Full episode transcript below. Beware of typos!
Nick Jikomes
Elizabeth Hobson, thank you for joining me. Yeah, thanks for having me. Can you tell everyone where you're calling in from and just a very short background in terms of your scientific training and what you're doing now?
Elizabeth Hobson 1:58
Sure, yeah. So I'm an assistant professor in the Department of Biological Sciences at the University of Cincinnati. So I'm, I'm currently in Cincinnati. It's becoming spring here. So that's exciting. I've got birds that are nesting right outside my window here. So in terms of my scientific training, so I did my undergrad at McGill University in Canada. And then I moved to New Mexico State University for my grad work. So I got my PhD there, working with social structure in parakeets. And then I had two postdocs after that. So I worked at the National Institute for Mathematical and Biological Synthesis, which is in Tennessee, and then moved back to a different part of New Mexico, for a second postdoc at the Santa Fe Institute, and then started my job here and started the lab in fall of 2019. So barely in the before times.
Nick Jikomes 2:53
Yeah. And you know, some of the listeners at this point, will probably be wondering why, you know, what we're actually going to talk about. So this is, you know, some kind of animal behavior person doing work in parakeets, apparently. And I want to go through all of that, and we are going to tie it back to, to things that don't explicitly have to do with non human animal behavior. I want to talk about your work with respect to humans, and the environmental changes we've gone through recently, I immediately started thinking about all those things as I was reading your work, and, and also computers and computer science and and network analysis. But briefly, can you describe some of the work in some of the animal models? You've used to study things like social behavior? You mentioned birds, but what exactly have you been doing with them?
Elizabeth Hobson 3:43
Yeah, so I think to start with kind of the the focus of the research, I'm really fascinated by the combination of sociality, and cognition, and trying to bring those together and infer something about cognition from the sociality and how individuals are interacting with each other. So I really like social species. So that that includes humans to an extent, but I've done most of my work on birds so far. So most of my fieldwork, and you know, actual, like on the ground data collection has been with a small parrot species called the monk parakeet. And they're really interesting because they're invasive, and they're super, super, super common in their native range. In South America, and several other countries, they actually call them le plaga the plague, right, because they're so common. And so this little bird has gotten into the pet trade and then become an invasive around the world, because of escaping or being intentionally released from the pet trade. And it's a really interesting model because they're, they're very social, and as a parrot, so they've got the big brains and you know, the potential for complex cognition and things like that. So I've done a lot of work with that particular species of parrot. But I also really like comparing, right? And that's, that's one thing that I started developing in the towards the end of grad school was like, Okay, well, now I know something about this particular parrot species. What does that tell us? If we compare it to what we know about elephants are hyenas or primates or something like that. And making those connections across species turns out to be, in a lot of cases really tricky. And so I really started diving into Okay, well, maybe I'll take off the blinders and not consider every animal as if it's a parakeet and and really kind of try to use this comparative perspective to see what does sociality look like for other species, and what kinds of generalities Can we start to discover from doing those kinds of comparisons?
Nick Jikomes 5:50
There's a very difficult chicken and egg type problem when it comes to thinking about cognition and social behavior and social complexity. Because on the one hand, you've got to be smart to deal with a complex social environment. But on the other hand, the complex social environment is probably a major driver for brains getting bigger and smarter and better. So how do you start to think about that, and unpack that as a researcher?
Elizabeth Hobson 6:15
Well, I think that's, that's, um, it's a perspective that people have had in the field for a while, right, like a lot of hypotheses are about this kind of ratcheting effect of you increase the sociality and you have to increase the cognition then, or you get a bump, a boost in cognition, and that allows you to develop, you know, more complex social structure. But I think there are other ways that we can get boosts in cognition and boosts in the complexity of, of social interactions, that they don't have to be tied together. And I think that, you know, for some of these species, and for some of the perspectives, the that kind of close relationship between cognition and sociality might have come from our own biases, right? Because we're social. And we have complex cognition. And so we're, maybe we're the pinnacle. Right? And so everything else is kind of going in that direction, potentially. And that really would shape a researchers perspective on how to formulate a hypothesis about how each of those potentially disconnected traits might have evolved in concert. So I think it's really important to take a step back and say, Okay, well, not everybody, not every species as a person, right. So like really thinking about their evolutionary trajectories? And how, how that might have evolved?
Nick Jikomes 7:36
Yeah, I mean, one immediately thinks about something like an ant colony, highly social, highly complex, in some ways, but I don't, you know, correct me if I'm wrong, I don't think anyone would argue that answer doing any particularly sophisticated type of cognitive behavior. It's, it's all or mostly, you know, direct sensory feedback mechanisms. And yeah, there's a complex society there.
Elizabeth Hobson 7:59
Yeah. And I think that, you know, the decision making process and ants can be very interesting and very complex in surprising ways. So there's some evidence for teaching, for example, you know, if you interpret the data in a certain way, you know, there are ants, where they appear to be teaching each other a route from one area to another area, for example. And so I think, you know, not coming into this with like, you know, a big brain is necessary, right, for interesting things to be happening is a good perspective to have on the whole situation.
Nick Jikomes 8:36
And a lot of your work in birds has focused on and and in the comparative approach has focused on dominance hierarchies, and aggression. So can you just define what a dominance hierarchy is? And why you chose to study those things in particular detail?
Elizabeth Hobson 8:54
Yeah. So a dominance hierarchy is when you can rank animals or individuals in this linear pattern, right? So you've got from the top ranked individual to a bottom ranked individual. And the fights or the aggression that they have among themselves allow you to do that kind of ordering. Where you have lots of unstructured fights. So for example, individual a up here that's maybe top ranked and a hierarchy beats B, B, C, and then B beats C and D, you know, that would be a very nicely structured hierarchy example. But you can get examples where those rules kind of break down and you get a Messier structure or the structure just doesn't, isn't even present. And so for example, if you've got individual a beats B, BB, C, but cbsa, right, that's a cycle there and that you can't easily linearly order. And so, dominance hierarchies have been studied in animal behavior and biology for almost 100 years now. And it's been a really fascinating topic, and Really interesting to see how the the history of work in that topic has progressed over the years. But I think that with the with the dominance hierarchies, often we're thinking of these these kind of linear structures. So how do they emerge? How do the individuals in a group, if they're all novel, and you throw them together? What happens? Does a dominance hierarchy come out the other end? How long does it take? What effect does rank have on things like reproduction and stress and things like that. So it can be really, really important to individuals that are living within those social groups.
Nick Jikomes 10:36
So you've done some work, teasing apart those things in the monk parakeets? So I'm interested to unpack, do they have dominance hierarchies? How quickly do they emerge? And then there's this question of how do they determine their rank? And then how do you as the researcher determine their rank by watching them?
Elizabeth Hobson 10:57
Right? Yeah, this is tricky, right. So with the parakeets and captive groups in like big semi natural flight pens, where it's much more possible to get really detailed social data than in the wild, for example, I tried that it did not work very well. But with these captive groups, they can, if you put a whole bunch of birds together, a dominance hierarchy can emerge pretty quickly. So within the first like three to six days, you can get pretty nicely structured aggression. And it's, it's easy by eye to see like, Oh, those are the top birds. And those are the bottom birds, everybody in the middle is a little bit trickier to figure out, you know, what the ranks are by eye. And so for that, and for, you know, formally establishing all of the ranks of the individuals that we use a lot of computational methods, and people use tons of different approaches to this. The field is adopted ranking methods from, from other from other fields, and also from sports. So one of the major ones that people use now is Elo rating, which was developed for chess, right? And so, it's really interesting to see how the methods have kind of come from other subfields and been incorporated.
Nick Jikomes 12:12
So it sounds like a key component of how you how you determine the rank is simply by watching and observing who's attacking whom? Mm hmm.
Elizabeth Hobson 12:22
Yep, exactly. So we just write down, you know, parakeet a, fought against parakeet B, and what time that happened. And then we can use our computational methods afterwards to come in and make sense of, of all of these fights, the parakeets are really active, and they can be very aggressive. And so there's just so much going on that by eye and observer has a real difficulty going in and saying, other than that, the obvious ones of like, Who's winning all the fights consistently versus who's like losing all their fights consistency, that middle part is really hard to, to eyeball and say like, Oh, yeah, that one's ranked like slightly above the other one. The other thing that's important about thinking about rank is are you just ranking them in order? Or are you giving them like a power score? Right, so is there a continuous measure of their the rank in the hierarchy? Or is it just the order that matters? And for different kinds of questions that that can be a an important distinction.
