Daniela Schiller is a Professor of Neuroscience and Psychiatry at the Mount Sinai School of Medicine, where she studies the neural mechanisms of emotional control and flexibility. In this conversation, we talk about her work on cognitive maps for social behaviour, the importance of power and affiliation for our social lives, the difficulties of measuring spatial navigation with fMRI, and potential psychiatric applications of cognitive maps.
BJKS Podcast is a podcast about neuroscience, psychology, and anything vaguely related, hosted by Benjamin James Kuper-Smith. You can find the podcast on all podcasting platforms (e.g., Spotify, Apple/Google Podcasts, etc.).
Support the show: https://www.patreon.com/bjks_podcast
Timestamps
00:04: Daniela's drumming
03:31: How Daniela started working on (social) cognitive maps
08:42: The 2 perspectives on the hippocampus: spatial navigation and episodic memory for relational learning and cognitive maps
15:22: Power and affiliation as fundamental social dimensions
19:24: Start discussing Daniela's paper 'A map for social navigation in the human brain'
28:45: The difficulty of measuring spatial navigation with fMRI
42:51: Clinical applications of cognitive maps
Podcast links
Daniela's links
Ben's links
References and links
The Amygdaloids: https://www.youtube.com/@theamygdaloids
Bellmund, De Cothi, Ruiter, Nau, Barry & Doeller (2020). Deforming the metric of cognitive maps distorts memory. Nature Human Behaviour.
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.
Jacobs, ... & Kahana (2013). Direct recordings of grid-like neuronal activity in human spatial navigation. Nature Neuroscience.
Montagrin, Saiote & Schiller (2018). The social hippocampus. Hippocampus.
Schafer & Schiller (2018). Navigating social space. Neuron.
Schafer, Kamilar-Britt, Sahani, Bachi & Schiller (2022). Hippocampal Place-like Signal in Latent Space. bioRxiv.
Schiller, Eichenbaum, Buffalo, Davachi, Foster, Leutgeb & Ranganath (2015). Memory and space: towards an understanding of the cognitive map. Journal of Neuroscience.
Tavares, Mendelsohn, Grossman, Williams, Shapiro, Trope & Schiller (2015). A map for social navigation in the human brain. Neuron.
Tolman (1948). Cognitive maps in rats and men. Psychological Review.
Yartsev, Witter & Ulanovsky (2011). Grid cells without theta oscillations in the entorhinal cortex of bats. Nature.
Daniela Schiller is a Professor of Neuroscience and Psychiatry at the Mount Sinai School of Medicine, where she studies the neural mechanisms of emotional control and flexibility. In this conversation, we talk about her work on cognitive maps for social behaviour, the importance of power and affiliation for our social lives, the difficulties of measuring spatial navigation with fMRI, and potential psychiatric applications of cognitive maps.
BJKS Podcast is a podcast about neuroscience, psychology, and anything vaguely related, hosted by Benjamin James Kuper-Smith. You can find the podcast on all podcasting platforms (e.g., Spotify, Apple/Google Podcasts, etc.).
Support the show: https://www.patreon.com/bjks_podcast
Timestamps
00:04: Daniela's drumming
03:31: How Daniela started working on (social) cognitive maps
08:42: The 2 perspectives on the hippocampus: spatial navigation and episodic memory for relational learning and cognitive maps
15:22: Power and affiliation as fundamental social dimensions
19:24: Start discussing Daniela's paper 'A map for social navigation in the human brain'
28:45: The difficulty of measuring spatial navigation with fMRI
42:51: Clinical applications of cognitive maps
Podcast links
Daniela's links
Ben's links
References and links
The Amygdaloids: https://www.youtube.com/@theamygdaloids
Bellmund, De Cothi, Ruiter, Nau, Barry & Doeller (2020). Deforming the metric of cognitive maps distorts memory. Nature Human Behaviour.
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.
Jacobs, ... & Kahana (2013). Direct recordings of grid-like neuronal activity in human spatial navigation. Nature Neuroscience.
Montagrin, Saiote & Schiller (2018). The social hippocampus. Hippocampus.
Schafer & Schiller (2018). Navigating social space. Neuron.
Schafer, Kamilar-Britt, Sahani, Bachi & Schiller (2022). Hippocampal Place-like Signal in Latent Space. bioRxiv.
Schiller, Eichenbaum, Buffalo, Davachi, Foster, Leutgeb & Ranganath (2015). Memory and space: towards an understanding of the cognitive map. Journal of Neuroscience.
Tavares, Mendelsohn, Grossman, Williams, Shapiro, Trope & Schiller (2015). A map for social navigation in the human brain. Neuron.
Tolman (1948). Cognitive maps in rats and men. Psychological Review.
Yartsev, Witter & Ulanovsky (2011). Grid cells without theta oscillations in the entorhinal cortex of bats. Nature.
[This is an automated transcript that contains many errors]
Benjamin James Kuper-Smith: [00:00:00] Yeah, I, I thought I could start with, uh, I guess with a slightly random question, but I guess I like asking, you know, as I said, these kind of side topics a little bit. So, um, when did you start playing the drums?
Daniela Schiller: I started playing the drums, uh, roughly when I was. Well, it was during, um, my service in the Army, . It's kind, it's a funny way to say it. Yeah, I just, um, in my Army service I was exposed to some, um, army bands and, uh, they had a lot of instruments and musicians. So I just, uh, you know, started playing the drums and there was, uh, A drummer there that treated me seriously and started teaching me, and that's how I got into it.
