Hugo Spiers is professor of cognitive neuroscience at University College London. His research explores how our brain constructs representations of the world and uses them to recall the past, navigate the present and imagine the future. In this episode, we talk about his work on Sea Hero Quest (with Michael Hornberger, former guest of this podcast), his new research project Taxi Brains, the difficulties and joys of working with more than one species, and cognitive maps in humans.
0:00:05: Dealing with email
0:04:42: Sea Hero Quest
0:25:53: Taxi Brains project
0:55:18: The difficulties and benefits of working with humans and non-human animals in the same lab
1:11:48: Discussing Hugo's review "The cognitive map in humans: spatial navigation and beyond"
Episodes mentioned during our conversation:
Michael Hornberger: https://geni.us/bjks-hornberger
Kate Jeffery: https://geni.us/bjks-jeffery
Bellmund, Gärdenfors, Moser, & Doeller (2018). Navigating cognition: Spatial codes for human thinking. Science.
Constantinescu, O’Reilly, & Behrens (2016). Organizing conceptual knowledge in humans with a gridlike code. Science.
Doeller, Barry, & Burgess (2010). Evidence for grid cells in a human memory network. Nature.
Epstein, Patai, Julian, & Spiers (2017). The cognitive map in humans: spatial navigation and beyond. Nature neuroscience.
Gardenfors (2004). Conceptual spaces: The geometry of thought. MIT press.
Gardner, Hermansen, Pachitariu, Burak, Baas, Dunn, ... & Moser (2022). Toroidal topology of population activity in grid cells. Nature.
Griesbauer, Manley, Wiener, & Spiers (2022). London taxi drivers: A review of neurocognitive studies and an exploration of how they build their cognitive map of London. Hippocampus.
Jacobs, Weidemann, ... & Kahana (2013). Direct recordings of grid-like neuronal activity in human spatial navigation. Nature neuroscience.
Lever, ... & Burgess (2009). Boundary vector cells in the subiculum of the hippocampal formation. Journal of Neuroscience.
Maguire, ... & Frith (2000). Navigation-related structural change in the hippocampi of taxi drivers. Proceedings of the National Academy of Sciences.
Newport (2021). A World Without Email: Find Focus and Transform the Way You Work Forever. Penguin UK.
O'keefe, & Nadel (1978). The hippocampus as a cognitive map. Oxford university press.
Solomon, Lega, Sperling, & Kahana (2019). Hippocampal theta codes for distances in semantic and temporal spaces. Proceedings of the National Academy of Sciences.
Solstad, Boccara, Kropff, Moser, & Moser. (2008). Representation of geometric borders in the entorhinal cortex. Science.
Spiers (2020). The hippocampal cognitive map: one space or many? Trends in Cognitive Sciences.
Tolman (1948). Cognitive maps in rats and men. Psychological review.
(This is an automated transcript. It's been lightly edited to include the key terms for search engine visibility, but there are many errors in here)
Benjamin James Kuper-Smith: [00:00:00] Actually, I wanted to start with something that, , as a slightly random topic, which is, , so after I talked to Michael Hornberger on this podcast he asked me basically whether he should set me in contact with other people, introduce me to other people who might be wanting to be guests.
And, I said I wanted to invite you next. And he said, oh yeah, you can, you can contact Hugo. I don't need to introduce you though. Just send him an email and like within like four hours, he'll say, yes, he's super good with email. And I thought, okay, let's, let's try this out.
So I sent you an email at, I think like 11:00 AM or something went for lunch still. Hadn't received an email, I thought. Okay, let's see. And then within, you know, just before the four hour mark, you responded to my email. So it was like, okay. Michael knows you well. Uh, I'm just curious, how do you, yeah, I think, I mean, we've.
So basically mail back and forth a bit because we changed the recording date and that kind of stuff. And you've always been very quick. Um, uh, how do you do that? Because from what I've heard, professors [00:01:00] get lots and lots of email, um, had lots of stuff to do, and most people are a lot, lot slower. So how do you do that?
Hugo Spiers: Yeah you just get into the practice of literally having to just hammer them out quickly because you don't have that time. And there's that classic sort of, um, you know, idea that professors typically the respond immediately or a year later as one of the
two, but it's, and that's what I find it's really tricky that you'll have a bit of a dialogue maybe and someone asks them, maybe takes a bit more thought and you never get that followup email quickly back.
So managing dialogue with people is a real trick, I think. Um, uh, but yeah,
I, I do spend a fair bit of my time. Uh, when I look at what is it I do as a professor, it is
a lot of emailing because I'm providing feedback, I'm commenting, I'm pushing things forward, I'm requesting, or often responding to people about things.
Um, and it's, you know, people sort of, sort of go, oh, Michael, actually saying, you know, what did you do? [00:02:00] He's like discussed, you know, what did you do today? Did you get work done or did you just email all day? And it's that kind of feeling that it's quite nice when you actually are not emailing back and forwards, but yeah.
so much of it happens.
Benjamin James Kuper-Smith: How'd you then manage that? Oh, like, do you mind that? Um, oh, is it more.
Hugo Spiers: it's frustrating at times, definitely, uh, managing it cause you know, Yeah, You think, I, gosh, I really better reply to these set ofemails, but I don't actually physically have the time today. So what I do is flag things. I, I had to keep a zero. I don't have any sitting in, I just flag things I've got to respond to and then go back through all the flagged email as soon as I can.
Um, but yeah, there's, there's a fair few. I don't respond to because absolutely I don't need to. Um, but yeah, it's hard. It's a general challenge. I I'm sure other people have way better. Some people are extremely efficient. I'm not one of these
Uber efficient people. Um, but, uh,
Benjamin James Kuper-Smith: put in the time doing it. [00:03:00] Okay. So yeah I don't know whether you've heard of Cal Newport, he's hada few books. He's a professor of computer science, but, uh, At Georgetown university, but he is, I think publicly much more known, a lot more known fairs, like productivity books.
And because he, I think basically he wrote a book called a world without email. I think he doesn't have any social media and all that kind of stuff. And, um, but as like a computer scientist, who's working on this stuff and, um, I don't know, like he suggests like, you know, using slack or whatever as an alternative, I've never actually used that.
So I don't know. I think I'm not really in the demographic that has email problems yet as a PhD student. But yeah, I was just curious how you kind of, um, okay, so it is just a bit of a slog
Hugo Spiers: I, I there's a slog and I think it's the flagging a capacity to flag emails and go back to those ones that are flights. I would never work. That's the only way I definitely find that other people have other strategies I'm aware, but for [00:04:00] me, that works. Um, but yeah, it's a lot of
emails. I mean, yeah. Lots of requests for things, you know, lots of all sorts of things that you, you have to quickly make a quick decision on what you're going to respond to.
Um, but yeah, it can just kill off getting really big, important things that need to be done can get delayed because you're just busy sifting emails. Uh, so it's just tricky. There's no doubt about it.
Benjamin James Kuper-Smith: Yep. Okay. I guess the main things we'll be talking about today are , the taxi brain projects, which is, , from what I understand a new project, just kind of starting now wish or have started recently. Um, and then you wrote a review paper in nature, neuroscience about spatial navigationin humans and in non-human animals and, um, kind of how you go from .
Space to abstract maps. Um, uh, before we'd do that, I thought we could maybe, you know, um, go slightly back to the conversation I had with Michael, because of course he mentioned [00:05:00] you a few times in the conversation because I guess you and him together you ran the Sea Heroquest game, which is this, , very successful mobile game that, uh, I guess lets you investigate, spatial navigation and memory, , is maybe that's the way to put it.
So the kind of question I like to start with here is that, uh, one thing Michael mentioned a few times is that you really had difficulties kind of getting acceptance about the project and the data that you got and that a lot of people criticize it as saying, well, this is just a game. You know, real science or whatever.
Um, you know, it's, it's a bit of a gimmick or whatever. And to some extent, that, to me seems like fairly lazy criticism and, um, I don't know what the specific criticism was, but that kind of idea that it's just a game to me, seems like just dismissing it without really thinking about it too much. Uh, but it seems to me that, of course there is a nugget of truth in that.
And you know, you do have drawbacks with this [00:06:00] kind of data with, you know, I mean, um, Michael described a lot of how you, how much time went into making it fun, you know, adding things, subtracting things that just, you would not do an experiment. So, you know, the obvious drawbacks, I'm just curious kind of what are some genuine criticisms of that, or maybe even a different way of phrasing it.
What are kind of your main problems when working with the data.
Hugo Spiers: Yeah, it's a great question about what, what are the sort of limitations and, and challenges and problems that you encounter. And , that project is as discussed in the previous podcast of Michael, just sprang out of an incredible perk in a phone call that came in to Michael and he phoned me.
Um, so it's really unusual. Um, the
Benjamin James Kuper-Smith: the way. Did you believe him? Like, what was your reaction when he said like, oh yeah, I got this crazy phone call, but did you go like, oh yeah, that makes sense.
Hugo Spiers: No, I believe Michael, I mean, in my career as well, I got an email once this sounded like some sort of African prince [00:07:00] offering me money and I ignored it. And then a week later I got a follow up email to say, um, I don't know if you received our email, but you're now only have two weeks left to apply for the $600,000. We'd like you to apply for, uh, are you, are you at all interested? And, and it was completely genuine. I then wrote an application and got $600,000 sent to me. Um, so I'd been used to this weird way in life. You can get funding. That's just, um, very odd. Uh, so, So this wasn't a surprise. Um,
Benjamin James Kuper-Smith: So just briefly that's I've I, to be fair, I'd never really heard of people getting for me, like from what I've heard, getting ground money is very difficult and basically a lottery. And, um, you seem to have two examples unless the opposite seems to be the case. Why? Like, why do you think you were contacted there or Michael in the other instance,
Hugo Spiers: They're both sort of their profile. I don't mean that the $600,000 was the James McDonald foundation where they had a process of selecting, who they wanted [00:08:00] to get to apply. So you had to be selected and then written to, um, and then you had like a 50, 50% chance I think, of obtaining the funding. So it was a very odd process and they've changed that I suspect.
and you're right, these aren't, these aren't normal, vast majority of grants awarded, uh, through, uh, government bodies or large scale charities, like the welcome trust, highly regulated and very organized. And, um, you know, I have received those as well. It hadn't run an entire career on crazy phone calls. but possibly the weirdest one was once getting contacted by somebody who said, oh, I've applied for a grant. The funding body said, um, they don't like my collaborator. My co-PI. Uh, they gave me a list of people they might think would be suitable. When you were at the top of the list, would you like to be the co-PI?
