[00:00:01.640] 
You're listening to Journal Entries, a podcast about philosophy and cognitive science, where researchers open up about the articles they publish. I'm Wesley Buckwalter. In this episode, lead author Jonathan Phillips talks about his paper, Knowledge Before Belief, forthcoming and Behavioral and Brain Sciences. Jonathan is an assistant professor in the program and cognitive science at Dartmouth. He's the recipient of the Stanton Prize and an APS rising star for his work in modality theory of mind, causal reasoning, moral judgment, and happiness.

[00:00:38.430] 
An important part of interacting with other people is treating them like they have a mind of their own. It doesn't sound like much, but when you stop and think about it, it's actually pretty fascinating. Researchers call our capacity to do this theory of mind. We observe how somebody speaks or acts, and then we theorize about what's going on in their heads, what beliefs, desires, emotions or feelings they must be having. That would help us make sense of all that we're seeing. An enormous amount of research in philosophy and cognitive science has narrowed in on one part of this, in particular, the capacity to represent what others believe or think about things.

[00:01:21.100] 
And because of all this focus, one might be tempted to think that this is the most basic theory of mind capacity we have. But is this really what the evidence shows? Jonathan and his co authors argue that it doesn't show that at all.

[00:01:39.740] 
This paper takes up a question, really in Theory of Mind, and it asks about which kind of theory of mind representations are most basic. And it argues that a s far as the mind goes, knowledge representations are more basic than belief representations. We try to say a little bit at the beginning about how you should understand knowledge and belief if you're going to try to figure out which one's more basic. And we point to a couple of features of knowledge that we think basically everyone agrees on.

[00:02:08.110] 
Knowledge representation should have that knowledge effective knowledge is more than a justified true belief. Knowledge has this interesting feature where you can have it can you can acknowledge Wat your knowledge how it for one way to give it is you can represent people knowing more than you know as well as knowing less than you know. And knowledge is not modality specific, so it seems to be the most general effective attitude. Okay. So then we say, okay, given that understanding of knowledge and the typical understanding of belief, which of them is more basic, and we try to lay out two views for which one you might think would be more basic.

[00:02:47.710] 
So I think there's lots of reasons I think belief would be more basic. That comes from I think both the sort of conceptual analysis project from philosophy where you believe as the most sort of basic attitudes that one could have a Representational attitude one can have. But also from a lot of the research in psychology which really kind of focused on demonstrating belief really early in life and belief attribution fairly in life and things like that. And then we give some reasons to think knowledge attribution would be more basic as well.

[00:03:12.080] 
I mean, one thing to think about is, like, knowledge attributions seem to be much more limited than belief attributions you. The set of things you can represent, people leaving is definitely going to be larger than the set of things you can represent a knowing. So maybe there's some way to thinking about knowledge attributions to the more basic processor.

[00:03:27.480] 
Okay.

[00:03:28.540] 
So then we basically asked, like, if you took a cognitive science approach to answering discussion of which of these two is more basic, what would it tell you? And we use a bunch of different tools from cognitive science to try to answer that question. So one is like, if you look at non human primates, so you kind of want to go sense for when is the point in our evolutionary past, did knowledge attributions arise in winded belief attributions arise? And you might think that's a clue to basicness would be whichever one sort of arose first in evolutionary history.

[00:03:58.400] 
And you can look at which nonhuman primate species have certain abilities as a way of estimating when that ability arose. We also ask when in untame or when a human development knowledge attributions and belief attributions come online, and that's another one. Try to get something, get a sense for basicness. We look at linguistic development in young kids. Like, when you start talking about beliefs, when you start talking about knowledge, we look at automaticity. So basically, is it more effortless to represent knowledge or belief in a human adults?

[00:04:33.090] 
And we look at cases of patient populations with the idea that, well, maybe what is going to happen is that more basic capacities are conserved even in the face of other disruption. So if you have some kind of disruption, maybe you still have the more basic ability. So maybe they could also give you a sort of clue to basicness. And then we go through each of those different cases, and we provide evidence that knowledge seems to be more basic using each of those different tools than to live up.

[00:05:00.000] 
We then turn to the question of we do a little bit of experimental philosophy asking, like, that work wasn't so much on the contest of knowledge and belief, but is the reason to think that in some possible are you really thinking about how people represent the concept of knowledge and believe that there's a picture that fits well with this approach for knowledge is more basic. We argue that there is one there. And then in the end, we kind of turn to this question of like, why would that be?

