Music Evolution & Empirical Aesthetics — Manuel Anglada Tort

Music can be magical, yet it is rooted in the material world and can be the subject of scientific, empirical study.

Does what we are told of a performer influence our appreciation of the performance? How do listening habits vary with the weather? How do rhythms and melodies evolve as they are passed along, as in a game of Chinese whispers?

Our guest is Manuel Anglada Tort, a lecturer at Goldsmiths, University of London. He has investigated all those topics. We discuss the fields of Empirical Aesthetics and cultural evolution experiments as applied to music.

Manuel’s website with PDFs and links to papers:


James Robinson: [00:00:00] Hi, Manuel Anglada Tort, welcome to Multiverses. Hi, thanks for inviting me here. So I always struggle to describe what it is that I do to my mum because I work on I guess a few different things. And it strikes me that, that you’re also someone who’s, um, got lots of different kind of research areas that maybe coalesce around a few themes, but how is it that you would just describe uh, I don’t know, elevate a pitch of, of, of what you do.

Manuel Anglada Tort: Yeah. Well, I have this problem as well. I’m not sure if my parents understand perfectly well what, what they do, but, um, In a very, uh, general sense, I guess I could say that I’m working in a fascinating field that people call it sometimes music psychology or the psychology of music. Um, and this is really just the scientific study of the psychological processes that allow us to create, uh, [00:01:00] perceive, process.

And respond to music. Um, so it’s really all about all the psychological, uh, machinery involved in, in this very weird and fascinating thing that humans do, which is, you know, creating music and being so moved and, and, and, uh, so moved by, by music of different sorts.

James Robinson: I think when, when people hear the phrase music psychology, it might bring to the mind some sort of like, I don’t know, means of treating particular, I don’t know, illnesses and neuroses and things.

But really, it’s much more about understanding what it is that, that makes music tick, I guess. Um, and. Yeah, what’s also interesting to me is that it’s kind of seems to sit within this field of empirical aesthetics, which is something that I’d never heard of recently. But then I see that many of your collaborators are based at Institutes for Empirical Aesthetics.

Do you have like just a kind of brief summary of what [00:02:00] empirical aesthetics means as well?

Manuel Anglada Tort: Well, I, before, before that, I guess something to say is that I think what is interesting about this, this field of music psychology is that it’s really bridging between different, many, many disciplines.

And one of them is more in the applied of things. So music therapy and how we can use this very powerful connection of, of music and the brain to treat in a very non invasive way, some, some disorders like Parkinson’s or Alzheimer’s but also, you know, help with all sorts of other transferable skills like reading, maybe learning a second language and all sorts of different, different skills.

That would still be part of, of maybe music cognition and music psychology, this part of therapy. And then another branch, as you say, is these empirical aesthetics. Um, and, and this is, uh, [00:03:00] kind of a broader field that is very interested in philosophical, big old questions of what is beauty and why are we moved when we see like a nice landscape or a pretty face, but also when we read a poem or, or when we listen to, to our favorite song.

And I guess it’s approaching this question using empirical methods, so controlled experiments that allow us to tell the causal role of different mechanisms. Um, and neuroscience and all of these scientific tools to, to kind of approach these questions of beauty.

James Robinson: Yeah. So it’s sort of less the armchair philosophy side of aesthetics, and more the cognitive science.

What are the experiments that we can do to measure people’s reactions to works of art or other aesthetic objects music, or just viewing a landscape, I suppose. And I suppose not just measure reactions, but also look [00:04:00] at the mechanisms that drive the production of those things as well.

It was really interesting looking through your various papers, just how diverse the, the things that you’ve worked on are. And I think we want to focus on the music evolution side of things, but maybe before we get to that there’s a whole bunch of things we could talk about.

Maybe let’s start with the, the quickest one, which is just like this genome wide association study. I think it’s the quickest to talk about because it’s probably where we both have the least to say. But I do think it’s a really fascinating study in itself. So maybe tell us about how genomics is being used to look at music.