I said, yes. He wrote, they wrote back the next day he goes, oh, we've got funded. You're
now a co BI on the ground. And, uh, he sent me the grant to look at, but it says, but this was an arts grant. So it was a very small budget and
not a [00:09:00] huge project to worry about. But again, the sort of ludicrous level of I've just said yes to a phone call and now you're a PI on a project, so things can happen like that.
So when Michael phoned , wasn't the most strange thing, but, uh, The way it panned out as, as, as, as Michael's described in your podcast. And as things move forward with that, we suddenly realized this is really big. This is very serious, and it couldn't believe the sort of entourage of people involved in, in the process.
Um, so yeah, I, going back to, you said your, your question was, you know, this is a course amazing that we've been able to test 4 million people worldwide and see cognition across the planet and, and dive into questions with unprecedented, you know, fidelity. Um, these are, it has been incredible. That's the positive sales side and we've got lots of work to that's coming out that we'll be publishing soon.
Um, the down sides are, um, you know, [00:10:00] it, it is, it is, it's a game. So it, we're trying to understand how cognition works, how people behave. And we assume that this taps into the real world, how to actually people actually navigate. Um, and the question sort of, you know, is a video game, factually valid, and we've done a fair bit of work to suggest that it is in many cases.
So, if you're bad at Sea Hero Quest, it will predict you being bad in the real world, we've done research to show that but if someone's really poor at video games, that validity starts to break down. So if you find it hard to control a boat in a virtual game, You know, whereas you can walk fine round a neighborhood, then you do have this obstacle in terms of that game help you predict.
And that's true of a lot of, a lot of different tests, you know, in neuro-psychology, if you're using it for clinical things, in many cases, people have to go to see, like, if you're using a Raven's matrices and example, you have to
see it. did you have you're blind? You can't do that task. Um, so, so the game has this sort of limitations as well, but [00:11:00] it's not it's nothing's ever perfect.
Um, but know the most frustrating thing, probably I would say about a project on that scale, uh, was that for sort of ethical. Positive reasons of like data security and the way it ran, you know, we couldn't identify and it was, you couldn't have any link to someone beyond their phone ID and it wasn't, it's not a code.
It could, you could trace back to them. It's scrambled by the company. So that, that was a good feature. And we were criticized in the guardian newspaper by collecting data and millions of people. And this is really bad, but actually it was the most data secure, uh, you know, uh, anonymous project I've ever worked on.
So it's really extreme, downside to that is you, you can't say much about anybody. They just, you have rely on their, their input within the video game. And that, that for us, if you look at a project on the scale where you've tested 4 million people, and you'd just like to know things like, um, you know, um, what year did they finish [00:12:00] their education or.
how much exercise they get a day, what's their ambitions in life. There's loads of things you'd be really interested in,
but you can't ask. You can ask them these questions, but in all other research projects, it's quite easy to just put a questionnaire in there.
Benjamin James Kuper-Smith: Sorryjust briefly, I mean, you do have information about what age, gender and that kind of stuff. So you do have some information, right? As I said, I actually played the game. I think, so my data is probably in there somewhere. I can't remember though. Was it just at the beginning?
You put in your age and gender , that's it, or,
Hugo Spiers: yeah.
So throughout the game, it kind of, again, these sort of, part of the careful design where it would get you playing it a bit and quite soon into the games he'll thank you for helping out dementia research. If you
can you help us more? The messaging was very careful by like, you know, you win a badge.
If you, if you do this now, uh, for your boat. And, um, Yeah.
we, we had nine demographic questions that were really fought over and they weren't all the top of the list because what we realized is we needed questions that would translate into every one of the 17 languages in the game. And that would not be confused by [00:13:00] people quickly on a phone.
So, if you want one question we'd love
to know, right. Is what, what, what, pro so the questions that we got, we decided to go for where, you know, what hand do you write with what are you left or right-handed because you can have a kind of symbol for that on the screen. And it's very clear what you're asking.
Um, but as you say, the most, the most key questions we looked at in the first piece of analysis was, uh, are you male or female? How old are you in which country. And, um, that country question unlocks a lot of information because if you know which country people are from, you can link them the database to GDP and a whole host of other features about different countries.
So that was the most important question we asked, I think in the game was which country are you from? And it was quite successful cause that one was like a passport entry. So you might, for all sorts of online documents, you're used to getting and going on, I select my country. So it just looked like that.
And I think that, that, again, for that reason worked really well, um, but so we do have fantastic data, but I just think, I wish [00:14:00] there'd been a way and we fought and fought as a way to try and get round this as a scientist to see if we could, could we link this to a questionnaire and in the time scales, as Michael described it as extremely like neck break speed, um, we couldn't, we couldn't, um, get there and, um, That's one of the limitations of the work.
Uh, and it's something we've been following up. We have now tested well over 800 people where we can know an enormous amount on these people and Sea Hero Quest,
Benjamin James Kuper-Smith: Like outside of the, in the lab basically.
Hugo Spiers: in the lab, well online, but via portal where you can, you can get consent. Um, but, uh, you know, as Michael mentioned in the previous podcast, you ran, we've now redeveloped, Sea Hero Quest.
So it's on the app store again. Um, but now you, if you want to play it, you need a code, uh, that's given to you by a scientist. And so we're
hoping to set up lots and lots of projects that are linked to projects. So soon we hope to have a campaign where people can just go get a code and go and help out, and then they can fill out a [00:15:00] questionnaire.
They could do all sorts of things. He could,
you know, so yeah,
Benjamin James Kuper-Smith: so like what, what is the, why couldn't you say, you know, for, let's say people already played the game for two hours or something. What, what prevents you from just adding a questionnaire there that and saying like, you know, an optional questionnaire, is it just that you think people are going to stop playing the game or because it seems like, you know, that's definitely technically possible.
Hugo Spiers: Yeah, that's right. It wasn't a part of the games design. I think it's more to do with limitations on the actual spots, the game, like, so it might've been something. We could do again, this goes back to the discussions we had, but then you're right. That it was a frustrating aspect of, there was a kind of, they weren't keen to just add extra questions at the end to keep us small.
And for all these reasons that ended up
being in that format. Um, But.
uh, yeah, as I say that you, the new version is on the app store, Sea Hero Quest research, um, is there to do that and [00:16:00]
anybody could use it. So this could be undergraduates in a department. They could go to their departmental supervisor And say, I've got access to this.
Would you could you supervise me to run a project? It doesn't have to be, uh, some fancy, impressive team. Uh, but of course we do have those, you know, researchers are in different countries who are, who are keen to follow up, but, um, yeah,
we really hope it gets used more as a tool, I could, you know, it's 10 minute tasks you can add.
So you could spend, as you said like two hours playing it, or you can do 10 minutes, 10 minutes is sufficient to give us really good data, really with that.
Benjamin James Kuper-Smith: And you said you need to, now, when you download it you need to code from an experimenter, um, is that then something that experiments create themselves or do they have to go through you
Hugo Spiers: I guess through a platform website. So they
go through, uh, they fill out what they want to do. It gets approved. and, then they get sent to a whole lot of codes that they then distributed to people. And when those
codes have been used, they get an email from this is Alzheimer's research UK, the charity that had very kindly supported this.
Um, they they've [00:17:00] organized the websites. And, um, so the data would come back to the researcher anonymously. Of course. Yeah.
Benjamin James Kuper-Smith: Okay. So can you say something, uh, I mean, with, with, uh, with Mike, I think we may need to talk about the development of game. of the initial papers that have come out that validated the quality of the dataset. Um, is there anything you can say about kind of the, the, so to speak on quote, quote, unquote, actual science kind of that's coming out soon, uh, like the exciting stuff that, um, Michael alluded to, or
Hugo Spiers: Yeah, I can, I, I can't tell you about some of the results. So some of it is, um, yeah, some of it's still at a situation where I,
Benjamin James Kuper-Smith: Or maybe you have questions
Hugo Spiers: review, but what I can, what I can point out. Yeah. So what I can point out things that are, I can put the two and two together, this, in the public domain, uh, in the sense of what's sitting there, anyone who's looked at it could, could figure out.
So, um, you know, this, this is a case requiring salami slicing. If we just took the entire [00:18:00] data and ran all the, you know, you would end up with a paper that has got 92 figures in notes or something. And, um, so we've really tried to figure out, uh, so we have papers we're writing up to this is, this is, this is not been put out before, but this is where we are Um, so the first time we've ever really talking about this, um, We have our paper, that's been under review and accepted on the impact of cities on, on people's navigation skill that's been discussed before. And, um, I don't know if Michael mentioned it in the podcast, but in that case, we were able to show that across the
set of countries, 38 countries, countries had really gritty cities.
If people grew up
in cities, in those countries, they were worse at navigating in the game. And specifically the more complicated and curvy the game levels in Sea Hero Quest, the worst those participants got. So there was a kind of sense of your experiences growing up in an environment shape how you then navigate, uh, [00:19:00] which I think is a really fundamentally interesting point.
The reason we were so excited about it is that nobody had really looked at it like this. Uh, as this is a core question, how does my development affect my cognition later on? And this for us as a really clear, clear result that because you've got, you know, last a half a million people's data to look over , then you can see these things.
And in our paper, one of the exciting things we've done for me, at least in terms of good science is to go and then replicate this effect. So we've tested, you know, all these 4 million people down sample that to really clear clean participants who got a half a million. And then, then you can then go and test 800 people, tiny, tiny fraction,
new, a new set, and find exactly the same effect size
in this new data. So we've, we've extended that in all sorts of different ways and that that should be coming out soon. Um, so readers will be embargoed by the journal. But that that's going to [00:20:00] be, you know, extend the story when the paper comes out much more substantially than I can describe here. Um, but that's just one of the, one of the nine questions.