Nick Jikomes 13:24
So one of the things, one of the tensions here that I sense with this type of work, it's a tension that's very common in the field of behavior generally. So you know, historically, there was a lot of field work that was done by I'll just wouldn't go out into the field and watch wild animals. The advantage of that is you're watching wild animals, you're seeing real behavior happen in real, naturalistic situations. The downside is it's very difficult, you can't control when things are happening, or how they're happening, or all of the infinite parameters that are at play. And so when, you know, when I was doing neuroscience, there was this big push to do more naturalistic, but quantifiable behavioral studies in a laboratory setting. So people would have mice and rats say in an arena, but they have quite a bit of freedom to move around. And you're using depth cameras and things like that to track their all of their movements. And you're using machine learning techniques to quantify, you know, down to the level of the bend of their spinal cord on a moment to moment basis. So where does your work kind of fall on that continuum? It sounds like you're doing semi naturalistic observations with human observers. But how, how quantitative Can you really get?
Elizabeth Hobson 14:36
Yeah, so the best way to collect this kind of aggression data in our situation, because we do have a relatively large area. And we do need to identify all the individuals by site. So you know, you could blanket the entire area with video cameras, but then you would have to have people go through and code those videos and that takes forever. You know, because it's not just the location or the presence of the individual is what's happening. And that behavioral context is really hard to get from video. The the new machine learning and neural network methods that are coming out are getting us closer to that. And it's really a very exciting development because I think it'll, it'll drastically speed up the just our ability to to get this data analyzed into a usable format, and then kind of do something with it. For example, when I, when I started my work with the parakeets, there was so much going on that I actually just voice recorded my entire field season. And then I took two years to enter all that data into electronic format, so I could analyze it. So it was a huge amount of time investment, just to get to the point of, you know, is there even a hierarchy. And so what we do now is a, it's kind of a combination. So we have human observers there. And they're staked out and blinds with binoculars and scopes, and things like that, we have all the birds marked, so we can tell who's who. And then everybody has iPads. And so we're entering the data into electronic format in real time. So then we have a pipeline in the background that does all the computational analyses and spits out rank for all the birds at the end of the day. And so that's drastically sped up our ability to get the data just into a usable format. I think the real computational part in my work, and the things that I've been developing with collaborators is on the analysis side and getting more into modeling and saying like, Okay, well, this is the real world structure that we see, you know, this is the pattern of the fights. And then we can go in with computational methods and either start to break little pieces of the real world and kind of put together this parallel universe that's slightly broken, right? And then you you push that far enough. And you say, when to those two worlds start to diverge? Like, have we broken something critical or essential in our model data that makes it diverge from the real world data? So that's how we can go in and start to see are there more structural rules in the data. The other approach is to build it up from the ground, right? If you have another parallel universe over here, with a modeling approach, you build in some rules that you think are important in structuring the fights and the aggression. And then you run that world. And again, you have this parallel universe, that you can compare it to the real data and see, have we coded in enough of the real structure that, you know, we get a match between what we're seeing in the model and what we're seeing in the real observed data?
Nick Jikomes 17:44
So can you give some examples of like models you've used, and compared to the actual behavior in the birds? To give people a more concrete sense of this?
Elizabeth Hobson 17:53
Yeah, so I had had one recently come out, where with collaborators, we did the The first method where you take the real data, and you break little parts of it, and then you see, you know, what changes? And so what we're doing there is taking the real ranks of all the individuals in the group, and keeping that the same, and then saying, Okay, if we go back, and we rewire the networks of aggression, that are the things that allow the rank to be calculated into emerge in these groups, what if we change who they're fighting with, right, so we keep the individual the parakeet that was the the winner of the fight. But then we change who the target of aggression was, who it went against. And in the model, what we did is say, okay, given all the ranks, right, if we've got an individual that's ranked midway in the hierarchy, all the individuals that are ranked below it could be potential targets. And so we say, Take away the the actual target, and reconnect that edge in the network to another potential target, randomly choose between that subset. So for each individual in the hierarchy, that subset is different of who they should be fighting with, if there's nothing else going on, in that structuring the fights other than I fight with individuals that are ranked below myself, right. And then what we can do is run those forward and see, okay, if we simulate 1000 times, you know, 1000, parallel worlds, do we consistently get ones that are overlapping and are similar to the real world data, suggesting that we haven't broken anything fundamental about their, their social structure by doing this kind of rewiring of the network?
Nick Jikomes 19:35
I see. So the idea would be, you know, you could have one setup where rank and aggressive behavior are basically of the form that the a bird is always going to attack only someone who's lower than them in rank. And they're going to do that indiscriminately. So they're just going to kind of be an asshole to everyone who's less cool than them. Yeah. And then what would be what would be an alternative structure
Elizabeth Hobson 20:00
Yeah, so one of the things that we started seeing with these models is we started finding some species that were not following this, this pattern, right. And there, there seemed to be additional structure on top of just this basic, you know, loosely follow the rules of the dominance hierarchy kind of structure. And so you can imagine, if you're an individual, maybe you're fighting, and you're, you're fighting with all the individuals that are ranked below yourself, but you're focusing your aggression on the ones that are ranked maybe at the very bottom of the hierarchy, or you're focusing your aggression on your, you know, your potential challengers for your own ranks. So the individuals that are only ranked like slightly below yourself, so that layering that extra structure on top of the rules, then in the model, if we break all the connections, and we say just like, indiscriminately, just fight with anybody ranked below yourself, then our parallel universe world is going to diverge from what we actually see in the parakeets, for example, or any other species?
Nick Jikomes 21:00
I see. So staying on the parakeets for a moment, what, is there a basic? Do they follow one of those basic structures? Or does it really depend on the individual? How much variability is there in that type of thing?
Elizabeth Hobson 21:12
Yeah, so for the kind of the, the way that that I was analyzing the the strategies with my collaborators is that the majority of the individuals in the group have to be following a certain pattern well enough that we can detect it. So if you have a big mix of, you know, social dominance patterns in the group or something, then it kind of defaults back to this, like, you know, the the simple pattern of just fight with anybody rank below yourself. So you have to have enough individuals that are doing these more kind of complex patterns, in order for us to be able to detect it in the data. And so with the parakeets, the two groups that I had studied a while back, now, they both follow this close competitors type social dominance pattern. And new results actually are showing just from last season, we had a short field season that unfortunately, got terribly truncated by all the COVID stuff happens, like right when we were first starting to get data, but the very initial analysis of that are showing that that group was following what looked like a bullying strategy. And so they can, that would be bullying would be formally described as fighting or picking on someone who's much lower than you in writing. Exactly, exactly. Yeah. And so, you know, one of the things that that I really like about this kind of like big picture comparative work that, for example, we did in this recent paper, is we found that we had data from multiple different groups of the same species. And then we can go through and say, like, how consistent are the strategies? Is that a characteristic of the species? Or should we consider that to be a characteristic of that group at that time in the conditions that they were under? And so that, like, opens up these ideas of like, how much can can groups change their strategies? Do they change over time? You know, like, what affects that?