Before that, I played the piano, but always wanted to play the drums, and that was my, uh, way in
Benjamin James Kuper-Smith: Interest. I just assumed that you'd started playing it in childhood or something and got . It's funny, I always get slightly jealous of people who play the drums because I always wanted to [00:01:00] play the drums when I was a child, but I wasn't allowed to. I mean, I played all, all sorts of other instruments, but uh, the drums were always too loud.
But I guess hearing that, I realized I can probably still start as an adult. Yeah.
Daniela Schiller: Yeah. Yeah. So I was, uh, relatively an adult, right? Kind of around the age of 18, but then I had my own salary and uh, I lived kind of in a separate unit, uh, the house. So, And I could just make my own decisions more independently so that, that allowed that, uh, me to just go ahead and do.
Benjamin James Kuper-Smith: Okay, so I mean, I saw somewhere, I think this might be on the Amic de Lloyd's website. I think it was, it said you started a band in your chartered called the Rebellion Movement, but that was on piano then I, I'm assuming, or was that on drums?
Daniela Schiller: No, that was after I played, I started playing the drums then. Then my first band, yeah, the one I joined was the Rebellion Movement, which is a play on, uh, words in Hebrew. in Hebrew, you say Nuta Mary. Rebel Rebellion is [00:02:00] Marion. We just had our lead singer. Her name was Mary, so, so that was our play on words,
Benjamin James Kuper-Smith: Okay. Is, uh, I was curious, what is, what exactly is Hebrew Rock? Is it just rock and Hebrew or? Does it have specific,
Daniela Schiller: rock. Yeah.
Benjamin James Kuper-Smith: so it's just the language. It's not, I was wondering like whether there's something like different about it other than the language, but
Daniela Schiller: Yeah. You just pronounce it like rock.
Benjamin James Kuper-Smith: okay. Yeah. Yeah. I mean, how did the, by the way, is the Milo still, are you still playing together or is that, I dunno, how.
Daniela Schiller: Yeah, in, in principle we were still, uh, a band, but, uh, in the last few years, uh, Joe Ledu was a band leader. He kind of went unplugged, and he is playing with, um, guitarist and singer songwriter. That actually I play with also, like we kind of split into small bands. Uh, so that has been easier for him, uh, in term during the pandemic [00:03:00] and in terms of traveling.
But every now and then we do our reunion. For example, one of the band members got married, so we played a little bit during, uh, the wedding or we had an anniversary. So every now and then there are events that we get together. So we, we never formally. Um, separated.
Benjamin James Kuper-Smith: it's never been a big breakup of the band. Uh,
Daniela Schiller: Yeah.
Benjamin James Kuper-Smith: yeah. I mean, I guess the, the, the main thing, uh, you know, what we, what we mainly wanted to talk about is kind of your work on, let's call it social navigation or. Cognitive maps for social, for social interactions or social behavior. Um, the, the one thing that really surprised me about this in some sense is it seems to me that you came that most of your, your previous research, like before I guess 2015, seemed to be very much from the kind of memory direction.
Um, I'm just curious, like how did you kind of go from the memory literature into now? I mean, you're still [00:04:00] continuing there, right? And your, your lab webpage says, um, the description is, Can we change our emotions? Our lab is interested in discovering the neuro mechanisms underlying emotional control and flexibility.
Uh, then you mentioned traumatic emotions and patients with anxiety and that kind of stuff. I'm just kind of curious like how, like how does the social space kind of work fit into that and how did you get to that from your previous research?
Daniela Schiller: Yeah, it, it emerged in an interesting way. I was, uh, studying traumatic memories and fear memories and how you can modify them. And uh, there was a period where I heard, uh, several stories, personal stories from people who underwent. Genocide, you know, during World War ii and in a few instances, um, it seemed that, uh, the people who survived and were highly functional had a very acute understanding of their social environment, which was surprising, um, for [00:05:00] their age and the, the conditions.
For example, you could have a, a teenager usually without parents. You know, all her brick SLS are surrounding them. Of course, no internet or radios, nothing surrounded by by enemies. Sometimes your neighbors are the enemies. And, um, the way they describe their behavior was through very good understanding of what's going on at the social level.
Who's a friend, who is an enemy? What are their motivations? And they could, uh, figure out how to interact. And that, uh, seemed to play a big part in their survival and how kind of they managed their environment. And, uh, I thought about it in one instance, and then I heard another story that popped out again, and I started thinking that it's actually really important for survival for you to Mac to understand your social environment.
And, uh, at that time, um, I had a, a postdoc that. . [00:06:00] Terrific. That's, uh, Rita Tavaris, and she, she made a huge transi transition. She actually came from immunology and she also, we kind of, um, we took some time, each of us kind of to think, uh, what are we in? We knew we wanted it to be something social but wasn't clear what, and she was also interested in, in social aggression and social defeat.
So we kind of quickly converged on the question of how one maps their social environment. and we, um, came up with the idea, can we assemble the team? And, uh, from the idea of how to analyze the data, we. came to the conclusion that it, it's, it might be using something like polar coordinates, you know, angle and vector, which was, which were prominent in other areas of the brain, for example, in visual cortex.
And you, you could see these types of encoding, said, okay, that makes a lot of sense. It's probably how it's done. And we assumed it's the, it's the hippocampus. And [00:07:00] we came up with the dimensions. Uh, it started kind of more intuitively, . Yeah, of course we map, uh, along the dimensions of power and affiliation, but then we also found a great deal of supportive, uh, literature.
So that was the, the initial idea. We kept it from the idea that it's important for survival and, uh, it, it allows you, it, it gives you some resilience. Even if you're the most inferior in the hierarchy, you would be better off because that type of representation allows you to understand and make predictions.