The other bits we've been looking at, we have a paper we're about to submit on sleep. So how does
the hours of sleep? You get relate to performance. And again, more importantly for us, just, just generally, how is sleep distributed worldwide. If you had to guess which countries sleep the most in which sleep
the least, would you be able to predict that,
Right. And, uh, is there a cultural clustering, you know, do certain countries
sleep similar lengths? You know? Um, so, and that's an interesting question where you have far-flung countries, where are they? You know, Western offshoots like Australia and New Zealand are fairly far geographically away from the UK, but maybe they've got a similar
sleep. You know, it'd be interesting to look at, but we see all sorts of fascinating, really fascinating patterns and sleep
aren't even related to a video game. It's just that it's a huge [00:21:00] sample of
just under a million people where we've got really good clean data. We can extract and discuss the lifespan effects of .
how sleep changes over the lifespan.
Benjamin James Kuper-Smith: that's an additional question you asked or like
Hugo Spiers: Yeah. Well, we asked.
Benjamin James Kuper-Smith: sleep or.
Hugo Spiers: Yeah. how many hours of sleep do you get on average? Yeah. So, and again, we've done some more followup too. There's a lot of more precisely based questions you want to know about people, but, and of course, you know, we don't know if this is the actual time someone says they sleep seven hours.
Maybe they only sleep six.
Uh, we don't know. Uh, but there are other studies showing very strong correlations between self-reported sleep and actual sleep on a population level. And this isn't this, you know, this is hundreds of thousands, but that's just two examples. Other examples we've been looking at is how good you think you are navigating compared to how you actually are.
And then the interesting question with that of course is again, world population, which countries are on the money, like which populations are. They get it right. You know, at a world average, they think they're not very [00:22:00] good and they aren't well, they think they're very, very good and they are, you know, where there are other countries that think they're really good, but not so, uh, and the neat thing there is, we've been able to predict using other worldwide metrics, you know, in the interview, large scale populations, about properties of what might, uh, you know, relate to this where you have been able to predict which countries overestimate their, their skill.
.Um, and so that's something we're, we're looking to submit very soon. Um, and then handedness, you know, as long story are left and right-handers different, uh, and here we've got, you know, 700,000, you know, people with different handedness scores and normally it's quite hard to find left. But here we've got a lot.
So we've been asking the old classic question of does it fit navigation skill? Um, and again, I won't tell the answer because
that's what the reading and the, the article is, but those are just a few, but there, there are other papers as well. Um, you know, linked to some of these questions, um, and [00:23:00] more mathematical analysis, everything we're publishing now is just, oh, how long did what's the distance people traveled in the game?
So someone who's very good will travel a short distance, but someone who, um, travels a long distance is bad. But what if we look at your, how wiggly you are, does that tell us something about your, your, um, your cognition? And so we've been teasing apart, these sort of patterns in the data.
Benjamin James Kuper-Smith: Yeah.
it's interesting that you're, um, getting also a lot of data. It seems on questions that don't seem like immediately related to what you actually set out to do, like the sleep questions and that kind of stuff. I mean, is that something you thought about bef for, or was this just a surprise because you had suddenly such a large sample that you went, oh wait, we can actually do kind of basic descriptions of some of the other things that haven't already out there so far.
Hugo Spiers: Yeah, we knew that as well. Now, when we set out, we were, the motivation for the questions was, was along the lines of what if we were to develop a tool for, uh, you know, [00:24:00] monitoring dementia, Alzheimer's dementia, or potentially diagnostic, Um,
tool. Um, then what would you want to know at the sole age and gender are really key.
We, our data shows very clearly. You need to know, uh, if you want to, if someone's navigating what their skill is, is really quite useful to know age and gender. But beyond that, what are the other questions that would be helpful in filtering with more precision, the diagnostic or the monitoring? And so those were all picked along those lines where sleep is a known factor for potentially memory and, uh, and dementia.
But as soon as we got these questions and it was very clear, there were going to give us some quite important results. And Michael described how we thought we might get a hundred thousand people after a year and within two days we'd had it. So I think it was quite clear at that point That we could start to ask questions as a scientist in terms of your podcasts and kind of thinking about scientific process and people's careers, and
you raise a good point in [00:25:00] a sense.
With a project that you start to spiral into areas you don't know enough about. Um, so we're looking at education and sleep and you know, so, so there , it's been absolutely fantastic to collaborate with experts, to get their input, to research articles, um, where they can give us their expert opinion and help draft the manuscript and process that I wouldn't normally do.
And normally be in my silo
working away. And I don't need someone else really, but I I've loved that. It's been a really fantastic part of science.
Benjamin James Kuper-Smith: That must be a similar experience that those collaborators get right. When they get an email from you saying, Hey, we've got data on a millions people sleep. Do you want to collaborate with us? Okay. Sure,
Hugo Spiers: Funnily enough, most of them say yes. Yeah,
Benjamin James Kuper-Smith: exactly. Okay. Um, is there anything else you want to add to Sea Hero quest? Otherwise I'd move on because we were about quite a while now.
Hugo Spiers: No, that's, that's covered most of the
Benjamin James Kuper-Smith: yeah. Okay. Uh, so I don't know how much you can say about the taxi brain projects, because I guess it's a pretty current or even future [00:26:00] project, um, even more than Sea Hero Quest to some extent.
but I wanted to talk a bit about it at least briefly. I watched parts of, uh, what was it? UCL mind's lunch hour lecture. That's on YouTube from 2013 that you gave. So I thought related to that, I'd asked something you explained there, which is what was the great exhibition.
In London and 1851. And how does that relate to the kind of science that, um, you and other people are doing today?
Hugo Spiers: That's a great question. And so that's a nice sort of curve ball one, because I'm not an expert on that historical event. What I know and
people could just look up on Wikipedia and probably find rapidly I'm wrong. But my understanding was that the United Kingdom decided, um, in London, they would have a great exhibition, like a large salary, one of the greatest advances in the age of arts and sciences and show off all these amazing things that have been created in the UK and get the world to visit London and see all this and show off.
And obviously [00:27:00] to increase wealth,
you know, that was there's ever these things, it's that economic capitalist approach to that. And what's amazing is if you look at the historical pictures there's a, you know, they built in a huge glass house in Hyde park. It's huge and entire building to hold this and then took it all down again.
And so the reason that's important for the story of cognitive science and indeed the hippocampus as a brain area is that in 1853 or 54, when this occurred people came to see it, of course, from all around the world. But anyone could just, at that point, be a taxi driver, just turn up and get a license and drive people across London and charge them whatever they wanted, as far as I know.
It may well have been regulated to some extent, and you can just be a taxi driver, but there was no tests. And
so there was huge embarrassment worldwide when people were lost all over London that rich people coming to see, it ended up in the wrong bits of London. So there was a stamp down to say, we can't have this, we've got to Institute some, some tests for London.
And so they tell [00:28:00] they develop, what's known as the knowledge of London, the knowledge it's just referred to, um, where if you wanted to be a taxi driver from that point forward, you needed to prove you knew enough about the streets of London. And it got around that. And this is what 18 in the mid 18, 18 40.
So. You know, in 2000. So in millennium later, Eleanor Maguire, Irish neuroscientist at university college London decided to, to have a look at the structural size of a group of these London taxi drivers brains, and compare them to non taxi drivers to see if there are any differences. And the reason Eleanor was interested in that is that had been a number of reports in different animal species, suggesting that there was clear variation in the size of this bit of the brain, the hippocampus with spatial ability to like even sort of seasonal changes when the spatial demand on retrieving a story nuts versus other periods.
So it's like across many different species. Do human show something like this was the question. And indeed she showed there [00:29:00] was a difference their posterior hippocampus in London, licensed taxi drivers is larger, uh, the non taxi drivers. And in fact, their anterior hippocampus shrinks. It's not, they have a bigger hippocampus.
That was what the S the sun newspaper ran a cartoon with this picture of a giant brain, uh, when it was discovered, uh, it actually is sort of like a redistribution along the axis from being increased at the posterior part and, and shrunken and the anterior. And it's quite a big discovery because it's a lot of textbooks and it's had a lot of impact on people's
thinking about brain.
It's sort of an extra pillar under what is all the evidence that the hippocampus has anything to do with memory and space. This is one of the pillars that sort of held up to say, not only X, Y, Z but also look at London taxi drivers.
So, so you asked.
Benjamin James Kuper-Smith: bachelor's in, I saw some 2010 and then there was, I think in like first year, you know, memory and cognition lecture, you learn about HM. And point then later you learn aboutEleanor Maguire's taxi [00:30:00] study. Maybe this was because we went London also. So,
Hugo Spiers: Yeah.
Benjamin James Kuper-Smith: slightly biased, but yeah.
Hugo Spiers: that's right. So H M of course is famous amnesic who had a lot more than this hippocampus removed surgically, but the key key area was the hippocampus. Um, you know, it was completely lost and, it's one of the key pillars held up. Um, so, so going back to your question
Benjamin James Kuper-Smith: I'm just.
Hugo Spiers: or sure.
Benjamin James Kuper-Smith: Uh, I mean, just Chris, like, uh, how, uh, you, um, because I mean, you've been for most of your life as your career at UCL but that was still before your time at UCL, right? This is
Hugo Spiers: I.
Benjamin James Kuper-Smith: the initial study.
Hugo Spiers: Yes. I just started my PhD with Neil Burgess at the time. So I remember exactly where and when Eleanor Maguire told me, uh, that she discovered this fact,
um, in a stairwell inside the Institute of Neurology , I remember her turning to me and saying, oh, you know what I've discovered, and
this is mind blowing.
Um, so yeah, I wasn't, I wasn't a fully fledged doctor by that point, but, um, yeah, it was, it was quite quiet discovery.
Yeah. So, so going back to your question about the [00:31:00] taxi brain's project, we've set up is really following on a, you know, 20, 22, but the time has passed. Um, but back around after Eleanor Maguire discovered that her research team and I was part of it, did a lot of work studying these taxi drivers.
They're absolutely fascinating group of interviews. To study. Um, so it's just a bit of background for them as well. It's worth going back to the, a great exhibition in London that they started this, but today, today, right? While we're podcasting now, while I'm interviewing with you, there's span to be hundreds of people on mopeds driving around London, but maps in front of them trying to memorize the names and lay out all these streets to become licensed taxi drivers.
