Organisms aren’t the only products of the evolutionary process. Cultural products such as writing, art, and music also undergo change over time, subject to both the constraints of the physical environment and the psychologies of those who make them. In recent years, the study of cultural evolution has exploded with new insights — revelations into the dynamics of how culture is transmitted, how it mutates under different pressures, and why some forms are remarkably resilient and stable across time and space. Just as in biology, patterns in the structures of our artifacts converge on universals and diverge to meet the needs of their distinct environments. Certain forces ratchet up complexity in culture, whereas others act like gravity and draw the works of different societies into shared basins of attraction. Finding the fundamentals behind both the unity and the diversity of cultures, and what cultural evolution does and doesn’t have in common with biological evolution, is a field of rich mystery. New research into structural and cognitive constraints on culture leads us into some of the most fertile questions known to science…
Welcome to COMPLEXITY, the official podcast of the Santa Fe Institute. I’m your host, Michael Garfield, and every other week we’ll bring you with us for far-ranging conversations with our worldwide network of rigorous researchers developing new frameworks to explain the deepest mysteries of the universe.
This week we speak to SFI Complexity Postdoctoral Fellow, Omidyar Fellow, AND ASU-SFI Center Fellow Helena Miton about her work on cultural evolution — namely, her recent Royal Society Proceedings B paper on "How material constraints affect the cultural evolution of rhythm" with Thomas Wolf, Cordula Vesper, Günther Knoblich, and Dan Sperber and the Current Anthropology pre-print she co-authored on "The predictable evolution of letter shapes: An emergent script of West Africa recapitulates historical change in writing systems" with Piers Kelly, James Winters, and Olivier Morin.
If you value our research and communication efforts, please consider making a donation at santafe.edu/give — and/or rating and reviewing us at Apple Podcasts. You can find numerous other ways to engage with us at santafe.edu/engage. Thank you for listening!
Check out Helena’s SFI Page, Google Scholar Page, and Twitter Account.
Join our Facebook discussion group to meet like minds and talk about each episode.
Podcast theme music by Mitch Mignano.
Follow us on social media:
Twitter • YouTube • Facebook • Instagram • LinkedIn
If you liked this episode, you may also like Helena's appearance on the Here We Are Podcast with Shane Mauss.
Machine transcript by podscribe.ai. This transcript edited by volunteer Ashley Baird.
If you would like to volunteer to help us edit transcripts, please email michaelgarfield@santafe.edu. Thank you!
Organisms aren't the only products of the evolutionary process - cultural products, such as writing, art, and music also undergo change over time - subject to both, the constraints of the physical environment, and the psychologies of those who make them. In recent years, the study of cultural evolution has exploded with new insights: Revelations into the dynamics of how culture is transmitted, how it mutates under different pressures, and why some forms are remarkably resilient and stable across time and space. Just as in biology, pattern in the structures of our artifacts converge on universals and diverge to meet the needs of their distinct environments.
Certain forces ratchet up complexity and culture, whereas others act like gravity, and draw the works of different societies into shared basins of attraction. Finding the fundamentals behind both the unity and the diversity of cultures, and what cultural evolution does and doesn't have in common with biological evolution, is a field of rich mystery. New research into the structural and cognitive constraints on culture leads us into some of the most fertile questions known to science.
Welcome to COMPLEXITY, the official podcast of the Santa Fe Institute. I'm your host, Michael Garfield, and every other week, we'll bring you with us for far-ranging conversations with our worldwide network of rigorous researchers - developing new frameworks to explain the deepest mysteries of the universe.
This week, we speak to SFI Complexity Postdoctoral Fellow, Omidyar Fellow, and ASU-SFI Center Fellow, Helena Miton about her work on cultural evolution — namely, her recent Royal Society Proceedings Paper on " How material constraints affect the cultural evolution of rhythm," with Thomas Wolf, Cordula Vesper, Günther Knoblich, and Dan Sperber; and the Current Anthropology pre-print she authored on "The predictable evolution of letter shapes: An emergent script of West Africa recapitulates historical change in writing systems" with Piers Kelly, James Winters, and Olivia Morin. If you value our research and communication efforts, please consider making a donation at santafe.edu/give and/or consider rating and reviewing us at Apple Podcasts. You can find numerous other ways to engage with us at Santa fe.edu/engage. Thank you for listening.
MG (2m 41s):
It's a pleasure to have you on Complexity, and especially so, after the flurry of technical difficulties leading up to this!
Helena Miton (HM):
Thanks for having me.
MG:
Today I want to talk about two pieces - one you lead authored, and one you contributed to as the third author. One will have come out in Proceedings, Royal Society B by the time this episode airs, and the other one is still in pre-print in Current Anthropology until next year. So, it's very exciting that we have an opportunity to discuss this paper while it's still, sort of, in the womb. Before we get into your research, I would just love to have you give people a little bit of your biography and your backstory. Like what got you interested in scientific research and your path into your PhD program, and to SFI?
HM (4m 1s):
I did a BA in sociology, first. And I kind of got frustrated with some of the shortcomings, to me, of sociology, at least as it was taught in the in the BA I was in, and I always had this interest into cognitive science, like since I was a teenager, and I ended up reading those books that are close to what is usually called cognitive anthropology. And I was like, okay, this is the kind of approach I want to have, and that's the kind of research I'd like to do. And I actually ended up working with most of the elders that inspired me to apply to a cognitive science program at that point.
So yeah, sociology, then cognitive science as a master’s degree for two years. And then I just followed up doing the same kind of things in my PhD program.
MG (4m 53s):
Where did you go for your PhD?