And so that was the idea. And uh, then we realized that we actually have a home. And that home was the, the home of the relational learning kind of, uh, Howard Eichenbaum at the time really opened the door for us and, and, and told us, yeah, you'll belong here cuz we considered it as spatial like, And kind of encouraged us to, to [00:08:00] frame it more as relational learning.
And we're specialist one case and we are another case of it. And it goes along with these many instances. And at that point that was nice because this is when we realized we do have a a home. Before that, we were just weird, you know, we didn't belong to the PTSD field, we didn't belong to the social field.
We didn't really belong anywhere. And that was the natural home. So the nice thing is that we really didn't come up from the cognitive map. We just arrived at it, you know, from a whole different direction. And, and since then, now we've been working in that direction.
Benjamin James Kuper-Smith: Yeah, I mean that's the, the, the aspect that I find really interesting because I guess in. Especially in your review in Neuron, uh, you mentioned, I think, I guess also in the review in Journal of Neuroscience, you mentioned very much these kind of two perspectives on the hippocampus. The one is kind of this, um, let's say more the memory side.
In psych, I studied psychology, so [00:09:00] likem and hippocampal lesions, that kind of stuff. It's super famous through that and the other side. Kind of spatial navigation place. So grid cell kind of area that you have these kind of two different interpretations of what the hippocampus is for and that these kind of cognitive maps might combine the two.
And I guess for me, it's particularly interesting to hear that you come from the memory side because it seems to me that, or most of the, let's say I basically, even though I studi psychology come more from the place in grid side. So I guess I'd especially like to learn a little bit more about kind of the memory side.
Um, I mean, could you maybe kind of just, I mean, I already mentioned kind of the two perspective. Can you maybe kind of flesh out, especially the kind of memory side, more on kind of what uh, yeah, kind of the key aspects from that literature that should be in, how should I put it that, I mean, if we want to kind of create a theory of what the hippocampus does, kind of what are the key elements of the memory literature that should be included in that?
I hope that question makes [00:10:00] sense.
Daniela Schiller: Yeah. Yeah. The two fields of, uh, studying the hippocampus, as you said. Spatial navigation and episodic memory. And interestingly, um, they very much studied apart by different people, different conferences, different journals and so forth. But, um, some people, especially Howard Iban, which uh, was a great influence, uh, for me, came up with a theory that there's a joint mechanism of relational learning.
That is, you organize information. in a map like, uh, organization, if you have several dimensions, uh, that are continuous and need to be linked, so that's how you organize them. You find the relationship between dimensions in memories. The way you can see it is with the sequence of episode, uh, just like you have, uh, sequences of steps that, uh, animal is navigating Here you have sequences of episodes and uh, [00:11:00] they could occur in different context.
and that, uh, type of sequence is what you represent. These are your data points that you organize just like you organized locations in space. And that type of, uh, relationships use the same, the same basic code of relational learning. The idea is that these are different cases of relational learning.
Other ideas are that, um, everything started from spacial representation and was built on it. These are. I think current, prominent, uh, ideas in the field, it's not really testable. I, I think it's like, uh, it's like theories with assumptions, right? Uh, There's no empirical evidence for that. Just a, you can consider what is the most plausible theory, but these are two interesting, uh, aspects of how to link and beyond these two.
So it started that the hippocampus had this, uh, main dichotomy of you either study episodic memory based on the patients who lost their, uh, memory [00:12:00] after lesion, and then fm r activations and, and so forth in relation to epi episodic memory. And then the other spa. Navigation now, um, or after in the last few years, it evolved to multiple domains.
Suddenly you find, uh, sound spaces navigation along a frequency domain and all faction and any manner of like abstract spaces. So now instead of having this dichotomy, it's like a new era, I think in the hippocampal research where you have this like multi-dimensional approach. , uh, sharing a fundamental code or way to organize information.
Benjamin James Kuper-Smith: Yeah, I mean, uh, you, you mentioned like kind of all these sound and all these kind of things, and I mean that, that's the part I think that I really liked about your review is that you had this section called, I think increasing abstraction or something like that. And I really like kind of how you like laid out all the.[00:13:00]
Yeah, I guess getting increasingly more abstract and showing that this is, you know, a very general thing and it's not specific to kind of one, you know, sensory modality, um, or one particular kind of of behavior. Yeah. I just wanted to say that I really liked that section, uh, because I guess I was also quite a lot of stuff in there that I didn't know about, uh, until I read it in your.
Daniela Schiller: Yeah. I think, um, I can say something about that, which is what I like most about the idea of the social space is that it's, um, I think the most extreme case of obstruction because, uh, if you start with the physical space, everything is concrete, right? It's based on, uh, Actual navigation in the physical world and, and sensory information.
Then you can go to other sensory domains like, sound like, uh, and olfaction. Then you can in increase abstraction by organizing rules in the environment like categories, but [00:14:00] it's still based on, um, visual categories. Uh, for example, you would see a group of items and there will. Clustered under a rule, which is abstract, uh, but it's still tied to some sensory information that you organize in a rule-based manner.
And also in many instances, uh, participants are tasked with learnings specific locations, and these locations are rewarded. Or they are required to map some organization in the social space. Um, there's none of that. Um, they really just interact with people. They're completely unaware of the dimensions of power and affiliation.
They're not instructed at all to map anything. They just interact. Uh, so it's, it's latent in that, in that sense that it's implicit. Not instructed and not rewarded. It's not a task that they need to perform. And from that highly naturalistic behavior, we can [00:15:00] extract these locations and the organization in the abstract space.
So in this sense, I think it's a very interesting support for the idea of this shared code. Uh, even when you remove all these elements that, um, might, uh, be alternative explan.
