what this one neuron does, and then you can model
Benjamin James Kuper-Smith: that. Exactly.
Koen Frolichs: Make it multiple.
Benjamin James Kuper-Smith: Exactly. Yeah. And I think, um, what, I mean, I guess it's to some extent also the interesting stuff about cognitive computational neuroscience, right?
Where this is just an open question, like how do you do this With the myths we have, can you, can you do approximate this? Or whatever, right?
Koen Frolichs: Yeah, totally.
Benjamin James Kuper-Smith: So I think to some extent it's also just a sign that our field, I guess, is not quite as far as others that started, you know, I guess like FMI has been around since 1990 roughly.
So, um, you know, whereas we've been able to record single cells for quite a while.
Koen Frolichs: Sure. So, yeah.
Benjamin James Kuper-Smith: Yeah, yeah.
Koen Frolichs: Yeah. I mean, yeah, I guess the, the field is maturing and we're, you know, we're borrowing a lot from, from other disciplines. Um, or I [00:19:00] guess we, we got along quite a way with borrowing, and now we kind of, you know, I, I'm not sure if physics has some way of like.
Yeah, it's like different, right? It's like it's, we're, we're measuring different things and we're trying to represent different things. Like, I, I don't know. I mean, I just don't know enough. I mean, talking outta my answer, but Yeah. Um, like, you know, we can't borrow. Yeah, yeah. No, no, but like, we can't borrow ideas.
Benjamin James Kuper-Smith: The goes straight from Konz as to your ears.
Koen Frolichs: Yeah. Yeah. Why, why do you always turn my my camera off? Like you don't,
Benjamin James Kuper-Smith: anyway, sorry, you were saying.
Koen Frolichs: Um, no, no, no. Um, like I'm just saying like we, like we, we got along quite well from with borrowing from other disciplines, right? Yeah. Like, you know, computational cognitive neuroscience is, is borrowing from mathematics and physics, you know, how did they model the world? Can we adapt it to neurons and stuff like that?
And. I guess this is not even modeling 'cause we can model single neurons, but measuring, you know, [00:20:00] vast amount of single neurons, you know, preferably in humans, preferably in deeper cortex or like, you know, not just like the motor cortex, which you do for like this, um, this ECCO patients. Right. Um, the epilepsy.
Yeah, right, that you can do like a little bit of human single cell recording or like, I guess, uh, yeah, I
Benjamin James Kuper-Smith: think that's usually like temporal cortex and that kinda stuff because for some reason Yeah, that's why those temporal cases originate. Well, you, it's not just like, you know, you don't have just everything, but apparently I think it's often temporal cortex and Oh, okay.
Koen Frolichs: Okay.
Benjamin James Kuper-Smith: Or temporal lobes, whatever you wanna call it. Mm-hmm. And then you, um, you know, you get some area around it, so you do get motor cortex and prefrontal cor, that kind stuff.
Koen Frolichs: Okay.
Benjamin James Kuper-Smith: And it depends on the patient. Yeah, totally. Um, but I, but I guess the point is also that to some extent our field is on the conceptual level, whereas the stuff that we just discussed in single cell studies, whatever, that's more the sub conceptual level.
Right. So to some extent, also, we just need a new. I mean, maybe that is what we're doing. Maybe that is what all these [00:21:00] FMI analysis are doing. They're just adding the appropriate new way of thinking about these kind of things and collecting and analyzing data.
Koen Frolichs: Mm-hmm.
Benjamin James Kuper-Smith: That you can't just take one-to-one from.
Other fields.
Koen Frolichs: Sure.
Benjamin James Kuper-Smith: Um,
Koen Frolichs: yeah, yeah, yeah. That, that, I mean, yeah, probably. Yeah.
Benjamin James Kuper-Smith: Yeah,
Koen Frolichs: yeah.
Benjamin James Kuper-Smith: But yeah, I guess neuroscientists I guess, are more like magpies where we just take the cool stuff from other fields and just like the shiny stuff and we just make it, um, sometimes, yeah. Maybe we have to have our own thoughts.
I don't know. Not we're screwed. We need our thoughts.
Koen Frolichs: Yeah. Peter, write a new book. Come on. We need it. Yes.
Benjamin James Kuper-Smith: Yeah. Can you please be a bit more applied this time?
Koen Frolichs: Yeah, yeah, exactly. Could you try to, yeah, so we can like implement it, you know? Got some good publications.
Benjamin James Kuper-Smith: I mean, I think that might be, I don't remember it now because, uh, now it's been a bit since I've read KO's, uh, Yakob Bowman's paper.
Koen Frolichs: Mm-hmm.
Benjamin James Kuper-Smith: I guess that is something that goes more towards making it applicable and understandable Yeah,
Koen Frolichs: sure.
Benjamin James Kuper-Smith: To people like [00:22:00] us. Yeah. Um, I mean that's, you know, a five page review or something, so obviously you're not gonna get too much from it in the, the, you know. It's an introduction.
Koen Frolichs: Yeah.
Benjamin James Kuper-Smith: Um, and a summary.
