WEBVTT 00:00:03.617 --> 00:00:06.517 This is the Convergent Science Network podcast. 00:00:08.537 --> 00:00:13.057 Leading researchers in the domain of neuroscience, brain theory and technology 00:00:13.057 --> 00:00:16.797 are interviewed by Paul Verschoor and Tony Prescott. 00:00:19.797 --> 00:00:23.277 This is Paul Verschoor with the Convergent Science Network podcast. 00:00:24.397 --> 00:00:29.077 And today I'm talking with Peter Gardefors, who's one of our speakers in our summer school. 00:00:30.017 --> 00:00:35.357 And Peter, you presented your work in generalizing a theoretical framework that's 00:00:35.357 --> 00:00:37.697 called conceptual spaces towards language. 00:00:39.557 --> 00:00:42.617 So what's this notion of a conceptual space exactly? 00:00:43.757 --> 00:00:47.497 It's a model of how you represent knowledge. I mean, I'm a cognitive scientist 00:00:47.497 --> 00:00:52.457 and I want to understand in some sense how our minds represent knowledge. 00:00:52.577 --> 00:00:53.917 And we have different traditions. 00:00:54.097 --> 00:00:58.557 I mean, you have the symbolic approach of the early AI and then you have the neural networks. 00:00:59.077 --> 00:01:04.717 My conceptual space is something in between, because I work with geometric structures, 00:01:04.837 --> 00:01:13.337 with metrics, distances, vectors, and that kind of notions to model different kinds of knowledge. 00:01:13.857 --> 00:01:21.077 So, can you give me an example of that? Well, the normal example I use is the color space. 00:01:21.717 --> 00:01:26.117 We perceive colors, but we organize colors along three dimensions. 00:01:26.117 --> 00:01:31.317 I mean, there is the hue, the color circle going red, blue, green, yellow, and so on. 00:01:31.417 --> 00:01:34.817 And then there is the dark and light. 00:01:35.117 --> 00:01:38.817 And then there is the intensity going from gray. So there's the three-dimensional space. 00:01:39.177 --> 00:01:44.557 And that's how we perceive colors. And it's fairly well established in psychophysics 00:01:44.557 --> 00:01:48.277 that we have this spatial arrangement of colors. 00:01:49.337 --> 00:01:54.917 So, but now in some sense, I could say, well, if I talk about an object having 00:01:54.917 --> 00:01:56.817 a certain color, like an apple being red. 00:01:57.644 --> 00:02:01.304 Then this apple might have a number of properties. So it will be round, 00:02:01.484 --> 00:02:03.744 it will have a certain taste, a smell, a color, and so on. 00:02:03.984 --> 00:02:06.184 So how do I bring that together in a conceptual space? 00:02:06.564 --> 00:02:10.864 Well, one of my key notions is that of a domain. I mean, we have several domains 00:02:10.864 --> 00:02:12.184 that we organize the knowledge. 00:02:12.304 --> 00:02:16.364 Color is one, size is another, shape is a third, temperature, and so on. 00:02:16.484 --> 00:02:19.464 We have a lot of, some of them are based on perception, some of them are based 00:02:19.464 --> 00:02:23.124 on action, some are based on our social grounding in the society. 00:02:23.124 --> 00:02:29.344 So, there is this general notion of domain that is used to sort up the information. 00:02:29.904 --> 00:02:34.024 So, you would not so much see it as, let's say, dimensions spanning this space, 00:02:34.204 --> 00:02:36.504 but a number of domains, or is it equivalent? 00:02:36.804 --> 00:02:41.084 Each domain could consist of a number of dimensions, and color domain is three-dimensional, 00:02:41.164 --> 00:02:44.884 and taste domain is four- or five-dimensional. 00:02:45.444 --> 00:02:49.804 So, are these domains like a Kantian prior? You're just born with it? 00:02:51.784 --> 00:02:55.224 Partly I mean it's my theory is neo-Kantian if 00:02:55.224 --> 00:02:58.044 you really want to bring in the philosophy Kant said 00:02:58.044 --> 00:03:01.124 we are born with space and time I say that we are born with 00:03:01.124 --> 00:03:06.184 we are disposed to have representations of space we learn time much later but 00:03:06.184 --> 00:03:10.124 then of course there are many domains that are dependent on us being in a culture 00:03:10.124 --> 00:03:15.264 and learning to discriminate new things and so on so our set of domains are 00:03:15.264 --> 00:03:19.604 expanding as we grow up in a society and they can 00:03:19.684 --> 00:03:22.484 change over time so it's a mixture if you want it's a 00:03:22.484 --> 00:03:25.324 mixture so there's a so do you is there really a 00:03:25.324 --> 00:03:31.364 a defined set of prior domains no no no no i mean we are born with certain sensory 00:03:31.364 --> 00:03:36.364 organs that that determine some of the basic domains like color perception like 00:03:36.364 --> 00:03:42.084 perception of heat and the tastes and and and so on but on top of this we build 00:03:42.084 --> 00:03:43.384 a number of domains i mean I mean, 00:03:43.424 --> 00:03:46.284 we extract other types of information. 00:03:46.824 --> 00:03:52.584 For instance, I mean, I have been emphasizing our perception of forces. 00:03:52.724 --> 00:03:56.224 We are quite good at seeing forces in people's emotions and so on. 00:03:56.284 --> 00:03:57.724 Right. Okay. So now we have our domains. 00:03:58.284 --> 00:04:02.044 And then you say, look, if I now see an apple, then I have a number of domains. 00:04:02.264 --> 00:04:08.444 I might have a visual domain and a haptic domain and olfactory domain and a taste domain. 00:04:08.864 --> 00:04:13.064 And now across these domains, apples are a set of data points. 00:04:13.364 --> 00:04:14.824 In that multidimensional space. 00:04:15.224 --> 00:04:18.524 So I have a cluster of data points and I say all these data points together, 00:04:18.764 --> 00:04:21.464 this cloud of data points, this is now Apple. Yep. 00:04:22.124 --> 00:04:25.444 It's a point in a very high dimensional space, but the space is organized along 00:04:25.444 --> 00:04:26.704 a number of domains. Right, exactly. 00:04:27.084 --> 00:04:31.664 But now, would there be a way to, let's say, reorganize that space? 00:04:31.844 --> 00:04:35.724 Like, for instance, you might at time, you might sometimes discover that, 00:04:35.764 --> 00:04:37.304 let's say, Santa Claus doesn't exist. 00:04:37.724 --> 00:04:42.684 So suddenly one organizational element of my reality is sort of, 00:04:42.684 --> 00:04:43.804 has disappeared, and now I 00:04:43.804 --> 00:04:47.304 have to reorganize all my data. So how does it work in a conceptual space? 00:04:47.464 --> 00:04:51.084 Well, first of all, you can add new domains. You can learn about apples having 00:04:51.084 --> 00:04:53.064 nutritional value or something like that. 00:04:53.891 --> 00:04:57.691 What happens quite often in learning new things is that you change the importance 00:04:57.691 --> 00:05:01.591 of domains, which are the most important in making the classifications. 00:05:01.911 --> 00:05:08.651 So in biology, maybe the behavior of an animal was more important when you classified 00:05:08.651 --> 00:05:12.571 something as a fish, but then came the biologists and said, no, 00:05:12.671 --> 00:05:16.051 it's the skeletal structure or how you feed your kids or whatever. 00:05:16.251 --> 00:05:21.951 Other variables that are more important in zoological classification. 00:05:21.951 --> 00:05:24.791 Classification yeah but then i was more worried about 00:05:24.791 --> 00:05:27.591 the case if i lose a domain not about adding domains but if 00:05:27.591 --> 00:05:30.831 i lose one that's why i thought about santa claus yeah 00:05:30.831 --> 00:05:34.411 but santa claus is not a domain i mean it's a fictional fictional object 00:05:34.411 --> 00:05:40.411 yeah i don't know any good examples of losing domain so yeah well imagine that 00:05:40.411 --> 00:05:44.971 i believe that uh let's say objects move physical objects move in the world 00:05:44.971 --> 00:05:49.391 because they have intentional states yeah and then i discovered that actually 00:05:49.391 --> 00:05:51.151 stones don't have intentional states. 00:05:51.271 --> 00:05:56.331 Now suddenly, the way I frame the world is changing. Okay, I would describe 00:05:56.331 --> 00:06:02.351 that as the domain that maybe does not disappear, but you assign it a very low value. 00:06:02.471 --> 00:06:06.751 It becomes a zero in your concept classification. 00:06:07.331 --> 00:06:13.631 To take an example from chemistry, you had some idea of the caloric dimension 00:06:13.631 --> 00:06:16.411 that was used to classify chemical stuff. 00:06:16.731 --> 00:06:18.871 And then suddenly, there was 00:06:18.871 --> 00:06:22.191 a revolution in chemistry and the caloric dimension totally disappeared. 00:06:22.451 --> 00:06:24.531 I mean, you can find examples of that in the history of science. 00:06:24.671 --> 00:06:27.531 In the history of human perception or in human classification, 00:06:27.731 --> 00:06:30.911 it's maybe more difficult to find such changes. 00:06:31.151 --> 00:06:35.091 So now I have this cloud of points in apple space. 00:06:38.443 --> 00:06:43.383 If you understand your proposal, and if you want the center of gravity of that 00:06:43.383 --> 00:06:46.283 cloud of data points is now my prototype of an apple, right? 00:06:46.463 --> 00:06:51.063 Yes, and that cloud is, I make the assumption that it's a convex region. 00:06:51.363 --> 00:06:57.103 I mean, a concept, Apple, corresponds to a convex region, meaning that for any 00:06:57.103 --> 00:07:00.043 two points in the region, all points in between are also there. 00:07:00.183 --> 00:07:06.043 And that is very useful in helping us understand how we can learn concepts to 00:07:06.043 --> 00:07:08.043 have this notion of convexity. 00:07:08.043 --> 00:07:14.223 But it does mean that you clearly have a very empiricist view on prototypes 00:07:14.223 --> 00:07:17.523 and concepts that are really derived from, if you want, data points obtained 00:07:17.523 --> 00:07:19.883 from the world. Yes and no. It's empiricist. 00:07:20.043 --> 00:07:26.703 And once you have the domains, it's data-driven how you divide the spaces into regions. 00:07:27.023 --> 00:07:32.983 But the domains are, well, the neocantrian position that they are partly given, partly learned. 00:07:33.243 --> 00:07:37.143 Okay. But now, how should I think about the development of these? 00:07:37.143 --> 00:07:38.743 Imagine I want to add a new domain. 00:07:38.903 --> 00:07:43.903 So the discovery of a new domain could be the result of just adding data points 00:07:43.903 --> 00:07:50.083 to my conceptual space and discover that I cannot organize these data points anymore. 00:07:50.183 --> 00:07:53.583 And I have to say, aha, I should now assume there's another domain. 00:07:54.043 --> 00:07:58.283 So how does that work? How do you go from single observations and their accumulation 00:07:58.283 --> 00:08:01.023 to this decision? Okay, new domain. 00:08:01.943 --> 00:08:06.503 That's normally tougher in individual. I mean, here in our society, 00:08:06.623 --> 00:08:10.503 we have scientists who say that we need this dimension, this domain, 00:08:10.583 --> 00:08:14.843 in order to understand the difference between this and that. I mean, so I... 00:08:17.847 --> 00:08:21.187 You can do it as an individual, but it's very tough. 00:08:21.367 --> 00:08:24.867 I mean, you're living in a culture where there are new ideas about what it means 00:08:24.867 --> 00:08:26.807 you need to understand something. 00:08:27.147 --> 00:08:31.867 Right. But do you see this in the same in relation to, let's say, 00:08:31.907 --> 00:08:36.327 the standard notions of assimilation and accommodation that Jean Piaget would 00:08:36.327 --> 00:08:37.087 talk about in development? 