Benjamin James Kuper-Smith: Yeah. Shall we maybe go a little bit into a little bit more detail about. The study you did on power and affiliation and you know, representing that in a social map. You mentioned earlier that a lot of your inspiration for this came from talking to, uh, people who had traumatic memories and that kind of thing.
I'm curious, is that why you chose power and affiliation as categories because they were like important from that context or, yeah. I'm just kind of curious why you chose those two dimensions as social dimens.
Daniela Schiller: Yeah, it started. Uh, the very first thought was intuitive in a sense that if you do need to come up with dimensions, um, usually it's, it's hierarchy and, uh, intimacy, right? You [00:16:00] can see across species that hierarchy is, uh, important for survival. Writer's, aggression or hierarchies built through aggression.
And then there are reproductive bonds, which are the affiliation part or familial bonds. and in our society there's, there's hierarchy and affiliation. Then if you think of other dimensions, they usually converge onto back on, onto these. For example, you can think about, let's say attractiveness, uh, trustworthiness, kind of different words you would find.
Uh, they usually end up being about affiliation. So we realize that these dimensions are clusters, clusters of, uh, these different terms. There. There's no commitment to terms and we found, um, a lot of support in social psychology theories, dating back to, you know, the sixties and the seventies, how people organize, uh, or understand [00:17:00] interpersonal behavior.
That was on the axis of dominance and love. And then, uh, Alex do of, uh, had a, a model of trustworthiness and, uh, uh, and dominance. These are, and this was data driven. He found that this is what people, um, get from facial feature. , uh, to evaluate these, uh, these faces. They kind of, uh, organize faces along these two dimensions.
Uh, another model was from Susan Fisk who talked about warmth and competence, which is, uh, how we assess people's abilities and then their, and their intentions or, uh, what they aim to do, and then their ability to act upon it. And depending on where you put them on that space. , we lead to some, some stereotypes.
For example, people who are incompetent and have, uh, let's say good intentions, then you would [00:18:00] feel sorry for them, uh, and so forth. So that was, uh, we found these two dimensions across, uh, a large set of theories. And there was also, um, the theory, uh, that relates to psychological, uh, distance, which again suggested we represen.
Psychological distance along multiple domains could be spatial, temporal, social, and these dimensions, uh, could interact and interfere with each other. But there will be a presentation from an egocentric perspective. So these are just examples of different theories all converging or providing support to, to that idea.
These are two fundamental dimensions, but we never thought it's just these two. This, for us, was just a, an approx. You know, you start from the fundamental ones and then you, you might increase the level of complexity, but to begin with, if there is any such a presentation, even if it's [00:19:00] multi-dimensional, then these two dimensions would be a good approximation of it.
Yep. So, so that was our starting
Benjamin James Kuper-Smith: Yeah. So basically like if, like if this social stuff is embedded in a map in some sort, then. These two features would be definitely part of that because they're two of the most important ones. And, um, okay. Yeah. Yeah. Maybe can you then, uh, I guess we've kind of been like tiptoeing around the paper a little bit for the last few minutes.
Um, just kind of briefly summarize kind of what you did in that paper and what the main findings were of that.
Daniela Schiller: Yeah. We created a, a story. We really wrote a story. We had a, a research assistant who was very talented and funny and used to write for Saturday nightlife. Uh, so we had, uh, a lot of fun writing the story and we came up with these interactions. There, there were a lot of decisions to be made about what is a power interaction, what is an affiliation interaction?
So it's a storyline that you go through as a character and then you interact with different characters. Uh, you come to a new city [00:20:00] and. have to get acquainted with the characters. And your goal is to find a place to live and to find a job. And currently you're jobless and you are living in a subject. So you interact with the characters, they approach you, and they come with a kind of different, uh, scenarios.
It's a continuous story. It's like a little movie, and then you, you make decisions and the interactions are divided into power interactions and affiliation interactions. For example, the first character is this person who used to be with you in high school, and she's like, Hey, what are you doing here? And kind of very happy to see you.
And then, uh, she goes for a hug. And this, so this is an affiliation interaction because it's about, um, physical warmth and, and physical touch. And then, um, you can decide if you go for a long hug or just give her like a brief, a brief pat on the back. And in this, in this case, you, you choose. Uh, if you can [00:21:00] imagine a game board with her on the, on the game board with these two axis, if you chose a brief pad on the back, at this point, she went one step away from you.
Uh, but if it was a, a long hug, then she would go one step toward you. These are the, these are the steps in this dimensional map. Um, power interactions would be, for example, there's this boss character that everybody likes, but kind of this, uh, you know, , extreme personality of it, . So that person will say, it's a, so how badly do you need the job?
You know? And uh, or it could say, I need you to stay late. And these are situations where you could say, yeah, sure, boss, whatever you need. Or you can say, well, are you gonna pay me extra? Or I can't, you know, find someone else and all that. So again, the person will, the character will go up or down in. and the idea that you really behave naturally, follow the story and based on your [00:22:00] choices, we translate it into this movement, uh, in the two-dimensional space.
And what we get out of it is trajectories. The characters move. They have this, uh, trajectory. They started from a, an arbitrary point of, uh, neutral power. and some distance from you from which they can go away or close to you, and then they move. So you can see like lines developing over time and this is what we measure.
That's the geometric, uh, model that we have for your social relationships. After you play the game, you recreate this map, and then that's a way to quantify your behavior. Now we have numbers for each interactions we have coordinates, we have the vector length, we have the orientation. We can look at the sum of these trajectories, for example.
if you change your mind a lot, let's say you give power, take power, give power, take power, or get closer and away, closer and away, the trajectory is wound developed. You'll just go [00:23:00] back and forth, which will end up with very short vectors. And overall, a very small space. Or let's say if you reject everybody, they will all go toward one quadrant and it is, it'll be a narrow space.