But yeah, I mean, this is a, a book, a philosophical book by a philosopher that Yeah. Seems to be most interesting to philosophers and linguists. But I think, I mean that's, that's what I find interesting, right. That also that you go this, like you, you really just, and think, okay, this is really cool. But almost to me, like the, the, the fascination comes almost from that.
It's not immediately obvious what you should do, right?
Koen Frolichs: Yeah,
Benjamin James Kuper-Smith: yeah. I guess, yeah. But there is a bit of mystery, a bit of like, huh, I wonder how this fits, or how, or whether it fits or,
Koen Frolichs: yeah,
Benjamin James Kuper-Smith: yeah.
Koen Frolichs: Yeah. I mean that and just taking you out of your comfort zone a little bit, right? Like. The, the trying to comprehend something that is not straightforward right away.
Like it's very related, but like you do have to like, kind of like have a different view angle or whatever. And I think that's very good for us to like, you know, look at stuff from a slightly different angle, [00:23:00] like almost literally, and like be like, Hey, but I. If I look at my stuff from, from this angle, will it like give me a new, you know, a new better viewpoint?
I mean, he even said it in the last chapter, like the most breakthroughs have happened in science. When people got a new way of representing data that gave it a nice structure. Like something like that. I don't know. It's something,
Benjamin James Kuper-Smith: yeah, he's said that several times also previously. Yeah. When he basically said, I think almost like what science is, is figuring out new dimensions or something like that.
Koen Frolichs: Yeah, yeah, yeah, yeah.
Benjamin James Kuper-Smith: Yeah. Anyway, so we talk about chapter eight, then, I guess we're already kind of moving, we've been moving it towards it quite a bit already.
Koen Frolichs: Yeah.
Benjamin James Kuper-Smith: Yeah. I guess, yeah, I mean, I guess chapter eight is just a summary. Um, I think the, the main, the two main points kind of, I, I wrote out here as a summary of it, are that conceptual spaces are bridged between symbolic and sub conceptual representations and that it's not.
One or the other, but it's kind of, you need all of them and they have to interact.
Koen Frolichs: Yeah.
Benjamin James Kuper-Smith: And the second point is, I guess something that we've discussed or Yeah. Mentioned [00:24:00] briefly a few times that this book. I think we, at first we thought like, oh, it's kind of cool that he mentioned stuff that he, that they haven't figured out yet.
But then in the, um, in chapter eight, he explicitly mentioned this is probably more like a research program rather than a theory. Yeah, yeah. And then I was like, ah, right. Yeah. Yeah. That's a good way of putting it. Because this is very much a kind of like. Here's a whole way of thinking and we can apply it to all this kind of stuff.
Koen Frolichs: Yeah.
Benjamin James Kuper-Smith: Rather than here's my theory of gravity or whatever.
Koen Frolichs: Yeah, yeah. This is like a finished thing and it applies to everything kind of, right? Oh, yeah. Yeah, exactly. Yeah, you're right.
Benjamin James Kuper-Smith: I mean, that was also like, that point was also kind of cool to me for my, um. Again, for my thesis, I mean, I haven't started writing it, so to some extent it feels kind of silly mentioning it, you know, and talking about it if I haven't really done anything yet.
But to some extent I feel like my, I also wanna do something a little bit in the idea of saying like, Hey, these transformation I'm interested in, this is the kind of stuff that's cool [00:25:00] and that could be cool for, to do the future. Yeah, almost. And using the stuff I have done in my PhD almost works as, as examples.
Koen Frolichs: Yeah.
Benjamin James Kuper-Smith: And. I don't know. When I read it, I felt like, I'm gonna remember this word. I'm gonna put this in there somewhere at the very least as an excuse for why I'm not precise enough.
Koen Frolichs: Yeah, yeah, yeah, yeah, yeah.
Benjamin James Kuper-Smith: At the very least. It's a good caveat at the end, by the way, if, if this wasn't precise enough, it's more of a research program.
Koen Frolichs: Yeah, yeah. It is like theory. Yeah. Exactly. Exactly. Imagine Christopher having two as Christopher supervisor having two of like both our um, um. Dissertation on his desk with both the same and then going like, yeah, it's more like, you know,
Benjamin James Kuper-Smith: also literally it's like he handed this in the same time. I dunno who plagiarized whom,
Koen Frolichs: yeah,
Benjamin James Kuper-Smith: exactly.
His thesis.
Koen Frolichs: Yeah. No, we just, yeah.
Benjamin James Kuper-Smith: But it is also, you know, I guess, I guess like when we think about, I guess when I think about. Science and stuff that people publish. I think I do still have this notion that you publish [00:26:00] your finished stuff, you know, you say, this is my experiment, this is what I found. Yeah.
And it's done. This is conclusive. Here's my fully fledged theory with formal model and everything.
Koen Frolichs: Yeah.
Benjamin James Kuper-Smith: But obviously that's not how it works, right? No, no. I mean, and I know like I've read papers where you, you read a paper and you go like, okay, this is like step one, basically, or something.
Koen Frolichs: Mm-hmm. Yeah.
Yeah. Which
Benjamin James Kuper-Smith: is kind of,
Koen Frolichs: yeah.