Manuel Anglada Tort: I’m not sure if that’s the quickest in that it’s a very complicated study with with many authors and , i’m just one of these many authors. So I I had like I played a role in this study. But that’s not one of the studies that I’ve been really involved in but, nevertheless, it’s a very nice example of [00:05:00] this connect this interdisciplinary of, of, of music.

So in this case, there is this very interesting question of thinking about how we humans synchronize to external cue events. And by this, I mean a musical beat. So if you think about it, this is very, it’s very remarkable, right? We do it without thinking. So we, if you listen to a song now, you can, even without thinking, or even you cannot even stop it.

You will need to, with your foot into the beat of the song, or you need, you need to dance. And even though this is very effortless for us. It’s computationally and cognitively very challenging and very complicated. How do we manage to do this? How do we manage to synchronize with someone when we walk next to them or synchronize to, to a song without even thinking about it?

And this turns out, turns out also to be a very unique, um, feature of, of, of, of [00:06:00] humans. Like very few animals can synchronize with the flexibility. And complexity in which we can, we, we can. So I guess this opens this question of, um, why is this important and how do we manage to do it effectively? And why are some people better, uh, at others, right?

And in this kind of question, there is always these two big components, uh, the nature kind of, um, biological genetic, uh, factors that might explain this, that might explain why some people are better at synchronizing to, to music than others. But then there is the exposure reasons, right? It could be that just by training and being exposed to different, uh, music or being, you know, forced to learn the piano that in many years can also make you, make you better.

So tell apart the, the, the contribution of these two big forces. One very powerful scientific method is this, [00:07:00] uh, large scale, um, GWAS, uh, studies, genetic, genetic studies, where you can take data from thousands of participants. And you get data from all of their, um, genetic architecture based on, on this genetic test and try to explain whether variation in the genes in this way, explain any difference in the behavior, in this case, in the ability to synchronize.

To, to, to, to music or not, and what, um, the lead authors of this paper show is that to some extent there is certain percentage that can be explained, that can be attributed to our genetic architecture or our, or genetic, uh, variation.

James Robinson: Um, yeah, I think it is really interesting as just an example of how. These kind of, uh, genome wide association studies are being applied to, to so many different things. [00:08:00] Um, you know, it’s, it’s, it’s, it’s something that one hears about, but just to see how it’s, being used in many different fields or such specific ways, I think is really insightful.

Um, some things that stuck out for me was just the size of the study. It was like over 600, 000 people, um, that you surveyed. And it’s, and it’s because, I guess, someone managed to get 23 and me to agree to have this very simple question, which was, um, How well can you clap in time to a beat, I think. Um, and then of course, you separately did a smaller study, but still I think of maybe around a thousand people or so, where you tested if that self reporting is accurate.

And so you got people to actually take an online experiment, um, testing whether, uh, listening to a beat and then tapping it out. Um, and. measuring up there, the quality of that [00:09:00] correlation between, yes, I am very good at tapping to a beat and actually being good at tapping to a beat. Um, so yeah, really wonderful illustration of how these techniques are being used.


Manuel Anglada Tort: And if I can say something about this is that, um, my involvement in, in this paper is kind of, kind of funny in that we, this has been potentially the easiest task or the easiest goal, um, Uh, scientifically, but the one of the most difficult technically that I have to do. So, um, this involvement in this study comes from when I was doing, I started my postdoc in the Max Planck Institute for Empirical Aesthetics in Frankfurt, in Germany, in the group, uh, led by Nori Jacobi, um, who he’s been doing a lot of amazing work on, on studying rhythm production and perception in, in the lab.

And the, um, uh, when Reina Gordon from the GWAS study [00:10:00] approached Nori, the, the, the goal was we have this situation where we have all of these large datasets. And we have this big, we have this self report question that we need to validate. Do people who say that can tap in time actually can tap in time?