Uh, others will have different types of varieties. So they will have a wider space. All these are quantities that we can measure. Um, and the important thing is that it's a way to quantify ongoing relationships, which we didn't have, uh, before. It just gives us numbers, uh Yep. So then we can take these numbers and make predictions out of them and, and say, well, if the brand is representing information in that way, then we should.
brain activation that correlates with these coordinates. We with these, these dynamics or changes and we tested that and found that the hippocampus was tracking the orientation of the angle. The posterior single at cortex was tracking the vector length. And, um, the interesting [00:24:00] thing was that the greater which the hippocampus was tracking, the coordinates correlated with people's, uh, self-reported.
social anxiety and, um, some, some deficiencies in personality traits. So it seems to be the case that if you don't represent the world in this way, you're, uh, worse off in, in, in terms of your social functioning, which was consistent with the initial intuition that you do need that to, to get along to survive.
Benjamin James Kuper-Smith: Mm-hmm. um, . So I just had to laugh because I guess you're in New York and sometimes that comes through through the background of, uh, police sirens and honking horns and all that kind of stuff. Um,
Daniela Schiller: Yeah. I also, you know, I'm in a hospital, so
Benjamin James Kuper-Smith: yeah.
Daniela Schiller: bound to hear multiple, uh,
Benjamin James Kuper-Smith: sometimes it seems like, it sounds like there's just mayhem behind you. Um, [00:25:00] um, yeah, I mean, so, uh, one kind of question I had whilst I was reading the paper, and to some extent, I guess you kind of already explained this to some extent, uh, but I guess I'll still ask the question is kind of what I kind of thought when I was reading the paper.
One thing was like, this seems like a. Uh, like an almost unnecessarily complicated task, right? Like if you compare this, for example, to the Constantin paper where they just, you know, they have the birds with a different, uh, neck length and leg lengths, and then they, you know, you move from one bird to the next and you have this like very simple, I mean, at least comparative, very simple task.
Like, one kind of question I had is like, why is it such. Story like task and such a, I mean, just like from, even from a practical perspective as a scientist, like why go the length to have, you know, all these stories and all that kind of thing, rather than just presenting people with different attributes on the screen or something like that.
Um, could you comment a little bit more on this kind of why [00:26:00] the. Um, yeah. Why this kind of like intricate story with, or what do you call it, choosing your own story, whatever it's called, um, kind of why you went with that route rather than something a lot simpler.
Daniela Schiller: So the complexity of the task. . It's interesting, uh, cuz my, um, entire, uh, research before that was based on fear conditioning. So you know that it couldn't be more controlled than that, right? You have stimulus shock, stimulus, shock, stimulus, stimulus, that's it. The most controlled possible. Uh, I think one, one aspect was it that I just started my own lab and, uh, it, it's something like you break free so you kind of do something wild, but.
Beyond that, uh, it didn't really scare us that, that it's, it's complicated, uh, quite, uh, the contrary because if you take such a noisy environment but carefully [00:27:00] isolate one component in particular, if the component truly exists, then it'll come out and everything will be canceled out. Um, you, in, in a way you, uh, kind of protect yourself from, uh, con.
because, uh, when you have it highly controlled, you, you really have to omit a lot of, uh, parts and some parts that you wouldn't omit could interact with it. Uh, also, um, the, the task, for example, esco, it's still, as I mentioned, it's still sensory. You still organize sensory information in the environment. So we had to do something that really reflects social interactions and that's it, right?
I mean, really the way to interact is to have. Person approaching you and you, and you're responding to that person. So the, the naturalistic aspect of it was, um, really required to get at the question. And the other consideration was that stories are very [00:28:00] engaging. We are kind of naturally drawn to stories and you look quickly and embed yourself in a story.
You don't need much, you just need a good story. Uh, and in this sense, we trusted that people will have that experience. Yeah. So I think about the approach is, uh, to go naturalistic as much as possible, but not to aim to model the mess, you know, or the con the complexity, but rather approach it mechanistically, uh, and isolate this one process or algorithm that goes through that noisy environment.
And it's supposed to come.
Benjamin James Kuper-Smith: One thing that's, um, slightly surprised me, I mean, uh, about this is that. About the, specifically about the findings with, at the, the angle, um, that, that were correlated, uh, the, the angle of movement through the social space correlator with hippocampal activity. I, I mean, what do you think the kind of. [00:29:00] I mean, it's not exactly what the mechanism is here, but I mean, one, one thing, or maybe I'll kind of backtrack a little bit.
Uh, in one of your reviews, you kind of mentioned that one really cool thing about having these two completely separate kinds of literatures is that you can draw inspiration from the other to have new ideas, new methods, all this kind of stuff. And, you know, taking inspiration kind of from the, let's say, special navigation literature.
Hippocampus is in this phase, is is the, you know, the place where you're at in this, let's say two-dimensional space in this case, but it's not, and you know, grid sales, uh, would be movement through it and these kind of things, but angle, I mean, that would be something more like a head direction cell or something like that.
Right? Or like, um, I'm just curious, like why did you, I guess okay, why, why did you. Go specifically for angle and look for correlates of that and not, yeah, I dunno. Um, I mean, for example, 1, [00:30:00] 1, 1 thing I was curious about this, um, because I knew about the content in Esco paper first. So when I heard about in your paper, I assumed, for example, you were also using.
Kind of this, um, hexa directional, um, analysis. Um, so I guess I'm kind of just curious that kind of why you chose this particular kind of fm, FM r i analysis.