00:08:37.927 --> 00:08:43.027 You can interpret it in that way. I mean, he didn't never say anything about 00:08:43.027 --> 00:08:48.247 how these processes work. He didn't have any model of assimilation and accommodation. 00:08:48.887 --> 00:08:52.027 Assimilation would have been just collecting data points, and accommodation 00:08:52.027 --> 00:08:55.107 would be changing the space, I mean, the structure of the underlying space. 00:08:55.287 --> 00:09:00.767 So I can model these processes that way here. Okay, but now at the core of your, 00:09:00.867 --> 00:09:04.847 and this is also what you said earlier, at the core of your approach is really 00:09:04.847 --> 00:09:06.667 a notion of a metric, a topology, 00:09:06.907 --> 00:09:10.287 and also a quantification of similarity, right? Right. 00:09:10.387 --> 00:09:17.347 So can that really generalize to, let's say, all different levels of, let's say, concepts? 00:09:18.127 --> 00:09:25.127 No, that's a good question, because our concept classification is very much 00:09:25.127 --> 00:09:28.127 dependent on our perceptual mechanism. That's the basic classification. 00:09:28.407 --> 00:09:32.727 But then we can introduce more abstract concepts, different means, 00:09:32.927 --> 00:09:40.127 and in particular via language. I mean, you have mathematics and in law you 00:09:40.127 --> 00:09:42.187 introduce new concepts by definitions. 00:09:42.367 --> 00:09:47.547 And we use language to construct new concepts. And there you may get out of 00:09:47.547 --> 00:09:48.387 the geometric constructions. 00:09:48.967 --> 00:09:52.947 I mean, I don't know if I can apply my methods there. But they are dependent 00:09:52.947 --> 00:09:57.127 on being grounded in these more low-level concepts. 00:09:57.427 --> 00:10:01.647 You can't start from the bottom by defining concepts. You have to ground them 00:10:01.647 --> 00:10:04.527 in some kind of perceptual domain. 00:10:04.867 --> 00:10:09.707 Okay, so you're saying the conceptual space is exposed if you want to both feed 00:10:09.707 --> 00:10:15.067 forward or bottom up influences perception, sensation, and top-down influences 00:10:15.067 --> 00:10:18.107 like high-level symbolic systems related to language. 00:10:18.407 --> 00:10:22.227 And in some sense, what you're saying is that what is most clear for you at 00:10:22.227 --> 00:10:25.387 this point in time is how this feed forward component actually operates. 00:10:25.587 --> 00:10:29.467 And then the top-down element is sort of for the future. Yeah. 00:10:29.587 --> 00:10:36.407 Well, I have one area that I think is very useful to create new concepts is by using metaphors. 00:10:36.747 --> 00:10:40.587 So if I mean, to take an example from science, when you introduce the notion 00:10:40.587 --> 00:10:45.387 of electricity, you have these notions of current and voltage and so on new 00:10:45.387 --> 00:10:46.727 dimensions that you don't perceive. 00:10:47.007 --> 00:10:50.587 And in order to understand how they work, you can then compare electricity with 00:10:50.587 --> 00:10:52.127 water running or whatever. 00:10:52.247 --> 00:10:56.267 And you get some kind of grounding in the perceptual domain in that way. 00:10:56.267 --> 00:11:00.847 But the electrical dimensions, they live on their own in one sense. 00:11:00.887 --> 00:11:02.167 You can formulate them mathematically. 00:11:02.747 --> 00:11:05.107 If you go to domains like, let's say, machine vision. 00:11:05.711 --> 00:11:10.531 There are lots of methods around for data association. So these are basic different 00:11:10.531 --> 00:11:14.851 kinds of methods, adaptive filtering methods and what have you, 00:11:14.951 --> 00:11:20.171 that try to explain how you can bring data points together in high dimensional space. 00:11:20.591 --> 00:11:26.091 Is that a rather direct expression of your conceptual space notion or is there a difference? 00:11:26.391 --> 00:11:31.491 Well, you must distinguish between technical solutions of dimension reductions. 00:11:31.491 --> 00:11:35.831 I mean, you can have lots of data sets and you can use multidimensional scaling 00:11:35.831 --> 00:11:40.571 or principal component analysis and so on to reduce and pick out the most important dimensions. 00:11:40.871 --> 00:11:44.071 But that's a technical solution. Then you have the more biological solution 00:11:44.071 --> 00:11:49.831 where you really try to connect the dimensions to some kind of perceptual mechanism 00:11:49.831 --> 00:11:52.831 or to your motor systems or whatever. 00:11:53.111 --> 00:11:55.331 I mean, and that's two different methods. 00:11:55.811 --> 00:12:00.591 Okay. All right. But they're not necessarily... You could imagine that any of 00:12:00.591 --> 00:12:05.811 these... Some of these methods would be rather directly implementing these notions of a conceptual space. 00:12:05.991 --> 00:12:08.971 Yeah, yeah. Okay, so there's no fundamental problem there. 00:12:09.031 --> 00:12:13.611 No, no, it's just a question of how biologically realistic you want to be in 00:12:13.611 --> 00:12:15.391 your implementations. Right, exactly. 00:12:15.711 --> 00:12:21.031 So now we have this idea of a conceptual space. 00:12:21.271 --> 00:12:23.691 And you've been working on that for quite a while. Yeah, yeah. 00:12:23.851 --> 00:12:28.431 And today in your presentation, you were emphasizing two, let's say, 00:12:28.491 --> 00:12:30.131 extensions of this framework. 00:12:30.131 --> 00:12:33.531 So on the one hand, you wanted to capture a notion of action, 00:12:33.691 --> 00:12:38.191 because so far you had looked more or less at static concepts, 00:12:38.211 --> 00:12:41.431 like apples, and on the other hand you want to go to language. 00:12:41.631 --> 00:12:49.131 So let's start with action. Well, I want to go to language as kind of application of these two. 00:12:49.371 --> 00:12:53.131 I mean, I want to use conceptual spaces to model the cognitive representation 00:12:53.131 --> 00:12:57.871 of actions, and I want to use them to model the cognitive representation of events. 00:12:58.451 --> 00:13:03.391 How do we think of events? What is the structure? Just like we have discovered 00:13:03.391 --> 00:13:04.851 the three-dimensional color space. 00:13:05.071 --> 00:13:10.371 I want to understand what is the action space in our minds. Right. Okay. 00:13:10.571 --> 00:13:15.471 So in a sort of boneheaded way, which I usually tend to follow, 00:13:15.591 --> 00:13:17.971 I could say, well, no big deal. 00:13:18.111 --> 00:13:22.051 I mean, also action I might be able to classify in some high-dimensional space. 00:13:22.807 --> 00:13:25.967 Let's say different limbs involved or whether it is 00:13:25.967 --> 00:13:28.907 just between agents or 00:13:28.907 --> 00:13:31.947 it's a single agent or an action on an object or whatever right 00:13:31.947 --> 00:13:35.387 so so why is that straightforward generalization 00:13:35.387 --> 00:13:38.327 not enough to classify action what's missing well 00:13:38.327 --> 00:13:42.167 what kind of data do you start from i mean you you can get a lot of data points 00:13:42.167 --> 00:13:46.687 about the human human body moving but that's a very high dimensional set you 00:13:46.687 --> 00:13:51.207 have to reduce the data somehow and you can do use as i said just technical 00:13:51.207 --> 00:13:57.267 ways of doing it But you could also use data from psychology on how we perceive actions. 00:13:57.587 --> 00:14:03.327 And then there are experiments showing that we have a kind of hierarchical representation of our bodies. 00:14:03.487 --> 00:14:07.867 I mean, there is the main body, then we have arms and legs, and we have forearms 00:14:07.867 --> 00:14:11.587 and lower arms and upper arms, and we have hands and fingers. 00:14:11.687 --> 00:14:13.007 I mean, you have a hierarchical representation. 00:14:13.467 --> 00:14:22.067 And an action is a movement in this hierarchical structure. So it's a fairly 00:14:22.067 --> 00:14:25.787 high-dimensional vector, dynamic vector that represents. 00:14:26.067 --> 00:14:31.447 And you need a more condensed way of understanding what an action is. 00:14:31.707 --> 00:14:35.067 I think that our brains need that kind of reduction. 00:14:35.547 --> 00:14:37.427 So it's interesting, right? Because that would mean that you're saying, 00:14:37.527 --> 00:14:41.727 well, the notion of conceptual space is like a core representational engine, 00:14:41.887 --> 00:14:43.707 which is modality independent. 00:14:44.387 --> 00:14:48.707 And then dependent on the perceptual filtering, I can now exploit this mechanism 00:14:48.707 --> 00:14:51.887 in different ways. So, this would be for action that you're saying, 00:14:52.027 --> 00:14:56.307 well, to map action into a conceptual space, it's basically how I filter action. 00:14:56.487 --> 00:15:01.287 So, I should maybe not look at action in static terms, but I should look at 00:15:01.287 --> 00:15:04.547 more at its dynamical property. So, how am I going to do that? 00:15:04.787 --> 00:15:10.087 I mean, if you see a person walking, take walking as a typical example here, 00:15:10.167 --> 00:15:13.487 you don't care about the clouds of the person, you don't care about whether 00:15:13.487 --> 00:15:17.407 it's hot or cold, you care about the movements of the person. 00:15:17.407 --> 00:15:20.267 So you abstract away a lot of the domains. 00:15:20.507 --> 00:15:25.207 And what remains in my mind is that you look at what are the forces that the 00:15:25.207 --> 00:15:27.867 person exerts on his or her body parts. 00:15:28.007 --> 00:15:32.087 So I see an action as a pattern of forces. 00:15:32.507 --> 00:15:40.307 That's my reduction of the domain for actions, to focus on the forces. 00:15:40.307 --> 00:15:43.867 Okay, so now, so also you showed in your talk that, for instance, 00:15:43.947 --> 00:15:48.707 humans actually need very little, few cues to extract motion, 00:15:48.967 --> 00:15:50.267 right? Like biological movement. 00:15:51.434 --> 00:15:54.854 And it's also been shown by others like Martin Gies and other people. 00:15:56.054 --> 00:15:59.494 So how does that help me to get to a notion of force? 00:16:00.194 --> 00:16:03.554 Well, I mean, first of all, these examples, I mean, it was Gunnar Johansson 00:16:03.554 --> 00:16:05.394 who started with this patch-like technique. 00:16:05.474 --> 00:16:08.974 You only put small lights on the joints of your body and that's enough, 00:16:09.074 --> 00:16:12.094 the information you get to identify actions. 00:16:12.174 --> 00:16:14.034 And you identify them extremely quickly. 00:16:14.174 --> 00:16:17.274 I mean, it takes 200 milliseconds to see what kind of action it is. 00:16:17.274 --> 00:16:21.334 It helps because it shows you that all the other features are not necessary 00:16:21.334 --> 00:16:24.114 you don't have to see the surface of the person moving, 00:16:24.374 --> 00:16:28.294 you don't have to see the color all the other features are gone so you see the 00:16:28.294 --> 00:16:32.334 kinematics and then of course you could say okay this kinematics that is, 00:16:33.714 --> 00:16:38.694 what you perceive in action but I go one step further and say that no it's the 00:16:38.694 --> 00:16:42.874 changes of the kinematics I mean if you talk mathematics there's a second derivative 00:16:42.874 --> 00:16:46.354 that bring out the forces forces. 00:16:46.714 --> 00:16:52.834 I would say that when we identify an action, it's more using the forces give 00:16:52.834 --> 00:16:58.554 us a more coherent and simpler representation of actions than looking at kinematics. 