Benjamin James Kuper-Smith: Yeah. Just reading it as a book like this just. Felt, I don't know. Yeah. The kind of thing you kind of already have is understood. Select 95%. Yeah. But then someone just hits it or like hammers home the punch. He doesn't hammer it, he just, he just mentions it. Right. Um, but that kind of just makes it, that, that little bit more obvious than it maybe was before.
Koen Frolichs: Sure. Yeah. Yeah. That makes, makes sense.
Benjamin James Kuper-Smith: Yeah. And maybe just a last point I have, uh, about this is that the questions at the end, and this is kind of cool and kind of disappointing. We kind of covered most of them in our discussions before it seemed to me. So he kind of at the end says like, oh, here's some interesting [00:27:00]questions for the future, and that kind of stuff, right?
Mm-hmm. Yes, yes, yes, yes. There was one in particular where I thought like, yeah, we've literally talked about this on the podcast. So for example, page 20 60, 1 of the pointers, which are the best dimensions, the best apology, the best metric to be used for, for representing a particular cognitive process. I think I literally asked you almost that question, like,
Koen Frolichs: yeah,
Benjamin James Kuper-Smith: two episodes back or something where I was like, you know what's really interesting?
Like it's cool to hear this, but like how do you know which one you should use? And that kinda stuff.
Koen Frolichs: Yeah, yeah,
Benjamin James Kuper-Smith: yeah. And yeah, so it was kind of weird like seeing the questions. Like a lot of them we covered, it's like, hmm. Like it's cool that we covered them, but also kind of wanted something new now.
Koen Frolichs: Oh, like that.
Yeah, yeah. I know. Yeah, sure. But I think this is also like, you know, because they kind of, the, the kind of broad, very broad questions that you can probably reapply to almost all problems you have, right?
Benjamin James Kuper-Smith: Mm-hmm.
Koen Frolichs: Like, okay, so I have like, you know, I have like this, this theory or like I have like this, let's say cognitive process.
How is it represented? Like, and then the next [00:28:00] question is, how is it, how are the representations between, you know, this cognition, like this thing of cognition and these other, like connected, how are they connected?
Benjamin James Kuper-Smith: Yeah, yeah.
Koen Frolichs: What are the dynamics, you know, how should it be modeled? Like, I feel like it's, it's also very broad.
Benjamin James Kuper-Smith: Yeah, they're, they're kind of generic questions.
Koen Frolichs: Very, yeah. Yeah. Kind of like, you
Benjamin James Kuper-Smith: know, like, yeah. Sorry,
Koen Frolichs: sorry. No, I mean, like, kind of like, like, you know, if you follow like this framework, um, quote unquote, um, of like conceptual spaces, that's kind of the questions you want to have fulfilled or answered at the end, right.
Um, of, of your experiment.
Benjamin James Kuper-Smith: I mean, that's, that's kind of what I find cool about this, is that he almost. He doesn't do this explicitly per se, but when you follow this book and you want to apply to your own stuff, it almost provides you with a bit of a, almost like a checklist in terms of questions you should ask.
Sure. Which is one dimensions.
Koen Frolichs: Mm-hmm.
Benjamin James Kuper-Smith: How are they measured?
Koen Frolichs: Yeah.
Benjamin James Kuper-Smith: Uh, how do you measure distances in your thing and all this kind of stuff. Right. Yeah. Whereas it's just like natural questions that you have now and that you can always ask when you're [00:29:00] doing something and you know, even if they're not really useful, they might at least make you think of something interesting or something new, you know?
Koen Frolichs: Definitely. Yeah, definitely. Definitely. And I think that's also for me that I think, I don't know exactly what I slightly more, in the beginning, I didn't write this down, kind of what I, what I wanted from this book. I think we asked, asked each each other, but that's kind of, maybe I've transformed into that, but.
Like that was also for media. Id just get a new viewpoint. Right? Get a, you know, have a different look. Something refreshing. Well, maybe not refreshing, but you know, just like another, another person's ideas. And I think, you know, if I can talk about the book as a whole, I think that kind of, I think
Benjamin James Kuper-Smith: we can move towards that.
Unless you have something about chapter
Koen Frolichs: eight. Yeah, no, I mean chapter 10 or chapter eight is, anyways, they kind of summarizing Yeah. Already
Benjamin James Kuper-Smith: that, yeah.
Koen Frolichs: But yeah, I guess that's. In itself in a whole, like, maybe not these questions specifically, but also like, you know, this distance metric, right? Um, you know, not just eu Euclidean distance is, is valid.
I guess, you know, I kind of, you know, had some ideas about this already, but like also the city block metric. Um, [00:30:00]
Benjamin James Kuper-Smith: yeah,
Koen Frolichs: that kind of stuff. Right. That's very interesting to just have added to my repertoire or, or whatever to think about.
Benjamin James Kuper-Smith: Exactly. Yeah. Yeah, I think that's a good way to think about it. A kind of repertoire of questions and, uh, things you can ask.