Interesting? And so? Well, I have a lot of questions on this stuff, the birds? So the answer to this question might be obvious, but I suspect maybe there's more to it than one might initially expect, when the birds are acting aggressively towards each other? Why do they do it? Are they taking food from someone of lower rank? Or is it? Is it something that obvious? Or is there something more complex and nuanced going on? Yeah, that's, I mean, it's a great question. And that's one that, you know, I've really struggled with, because in these captive populations, right, they're in a flight pen, they've got as much access to food and water as they want, they've got a lot of space where they can get away from each other, you know, the pen is big enough that they're not forced into these, you know, artificially dense groups that might increase artificially increase aggression or something like that. So I don't, I don't know that they seem to spend a lot of time and energy doing these kind of like, they're not, they're not aggressive fights were generally the parakeets are injuring each other. So it's kind of a low level, like pecking kind of thing. They're spending a lot of time doing it. And they seem to be spending a lot of time not just processing and acting as if, you know, they, they know something about their own rank and the rank of others in their hierarchy. But they also seem to be watching each other and putting together information of like, oh, he just bought B. And if I'm C, and I'm watching that I know something about those two individuals without having to actually fight with them directly. And it appears that the parakeets are doing that based on both the computational analyses and also just the observations. I mean, if a big fight breaks out, you can see a lot of the birds in the in the pen will like all fly over and they often these big fights are on the ground. Until they come running over, you know, they're craning their necks and they're just like, you know, really interested in what's going on with the fights. So it's pretty hilarious to actually watch it, you know, they look like brawl breaks out or something interesting. So they're spending a lot of time, what might at first blush seem like excessive time fighting over? Nothing. But you know, perhaps, you know, if they're consciously aware, I'm going to I'll just use that phrase, if they're consciously aware of their rank, and they care about it. How do we get at the reason that they care about it? Is there, you know, the first thing that comes to mind is do they get a reproductive dividend for being high ranked?
Yeah, I mean, I would hope that there's some kind of purpose to all this fighting, right, it would make more sense if there if there was, and I think that the tricky thing, in order to address that you really need a long term study. parents live forever. So that complicates things, and you really need it to be in the wild, right to understand the fitness effects of being high ranked. So what it looks like in captivity is it's kind of this micro improvement in their micro habitat. It's like, if you're sitting over there, and I want your perch, you know, then I get I get to take it. And so there, it's possible that there's some like short term, you know, benefit from from that kind of thing. But yeah, it's a it's an interesting society, because in a lot of other species, it's much more obvious what's going on, right? Like males will be fighting for access to reproductive opportunities, or to potential mates, right, or defending territories, or, you know, the classic one is, you know, a predator standing on top of a carcass, you know, like, monopolizing that food resource. And so the parakeets are kind of interesting, because, you know, they, they don't seem to be monopolizing, things like that. So in the wild, the monk parakeets are the only ones of the about 375 parrot species that build their own stickiness, and they build them anywhere. So in the invasive range, you know, it's really common for them to build them on telephone poles, cell phone, towers, trees, all sorts of places. And they even do that in the native range, too. You'll see them on windmills, and, and Transformers on power lines and things. And so, you know, they don't seem to be constrained in terms of where they're building their nests. Right. So that's one of the resources that animals will often fight over is, you know, areas to reproduce or raise their young. They also, especially in the winter, they forage on on grass. And they're in Argentina, for example, where I was working in the native population. It's just like giant fields of grass, right, there's no way that a parakeet could come in and be like, yeah, this is my field and like, you know, defend it from other individuals. So the monopolization part is, is a really classic part of how we approach dominance hierarchies and why they're fighting and and all the competition between individuals. But I think that's one of the things that makes the parakeets really interesting is they don't, they don't seem to follow one of those, like really classic patterns that we we come to expect from some of these species.
Nick Jikomes 28:18
So in thinking about things like aggression and dominance hierarchies, I guess my question is, is aggression best characterized as a true motivational drive. And so we should unpack these terms for people. But if you're not familiar with this language, if you think about something like hunger, right, if you if you just stop eating, and you wait, and you wait, and you wait, the hunger will build up inside of you. And obviously, we consciously are aware of that. And you'll be very, very oriented towards getting food, the longer you're deprived of that. So same thing with sleep. Same thing with the other core behaviors that are out there is aggression the same way. So if a bird in this example, simply hasn't been in a fight in a long time, are they more likely to engage in one? Hmm,
Elizabeth Hobson 29:03
yeah, that would be that's an interesting perspective. Um, I think probably not. And I think I'm basing that mostly on how you see pairs interacting. So for example, birds that are housed long term in a cage together, you know, you don't see these like wars, just like all of a sudden erupting between a pair of like long term cage mates or something. But it is a possibility, you know, some individuals or some species have kind of these bursty aggression patterns, right, where it'll be peaceful for a while, and then kind of like, fights breakout, and then it gets peaceful again. And so that might be an indication that there might be some kind of just drive. I think the interesting part to think about is it could also be a signal, right? If one fight breaks out, you've got a winner and a loser in that fight. And those to individuals might have motivations to go on especially the loser right to go on and when it's next bite, so it might start a fight with somebody else that it thinks it could beat essentially, to, you know, signal that, you know, yeah, just lost, but it just one so it counteracts a little bit. So it depends a little bit on what the animals are paying attention to and how they're cognitively processing rank. And what rank actually means. Does rank mean that you just win lots of fights? Or does it matter who you win the fights against, right, and that can change the decision making process there. But I think, especially in the case, if it's a signal, right, you have one individual that loses a fight, all the other individuals might then be attracted to try to win a fight against that recent loser. So the reason loser might have incentives then to go out and win its own fight and reset that essentially. So I think thinking about just the decision making process, and how complicated is it actually can help you understand, you know, just how these structures are organized.
Nick Jikomes 31:08
And when you guys are determining rank, so you mentioned that you're using techniques that are used in online sports for ranking, or chess, I think you mentioned. And I also read that you're using variations of the PageRank algorithm. And so it's, it makes sense when once you think about it, but it's kind of cool when you first hear that, that the same approaches are being used. So can you talk a little bit about how something like PageRank actually works for people who don't know what it is? And then how you guys are using versions of those search algorithms to actually do your work?
Elizabeth Hobson 31:40
Yeah, and I think this, this really highlights a cool part about networks, right? So networks really kind of burst onto the scene, you know, and started and a lot of started really being important in a lot of subfields. And so you have networks as graph theory. In mathematics, you have networks as social networks in, you know, anthropology and behavioral ecology. And you have networks in how websites are connecting and linking to each other. Right. And so PageRank is was developed, correct me if I'm wrong, but it was the Google guys, right, the developer
Nick Jikomes 32:15
page. Yeah, yeah.
Elizabeth Hobson 32:16
Yeah. And, you know, and that's a search algorithm to say like, Okay, well, what are the really popular pages? What should that search algorithm return? And it's all based on a network, right? If you've got pages that are linking to each other, that's the same mathematical structure as parakeets that are fighting with each other, you know, those are both networks. And so one of the things that I think is really exciting about the network perspective, and the network approach is that it's really emerging as a common language for interdisciplinary work. So we can go to something like an algorithm that's developed for search engines and say, Okay, well, we can port that over. And does it make sense to use it for parakeets, and in a lot of cases, because it's a network structure and a network structure, you can really kind of adopt tools from vastly different fields. And so what the PageRank does, to go back to your kind of mechanics, part of the question is, I really think of this as you know, you can picture it as like percolation through the network. So if you've got a network, lots of individuals are connected, and then there's a fight, or there's a link between them. And so you get some kind of power, if you're thinking about fights gets passed to that node, and then that individual has a fight, and that gets passed on. And that all collects down into four websites, the the ones, the sites that are linked, have many links to other sites that have many links to other sites, for the parakeets. It collects down at the individual that has the the lowest rank in the group, right? So it's, it's connected then to individuals that are winning against it, and also connected to other individuals that are winning against those individuals. So all the aggression is kind of percolating down and coalescing around the bottom ranked individuals.
Nick Jikomes 34:12
So the end of the day, it's it's all basically the same math, you just have different entities, it could be websites, it could be parakeets, it could be humans, you're counting literally just counting the number of things that you think are relevant. So in the case of Google, how many times does your website get a link from other websites? Probably weighted in some way towards the quality of the website. Right? So if I'm getting a link from some guy's random blog, it's maybe one point but if I'm getting a link from whitehouse.gov, it's 1000. points.
Elizabeth Hobson 34:43
Well, and that's, that's you want to think about, it's not really the quality there, right? Or like how much you trust those websites. It's how many other websites are linked to that website,
Nick Jikomes 34:54
right. That would be the proxy for trust, right is how many links it has.
Elizabeth Hobson 34:58
Mm hmm. Yeah. And this, this Example. Yeah, you could you could use that as a proxy for trust. But it's also a proxy for like, popularity or impact too. So it kind of depends, like how you twist it.