Daniela Schiller: Yeah, the, the angle, uh, It is an interesting choice compared to the rest of the literature. We wanted to find a, a way to quantify location and polar coordinates are convenient in that sense, um, cuz you have get the angle and the length and that gives you the location. So with these two, you can find where people are and that's what we were going for.
We didn't go for a grid-like representation at that. Um, this requires, um, many, uh, samples kind of sampling in that [00:31:00] environment to fill out the grid and find correlates. Um, so that was kind of our, our initial approach. And it is true that it's something like head direction and we thought about it like orientation, uh, because, um, if, if it is a space you kind of orient toward one direction versus the other, and then there's the.
From you. And we also assumed it should be egocentric because of that literature and psychological distance and just the, the idea that when you interact with characters, it's relative to yourself. Uh, for example, you will have only half the, the field it's, uh, 180 degrees because there's no people behind you in that space, right?
Uh, unless you conceptualize. or history, you know, of relationship is, is something that goes behind you. Uh, so that was really just the initial, as I said, like it was so, uh, wild that we, we kind of went with these, these approximations and it [00:32:00] says, okay, if, if it is true, then we find something like this. It wasn't, um, easy to interpret because, uh, the more, uh, the angle was kind of toward one, one direction, which.
Usually goes with high power. Then we saw more, uh, activation that is higher amplitude, which is not easy to, it's not like play cells,
Benjamin James Kuper-Smith: Yeah, exactly.
Daniela Schiller: fire the same just in different locations. Um, so, so, um, that wasn't definitely the end of it. We, we continued analyzing, uh, and definitely needed additional methods to fully flesh out this idea.
But that was the initial intuition, the idea that it should be from your perspective, uh, egocentric and it's, it's a form of orientation. Since then, we did, um, additional analysis that are closer to place, like representation. We use the representational similarity analysis and the idea is, , if [00:33:00] you have two characters in your space, if they're close, then the neural activation should be more similar to these characters as opposed to characters that are far apart across the trials during the task.
It's not published yet, but it's on bio archive, and that gave, uh, additional, uh, evidence or more direct evidence for location per se rather than orientation. , but we we're still working on it. Uh, we want to see whether there's overlap between the representation of the location and the angle. Um, because, uh, I think these are different types of information that are represented and somehow, uh, integrated in the pka, you have the egocentric and the acentric and, uh, it made, it may be even, uh, a more complex type of representation, uh, that we, we were looking at.
Benjamin James Kuper-Smith: Yeah, but it just seems to me that the one unfor unfortunate. FM I is that play cells are just [00:34:00] really, I mean, from what I can tell, basically impossible to really get at . Um, uh, yeah, I mean a grid, I guess grid cells are to some extent, um, through the, the, the analysis that cost introduced or some, you know, some other things.
Um, but yeah, it just seems to me that play cells are particularly difficult to get at with F M R I.
Daniela Schiller: Yeah, we are looking, uh, started collecting data from, um, epileptic patients
Benjamin James Kuper-Smith: Right. Yeah. Yeah.
Daniela Schiller: recording. So this, I hope, will get us closer to the animal electrophysiology literature.
Benjamin James Kuper-Smith: Yeah. But that's kind of cheating No, I mean, no, but that's the, no, no. I mean, obviously that, that's fantastic. But, um, yeah, I guess like the, the thing is just, it's, yeah, I guess it's very different kind of research and I guess ideally you'd have something what, that you can do with, you know, healthy people.
Um, but, uh, yeah. I fully, it seems like kind of intracranial recordings seem to be the, [00:35:00] the only real way of getting at them right now.
Daniela Schiller: Yeah, we're also now analyzing data from seven Tesla, so that gives you better signal to noise ratio and, and greater resolution. , um, yeah, still, still not, uh, neuron level. It just like you need really converging evidence
Benjamin James Kuper-Smith: Yeah, yeah. Yeah. I mean, I had one kind of general question I had that uh, maybe I can ask now is kind of like how to think about. I mean, one thing I really wondered about with this stuff is kind of the, not necessarily the reproducibility of these studies, but more the kind of, how should I put it? I mean, it seems to me very intuitive that, especially given all this other literature that already exists from animal literature and all this kind of stuff, um, it seems very intuitive to me that we would have something like, uh, we should be able to measure something like grid cells or play cells.
In humans, you know, if we can do so, but that in principle, they should be there. [00:36:00] So, and one thing I wonder about is, is maybe a kind of, maybe it's more like about publication bias or something like that, where, you know, it just seems to me every time, if, if you, if you don't find this kind of encoding of something, then it prob, it's more likely that it's a methodological problem rather than that the phenomenon doesn't exist.
and I'm just, yeah, like one thing I was just wondering about is like, to what extent kind of, there are lots of no findings maybe out there that people quite reasonably maybe assume is more like a methods problem rather than a, that the phenomenon doesn't exist, whereas maybe something isn't actually incurred in humans in this way, but that yeah, we kind of make this false interpretation.
Um, Do you kind of, do you see what I'm trying, what I'm trying to say? Um, I, I guess my, it's not really a question, but like kind of how do you, I mean, how do you deal with that kind of thing? How do you think about that when you do your own studies? Because Yeah, I dunno.
Daniela Schiller: [00:37:00] Yeah, I, I actually did my PhD in animals. I did animal models of schizophrenia, which were also based on fear conditioning. Um, so, so I did start with the basic neuroscience approach before I switched to humans. . And, um, when you're in the field of human neuroscience, the idea is not to accept any findings in animal for granted.
Um, that's why I, I think it's important to have that layer in between animals and the clinic before you go to patients to see how it is in the neurotypical human brain. and indeed it could be different. Not ev, not necessarily in humans. Um, I think OVS found, uh, some different, um, ethics, something related to Ations or I don't, I don't recall now, but, but something that was different actually with, from the like mic, uh, findings.