HM (4m 55s):
I did my PhD in Budapest at the Central European University, in the department of cognitive science. And it is a department that has four main labs: One is a pretty famous one for developmental psychology. Second one is the joint-action one - so, a lot of experimental approaches to how people coordinate in two tasks. Another one, is more visual perception; and kind of the remainder, was this cognitive approach to culture and communication, mixed with a bunch of behavioral economists, kind of experimentalist people, as well.
MG (5m 32s):
And then how'd you get from Budapest to SFI?
HM (5m 37s):
It was a bunch of kind of coincidences. I met Vanessa Ferdinand, who used to be a post-doc at SFI, at a conference, and she was very strongly pushing me to apply here. A couple of my supervisors were also pretty pushy about it being a good place for me, and I kind of just trusted them - but mostly after my PhD, I really wanted a pretty autonomous type of fellowship. So that was pretty good. And I've always worked between disciplines, so it was kind of nice being in a place that was really designed for it for once, instead of one where people are trying their best but you don't really get the kind of project of being interdisciplinary from the get-go.
MG (6m 21s):
Right on. So, the first paper I'd like to discuss with you is dear to my heart, and I imagine the hearts of many listeners - you know, music being such a broadly applicable strain of the human experience. This piece, “How material constraints affect the cultural evolution of rhythm", was just such a delightful read. And as we'll kind of unspool in this conversation, seems to have links to so many other projects going on at SFI and elsewhere - not only in cultural evolution, but in biological evolution. I'm just really excited.
So I would love for you to start with a little bit of the background about prior research, and what set the stage for you and your coauthors to ask the questions that you decided to ask in this paper - because this is built on like, a pretty robust literature on cultural evolution already.
HM
There's a bunch of kind of branches and threads we try to tie, so one of it is the kind of general theoretical framework, which is called Cultural Attraction Theory. And previous names, when it was first started by Dan Sperber, who was one of my PhD supervisors. It was called Epidemiology of Representations.
This label didn't encounter as much success as Cultural Attraction Theory, which you can actually call CAT for short...for pretty obvious reasons. So, this theoretical framework, it's now almost mainstream in cultural evolution, but it really wasn't when I started working with it. And I started working with those kinds of concepts during my master’s degree. So actually, by the point I was 19, and eight years ago, sorry, that goes back a while.
So this is like a first bunch of things where culture is really considered like a dynamical system to a lot of extent, and like something that is pretty probabilistic, and that you should think in terms of which kind of mutations are not going to be random, because if you get non-random mutation, that's a pretty good way to get things to be culturally stable.
Because even if you don't get some kind of good, highly faithful transmission, you're going to have a mechanism that is going to correct for kind of changes. You're going to get back to those kind of stable forms. So that's one part - other parts are kind of developments in terms of, which experiments we are able to run to mimic cultural evolution. So, this paper uses a transmission chain experiment. That's like the telephone game which you can play with kids, where one kid is going to have one sentence to say to the next kid, and so on and so on. And by the end of the chain, basically you get a very different sentence than what the first kid had to say to the second one.
That kind of experiment became like a pretty well-spread tool to study cultural transmission over maybe the last 15 years. And that was a methodology, I already wrote a review paper on, and I already used it in previous work as well, and the last thing was we wanted a kind of good domain where you could get very precise measures and precise predictions. And we were lucky to be in this big European project where we’re really encouraged to work with the different labs we had. And part of the joint action lab at the CU is pretty specialist in doing experiments with music. So, we had a proper music lab that was kind of ready for us to play with.
MG (9m 54s):
So just to clarify the concept of cultural attraction theory, it sounds basically like it's the same kind of model as the idea that there are attractor basins in a biological… I mean like if you turn an evolutionary fitness landscape upside down and you drain things into it, right. It's like there are paths through that landscape that are more probable than other paths, and that these transmission chain experiments are basically like running massive parallel fruit fly lineages in a laboratory.
HM (10m 34s):
That’s right. It's kind of like that. I mean, the idea is pretty similar, so we definitely cannot run as many participants as you can run fruit flies, like I would love to, but that's not really possible! It's pretty similar. Some of the people I’ve worked with on it definitely resist to just collapse the both of them mostly because Cultural Attraction Theory for cultural evolution is really meant to be a bit more stochastic than attractors and dynamical systems are.
MG (11m 2s):
Huh. Are you going to more detail on why that is? No. Okay. Yes, it is one of those things that's like a general theme, it seems like, in SFI - this question about how safe it is to draw these analogies, right? And then people tend to get a little defensive about their discipline, and then other people are like, really eager to just knock down the walls.
HM (11m 24s):
Yeah, I think in that case, it's also a lot due to the fact that we've really started to formalize Cultural Attraction Theory not that long ago. So, like the oldest models or the ones done by Nicolette Lydia, and they're maybe 13 years old and we're like not a big team or school. So, we're kind of still working on what are proper formalisms of Cultural Attraction Theory. And I think that's the kind of things we haven't figured out completely yet, how much we can borrow from attractors in other fields.
MG (11m 56s):
All right. So, you make a really important distinction in this paper between ecological factors and psychological factors, and cultural evolution. And I know from before we started this recording, that it's both, important to make that distinction; but then, it's also kind of important to criticize that distinction. And I'd love to hear you unpack what the difference between those two things are, meaningfully, for the sake of this particular study. And then maybe we can go into a little more detail about ways in which those distinctions start to fall apart.
HM (12m 33s):
Sure. So, maybe first - why is it important to distinguish them? I think it's just in general, if you want to study a phenomenon, it's pretty good if you have a way to parse out what are going to be your different types of causal factors that are going to be at play with it. So that's definitely one first point, and that's definitely part of the things I really like in Cultural Attraction Theory, is it's a pretty good framework and methodology to start playing around with different types of factors. It's pretty flexible for dance. So psychological factors, or cognitive factors, or the ones usually related to how the mind works, roughly; and ecological are pretty much everything that is going to be outside the mind.