Nick Jikomes 35:09
Yeah. So you can do something similar with the parakeets going to your comparative work. And thinking about what Just tell me if this is too simplistic, but thinking about the smartest social animals like parrots and primates, and the dumbest social animals? Like, I mean, I don't even know what you would pick prairie voles or something? Oh, gosh,
Elizabeth Hobson 35:31
you're gonna get me into fights at the next conference I
Nick Jikomes 35:34
go to? Well, my question is really? how similar are the structure of dominance hierarchies across species with different levels of cognitive sophistication? Are there some general structures that tend to emerge in most species? Or is there a lot of variability?
Elizabeth Hobson 35:52
Yeah, so I think it depends at what level you're looking at. So if you're looking at the micro level, right, the the tiny little structures that then allow the dominance hierarchy to emerge and be structured in this, this kind of linear pattern, those are remarkably similar. So I have a friend Daya Shizuka that did this amazing project where he trolled through all of the old papers going back to I think 1937 was the oldest one. And people have been reporting dominance, hierarchy, data and aggression since then, and even earlier. So he went through and he blew the dust off all his old data sets, and then came in with new computational methods to kind of deconstruct the networks. So you can think of that as breaking apart the the networks of aggression down to each three individuals. And so you're just looking at these triads then. And looking at the structure and the patterns just within those triads, and then saying, Okay, how many triads have certain kinds of structures do we have in the entire network as a whole. And when you do that, kind of, like, break the network down to its its tiniest pieces, you see a ton of similarity between species that are very different. So you know, ants, octopus, fish, lizards, primates, elephants, hyenas, you know, they all kind of have the same kind of micro level structures. And so that that analysis really set the stage for the project that I just completed with my collaborators, Simon de daioh. And Dan Munster, and we came in and we took the opposite approach. So instead of like, breaking the network down to its tiniest pieces, we came in from the top and said, like, what are the highest level structural patterns that we can find with these social dominance patterns? And are those consistent across species? So we were looking for, you know, some indication of maybe phylogenetic restriction of certain kinds of these more complex strategies. So for example, you might expect that attacking your or focusing your aggression on your close competitors, right? The individuals that are potential challengers to your own rank, might be harder to figure out, who are those guys that I should be focusing on? Right. And so you might be more likely to see a close competitor strategy and species that have larger relative brain sizes. But what we found is that these strategies are and the social dominance patterns are just sprinkled all across the entire data set. And so we see them cropping up all over the place, and they don't seem to have any strong indication of any kind of like evolutionary history. I see the patterns.
Nick Jikomes 38:38
It's not like primates have one type of social strategy, birds have another one, and so on and so forth.
Nope. Interesting.
The one of the questions I have is related to change changes in behavior. So you can imagine, and you talk about some of your work changes that can be driven by the individual. So individuals start to do different things, versus the environment, the environment changes in some way. And that can actually change, you know, all of the forces, motivating individuals to act in certain ways. So how do you start to study that and think about that the impact of the environment on the social dynamics of a group?
Elizabeth Hobson 39:20
Yeah, so I'm glad you asked that, because this is the focus of some of our new research in the lab is okay, now we have all these computational methods to go in and detect the interesting structures. And we have the study systems kind of set up where we can do this kind of work. And now we can go in and say, like, okay, form your social system, and then we can essentially come in and kick it, right. And we see like, we perturb it in certain ways. And then we see, did we break anything? Does the social structure, you know, reform after that perturbation, or is it robust, right and robustness here would be, you kick it, and now it's like pretty much stays the same. and resilient might be you kick it and it changes. Maybe the for example, the para keys would change their social dominance pattern. But the dominance hierarchy is still there. And so the basics of the social structure would still persist in that group. But the perturbation would change kind of the flavor of, you know, how they're they're using certain rules to choose their fights.
Nick Jikomes 40:23
Yeah. And I feel like I should say, right here, this isn't a purely academic thing. This is actually happening to us every day right now. So if you think about your Facebook usage, or your Google usage, they are perturbing your digital social environment all the time to see how it impacts our behavior, and it definitely does.
Elizabeth Hobson 40:45
Yeah, which is really interesting to think about, too. And then also, you know, how plastic is the response? Right? If you, if you perturb things, write in an animal or in humans? Can people respond? And can people change? Or, you know, if you perturb it enough? Do you break out of you know, the robust pattern and really break things fundamental about the social structure? And so I think that, you know, some of the work that we're doing in animals is a way to get insight into potentially how human structures might also respond to different kinds of perturbations.
Nick Jikomes 41:22
Are there any before we get to humans and and thinking about our own behavior more? Are there any examples in non human animals have shifts in social structure, so some sort of environmental change happens and the dominance hierarchy shifts in a big way, or the pattern of behavior in some other capacity shifts in a big way and become stable in that new configuration?
Elizabeth Hobson 41:46
Hmm, yeah, I mean, you can have a lot of things. So you can imagine just something that changes the amount of aggression in a group, right, and like dials up the amount of aggression, that can cause the group to lose its social cohesion and kind of fracture, and individuals that are in one of these groups and in a dominance hierarchy. One of the one of the questions is, why do the lower ranked individuals stay in that dominance hierarchy where they're like not getting the benefits of being top ranked, maybe they're getting, like bullied or something. And so I think that one of the interesting things that you can do is come in and say, Okay, we're going to change the level of aggressiveness. Does that make it just totally unfeasible for those lower ranked individuals to stay in that group? And do you get the group just fracturing? So that could be an interesting way to approach that kind of thing. And I think that temperature is one way that in a lot of species, you can really crank up the aggression, right? We all get grumpy when it gets like hot and humid outside. And, you know, does that cause some of these groups to start to fracture? So another case is things like pollutants, so especially pollutants that depress aggression, right, so you've got these, like estrogen, kind of mimicking pollutants that are getting into our water supplies, is that decreasing the aggression in something like a fish to the point where, you know, the dominance hierarchy just kind of crumbles and dissolves, and everybody's kind of just peaceful around each other. And that would be that would be a major change in the social structure of how those groups are, are put together.
Nick Jikomes 43:25
So you sort of hinted at it or indirectly mentioned that the level of aggression, the types of behavior that are happening, seem to be closely related to the size of groups. So groups can grow, they can split, this happens all the time, this happens in the human world, you see this in the animal world? Can you unpack a little bit more about fission, fusion dynamics of groups? Like what dictates? What are some of the big things that are dictating group size?
Elizabeth Hobson 43:54
Yeah, so resources, for sure, is one of the major structuring features. If you don't have enough resources in an area, then what are your choices, I mean, you can dive right or you can leave. And so some of those kinds of decisions that are really fundamental level can cause these splits. Often for the fission fusion dynamics, we're thinking about more of like a shorter timescale and a much more fluid social system. And so you can imagine a big group of, for example, pair keys that are all roosting together in a colony. And so they're in a big group overnight and in the morning, but then they split into smaller groups to go foraging. And then they meet up when they're foraging and then they come back. And so you've got these kind of subgroups that are forming, coalescing, and then splitting apart and then coming back together with a different kind of composition of individuals. So often, we'll think of that as like a fission fusion type society. And that can happen on a really short timescale too. So it kind of depends on on the species, but, you know, for the parakeets there Just they're constantly seems like they're constantly in action. And so they're, you know, flying over and forming a little short term foraging group then splitting apart coming back together and a different composition. And so just tracking all of those processes can be tricky.
Nick Jikomes 45:16
And what are the trade offs for ascending and rank in a group? So it would strike me that you want everyone ostensibly wants to be of higher rank, because you get something. But I imagine there's also a cost to being higher in rank. So can you talk a little bit about stress and and things like that with respect to where you actually fall in a hierarchy?
Elizabeth Hobson 45:37
Yeah, it's a good question. And that's something that people have been working on in lots of different species. It really stress and rank interact in interesting ways. So it really depends on like, which species you're actually looking at. But you could think of it in different ways, right? If you're high ranked, and everybody else wants to take over your rank, then you're going to be the subject of a lot of challenges. So that can be a stressful situation. If you're in a society where the top ranked individual is so much more powerful than all the individuals below it, maybe it doesn't get challenged, and then maybe you just sit up there at your top ranked position, until you know, you get an injury or you die. And somebody else kind of like ascends into the rank.
Nick Jikomes 46:25
So I think why and model?