00:16:59.034 --> 00:17:04.674 Okay. This is interesting. I want to say, look, we have to have a direct perception 00:17:04.674 --> 00:17:08.234 of change in the kinematics, that's one thing. That's the second derivative. 00:17:09.314 --> 00:17:12.394 The direct perception of the change you call a force. 00:17:13.274 --> 00:17:16.034 But if I would look at that from, let's say, an engineering perspective, 00:17:16.294 --> 00:17:22.054 I could say, hey, wait one moment, but that would not map necessarily onto the 00:17:22.054 --> 00:17:26.974 forces I'm really exerting on the degrees of freedom of that walking human, right? 00:17:27.054 --> 00:17:29.954 So, how do I relate these two now? 00:17:30.154 --> 00:17:36.794 Okay, that's a good question because you have the perceptual inputs when seeing 00:17:36.794 --> 00:17:40.934 somebody walking and then you can exert the forces on the body limbs. 00:17:40.934 --> 00:17:46.674 On the other hand, you have the kinesthetic experiences of your muscle control, 00:17:46.874 --> 00:17:51.554 and you know that you have to stretch your arm in a certain way to push a door or whatever. 00:17:51.974 --> 00:17:58.474 And I'm making a tacit assumption here that these two systems map onto one another. 00:17:58.694 --> 00:18:04.054 This is basically like when I understand when you're talking, 00:18:04.254 --> 00:18:08.114 I'm somehow representing how you produce the sounds, but I also have to map 00:18:08.114 --> 00:18:10.134 them on how I produce the sound. 00:18:10.134 --> 00:18:14.994 It's the same kind of perception and motor action mapping. Right. Okay. 00:18:15.594 --> 00:18:20.474 But that would mean I would have great difficulties to understand the walking pattern of a bird. 00:18:21.466 --> 00:18:24.166 Because the bird has a rather different kind of body than I do, 00:18:24.266 --> 00:18:25.666 at least according to me. 00:18:25.946 --> 00:18:31.786 So now the interpretation, the grounding of the forces that you call them, 00:18:31.806 --> 00:18:32.946 I observe, will be difficult. 00:18:33.326 --> 00:18:35.666 So how do I circumvent that problem? 00:18:36.286 --> 00:18:41.746 Well, we understand the movements of other animals. We do understand ourselves. 00:18:41.926 --> 00:18:45.966 So there is this generalization problem. On the other hand, once your brain 00:18:45.966 --> 00:18:50.926 has learned to extract the second derivatives of movement, you can then see 00:18:50.926 --> 00:18:52.166 the patterns in movement. 00:18:52.366 --> 00:18:56.886 You can see that the patterns of sparrows flying are similar. 00:18:57.126 --> 00:19:03.486 They have much quicker wing flapping than if you look at an albatross flying. 00:19:03.626 --> 00:19:08.646 It's very slow and much more forceful wing flapping than in a sparrow. 00:19:08.806 --> 00:19:13.106 There are differences in the patterns, even in birds, and we can learn to identify them. 00:19:13.606 --> 00:19:16.906 But now, wouldn't there be an other way to interpret this where we say, 00:19:17.006 --> 00:19:20.326 well, what we really learned to classify are these derivatives. 00:19:21.526 --> 00:19:27.526 And we can call them, let's say, kinematic dynamics, something like this, or kinematic change. 00:19:28.786 --> 00:19:30.766 And these changes will, in the 00:19:30.766 --> 00:19:35.246 end, be brought about causally by forces operating on degrees of freedom. 00:19:35.326 --> 00:19:38.166 But from a perceptual perspective, I don't care about these forces. 00:19:38.326 --> 00:19:42.446 I care about kinematic change. Okay. So what would be wrong with just saying, 00:19:42.486 --> 00:19:44.486 look, why don't we call this kinematic change for now? 00:19:46.846 --> 00:19:49.866 It's a bit of a tough question, because now we've been talking about biological 00:19:49.866 --> 00:19:53.026 movements as the only type of actions. 00:19:55.216 --> 00:20:00.676 And we humans are not very good. Well, we are good Newtonians in one way that 00:20:00.676 --> 00:20:02.616 we can do these second derivatives. 00:20:02.976 --> 00:20:08.356 But our minds are also full of other types of forces, like in social interaction. 00:20:08.676 --> 00:20:14.836 I mean, I know that somebody is my superior. He or she has a force, can control my actions. 00:20:15.736 --> 00:20:25.056 I know that I'm attracted to a certain woman. and we describe attraction as a force as well. 00:20:25.156 --> 00:20:33.056 So in our understanding of what causes actions, there are other types of forces 00:20:33.056 --> 00:20:35.696 than the traditional Newtonian forces. 00:20:35.976 --> 00:20:39.536 That's why I want to use forces rather than just the movements, 00:20:39.656 --> 00:20:42.796 the kinematics or the dynamics involved in biological motion. 00:20:43.236 --> 00:20:47.436 But that's interesting, right? Because then what a prediction could be, 00:20:47.436 --> 00:20:52.036 You say, look, our perceptual systems, as applied to, let's say, 00:20:52.116 --> 00:20:54.856 many different phenomena, different levels of complexity in the world, 00:20:54.976 --> 00:20:57.256 are, if you want, assigning a notion of force. 00:20:57.496 --> 00:21:04.516 They are just inventing, let's say, a pseudo-causal relationship behind the 00:21:04.516 --> 00:21:08.956 change in the world we observe, which can be completely beside the real causes 00:21:08.956 --> 00:21:10.836 of what we observe. Sure, sure. 00:21:12.096 --> 00:21:14.796 That's a good interpretation because our brain is adding these, 00:21:14.916 --> 00:21:17.896 let's call them theoretical variables or hidden variables. 00:21:18.196 --> 00:21:23.316 But this is just like in visual perception. I mean, our eyes are adding the contours of objects. 00:21:23.536 --> 00:21:28.116 I mean, on the retina, there are no contours, but our brain is adding that in 00:21:28.116 --> 00:21:29.476 the interpretation of the world. 00:21:29.676 --> 00:21:34.776 So when we are looking at an action, we are adding the forces as a kind of contour, 00:21:34.916 --> 00:21:38.876 if you like, of the perception you have. 00:21:38.956 --> 00:21:41.416 It would be like a dynamic contour. It's a dynamic contour, exactly. 00:21:43.270 --> 00:21:46.750 So then that means we should not get too confused about this notion of force. 00:21:47.090 --> 00:21:50.570 No, no, you shouldn't restrict it to the physical force. 00:21:50.810 --> 00:21:53.650 But there's an interesting prediction there, right? Because it really means 00:21:53.650 --> 00:21:58.530 that this might be a way in which the brain is also imposing an interpretation of the world. 00:21:58.650 --> 00:22:04.090 And by virtue of that, that we are so good at recognizing biological motion with minimal cues. 00:22:04.470 --> 00:22:07.490 Yeah. Okay. No, I mean, since you mentioned Kant earlier, I mean, 00:22:07.510 --> 00:22:10.490 he was saying that we can't help by seeing causes and effects. 00:22:10.570 --> 00:22:16.930 And I say we can't help by seeing forces. I mean, that's part of doing the causal effect relation. 00:22:17.190 --> 00:22:22.230 Exactly. Yeah. No, that's very good. So now we have an idea about how to deal 00:22:22.230 --> 00:22:24.770 with motion perception of action. 00:22:26.030 --> 00:22:29.210 But in some sense, it's more like a classification, right? 00:22:29.250 --> 00:22:30.490 Because now we have these changes 00:22:30.490 --> 00:22:34.130 in my posture, let's say. I change my posture. I call this walking. 00:22:34.470 --> 00:22:39.450 And it has a certain gait. and then you say, well, and I can detect and classify 00:22:39.450 --> 00:22:44.770 it because I know how this gate was changed over time, right? 00:22:44.830 --> 00:22:47.050 I lean on one leg first and the other and so on. 00:22:47.770 --> 00:22:51.970 But then I could say, yeah, but that's not action because to talk really about 00:22:51.970 --> 00:22:54.910 action, there must be an intentional component in this. 00:22:55.090 --> 00:22:57.610 This is just movement. So how do we get from movement to action? 00:22:57.970 --> 00:23:02.070 No, I don't. Okay, that's a philosophical point. I mean, some philosophers would 00:23:02.070 --> 00:23:06.690 say that an action involves an intention. I don't use that. 00:23:06.810 --> 00:23:10.430 I don't restrict actions, the notion of an action to intentional actions. 00:23:10.770 --> 00:23:17.870 I mean, no, I don't. I mean, so I have a more general, maybe the term is not 00:23:17.870 --> 00:23:19.510 appropriate, but that's how I use it. 00:23:19.710 --> 00:23:23.130 Okay. Now, look, I'm not actually going to bicker over it. I just want to understand 00:23:23.130 --> 00:23:26.270 whether you saw a division there between, let's say, movement pattern action, 00:23:26.370 --> 00:23:29.330 but basically you say, look, as long as we're changing the body, 00:23:29.490 --> 00:23:30.850 that's what I want to recognize. 00:23:31.150 --> 00:23:37.230 Yeah. Okay. So now we have a way to directly perceive features of movement so 00:23:37.230 --> 00:23:39.070 that we can map them in the conceptual space. 00:23:39.830 --> 00:23:45.390 Now, what are the domains of that conceptual space to organize representations of action? 00:23:46.871 --> 00:23:49.451 So what are the domains? The domains are the force patterns. 00:23:49.691 --> 00:23:50.871 I mean, these are the forces. 00:23:51.131 --> 00:23:55.471 They can be complicated because you have a body moving as a complicated structure 00:23:55.471 --> 00:23:58.091 of forces. So it's a pattern. 00:23:58.231 --> 00:24:02.951 But the domain is still this more general notion of force. 00:24:03.191 --> 00:24:08.211 And my claim is that when we classify actions, I shouldn't say it's only that 00:24:08.211 --> 00:24:10.751 because there could be constraints coming from other things. 00:24:11.411 --> 00:24:17.671 If I hammer something, I use an object. I can't be hammering with my nose. 00:24:17.751 --> 00:24:22.231 I can be hammering with my hand, but that's a metaphorical use of hammer. 00:24:22.391 --> 00:24:28.071 Hammer means that that kind of action is involving an instrument. 00:24:28.431 --> 00:24:32.311 It's constrained by having an object that functions as a hammer. 00:24:32.671 --> 00:24:35.911 Maybe not the best example, but something in that direction. 00:24:35.911 --> 00:24:38.811 But still, what are the domains? 00:24:39.351 --> 00:24:43.371 What's the dimensionality of this space? Is it the limbs I'm using? 00:24:43.591 --> 00:24:48.851 Is it the direction of movement in some Cartesian coordinate system? 00:24:49.471 --> 00:24:52.531 What are these core domains? Is it body types? 00:24:52.951 --> 00:24:57.051 Okay. We've been talking mainly about biological motion. 00:24:57.291 --> 00:25:04.791 A car has a sort of pattern for you to accelerate and decelerate the car and so on. 00:25:04.991 --> 00:25:11.891 Of course, in physics, there are generators of forces, motors or muscles or whatever. 00:25:12.471 --> 00:25:18.031 But I think that when we perceive actions, when we categorize action, we extract from that. 00:25:18.231 --> 00:25:25.051 A robot walking has a totally different system of generating the forces than 00:25:25.051 --> 00:25:28.231 a human walking, but we would still classify it as the same action. 00:25:28.611 --> 00:25:33.731 My claim is that we can abstract distract away from the mechanism behind the 00:25:33.731 --> 00:25:35.411 forces and just focus on the forces. 00:25:35.751 --> 00:25:41.971 Okay. All right. So, but it still remains to be seen how that force space is exactly structured. 00:25:42.231 --> 00:25:46.531 Yeah. Yeah. Okay. So now we've actually… Well, it depends on the parts of the 00:25:46.531 --> 00:25:49.171 acting individual, the acting object. 00:25:49.351 --> 00:25:53.351 I mean, how many force generators there are and how they are related. 