So this is a distinction Dan started to make I think in 1996, and we've just like tried to work on it a bit more, but it's definitely to be thought of as a continuum with things that are kind of, more in the mind, and others that are more in the environment. An example of psychological factors would be, for instance, the paper I got published earlier this year on the spatial composition in portraits.
MG (13m 41s):
Oh yeah. Talk a little bit more about that, because that's actually a really cool paper too.
HM (13m 48s):
I really like it, because it's also very simple to explain. A lot of people tend to look at profile-oriented portraits of pictures differently, after I told them about the story. So, it's pretty fun. For bunch of reasons, humans tend to prefer and find more aesthetically pleasing, compositions if you have an agent that is oriented in profile: you would have more space in front of the agent than behind. So, we just wanted to check out whether that's actually true in European portraiture. And we collected data from Wiki Art from and Art UK. Actually, that's a bit makes that as such that - from the 15th to the 20, almost 21st century - basically, it's a bias you can observe in a lot of paintings. And another nice thing is, it's actually a bias that becomes also stronger over time. So, you can see just the frequency how much eccentric, how much more space you’re putting in front, rather than behind your character, you're starting to have over time. So again, that's it for like one example.
MG (14m 51s):
I wonder if that's related to, you know - I used to do dinosaur field work, and in Wyoming, occasionally as we're driving out to the dig site - this is a kind of a lark - but we would find these pronghorn antelope, but they would never run in front of the car! They would race field vehicles, but they would never get in front of us because presumably, this sort of questionable evo-psych hypothesis about this was, that they're used to being chased by American lions and cheetahs in their evolutionary history. And so, they didn't like having something behind them, but they liked outracing something! And I wonder if something similar is at play in the notion that we visually prefer there being space in front of the subject, because like, the notion of like standing with your face to the wall is aesthetically unpleasant to us.
HM (15m 43s):
There's a bunch of reasons that have been advanced - In the way we wrote the paper, we're pretty agnostic on what is the good reason; or just like, there's a bunch of evidences and hypothesis that have been proposed by people in the literature - we're not taking a side on it. Part of it is, actually being able to parse out what is the interior and posterior side of an agent - is very useful to predict its action. And that’s something you can find in a lot of animals, actually. So, you know, not even related to them being a prey or not, that's kind of independent of being chased - but you would still need to know, what is the most likely direction of movement for most of the other agents you’re going to encounter in your environment.
So that's one. Related to that, is the idea that it's better fraction anticipation. If you actually have more space in front of the agent, because this is where you would assume they might move - and kind of the last line was about gaze following. Because humans are particularly good and interested in following the gaze of other people. So, if you want to follow the gaze of a human that is depicted in profile, it means you would need to have more space in front of them.
MG (16m 51s):
Interesting. Yeah. There's so much there on also like, what makes a particular piece of cultural media popular, right? It's like the gaze following that.
HM (17m 1s):
It's like in that case, you have several factors that would all play in the same direction. That's pretty nice because you don't have anything that would contradict or like, push in the other direction. Although, one factor that at least for some of the history of it played a role, was conventions for centering agents in portraits, which made sense. And one of the things we have evidence for is that, whenever you get this norm to be relaxed, your agents start to slide off. And so, you get this centering effect that increases. So, you can kind of counteract those natural, or just psychological, biases with explicit cultural norms. But if for some reason, your norm just weakens, then you're going to get more strength on the psychological bias side.
MG (17m 51s):
Awesome. So, to come back to the drumming paper!
HM (17m 54s):
Sorry, that was a big parenthesis!
MG (17m 55s):
No, no, no, no. We love non-linear tangential conversations here! At least, I think we do. And no one has told me otherwise yet! So, again, this is a paper that's built on a really interesting, robust literature on cultural divergence and convergence, and musical production. And you talk in this paper about prior research that's [revealed existence of statistical universals]? I think a lot of people are familiar with the notion that people can recognize certain kinds of music like lullabies across cultural lines. So like, what else are you standing on for this study, and how are you trying to differentiate this particular piece of research from what's already been established?
HM (18m 38s):
I think one of the main differences - a lot of the previous studies are actually kind of from out “in the wild”, the data; this is an experimental study, which is actually fairly unusual for me. I am, I tend to be, also like large cultural data sets. So that's the first thing - like the approach and the methods we use are pretty different. There was another experiment that used the same methodology to get to universal characteristics as well, but that's not what we wanted to do. We wanted to try to have, actually, kind of different traditions within the lab. So we had a between-subject-with-different-conditions paradigm, and yeah, that design just allowed us to put different constraints on those different conditions.
You can just see them evolve in different directions. So that was one of the main things. The other thing is, in terms of experiments, usually people start with very complicated things that [they have] their participants to reproduce. We started with the simplest thing we could think of, which is basically a metronome. So, that has also implications in terms of which direction you predict your movement to go.
MG (19m 45s):
So, getting into the meat here, after a lovely appetizer, this is a really interesting and clever experimental design. And I think it has a lot of implications for, you know, anybody listening who is thinking about the design of environments, or of like, software interfaces. And maybe that's getting a little ahead of things, but I would love for you to talk about how you actually structured this experiment, and then the hypotheses that you intended to test with it.
HM (20m 16s):
Do you have an example of what kind of answer you're expecting? Cause that's like a very broad!
MG (20m 21s):
Yeah. So, this is a study about laying a drum set out in different ways. And so like, you know, it's interesting - I mean, just from my years in, in music, there is remarkable convergence and universality in the like rock and roll drum kit, you know, but since the advent of highly modular and customizable midi setups, you’re seeing massive personalization of people's interfaces for musical control. And it seems like this is an in-the-wild example of how you were setting up this experiment - and it's the hypothesis that you confirmed, about how setting up a tool kit environment changes.