Elizabeth Hobson 46:26
Yeah. So or like the hyena model, right. hyenas are fascinating. And I think that, you know, really depends, and one of the things that would be really exciting is one of the ideas that we've been working on and discussing in the lab recently is, what if, what if there's information processing going on? And so the top ranked individuals would have an easy time because they went on their fights, they say, Yes, like, I know, I'm top ranked, I don't have any doubt about that. The bottom ranked individuals are saying like, yeah, I'm bottom ranked like no doubt about that. The middle guys, as I mentioned before, it's it's hard by AI to sort out who is ranked slightly higher than another individual in the kind of the middle of these hierarchies, oftentimes, especially with larger groups. And so what if the individuals there don't know, aren't certain about their own ranks, and that uncertainty then causes them higher stress. And so you could see these like, really interesting interactions between why they're stressed, and the information processing that's going on in the background.
Nick Jikomes 47:36
Interesting. Um, another thing that I was reading about as I was researching you is the concept of a social tipping point. And this is a fascinating subject, because well, I think it's intrinsically fascinating, because it's a radical shift in how groups behave. And that's, that's almost always interesting. But also because it immediately was evocative of the past year and human events. So can you start to talk about tipping points, what are they? And maybe let's start with some examples from the nonhuman world and what we know about how these tipping points are actually triggered?
Elizabeth Hobson 48:13
Yeah, so this is a very new area of research. So we don't have many very good clear empirical examples right now. But it's a really, as you said, it's a really intriguing idea. And so the idea here is that you can do these, these micro perturbations of the social system of a group and push it through this gradient, right. And at a certain point, when you get far enough on that gradient, maybe the social structure flips over to something completely different, right. And that was a tipping point should happen relatively rapidly, right? It should, should really do this kind of flipping motion. The interesting thing is, then if you push it the other direction, through that gradient, you should get the point at which it flips should be slightly different, or radically different. And that's the idea of history thesis, so that there's memory in the system. And then it depends which direction you're pushing the system through that gradient. And so here, you know, you might push a system, and you might push it really far until it finally flips to a different strategy. But then to reset that system, you might have to then not just push it back a tiny bit to where it flipped initially, but so much further back in order to get it to reset to the original strategy. So
Nick Jikomes 49:33
you gave an example of this using social spiders. Can you describe that? I think it'll make I think it'll crystallize this concept for people.
Elizabeth Hobson 49:41
Yeah, so there were some experiments with spiders and heating up the temperature. So like literally just cranking up the temperature, and seeing when aggression kind of turned on and when it turned off when you're manipulating the temperature. A lot of those empirical studies are now Under extra examination, there were some questions about the data quality underlying some of those, those studies. So what we need to do is definitely more empirical work to really understand, you know, are those patterns real? You know, do we really see them in in groups that are in naturalistic habitats, and how widespread is that kind of pattern across lots of different species.
Nick Jikomes 50:26
But the idea would be something like you've got a group of individuals, the temperature starts rising slowly, you get to some point, everyone gets cranky, and they start fighting with each other more, and then keeps heating up and they're fighting more and more, but then when you cool it down, when you get back to the temperature, they started fighting out, they don't actually stop, they actually keep fighting, and you have to go even lower to overcompensate for what happened somehow.
Elizabeth Hobson 50:51
Exactly, exactly. Yeah. And so if you're coming at the system from, you know, a traditional, maybe a field biologist perspective, you might only look at one gradient, and say, yeah, you know, we have this, we push them through the gradient, and this is the point at which they change their behavior. But if a different person comes in and looks at the other direction, the gradient, they would have a completely different take on when that flip occurs. And so I think this, this idea of memory in the system is a really cool one for integrating into animal behavior, because it really makes us think about the past, right? What did that animal already experience or that group? What is it already experienced? And how that how might that past interaction or past history affects the current behavior in ways that are, are not obvious? And especially if we're parachuting in and we don't know, the past behavior? How do we deal with a situation like that of understanding current behavior? And I think that some of these perturbation experiments, you know, that people in my lab are, are planning and other labs are getting excited about are looking just, you know, specifically at this of like, you know, when you change things, what happens? Like how far can you change things before? How far can you change conditions before the behavior is, you know, radically shift?
Nick Jikomes 52:14
So what are some other examples? Any salient examples of tipping points and social animals?
Elizabeth Hobson 52:20
Yeah, so we have some examples from, for example, in humans, a lot of examples in ecology. So for example, you know, changing the chemistry of lakes, until they, they flipped to a new kind of ecological condition. But we really need a lot more tests in animal behavior in order to really identify, you know, really clear examples, and also to just see how general these these kinds of patterns are. Is it something weird that, you know, only a couple species are prone to? Or, you know, I suspect that it's, it's relatively common, because this, this idea of that past, interacting current behavior is one that I think is most likely pretty common in a lot of species.
Nick Jikomes 53:10
So it sounds like there's clear examples in, in non animal systems, so like, a lake can suddenly Can you unpack that that version for people? Because I think it's an example people will be familiar with, they just don't quite realize it's an example of a tipping point.
Elizabeth Hobson 53:26
Yeah, yeah. So I think anytime you've got this, like, drastic change, it could be a tipping point, tipping points outside of like, mathematics are, it's tricky to say, you know, is it actually a tipping point? Do you have a bifurcation in the system? And that's one of the the tricky points or tricky parts of trying to study these is that the mathematics get complicated. And it is really hard with our empirical data, which is messy and not clean, like you would see in, you know, dynamical systems and the mathematical world. It's really hard to go in and say, you know, with limited data, and messy noisy data sets, is this like for sure, a tipping point or not? And so I think that some of the characteristics, the basic characteristics are these like big shifts that happen over a short amount of time. And this idea of history says that the memory is really important and the direction that you're looking at when the shift is occurring, if you can find both of those, then I think you're on strong, you have much stronger evidence of saying like, yes, this is a tipping point. But this is is that is an interesting example of taking a mathematical concept that's used a lot in dynamical systems and things like that, and translating it over for use in a completely different field where the assumptions are different. And so we when we were working on this project, we really had to dig into that and say like, Okay, well, what can we actually feasibly do on the animal behavior? Or the behavioral ecology side to detect, you know, indications of things that would resemble tipping points? If you're thinking about it from the mathematical perspective?
Nick Jikomes 55:17
How did you start to think about human behavior, because it was really natural for me as I was reading your work to start to, you know, start to think about parallels to human behavior. So, so for the last, you know, year, we've, the environment has been perturbed in a very big way. And people are very stressed. And every different way you can imagine, you seem to be seeing, in certain cases, in metropolitan areas where there is literal unrest, there's an increase in aggressive behavior, we can't measure it with the level of fidelity that you can measure something as a biologist, but but it seems to be going up, do you think that we're at risk, or potentially have crossed some kind of social tipping point in human society
Elizabeth Hobson 55:59
right now? Yeah, that's, it's, it's an interesting thing, like an interesting thought experiment. I think it's hard to say, because we're in the midst of it right now. Right. And, you know, if you were coming at this as a scientist, then you know, you would structure this with a control, right? So that you could say that these are above and beyond, you know, natural variation that you would expect, given this time in this place, and, you know, try to pull apart what are the the social effects of a pandemic, versus the social effects of, you know, racial justice and injustice, you know, like, Can we pull those apart and, and really identify that different effects are coming from different kinds of factors. And with the messy human world that we have, that's really hard to say, it's like, okay, you're going to experience a pandemic, and you are going to experience you know, this other thing, and then see what happens. And so the human systems are really hard to use that experimental approach with in the real world, so you can give them you know, computer games in the lab and see, you know, hypothetically, how would you respond to, you know, this kind of perturbation, and that gives us some insight into potentially how people would interact in the real world. But it's a it's a huge open question to see, you know, how, how are, how is this past year, more than a year going to affect things? Long term? And, yeah, we just we don't know yet.
Nick Jikomes 57:36
Interesting. So in the context of tipping points, you know, to the extent that they're actually occurring with any, with any reasonable frequency in the animal world, what would be the purpose of these in the first place? You know, what, what immediately comes to mind for me is, okay, for tipping point is really just like a phase shift. Right? Moving from liquid to solid, it's the sudden change in this case in behavior. Is there an advantage? You know, it's like, why would behavior be structured such that you could, and you often do toggle between two very distinct modes of behavior, versus just continuously and smoothly transition between behavior generally?