00:25:53.511 --> 00:25:56.371 Like our muscles in our body, for instance, which is complicated. 00:25:56.791 --> 00:26:01.551 But it's interesting, right? But if you take the car example and let's say ourselves 00:26:01.551 --> 00:26:05.771 motoring around, if you want, or navigating the world, indeed, 00:26:06.011 --> 00:26:09.831 at an abstract, more Cartesian level description, you say, okay, 00:26:09.851 --> 00:26:11.331 I'm changing my position in space. 00:26:11.631 --> 00:26:14.691 And that's the level where these forces act. And I don't really care whether 00:26:14.691 --> 00:26:16.871 the wheels are turning or the legs are moving. 00:26:17.231 --> 00:26:19.931 So this would make that point about abstraction. Yeah. Okay. 00:26:21.071 --> 00:26:25.131 So now we have an idea of conceptual space, how we can use this to... 00:26:25.725 --> 00:26:30.225 To understand action or describe, represent action, heavily relying on, 00:26:30.245 --> 00:26:35.765 let's say, a transformation of action in the real world to an internal sense of force, if you want. 00:26:36.545 --> 00:26:42.025 But now you also elaborated the same framework towards the notion of an event, 00:26:42.685 --> 00:26:45.825 which is rather critical in how we deal with the world because we don't only 00:26:45.825 --> 00:26:47.625 have dynamics, we also have an event. 00:26:47.785 --> 00:26:51.105 I have not been able to find very many cognitive theories of events. 00:26:51.245 --> 00:26:54.585 There are lots of philosophical theories, but if we think about the cognitive, 00:26:54.745 --> 00:26:58.445 I mean, The most naive description of an event is something happens to something. 00:26:59.025 --> 00:27:05.165 And then this something I call a patient. This is what is in the focus of the event. 00:27:05.685 --> 00:27:09.505 And then there is something that causes a change. And there is a result of this. 00:27:09.645 --> 00:27:12.445 I mean, well, in most cases, something happens. 00:27:12.885 --> 00:27:17.245 So I divide an event into two vectors. 00:27:17.425 --> 00:27:25.905 One describing the forces that apply to the patient. and the other vector describing 00:27:25.905 --> 00:27:27.325 the change in the patient. 00:27:27.465 --> 00:27:30.645 Sometimes the change is null, but it can be a state. 00:27:31.065 --> 00:27:35.985 So my basic model of an event are two vectors, one force vector, 00:27:36.105 --> 00:27:38.585 one result vector, acting on a patient. 00:27:38.865 --> 00:27:44.325 Very often there is an agent generating the force vector involved in the event, but that need not be. 00:27:44.445 --> 00:27:52.225 It can be just gravitation or some other non-object generating force. Right. 00:27:52.365 --> 00:27:57.365 But now, so this is nice, right? So we have the notion of event and we have 00:27:57.365 --> 00:28:01.065 decomposed it in, let's say, an agent and a patient, 00:28:01.145 --> 00:28:08.345 which let's say are some core entities who are sort of physically present, defining the event. 00:28:08.665 --> 00:28:11.645 And then we have exchanges between them, which are the forces. 00:28:11.705 --> 00:28:15.485 And these are the two vectors, right? So we have a cause and effect. fact yeah um 00:28:15.485 --> 00:28:18.205 so but are these is that really the 00:28:18.205 --> 00:28:21.345 minimal description of an event we might 00:28:21.345 --> 00:28:24.805 have events where there's no agent that is causing anything 00:28:24.805 --> 00:28:27.965 yeah the minimal is is a patient and 00:28:27.965 --> 00:28:34.365 and and and a force and a result vector that's a this is my hypothesis i mean 00:28:34.365 --> 00:28:40.505 this is a theory and there has some some consequences uh in how we perceive 00:28:40.505 --> 00:28:45.805 actions for instance i mean we would We'd be very surprised if something happens without the cause. 00:28:45.965 --> 00:28:48.085 I mean, that's the Kantian notion again. 00:28:48.265 --> 00:28:52.685 I mean, if we see something happen, we presume that there is some kind of action 00:28:52.685 --> 00:28:58.085 pattern going on behind the screens. Right. But if nothing happens? 00:28:59.484 --> 00:29:03.344 We can still describe an event in which, in some sense, nothing happens. 00:29:03.644 --> 00:29:08.824 Well, there are two kinds of nothing happening. One is just that there is no 00:29:08.824 --> 00:29:12.924 force, and consequently there is no change either, no result vector. 00:29:13.084 --> 00:29:17.424 So it's just a state. And that's a special case of an event, 00:29:17.604 --> 00:29:19.144 a fairly boring case of an event. 00:29:19.364 --> 00:29:23.564 But then there are also events where there is a force and a counterforce that 00:29:23.564 --> 00:29:25.664 balance each other, so still nothing happens. 00:29:25.664 --> 00:29:31.224 So when I'm pushing a door and the door doesn't open, I mean, 00:29:31.244 --> 00:29:35.904 I'm exerting a force and the door exerts a counterforce. 00:29:36.744 --> 00:29:41.064 Nothing happens, but there is still an interplay of forces and there is still 00:29:41.064 --> 00:29:43.484 an action from my side of pushing the door. 00:29:43.584 --> 00:29:47.864 But the counterforce brings out no effect. 00:29:47.864 --> 00:29:52.344 But if the event is, let's say, more abstract, where we say, 00:29:52.484 --> 00:29:59.784 in 2012, the Summer Olympics happened in London, took place in London, right? 00:30:00.104 --> 00:30:04.544 So does that mean that I would have to decompose that in all sorts of microscopic 00:30:04.544 --> 00:30:08.384 elements where I can now again recover these forest relationships? 00:30:08.384 --> 00:30:11.824 Relationships in this case it's a lot of intentionality going 00:30:11.824 --> 00:30:14.564 on i mean lots of people involved in this event and their 00:30:14.564 --> 00:30:19.464 joint intentionality constitutes the creation of this summer olympics it's a 00:30:19.464 --> 00:30:26.444 very complex event involving i mean a complex factor of of mental causes not 00:30:26.444 --> 00:30:31.884 physical causes generating this event i don't know how to analyze this in detail i mean i stick to 00:30:31.924 --> 00:30:34.304 the more concrete actions. 00:30:34.624 --> 00:30:39.284 Right, okay. No, but this is interesting, right? Because this issue of generalization 00:30:39.284 --> 00:30:41.424 is a challenge right now for the framework. 00:30:41.564 --> 00:30:45.244 Yeah, of course it is. I mean, going up to these more abstract types of events 00:30:45.244 --> 00:30:49.064 that we talk about that are generated by our societal structure, 00:30:49.244 --> 00:30:52.504 by our interactions with other people. I mean, I don't really have a good analysis. 00:30:52.704 --> 00:30:56.164 I mean, this is the kind of program I have. I hope to be able to extend it. 00:30:56.264 --> 00:31:01.584 But I start with the more basic concrete actions which involve physical forces and so on. 00:31:01.884 --> 00:31:06.084 No, but moreover, of course, it's sort of an easy game to sort of throw these 00:31:06.084 --> 00:31:08.144 pot shots at you and say, oh, here, I have an exception. 00:31:09.284 --> 00:31:12.384 So that's, I think, not really the point, right? It's really the point to try 00:31:12.384 --> 00:31:16.264 to understand where would be, let's say, principled transitions of the approach. 00:31:16.384 --> 00:31:20.344 That's why I thought Olympics might be sort of abstract enough that that could be challenging. 00:31:20.604 --> 00:31:27.484 It is. So we can now decompose, let's say, events in a patient. 00:31:27.484 --> 00:31:32.224 Why did you use the word patient for, let's say, the core object in the event? 00:31:33.125 --> 00:31:36.505 Well, that's something that undergoes the change, so to speak. 00:31:36.685 --> 00:31:41.005 I mean, sometimes the agent is identical with the patient. When I'm walking, 00:31:41.185 --> 00:31:44.665 I'm exerting a force on myself and I'm changing my own position. 00:31:44.965 --> 00:31:48.505 So there are cases where the agent and patient are identical. 00:31:49.745 --> 00:31:53.525 But in many cases, we can separate them, or in most cases. 00:31:53.685 --> 00:31:59.505 Okay, so is there any psychological or neuroscientific grounding for this decomposition of an event? 00:32:00.905 --> 00:32:06.485 That's a very good question. I don't know a very good answer to it yet. 00:32:06.545 --> 00:32:10.125 I mean, there are some people who are speculating about, I mean, 00:32:10.145 --> 00:32:14.925 we have two visual pathways in the brain. 00:32:15.005 --> 00:32:18.565 I mean, the more dorsal is going for the motion patterns, and you can think 00:32:18.565 --> 00:32:21.305 of that as picking out the kinematics. 00:32:21.525 --> 00:32:26.585 And maybe, I mean, maybe, I don't know if you can find some correlates of picking 00:32:26.585 --> 00:32:29.165 out the forces, but I have no idea. 00:32:29.165 --> 00:32:33.925 And then you have the more eventual pathway going to object identification, 00:32:34.165 --> 00:32:38.805 which is more involved in the static properties of all the objects. 00:32:38.965 --> 00:32:43.385 So you have a little bit of division of the forces and the objects. 00:32:43.605 --> 00:32:46.905 But I mean, that's on a very rough and general scale. I don't know if there 00:32:46.905 --> 00:32:49.905 is more detailed knowledge here. 00:32:50.225 --> 00:32:56.045 But now, so in the face of that challenge, in some sense you're using very physical 00:32:56.045 --> 00:33:01.005 metaphors, right? With force, cause, effect. And is that not limiting the framework? 00:33:01.145 --> 00:33:07.665 Because as soon as we talk about these more, let's say, metaphorical causes or about intentions. 00:33:10.602 --> 00:33:14.282 It might become confusing if we want to think about this as a cause. 00:33:15.022 --> 00:33:22.102 If I ask you, please give me this cup of coffee, Peter, to really start to think 00:33:22.102 --> 00:33:24.982 about that in terms of causes can also get very confusing. 00:33:25.062 --> 00:33:30.822 My larynx is doing stuff, sound pressure waves, hit your cochlea, et cetera. 00:33:31.342 --> 00:33:35.122 And then so, oh, but Peter grew up in this culture where these words mean certain 00:33:35.122 --> 00:33:42.462 things. But it sounds like a very, very unfruitful way to pursue the question. 00:33:42.882 --> 00:33:48.702 So wouldn't it be more useful to replace the notion cause and effect to something a bit more neutral? 00:33:49.162 --> 00:33:53.242 Well, that's what I'm trying to do. I mean, by having these force vectors or 00:33:53.242 --> 00:33:57.442 result vectors, that's, in my opinion, more. 00:33:57.642 --> 00:34:02.142 It's a way of reducing causes and effects to something that I can describe in 00:34:02.142 --> 00:34:04.582 semi-mathematical models. 00:34:04.962 --> 00:34:12.242 Okay. But then force would then be broadened to something like a metaphorical force. 00:34:12.402 --> 00:34:16.922 Yeah, well, as I say, social forces, emotional forces would be included in this 00:34:16.922 --> 00:34:18.822 notion. But are there other concepts we could consider for this? 00:34:18.862 --> 00:34:23.022 Maybe it's a bit of a boring question, but it can lead to so much confusion, 00:34:23.122 --> 00:34:25.382 right? If we use these physical metaphors. 00:34:25.482 --> 00:34:29.682 Yeah. So I'm just wondering whether there are other candidate concepts for this 00:34:29.682 --> 00:34:31.422 we could consider or not really. 