MG (21m 9s):
The affordances change the options that we have, and then changes the way that we navigate those; and the kind of cultural products that result from that. So like, what were you hoping to find out from this experiment? And then how did you set this up in a way with your test subjects to produce the results that tested that?
HM (21m 30s):
The first thing is, we actually have the same physical setup for all participants. What changes is, how are we asking them to use it? And so, it's not exactly the same, but it's totally true that it changes which affordances they have, and what is going to be easier to produce. I think in terms of “out in the wild” things, there's something even simpler about music, but the way guitars can be tuned is also one thing that makes some melodies much easier to play than others. So, I think it's, it might have been kind of much older traditions in music making, to arrange instruments in a way that is actually making whatever you want to produce easier.
I don't know if you want predictions for instance, but I guess what this kind of things would predict is that people are first probably going to set up in a way that makes it easier for them to produce what they want to. But that might also mean it's going to be harder to produce other types of productions that are not the ones they had in mind when they figured out this setup.
MG (22m 38s):
There's a piece of this that's about where the complexity and musical patterns is coming from, here, that I think is really interesting. And I'd love to hear you speak to that.
HM (22m 50s):
Part of the previous studies tended to focus on an increasing complexity; but measured as, from moving from a random kind of sequence, to one that is more ordered - which, you know, depending on which version you have, of how to measure complexity - it might actually, it won't be an increase because for instance, for description length, it's going to be actually a decrease. If you get something that is ordered, it's easier to describe than something that is fully random. So we actually wanted to show that if you have this kind of mismatch between the physical setup, or your affordances and the very simple reason, you're going to be pushed in this region where the rhythm that is easier to produce is also going to be a more complex one.
And I think maybe I should precise something about, the whole experiment was really thought of as a proof-of-concept, and can we obtain those kinds of effects? So, that's definitely one of the effects we just wanted to prove is actually possible - that if you're stuck in a kind of suboptimal level of complexity, you might be pushed towards more complex versions.
MG (24m 2s):
So basically, the setup is the same. The layout is the same, but you're manipulating the order in which the drums have to be hit. So, people are having to move more or less in order to play the same pattern.
HM (24m 16s):
Yes. So, they're all start having to do the same pattern, that is this metronome. And then we have four different conditions. In two conditions they are, I think, only one type of distance to cover between pads. And it can be either small for the one of the condition, or large for the other one. And the large distance is basically twice as large as the small one. And then we have two - what we call the unequal movements conditions - which are a mix of those two types of distances. So, it would be either covering like a small distance, another small distance, then a large distance; or the other way round, starting with the largest distances, and then having small, small distances.
MG (24m 55s):
And so, there are five hypotheses that you test in this paper - and this is really cool, I thought, because this really shows just how careful and granular you are as a researcher, and how delicately you manage to differentiate between all of these different effects. So, could you get into the weeds here with me on that?
HM (25m 20s):
Sure, sure. Part of the reason we have a pretty detailed hypothesis is, all of the study was pre-registered - and that's something I do on pretty much every study I lead. It means you're going to have some kind of record of what were your hypothesis, methods, and analysis plan, before running the study.
It usually is considered to be a good way to avoid [having] people just kind of “bricolaging” their results. And I think it was a pretty good exercise here, to figure out what kind of effects we wanted to observe exactly, and what we were predicting. One of the reasons we just have so many hypotheses, is because there are hypotheses that pertain to comparison between two of the conditions, and hypotheses that are more general.
I'm not sure in which order that I put this on the paper anymore. One is, this general divergence between conditions. Because you have those kinds of different physical constraints in each condition, we predict that over this passing of generation, they're basically going to go into different possible rhythms. So that's one; and then there's kind of more nitty-gritty ones about, what are going to be the specific adjustments you’re going to find in each condition? One is that, the two conditions that have equal movements are still going to be able to remain around this kind of metronome-like rhythm. Whereas, the ones that have unequal movements are going to move into different types of rhythm, but they're going to not be those kinds of [?] of regular beat anymore.
And then you have like, specific, between the two equal movements. So, if you're doing just small movements, we were predicting what is called the inter-onset interval. That's that, the time between two beats was going to be smaller than if you have to do only large movements to reflect, again, what are those physical constraints for the two unequal movement conditions. Because you have this one distance that is larger to cover than the two others, and it's not in the same place in the two conditions, we were predicting that you would get this larger inter-onset interval at different points in the sequence they were asked to repeat.
And the last thing is actually, that's just something that people would usually predict about most of those transmission chain experiments - that people would become, just make less mistake in coping rhythm, because the rhythm they are copying should be more adjusted to the constraints that they were in.
MG (27m 50s):
And that's the one that caught you off guard, right?
HM (27m 54s):
Yes. That's the one that didn't work out. Yeah. That was slightly surprising. So what we assume, or what's our hypothesis we cannot really test, is that we might have just been at a, some kind of threshold, of how bad people are at coping that - because we were using naive participants, so it might just be that they cannot get any better at that.
MG (28m 18s):
Which actually, that's a good loop out to talk about how you recruited people for this. And you know, you had a kind of an interesting anecdote about that recruiting - I am totally leading you into this, but yeah, like how did you find the people for this, and what kind of people were you using in this study?
HM (28m 41s):
Our criteria for recruiting people, we wanted them to be right-handed, because we need them to have at least that part of constraint - similar and not varying - whether you were going to use your right or left. And we recruited them mostly through the system we had with the joint-action lab, because we had that database, it turned out. We didn't manage to find enough participants in that database, so I ended up recruiting people off Reddit. You do what you have to do to finish your PhD! So, we asked all of them to have no experience in music - just ask to not have any formal training, or having played an instrument.