Elizabeth Hobson 58:20
Yeah, it might be the case that the social system is set up to be robust to changes and to perturbations, to an extent, right, and so you can push that system, you can kick it, you can perturb it, and it stays relatively stable until you push it too hard. And so it could be a sign of major things in the social system, where it's pushed past his breaking point, essentially. And the tipping point could be an indication that like, something has gone terribly wrong, and something is broken. So I think that understanding better whether that's a one way shift, right, can you push things till it breaks? And then can you push it backwards? And can you eventually reset it? Or is it like, Is it broken? And I think with behavior and social systems, one of the things that I think is interesting is that they are so responsive, and so plastic and so adaptable. And so I think that it's it's definitely possible that we can push a system, you know, past this breaking point, and then push it back and and it, it could be able to recover. But that's that's a big open question right now. But I think that that might be might be something where, you know, the tipping point, if it's there is not necessarily a feature of the social systems like a beneficial feature, or an evolutionary advantage. It might be an indication that things have gotten so bad that the system can't continue as it was before.
Nick Jikomes 59:49
What about analogs this type of thing at the individual level, so sudden changes in behavioral state things that come to mind for me are, you can imagine the fight or flight response. bahnsen an individual, you're perfectly calm and anxiety free, you see something dangerous and you're suddenly transitioning to a new state of heightened arousal. There's clearly history rhesus there. And also things like, you know, people who study animal foraging behavior, have things like explore versus exploit models, right, where an animal is going to look for food locally, and then relatively, suddenly take some time to move and go explore new area, and then suddenly, again, transition to actively looking for food. And it starts to feel like that there actually is an adaptive advantage at the individual level two, being able to suddenly switch between two modes like that, rather than just sort of like halfway doing things.
Elizabeth Hobson 1:00:42
Mm hmm. Yeah, definitely. And I really like your like fighter flight kind of example, there, right like, and that's like to dig into that a little bit. That might be you know, you're going along, you're very calm, and you get a stress response, right, you get frightened, and then you know, your heart is racing, and you're out of breath, and you need to really calm back down quite a bit more in order to go back to like peaceful state. And so I think it's interesting to think about, you know, the individual reaction, like you were saying, it might be really adaptive to be able to flip between those. And to do it more quickly. And without a whole lot of the memory in the system, right? If I can go from peaceful to running away to back and peaceful really easily, then maybe that's better for my overall stress levels, that's better for my health, maybe that's better, if I'm an animal for, you know, getting back to foraging or caring for young or something like that. So it could definitely be adaptive to be able to flip between those without a huge range of histories is in the middle, or memory in the system. Um, I think that, you know, going from individual level effects, up to the group level effects, you're looking at the, at the group level, it's more of like an emergent structure, right. And so it's, rather than all the individuals just doing their own thing, it's more of how the individuals are interacting with each other. And so it just becomes this kind of, you can think of it as a network or, you know, an emergent property. And a lot of times we think of that as kind of, you know, more than the sum of what the individuals are doing, there's something else going on in the system.
Nick Jikomes 1:02:23
So, we've talked a little bit about humans, and you know, humans are complicated, and it's really, you know, you can't control all the variables you would want to control. Is there anyone doing research on human behavior? Perhaps using something like, data from a social network?
Elizabeth Hobson 1:02:40
Oh, yeah. I mean, there's, there's tons of work on networks in humans. And that goes back quite a ways into, gosh, I think people were using networks even in like the 60s and the 70s, to understand interactions between human groups. And so you know, some of the really interesting applications now, or, for example, Elizabeth Brooke is doing some really cool work with dating networks. So she's looking at messaging patterns on these online dating platforms. And like who messages who, you know, what are the structural features that, that describe that at a network level? And, and doing these kinds of ranking methods based on the messaging patterns? And so I think you can apply these kinds of network methods to lots of different human interactions.
Nick Jikomes 1:03:33
And I mean, is there any insight we have from that type of work into the question of whether are humans special in any way in terms of our dominance hierarchies? Or do the patterns we see in other animals? Are they also the patterns that we see in humans?
Elizabeth Hobson 1:03:46
Yeah, that's, that's a good question. I mean, I haven't gotten human data and applied, you know, the social dominance pattern type analysis to it to it yet, I had a project again with Simon to do and Dan Munster, who I worked with on the animal project, where we had people coming into the lab, and they were fighting virtually over these, with these computer games. And then we were trying to see, you know, does that kind of structure still emerge in human groups that are doing the virtual fights rather than, you know, like being in there and actually fighting with each other? So I think there's, there's some fun ways that you can think about adapting some of the methods over into humans. But you can use the same kind of approach and the computational methods to understand, for example, like editing patterns on Wikipedia, right, like who reverts who could be a sign of aggression, in some cases, who interrupts who in a meeting can be a sign of aggression. with people, you really have to look at lots of different kinds of things that could be aggressive or microaggression type events, where the parakeets, you know, they're obviously they're coming over and they're bashing each other. Picking each other You know, with humans, sometimes it can be more subtle. And so I think that looking at human interactions can really be interesting for translating that over. And I think Twitter is just a fascinating example, also, of the kind of like the Twitter fights that breakout, you know, and you can see examples of this where people with a huge follower count will sometimes, you know, be very aggressive towards an individual that's maybe new to Twitter or has like a very small follower count, and not a lot of power, right. And in the animal systems, we'd call that bullying, right, a high ranked individual attacking like a very low ranked individual. What does that high ranked individual? What do they get out of that? Is it just like, you know, some kind of rash that they get from being aggressive and like winning a fight? Are they signaling to their followers that they are like an aggressive person and can win these fights? You know, the, the combination of what does the individual get? And what are they trying to signal to other individuals, I think is just a really fascinating connection that we can we can think about, both in the animals but also with the humans.
Nick Jikomes 1:06:12
Yeah, I, I'm interested in this concept of signaling. You know, oftentimes, it certainly appears to be true. And I'll, you know, ask you to describe if this is actually demonstrated, or known in more detail, but a lot of times behavior, human behavior, it seems, it's, you know, the behavior is not really the behavior, the behavior that someone engages in, is really the signal to someone else. And there's probably lots of examples we could go through. But how common do you think that is? where someone does something not for the sake of doing it, but for the sake of communicating something else to someone who's watching?
Elizabeth Hobson 1:06:47
Yeah, so I think in a system where the watchers are important, that should be really common, right? If it if you gain status from, you know, being seen to do something, or, you know, you gain prestige, or popularity or something like that, and you have these watchers that are attentive to those kinds of behaviors. You know, in the animal world, we think, a lot about attentiveness. And recognition. So are the watchers watching, right? That's a fundamental thing. But are the watchers watching? And do they recognize? Do they have individual recognition? Where do they know like, who's who in this kind of situation? Or are they just watching and getting context, a fight was just one, right? Because then they can process the information in different kinds of ways. So you can almost think of that as like a filter, right? Of how individuals what kinds of information, individuals can actually incorporate and retain, then really constructure, how they're able to move forward and potentially use that information in different ways. So for example, if your your system, it's just important to win fights, or to look like a dominant individual, then it doesn't really matter who you're fighting with, or who you're gaining popularity or strength from, it doesn't matter. It just matters that you you did those events. Whereas if it really matters, like who you're interacting with, and there's more of this kind of network approach, then you know, it, you'd have to be much more selective in terms of you know, who you were interacting with? And also, who is there in the audience, right. So we see audience effects, for example, in many animal species, where individuals will interact with each other differently. If they're there, or watchers present, then if they're alone, you see this with humans all the time, too, right? Yeah. And that's an indication that the audience is is important in some way. So I think digging into that more also is a really exciting area of research.
Nick Jikomes 1:08:53
Yeah, I've had a few instances in my life where I'm, for professional reasons, we're doing something, there's a group of us, and the camera goes on. And now you know that you're, you know, that there's people that are going to be watching and you can literally see people shift their behavior, sometimes in very drastic ways. It's kind of funny. So one of the things I'm interested in, too, is changing social rank. So it's almost true by definition that an individual will want to rise in social rank for some reason, I imagine that there's a difference between being at a particular rank and ascending to that rank versus descending down in rank. So can you talk a little bit about the the effects on physiology and other aspects of behavior for the individual, when that individual goes up or down and rank? And then I want to get into maybe some of the brain networks that mediate some of these responses.