And I mean, to some extent, of course, this stuff might be almost too much detail because it, like, another thing I had that's like, okay, so now you have these. Kind of Cartesian space is then also, uh, uh, in that sense, I guess at least you think about linear usually, right? You just add up stuff or not. Um, but it can obviously also be non-linear, but then it gets very complicated very quickly and just it's becomes too difficult to, to deal with.
Koen Frolichs: Yeah.
Benjamin James Kuper-Smith: Um, but yeah, I def so it might, what I mean is like, there might be like. You, we already have way too many questions. Like each, like the first one enough is almost like one of the dimensions and how do you measure them is almost like that's gonna take you like a few years or something. Yeah.
Satisfied.
Koen Frolichs: Yeah.
Benjamin James Kuper-Smith: Yeah. Um, but yeah, I, I, I really like this [00:31:00] stuff and it feels like. I guess it helps you situate or, or contextualize what you're doing to some extent because it's not just, here's my question I'm interested in, but you can say like, here's the question I'm interested in, and we can measure it in these different ways and they have these, this or that implication if we do it this way and then it just becomes like, I feel like a bit more fully fledged, if you wanna put it that way.
Um. Versus before maybe being slightly unaware of exactly the context in which you're doing this or the, it's not exactly the word I mean, but yeah, I can't find the precise thing what I want to say. But I think, I mean, I think for me, this is definitely gonna be something that, as I said, uh, in the right, in the beginning of today's episode.
I definitely want to spend some time with this in my, for my thesis, and, um, I think I would definitely include this, you know, not just in passing, but at least as a, like, minor section somewhere.
Koen Frolichs: Yeah.
Benjamin James Kuper-Smith: And, uh,
Koen Frolichs: like, as like, like truly as a separate [00:32:00] section, like the conceptual space of um,
Benjamin James Kuper-Smith: maybe, I guess it depends like how it, yeah.
I mean, I guess so what I mean is like I'm, it's, I think this is definitely going to be more than just blah, blah, blah. Uh, metrics. Bracket, Gaden first 2000, close bracket. Right? Uh, it's, it's gonna be more than that. It's Stephanie gonna be mentioned in like a paragraph or something.
Koen Frolichs: Okay.
Benjamin James Kuper-Smith: Um, but who knows maybe as an entire like chapter or, I mean, I have no idea.
Like what structure. Yeah, yeah, yeah. I mean, it's a stupid
Koen Frolichs: two to
Benjamin James Kuper-Smith: Yeah, yeah, exactly. Also, depending on how you choose, you might have like four chapters or 20, like, you know, I have no idea. Like what, yeah, no, like,
Koen Frolichs: yeah, it's a lot of
Benjamin James Kuper-Smith: stuff, but I definitely think this is gonna be. I'm gonna mention this, and to some extent, this might build the kind of theoretical backbone, maybe to some extent in the same way that MA builds a kind of theoretical backbone.
Koen Frolichs: Yeah. Yeah.
Benjamin James Kuper-Smith: So I think for me, depending on how, I guess the, I think about this in the next few months, I think this might be one of the, like, uh, one of the, yeah, [00:33:00] the, the theoretical background within. I discuss the things I'm interested in. Maybe that's the way of putting it.
Koen Frolichs: Okay. Well, very interesting. I, I just, 'cause, 'cause I just want to ask, like, do you think you'll ever completely reread this book or do you think you'll maybe like pick sections now?
Benjamin James Kuper-Smith: Uh, probably. I mean, I'm probably not gonna read the entire thing through.
Koen Frolichs: Yeah.
Benjamin James Kuper-Smith: Uh, you know, I'm not gonna, there's no point in reading the entire Semantic Strep again. Right.
Koen Frolichs: Yeah, yeah, yeah. I mean, I was, yeah.
Benjamin James Kuper-Smith: Yeah. I, I would definitely read. I, I mean, I don't know. Let's see. I mean, I definitely pro, I'm pretty sure I'll read the first two chapters again.
Yeah. Also, now that you've, we've read everything because they are, yes. Yeah. You know, also kind of fairly general. I could imagine reading chapters three and Yeah. And then I probably like, I think the first four chapters really are the meat and bones of the book, Uhhuh. I think the rest, maybe like chapter seven and eight have some good stuff in it.
And actually I wanted to mention this last time. I forgot to some extent, I feel like it's good that I've read chapter five and six. Yeah. Even if I'm not gonna take [00:34:00] anything directly from this. It, I was, I was reading it. It felt like I. To some extent, this probably is gonna influence my thing or like some ideas that are in here that I haven't thought about quite so much so far and that I'm probably not picking up a hundred percent.
I dunno. It just felt like this might somehow like, I dunno whether, you know what I mean? But like the kind of idea that you're not using it now and you might never use it directly, but somehow it's, it's, it's influencing your thinking a little bit.
Koen Frolichs: It's there sub conceptual spaces you're talking about.
It's the next book. Volume two. No. Yeah, I know
Benjamin James Kuper-Smith: what you mean though. I had that with semantics introduction where I thought like, Hmm, I might like, have similar thoughts. Its in the future and pretend I'm my own, um,
Koen Frolichs: as a, a good scientist.