00:34:31.682 --> 00:34:36.302 I don't know. No, I mean, some people talk about powers rather than forces, 00:34:36.362 --> 00:34:39.942 social powers and so on. I don't know whether that's better or not. 00:34:40.102 --> 00:34:43.502 I mean, I don't care very much about the terminology here. I mean, 00:34:43.502 --> 00:34:45.722 I care about what kind of models I'm using. 00:34:45.862 --> 00:34:49.162 And as we talked about earlier, I mean, my use of action may not fit with the 00:34:49.162 --> 00:34:50.482 philosophers' uses of action. 00:34:50.862 --> 00:34:53.442 But that's not your main concern. That's not my main concern. 00:34:53.722 --> 00:34:57.562 Okay, very good. So, okay, so we got events sorted. it. 00:34:59.102 --> 00:35:03.742 And also, you would make the point that this decomposition of events would hold, 00:35:05.039 --> 00:35:10.179 in principle, at any level of description. So whether I talk about the cup falling 00:35:10.179 --> 00:35:13.619 from the table or me writing a paper, 00:35:14.019 --> 00:35:19.839 the basic decomposition of agent, force, patient, and then a resultant vector 00:35:19.839 --> 00:35:23.359 or force would hold. Yeah, that's a minimal description. 00:35:23.579 --> 00:35:25.899 Then many events involve more components. 00:35:26.159 --> 00:35:29.299 I mean, if I hit something with a hammer and I have the instrument that's 00:35:29.299 --> 00:35:32.299 between my force exertion and the force that happens to 00:35:32.299 --> 00:35:35.279 the object I'm hitting uh and i can have 00:35:35.279 --> 00:35:38.239 uh when i give if i give something to 00:35:38.239 --> 00:35:41.359 you i mean there is a physical movement of the cup if 00:35:41.359 --> 00:35:46.419 i give you my coffee but there is also the the transfer of possession i mean 00:35:46.419 --> 00:35:53.039 which is a much more advanced and intentional part of of the of the action so 00:35:53.039 --> 00:35:56.499 there are there are different details so different components you can add in 00:35:56.499 --> 00:36:00.919 in a in an event description but the The basic ones are the two vectors and the patient. 00:36:01.039 --> 00:36:04.659 I mean, and I would say that this is a cognitive theory, so that these parts 00:36:04.659 --> 00:36:06.759 appear in all our representations of theories. 00:36:06.939 --> 00:36:09.819 Then you can go down on finer details and add more components, 00:36:09.959 --> 00:36:13.339 depending on what level of description you're aiming for. 00:36:13.859 --> 00:36:20.879 But now, if we decompose events in these terms, you could imagine that we could 00:36:20.879 --> 00:36:24.639 expose humans to, let's say, event descriptions. We put them in a scanner, 00:36:24.759 --> 00:36:28.299 fMRI, and we see which areas of the brain light up. 00:36:28.379 --> 00:36:32.159 And then hopefully you see four different areas where you can say, 00:36:32.199 --> 00:36:35.179 okay, agent and two forces. And is there anything like that? 00:36:35.419 --> 00:36:40.419 I think so. I mean, people have been looking at how verbs are represented in 00:36:40.419 --> 00:36:41.979 the brain and we are getting into verbs now. 00:36:42.359 --> 00:36:47.359 I would look for differentiations between these force vectors and the result 00:36:47.359 --> 00:36:51.639 vectors. If the brain lights up in different areas, depending on whether you 00:36:51.639 --> 00:36:54.039 talk about the causes or whether you talk about the effects. 00:36:54.399 --> 00:36:58.279 I don't know if we can find any clear results. I've taken part in a small experiment 00:36:58.279 --> 00:37:02.019 on this where there are some indication that we can make such a division. 00:37:02.319 --> 00:37:06.739 But there is a lot of research to do on this. I mean, what happens and does 00:37:06.739 --> 00:37:10.479 the brain distinguish between causes and effects in its analysis? 00:37:10.699 --> 00:37:13.919 That's, for me, a very interesting question, but I don't know anything about 00:37:13.919 --> 00:37:18.899 the answer. Well, actually, very basic learning mechanisms like classical conditioning 00:37:18.899 --> 00:37:24.079 have been interpreted in terms of the brain extracting cause-effect relationships from the world. 00:37:24.139 --> 00:37:28.779 Because you get exposed to, let's say, the tone and there comes the foot shock. 00:37:29.519 --> 00:37:35.139 And in some sense, you might interpret that as a causal relationship. That's true. 00:37:35.519 --> 00:37:38.679 So there have been relatively interesting theories about this. 00:37:38.699 --> 00:37:40.819 And it would be very consistent with what you're proposing. 00:37:40.899 --> 00:37:43.199 Okay, but that's an area I don't know very much about. Okay, 00:37:43.219 --> 00:37:46.519 yeah. So, indeed, but now you mentioned language because this, 00:37:46.559 --> 00:37:48.339 I think, is sort of if you want. 00:37:50.185 --> 00:37:53.745 The long-term ambition of this program is also to account for language. 00:37:53.805 --> 00:37:58.105 Yeah, I mean, that's, I shouldn't say the main application, but the application 00:37:58.105 --> 00:38:00.825 I'm working on right now. But why do you call it an application? 00:38:01.425 --> 00:38:05.405 Because I use this model. I mean, it's a model. I can use it to analyze causes 00:38:05.405 --> 00:38:07.165 and effects, how we perceive causes and effects. 00:38:07.325 --> 00:38:12.565 I can use it to analyze how we use verbs. I can maybe use it to analyze social interaction. 00:38:12.705 --> 00:38:16.525 I don't know what. But at the moment, I'm focusing on how this maps onto our 00:38:16.525 --> 00:38:17.885 understanding of language. 00:38:18.305 --> 00:38:21.965 Okay. Okay, but before we go to language, there's an interesting implication of that, no? 00:38:22.045 --> 00:38:24.845 Because then you're saying, well, from a cognitive perspective. 00:38:25.325 --> 00:38:29.345 There's a core meaning system organized in conceptual spaces. 00:38:29.605 --> 00:38:34.305 And this meaning system is sort of modality independent, and now it can be expressed in language. 00:38:34.445 --> 00:38:38.425 It can be used in how we interpret the world through haptics, 00:38:38.425 --> 00:38:39.665 vision, blah, blah, blah. 00:38:39.765 --> 00:38:42.625 But this is like an independent meaning module. 00:38:42.905 --> 00:38:46.565 Yes, well, independent meaning, but it is a kind of core module. 00:38:46.565 --> 00:38:51.565 I mean, the event representation, I think, would be at the core of our understanding of the world. 00:38:51.685 --> 00:38:56.265 And it depends on, I mean, you can have it to analyze perception. 00:38:56.465 --> 00:39:02.305 You perceive an event, but you can also use it as a basis for formulating linguistic expression. 00:39:02.465 --> 00:39:05.325 You can use it for different things. And as you say, I mean, 00:39:05.345 --> 00:39:06.665 it's modality independent. 00:39:06.905 --> 00:39:11.685 I mean, I don't assume a separate module for language semantics and another 00:39:11.685 --> 00:39:15.265 module for representing perceptual concepts. I mean, for me, 00:39:15.345 --> 00:39:17.225 they are on the same system. But that's interesting. 00:39:17.525 --> 00:39:22.685 The really important consequence is I can use this meaning system to also make predictions. 00:39:22.965 --> 00:39:26.825 Oh, yes. And I can, because it's modality independent, I can make cross-modal predictions. Yes. 00:39:27.045 --> 00:39:31.465 Right? So I think that… And these cross-modal predictions show up in language 00:39:31.465 --> 00:39:33.685 in terms of metaphors. Right, exactly. 00:39:33.945 --> 00:39:39.865 And they show up in body movement, in how we gesture or how we mimic things or stuff like that. 00:39:39.925 --> 00:39:43.785 Okay, so given that we're both very enthusiastic about this program…. 00:39:44.271 --> 00:39:47.331 How far did you get in accounting for language with it? Well, 00:39:47.331 --> 00:39:52.131 not to the details. I mean, I've got into some of the rough stuff. 00:39:52.311 --> 00:39:56.251 And if I start at the very basic level, I mean, all languages, 00:39:56.491 --> 00:39:59.831 I mean, the syntactic structures of languages are quite different, 00:40:00.071 --> 00:40:03.791 but all languages seem to have noun phrases and verb phrases. 00:40:03.951 --> 00:40:07.671 I mean, that's the basic thing you can have. And that division, 00:40:07.991 --> 00:40:12.091 I mean, linguists never explain why noun phrases and verb phrases exist. 00:40:12.091 --> 00:40:14.931 I mean, why are these the basic building components of language? 00:40:15.091 --> 00:40:16.271 They are taken for granted. 00:40:17.031 --> 00:40:22.931 Now, given the model of events that I have presented, known phrases map onto 00:40:22.931 --> 00:40:25.331 agents and patients primarily. 00:40:26.031 --> 00:40:29.891 They might map onto instruments and other stuff. But basic uses of known phrases 00:40:29.891 --> 00:40:31.571 is to denote agents and patients. 00:40:32.011 --> 00:40:38.351 And basic uses of verb phrases is to denote force vectors and result vectors. 00:40:38.351 --> 00:40:43.551 So this division between the two objects and the two vectors maps onto the division 00:40:43.551 --> 00:40:45.051 between noun phrases and verb phrases. 00:40:45.371 --> 00:40:47.671 Okay, but how clean is that mapping? 00:40:48.031 --> 00:40:53.471 I don't know how clean it is, but if it serves as a grounding for learning language. 00:40:55.191 --> 00:41:01.031 I don't think I can explain all features of the semantics of a human language 00:41:01.031 --> 00:41:02.951 because that's very rich. 00:41:02.951 --> 00:41:08.051 But if I can say something about the general principles of how the semantics 00:41:08.051 --> 00:41:11.951 is structured, I can use that to explain how children can learn a language. 00:41:12.311 --> 00:41:17.511 Because the data you get as a child is not very rich. You need some kind of constraints. 00:41:18.011 --> 00:41:23.171 And these event structures, if my model of the cognitive representations of 00:41:23.171 --> 00:41:29.411 events are correct, they give you a structure that constrains what words can refer to. 00:41:29.551 --> 00:41:32.931 I mean, I make this basic distinction between noun and verb. 00:41:32.951 --> 00:41:33.911 Phrases and verb phrases. 00:41:34.391 --> 00:41:40.651 But now, of a verb, you said, look, a verb is essentially a convex region in a single domain. 00:41:41.091 --> 00:41:44.631 Yeah. Right? So that seems a rather strong statement. 00:41:44.931 --> 00:41:49.411 Yeah. I mean, before we get to that, I mean, let me start by saying that a verb means either. 00:41:49.491 --> 00:41:55.191 It either refers to the force vector or it refers to the result vector. 00:41:55.311 --> 00:41:57.931 That's right. You don't have a single verb. I mean, you can have compositions 00:41:57.931 --> 00:42:02.231 with prepositions and stuff like that where you get. But the root of a verb, 00:42:02.291 --> 00:42:06.051 is my hypothesis, refers to either the force vectors. 00:42:06.231 --> 00:42:09.111 And these verbs are what we call manner verbs, how you do things. 00:42:09.191 --> 00:42:10.651 You hit things, you pull and you push. 00:42:10.831 --> 00:42:14.091 And then you have verbs that refer to the result vector. 00:42:14.271 --> 00:42:19.691 And these are, well, results like painting or heating or whatever, opening. 00:42:21.311 --> 00:42:25.411 But wait, if it's a verb, it always requires an agent. Right. 