[00:05:27.580] 
Why would knowledge be more basic and belief? And the answer that we give is if you think of the features that knowledge has, you know, that there are cases where it's not just true belief. Stay. You can represent you can fail to represent someone who knowing something when they don't seem to be sort of have the belief that they have in the right it's not connected in the right way of the world. Or why did knowledge have this feature of being able to represent someone who knowing more than you?

[00:05:57.680] 
That's kind of a weird feature. So why would the representation that has these features? Why would that be a basic one? And we argue basically, in contrast to a lot of the perspectives that have historically been the case in theory mind research, that theory mind isn't probably so much. Or at least this basic form of theory mind isn't probably so much just about predicting someone else's actions, but it's more about being able to learn from other people about the actual world. And so a lot of these features, the activity of knowledge, or that you want the person's, like, sort of mental state to be connected in the right way to the world, or that you could represent others knowing more than you make a lot of sense onto that perspective where you can sort of use other people's representations of the world as a way of learning about the world itself.

[00:06:38.070] 
And that's basically where we end the history of theory of mind research. It's interesting Cognite science piece of history itself because you had it's a very real theory mind research with Pre Mac and Woodrive paper. Does the chimpanzee have a theory mind? Publishing Behavioral Brain Sciences, same paper, same place as this paper. So that BBS or Behavioral Brain Sciences has this structure where there's a target article, and then there are a bunch of responses to the article, and then the authors respond to the responses, and it's also published at once.

[00:07:08.830] 
Okay, so the original does the chip fancy you have a theory of mine? Paper is like 70s, maybe early Seventies, late Seventies. I'm not sure. 78 maybe. They said the chimpanzees do have a theory mind. Here's why they can figure out sort of the piece of the puzzle of a problem that someone else is trying to solve. So if someone else is trying to open a door, they can note that they need to give them a key. In the commentaries that paper, a bunch of philosophers actually push back against that being good evidence for a theory of mine.

[00:07:42.300] 
And the reason they push back against it was because they said chimpanzees could be using their own representation of the world, like the fact that maybe a key goes with the lock rather than attributing really representing the mind of the other person independently of their understanding of the world. So they number them like Jonathan Bennett, Jerry folder and been in position. All of the testing you to do is a false belief representation. If you want to show that they have good evidence that they have a theory of mind, they need to be so that they can represent fossil so which are sort of necessarily independent grow.

[00:08:20.840] 
Okay. But if you think about where that perspective came from. So a lot of the philosophers maybe especially denied it's a perspective on theory of mine that seems to be really focused on, like, our own way of making sense of other people's minds. So we can explain or predict their behavior, really focusing on belief representation, which has if we're really engaged in the intentional stance sort of picture, that's going to be one where the content that someone's belief can just sort of vary independently of how that person is actually connected to the world or something like that.

[00:08:53.310] 
Any part of the way in which this paper tries to push back against that whole paradigm is to think about, like, look, in our ordinary lives, like when we interact with each other, most of what we're doing is kind of like, actually just coordinating on the actual world. Very little of the way in which we make sense of other people's minds has to do with representing false beast. That's like a very special case. Most of the times in which we interact with one another, they're about like, did you also see that movie, or do you know her?

[00:09:22.120] 
Or did you hear what she said about him? Can you believe that? Or whatever all of these cases, they at least involve active mental state representation, just things where I take the world to be a certain way. And I'm asking, do you also understand that piece of the world to be that way? And so I think this paper tries to push back against the basic former theory of mind being one word is really focused on belief, which is kind of like my own special thing that I'm doing to make sense of you or get to your behavior and rather thinking about a sort of jointly coordinating on the world.

[00:09:57.980] 
The origin of this project? No, I think, like, seven or eight years ago was one of the really long term projects. But back in grad school, I was got really interested in this research showing automatic false belief representation. This is worked by Agnes Covach and colleagues, and I tried to follow up on it and replicate that work and then kind of extend it, and we ended up having a lot of trouble replicating that workers. It's not that we couldn't replicate the original effect is that we could replicate the original effect, but it didn't seem to work in any of the ways that we should if it were really a belief representation.

[00:10:34.420] 
So we ended up doing, like, 17 studies and figured out that at least largely was due to this compound and some really boring timing of an attention checks that was different between the conditions. And it was kind of this really dark time. I couldn't believe I spent that much time and effort to find out about the compound at the same time a yell, Laurie Santos was doing some interesting work on monkey theory of mine. They've been really interested in that. And Josh no was also really interested in thinking about some of that stuff is like a knowledge representation.