And do those who say no are actually bad tappers? So the goal really, our goal was just to come up with a measure to validate this scientifically. So in theory, it’s very simple, but technically it’s very complex because to collect data from a thousand participants. On beat synchronization. These are experiments where people tap in time with their finger, uh, to like a, a music or to like some external metronome, and then you measure the, the synchronies in like millisecond precision to do this online.

It’s actually very complicated because when, you know, online calls or online software and stuff, it’s very [00:11:00] inaccurate. It has all sorts of latencies and timing accuracies. So our task was to come up with this method to be able to run this large scale tapping study so we could just validate this very simple question.

James Robinson: Yeah, yeah. Now I, I’m really familiar with all these problems of, uh, using the messy hardware of reality to measure things. That’s sort of my bread and butter is, uh, Yeah. Measuring mobile networks. And so we spent a lot of time thinking about how to, how to measure latency. Um, and, uh, yeah, yeah, really interesting.

And it’s quite, I mean, it’s just a fascinating result in itself that actually people are fairly good at assessing their own, uh, abilities in terms of, uh, of, of clapping, um, you know, rhythmically. So, um, you know, that on its own is, uh, an interesting result. One thing I couldn’t figure out is, um, And we chat about this over email.

I, I don’t think you have an answer either as the researchers did [00:12:00] find that there was, uh, you know, dozens of genes, I think over 60, 60 areas in the genome that are associated with the ability to clap rhythmically, or at least. associated with a positive answer on, yes, I can clap. And I’m sort of wondering if, if some of those associations are, you know, real, but maybe there’s a few people, you know, some proportion of people are just arrogant and will think that they’re good clappers and aren’t.

And so maybe there’s also some like arrogance genes, uh, mixed in. Um, but there must be, If this study has not already been done, I’m sure someone’s looking at whether there is an overconfidence

Manuel Anglada Tort: set of genes. This is an excellent, an excellent question. And we do, um, we did think about this in many ways, because when you, kind of develop these experiments, of course, when you need to really make, make sure that you can trust this self report or not, then you, and you spend [00:13:00] all of this time and resources to collect this data.

You really want to think about trying to make sure that you can validate these. And one way, um, perhaps it’s not perfect, but one way in which we do that is that we not only correlate this behavioral tapping data with the self report, the original self report of yes and no. But we also correlated them with other measures of self report, and one of them in particular is something like how confident are you that you can tap in time to a musical beat from one to seven.

Um, another, um, and here we, as well, we see this, uh, good correlation with tapping behavior and your self report confidence, for example. Um, another way in which we do it is that we correlate this tapping data with, um, your musical years of musical training. And here as well, we see a correlation. So, um, participants who have more years of music, formal musical training also, uh, [00:14:00] show like a higher, uh, accuracy in, in their, uh, tapping behavior.

So these are like little things that we can use to build more confidence in, in trusting the results. But of course, um. It’s true that there is always this problem when we use a self report, is that there is always, uh, some bias that might enter there for different response, uh, response biases of participants.

James Robinson: Anyway, I, yeah, it’s such a lovely study and I’m intrigued to see if at some point 23andMe starts saying, oh yeah, you should, you should try becoming a drummer or something or maybe you have all the traits to, uh, to do this. Um, yeah, maybe we can talk about some of the, the other kind of, um, yeah, intriguing Things in or angles on this.

So, um, yeah, another very large study or sort of in terms of the amount of data was you looked at correlations with with weather [00:15:00] and the kind of music that that people listen to, which I thought was, you know, it’s clearly something that we all. experience, like our mood changes and our preferences change with the weather, uh, at least anecdotally.

Um, but you sort of managed to show, yeah, this is, you know, something that really does happen. And some of the results were quite, um, yeah, maybe a little bit surprising as well. Uh, so maybe run us through that.

Manuel Anglada Tort: Yeah. So to put this in, in context to try to kind of related with the first, um, study, um, is that.

In this first study, we look more at this, uh, focus on how much our musical behavior can be explained with, uh, innate factors like our genes. This second study is a completely different extreme where, um, it asks these questions of how much of our musical behavior, in this case, music [00:16:00] preferences can be explained by very broad, uh, Uh, environmental factors in this case, such as weather conditions.