Um, so there's a distinction in London between licensed taxi drivers
and mini cab or Uber drivers. Um, and the distinction there is obviously a pass this test, or we'll talk about in a moment, but they're the only ones who are allowed to actually pick up someone who's held their hand in the air.
It's illegal to pick up somebody
otherwise, unless you've got this [00:32:00] license.
Yeah. For safety reasons safe, it's all regulated and monitored.
Benjamin James Kuper-Smith: I'm assuming you only allowed to use a black cab, if you have that. Okay.
Hugo Spiers: Exactly. So I think incredible group of individuals who do this and I always enjoyed, they're absolutely fantastic to talk to.
Um, because for me, the reason is that they they're one of these rare groups of people are really using their long-term memory to do something. Most people don't, most people use GPS to find their way around. Like if it's somewhere you've not been in a while,
you know, you rely on GPS or you've never quite made a journey, you'd use GPS for most people, but they're, they're literally every day solving a massive jigsaw puzzle in their head.
Every time somebody asks them for somewhere. Um, so it takes two to four years to memorize. 26000 streets. I mean, there's, there's 53,000. Somebody told me in the whole greater London area that you could memorize, but for the main passing that that test you've really got a very good grasp at [00:33:00] 26,000 street names.
Benjamin James Kuper-Smith: How many zones is that on the tube system? Like it's, to me, it's not all of it. Right.
Hugo Spiers: it's depends what? So it depends what budget. There are two different badges and
that the green badges that really extended one most are green badge holders. And, um, Yeah. they need to know all these zones. All it's huge. It's insane. Um,
so they learn in a very
Benjamin James Kuper-Smith: people that haven't been to London, I mean, that's such a big area.
Hugo Spiers: Yeah.
it's, it's ridiculously big. It's a ludicrous proposition that you say or proposition, sorry that you would say, uh, now go and memorize 26 times. And, uh, I think, um, for, for, for me as a cognitive neuroscientist, this is fantastic because you've got a task that you could never do in the lab, the cost to train someone, to memorize that information and to know that.
That everyone you're testing has done what you are required. And then crucially, as part of this, going back to the great exhibition, that to sit an exam and prove they did actually know it. So the exam is just being given to random places in London, and they have to tell the examiner, the [00:34:00] exact route each street they would take to go from the origin to the destination, indicating what turn they would take.
Is it forward as it left, right to comply with what roundabout regulation,
everything they need to know. It, they can't just, they also don't. There are also some names that are not given street names. They're just given hotels. So they, they need to know thousands of hotels and venues and you know, all these different locations that could be asked to go to.
So it gets extremely elaborate. And there's something fascinating talking to them about, uh, we published a paper just at the beginning of this year in hippocampus on how do they do this?
Um, so that was, that was the first real step in
Benjamin James Kuper-Smith: So the learning or The
Hugo Spiers: The learning,
Benjamin James Kuper-Smith: it, the navigating.
Hugo Spiers: the learning, the learning process of how they
learn. We did a lot of Like basically a field study looking at this. so that kind of takes me into the, the taxi brains, uh, story. But I guess before I get further into what we're doing in that, in that paper, which is published in [00:35:00] hippocampus, if you go to my Google scholar page, like one of the recent papers sitting in there, it's it's open access, so anyone can read it as well.
Um, that also reviews, what do we know up to 20, 22 about London taxi drivers? So I cover it if I want someone's to read further on this, I've tried to cover everything that's in the field, every published paper on them. And then we dive into, how do they do that? How do you memorize 26,000 streets and thousands of points of interest?
And there's a lot of detail in that paper about all the techniques and things they do.
Um, but by government, exactly,
Benjamin James Kuper-Smith: or,
Hugo Spiers: exactly, exactly. A bit written up, but this is the first kind of documented article that lays out. How do they do this? Because I thought it'd be very useful to, to document that in terms of
how might we help develop better tests, tools for training people, if you want to learn a new city or,
something, what would you do?
Could you do it like a London taxi
driver? What do they do?
Benjamin James Kuper-Smith: So brief question, uh, do these, these London taxi drivers are most of them native [00:36:00] Londoners who kind of grew up in the city already, or are they actually people who came from outside and then started at zero knowledge? Because that obviously, um, I mean, that would be the most interesting, I mean, for, for the learning part, I think the most interesting comparison maybe might even be people who already have, or there's not as versus people who don't know that much yet about
Hugo Spiers: you, you, your rights? Um, no, it's, it's a good question. Um, we do our follow-up work. We've been asking about this sort of thing. What's their background and, um, yeah, my 90% of them will have
more than that. We'll have been in London doing something. They got excited about London. Most of them really are like, they love London.
They just love being in
London and going round London. So it's the idea of being a taxi driver is really appealing because it's like I get to be paid to drive around London and gets in and really get to know under the lid, like lift the lid and
they have to learn all about these venues. And I just love it.
I love driving of course is the other key aspect of it. Um, but, uh, Yeah,
you're right. It'd be fascinating to study someone who had no knowledge. And then, and then did all this, one of the [00:37:00] nicest stories. Uh, PhD student at the time. Um, so on the types of drivers to, to marry his wife, uh, he had to, he had to sit there, do the knowledge, his father-in-law said you can't marry my daughter, nevertheless.
Benjamin James Kuper-Smith: Was he a taxi driver
Hugo Spiers: he was a taxi driver. Yeah, he was,
Benjamin James Kuper-Smith: a random person
Hugo Spiers: no, he said, I want you to do this. or if you're good enough, then you can do that. And he did he succeeded, married, this, married his wife. So it was very sweet. Um, it's like thrown down the gauntlet to, to do that so that's the background.
And then this, the project taxi brains was really a project begun because I decided enough time had passed that there was a number of questions that are not been answered. And I was always, I had really enjoyed working with Elena Maguire when I was in her lab with the participants, the taxi drivers, and thought, well, I'm going to go dive back in again.
And, um, I was fortunate to have a really motivated PhD student, Eva Maria Griesbauer who joined my lab under a ups or in one of these economic, sorry, the, the [00:38:00] engineering, physical sciences research council funding. And she was also co-funded by ordinance survey. So they were keen to support work on taxi drivers and
mapping and understand that. So this is, this is the government, uh, Regulated don't run company that does all the mapping for the UK. So they,
they keep them, they keep updated maps for the entire layout of roads and highways and all this. But of course they make money by selling maps. OAS maps are very profitable so they've been fantastic to work with wouldn't survey. So they, they began the process of helping and get going and ever dived into, uh, you know, every Maria was just, she'd been native at the taxi drivers attending all these knowledge schools. So they learn, uh, studying how they do this so that was one of the things she did.
Um, another thing was she did that we published was looking at boundaries, where do they think boundaries are in London? Because to me, it seems obvious that this is like the river Thames , and then there's places like Soho and Mayfair and, uh, in a Bloomsbury clearly there's Bloomsbury and [00:39:00] there's the city.
know, it just, I know these labels for London and I kind of assumed taxi drivers would have a really good idea of where they start and, and, and, and stop. um,
and what we found was that's not true. They really don't. They have very different perspectives on where they individually think a lot of these boundaries fall.
A good number of them are absolutely consistent, but it's small. So things like Soho. There's no dispute there distinctly an area. So if you, and the reason we were interested as is kind of that hierarchy of, of, of navigating. So if you're driving to Soho, if you get a street there, you don't think about what you do inside Soho.
You just go to Soho. If that makes sense, same with Mayfair. Um, but if you're going to somewhere in say Bloomsbury , they might not do the same thing because it's more of a open area.
It's not got a confined in a band reset to it. Um, and you know, each taxi driver might have their [00:40:00] own distinct street.
They think of as being Bloomsbury and particularly the city of London is a complete mess. Its boundaries are very irregular and they don't, they don't have an idea. Most of them have where it precisely for them. Well, it's not so much that they do know it. Many of them will have to know that, but they don't think about it.
Consciously, but that's a very niche question and it was necessary for the next thing we looked at, which was, let's explore that exam. Let's go and test them on. Can they tell us what streets they would take going from a, to B on so qualified and what could this tell us? So the really cool thing I felt about this approach was that you can't normally in a normal population.
I can't say to you, right. Ben tell me every street, you would take an exact right? order from your house to, uh, somewhere on the other side of town, that's a bar and you would just wouldn't normally know them or to tell me,
but with this, I can test someone 30 streets. I can have someone go through and plan it or, and see what route they're taking.
And for [00:41:00] me, an exciting thing around that is going back to previous collaborations we've had of this is exciting to look at the cognition of that in relation to networks. Because they're plotting their path through a network of connected nodes and edges in the graph. And can we predict them when they take a long time to think about the next step?
Or a short time in the real world. So taxi drivers, aren't robots. They're not, they're not, uh, an amazing DeepMind AI. They are human. And there it is. Um, but in some ways like memorizing 26,000 streets and efficiently in three seconds telling you it. And I think a key thing is they can do all that.
They can give you the answer to this question and they really have to solve it fast because they could be sitting in the middle of Euston road at the highway with cars, zooming around them. And someone says, can you get me to Peter street in Soho they can't sit and have a good old think about it?
Well, a line of, you know, buses behind them are honking out. So they have to solve that fast and they can do that on the basis of having [00:42:00] eaten a sandwich and a cup of. Not on huge amount of trainings, all sorts of, I think this is one of the things we underestimate is that the human brain is running, running on the basic food and drink.
We're eating. It's remarkable with the energy resources has got so, so they're doing that, but that's been really exciting to think. Can we predict where they'll go on when they'll make the jumps in the space? So looking at mental simulation in the real world with a group of people that have a fast space state space that th the factorial number of possible jumps is
bigger than the life of the universe, or the number of stars in the universe is far smaller than the number of transitions.
Those taxi drivers. In London and their head. So anyway, that's one of the projects, but we then put them in the MRI scanner as part of taxi brains. And of course in 2022, we want to know, can we still see these differences in their hippocampus size?
And I don't have the answer yet. We're sitting on the data.
We have to crunch
it. My next meeting after this, in fact, after this [00:43:00] podcast, we just to follow up and see where we are.