So, um, yeah, these are empirical questions. They are, need to be, they should be tested. [00:38:00] Uh, but you do need to take to account the. Uh, before you, you reach conclusions? Um, actually in humans there are, there is some evidence from that, uh, electrophysiological recordings in patients with epilepsy for some, uh, grid, grid cells and play cells, there is some, yeah.
Benjamin James Kuper-Smith: Yeah. Yeah. I mean, the, the, my point, I guess I'm, I'm not, that's kind of the whole point kind of. I'm not really doubting that it exists in humans because it seems like, you know, in more or less similar ways, it seems to be the case in different species. And there, there are these few patient studies in humans.
I guess, at least I've noticed in myself this kind of. Assumption that it was, it's probably gonna be more or less the same in humans if it's that way in all the other animals and that maybe I'm less critical, maybe sometimes when reading some of these studies because I go like, oh yeah, it should exist.
It kind of makes sense. It fits into the literature.
Daniela Schiller: Yeah, you know, the, uh, the literature of, uh, fear conditioning and the, the circuits [00:39:00] of, um, uh, acquisition of that learning, uh, are re remarkably similar across species. So you, you could, uh, assume, you know, for, for some circuits that they're, they're highly. , uh, but probably there are some additional layers on top of them or the inputs are different or something like that.
But yeah, these are I think, empirical questions.
Benjamin James Kuper-Smith: Yeah. Yeah. Where someone with some very good methods expertise can really make a difference and figure out how to, how to get at this in fm, I in particular, or even emg, who knows? Um, yeah, I mean, uh, to, to, to slightly kind of broaden it from, you know, from your paper into more general discussions about this.
I mean, I guess we've already been doing this in part, but I'm curious is do you think, is there any, I mean, are there. Any social aspects or aspects of social behavior that you think in principle wouldn't [00:40:00] be encouraged in this kind of way? Because I guess the whole, the, the gist of the cognitive map, let's say hypothesis, is that pretty much any feature that's continuous, let's say, or doesn't even have to be continuous actually.
Um, but that pretty much any aspect can be represe. in a kind of map, in a map like way. Um, is there anything that in principle you think shouldn't actually be represented in this?
Daniela Schiller: I think it depends on the, the task, uh, and relevancy. I think they mentions emerge if they are relevant. So wouldn't expect, uh, all information that objectively exists in the environment to be encoded, uh, in that way unless it's relevant for you to organize it in that way. Um, yeah, [00:41:00] it's, uh, so what other, what other aspects, uh, you imagine that that would be,
Benjamin James Kuper-Smith: Sorry, say again?
Daniela Schiller: it's a very broad
Benjamin James Kuper-Smith: Yeah, yeah. It is very broad, I guess.
Daniela Schiller: because I, I'm not sure, uh, I can pinpoint a response really.
Benjamin James Kuper-Smith: yeah. No, I mean, yeah, I mean, I guess I was just, because I guess in, I guess I'm just imagining that more or less anything that people think about, um, Could be encoded in here. Um, and I can't really think of anything. I guess, I guess, um, I go along roughly your lines that like, obviously not everything is encoded all the time in the, along this way.
Um, I dunno. I was like, I was, for example, curious whether something like a binary variable, something that just has to be binary by definition. Whether it's something like that, you know, you're not gonna put that on a, on a map, on a relational,
Daniela Schiller: I think that's could, uh, yeah, this might be, that's kinda how we were thinking [00:42:00] about mental dysfunction. Uh, if it's binary instead of the, the map, like then you're in trouble, right? You don't have that, uh, extended representations. That gives you the flexibility and ability to interpret and make predictions.
If it's b. . So that could be, um, uh, an impairment if you don't do that. But for some variables, um, I imagine there is, I mean, I don't think there's, um, one code that explains everything really depends on, on the case and, and the information. And what is it that you need to record. I think it would just go by the efficient representation.
If it's efficient and relevant, then you would use it.
Benjamin James Kuper-Smith: Mm-hmm. , I mean, you mentioned, uh, impairments in patients, and I think in one of your reviews, don't you also mention somewhere that, um, people with depression or schizophrenia have hippocampal impairments. Is that something you're looking at at all, [00:43:00] or are you right now staying with the. Yeah, non-clinical research in this.
Daniela Schiller: Yeah, there was, um, something interesting happened, uh, here at Mount Sinai cuz I work in a medical school. Within the psychiatry department and through conversations with the psychiatrists, uh, that finding was received, um, with great enthusiasm. It felt like they were really looking for something like that because the social, social problems, uh, you see them in many psychiatric disorders.
Many of them said that they felt that this was very fundamental and sometimes even a precursor or, or, or a cause for everything, but was, uh, in many cases, um, really regarded as like, uh, some side residual symptoms. So that was, uh, it was really well received and then they embraced the task and now it's running in several clinical populations, rather in autism and borderline [00:44:00] personality and avoidant personality, also, addiction.
which, um, social aspect of addiction is very interesting and, uh, it, it's sure that many of disorders, there's also hypo dysfunction in autism. You see that in schizophrenia, in depression. Um, so you have all the components, they just need to be linked. And we are working on that data is in collection, so I can't really give you funding.
You know, preliminary results to the extent we, we can trust them now. Uh, seem to indicate some, some relationship within the patients, but, uh, we need more data.