Benjamin James Kuper-Smith: Yeah. But no, I, I think I would definitely read, um, at least one or two chapters of this again, probably more or less.
Koen Frolichs: Yeah.
Benjamin James Kuper-Smith: Obviously in the skim stuff, but, uh, yeah, I think so.
Koen Frolichs: No. Very good. Cool.
Benjamin James Kuper-Smith: I'm assuming you too, or,
Koen Frolichs: yeah, I mean, I think, I think very, very similar to you actually. Um, I mean, in [00:35:00] the beginning I was also just very excited, right? So I guess there's also like, um, but I do think, yeah, chapters one, two, definitely maybe three, four to kind of, I mean, I think, I think what, what, where's great merit?
I mean, I, and I've said this before now, so, um, it's just like. Seeing how other people do stuff a little different. And I think, you know, I might, I think I sometimes, you know, I get a little bit stuck in, in the, the current paradigm, the coolest way of
doing
Benjamin James Kuper-Smith: things.
Koen Frolichs: Yeah, yeah. Um,
the
Benjamin James Kuper-Smith: faultless paradigm,
Koen Frolichs: oh God, it's such a.
So just stubborn.
Benjamin James Kuper-Smith: It's a very vague
Koen Frolichs: Yeah.
Benjamin James Kuper-Smith: Ill-defined.
Koen Frolichs: Ill-defined. Yeah. Yeah, yeah, yeah. And you just talk a lot. It's, you call it fu obfuscation. Um, you just, just talk a lot to people. Just don't know anymore what you're talking about. No, no. But you're right. I mean, I, and I do enjoy it some ways of thinking about it, and I think that's how I will use it.
I'm not sure, 'cause I am kind of, you know, in the conceptual space already. So I guess, you know, for me to say, like I'll add it is kind of a. I, I hope I [00:36:00] will. I, I need, like, talk about spaces and representations. Um, but yeah, how exactly, like, I, I feel like you're also thinking more about your thesis already than I am.
Like in a different way maybe? Um, um,
Benjamin James Kuper-Smith: I think maybe in, I think I'm thinking about a different way because, I mean, this might be completely wrong, but to me it seems a bit as if from the beginning, your thesis, what a bit clearer or your, your topic of your PhD. Sure. To some extent because it was about learning about personality traits, that kinda stuff, which of course gives huge amounts of flexibility, how you frame it, et cetera.
But I think for me it was fairly much a kind, I think for me, the beginning was. Okay, we're gonna do corporation and social decision making, and we kind of want to go away from like just using the standard paradigms. But that was, it was pretty vague from the beginning. Yeah. I think I, I really had to, we need a bit
Koen Frolichs: more direction.
Benjamin James Kuper-Smith: Um, I mean, not needed, it's just like, um, you know, that's what I wanted to do and it's just, for me, it's been a. [00:37:00] 'cause I didn't have really a, a theme, so to speak.
Koen Frolichs: Yeah.
Benjamin James Kuper-Smith: It's just been something I've been thinking about pretty much since the beginning.
Koen Frolichs: It makes
Benjamin James Kuper-Smith: sense. Yeah. Yeah, that makes sense. It's like, what, what is my thesis?
Like where is this, where, where are, where am I going here? Where
Koen Frolichs: am I? Am I
Benjamin James Kuper-Smith: just running in circles or uh, and yes, I am. Yes. We often
Koen Frolichs: find Ben outside screaming
Benjamin James Kuper-Smith: with his shirt up in a sec. Yeah. Those are the good days though.
Koen Frolichs: Yeah.
Benjamin James Kuper-Smith: Yeah. Yeah, it's weird. To some extent it feels like I know roughly what I want to do for my thesis, but then again, you know, once I start writing, I finally realize it's all bullshit.
Now I have to start again.
Koen Frolichs: Yeah, no, yeah. But that's, that's fine. That's really fine. I mean, it's just, I think it's, you know, it's, it's fun and it's good to read and think about it. Right? That's how you get there, right? Um, you won't, yes.
Benjamin James Kuper-Smith: Yeah. So we'll see how much of this makes it into it. But yeah, I think also as you kind of mentioned, like to some extent we already are in the conceptual space.
Mm-hmm. I guess we just weren't quite aware of it, or at least I wasn't.
Koen Frolichs: Yeah.
Benjamin James Kuper-Smith: [00:38:00] Um, right. Like this to some extent. This is kind of what we're doing already. It's just we never quite made it as explicit and weren't quite as aware of. Right. That's,
Koen Frolichs: yeah. Yeah. No, don't, I, I, I totally agree. I'm, uh, yeah. I, I, I, yeah.
No, I, I think that's, I think I just agree with you. That's why I wasn't saying anything. I could only think about
Benjamin James Kuper-Smith: Okay.
Koen Frolichs: Some stoner going like, yeah, man, there all the time, man. Yeah.
Benjamin James Kuper-Smith: Well, if that's, if that's who you are, then just do that. Say that then.
Koen Frolichs: Yeah.