Benjamin James Kuper-Smith: So, one thing I was curious about is that , as you mentioned, there's been quite a lot of work on taxi drivers and I was just curious, is there anything here that you basically couldn't have done in 2005 or whatever, is it, or, or is it a matter of just you or a student got really interested in it and you just said, oh, well we still have outstanding questions that we want to ask, or is there really anything where you went like, oh, finally, now we can do this or that.
Hugo Spiers: Uh, there are two things we can do now. I mean, most of it is just, we didn't do this in the past. And I think for me, that capacity to exploit their recall is, is incredible. I obviously love watching their recall or these routes because many times there is No.
gap. They'll just rely on a sequence of places, five streets in a row that they can think through where the gap is less than 500 milliseconds and their response time, other points, 20 seconds, a sitting thinking.
What move do I make [00:44:00] next? Why is it taking 20 seconds to choose the next streets? You know, that's rare. Most of it's, most of it is around three seconds. So that's something we could have done, you know, he could have done it in 1853. Right. , what's new, the two things new we've got with this project, um, one is the FA with the MRI data, we can go much deeper.
So we've got much more resolution and MRI scans and we can now decompose the hippocampus into all the sub-regions. So it's the posterior hippocampus is what's been measured in the past, but is it CA3? Is it CA one, there are a number of sub-regions to the hippocampus that may be where there are differences in size, uh, and to discover that will be quite important, I think, to, to clarify.
And let's say even just, if we can replicate the effect, it will be important because it's a textbook store. But if we can't replicate it, it doesn't mean it was definitely wrong in the past. It may be that things have changed with a number of aspects of, of, of what's going on. I wouldn't, I wouldn't rule out that [00:45:00] it's, uh, um, but my prediction and given the number of studies on this is that we will replicate that, that prior finding.
Um, so we can do more precision. The final thing we can do new is that it turns out, Sea Hero Quest is quite helpful here
because we can look and see how these taxi drivers compare to the rest of us, the 4 million other people, and they should be fantastic in navigating. Uh, what is it they do? Is there something in particular they do that, um, can tell us?
And if you put those two things together, there's an interesting story here that starts to build the story of the taxi brains. Is that for taxi drivers, they're good at navigating. There should be exceptional at Sea Hero Quest and their hippocampal size is larger. Then you and I, on the other end of the spectrum, you have people who are the early stages of Alzheimer's dementia, who have a smaller hippocampus and are worse at navigating.
And how does this manifest? So we've been looking at Alzheimer's cases with Michael Hornberger , and [00:46:00] understanding the impact of, of brain damage and dementia on navigation with taxi drivers. It's the reverse is thinking if you become very good at navigating, what does it show? What patterns emerge in the data there.
And that's the reason this is possible is that Sea Hero Quest , benchmarks navigation in a way we never had before.
So if you have a 54 year old male from the UK and you want to test, we have hundreds of sizes of 54 year old men from the UK to compare them to. So there's loads of analysis that can be done with that.
Um, but that's partly, what's motivated this. So the taxi brains is then this sort of wonderful collaboration between ordinance survey who supported the initiation of the project, continued to be involved. And Alzheimer's research UK who supported the development of Sea Hero Quest who are now, you know, interested in is thinking about driving a mobility.
And, and these taxi drivers is, you know, what can they take? What could they unlock about the human brain that we don't [00:47:00] know? Um, but for sure it will be doing more. We've also run a functional study , looking at how they do that planning. So when you give them an origin destination, you know, do we see grid-like patterns in their brains as they put.
Do they, how do they activate, how, how to, what extent do they use their prefrontal cortex? And that is that when you say, all right, I'm going to plan over a state
space at 26,000 states that should involve a lot of prefrontal cortex and a key question for us is they can't be doing that on a model based process.
So, uh, you know, they
can't be, it's impossible. So what are they doing? And what is the, uh, and already preliminary, we can see some really nice results about Y you know, where they, where their prefrontal cortex gets dampened, you know, where they can do some very clever stuff. It just doesn't need to use it.
Um, so, so that's been a really key question for me, as well as getting a planning and the brain looking at these. So it's a diverse set of questions and we're, we're building it. It really isn't just one study. It really is now [00:48:00] a sort of increasing project where we're interested in all aspects of, of, um, London taxi driver cognition, really, and what it can tell us about the brain and then help with dementia. a bit of a
long answer, but Yeah.
Benjamin James Kuper-Smith: Yeah. That was one thing that it was that half surprised me half seemed obviously true. And that's when you mentioned that the taxi drivers , you would expect them to be better at the Sea Hero Quest game and initially I thought, yeah, probably. I mean, you know, that's basically that job, but then on the other hand, You know, they've, they've learned, they've spent lots of time practicing London specifically, and, you know, I'm assuming they all stay.
They've used physical maps that they've looked at to learn or drove around or walked around or whatever, or walking would take a bit for, to driving around London. Um, whereas in Sea Hero Quest your, if I remember correctly, you know, you do see a map at the beginning sometimes. Right? I can't, I can't remember.
It's been awhile, but often at least you you're in an unfamiliar environment, right? [00:49:00] Or, yeah, I guess my question is kind of maybe more on interaction between, um, that they, they might be good at doing the navigation, but only once they've learned the environment, whereas in the beginning where they're maybe not as familiar with the environment, it might not make that much of a difference.
Hugo Spiers: yeah, they might be, they might, if they're not good, they might improve more rapidly than
everyone else. That's one possibility that is something we can look out with the game. Um, and you're right. So Eleanor Maguire had run a studypreviously. Very, I was keen to get back to this because previously she's tested them and compared them to very small sample, maybe another 20 participants or so on their ability to watch a number of movies going in their own route and understand and work out what sequence people had seen landmarks and, and they were better at.
Um, but so that's not telling us a huge amount where here we can look at the trajectories. Why do they make choice this choice or that choice? Because see your request affords us the capacity to analytically. Look at when the trajectory takes a certain shape to [00:50:00] it in the path. Um, what is it people are choosing to do?
Are they heading for open water, occluded edges, and you played the game yourself? Uh, it was designed with the games designers and an architect, Ruth Dalton, who spent her career designing virtual environments to test and probe kind of these sorts of questions. So there are some levels where is a very clear map and it's quite an easy map.
There are levels where you can clear map and it's really complicated. Five places to get to, and it's all it needs to be planned carefully. And then there are levels where there's maps, just all water damaged and you can't see anything. So taxi drivers might be very poor when there's no map like they, maybe they really are good.
Their excellence comes out in their strategies for map use. Uh, your rights are that, that isn't another exciting dimension. It's not quite as simple as saying, well, of course there'll be better. It's more nuanced than that. And indeed, another aspect of this is they've got a smaller anterior hippocampus than the rest of us, according to the data.
And [00:51:00] so that's thought to be part of a system for doing new encoding of memories. So they should be worse at navigating if they want a
smaller hip account. So they do present this interesting conundrum to me that I I'm hoping we'll find out some really nice results from but that that's part of it.
These are the things we couldn't do. Uh, you know, five years ago or so maybe a bit.
Benjamin James Kuper-Smith: Yeah. Yeah do you, by the way, how many London officially people with the badge cabbies are there
Hugo Spiers: Oh, I keep forgetting. It's it's uh, I did see it on a website. It style. Oh yeah. Many thousands? many, many thousands. Um, yeah, I forget. And it fluctuates of course, but it's, it's a lot. And if you walk around London in certain areas, we'll see it's flooded with taxi taxis. Um,
Benjamin James Kuper-Smith: Yeah. No, it was just one thing I was just wondering about was like how many, what to get like decent sample size across several experiments. What percentage of London taxi drivers have to take part in your study, but it's not a high percentage then.
Hugo Spiers: that that's a beauty of this. As [00:52:00] you know, we've, we've currently got a cohort of 80 participants and in the past that was just so hard to get them because of Twitter.
And a lot of media we've been able to recruit many more people than we would have. Our power calculation suggests we need 34, brains 34 MRI
scans, uh, should dislike more than double the previous sample size to find the effect.
It's very clear. Can we come and replicate it? So we don't need a hundred taxi brains to test these predictions. But the more we can work with the better, and there's a whole range of questions we're coming up with. Some may require a lot of taxi drivers. Some may
be not so many.
It just depends. But yeah. again, social media is a good example where that's made these projects possible. It really is really helped enormously switcher.
Benjamin James Kuper-Smith: yeah. One thing I just read as well. So. Oh, so we were talking about this is that , as you said you have two separate let's say extremes here where people have dementia who are very bad at this and the cab drivers are very good at this. And, I think often it's very interesting [00:53:00] to study people who are world-class at something who are really good at a specific task, but that's very difficult to do because often number one, there aren't many people of those in a specific area.
And number two, most world class people earn enough money or a lot of money. So they wouldn't take part in a study, but it seems to me that you've really found a, um, I'm going to necessarily use specifically found it, but like this kind of research area has found. An elite level of function in something where there are lots of people around, at least in London who could actually take parts in, it might even want to.
Hugo Spiers: yeah, that's true. And, and the ones who'd come forward are much, you know, they're so much more enthusiastic to do the science than the average undergraduate who wants to.
do a psych study, you know,
they're really motivated. Um, so, uh, Yeah.
I mean, there, I would say that if your job you've chosen is to try and every other, you know, every sort of maybe 15 to 30 minutes plan a route across an
entire city, uh, then [00:54:00] you're an unusual person.
Who's got that mental challenge in mind. Um, it's pretty impressive.
I think you're right. It's telling us something about human expertise as well, or an interesting model for, for how do you compress a massive amount of knowledge? And that's one of the things with the taxi brains that was so interesting, really.
Benjamin James Kuper-Smith: Yeah. . I remember like once someone did, I can't remember who did a study about, I think football players and something like spatial decision-making or something. I can't say exactly what it was, but it was something like if you have a football player, especially let's say someone like a midfielder, um, let's just when, like Iniesta or whatever, right?
Like they have this incredible ability to know where people are and how they're moving through space and how they can, you know, get to different places. But the problem with football players is number one, there aren't that many elite players and number two, they earn millions. They're not going to take fun to study, whereas yeah, cab drivers
your, for you, fortunately don't unbalance, they might see [00:55:00] differently, but luckily for yours.
Hugo Spiers: Yeah.
it's it's we paid them at the same rate. They're driving a taxi around an hour's worth of experiment. We get quite expensive.