[00:11:09.230] 
Maybe there's parts of the kind of philosophical background and what knowledge is that maybe could capture some of what Laura was finding with monkeys and not even primates. So we decided to put together this workshop or conference on cognitive science of knowledge representation and included a bunch of people who ended up being authors on the paper. So Wesley Buck, Ulter among him. But John Terry and Ori Friedman, Alia Martin, Lori Santos, Safari, Chrisman. And the idea is to bring together a bunch of different people who had different pieces of the puzzle about how knowledge representation might work.

[00:11:45.040] 
So some people are really working on response time kind of stuff, like how quickly can someone decide whether or not or attribute knowledge to someone or attribute belief to someone? Some people are working from the developmental perspective on all these weird features of how and when kids succeed in attributing knowledge, they'd be working infant stuff like Ala Martin, and some people Woking on primate stuff like Larson's. And some people are doing Superman philosophy of knowledge representation like west and John Trey. We try to get all this perspective together in a single room and just chat.

[00:12:17.630] 
And I think one of the things we noticed was that there was kind of a striking convergence from all of these different cognitive science respectives. And that was sort of the birth room of the project, which then took about eight years to fully conceive. Doing traditionally research in cognitive science, I think, is really hard and just takes a lot of time and patience is my main experience. I think if you're looking for projects, if you're interested, doing kind of introduce where you're looking for projects. I think the main thing to start looking for is like, you know, basically arbitrage situations.

[00:13:01.080] 
There are situations where you can kind of see what insight from one field and you think, oh, that might actually really fit well with something that's going on this other field. And that's kind of the easy part. I think you can find a lot of points of that where progress has been. Two fields or three fields have been working on kind of a similar question without talking to one another, you can find discoveries from one field and offer those to another field and vice versa. And that's where you'll find really good introduction projects.

[00:13:26.250] 
The hard part is getting those people to understand what in the world the other people are saying. I think a lot of the work is basically trying to come up with a common language that people can speak to each other and have a sense of what the other people are saying so that they can figure out where they do and don't disagree. And my feeling about that is that you just have to have patience. Like, I think all of the truly in traditional projects that I've tried to work on this knowledge, one or one in a modality are thinking about how people represent possibilities.

[00:13:56.250] 
They just take, like, years and years. I mean, there's a project with an Colchester. I think we spent five years literally just trying to understand each other before we could really write a paper. But it's kind of worth it. I think in the end, too. So if you have time and patients, it'll pay off. If this paper has one key feature, just the overwhelming amount of evidence that we try to include in this paper. So everything from my math to infant studies to linguistic corpus studies to studies of four year olds to patient populations, to questions automatise with adults, experiment policy studies, and so on and so forth.

[00:14:40.890] 
So let me try to provide rehashing all of that. Let me try to provide two pieces of evidence that I think are part of that picture is that I think you're really nice. So one of them is kind of simple, and I think it's a good way to wrap your head around just the kind of APIs we're providing from human adults. So the way that this study work is research sided was Josh Non and Fiery Fishman and also been extended in a project with Brent Strickland and some folks in France.

[00:15:08.480] 
Okay, people rather than yet about an agent. And it was just like, these are pretty boring things. Like someone reading a book of where the stars are because they get a new telescope, pointing their telescope at the guy, and then looking at a star and then thinking that they're looking at a certain star. So there are some cases in which that would be true. It would be a case of typical knowledge. They correctly read the book. They just look through the telescope at the stars that the book told them would be there and how they install it.

[00:15:39.590] 
Some cases it would just stay. Would it be ignorant of what that star was, but they would just ignore the book, look through the telescope, see a star, and be like star. And then in some cases, they would have to be kind of like a false police case. So they would actually say, be looking at NEP soon or some other star and think they were looking at something else. Okay. Then after they would add one of those versions of any, we would ask them to decide whether it was true, revolve that either that person knows that they're looking at a certain start or they think they're looking at a certain start.

[00:16:13.720] 
So the question of how quickly you can decide whether or not someone knows something or whether or not someone thinks something and you think as a proxy for belief. But think is probably better if you're going to do a response in study because it's much more frequent than believe. That's just like a word. Fosters use more. The pattern that we found was both for when you're attributing knowledge correctly and denying knowledge correctly in the easiest cases here to think about are like the knowledge case in the ignorance case.