Benjamin James Kuper-Smith: Okay. Uh, so I guess we're. To some extent coming to the end, but of discussion of the book in general and the, I think so, yeah, the club discussion.
Uh, so I think maybe as a kind of summary, I think this is a really cool book. I mean, I guess we've kind of, it's funny, like, I guess a lot of the stuff we talked about was kind of critical. It's like, yeah, this kind of sucked, or this wasn't clear and it's a bit abstract. I don't really know what he meant.
But I think the, in general, I think we both like the book quite a bit.
Koen Frolichs: Yeah.
Benjamin James Kuper-Smith: Yeah.
Koen Frolichs: Yeah.
Benjamin James Kuper-Smith: Uh, yeah. Obviously [00:39:00] it's, you're not gonna like everything about it. That's fair. Yeah. But yeah, as we said, I think it's, it's a really cool perspective and clarifies or makes you, or makes it, forces you to clarify things yourself almost more.
Koen Frolichs: Mm-hmm.
Benjamin James Kuper-Smith: Yeah. So actually one thing, maybe my last kind of point is that, you know, I mentioned earlier kind of my bold prediction of ai.
Koen Frolichs: Yeah.
Benjamin James Kuper-Smith: And to some extent I think the, the same is happening in cognitive neurosciences, right? Where. You know, as you mentioned, you also mentioned this last time, like a lot of the early stuff was about language and semantics and that kind of stuff.
Yeah, yeah. Mm-hmm. The semantic stuff he cites and he's usually the older stuff from like the sixties and seventies or something. Sure. And you know, then I guess AI and cognitive news science have gone somewhat in parallel. Then you have, you had like, you know, back propagation and that kind of stuff in the eighties and nineties and that kinda stuff.
And um, and it's obviously still going of course, but it seems to me, uh. I guess, uh, the, the question I wrote down is, are we entering the age of conceptual spaces?
Koen Frolichs: Oh boy,
Benjamin James Kuper-Smith: that's a good way of [00:40:00] putting it. Right? Yeah. Um, and I guess to some extent that's the question I asked for ai, but I also wonder whether that's the, the what's happening basically in cognitive neuroscience where we.
We've kind of been doing it for like 10, 15 years, I guess, but maybe a lot of people are not quite as explicitly aware of that that is what's happening.
Koen Frolichs: Yeah, I mean, it's a hard, I I, I, I'm a little bit stunted by that question 'cause I dunno from where to approach it, right? I mean, yes. Like I think for the past 10, 15 years, I think it's been kind of like, you know.
These represent, I think everyone has always been interested in how is stuff represented in the brain. Right. And I think also with the new methodologies and stuff like that, we are able to start to have a look at it, you know, as we just said. But then FMRI is very cos right? EEG. Like you can't really use it for that.
I as much I would say, uh, but I'm also not really probably, yeah. I mean you probably can, you can do some stuff, right? Um, depends. Yeah. But yeah, I mean we might be, I think now. But I think we might even be like, conceptually we might be ahead of what we can [00:41:00] actually measure maybe, right? I guess there's these dynamic, um, what's it called again?
Um, dynamic conceptual, no. Okay. Ah, do you know what I'm talking about at all? Like where you basically like look at groups of neurons coding and then you try to like do a dimensionality reduction on the coding. So you then look at like how neurons kind of behave in a less. Oh God. Yeah. I can't, I can't, like, I know already, like it's not in the tip of my tongue anyway.
Like, anyway, so we are kind of getting there. I mean, I, you can, I think you can keep it in like, um, we, we are kind of getting there with like new ways of looking at like new ways of like using mathematics and stuff like that. We are trying to get there, but I think what we also like would really benefit from is, is you know, more fine, you know, better ways of, of measuring the brain, right.
Non-invasive. Um, and
Benjamin James Kuper-Smith: I think that's. To some extent always been the case. Right? I mean, I think most, a lot of the, I mean, that's pretty much the case, I think with most of science actually, that most of the breakthroughs [00:42:00] happen after someone invents a tool with which you can measure something. You could measure before.
Sure,
Koen Frolichs: sure.
Benjamin James Kuper-Smith: I think the, the best example here whilst we're talking about spaces, is that John O. Keith was just, it was a few years after someone, I think it might have been James rank or I can't remember, but someone developed electrodes where the mouses could move.
Koen Frolichs: Yeah, sure. Yeah, yeah, that makes sense.
Benjamin James Kuper-Smith: And then John O'Keefe just put a,
Koen Frolichs: put 'em in a la on a, in a ma.
Benjamin James Kuper-Smith: I'm sure this is all that happened. You just put a mouse in a box or a ratchet box and said, come on, move
Koen Frolichs: to your tank.
Benjamin James Kuper-Smith: No, we'll prize. That's how it works. Yeah. But that, I mean, that happens with most of the things, uh, I think Right. A lot of, to some extent it's, it's, it's kind of cool and also kind of frustrating to me.
It seems that, it seems to me like most scientific advance hap or a lot of scientific advance happens just. 'cause people figure out how to measure something,
Koen Frolichs: which I guess is fair. But then again, like, I mean it's also like this question already. I mean, I'm sorry, I'm kind of, um, um, drifting away from your question, like what should we measure now, right?