00:42:27.149 --> 00:42:30.029 Um but if it's an agent it 00:42:30.029 --> 00:42:34.009 requires that um but look 00:42:34.009 --> 00:42:36.989 if we have to the patient now and the patient 00:42:36.989 --> 00:42:39.929 undergoes a change let's say uh you 00:42:39.929 --> 00:42:42.869 you hit my finger with a hammer my finger turns purple 00:42:42.869 --> 00:42:46.349 the result vector is 00:42:46.349 --> 00:42:49.169 not necessarily an action as i can 00:42:49.169 --> 00:42:52.889 capture in a verb no no right well you say 00:42:52.889 --> 00:42:55.829 turns purple i mean that's uh you have turning here means 00:42:55.829 --> 00:42:58.909 and then you have the color term so what happens is 00:42:58.909 --> 00:43:01.629 that turn it means a change or a movement no but 00:43:01.629 --> 00:43:04.709 you agree i could describe it in a way that there's no verb involved i 00:43:04.709 --> 00:43:07.589 said now yeah my finger is blue 00:43:07.589 --> 00:43:11.289 well the property change yeah there's a property change there's a property change 00:43:11.289 --> 00:43:16.689 so how so how do i well on it this is not in itself an action of the patient 00:43:16.689 --> 00:43:20.729 right it's a property change yeah it's a property but it's that's the result 00:43:20.729 --> 00:43:25.049 vector and it's described i mean you can say turns red turns purple, 00:43:25.229 --> 00:43:26.269 you can say becomes purple. 00:43:26.449 --> 00:43:32.049 We have those verbs that help us in saying that there is a change in a property, 00:43:32.189 --> 00:43:33.409 and purple is a property word. 00:43:33.689 --> 00:43:37.529 But you're saying it's going from some other color into the region of purple. 00:43:37.749 --> 00:43:42.869 So there is a change in the result. The result vector is going from some region 00:43:42.869 --> 00:43:45.469 into the purple region. That's the result vector. 00:43:45.789 --> 00:43:49.249 So what I was after, I was just trying to understand whether you're saying verbs 00:43:49.249 --> 00:43:53.229 describe change or verbs describe action. 00:43:53.369 --> 00:43:56.949 No, no. No, there are two kinds. There are verbs that describe actions. 00:43:57.109 --> 00:44:02.509 These are the manner verbs. And then there are verbs that describe changes. 00:44:02.849 --> 00:44:06.269 And I say there is a fairly tight division between them. There are some verbs 00:44:06.269 --> 00:44:09.749 that can be used for both topics, but in a particular sentence, 00:44:09.909 --> 00:44:13.609 they're used either to describe the manner, the force vector. 00:44:13.629 --> 00:44:18.609 Right, but I was trying to understand, let's say, the overlap between these two domains. 00:44:18.789 --> 00:44:22.829 Because you could say, look, you lift the hammer and now the patient ran away. 00:44:22.829 --> 00:44:29.849 So now I'm acting so I'm not turning blue anymore so how unique is that mapping 00:44:29.849 --> 00:44:34.929 of these two kinds of verbs to either the cause or the effect, 00:44:36.609 --> 00:44:43.489 I would say it's fairly of course there is a particular action can have very 00:44:43.489 --> 00:44:47.809 many different outcomes as you say depending on the situation on the context 00:44:47.809 --> 00:44:52.469 but we don't have verbs that summarize causes and effects I mean this is This 00:44:52.469 --> 00:44:55.189 is my constraint on how we learn these verbs. 00:44:55.289 --> 00:45:00.129 They either express the cause, I mean, the force vector, or the effect of an event. 00:45:00.349 --> 00:45:03.009 So that's a kind of cognitive prediction here. 00:45:05.069 --> 00:45:11.109 For instance, you could be running towards me, and I'm running away. So we're both running. 00:45:11.329 --> 00:45:13.329 The agent and the patient are both running. 00:45:13.649 --> 00:45:17.709 So the verb we use to now describe the cause and effect is the same. 00:45:18.705 --> 00:45:22.745 Is that a problem? No, because you use the preposition away, 00:45:23.045 --> 00:45:25.785 which introduces a result. 00:45:26.785 --> 00:45:33.905 Run away is, you modify the manner by running away that introduces a change 00:45:33.905 --> 00:45:35.965 in space, which is a result vector. 00:45:36.245 --> 00:45:39.605 By having this preposition added to running, you're changing it from a manner 00:45:39.605 --> 00:45:40.845 vector to a result vector. 00:45:41.325 --> 00:45:43.845 Okay, but that could also say Peter's running and Paul is running. 00:45:44.145 --> 00:45:46.145 Yeah, but then you don't have a cause and effect. 00:45:46.145 --> 00:45:49.025 Okay but then 00:45:49.025 --> 00:45:52.165 then it's two events but you did bring in the proposition now 00:45:52.165 --> 00:45:55.465 so then it becomes let's say a conglomerate yeah it's 00:45:55.465 --> 00:45:58.305 not only the verb anymore no no no it's it's i mean there 00:45:58.305 --> 00:46:04.005 has to be at least one vector and one one uh editor patient i mean so that's 00:46:04.005 --> 00:46:07.365 the noun phrase and the verb phrase are the minimal elements of of an event 00:46:07.365 --> 00:46:11.465 description okay okay so it's a phrase it's not just a single word no no no 00:46:11.465 --> 00:46:15.785 it can it can i mean then you add grammar but i mean you're and all kinds of 00:46:15.785 --> 00:46:16.885 compositions of descriptions. 00:46:17.285 --> 00:46:22.745 But what is denoted by these composite expressions are agents and patients. 00:46:23.085 --> 00:46:30.665 Okay, so now we have these two kinds of verbs, and then from that you came with 00:46:30.665 --> 00:46:34.385 the prediction, if you want, because, okay, if I look at that from the perspective 00:46:34.385 --> 00:46:35.385 of the conceptual space. 00:46:36.065 --> 00:46:42.945 It means that if every verb is indeed in the convex region, they cannot sort 00:46:42.945 --> 00:46:45.285 of be in two disjoint regions at the same time. 00:46:45.405 --> 00:46:50.545 And therefore, let's say cause and effect verbs, there are no occurrences of 00:46:50.545 --> 00:46:52.585 verbs that are both cause and effect. 00:46:52.765 --> 00:46:56.385 No. And as a matter of fact, you can make it even stronger because among the 00:46:56.385 --> 00:46:59.325 result verbs, they only refer to a single domain. 00:46:59.565 --> 00:47:01.805 So when you're heating something, you're changing the temperature. 00:47:01.825 --> 00:47:04.245 When you're moving something, you're changing the spatial location. 00:47:04.585 --> 00:47:07.365 And when you're painting something, you're changing the color domain. 00:47:07.605 --> 00:47:12.565 I mean, I had previously a theory about but adjectives that say that adjectives 00:47:12.565 --> 00:47:15.345 refer to regions in single domains. 00:47:15.525 --> 00:47:19.045 So you have the color words, color adjectives that refer to the color space. 00:47:19.145 --> 00:47:22.985 You have the temperature, hot and cold, I refer to temperature, and so on. 00:47:23.145 --> 00:47:27.145 Now, I'm generalizing this single domain hypothesis also to verbs. 00:47:27.465 --> 00:47:31.445 And there are lots of potential counterexamples. 00:47:31.485 --> 00:47:36.185 I don't know how well this hypothesis will hold up. But if it does, 00:47:36.265 --> 00:47:40.745 it shows a very nice parallel between the structure of adjectives and the structure of verbs. 00:47:40.885 --> 00:47:43.265 Right. Which would be nice from a cognitive point of view. 00:47:43.625 --> 00:47:47.405 From the point of view of cognitive economy, I should say. 00:47:47.685 --> 00:47:51.325 On the other hand, you have some wiggle space, right? Because you have a free 00:47:51.325 --> 00:47:55.485 parameter in your model, which is the domain itself. Okay, so that... 00:47:56.029 --> 00:47:59.889 So the point is, even if you find counterexamples, they raise interesting questions 00:47:59.889 --> 00:48:01.889 about really the structuring of these domains. 00:48:02.249 --> 00:48:07.809 Well, I'm saying that at least that the same domains apply to adjectives that apply to verbs. 00:48:08.009 --> 00:48:14.829 And we find these nice mappings between movement verbs and position adjectives and stuff like that. 00:48:15.049 --> 00:48:21.069 But now one problem I had with that prediction is that I could also argue, 00:48:21.169 --> 00:48:22.489 well, look, yeah, sure thing. 00:48:22.489 --> 00:48:27.469 This is not a big surprise because, let's say, in general, words used in a language 00:48:27.469 --> 00:48:33.989 that are ambiguous and confusing will just die out because they don't help for pragmatical terms. 00:48:34.529 --> 00:48:41.349 So in that sense, could I say, well, this is like, this is the truism. 00:48:41.689 --> 00:48:46.549 Because, of course, words in our language are only there because they're not ambiguous. Yeah. 00:48:46.669 --> 00:48:51.809 I would say it's, well, you can think of it as a truce, but the reason is that 00:48:51.809 --> 00:48:55.809 if you have two complicated words, I mean, words with two complicated semantics, 00:48:56.189 --> 00:48:59.109 they would be very difficult to learn. 00:48:59.569 --> 00:49:05.009 And having them sorted up into referring to a single domain makes them easier to learn. 00:49:05.169 --> 00:49:09.629 I think that this is a good constraint that explains how children can pick up 00:49:09.629 --> 00:49:10.989 language so quickly as they do. 00:49:11.189 --> 00:49:14.689 That's one of the constraints, and it's not the only one. But your prediction 00:49:14.689 --> 00:49:18.249 is that language have evolved the way they did, also to facilitate learnability. 00:49:18.629 --> 00:49:25.029 Oh, yes. Oh, yes. And, I mean, there are these famous Chomsky arguments of the 00:49:25.029 --> 00:49:28.289 positive of the data. We don't get enough data. 00:49:28.389 --> 00:49:30.489 We have to have innate structures. 00:49:30.929 --> 00:49:35.129 I don't think that's a very strong argument because he only considers syntax in his studies. 00:49:35.169 --> 00:49:39.249 If you look at the interplay between perception and language, 00:49:39.449 --> 00:49:44.049 as I've been doing in my studies of actions and events, I mean, 00:49:44.049 --> 00:49:46.829 you can generate a lot more constraints on learning. 00:49:46.929 --> 00:49:52.589 And that's what one of my aims is, to identify these constraints on how the 00:49:52.589 --> 00:49:54.969 semantics of words are structured. 00:49:55.349 --> 00:49:58.229 Yeah, but it's interesting about this argument about the poverty of the stimulus 00:49:58.229 --> 00:50:00.209 originating in the 1950s. 00:50:00.962 --> 00:50:04.102 That in some sense the data is still out there. 00:50:04.262 --> 00:50:11.022 I mean, the data has not been conclusively summarized in some way that indeed 00:50:11.022 --> 00:50:17.662 the sensory stimuli the growing child is exposed to is that restricted or not 00:50:17.662 --> 00:50:20.502 structured enough to allow learnability. 00:50:20.802 --> 00:50:24.182 But what kind of data do you have to support your position here? 00:50:25.162 --> 00:50:31.522 No, I mean, the data I have comes from, to the extent I can confirm my models here, 00:50:31.702 --> 00:50:36.622 that I have to look at, for instance, the verbs, to see whether this classification 00:50:36.622 --> 00:50:41.662 between manner verbs and result verbs hold water, and that's a hot topic in 00:50:41.662 --> 00:50:43.782 linguistics at the moment. 00:50:44.102 --> 00:50:45.822 And if it does, then we have some 00:50:45.822 --> 00:50:50.682 data showing that there is this kind of division of meanings in verbs. 00:50:50.962 --> 00:50:55.462 And that division of meanings would help me, would generate a constraint that 00:50:55.462 --> 00:50:59.222 can help me in explaining meaning why kids can learn language so easily. 00:50:59.682 --> 00:51:03.982 Okay, but that would mean that the structuring of the conceptual space of language, 00:51:04.502 --> 00:51:07.922 should then coincide with the progression of language learning. 