Uh, but luckily they're there. They've been fantastic at helping out.
Benjamin James Kuper-Smith: Okay. So one topic I like to talk about, which is kind of, I mean, it's kind of related to all you, all of the stuff you do, but it's not directly related to the stuff we've been talking about.
Is that, from what I can tell, um, you are one of relatively few people who work, who do studies, both with humans and with non-human animals. I'm sure there are other people, but it seems to be, most people either do work with rodents. I mean, usually you focus on one species usually. Um, and, um, you've done work with rodents and you've done work with humans.
So just like to talk about that, uh, for a few minutes, at least, um, kind of what that's like and what the advantages disadvantages might be. I thought we can maybe start by saying, how did you start out? Like, did [00:56:00] you learn FMRI first or did you learn rodent work first or kind of, what was your trajectory in working with, um, these different.
Hugo Spiers: that's a great question. I started out with a neuro-psychology so my undergraduate degree at UCL was in neuroscience, but I got involved in a study at Kings college, London looking at the Institute of psychiatry, temporal lobectomy patients, um, testing their navigation, their path integration ability, uh, by
then. Yeah, that was my introduction. And I loved the project. I love all sorts of volt, immersive virtual reality. In 1997, we were doing that. And then they were really expensive and unwieldy and virtual reality headsets. So it never worked out. Um, so I worked, I started working with humans in that, and I went into a PhD in neuro-psychology with Neil Burgess who got really into
uh, starting a line of work along on that back. And, uh, you know, this is 1998. And throughout my [00:57:00] PhD, I was excited to get into brain imaging. It was really on the rise, particularly functional magnetic resonance imaging. It really exploded as a
Benjamin James Kuper-Smith: at UCL.
Hugo Spiers: at UCL.
so that the brain of the FIL was just discovering all sorts of major discoveries in the late nineties.
And slowly sort of worked up to the point where I was getting involved in MRI studies and you know, uh, Yeah.
carried on doing your psychology for a fair bit. Um, but eventually joined Eleanor Maguire's lab back in use having gone to Cambridge and then coming back, uh, and did that for three and a half years in her lab.
We talked to her, the taxi driver where we, a lot of that was MRI. So crucially will happen at the end of that period. And my career trajectory , the grid cells have been discovered in 2005. And there was this explosive, these, these poor people that aren't aware of these, that these cells in the, mostly in the medial entorhinal cortex of rodents have been recorded from predominantly that provide this hexagonal tessellating pattern of activity across the space.
Although the rats running random or rain searching for [00:58:00] it for rewards, the signal is absolutely stunning that I think they're one of the most important discoveries
in neuroscience, basically, grid cells. So just draw jaw-dropping and I thought, wow, I really made a mistake. Didn't I, I should've been, I should have been in this field
Benjamin James Kuper-Smith: mean, you were kind of in the field, right? I mean,
Hugo Spiers: I were, but it was frustrating. So you're right. It says I was in this field of spatial navigation. I was enjoying it, but I realized is you if you're in that field, The precision and the level and the
questions you can ask with rodents third, just on, you know, and I, you know, I have to say credit to, to particularly say Christian Doeller and Neil Burgess and Caswell Barry when they started thinking about ways to look at some of these things in MRI data that sparked a whole lot of ways of re looking at human data, we, you know, moved on, but still there is that sense of you're at the coalface, you know, with the cells and the neurons.
And, uh, I realized so on, if I'm right?
it was like a Thursday. I noticed , you could apply for a grant. [00:59:00] And the Wellcome trust that would give you three years of personal funding fellowship, but no, just your salary search costs, but no teaching. But you could get it only if you completely changed what you were doing to something else.
So this is an advanced training fellowship, uh, and I thought, Ugh, be great too. To to do that. That would be amazing. I can apply to do this and that's what I did. But the problem was that the deadline was Monday. So I had to develop an entire grant over a weekend, get it costed and everything organized the supervisor, the entire thing in three days for a grant
deadline, absolute hell, but it transpired.
I know I went through and got awarded that. So that was a real complete about shift. Uh, you know, paper-wise and reading didn't change. I'd always been reading these things. I just went from pressing buttons and MATLAB and putting people in MRI scanners. Oh, my goodness. The amount of training you need to do, rodent physiology was huge.
And I always think I, for sure did it the [01:00:00] wrong way. You would want to do the rodent physiology when you're a young student, because you have a bit more time and you can, you can do that. I had a small child at home by that point, and that is not a good combination of single unit physiology and small children.
Benjamin James Kuper-Smith: So just briefly where you at the time, what was your position where you elected to give lectures where you lecture or was that, did that pay for you to not have to do anything
Hugo Spiers: paid, not have to do anything else See, I
Benjamin James Kuper-Smith: so you could actually focus on it. Wasn't like you had to give lectures every other day or something. Okay.
Hugo Spiers: but I'm really? grateful. The Wellcome trust has set that up. They, they disbanded that fellowship a few
years later. They weren't. I, yeah, as a charity constantly rethinking about where the most useful use of the funding was.
And they decided that that wasn't producing the ideal results. And I think, you know, one perspective would have on that was that I suspect like me, many of the other people who've gone through this advanced training three year fellows, she was captured three years is how the hell are you going to then apply for the next bit?
Because you've got [01:01:00] no, I couldn't generate in a successful papers.
Three years of learning electrophysiology. It wasn't going to happen. So I didn't, I couldn't, I literally was disadvantaged to extend my career at that point. And there wasn't any kind of leeway. Uh, they were very sympathetic, but there wasn't that kind of leeway to say, we can take a gamble on you with no evidence.
Uh, that just doesn't, it doesn't work like that
Um, but that's how I got into
Benjamin James Kuper-Smith: on your CV said, didn't it help to have that fellowship on new CV as, I mean, you said you were at a disadvantage, but to some extent you all said we
Hugo Spiers: Yeah. I'm certainly, yeah, certainly. Unless you've got a personal fellowship. So back in this was 2007, I was able to negotiate a proleptic appointment and the university, which has become much more difficult now. But back then, that was an accepted kind of that you've required significant grant funding for your salary.
You're clearly kind a track record that suggests you'll go forward and do that. So the university say when your fellowship ends, we will appoint you as a lecturer. And [01:02:00] that's indeed what happened to me? Yeah, that was an impressive, and I'm very grateful to Kate Jeffery, who took me into her lab to train
I had as well, was the lab closed immediately. And, uh, they had, you know, building works and all sorts of things going on at the beginning
of my, but that's true. I mean, almost everyone's had these sorts of challenges in their career where you have to readjust and do things. That is the challenge of being successful in science, I think actually is adapting to no end, no end of problems that get sent your way, things not working. um,
so that, so your question was more than just how they end up doing it. It was that w how does it work out? How do you do that?
Benjamin James Kuper-Smith: yeah, I mean, first I was just curious about like how, because it seems to me that there are some people who switched or who do both, and it seems to me, most of them start off with the physiology and that switched to humans. So I guess that's what you also said might be the better route.
Hugo Spiers: it is. But I have seen other people, uh, Helen Baron is another example where she was in human FMRI with Tim Behrens and, uh,
switched to working with a [01:03:00] team in
Oxford to single unit recording. So I'm not the only one. I was
Benjamin James Kuper-Smith: yeah, yeah,
Hugo Spiers: she came forward to say, I was going to do this, did it work?
But it's hard. It really isn't, it's not a trivial thing to do, um, on indeed than running. Like you said, the question of how do you run a lab that does multi-species, um, work,
Benjamin James Kuper-Smith: yeah. I mean, maybe one question here I had is, are you doing it like fully in parallel or does it kind of come and go that, you know, you have a student who has funding to do animal work, so you're doing what, and then work the next three years or whatever, or are you ready? Yeah. How does that balance kind
Hugo Spiers: It's very dependent upon what successful grants you're getting. So in the last few years, I was fortunate to have a human brain project grant funded for a two year postdoc to do recordings. So we published a key paper from that. Last year on how to rats navigate, uh, you know, I have a multi compartment environment with different doors opening and closing.
And, um, so that was part of.
a key project, uh, with that with post-doc. [01:04:00] And that's been great, but that was very much like that it was supposed to link up with other aspects, very difficult across the grant and that sort of timescale. But what's been really fun in the last , three years has been another EU grant where we were in.
Put together, two students together, two very talented students signed them, hiring Chris Gahnstrom, Nils Nyberg who both Swedish, as that turned out. It wasn't by selecting for Swedish brilliance,
but they are brilliant. So these two, particularly talented students join my lab to, um, Neil's ended up working on rodents and, um, Chris looking at humans in the exact same paradigm.
So we wanted to have humans and rats running round, trying to find hidden goals and going to random occasions to get rewards , and look at the neural activity and rats alongside the neural activity as recorded with FMRI and humans and then relate those, integrate them to with reinforcement learning models that would predict what the patterns and the.
The two species where, and that's still underway. We've got the, [01:05:00] as,
you know, as you can imagine, the FMRI data is somewhat easier to acquire.
Um, the road and data is ongoing and, but it's been amazing. It's been a real joy to have gotten to that point in my career where I could see, you know, this actual integration, the lab of the human and rodent work in one paradigm.
Um, and we've been building that up for five years before with, with a whole team of people looking at rats and humans. And one of my intentions going forward is not just to stick to two species, but it would be great fun to, to work with more species where we can explore other phenomenon and look at that, but that all these things take time.
Well, bees bees are interesting
Benjamin James Kuper-Smith: I'll be okay.
Hugo Spiers: But there are others such as drosophila I haven't worked with yet, but there's a lot of exciting questions you can look at with other species where they have different ways of planning. So humans and rats differ, right. And
rats, vision's different there.
They use their whiskers a lot. They have to groom. They have to pause the groom, a life and humans, you know, [01:06:00] blink and they're all sorts of things that are in get bored in ways that rats. So, so this is interesting questions about sampling across lots of species and trying to understand core common mechanisms and things that make them different.
But these things are all challenged. And I think one of the things I have encountered that, that, um, I hadn't anticipated, but it's blooming obvious when you think about it is that if you are running a lab at all these speeds in different questions and you're integrating over it, I've got all that experience to draw on.