And yeah, we've managed to find 120 of those people.
MG (29m 29s):
I'm sure there are more of them out there.
HM (29m 31s):
I hope so.
MG (29m 33s):
So, you know, again, I am one of these people that errs on the side of consilience - and I read a paper like this, and to me it sounds a lot like when you think about the way that certain errors seem more likely, because of the distance between two genes, and like the way that copying errors happen as stuff gets shuffled around on chromosomes that - this is not necessarily, and again, this, this sort of gets to that question of like - where you draw the line with so-called ecological constraints. Because some of them might be actually like, inside the organism, right? Like [what] are usually [called] multi-scale systems.
And so I'm curious - just kind of more speculative question - what links you think there might be - or what insights do you think this suggests - in terms of, how cultural evolution is related to biological evolution, and how that is linked to like, for example - I think I sent you the totally surreal and terrifying, I think it was in Popular Science - article about convergent evolution and crabs; and how like, different groups of crustaceans keep evolving into crabs! And it's like this, you know, this thing about, they're just sort of easier pathways through the evolutionary landscape.
I don't know. What are your thoughts on all of that?
HM (30m 60s):
One thing about this distinction between ecological, and what's inside or outside the organism - maybe that's a funny side note - but, we actually had pretty vivid debate between the co-authors of this paper; because, you still do need a mind to perceive the environment, and the environment is always mediated through your body and your cognition still -- So, that's something we haven't really solved. I'm not gonna lie - We just went with what is making the point come across easiest! So, that might not be a very good answer. I just have to say, I'm not a biologist by training. So, I would tend to refrain from having very strong opinions on how it works for genetic evolution - but there's definitely convergent evolution both for biology and culture.
MG (31m 48s):
And then sort of relatedly, there's this question of divergence and the emergence of new complexity, right? And so, the sort of path-dependent histories of different lineages, and gets to the heart of one of the most - what is a core question in both studies - of biological evolution, and cultural evolution; which is - Where does all of this diversity come from? What are the constraints that lead to more complexity in an organism, or in a civilization over time? So, I mean, maybe you might have the same answer to that question as you did the last one, but…
HM (32m 26s):
Yeah. I mean, that's a very broad question! So, there's a bunch of links for sure. Yea, I think it's mostly, you have very different constraints and problems between biological and cultural systems. And one of the things is for culture, is actually much harder to maintain stable types because the mutation rates you have are like, very different from the ones who get in biological systems. So, in terms of, you know, more complexity science kind of concepts - that that would mean you would kind of get into error catastrophe pretty quickly. The kind of enigma for culture is more, How do we maintain stability? Whereas I guess for biology, you have a bunch of mechanisms that we know are able to maintain stability, because DNA transmission is pretty faithful. It's just like not even the same order of magnitude, in terms of errors you get in cultural systems.
MG (33m 17s):
That said, I think this is the right time to give a shout out to, I think it was episode 17 that we had with Chris Kempes, because he and I talked in that episode about his research into copying times for genetic information in cells as they scale; and how you get the advent of multicellularity is like a response to the threat of an error catastrophe, as the amount of DNA required to regulate the contents of the cell gets bigger and bigger, as the cell is growing. And then, you get to a point where it actually makes more sense to have a multicellular organism. And so again - like I think about this in terms of one of the caveats that you give in this paper - is that, most musical production in human beings is in groups now.
And I wonder if again - I'm being totally irresponsible here! - but I wonder if this kind of divergence and increase in complexity that you're seeing in this study, is like watching the beginning of an evolutionary transition into collective musical-like practice and performance. Like, you're watching the length of the code start to grow, until like eventually you just need a form a band.
HM (34m 38s):
I like the idea of that. I think people who were, you know, much more involved in debates on what are the evolutionary origins of music, have other hypothesis as on, why music making is often happening in groups. And then, even if those people tend to disagree, mostly it's like either because music is a good signal of your ability to coordinate within groups; so, it would have emerged in the context of inter-group warfare; or, because it's simply a good way to bond with other people. So, you have those kinds of other hypothesis, on why you would get music-making in two up, and in groups, more than alone.
MG (35m 14s):
So, where else do you see examples of this kind of dynamic and action in cultural evolution, where ecological factors lead to different stable cultural items from the same seed?
HM (35m 29s):
I'm not sure in the real world, [that] it happens exactly from the same seed, like that. That's the part where, you know, that's kind of an artifact of having done that in the lab; then you get this kind of path-dependency in the real world that you just talked about. One domain where you're very likely to observe it - is cooking - because your availability of raw products is very different, depending on which environment you're in. But there are other things just like availability of eating sources, that have impacted a lot of where cooking, and like eating your food, would happen. So, one of the places and ways it could occur is, if you have actually hot springs around - because you can just like put your food in some kind of [container] and use the hot spring, or whatever water is warm nearby, for you to cook your food.
To some extent.
MG (36m 22s):
Another example would be the availability of different spices, right?
HM (36m 27s):
The kind of ingredients you have is also going to impact which kind of food you're going to be able to prepare. And I guess there might have been also major shifts with foods, and staples being introduced, in different places.
MG (36m 39s):
So, no offense to my ancestors, but this has to do with the notoriously bland English cuisine, you know - just it's like it's too far to reach! That explains kind of why they went to such an effort to establish transoceanic trade routes. Right?! Anyway, we're getting way off the subject. [Laughter]
HM (36m 58s):
There is a Lovely book called Consider The Fork, about different, just like, the story of cooking and cooking instruments; and the author, at some point argues that British food was probably not as bland as it looks, because it was cooked in much smaller pots. So, it was actually not as over-cooked as the timing on the recipes look like, if that's any consolation! [Laughter]
MG (37m 26s):
That’s sort of an origins-of-life thing, you know - with the hot springs hypothesis, you know - collecting organic molecules, rather than … anyway!