I mean Yeah, sure. Like, like single cell recordings, like, [00:43:00] but not invasively like, or like it seems also abstract. Can I think that's
Benjamin James Kuper-Smith: a good thing.
Koen Frolichs: Yeah.
Benjamin James Kuper-Smith: What we want to do doesn't, it doesn't necessarily make sense.
Koen Frolichs: Yeah. I mean like, unless you can do like single cell recordings for every neurons in like a.
Like a patch, like a, a piece of cortex? No,
Benjamin James Kuper-Smith: I mean more in a con, I mean like more in a conceptual way that, not a conceptual basis way, but con concept of the like generic term. Yeah. Know, the kind of stuff we're interested in isn't necessarily the stuff that you will find represented in individual neurons.
Koen Frolichs: Mm-hmm.
Benjamin James Kuper-Smith: I wonder, you know, because the, okay, sure. You have, that's how good sales play, sales, blah, blah, blah. And even the concept sales that, you know, Jennifer Aniston sell on whatever you wanna call it.
Koen Frolichs: Mm-hmm.
Benjamin James Kuper-Smith: You know, you can do it. But to some extent there's also the question of whether some of the stuff.
It's represented more in terms of large numbers of neurons interacting with each other. Yeah, yeah, definitely. Definitely a single neuron coding for something.
Koen Frolichs: Yeah, yeah, yeah, yeah, yeah.
Benjamin James Kuper-Smith: So for some of the stuff we're interested in, single neurons don't even make sense to measure.
Koen Frolichs: No, no. I just meant [00:44:00] it as in like FMRI, but then you can measure single, all the single neurons.
I guess I would give you information overload, but
Benjamin James Kuper-Smith: Okay. Yeah. Yeah. Yeah.
Koen Frolichs: I mean, yeah, I went all the way.
Benjamin James Kuper-Smith: Yeah. Kun, I have a 1 billion, a hundred billion to
Koen Frolichs: get petabytes.
Benjamin James Kuper-Smith: Like my vector of, uh, neural activity is a hundred billion long. What do I do with this?
Koen Frolichs: Yeah, you transpose it. And
Benjamin James Kuper-Smith: Yeah. Then you wait a week.
Koen Frolichs: Yeah, yeah, exactly. Yeah. And then Matlab like sends you an angry email. No. Yeah, no, you're right. I mean, I, things has have to go hand in hand. Um, to come back to your question, yes, I do think probably we're in the conceptual spaces, era. A little bit like we're trying to figure it out. But yeah, I guess there's a lot of stuff going on in, in general.
I think there's a lot of people doing a lot of interesting stuff.
Benjamin James Kuper-Smith: Yeah. Yeah, of course. I just liked, I liked that he had this phrase, I don't know what exactly how he put it, but I just wrote down the phrase, um, hunt for hidden conceptual spaces. That's kind of what researchers should do, you [00:45:00] know, should hunt for your hidden conceptual spaces and discover new ones.
Uh, and show how to relate to other things. I like that idea, and I guess we'll see whether that's what we'll do or not.
Koen Frolichs: Who knows?
Benjamin James Kuper-Smith: Maybe we start doing it, realize it's all nonsense. I have no idea right now we haven't actually used any of the stuff we've read about.
Koen Frolichs: Yeah, exactly. Yeah. No, I mean, yeah. Let's see, let's see.
Maybe in, in 10 years, we'll we'll know.
Benjamin James Kuper-Smith: Okay. Let's do, okay. Part five. In 10 years we'll find answer.
Koen Frolichs: We should do part five Is the end of your teases. Has it, you know, has it been. Integrate it into, into your thesis and then 10 years, it's just gonna be 10
Benjamin James Kuper-Smith: seconds where it say, so part five and has it in and, uh, no, no,
Koen Frolichs: no.
Yeah, it hasn't. No thank you. I'm kind of like, I'm crying a lot. Um, I'm very frustrated.
Benjamin James Kuper-Smith: This book was a waste of time.
Koen Frolichs: Yeah.
Benjamin James Kuper-Smith: Yeah. It wasted three months trying to make, put this into my thesis. It didn't work. Now I'm homeless. Anyway, um,
Koen Frolichs: that's a great ending.
Benjamin James Kuper-Smith: Yeah, I mean, uh, maybe, uh, apart from that, [00:46:00] maybe we can, one brief thing, you know, that I always do, uh, is very briefly discussing what it was like to discuss this book.
Koen Frolichs: Oh
Benjamin James Kuper-Smith: yeah. On the podcast. I mean, so you've done a previous one now? Yes, yes. It was slightly different, uh, which was, uh, I mean for those who haven't listened to it, it was a jacker book, which is a kind of 500 page thriller, I guess that is very plot driven and very easy to read. Yeah. Um, and. This is a, you know, last time I think we did a hundred pages a week and then said probably should on 150 to 200 mm-hmm.