[00:16:38.800] 
So whether or not you're saying it true that they know it or is false that they know it, those are both faster then participants responses, and they're saying it's true that they think and it's false that they think in those two cases, respectively. What's interesting about that is it seems like, well, here's what they can't be doing, if that's right, they can't be deciding whether or not someone knows something by first figuring out whether or not they think something that patterns, like super inconsistent with that.

[00:17:04.350] 
So it seems like there and they have some way of determining both whether or not is true, that someone knows something and whether or not fault that know something without having processed at least linguistically, whether or not they think that thing. So it's a one piece of evidence. I think that knowledge representation seem to be sort of more basic or at least easier to compute into human adults, linguistically, totally different pieces of evidence, which is one of my favorite studies right now, was done by Laurie Santos and Daniel Farshler.

[00:17:31.960] 
And I know they have another one or two colleagues that I'm totally blanking on. And I'm sorry to those colleagues, but it published in Cognition recently, so you can look it up. Okay. So what they did was they're interested in these cases where it's basically like Laurie Montes cause is like a funky monkey condition or something, but it's basically like a getter style experiment on monkeys, and they're comparing a getter style case in a way to a typical knowledge case. So let me try to explain how they set it up.

[00:18:11.680] 
So the there's an experimenter who is watching a ball move between two boxes, and I think it's actually a piece of fruit because monkeys are typically more interested in food. But whatever says it's old women in the typical knowledge case. Well, actually, let me start with the Getty case, because the things that don't make more sense. So the Getty case, what happens is the experiment is watching the ball moves, it moves into the box, they move into the box on the right. There is an include that comes up that blocks the experiment is perspective on what's happening, but the monkey can deal with see what's going on.

[00:18:46.630] 
While the experimenters is blocked, the say lemon moves out of the box and it goes into the middle of the table and then move back into the bus. Okay. So if you think about the kind of representation monkey could be attributing, or maybe I should be attributing in that kind of case be a true belief. Maybe they don't have knowledge in the same way that some of the Gener cases seems like they kind disrupted the typical you have connection with the world, but they just have a true belief in that case.

[00:19:12.910] 
Right. The experimenter laps all the lemon in the box on the right, the limit is now in the box on the right. They have belief about the limited at right. That's true. Maybe you think they're justified and having that belief because they actually perceptually saw going to that box. But it turns out like the thing that makes it true wasn't the original evidence. Right? It was like the fact that it went back in while their view was in this sort of get your style way. Okay.

[00:19:32.970] 
They then look at where does the monkey expect the experimenter to reach in that kind of case? Do they expect them to reach in the box on the right or to reach in the box on the left? Basically, what they find is that monkeys don't have any expectation for the experimenter. In that case, they don't expect them to reach on the box of the rate, which you might think they should if they are able to represent true beliefs. So if they're able to correctly think that person believes that it's a bacon rate, then they would expect them to reach it from the proximate rate in the comparison condition, which is the standard knowledge condition.

[00:20:02.490] 
Does this really beautiful thing where nicely matched? I think so. The setup is the same. The experimenter, the lemon goes in the box on the right, an includer comes up instead of the lemon itself. Moving the box that's over the lemon moves off the table and then moves back onto it. So there's still something that the experimenter didn't have access to. There's still some change in the theme the experimenter didn't know about, but it shouldn't be relevant for their knowledge about the location of the limit.

[00:20:28.680] 
So then the includer goes down and they ask, where does the monkey expect experimenter to reach? The experimenter expects a monkey to reach in the box on the right. And so I think that's a really nicely tightly controlled. They also replicate it, which is good. It's a nice, tidy control, I think. Good solid evidence that one kid came to succeed in these cases where you could really use the knowledge representation, but they seem to fail in nearby cases where this belief would still be true. But you wouldn't actually typically attribute knowledge.

[00:20:56.400] 
And I think that's another piece of evidence that I think it's so nice. And it really illustrates that some of the features of the knowledge representation that we care about, I say not just being true believe. You seem to find evidence for that feature, even in non human primates. So not just in humans that are conceptually sophisticated. So that's evidence that not human premises are attributing knowledge. It's seems like it's also evidence that they're not in a very match case, they're not succeeding and attributing true belief, the case where they should be able to attribute to beliefs that they wanted.