HM (37m 34s):
[Laughter] So you're going to have to edit this part!… but I grew up learning to drink wine, like South France wine, and like, very specific species. And that definitely has a major impact on which wines I appreciate now still! And wine is probably also one of the kind of domains, where you have very strong dependence on what your final product is based on; what is the species of grape you have.
MG:
What is it they say? The smaller, the berry, the richer, the juice - or something like that? It's like, the small dogs seem to have as many neuro-transmitters as the big dogs!…
HM (38m 12s):
I don't know. [laughter]
MG (38m 16s):
That’s okay… so, there's a very low-activation energy transition that I want to make here - into this other pre-print for Current Anthropology, that I was really excited to read about. Also, the predictable evolution of letter shapes and “Emergent script of West Africa recapitulates historical change in writing systems” - Piers Kelly is the lead author; James Winters, who fans of cultural evolution study will probably recognize as a big name; and then Olivier Morin. So, this paper feels like a natural transition - just because we're talking about the same kind of processes at play; but, about the evolution of writing systems over time.
Can you take us on a little tour of the history of the thinking about the evolution of writing systems? You know, actually, I highly recommend people read this, because this is a very rich and detailed piece of the paper - but just a 30,000 foot overview would be great!
HM (39m 22s):
I should start though just by, I liked it because this was a really dream-team to work in; and I think it's probably also what makes the paper that great. Piers Kelly is a linguist and anthropologist, and is probably one of the most knowledgeable people about the Vai script of Liberia , which is the one we were studying in this paper. Olivier Morin already worked a bunch on what kind of shape writing systems take. So, in terms of factors of attraction - we were talking before, in like ecological versus psychological - it is an amazing work at showing that you get cognitive factors about, what kind of shape characteristic; and mostly, that have cardinal or oblique orientation, in terms of strokes, a lot more than you would predict just by chance.
And that's one of the way it's really tuned to our visual system. And James Winters is mostly linguistic evolution, and cultural evolution in general. And actually, the methods we used in that paper are methods I first used on my paper on the graphic complexity of motifs in heraldry. So that's all the kind of influences that went into that paper, you know, as a start. Now, back to - what our kind of hypothesis about the evolution of writing? There's a lot of writing scholars that I've suggested, that you get this kind of almost natural history of writing, that comes from first having very complicated signs that look almost drawing-like; and having them progressively simplify. And he would also have this kind of transition from pictorial systems, to things that are more, with smaller linguistic units.
MG (41m 10s):
The example that most people would probably be familiar with is Egyptian hieroglyphics, right? Going from this, is an image to like, the hieratic script, which is on its way to just becoming phonemic letters. Right? So it's like, symbol, to grapheme, to phoneme?
HM (41m 30s):
Yes, I think it also has this kind of interesting, almost sociology of science thing, where it's a hypothesis that I think the paper is usually cited for - it is from [Ignace] Gelb I think, 1960s. And I think a lot of people worked on kind of, more localized system, would disagree with it. But for some reason, this disciple still has some kind of a ground. So there's that, and there's a bunch of work that are more on the experimental side from linguistic evolution, where if you ask people to do some kind of guessing Pictionary game in labs, you can also get those kind of simplification effects where, you know, they started doing something that looks a lot like a drawing, and they end up having things that looks a lot more like symbols.
And you can also get those kinds of simplification. If you ask them to reproduce a character from memory, like you get differences. So, if you're asking them to reproduce them with the model in front of them, you can maintain pretty complex kind of scribbles if you do that. But ask them to reproduce them from memory, they're going to simplify a lot more and that's work from the Center for Language Evolution in Edinburgh. That's actually where James Winters did his PhD.
MG (42m 44s):
Cool. So you make a really important distinction in this paper - and that's the distinction between simplification and compression. And I think that this really gets at - to spoiler alert this - this gets at, I think a much bigger issue about - I am constantly talking about physicist Mark Buchanan's 2018 Nature Physics editorial - on a bias for simplicity in physical theories. And I know that you work with Simon DeDeo; and Simon has recently done some really interesting work on the sort of, guiding like, Occam's Razor, and like, consilience; and this trend, that's not a consistent trend.
But it's interesting to note the contexts in which we move toward simpler, more unifying theories in science - or we look for this is sort of like - the ghost haunting this entire conversation - which is, When is it important to make a distinction? And when is it okay to bundle things together? In taxonomy, you've got “lumpers” and “splitters.” And so, this research gets at this question of languages. When are they getting too simple? When do the letters start to blur together? So I guess I would love to hear you talk a little bit about what you mean specifically, by talking about compression versus simplification. And then, what are the constraints that you are suggesting in this paper, about when a visual writing is going to get simpler over time; or why it's getting simpler? Like, when does it stop getting simpler? When does it have to start becoming more complex again? You know, when does it become - when does it all go crabs?! - and then, you know - When do we have to assign some different roles to the basketball team?
HM (44m 43s):
I think maybe we need to start with, What do communication system do first? Because we need to get from, What are the kinds of constraints that operate on those cultural items, before answering all of those questions. And one of the first things is, a writing system is really a set of written symbols of characters that codes for a set of meanings. So, simplification can occur in different ways: You could just lose characters and meanings, together. And it just means you have less complex stuff -- Compression means you're still managing to communicate the same amount of information, but with simpler symbols, usually. And that's the way we use it in this paper.