Um, and this is maybe not so obvious at the beginning, but it makes sense when we think about music preferences, because we know that to understand, you know, music success or popularity dynamics in the music industry, the music itself, only it explains only a little bit of these dynamics. Right. So what makes a good music like famous or like a global heat, it’s explained only partly based on the music itself, but there, there are all sorts of other factors that are not musical that will determine, determine the success of this song.

Um, so of course, one factor could be the distribution network, you know, like how much money do the artists have to pay to radio stations and audio streaming services and, and, you know, market their products. But, [00:17:00] um, and there are many other factors, but one of one factor that has been ignored, uh so far is this idea of like our environment and this like very broad Conditions that happen in our environment like the seasons of the year or the weather and I guess This was a bit the motivation of this study was can we try to explain?

success in the music market based on something as simple as the weather conditions in this case, in the UK, which is, I think, a very nice testing ground for the effect of, of weather, because the weather here is kind of Yeah, it’s

James Robinson: pretty, the polite word would be variable, I guess, right? Like, um, yeah. So certainly there’s a lot of different conditions under which music is listened to in the UK.

Manuel Anglada Tort: Yeah. So to approach this question in this case, I use the big data approach. [00:18:00] Uh, so we don’t use actually. Any, any participants here? But what, what, what we did is that we collected all the popular music that made it to the charts in the UK in the last, uh, 60 years. So that’s a lot of music. I think we had over 20, 000 unique songs that made it to the top, to the top charts.

Um, and then we get all of this music and nowadays with all of these new technologies and, uh, machine learning techniques, we can extract music features that describe this music. So for example, for each song, we have a feature that tell us What’s the emotion that this song might evoke? Is it happy or sad?

We have another feature that will tell us information about the tempo. Is it fast or slow? Another feature that might tell us how danceable this song is. If you listen to this song, do you want to dance or not? So we can extract all of these features and then we can just see whether these different features [00:19:00] that describe the music.

change in some systematic ways over the years based on weather. Do these features of the music change if there was a month that it was very rainy and miserable compared to a month where it was very sunny? Um, do we see like these changes in the, in the music that becomes popular during this month? And that’s what we study.

Um, with these, with these, all of these songs and we indeed find very clear associations of, of weather and particular sets of, of music. So to put it, uh, to summarize it, uh, to put it simple, really, really what we find is that, um, features that reflect positive music, happy music, and, and, and a bit music, um, and music that tends to like make, makes you want to dance and so on, um, are strongly associated with, um, good weather conditions and negatively associated with, with rainy, with [00:20:00] rainy months.

James Robinson: And there was some like, particularly interesting tidbits in this study for me, which was, I mean, one was that the association was particularly strong in like autumn and winter and when, you know, when you suddenly had a nice day. In, in autumn and it’s been like rainy all around and stuff. Like that’s when the effect of, um, the weather is, is, is most pronounced in terms of, Oh, people certainly, you know, reach for the, the more positive, uplifting, danceable, angelic tunes, uh, in those conditions.

Whereas I guess if you’ve had like a month of sun, you get kind of used to it. Um, so yeah, that, that, that

Manuel Anglada Tort: was

James Robinson: a really

Manuel Anglada Tort: nice insight. This is a, um, yeah, a very good observation in that. I agree that. That’s, I think, what makes this study interesting because to some extent you would expect, right, that good weather correlates with positive, with like a [00:21:00] beat and music that makes you dance and bad weather with music that is sad and so on.

But there is this nuance in the paper that I think is very interesting, I agree, and it seems that this contrast, basically there is this, what is the effect of weather on in our behavior? So as you say. If it’s been, um, sunny for months, one day more of sun will have no really impact. But if it’s been rainy and then suddenly it’s sunny, that day, you know, we know it very well in the UK.