Elizabeth Hobson 1:09:50
Oh, yeah, that'll be exciting. I don't know much about the brain network side of things because I've been so focused on the social interactions and the behavior so it would be really exciting to dig in. That more. But in a lot of species, what you'll see is the signals will change. Right? So if an individual ascends in rank, then often if there's a signal of rank that is signaling to other individuals, so in birds, this could be like a brightly colored feather patch, in fish like cichlids. This is a drastic change in the coloration. In some birds, it's like a kind of like a fleshy crown that gets vascularized and gets much redder when they ascend in rank. So there are a lot of different ways that animals can update their signal about like, Oh, yeah, like I'm top ranked now. And that can happen on different timescales, too, which is really interesting. So, you know, for example, birds, they will bolt, usually most species will molt once or twice a year. And so they only have those opportunities to update their signal. And so if rank is changing on a faster scale than that, then in that kind of system, their signal would be out of sync with their current rank. And what you'd expect over evolutionary time is that that signal wouldn't wouldn't be very informative anymore. And you might see that signal, kind of not be used anymore degrade out of the population. And so I think that, you know, the timing of how rank is changing, interacting with the signals is really interesting. But I also, you know, to get at your other point about how does that happen? How does this rank change happen over time, some animal systems have a queueing system, right, where you've got individuals kind of lined up, and they're waiting for the top ranked male, in some cases, to die off. And then in a very orderly fashion, everybody bumps up one rank. In some cases, you have the loss of the top ranked individual. And that can lead the group to just fall into chaos. And individuals have to just fight it out, again, to see like, okay, what's, what's the new dominance hierarchy that's going to emerge. And so animals can do deal with these in in very, very different ways, depending on the species. And I think also the conditions that they're under, you know, if it's a resource rich environment, maybe rank isn't as important for them, it's not super, super critical for them to be top ranked. But in cases where resources are really constrained, and for example, only the top ranked individual is the one that reproduces, then there is a lot more pressure on the individuals to try to ascend in rank.
Nick Jikomes 1:12:28
So it sounds like in many species there are, you know, something happens, you change rank, so your your rank increases, there's a physiological change that leads to a morphological change, that is purely a consequence of the physiology. So it's not like the animals just strutting around more. But there's, there's really physical changes that cannot be faked. And that strikes me as a key thing, right? You can't fake something like that.
Elizabeth Hobson 1:12:52
Yeah, so there, there are some, like, really cool experiments where the experimenters go in and they say, like, okay, let's fake it, and then see, like, what happens, do the other individuals know that it's fake? Do they treat that individual differently. And in some cases, you know, if you fake signal of rank for an individual, the other individuals will treat that individual as if it is higher ranked or lower ranked, then then it actually should be, in some cases, they can kind of like, eventually they figure it out. And then that individual can be subject to a lot of aggression, if it was displaying like a signal that is much higher, indicative of a much higher ranked individually than it actually is, then they get a lot of aggression potentially. So there is, there are costs to a mismatch, like an intentional mismatch, or a cheater, in these kinds of systems. Sometimes that interacts with what the individuals know about each other. So there's some fascinating work in sparrows, where you can go in and with strangers, you can manipulate the signals, and then they'll respond to that updated signal, as if that individual is higher in rank. But if they know each other, and you change one signal, the other guy is not fooled. And so keeps interacting with that same individual, the known individual, as it was, you know, had his previous signal. So there, it's just like, you know, it's a fascinating example of like, what do they know about each other? And how are they How are they using that information to make decisions about how to interact and how to behave?
Nick Jikomes 1:14:28
Yeah, that seems like a key thing. One would want to be able to detect if you have the cognitive apparatus for when Is there a mismatch between the the outward display of something and the actual, legitimate behavioral state of that animal? So someone's acting, or displaying somehow visually that they're of higher rank, but they're not quite acting that way. It's actually a huge opportunity for the people that are right around that rank for real.
Elizabeth Hobson 1:14:56
Yeah, to maybe test things out. It's like are you should you really be That rank, like maybe I'll challenge you a little bit more and see, you know, if I could take over, for example. So yeah, that could definitely be a strategy that the individuals of us have kind of like testing the waters or, you know, doing these, it depends on the cost, too, right? Like, if you're testing the waters and saying, like, are you really ranked slightly above me, and you know, I go and try to fight you. And then you know, you respond really drastically. And you know, the the second in individual gets really injured in the fight or something, there can be cost to doing these kinds of testing the water type behaviors. And so, depending on the social system, and depending on how common those drastic responses are, that lead to injury and things like that, they would be weighing the costs and the benefits of doing those challenges in different ways.
Nick Jikomes 1:15:53
Interesting, and how have you, you know, shifting gears a little bit, one of the challenges in in the current context, as our environment has shifted the past year has been? Well, a you mentioned, your research has been impacted. But I also imagine that your teaching has been impacted. And I know that, you know, everyone with kids, or who's in college, or whatever, is having a challenging year, because we've suddenly had to shift from what we're used to this completely new environment of online remote learning. And so anything interesting that you've been doing in that regard?
Elizabeth Hobson 1:16:26
Yeah, so I had a crew, you know, in the field doing parakeet research, and we had to just like, add an almost a moment's notice, just evacuate everybody and close everything down. So that was, that was really sad. Last spring, but everybody safe. So that's good. What I did with the lab, you know, it's a, it's a new lab, that was our first field season that we were collecting data on this, going back to work with the parakeets. So as a real disappointment, but what I did is just do this drastic pivot and say, okay, all the people in the field, we're gonna do computational stuff. Now, we did it all online. And I really got everybody, you know, into these computational questions. So we shifted over to thinking about, you know, okay, we can't collect new data this season. So are there modeling approaches? Can we get at the same questions with a kind of more of a computational angle, and we had a cool paper come out of that, where we were looking at, with the limited parakeet data that we had, our question was computationally are two similar looking types of aggression, that we coded as distinct behaviors in the field season? Were those functionally the same? Could they be considered interchangeable? So we did this big computational analysis to to kind of test for that, and also to build a framework for other people to test similar kinds of questions. So it was really fun to take the empirical question that we had, and then shift that to this tractable computational problem, and then get everybody trained up in computational methods, and work together on this this big project. So it was a really good turned out to be a really good lab bonding experience. The lab is really new. So it's a it's a brand new group that was just brought together relatively recently. So it was a good opportunity for everybody to get to know each other, while they're building up the skills and things. I really carried that forward. So a lot of the computational stuff that I do, is really amenable to this kind of like virtual research environment. So I, for example, had a lot of undergrads working with me in the fall to analyze movement data from a penguin population at the Cincinnati Zoo, which is just down the road from me. And they had all this data on their penguins. And I came in and said, Oh, have you thought about doing something with like, the social behaviors and the social interactions, and they had been looking mostly at swimming. So this was a cool project started by Katie Calafate, who I'm now collaborating with. And she was looking at how much do each of the penguins swim. And so what I did is come in and say, like, Oh, we can use all this stuff to look at the social interactions. And that got a lot of the undergrads really excited. And so they joined this group that I put together to help them learn how to code and learn how to do the social analyses. So we did that all virtually. It worked out pretty well. And I really used it as a test case for, you know, transitioning some of my courses. You know, I'm teaching a course this semester on how to learn how to code for undergrads that have no experience with coding, and we're doing it all virtually. And we've got, you know, the penguin data again, to kind of give people the the excitement of working with data from a real species. That's curious matter. And to also get them through the learning curve. And the frustration of learning how to code is like, you know, when they get frustrated, I can always say, but the penguins Don't you want to learn about the penguins, you know, and they come back and solve the problems. So in some cases, it's working out really well. In other cases, I mean, we did have to delay, you know, building up the two other study systems that we're working on in the lab. So that's, that's stuff that we were hoping to have up and running last spring and last summer, and that's still we're hoping maybe this summer to get that up and running. But yeah, it's it's been an interesting experience responding to the pandemic, for sure.