Benjamin James Kuper-Smith: Yeah, and I guess as you said earlier, it would probably be a bit, uh, naive to assume that, you know, all of this is explained with some people can't put dysfunction. Probably all of that's a bit more complicated. But yeah, I guess it is really interesting because I guess the. I guess kind of one, one consequence of the cognitive MAB hypothesis, you know, would be that it would be implicated.
I mean, not exactly, but it [00:45:00] almost means it should be implicated in some clinical disorders just because the hypothesis is that it's so fundamental to so many aspects of behavior. So it would almost be odd if, you know, dysfunction in this aspect would not influence all sorts of things in all sorts of different ways.
Daniela Schiller: Yeah, definitely. I think the question is, assuming that it is, then the question would be how. , uh, there's a part in Edward Tolman's Thomas's review from 48. If, if you go to the very end of the, that review is kind of funny, you know, he says in, he says, if I may be allowed to serve as, like, as a, a rent psychologist.
Uh, and he started kind of doing kind of interpretation of, of behavior and he talked about neuroticism and, uh, different types of. , uh, dysfunction that he kind of speculated how they seem as a map. So, for example, a map that is too narrow or, um, Kind of missing some, [00:46:00] some or some distorted representation.
Uh, so he already made that, that intuition? Uh, I think it could be, I can think of several ways that I would want to test. One is, uh, either the is intact, but the regions that feed information into it or impaired. For example, you have a visual cortex, uh, for spatial navigation. And other sensory navigation.
And you have the prefrontal cortex for social information, which is like, you know, the visual cortex of, of the social, uh, brain. So if, if impairments are in diff different regions, then uh, that might impact these representations. Another option is that you have the map representation, but it's distorted.
For example, by emotion. Let's say you are very. , um, sensitive to power, you know, very much afraid or anxious. Then maybe people [00:47:00] that are located in the area with high power because of distortion in this space, they would even seem higher in power. So it would be a non-linear representation, which, uh, I kind of, uh, my sense is the like, like the future of what we need to study with the maps going to shift from the.
Yes. to non-linear representation.
Benjamin James Kuper-Smith: Yeah, definitely. I mean that, that reminds me of, I talked to Yako Belmont about his study where they use a, this is a behavioral study in humans where they use this kind of, uh, Trapezoid shape or whatever. It's, it's not a square box, basically. Right. It, it kind of narrows towards one end. Massively. Uh, and I think they kind of link that, that kind of, these, this kind of, that kind of slightly distorts the space because you suddenly don't have this like neatly uh, angled space that you usually have.
And I think, I think it's something like that that then leads to slight memory impairments in those areas. I can't remember exactly what it is, [00:48:00] but. . Yeah. I'm really interested in these kind of like, um, how you can basically, yeah. To some extent, like transform these maps maybe in ways that people hadn't intended.
Um, or that,
Daniela Schiller: Yes. Uh, yeah. I think you, they also see distortions, uh, at the borders of the map, which is also very neat. Yep. This is a, a really fascinating direction.
Benjamin James Kuper-Smith: yeah. So is that, is that mainly what you're doing in the next few years, or by the way, are you, are you still doing a lot of the kind of. Fear conditioning memory kind of literature or, uh, are you
Daniela Schiller: Yes.
Benjamin James Kuper-Smith: Okay? So it's kind of both
Daniela Schiller: yeah. This is, uh, yeah, it's still, uh, alive and kicking, uh, in the lab. But, uh, interestingly, we also, uh, we're also, uh, venturing into naturalistic. representation, um, because after, again, many years of fear conditioning studies, which, which are very informative, one thing that that will remained unanswered is to what extent these types of processes and, uh, [00:49:00] even if you use computational models and then you do find correlates to psychiatric symptoms or PTSD severity, but it still doesn't touch with the core of the trauma, which is a very specific personal event.
So one idea we're exploring now is finding ways to analyze neural data while people listen to their own memories or even freely recalling them. So that requires, um, some sophisticated analysis. But again, it's the same approach, which is you do approach it mechanistically. You expect, uh, you have a very specific algorithm in mind that could.
explain aspects of, of that behavior, why they listen to that complexity. And even though it's very different across patients, because each has their own personal event, there should be something common that reflects the fact that it's traumatic. So that type of approach, we're trying to get it. Uh, what makes it traumatic [00:50:00] or in what way?
It's different because it's considered traumatic. Is it just a matter of degree because. Currently, our working hypothesis is that because we use this model is that in trauma, it's the same model. It's same processes that we are identifying now only at a, at a much more extreme level. But it could be that it's a whole different representation altogether, kind of in a, even an alternative type of representation, maybe, maybe even not memory per se.
So, uh, I think side by side, uh, the highly controlled studies, , we would like to do the more personal ones and, and see if we can compliment or maybe find some aspects that, that capture that personal experience.
Benjamin James Kuper-Smith: Cool. Uh, . Sounds like I have to have you back in a few years when , when all the results from all the different, uh, studies are in and. Uh, I don't know. Maybe we'll be, maybe we'll find out that especially the cognitive [00:51:00] maps can explain, you know, you've, you've figured out how they go wrong or what work doesn't, doesn't work in the different clinical groups.
Or maybe not, I dunno, but it sounds like there's some interesting stuff.
Daniela Schiller: Yeah, I think it will merge, uh, at one point because. , if you look at the trajectories of the characters, you are really looking at the storyline and uh, at a narrative, which is me memory, right? It's a series of events that people remember and it gives it structure. So imagine how that can be distorted after a traumatic event.
Uh, so I do see a way to kind of link everything, which is the emotion and the spatial light or the map, like representation and the social aspect of it. Uh, I hope eventually we'll, we'll,
Benjamin James Kuper-Smith: Yeah, , that doesn't sound like the project you're gonna do in the next two years. sounds like quite the undertaking.