MG (45m 26s):
You mentioned the techniques by which you actually measured the complexity - so I guess first of all, you mentioned that you had developed some of these techniques on an earlier paper, that we'll also link to in the show notes, on heraldry, which is really cool; but I would love to hear you talk about like -because this is something that comes up, for example, in the Facebook group all the time - which is, people arguing about these different measures of complexity - and this is a bigger kind of argument surrounding complex systems research, in general - but you came up with a really finely-specified and concrete-actionable set of measures for how you standardized your data set, and then operated on it. And I think that was really cool. And then, talk about the data set too - because the language that you studied for this, is fascinating.
HM (46m 17s):
The language is this Vai script of Liberia; it was invented early 19th century. And what's interesting is, you don't get that many independent inventions of writing. So, it's not exactly independent, because part of the group that invented it had been exposed to writing in other forms. It's those kinds of almost de novo creation of writing. And there's hardly a dozen of those that happened; and so that was why [it] was partly interesting, and it's also one that is really well-informed. It's well informed also, because purists did a great job at kind of tracking all the people who had done inventories of the signs and sounds that we’re using in it.
So in that case, we had that, I think was in 1834, if I'm not wrong. And the last one is 2005 from Unicode. And we have, that appoints this kind of census of the characters in the script almost every 10 years - in between like 10 to 15 years, I think - we had like 16 time points, or something around that.
MG (47m 31s):
So, it's like geologists getting to watch an island form!
HM (47m 36s):
It's a bit like that - it’s really rare data - it’s a really amazing data set for sure!
MG (47m 42s):
Then talk about how you actually standardized - and then the measures; because there's two different measures of complexity that you're studying here. I think those are both really interesting, because that spills out into a conversation I had with Geoffrey West about fractals and the history of complex systems, and how you actually measure coastlines, and all that kind of stuff. Again, I’ve got to stop this! But yeah, there’s really, really interesting methodology here, I'd love to hear.
HM (48m 15s):
We actually mostly borrowed it from this transmission chain of symbols that [Simon] Kirby & [Monica] Tamariz did. And there are two types of measures we're using. One is called, I think usually, algorithmic. It’s really just taking your picture, compressing it as much as you can, and see what is the size it still has on your hard drive. It's kind of, a bit agnostic one to some extent. The second one is called parametric, and it's a ratio of what is your in-surface, divided by what are your parameters around it; and it was used in psychophysics experiments. So, we know it's actually a good proxy for how hard it is to remember, and kind of learn, characters.
That's one of the nice things, actually, even the algorithmic types of measures, seem to correlate to some extent with human performance - At some point, I went down the rabbit hole of looking at ergonomics studies that happened 2000’s, about - So, that's where we kind of know that usually, more complex symbols are going to be harder to recognize and to learn. That's one of the constraints you get, that should drive those kinds of simplification of character or compression of the script.
MG (49m 29s):
So, to draw that line to ergonomics, it's like, why you don't have any pretzel chairs, right?! Like, the amount of time it takes to get in and out of a chair determines how likely that chair is to exist!
HM (49m 41s):
I'm not sure! So like, in terms of ergonomics, I was thinking more about, kind of, cognitive or task-oriented performance. Not necessarily like physical.
MG (49m 50s):
Well, there's a link between this paper and the drumming paper in terms of keyboard layouts, right? And the convergence on a QWERTY keyboard kind of; it seems like it has a lot to do with the distance between keystrokes.
HM (50m 6s):
Keyboards are actually like those kinds of super path-dependence story where, you know, they were optimized from typewriters. They were made so, I think it was just easier for, the mechanism of them to still work - so it actually minimizes, like, crossing of something in the machinery, in the first place. That's why they don't actually make a lot of sense for typing now! And fun fact, I have an SRT keyboard as a French person, not a QWERTY one. So, I'm pretty sure if I'm trying to type quick, I'm going to make different typos than you would on the QWERTY keyboard. That's this kind of, your material environments is going to change.
MG (50m 43s):
Because it was locked in, but I guess, it was originally a perimetric complexity. Right? Because you said, it's like the bars inside the typewriter.
HM (50m 52s):
Yeah, I think now the optimal would be like the Dvorak, probably.
MG (50m 57s):
Yeah, I had a friend who changed his keys out to the Dvorak.
HM (50m 59s):
One of my supervisors did.
MG (51m 4s):
So, what were the predictions that you tested in this study?
HM (51m 8s):
We wanted to test whether you find this general simplification of characters over time. And then we were interested in the what's like, Which characters in the distribution are actually going to change most? So, our idea was, if you actually get a system to be more compressed; what is going to change is mostly, the more complex characters are going to become simpler, because they're just like, harder to process. So, you have constraints to get them to be simpler; but really simple characters wouldn't get any simpler. If they're, you know, already at this kind of optimal, between being able to distinguish and remember.
MG (51m 45s):
So that would mean that overall, between 1834 and 2005, when you're looking at the evolution of the Vai script, that it ends up kind of becoming more homogeneous, right? Like in the complexity of different letters.
HM (52m 2s):
Yeah, that's the consequence if you get only your more complex letters to become simpler, but your simple letters kind of stay at the same level. It means you're just gonna be, have less variation, in terms of how complex your characters are. So, it gets more homogeneous.
MG (52m 18s):
I think again - like I kind of jokingly alluded to this, with the imaginal crab scape - but you know, when we had Jennifer Dunne on the show, in episodes 5 and 6, talking about trophic networks, one of the most interesting things about her work is, how conserved those food webs are in completely different ecosystems, across like, hundreds of millions of years. And you know, it really suggests that the niches have been established; that when you have an ecological collapse, like a mass extinction, then there's a moment. And we talked about this in the Transmission series; also in Episode 29, about how like, mutation rates go up after a mass extinction, and things, you know, radiate to fill the lost niches.