I mean, I’m not from here, but that day that is sunny, everyone is happy. You force yourself to leave, you know, the room and go for a walk. And you can see it here when it’s summer in the UK, there is such a big, um, a big, um, Consequence in people’s behavior. And if you compare this with Spain, where I’m from, actually, people take good weather for granted there.

So a sunny month [00:22:00] in summer probably will not create this, uh, weather effect and behavior so much because you have this literally kind of almost every month through the year. So it’s this contrast or this impact of weather and these extreme conditions that I think Yeah,

James Robinson: my wife is Argentine and, uh, actually she recommended that, um, uh, we speak actually, cause she, she came across you, um, she’s, she’s studying here at Edinburgh, um, where you came to give a talk recently, but that’s, that’s sort of by the by, but, uh, she, she doesn’t have many, uh, positive comments on the UK weather, um, and particularly not the Scottish weather.

But one thing she does say is like, it is great. Like, as soon as you get a sunny day, everyone just goes crazy. Um, yeah, there’s like none of that kind of habituation or taking for granted of good weather. Um, I thought the other, the other kind of nuance that I noticed was that while there’s a kind of.[00:23:00]

Positive correlation with, um, the nice weather and, um, happy tunes for one of a better phrase, the opposite isn’t exactly true when you have a kind of really bad day, it’s, it’s not that people go listening to radio head or something, um, there’s just kind of the same mix of music there. Um, which maybe that’s just, I, I I’m curious, like maybe that’s a UK thing because The not great weather is the norm.

So I don’t know, maybe in Spain or something, when you get that rare day of bad weather, maybe it does have the other effect.

Manuel Anglada Tort: It’s a tricky one in that, of course, there are some limitations, important limitations in this study. So that’s a good question. Correlate correlational study. So it’s very hard to really tell what you cannot really tell any causal effect here of underlying [00:24:00] mechanism.

So all, what we can do is to kind of speculate in different ways, but it’s true, um, that some, there’s something quite interesting about this because, um, as I said, Not all combinations of features correlate with weather. We only find this correlation with these features that relate to energetic and positive music.

Now, a different combination of features that reflect music that is sad and more relaxed and more acoustic, in this case, in this, in this combination of features, we find no relationship whatsoever with, with weather. Um, and this, I think it’s, Within this speculative, uh, world. It’s quite interesting because I think it says something about when do our preferences, affective preferences, when they, they might be influenced by external factors.

So it could be that weather is weather conditions is better at affecting music preferences that are positive, [00:25:00] but actually negative affective preferences when we are sad or upset. are way more influenced by individual, uh, personal factors. Like if you broke up with your girlfriend or something happened in your life.

So it could be that these choices of music based on negative or sad moments are way more based on these personal circumstances. than, than these general external, external factors. That’s one possible explanation for this finding, I think.

James Robinson: Yeah. Yeah. That makes sense. Um, yeah. And again, it’s, it’s a great illustration of what, what a creative academic or team of academics can do with, you know, these platforms that are out there already and have So much data.

Um, there’s just like great hunting grounds for, for, um, association studies like this. Of course, we should [00:26:00] say that we don’t know that it isn’t the weather that’s affected by the popular music. I mean, that’s kind of sci fi, but, uh, you know, who knows, maybe the correlation, maybe the causal effect is the other way around.

Manuel Anglada Tort: Well, and that’s one thing. But then the other thing that, um, people don’t, I’m not aware a lot when, when they do big data studies is this role of, um, algorithmic confounds or in the case of the music industry, commercial gatekeepers, right? So I’m doing a lot of work now with, with big data of music from Shazam and Spotify, and you can use this to study how.

music, uh, spreads across the world and so on. But there is always this problem of, um, is this people who are making these choices or is this little algorithm who’s telling you to listen to that because it’s sunny or [00:27:00] industry gatekeepers who like promote different songs in different times of the year or in different places in the world based on commercial, uh, factors.

So. There is a trade off of, you know, controlled behavioral experiments in the lab versus big data in that in the big data space, we don’t have, it’s always, we don’t, we will never be able to know how much of that it’s real behavior versus algorithms.