I've done all those things, and I know
how much how beneficial it is, but if you arrive as a brand new student
engaging in this, you've got to learn your trade. You can't afford to do all that. So you've got to pick one and then you end up being a bit more siloed as a student. And I think that then makes some challenges running a lab where people do, they need really detailed feedback and other people just have no idea what you're talking about in that
part of the lab. Um,
Benjamin James Kuper-Smith: I haven't had to figure at UCL, you have enough people to, who can help out from other labs, but it has to be [01:07:00] collaborative than.
Hugo Spiers: it has to be, and I would really highlight the fact, the only reason I've been successful at making this work is that I'm surrounded by just wonderful colleagues. Who've completely put up with me not being able to pull the same weights. They've been just wonderful. So these are particularly people like Kate Jeffery, who's run the Institute of behavioral neuroscience for open able to work as an absolute star in terms of making that work.
But it's not just her, it's all the other PIs that I work alongside of really, um, made it. I just couldn't do what I'm doing at all, uh, in most places. And just collaborators who help work on more detailed analysis. And so this is the great, the great thing in science about the further you go through.
The process you get to collaborate with lots of people and throw ideas around and make jumps and leaps that you see as the PI that you don't really get earlier in your career. Um, I think people just get this continuous negative narrative about how horrible science is a lot of the time. Um, but that side of it's [01:08:00] really good.
Benjamin James Kuper-Smith: Yeah, just as an aside, Kate Jeffery has also been on the podcast. So if people want to know more about her and her work, um, there's an episode about that, that I can put in the description. Um, but one thing I'm just going to see is. So this is something that I also, I mean, I'm kind of interested in animal work.
I'm not sure I ever do it. But I think the more general question here is that of the common dilemma of specialization versus depth versus breadth, do you feel like sometimes you, you know, you just don't really know either the things they're talking about because you're comparing yourself, so not comparing yourself, but you're speaking to people who've spent 20 years on one thing, whereas you've spent, let's say 10 on each, um, just to, you know, give an example, is that a problem to you or do you think you just win more perspective or yeah.
Hugo Spiers: I think it's the latter actually mulling over it. Maybe send to, Yeah.
I mean, certainly. So generally it's the last year that you gain more perspective? I feel [01:09:00] like, Yeah.
that's, that's the feeling I have, right. Is that
how it actually is? If someone metrically lifted, it might be a different thing, but the, the area where maybe I feel, I wish I had more time or more skill in was the analytical, uh, you know, to, to,
So I really, I've been extremely fortunate again, to have people who spent their career. Really thinking hard about analysis methods and so on. But, um, I just benefit from, so I have a whole range of it over my career of understanding what different ways you can look at data and tease it apart. And I read up on methods, but actually, going into the fine detail of figuring
out what is, and isn't going to work and guiding a student rattly through it.
That's the area where I think that is increasingly hard. And I've certainly had supervisors throughout my career who also didn't have that skill. So someone like Neil Burgess is unusually
amazing and that he [01:10:00] has,
Benjamin James Kuper-Smith: Yeah.
Hugo Spiers: he really does know these things and this really, you know, so there's that kind of thing.
But these things keep, the landscape keeps changing. So even if you are utterly brilliant, you just do not have the time to sit down and really figure it out. Often as a pie, you've got all those emails to answer that we started the podcast or, you know, so that's probably the frustration of. Um, uh, spreading, I think maybe one side of what I'm doing is that it does require jumping, changing hats a bit, and that has a cost.
You kind of get caught up thinking, is this a project I should be involved in or not? Is this a risk to do this line of project? And so, Yeah.
it's hard, but I do see it, my intention is to completely continue doing cross species work, but it does, it's entirely dependent upon funding bodies who allocate money to me to be successful enough at making both of that work.
But, you know, we'll see, that's partly the challenge.
Benjamin James Kuper-Smith: Yeah, let's see if you get your third species [01:11:00] or fourth species on. I mean, it seems like you're going, you know, further down the hierarchy of complexity, so to speak. Um, I don't know, maybe one day you would just be doing mold or
Hugo Spiers: yeah, yeah, yeah. That's a slime mold has been
on that one application in the
past year that, yeah, yeah, yeah. They don't slime. Molds can do some pretty impressive things.
Benjamin James Kuper-Smith: Yeah. So at some point you're just stretch the entire old families of living organisms or something.
Hugo Spiers: Yeah. Yeah.
Benjamin James Kuper-Smith: Um, okay. Um, so you mentioned something earlier, which I think leads quite nicely to the final topic I wanted to talk about. And that's, you said the kind of, you're working to some extent on the integration of rodent and human work.
And I think your, the review paper that I wanted to talk to you about is a pretty good example of that because I guess you're kind of doing two things on that paper. The one is saying, we know all this literature from Rodens. Does it also, like, what do we know about humans basically? And the other part is going from physical navigation to the more abstract cognitive maps.
The, [01:12:00] the paper I'm talking about is your 2017 review in nature neuroscience. And, um, because I haven't mentioned this against, uh, for new listeners, I put references to all the papers we talk about in the description. So that will be in there. Maybe I started a slightly, uh, Not random by, but yeah.
Study random point. So I was just curious. So you were at, you mentioned, you know, 2005, there's the grid cell paper that came out. And then in 2008, Christian Doeller, Caswell Barry, and Neil Burgess's paper came out showing that you can also find grid like activity in humans. I'm just curious. What was that like for you when you were working on this kind of stuff and especially because I guess you'd been in Neil Burgess's lab relatively recently before that, or maybe even at the time.
How did you hear about this and what did you think? Uh, once they said like, Hey, we can actually do this in humans,
Hugo Spiers: I was very excited to hear that. I remember, um, I mean, I was in, I was in UCL, so this is 2010. So it was in my [01:13:00] fellowship
to lectureship and, um, Yeah. I remember hearing that they pulled this off and thinking, wow, that is very clever. That is, that is a clever move.
Benjamin James Kuper-Smith: as you were moving, as you were learning road and
Hugo Spiers: I was switched.
I just started my, it came out of the year. I'd switched to being a lecturer. Um,
Benjamin James Kuper-Smith: So then they showed actually you don't need rodents.
Hugo Spiers: I don't, I wouldn't say that. I mean, not
that paper is, is a good point, but that paper includes a whole rodent analysis from Caswell. So he looked to
prove one of the reasons this is a great paper is that it is genuinely combination of recording and rodents and showing you get these patterns.
And then let's go look at the FMRI signal. Um, and it has been replicated a number of cases. And of course, crucially, Josh Jacobs had a nature neuroscience paper showing he could get grid-like patterns in from single units in humans recorded from patients. So it's been followed up and there's a number of follow follow-up impressive studies taking this [01:14:00] line.
It was a real thing in UCL. And of course, the thing you will want to consider is what's described often as the stretchy birds paper,
um, Constantinescu and, uh, in Tim Behren's lab. And, uh, Yeah.
She did an amazing job with that paper and that was a real, I think that point was, I can't quite believe this, not as in the skeptical, these are bad sciences pullings on the envelop, but this is a real step change, um,
Benjamin James Kuper-Smith: I mean, what are you thinking about this at the time? Like, I
Hugo Spiers: Not at all.
Benjamin James Kuper-Smith: grid cells in humans or
Hugo Spiers: think Tim. Yeah.
I hadn't been thinking about this perspective at that point when the work was just come out. Um, I hadn't even, I mean, when this line of thinking emerged, I was unaware that Tolman back in, uh, the 1930s was actually first described in a rat's navigating amazes and example, but he'd considered the, the content map might extend more generally that it was even in Tolman's
original thinking. But I was always aware, [01:15:00] always aware as an undergrad. Even as a sort of young PhD student reading the cognitive map, by John O'Keefe and Lynn Nadel, the, in chapter 13 of that book, they go wild and exp expand the ideas into language. Uh, and
it's an incredible chapter, a piece of writing in 1978 that predates All this, this work, um, thinking about how the map might extend beyond space.
So a lot of this thinking was sitting there for a very long time, 1930s, 1970, um, and our reason it was swept away as the precision of the spacial coding has just been phenomenal and that, you know, that place cell literature is just so exciting and evocative that it, it took precedence over this, this angle.
Um, so Yeah.
in that review in 2017, Russell Epstein, Josh, Julian, and Zita Patai, we were looking at. Put forward the argument that we've got a lot of good evidence now in 2017 of, of similarity across humans and [01:16:00] rodents. But like you said, where does this go? Beyond space, you know,
Benjamin James Kuper-Smith: Yeah. So I guess, in that paper, the three main points that you, Highlight also in the abstract are basically that the hippocampus and entorhinal cortex to support this map lab coats.
Um, then you have para-hippocampal and retrosplenial cortices, help anchor maps to environmental landmarks. And then frontal lobes are used to plan routes. So you kind of have this, if you want to put it this way at the back of the brain, you kind of have sensory inputs that create a map and fixated at the front.
You have the planning into whatever you want to do in the middle. You have the map itself. And, because it was slightly running short of time, I think maybe we can, I guess kind of skip the spacial stuff. I mean, I've have, I've had quite a few episodes already on what grid cells are and places are and that kind of stuff.
And maybe go straight towards the more abstract of, um, before we do the actors, one question, which is something. Uh, I'm not, I don't actually know the [01:17:00] answer to what exactly is the difference between a boundary vector cell and a border cell I just read that somewhere in your thing. I felt like, wait a minute.
What is the difference? I thought there were the same.
Hugo Spiers: that was discussed on Twitter recently with a back-and-forth dialogue.
It's, uh, it goes back to what you're analyzing in the, in the data. So the boundary of actor cell as described by Neil Burgess, when he postulated they exist, um, are cells that fire when a boundary, a wall or some feature that restricts movement is it a particular distance and orientation in the Cardinal or the, the centric space?
So I walk into a room and there's a wall that's to this two meters to the Southwest of me. There'll be a border cell that gets excited by this feature. And it doesn't matter where I go in the world. It gets excited by. Two meters to the Southwest, that same particular boundary vector cell boundary cells is described by the Mosers and their paper where cells, where you were quite proximal to the boundary, you had to be close to it.
Um, and that were [01:18:00] not so vector driven. So when I'm near boundaries, they would fire. They may be more modulated by boundaries in some directions, but generally that these cells
are thoughts of fire. The thinking is that probably something very similar. It's just the way they've been analyzed and presented.