Because it's so easy to read this time we did, you know, 60 pages a week pretty much on average. Yeah. Pretty much. Exactly. 60 pages per week on average. And, yeah. So what, I dunno, maybe I'll just ask you what, what do you, was it easy to discuss this on the podcast? Or how did you
Koen Frolichs: Um,
Benjamin James Kuper-Smith: yeah, other, 'cause it's very different than the other books I've done so far.
Koen Frolichs: Yeah. Yeah. I, I, I did. Struggle a little bit like, you know, especially like how, like what will you [00:47:00] discuss where you discuss the content directly, right? Like. Specific, you know, things he explains or he tries to get you to understand or do you talk more, you know, a little bit more global, which I think, I feel like that's what we did more now.
Benjamin James Kuper-Smith: Yeah,
Koen Frolichs: right. A little bit more. Um, which also makes sense because like if you go into detail it just, you know, you get lost probably, and it takes way too long. Yeah. But I mean, overall that fun. Um, I think, you know, a book like this, which is just a little bit difficult sometimes. You know, it's also just good to have, you know, the reason to, you know, you have to finish it, so might as well read it tonight.
Kind of, right?
Benjamin James Kuper-Smith: Yeah, exactly. Um,
Koen Frolichs: but, you know, I, I
Benjamin James Kuper-Smith: mean like, what I just found difficult, I guess, is that in the three books I've done so far, I mean, two of them were novels and the third was a biography. All of those are very plot driven. Stuff happens, you know, their actions. Yeah. For example, even in the last one with the BOL biography, you know, we could.
Laugh at homeboy. Basically being bitten by mystique, by mosquitoes all day when he was in the Amazon rainforest and just having an absolutely [00:48:00] miserable life or of, you know, him like wanting to go on a great adventure again. And basically he had to wait like 20 years or 30 years because countries were at war with each other all the time.
You just can like make it happen. Yeah, because if you wanna like make a far journey. And you German makes sense. And the country you want to go in is that war with Germany? Then maybe don't go there.
Koen Frolichs: Yeah, yeah.
Benjamin James Kuper-Smith: If it's like 1,820 or whatever. So you know, you have like these things that are so easy to understand, so obvious, whereas, you know, this is something that's very much, he's trying to make a complicated point and Yeah.
Um, yeah, he's trying to make it easy, I would assume, but it's still complicated. So, yeah, I don't know. I mean, I guess we dunno whether what this, what it's like to listen to this, but hopefully it's, it helps to some extent,
Koen Frolichs: which is probably horrible.
Benjamin James Kuper-Smith: Yeah. It's just misery.
Koen Frolichs: This, this will be like in Guantanamo Bay when, you know, sort of like Metallica, they'll like have us jab on like, or.
Um, yeah. Laban for an hour.
Benjamin James Kuper-Smith: Yeah. There's just three people, or like, and there's probably like two or three people [00:49:00] who just, I don't know, are in ton syndrome and someone accidentally left left, like the, the, the, the, the podcast like from a previous episode on it just runs through. Yeah, yeah, yeah. So someone's just lying in hospital bed, go, oh God, please make it stop.
Koen Frolichs: No, I can. Yeah. Yeah. It's. Especially if you're like, um, oh no,
Benjamin James Kuper-Smith: there's part four.
Koen Frolichs: There's part four. And like, especially if you're like a semantic philosopher, right? And we just like shadow on and you're just like, no, you just misunderstanding everything. You bunch your idiots. Yeah,
Benjamin James Kuper-Smith: exactly.
Koen Frolichs: When I get outta
Benjamin James Kuper-Smith: this lockdown syndrome, I'm gonna fuck you up.
Yeah,
Koen Frolichs: yeah. I'll, I'll find you and I'll hurt you. Yeah. Um, if, if so I'm sorry, um, for your predicament. Um, yes.
Benjamin James Kuper-Smith: I
Koen Frolichs: wish you all the best.
Benjamin James Kuper-Smith: And for those who, dunno what it's like. There's a great book about lockdown syndrome called The Butterfly and the Bell by, oh yes, what's it called? John Dominic Bobbi.
Something like that. It's a big French,
Koen Frolichs: isn't it?
Benjamin James Kuper-Smith: Yeah. Yeah. French. Also a very good film by Julian Sch. Uh, anyway, [00:50:00] I think if we're starting to ramble that farther, maybe we should just stop. We should end
Koen Frolichs: it.
Benjamin James Kuper-Smith: So yeah, thank you for listening and, uh. Yeah, I guess go read the book. It's interesting. And
Koen Frolichs: yeah, it's worth 33 bucks, however much we
Benjamin James Kuper-Smith: paid.
Yeah, I think so too. Well, for you was 30, because you only gave me 30 bucks, but yeah.
Koen Frolichs: Surprise of my, um.
Benjamin James Kuper-Smith: Right. That, that's no loan for. Exactly. That's, that's your, the hourly wage is not good.
Koen Frolichs: Yeah, my
Benjamin James Kuper-Smith: hourly, three euro, I'm a PhD. Half hours of recording. Yeah. It's still better than your PhD salary.
Koen Frolichs: Yeah. Okay, Ben.
Benjamin James Kuper-Smith: Anyway,
Koen Frolichs: it was good talking to you.