[00:21:29.730] 
So then if you think about the question of basicness, the idea is that, well, look, these monkeys they're with their most common ancestor with was quite a long time ago. So if they're able to represent knowledge in the same way that we're able to represent knowledge, probably that common ancestor had a kind of something like that knowledge ability node attribution ability, they're not able to successfully represent belief in the way that we would be able to represent belief in that case. And what that suggests is that at a later time point, there's probably some is further evolved an investor of ours, but not of sake fiction monkeys that developed the ability to do belief representation.

[00:22:08.160] 
So this kind of tries to get the phylogenetic question of when did these abilities arise? And it's a piece of evidence that knowledge representations, that the capacity to represent knowledge probably arose earlier in biology than the capacity to represent belief. That's the basic inference. One of the cool features of being able to write a paper for BBS or getting in is because then you get to hear basically how the whole field thinks or think about your work, whether it's good or bad. And so we got 36 comments short, like one0 word papers on this paper, and we had to then write a response where we responded to all of them, which is challenging for sure.

[00:22:52.370] 
36 response comments A huge array of different ideas is really hard to write a response to you, but it's also fun. And I think in a way it helps. It certainly helped me at least I can't speak from the co authors, but helped me think through what was really different about this approach. We got pushed on a lot of different parts of the proposal. So I think one general objection that people had was that this picture where knowledge is, you know, thinking about someone's relation to the actual world, it kind of has its factor thing isn't really a theory of mind attribute representation in the first place.

[00:23:32.010] 
Like, some people are really committed to this idea. That for to be a theory of my representation, it had to be a representation of someone's understanding completely independent of your own understanding of the world, or had to be meta Representational in the same way. Belief is another set of, you know, arguments, or basically, I think they're trying to find, like, aren't there other ways or other ways to explain the kind of basic representation? Basically. So maybe it could be something as simple as, like, perceptual access?

[00:24:06.000] 
Isn't that all you're finding is perceptual access is simpler? The fair the belief representation is that really surprising. Their way of pushing back against, like, really be a knowledge representation. Another one was like, maybe what they have the ability to do is represent kind of acknowledge how or something like that, but not really propositional knowledge representation. So I kind of collectively think of those is thinking about the restrictions on the basic form of mind coming down to the restrictions about the content you can represent not only a visual perspective or only certain kinds of acknowledge.

[00:24:42.690] 
How not that it's really a difference between knowledge and belief. It really boils down to content rather than attitudes. A third group of people basically objected this picture where theory of mind really isn't about predicting and explain a behavior, but it's about trying to coordinate on the actual world or learn from others about the actual world. And they just let that the wrong way to think about a theory of mind. And then you know nicely, there were some people who love the idea. I love the basic picture and then wanted to extend it in a bunch of different ways.

[00:25:15.410] 
So there were ten or 15 papers, I think pointed out really beautiful ways of trying to take this idea and then run with it in a bunch of different ways and ask interesting no questions. I think one place that people who definitely push back was we paint this picture on which belief has really been the central focus of theory of mind research. And maybe people are committed to live being basic. I think fairly from this, like certain parts of psychology that they wanted to object to that.

[00:25:45.090] 
And they wanted to say that no one was committed to belief being basic. We basically thought of belief as a good way, sort of the right litmus test for a theory of mind and be that like we have been, in fact, studying knowledge. And I think we try to point out in the paper that the clearly he has been good work on knowledge and factor representations. We're not saying that there was no work on that, but I think we wanted to say if you think about comparatively how much work has been done on belief representation versus how much has been done on knowledge representation, it's Willy out of proportion.

[00:26:19.860] 
I think, especially if you think about the role of those two things probably play in most of our ordinary theory of mind won't be interact with other people. So I remember doing a Google Scholar search, which I think has been a footnote in the paper very early on. And it was like, if you do something like false belief test or falsely task, where it was knowledge task or knowledge task or ignorance task or ignorance tasks to incent is I had really been focused on believe falsely.

[00:26:43.440] 
In particular. It was like an order of magnitude war. It was like maybe eight0 citations, four things and talk about knowledge tests or ignorance tasks or tasks or whatever. And then like 90,000 papers or something that talked about it was like something of that magnitude. It was crazy. And I think that's out of proportion. And I think basically I think what I want to say is not. I think I think psychologist weren't certainly at least a decent portion of developmental psychologists weren't committed to this conceptual claim that belief is supposed to be more basic than knowledge.