00:51:08.962 --> 00:51:12.462 Yeah, I mean, that's a give and take variation, yeah. 00:51:12.622 --> 00:51:15.522 So how, but how do you really see that feedback mechanism? Because there's some, 00:51:15.582 --> 00:51:17.302 there must be some bootstrapping in this, right? 00:51:17.362 --> 00:51:21.062 Because the argument is always, look, if it's just feed forward and you grab 00:51:21.062 --> 00:51:22.362 it from the world, it's not enough. 00:51:22.542 --> 00:51:27.082 No. So what are these key rules that would help the bootstrapping in this? 00:51:27.142 --> 00:51:28.262 Do you have an idea about that? 00:51:28.342 --> 00:51:33.602 I have thought a little bit about the ordering in which you learn domains as a child. 00:51:33.802 --> 00:51:37.562 I mean, you learn the shape of objects quite early. You interact with things 00:51:37.562 --> 00:51:39.602 with your hands and your mouth and so on. 00:51:40.242 --> 00:51:43.182 Color comes perhaps later. You learn about spatial relations. 00:51:43.582 --> 00:51:46.882 But then there are lots of spaces where you only learn things later. 00:51:47.122 --> 00:51:52.282 And I looked a little bit at child data on when certain domains are learned. 00:51:52.722 --> 00:51:56.082 So color and shape comes very early. But for instance, all the... 00:51:56.570 --> 00:52:03.630 All the domains related to knowledge. I mean, knowing, believing, and lying, and so on. 00:52:03.870 --> 00:52:07.070 Children normally don't learn them until they're about three or four years old. 00:52:07.230 --> 00:52:10.790 So that's a more abstract domain that comes later. 00:52:11.150 --> 00:52:18.310 Or take economic terms. I mean, like a loan or inflation, to take a tough word. 00:52:18.470 --> 00:52:21.250 I mean, you can't teach a child that. Bail out. 00:52:21.390 --> 00:52:27.210 Because for kids, money are coins and bills. And they don't know about this 00:52:27.210 --> 00:52:28.950 abstract space of economic relations. 00:52:29.330 --> 00:52:34.550 That's something that they learn much later. So my idea of some kind of progression 00:52:34.550 --> 00:52:39.550 of domains would be another constraint on language learning. 00:52:39.810 --> 00:52:44.530 Some domains come earlier in their development, and then via learning a language, 00:52:44.590 --> 00:52:47.930 via being part of a culture, you learn further, more abstract domains. 00:52:48.230 --> 00:52:51.330 But then your prediction is that there must be a hierarchy of conceptual spaces. 00:52:51.730 --> 00:52:57.530 A hierarchy of domains, yes. Of domains. Yeah. Okay. And would that hierarchy, 00:52:57.670 --> 00:53:01.790 let's say, also be an explosion in dimensionality or not necessarily? 00:53:02.130 --> 00:53:07.250 Well, each new domain adds dimensions, so it's bringing out more dimensions. 00:53:07.390 --> 00:53:12.550 Yeah, but it could also collapse dimensions of preceding domains, right? It could, yeah. 00:53:12.990 --> 00:53:18.410 No, I think in general, our minds are growing in the number of dimensions we 00:53:18.410 --> 00:53:19.950 are using. Okay, all right. 00:53:20.050 --> 00:53:28.810 So then you had a number of predictions, if you want, about verbs, right? 00:53:28.890 --> 00:53:34.270 For instance, that similarities in verb meaning, or the subcategories of verbs, 00:53:34.490 --> 00:53:38.050 or the subregions in the notion of walking. 00:53:39.169 --> 00:53:44.229 So what are these predictions exactly? Well, we categorize. 00:53:44.289 --> 00:53:48.289 I mean, one idea behind the concept of spaces is that similarity plays a great 00:53:48.289 --> 00:53:50.869 role in categorization. 00:53:51.209 --> 00:53:57.449 So we categorize colors as yellow because they are similar in our perception. 00:53:57.669 --> 00:54:01.409 And we categorize actions because they are similar in our perception. 00:54:01.409 --> 00:54:06.449 So I would say that running is very much similar to walking, 00:54:06.629 --> 00:54:12.549 or more similar to walking than running is to flying, for instance. 00:54:12.649 --> 00:54:14.129 Yeah, that's a good example. 00:54:14.469 --> 00:54:19.729 And we use these similarities. I mean, I don't have a measure of distance between 00:54:19.729 --> 00:54:24.109 force patterns, but somehow the force pattern of running is more similar, 00:54:24.229 --> 00:54:27.269 closer to that of walking than it is to flying. 00:54:27.269 --> 00:54:33.669 And we use these similarities when we judge the similarities between verb meanings. 00:54:34.049 --> 00:54:37.269 And also, if you can generalize, I mean, take walk again. 00:54:37.469 --> 00:54:41.129 There are a number of subdivisions of walking. You can be limping, 00:54:41.169 --> 00:54:43.049 you can be marching, you can be strutting. 00:54:43.669 --> 00:54:46.509 I mean, there are lots of kinds of walking. 00:54:46.709 --> 00:54:51.169 But they all fall under the general notion of the walking domain of possible 00:54:51.169 --> 00:54:54.049 walking patterns, force patterns. 00:54:54.389 --> 00:54:59.709 But the subdivisions then are sub-regions of this action space. 00:54:59.949 --> 00:55:03.909 Right, and that would then coincide with, say, our ability to distinguish these 00:55:03.909 --> 00:55:06.949 movement patterns. Yeah, this is just like we have subdivisions of nuns. 00:55:06.949 --> 00:55:09.149 You can have an animal, a dog, a terrier. 00:55:09.349 --> 00:55:13.069 You have this hierarchy. We have the same similar hierarchy of action representations. 00:55:17.842 --> 00:55:23.042 But you also had, let's say, a more detailed decomposition of, 00:55:23.082 --> 00:55:27.362 let's say, how verbs translate to action in the world. 00:55:27.462 --> 00:55:33.042 Because if we take the example of push that you talked about, 00:55:33.222 --> 00:55:39.242 to push an object in itself is not necessarily the complete story, right? 00:55:39.282 --> 00:55:43.342 Because in the result vector, different things can happen, right? 00:55:43.342 --> 00:55:48.502 So does that mean I also have to think about then subcategories or let's say 00:55:48.502 --> 00:55:54.322 decomposition of push into different kinds of resultant vectors or not? 00:55:54.482 --> 00:55:58.542 You must make a distinction between what the verb represents and the event you're 00:55:58.542 --> 00:56:00.942 describing because the verb represents a force vector. 00:56:01.002 --> 00:56:06.302 Pushing is a force vector normally horizontally towards an object. 00:56:06.522 --> 00:56:13.082 But then the object may be stuck or may be heavy or so nothing happens. 00:56:13.342 --> 00:56:16.582 But that's part of the event I mean that's the result vector but 00:56:16.582 --> 00:56:20.182 the push it as the verb 00:56:20.182 --> 00:56:23.122 only concerns the force vector and the 00:56:23.122 --> 00:56:26.942 result may be a complicated story depending on how the world looks like but 00:56:26.942 --> 00:56:30.482 how do I segment this now because imagine so you push me and I'm pushing you 00:56:30.482 --> 00:56:35.982 back so I'm your patient and you're my patient yeah then there are two events 00:56:35.982 --> 00:56:42.422 okay exactly right so the segmentation of let's say um. 00:56:44.777 --> 00:56:50.637 Real-world events, this segmentation would fall along the lines of the forces 00:56:50.637 --> 00:56:51.457 and how they're exchanged. 00:56:52.257 --> 00:56:56.997 But would that decomposition be unambiguous? No, not necessarily. 00:56:57.177 --> 00:56:58.357 Does it need to be unambiguous? 00:56:59.497 --> 00:57:04.077 You can have different levels of description of events. 00:57:04.417 --> 00:57:08.277 I mean, if you take wrestling, there is a continuous exchange of forces, 00:57:08.297 --> 00:57:10.817 but we still generalize this 00:57:10.897 --> 00:57:18.957 general pattern of two people pushing against or pulling against each other 00:57:18.957 --> 00:57:21.357 as a general type of interaction. 00:57:21.817 --> 00:57:26.037 So that's a high-level description. Then, of course, I can have a more detailed 00:57:26.037 --> 00:57:28.617 description of a wrestling match. 00:57:28.777 --> 00:57:33.877 I can have, I'm bending your knee or bending your arm or pushing there or lifting there or whatever. 00:57:34.117 --> 00:57:37.337 I mean, there are lots of sub-events in this global event. 00:57:37.337 --> 00:57:42.017 But is the prediction of this possibly that if the forces cancel each other 00:57:42.017 --> 00:57:45.157 out, or if the roles of patients and agents cancel each other out, 00:57:45.297 --> 00:57:51.657 that that is a necessary reason to step to a next level of abstraction? No, no, no. 00:57:51.737 --> 00:57:56.437 I mean, it's just as when you're describing a scene around you, 00:57:56.477 --> 00:57:57.737 you can say, well, I'm in a room. 00:57:57.857 --> 00:58:01.137 But I can also say I'm in a room with this and that pieces of furniture. 00:58:01.337 --> 00:58:04.897 I mean, I can go down and talk about the details. 00:58:05.057 --> 00:58:09.457 I mean, it's a level of attention we have on a particular situation. 00:58:09.457 --> 00:58:12.297 Situation okay so so now you you came 00:58:12.297 --> 00:58:15.297 a long way with the conceptual spaces approach to understand 00:58:15.297 --> 00:58:18.317 language right so in some sense you're saying nouns are categories 00:58:18.317 --> 00:58:25.217 yeah right adjectives are properties um prepositions are are force and spatial 00:58:25.217 --> 00:58:31.357 relations sort of relationships yeah verbs now are forces and results causes 00:58:31.357 --> 00:58:36.617 and effects vectors yeah right and then adverbs verbs are like modifying vectors. 00:58:37.077 --> 00:58:41.077 Yeah. What would that exactly mean? Well, if I push something, 00:58:41.357 --> 00:58:45.557 I mean, I'm exerting a vector, but I'm not saying very much about how strongly I push. 00:58:45.717 --> 00:58:50.837 I mean, the strength. If I say I push strongly, I'm saying that the vector I'm 00:58:50.837 --> 00:58:57.137 exerting belongs to the, well, with stronger forces, I exert more force. 00:58:57.297 --> 00:59:01.237 So it's like a multiplier or a modifier 00:59:01.237 --> 00:59:05.677 of a vector. Or I can change the direction. I say I push downwards. 00:59:06.217 --> 00:59:11.337 That means I'm changing the direction of the pushing. 00:59:11.637 --> 00:59:17.517 Okay. So that's a very complete, if you want, decomposition of this very basic 00:59:17.517 --> 00:59:20.297 notion of an event in the atoms of language. 00:59:21.517 --> 00:59:27.497 And then your proposal is that sentences as such, as a unit, capture the event. 00:59:27.657 --> 00:59:32.097 Yep. But now to go from these linguistic atoms to a sentence, we need grammar. 00:59:33.066 --> 00:59:36.786 Not necessarily. I mean, we can communicate without grammar quite a lot. 00:59:36.946 --> 00:59:40.506 I mean, there is this thing that's called proto-language, that is just meaning 00:59:40.506 --> 00:59:41.986 components without any grammar. 00:59:42.126 --> 00:59:47.586 I think grammar is needed to disambiguate a lot of things here. 00:59:47.586 --> 00:59:56.506 I mean, in English, for instance, there are a lot of words that can mean either a verb or an object. 00:59:56.726 --> 00:59:59.666 Like hammer. I mean, hammer is a verb or it can be an object. 01:00:00.066 --> 01:00:03.946 And the word in isolation doesn't tell you what it is. 01:00:04.306 --> 01:00:11.866 But if you put it in a context and add syntactic markers, tenses or third-person 01:00:11.866 --> 01:00:16.846 markers or whatever, then you can see whether it's a noun or a verb. 01:00:17.586 --> 01:00:23.746 So syntax for me is a tool to disambiguate the uses of words. 