But for simplicity, people often say border cells and boundary vector cells. And
it may be one way of thinking is the border of cells or a subclass of boundary vector cells. But I think that maybe even that's too simplistic to be a bit careful.
Benjamin James Kuper-Smith: Yeah, it's just, I mean, I guess, you know, for me, this is something I'm really interested in, but something I don't always read to cafes. So whenever I read like boundary or something, for me, that was just the same thing. Um, until I, until you kind of put them side by side or something, I was like, wait a minute.
Um, so for this very specific question I have a more general question about the metaphor of a cognitive map, because I think one thing. Easy to forget when you kind of get swept up in spatial navigation, and there's a cognitive map and all this stuff is that the [01:19:00] map is of course a metaphor.
And I'm not criticizing metaphors. There are useful, and it's basically impossible to think without them. But I was just curious, what do you think are some potential problems with this metaphor? Um, so, you know, we think of a map and then we often, I often imagine a physical map let's say of London in front of me, and then you can use that to create new ideas or whatever, but where do you think this metaphor breaks down, uh, for space or even for abstract things and what problems might it cause in terms of people thinking too closely in terms of the metaphor, rather than the thing that actually trying to study
Hugo Spiers: Yeah. no, it's, it's a core. It's one of those slightly deeper, important questions in the field about how the language we use then impacts the kind of way we interpret the data. And it's a good one to have. It's good to think about, um, yeah, I'm always aware. I mean, I think, uh, I've always since reading about it conceived of, if you go back to the place cells and the first [01:20:00] suggestion, so how you link brain activity to map.
So the dark very long, 500 page book by John O'Keefe and Lynn Nadel lays out. The case in the 1970s is why you would use the map metaphor to describe place cells. And the idea there is this a sort of moving bump of activity in this bit of the brain. That is a kind of a, you are here point it's constantly roving ran.
So it's not. As you said it's not something that can be conceived in one image. It's a thing it's a kind of a signal that allows spaces to be integrated. So even at the understanding of what the cells are doing is a challenge. There's something I've been keen to actually write more generally.
I don't think, I don't think there's a good article out there. If you, maybe you've been studying, thinking about this as well, you could point me to that, but I haven't come across one that really tries to. Hi, this is operating for a general reader because it isn't true. So the brain has a map in it. Yeah, And then there's these place [01:21:00] cells, but what does that mean? And your question is getting to that metaphor idea. So it's not a map in the sense of the cartographic one, it's a system for mapping space. but it's not an egocentric one. It's an allocentric one, and then when you start to get to, that was out of century, cause the other centered is fixed on the environment.
That is kind of a map as we understand maps. So I'm reusing the words and not really defining them and getting you much further in that. But I'm getting through my thinking, one of the most important features of the debate around this and I've written about it wrote a short, uh, commentary and TiCS a spotlight, um, which entitled something like.
Ah, I should go back and check what the title was, but it was trying to be a bit provocative. Like,
is there a universal map? Like, is there actually
a general cognitive map? Uh, I got a bit frustrated by slightly lazy thinking, uh, in some of the papers that raid. So I went back and, you know, highlighted, it's a small number of citations in there.
What a highlight in that think about your question is the distinction between physical spaces where you can have vectors. [01:22:00] You know, I can choose a million ways to move across the room. And I can think about the Euclidean geometry to solve problems in the space is different to a social network or a stimulus network or anything that isn't actually.
Set within a real space, it's connected by gradients. And, you can, you know, uh, it came about to say that the stretchy birds paper, which we didn't, we didn't walk our way through, but this idea of you can have, and there's some lovely work by Christian Doeller's lab as well. Extending that to hippocampal activity.
You can have a kind of continuous stimulus space. Um, but a lot of the world is, is not necessarily like that. You know, semantic knowledge is not so smooth. Um, so th this lack of smoothness and, and, uh, Euclidian spaces is, is a really interesting.
feature of, of that. There's a nice paper as well. Um, Michael Peer and others in Russell Epstein Nora Newbome, and Iva Brunec I think it'd be authors on cognitive maps and cognitive graphs.
So that even in
space do, does everyone [01:23:00] actually, Yeah.
Does everyone actually build a map and that the, you know, the data suggests, some people are more graph, like in the way they build their, their understanding of space. So, so even in physical space or a questions about the way it's represented as it truly map, like,
so I think these are all questions that are at the heart of my short review.
Postulating that this really attractive to have this, this universal cognitive map perspective that Tim Behrens and many others have put forward. But it struck me that there are enough differences between navigating a social network and a physical space and fundamental differences in them that it seems plausible.
The brain may have specialized the hippocampus in some ways. And in particular data on lateralization in patients where you tend to get like, um, certainly work. I started up my career, right? Lateral lateralized temporal lobectomy is resolved. Quite severe damage and navigation ability. But if you damage the left hemisphere, you don't get these same
effects that are pretty normal.
And this goes back to Brenda Millner and [01:24:00] case H M she was doing a lot of unilateral temporal events, memories, similar story. So if that's true, it may be that the brain has slightly specialized systems within there. Isn't one universal map that just does everything. It can do space like the equivalency of space to other, other features is maybe overblown.
It was my, my thought, but I think what we wrote in that 2017 paper, I would stand by many people who argued is that. If you look at the hippocampus and the way it's structured, anatomically and physiologically, and the fact that we know it's needed for these things, the social navigation, then, then it seems extremely likely that there's a universal computation.
That's developed as vertebrates. We've developed as hippocampus to navigate spaces, but also life, you know, uh, Neil Cohen and Howard Eichenbaum I don't know, the campus is not just for navigating space, but for navigating life was, was the argument made, which I thought was too vague. But this precision of saying you can have a grid light codes for [01:25:00] non-spatial that Tim Behrens has developed in his papers.
And, um, there's other nice word from Dory, Dory , um, that kicked off some of this, um, before, um, that that's really like taken us to some really nice directions for thinking about how computations from spatial coding could be used in non spatial space. But
it was a really great question about this vectors and how our brain deals with the smooth planes versus grid-like environments, or would like structures of knowledge.
Benjamin James Kuper-Smith: Yeah. Then there's of course fairly recently the nature paper by the Mosers about the torus. I haven't read it yet, but that seems to know that exactly what it does, but it seems like that's changing quite a bit, how we think about grid cells anyway, but I mean, I didn't also want to like, uh, you know, I'm not criticizing the metaphor itself and I think it can be really useful.
And that's maybe the last thing we can end on is then what are, for example, landmarks in abstract conceptual space. Cause that's something that, to some extent [01:26:00] makes sense to think about it that way. Um, but it doesn't always lend itself to an obvious answer. And sometimes I'm wondering like, is that because I, I just can't see what it is or it doesn't just not exist.
So maybe the question of like what a landmark and abstract conceptual space is, maybe briefly, like we use landmarks to navigate and maybe what are some of the key features of a landmark and how does that maybe help us understand what landmarks would be in conceptual space?
Hugo Spiers: Yeah.
that's, that's a great question. I've actually a question I've never heard posed before, which is really nice. I may have imposed in, um, uh, you know, you'd been running over, um, the recent book, uh, you know, thinking about these metaphors more, but I, I don't have a really great answer for that as well.
Uh, cause I was just thinking about the nice work by Michael Kahana's lab, um, university of Pennsylvania. Keyson where you, you know, you're looking at factor, word space, the word to vac, you know, K you know, estimates of, uh, [01:27:00] relationships between words. So you can move in these different multidimensional word spaces for Symantec knowledge.
Um, but what's a landmark,
as you said, or you're moving along these dimensions. So I'm going down here and I'm got to take a left on that plane. What would provide a landmark to allow you to turn left to the next word orientation and the word space? I think you're right. The first step is that. Seem like there's an obvious, you know, correlative, that's getting into this abstract space for that phenomenon.
There may be another scenarios that may be, um, um, aspects. Um, I mean, one thing we've been playing with and I'd love to play more where there's the, the graph structure is so properties are graphs. So when
you think about, you know, really central nodes, like in a city, there'll be a central roundabout or somewhere like in London, you know, Leicester square or Oxford circus, where they're really important points where multiple roads come together that [01:28:00] also will exist in social networks.
And there's just that one person who knows everyone, uh, you know, they're really connected. They're
a really important node. And so there's interesting aspects of, of looking at that, that isn't landmarks that isn't answering your question directly, but it's, it's highlighting their assumptions features of the.
Physical and non-physical spaces have , nodes and, uh, you know,
important properties within the space. Um, so Yeah.
Benjamin James Kuper-Smith: Yeah. I mean,
Hugo Spiers: been with,
Benjamin James Kuper-Smith: it seems to me also, like, it just, it really depends on the context you're looking at and some it's fairly obvious. I mean, there's also the, you know, I've talked to Jacob Bellmund on the podcast and about his review paper. And, um, since then, now I've read Peter Gardenfors's book, um, conceptual spaces.
And I think that, um, you know, obviously cons summarize it here, but, um, I think that that kind of the idea of having stereotypes or prototypes of different concepts, um, goes maybe [01:29:00] somewhat towards that direction. Um, and there's, I mean, I, I found Peter Gardenfors's book really interesting. Um, and I think there's lots of stuff to be done that he kind of just briefly mentions here and there.
So in some contexts it is quite obvious, but again, others, it feels like the whole thing might not even work or Yeah.
Hugo Spiers: yeah.
Yeah, no, I agree. It's a, it's a hot topic in the, in neuroscience and cognitive neuroscience is understanding more about this. A lot of people are very excited about the extension of models and ways of thinking about space into non-space and exploiting, uh, all the rat of Matics people were described in the road and field of spatial
navigation. So many ways of analyzing. So I think that's, that's, what's driving some of this.
Benjamin James Kuper-Smith: Yeah. It's funny. I've mentioned this before in one of the interviews, but I remember when I, um, took Neil Burgess and Caswell Barry's course I felt like, oh, they figured this out. Like, you know, [01:30:00] there's grid cells, there's place cells. I was like, okay. We don't quite know how it works, but we've pretty much figured it out.
But then, yeah, it's just the start. There's still a lot of interesting stuff to be done.
Hugo Spiers: Oh, for sure