[00:27:17.580] 
There are some who really want to push this kind of, like, like core theory of mind ability that emerges like super early in development. Like young infants, like eight months old, will have an ability to do fossil representation. And if you think that belief representation is part of that core concept, then probably you are committed to believe being a very basic, at least as basic as knowledge, not less basic. But in that small group of people outside, I think most people probably weren't committed to the idea that I believe really was basic.

[00:27:46.080] 
That was just what they thought was the right way to test it. Like, part of the consequence, though, of thinking that the belief was the right way to go for testing. Gary mind was that we know a huge amount of belief, and we actually have a relatively poor understanding of how knowledge Attribution works. It's like we have no understanding. People have done really good work on it. Like, Paul Harris is someone who's done fantastic work on Nodge knowledge attributions and young children, how that develops linguistically, all these evidence.

[00:28:11.860] 
And there's all this beautiful work on this literature called Trust and Testimony, where young children are basically trusting what other people say, depending on whether or not they represent them as knowledgable or not. Truly fantastic work. We try to draw a lot of that to use that in our paper to argue that that's kind of how knowledge works. But aside from that and some other people worked on knowledge, we don't really know that much about it. So I think a lot of the comments in the paper ended up being like, okay, wait, exactly how do you think ignorance representations work or something like that?

[00:28:42.900] 
Exactly how what's the format of the reference this knowledge representation? And I think if paper is going to serve any purpose, I think the hope is that it'll kind of push people to really begin focusing on the processes involved in knowledge attribution. And then maybe we can have a much better understanding of knowledge tribution in the same way that we have a good understanding of belief attribution whether or not knowledge is like the most basic theory mind, our capacity to be knowledge is the most basic theory.

[00:29:12.250] 
My capacity we have, or whether or not it's just more basic than belief and somewhat basic. That's a good question. I think I'm not committed to knowledge being some held the most basic theory of mind representation that we have. I mean, I think if you look across the things that visual perspective taking is super interesting one to take a sing representation, I think this is a good chance that that's certainly as basic as knowledge representation may be more basic. Young children seem to start using the word C or words that her perceptual kind of access things before they start using the word no, at least successfully using the word no.

[00:29:52.200] 
I think there's probably an argument to made that there are even more basic to your my representation. I think the data will have to barrel whether or not that's the case, but I will say a feature that is really important of all the all the basic ones is that they're factive, like every single attitude that comes on early as effective attitude. And I think that even if it turns out knowledge in the form that we testified isn't the most basic, I think it points to a different way of thinking about theory of my attitudes, that we should really be studying the fact of ones and how those develop.

[00:30:27.680] 
And those probably are the most basic theory of mind representation. Paper ends with a call to Arms to Strong section, header title. But basically the idea that people have been studying belief a ton maybe we think a bit too much. And the idea would be to start studying knowledge in that same way to really understand how knowledge work. So it's sort of a call to Arms for the study of knowledge rather than Bull ship focus in theory, mind research. I mean, some of the areas that I'd be really excited about going or, like figuring out the signature limits of knowledge.

[00:31:08.630] 
That's like one thing that I think I Super interesting. We point to these some of the limits that we think these features, not representation, should have. They should be limited to things that you don't think are inconsistent with your model of the world, like the spec Tivity constraint, or, like, you know, that you should be able to represent other people's knowing more than you. But like that ability to represent other people knowing more than you. Like, we don't really understand how that's done very well. We don't understand what's actually the base that even Western Jennifer Nagel and Dire having right now.

[00:31:40.210] 
Like, what's the kind of content you're attributing to someone and you represent them as knowing something you don't know? Do you represent them as knowing if that possible answers? Is there some kind of different attitude, like questioning attitude or just some more basic tagging connection that you have? You represent them as having to the actual world, even if you don't know what it is. I think there's all these really cool questions about if knowledge representation works this way, and there's so many avenues to pursue. What are the limits of it?

[00:32:09.770] 
What's the format of it? How do you do these weird things where it doesn't seem like clear what the content of the representation is supposed to be? Yeah, it'd be really fun to see research, try and spell out up the details for that stuff.

[00:32:25.190] 
That's it for today's episode. Funding, in part, was provided by the Institute for Philosophy and Public Policy at George Mason University. Visit our website at Journal Entries Fireside FM for more information about Jonathan Phillips, his work, and some of the resources mentioned on Today's episode.