01:00:23.946 --> 01:00:29.326 And then syntax is another constraint on how to learn a language because syntactic 01:00:29.326 --> 01:00:35.926 markers will help a child to see whether we're talking about an action or we're 01:00:35.926 --> 01:00:36.886 talking about an object. 01:00:37.686 --> 01:00:43.526 Yeah, but what I find surprising about this, and that's probably because you're 01:00:43.526 --> 01:00:47.306 more aware about the intricacies of language, But I would expect that you would 01:00:47.306 --> 01:00:51.086 have said something like, well, I have my basic decomposition of an event. I have an agent. 01:00:51.326 --> 01:00:54.586 I have my two vectors and I have a patient. It's a cause and effect. 01:00:55.086 --> 01:00:59.566 And it's that order of an agent cause and effect on the patient. 01:00:59.646 --> 01:01:01.426 It's that order that translates to syntax. 01:01:02.146 --> 01:01:09.806 No, not necessarily. Why not? Because there is an event or an element in between. 01:01:10.046 --> 01:01:15.686 There is an event. But there is also you looking at the event and you focusing on something. 01:01:15.946 --> 01:01:20.746 I mean, you're looking at what the agent is doing or you're looking at what the patient is doing. 01:01:20.806 --> 01:01:24.386 That's two different foci on the event. 01:01:24.506 --> 01:01:29.346 And that attention focusing shows up in language. 01:01:29.646 --> 01:01:37.946 So in a large class of languages, the subject turns out to be the thing you're focusing on. 01:01:37.946 --> 01:01:43.366 So if I say that I'm hitting you, I'm focusing on me doing the action, 01:01:43.486 --> 01:01:46.066 but if I'm saying you are being hit... 01:01:46.677 --> 01:01:51.297 I'm focusing on what's happening to you. I'm making you the subject. 01:01:51.817 --> 01:01:55.957 And in both cases, I'm the patient. I'm the agent. You're the patient. 01:01:56.317 --> 01:02:02.017 So what becomes the subject of a sentence is determined by what is the focus 01:02:02.017 --> 01:02:03.817 of attention on the event. 01:02:04.217 --> 01:02:07.117 Yeah, but exactly this example, I think, is a good one. I mean, 01:02:07.137 --> 01:02:10.877 you might have reached a stage in this discussion that you want to hit me. 01:02:10.977 --> 01:02:15.857 But still we say Peter is hitting Paul. all. So first we have the agent, 01:02:15.937 --> 01:02:17.997 then we have the cause, and then we have the patient. 01:02:18.357 --> 01:02:20.717 And if it's the other way around, we say Paul is hitting Peter. 01:02:20.897 --> 01:02:25.917 So it's that ordering in the language that is rather directly correlated with 01:02:25.917 --> 01:02:27.397 the ordering of your event decomposition. 01:02:27.857 --> 01:02:31.697 No, not necessarily. First of all, there are all kinds of orderings between 01:02:31.697 --> 01:02:34.897 subject, object, and verb in languages, so that doesn't. 01:02:34.997 --> 01:02:38.477 But I think the most important thing is just what is your focus of attention? 01:02:38.717 --> 01:02:43.637 That's what determines the ordering, because the thing most focused on comes first. 01:02:43.957 --> 01:02:48.797 Yeah, but isn't that now a tricky direction to take? Because now you bring in 01:02:48.797 --> 01:02:51.237 for the first time, let's say, a point of view. 01:02:51.697 --> 01:02:56.437 And you say, depending on the point of view, you reorganize the conceptual space, 01:02:56.657 --> 01:02:59.457 or you reorganize the event structure. 01:02:59.717 --> 01:03:02.877 But it shouldn't really matter, because independent of the point of view, 01:03:03.097 --> 01:03:07.017 the event structure always has the agent, the cause, the patient, and the event. 01:03:07.077 --> 01:03:10.677 No, you don't reorganize it. You focus on it, And then you turn this focusing 01:03:10.677 --> 01:03:11.977 into a linguistic expression. 01:03:12.417 --> 01:03:19.177 So this is an intermediary step between your event perception and how you describe it. 01:03:19.317 --> 01:03:23.857 And between your perception, there is the focusing of attention. 01:03:24.077 --> 01:03:30.197 And that determines partly how you assign syntactic roles to the parts of the 01:03:30.197 --> 01:03:31.777 event. Okay, so this is interesting, right? 01:03:31.857 --> 01:03:35.117 So I find this interesting. interesting, because on the one hand, 01:03:35.137 --> 01:03:40.037 you're saying, look, from, if you want a cognitive representational perspective, 01:03:40.257 --> 01:03:43.597 I can tell you how an event is represented, is decomposed. 01:03:44.157 --> 01:03:46.977 So there we go. We have the agent and our cause, the patient effect. 01:03:47.657 --> 01:03:52.337 Then you say those components, I can map on the components of language. 01:03:52.577 --> 01:03:56.557 Yep. Okay. Yep. So we're still doing fine. So this is like we have our conceptual 01:03:56.557 --> 01:03:59.337 space, the domains, and there we go. 01:03:59.597 --> 01:04:02.177 And we map this to, let's say, components of language. 01:04:03.157 --> 01:04:08.017 And then now we want to make a sentence. And you're saying the whole sentence 01:04:08.017 --> 01:04:10.837 describes the whole event. No, I'm not saying that. 01:04:11.057 --> 01:04:14.697 I'm saying the sentence describes an aspect of an event. You have to pick out 01:04:14.697 --> 01:04:16.417 a certain aspect of an event. 01:04:16.517 --> 01:04:20.497 And you can start very basically with a subject and a verb. 01:04:20.637 --> 01:04:23.357 But then you can add prepositional phrases. You can add modifiers. 01:04:23.617 --> 01:04:28.157 You can add, I mean, I can say I hit you, but I can say I hit you strongly with a hammer. 01:04:28.277 --> 01:04:32.677 I hit you strongly with a hammer so your eyes became blue and so on. 01:04:32.717 --> 01:04:39.197 I can expand the description of the event by adding more and more parts of the event. 01:04:39.377 --> 01:04:42.677 So that means that I have now a transformation. I have to look at some transformation 01:04:42.677 --> 01:04:49.397 that goes from, let's say, a neutral mapping into a conceptual space of an event, 01:04:49.437 --> 01:04:50.637 as it is in the outside world. 01:04:51.037 --> 01:04:54.057 Then I want to translate this to language, and for that, I have to make a transformation. 01:04:54.357 --> 01:04:56.097 And this transformation depends on the point of view. 01:04:56.437 --> 01:05:00.517 Well, yeah, you make a selection, basically. But in making this selection, 01:05:00.897 --> 01:05:04.677 I still map it into your basic event structure. Yep. 01:05:05.240 --> 01:05:09.700 No? Yes. And to map that basic event structure in language, I still face the 01:05:09.700 --> 01:05:14.360 same problem, which means there's an agent to cause a patient and an effect. 01:05:14.420 --> 01:05:17.040 I mean, the event structure will constrain language. 01:05:17.260 --> 01:05:20.520 It will not determine what you say, but it will constrain the structure of your language. 01:05:20.880 --> 01:05:23.920 But then where's the magic of coming in for the syntax? 01:05:24.960 --> 01:05:28.820 Because still, this is how I have to map it to my sentence. Yeah. 01:05:29.520 --> 01:05:36.420 But, I mean, of course, as I said, that you have to have some way of disambiguating the role. 01:05:36.580 --> 01:05:41.880 So you have to decide, know whether something is a verb or a noun, 01:05:41.980 --> 01:05:46.980 for instance, or whether something is talking about spatial relations or whatever. 01:05:47.640 --> 01:05:54.060 You can do a lot with proto-language without syntax. And when children are learning 01:05:54.060 --> 01:05:55.700 language, they start with the proto-language. 01:05:55.860 --> 01:05:59.280 The grammar comes, most of the grammar comes much later. 01:05:59.580 --> 01:06:03.420 Right. And if you start communicating in another foreign language, 01:06:03.480 --> 01:06:07.240 you start without the syntax. I mean, you have a vocabulary and you try your best. 01:06:08.020 --> 01:06:11.980 So, I mean, this is the proto-language structure of communication. 01:06:12.260 --> 01:06:17.980 For me, syntax is just a tool for helping us in getting more better understanding 01:06:17.980 --> 01:06:22.280 how you made the transformation for the event description to the linguistic expression. 01:06:23.120 --> 01:06:27.180 So, Peter, to wrap up, I have two questions. 01:06:27.180 --> 01:06:32.600 So, I mean, you've gone around in this sort of very cognitive view on the mind, 01:06:32.680 --> 01:06:34.140 if you want, and its different capabilities, 01:06:34.380 --> 01:06:41.780 proposing this notion of conceptual spaces for a while now, looking at its consequences, 01:06:41.920 --> 01:06:43.700 and that seems to be panning out rather well. 01:06:44.080 --> 01:06:49.000 So what would be Peter's law that we have to adhere to in trying to understand the mind? 01:06:51.260 --> 01:06:56.120 Well, one important thing is that our mind divides information into domains. 01:06:56.120 --> 01:07:03.500 I don't have a crisp definition of what is a domain, but I can give you examples. 01:07:04.180 --> 01:07:09.240 And this domain structure shows up in a lot of circumstances. 01:07:09.900 --> 01:07:13.720 So that's one of the big messages I have. Okay. 01:07:14.100 --> 01:07:18.140 And then the second one is five years from now I'm going to go visit you in 01:07:18.140 --> 01:07:22.860 Lund and say, okay, you gave me this prediction five years ago and now I want 01:07:22.860 --> 01:07:26.300 to know whether it really happened. So what's the one prediction you feel most 01:07:26.300 --> 01:07:27.160 passionate about today? 01:07:29.402 --> 01:07:34.982 A prediction, testable prediction that you're going to see tested and validated 01:07:34.982 --> 01:07:37.702 five years from now. I have a lot of smaller predictions. 01:07:37.902 --> 01:07:42.082 I mean, I have predictions about how the semantics of children develop. 01:07:42.282 --> 01:07:45.582 I mean, I can tell you about the order of objectives that we learn, for instance. 01:07:45.722 --> 01:07:50.662 I have predictions about what will be easy to teach to a robot and more difficult 01:07:50.662 --> 01:07:51.922 to teach in terms of verbs. 01:07:52.642 --> 01:07:54.522 Give me the most exciting one. 01:07:56.302 --> 01:08:00.482 With the highest chance of failure. And I haven't thought about formulating. 01:08:00.642 --> 01:08:06.482 Okay. I mean, yeah. Okay. The highest rate of failure is that nouns in general 01:08:06.482 --> 01:08:07.662 involve several domains. 01:08:07.902 --> 01:08:12.322 And when we talk about dogs, they have sounds and smells and sizes and shapes and whatnot. 01:08:12.802 --> 01:08:20.022 Other words, adjectives, prepositions, verbs normally refer to single domains. 01:08:20.022 --> 01:08:27.902 So, if I can validate this hypothesis, I will have a fairly strong toolbox to 01:08:27.902 --> 01:08:31.682 generate further, more detailed predictions about how language works. 01:08:31.702 --> 01:08:37.462 I'm going to have this kind of program of extending this single domain hypothesis 01:08:37.462 --> 01:08:38.822 to different word classes. 01:08:38.982 --> 01:08:40.962 That's a fairly general hypothesis. Exactly. 01:08:41.442 --> 01:08:45.742 That's a good one. And if that works out, I… You'll be very pleased. 01:08:45.802 --> 01:08:46.842 I will be very pleased, yes. 01:08:47.022 --> 01:08:50.282 Okay. Peter Gardefors, thank you very much for this conversation. Thank you very much. 01:08:52.682 --> 01:08:58.562 The CSN podcast was produced by the Convergent Science Network of Biometrics 01:08:58.562 --> 01:09:05.162 and Biohybrid Systems, a project funded by the European Sevens Research Framework Programme. 01:09:05.200 --> 01:09:33.296 Music.