Nick Jikomes 1:20:43
So it sounds like you've I mean, you guys rose to the challenge. And and you adapt to your circumstances. When you're teaching people who don't know how to code, the undergraduate say, where do you start? I this is an area where, you know, on the one hand, in the average person can't write computer code. But on the other hand, it's becoming more and more important, such that, you know, I almost think of it like learning how to write at this point where we should be teaching children how to do this, in some cases we are. So how are you approaching? A group of people like that, that has no experience? We're like, how do you? How do you get them started? How do you keep them motivated?
Elizabeth Hobson 1:21:19
Yeah, well, the penguins are a great way to keep them motivated. Right. So that that's been incredibly helpful, actually, to have kind of a cohesive topic and a cohesive kind of set of questions that I'm trying to get the students to be able to address with the coding approaches. I think that, at least for me, the way I've approached it is to try to make it less abstract, right, like a lot of the intro to coding courses for, you know, a biology major would be biostatistics, right, which is combining learn to code with like, learn to stats, and that combination can be really tricky. Because if you get an error there, the student doesn't know, is it because I missed a comma? Or is it because I don't understand a T test or something like that, you know, it's just not, it's too abstract in some cases. And so what I've done with the students is try to really start them on things like plotting, right, so if you plot a data set, and you say plot it in green, and it plots in blue, like something has gone terribly wrong, right. And so it's a really immediate feedback on something that's, that's much more tangible. And I think that's, that's been a really nice place to start, especially with students that, you know, are not only not experienced in coding, but are worried about it, right. And like, they know that coding skills are important and can be super critical for moving forward in their careers. And as a skill set, but they are extremely stressed and scared, in some cases of getting into the coding. And so I think just making it tangible, and making it a friendlier introduction, is really helpful for getting people involved in coding that that might not necessarily like go out and try to learn it themselves.
Nick Jikomes 1:23:12
And do you think I mean, do you think something like learning to code, at a basic level is something that's so critical now that it should be part of like standard public school curriculum?
Elizabeth Hobson 1:23:24
Yeah, I mean, I think so. And I think that, you know, we're realizing that in the field of biology, as we get more and more types of data that are collected automatically, we're running into a situation, like you were saying before, of, you know, like starting out as a field biologist, and then all of a sudden, you're, you have a big data problem where you're drowning in data. And if you, as the researcher don't know how to deal with big datasets like that, or, you know, can't really even manage the data, then it's really hard to find patterns in the data and, and figure out what's going on with your system. And so I think, yeah, it's becoming becoming really critical. I think, also, you know, with the, there's constant development of new methods. And some of the Open Science platforms are much better at really kind of being the cutting edge, right? I can write a package and post it and people can use it tomorrow, right? They don't have to wait for a company to be convinced that that's like a cool thing. And like 20 years down the road, like integrate it into their, their, their package or something. And this is something that, you know, I ran into in grad school was, you know, I was a pure field biologist until about halfway through grad school, where I really wanted to get into the network analyses and was realizing that the out of the box methods that were available at the time, were only asking, you know, like my stage zero question, right? It wasn't getting to me The questions that I consider to be interesting. And so I had this realization at one point, I was like, oh, gosh, I'm, I'm gonna have to learn how to code. And that was, you know, for me was was the the trigger that, you know, really started me on this computational journey is just getting backed into the corner and be like, No, I'm going to ask these research questions, because they are interesting. And even if I had to learn how to code in order to do it, you know, that motivation was strong enough to get over that initial learning curve. So I think for students, especially they're they're starting grad school, often they're running into the same kind of mentality where it's like, I'm being forced into this, because my question is interesting, and I'm so excited about it that I will, I will do this, even though it wasn't in the plan. And I think for the undergrads trying to get them some experience earlier than that, rather than like, waiting to the point where you're backed into the corner, right, is, is really helpful for just the learning process, and making that that learning curve more more approachable.
Nick Jikomes 1:26:01
I see that makes a lot of sense. So what what are you working on these days? What are the latest research questions that you're asking?
Elizabeth Hobson 1:26:08
Yeah, so with the the parakeet system, we're working on the the perturbation studies. So my crew is, is going back to do more work with the parakeets. This spring, if everything goes, Well, we've got our COVID safety plan in place. And hopefully, they're able to do things safely. But what we're going to do is, is start to perturb the system and repeatedly perturb the system in different ways. And then see, you know, how does it respond? Does it consistently respond in the same way to different kinds of perturbations, or do different kinds of ways that we're kind of perturbing the system result in different kinds of changes. So that's been a really exciting part to kind of get up and running. In the lab, we're gearing up for bringing in bobwhite quail to do social experiments on in the lab, which will be really fun. So I have a student, Sanjay Prasher, who's going to be focused on the social learning and who learns from who? And does learning affect your position in the network? If you are kind of the producer of information, does that make you more popular in your network? So we have some experiments planned where you're going to try to manipulate you know how, how much individuals can have access to information about things like foraging, and how that changes? How other individuals are interacting with them.
Nick Jikomes 1:27:38
And what do you say? What's sort of the why behind your research efforts? So if someone asks you whether whether it's a grant agency, or it's just a curious person, you know, they might ask you, why are you doing this research in these species? what's the what's the why behind that?
Elizabeth Hobson 1:27:54
Yes, I think it's, it's the really fascinating combination, combination of sociality and cognition is like, I really think that trying to understand what animals know, right, how they process that information, and how that affects their decision making can help you understand individual behavior on a day to day basis, group behavior on kind of the scale of day to day two lifetimes, evolutionary trajectories. So I think that that's a really exciting question of just like, what do they know? And how can we detect it from how they're interacting with each other? And so I think that this is a really neat area for lots of research.
Nick Jikomes 1:28:37
Yeah, are there any final thoughts that you want to leave people with about, you know, anything on this topic, whether it's social behavior, and trying to anticipate the ways that can change things to look out for as we interact with one another?
Elizabeth Hobson 1:28:51
Yeah, I think, um, you know, once you start looking for aggression, you see it everywhere, right? You can see little microaggressions and I think taking a step back, and in human system, saying, why, why is that person doing what they're doing? You know, are they signaling? Is it me that they're signaling to? If I'm watching this happen, you know, is that what what's going on in like, the broader fabric of the social world is, is a really nice way to kind of step back from from some of these systems and say, like, Do I want it to be this way? Or can I be the agent of social change? And can I be the one that disrupts the system and changes it for something maybe that, you know, is is a friendlier situation to be in. But you can also see aggression in lots of animals. So people have gotten especially in the pandemic, much more excited about watching just what's going on in their backyard. So you can see, you know, different species of birds interacting and fighting and aggressing all the time and I think taking a step back there and saying to like, Why are they aggressive? And what's the purpose of this? And this trying to try to understand the drama of the social interactions and how people exist within their social worlds?
Nick Jikomes 1:30:12
I actually did have one more question that I forgot to ask, which is, you seem to see hierarchies everywhere in nature, when you look at social animals. And, you know, it's sort of a fuzzy question, but a, are hierarchies inevitable, or dominance hierarchies inevitable? Do you basically see them in all social animals? And are they This is the big part? Are they good? Like, should there be some hierarchy? Or should we, you know, it's very natural to think of them in almost a negative connotation?
Elizabeth Hobson 1:30:42
Yeah, I think for the inevitability, you'd want to think about how you're analyzing the network, and the the aggression, right? If you're using one of these network approaches, in some ways, you know, you can always come up with continuous measure of an individual's position within the network, and use that to rank individuals, even when the hierarchy is like, very poorly structured. So trying to think about, like the computational side of things. And, you know, how convinced Are you that there is a hierarchy that this is like a real thing is is a good question to ask. whether or not they're inevitable, is, is a tricky question, too, I think, you know, in some cases, it can be helpful to have structure. So that increases the predictability of how for animals how individuals are interacting with each other. And that increase in predictability could cause like greater group cohesion could lower stress for everybody, and reduce the risk of, you know, challenging the wrong individual and getting, you know, injured for that or something. And so there, there could be an increase just overall in the predictability of the social world, and that could be a benefit.
Nick Jikomes 1:31:59
Interesting. Well, thank you for joining me, Elizabeth. I think your research is fascinating. I'm definitely gonna look forward to seeing more of it come out in the coming years. Cool. Thank you. Yeah. Thank you for joining me.
Elizabeth Hobson 1:32:11
Yeah. Thanks for having me. This was this was really fun to dig into all this
Comments