So this kind of work suggests something similar, which is that the amount of changes - the degree that you see a visual language changing and complexity over time - has to do with the number of characters involved, and like, how much differentiation there has to be between those things, in order to be recognized as distinct characters. And do you think that this is a fair analogy that like, basically for the same reasons, or for similar reasons to like, an evolutionary radiation in a collapsed food web? - you know; that there are, in a really thick, dense language - like Chinese written language, where you've got like hundreds and hundreds of characters - There's less room to move?
HM (53m 57s):
I actually have some work, but like not out yet, on that - You're going to need to invite me again, like mad! and then scannable like and say - but yeah, you definitely get some kind of pressure from distinctiveness; actually in another case study we did. And, we actually also did it with Olivier Morin, about distinctiveness in - is also on the heraldry. And we did this kind of model to show that people kind of, apparently, try to maximize distinctiveness, in terms of which combination of features they put on their coat of arms. You kind of get those frequencies of each type of coats of arms that are more based on the frequencies of their elements, than by copying them.
MG (54m 44s):
Heraldry is a really interesting example. And, and here I'm stepping completely off the cliff into something I know nothing about, but like - you know, because it's a symbol made of other symbols, then it sort of seems like it loops us back to questions around the evolution of syntax, and how you are able to generate distinctive sentences, rather than having to remember a bunch of unique components, right? A complex family crest seems a whole lot easier to assemble than a completely unique letter, you know, that’s supposed to represent your house - right?
HM (55m 26s):
To be fair, what we did was using fairly simple coats of arms only! So yeah, a word that I said was basically, pretty much all coats of arms that could be kind of coded down as, one color, another color, and one motif or, you know, type of shape. So, it doesn't say much about like, very complex ones.
MG (55m 47s):
So, all three of your hypotheses were confirmed by this study. Did anything about your results surprise you?
HM (55m 55s):
Well mostly, how well it worked - I think that that's something that people forget, but as a scientist, are also very surprised when things work! I actually, yeah, you probably picked the two papers I have worked, predictions are the most well confirmed! So yeah - I know, the portraits too - but otherwise, usually things don't really work that well! So yeah, I think we're, weren't too surprised. We're just like, this is really a beautiful example, and we're really happy with it.
MG (56m 24s):
Just out of a commitment to thoroughness here, this paper has towards the end of it, a really intricate section on the context in which these symbols could end up being compressed. And it was really interesting, just to rattle them off - and then give you the opportunity to go deeper on this stuff - You differentiate between the possibility that compression is a solution to a coordination problem, or the process of institutional standardization; or, that it has something to do with the movement of language from one medium into another medium. And I'd just be really curious to hear you give a little more detail on that, and then how you and your co-authors were able to distinguish between the effects of those different possible causes; and where you see them explaining, or not explaining, the results that you got.
HM (57m 22s):
That's like all possible explanations, but I think that's kind of the tricky part. Like, you really need so fine-grained data on a lot of things, if you want to be able to like parse all of those out.
MG (57m 37s):
So, I mean - given everything we've discussed today, What are some take-home insights that you feel that you can offer people, in terms of actually living in-the-mix of cultural evolution? And I know a lot of people are concerned in this time with the, you know, the collapse of civilization. What does your research suggest about the interplay of all of these factors? I think about people that are at work on archival or cultural preservation projects -- I think about people that are interested in pioneering new modes of linguistic communication, to try to match the complexity of our evolving media landscape.
And what advice, if any, do you have for people that are looking to sort of play with these forces, and actually act upon cultural evolution in meaningful ways?
HM (58m 36s):
I think one first thing - which is kind of more a property of culture I tend to using the way I work - is this idea that sometimes cultural practices almost go kind of out of equilibrium, or like in kind of places that are not necessarily optimal, to match the different technological or psychological factors. And it's kind of a property of it and of this, you know - [the] idea that culture operates mostly through those directed types of transformation that, with each passing generation, it kind of means it's going to get back to something that is more optimal, if it doesn't feel like it is optimal right now. I think it's a pretty optimistic - or at least kind of a bit heads off away - to see how our culture evolves.
I think it was, if it's a dynamical system that has those kinds of preferred rate; like parts of the landscape - it would ultimately move back to, yeah; it's kind of double-edged, because you can also think it's meaning, we’re doomed as well - because here, we cannot do much about it. Yeah. I think it's also just a good method or way to see the world, to try to put aside what are different factors in those cultural phenomena, especially when they feel a bit overwhelming. I mean, at least for me, I know usually understanding things makes me much calmer by them! So, I guess that's probably more my kind of take-home is, like just trying to understand, is usually helpful.
MG (1h 0m 1s):
I keep hoping that we have fewer messaging apps. That would be a beautiful instance of collapse, like that the letters are too similar to one another, right?!
HM (1h 0m 12s):
But I mean, it also has this implication of usually you're going to get what you need out of culture. Like if you have those kinds of niche for one function, you're going to have something fulfilling it. So ultimately, you know, it looks like it's not actually happening now. It's going to happen at some other point.
MG (1h 0m 30s):
I guess there's money for all of those startups. So, something is feeding all of those crabs! Well, okay. I want to thank you so much for taking the time with this and thank you so much for co-authoring such interesting research and for being so patient in this conversation.
HM (1h 0m 49s):
It's my pleasure. Really.
MG (1h 0m 50s):
Anything else that you want to offer people before we bounce? Anything you want to point people to beyond your Google scholar page and your Twitter account, which people should be following. And if you're not, I’ll find you…!
HM (1h 1m 3s):
Apart from that, not really. Yeah. Just feel free to read any of my papers. Like, I mean, right now, I'm also pretty free to come to lab meetings to talk about any of them. So, feel free to reach out in any case!
MG (1h 1m 16s):
Awesome. Thanks a lot!