James Robinson: Yeah, that is a wonderful point and, uh, presumably something that’s going to come become more and more pronounced, uh, you know, even work like this will influence those gatekeepers and they’ll think, okay, well, you know, maybe there’s some objective reality to, to, to people preferring energetic, uh, tracks when, uh, the weather’s good and they, they might actually, even if this wasn’t in the algorithms previously, uh, This sort of work could, could actually influence what we end up listening to.

Um, [00:28:00] yeah. Um, I want to talk about one more thing before we move on to language evolution, which was, uh, yeah, again, a very different study, but on the same themes of what makes us, what influences our, um, our consumption of music or our appreciation of music perhaps. Um, and it’s, It dates back to this really interesting, uh, event on a radio station in Germany in the seventies, I think, um, which is how you introduce your paper on, on, on the topic where they played the same symphony, um, three times.

And in fact, the same, the same, exactly the same excerpts of the same symphony. So precisely the same piece of music, um, you know, same recording, um, and, uh, and so on. They pay it three times, but saying that it was, uh, performed by three different, um, Bye [00:29:00] composers, at least, and presumably orchestras. And, uh, they got hundreds of calls afterwards with most people saying, Oh yeah, like, uh, those were indeed different pieces of music and I had my preferences and so on.

Um, so this really pulled the wool over people’s eyes. Uh, just telling them that they were listening to different pieces of music seemed to make them believe that they were listening to different pieces of music, even though the evidence, uh, coming over their ways was that, no, it was exactly the same. Um, Obviously that’s not, you know, that wasn’t a particularly scientific experiment.

Um, but then you, you repeated this in a much more controlled manner, um, and, and, and, and kind of managed to substantiate those results. And again, there’s lots of interesting, uh, nuances that came out. So, yeah. Um, yeah. What, what got you interested in this effect and, uh, take us through this study?

Manuel Anglada Tort: Yeah. I love that you [00:30:00] mentioned this study because now it feels like, Yeah.

Well, some years passed since then, but this, but this is like the first study. I, this is the study that I guess helped me decide to pursue a career in music cognition and music psychology. And this was actually during my master’s at Goldsmiths. where I’m working now as a lecturer. So I came, you know, come back full, full cycle.

And now I’m, I’m working in, um, uh, as a lecturer here in psychology, but during my master’s, I work with, uh, professor Daniel Mullen Siffen, who is from Germany. And he came across this study and we kind of discuss it in one of our meetings. And I was so fascinated by this phenomenon because I couldn’t believe that that was true because when you explain it, everyone will tell you, no way.

If it’s me. I will, I will be able to tell you that it’s the same song. So we just designed this experiment in very controlled settings in the lab, [00:31:00] where people were tested in cubicles with very good audio conditions and professional headphones. And then we used, um, two different conditions, but one of them was actually a very familiar song.

That’s Jailhouse Rock by Elvis Presley. Yeah. Um, you can hear the audience clapping. It has so many distinctive features to identify that it’s the same recording. Um, and in this condition, which is the one that you would expect that most people would be able to tell that it’s the same, we created different fake contextual texts.

that provided different information about the Elvis impersonator. And this information was manipulated, um, so it evoked different prestige. So in one condition, that was, uh, Tommy something who’s like an amateur impersonator of Elvis, who has not been doing very well in his career. And then you read this information and then you listen to the original recording of Gerhard Rock.[00:32:00]

And then you evaluate it in many different ways. After this You are told that someone else, you’re going to listen to another performance from the same composition of jailhouse rock. But in this case, we have this very prestigious impersonator who work in Las Vegas and is very good and makes lots of money doing this.

And then you listen to the exact identical same recording and evaluated in the same way. And. In this study, what we find is, um, at the end, 75 percent of the participants believe that they have heard different musical pieces. But not only that, um, but also they provide very different verbal, uh, experiences of this music.

So some, so some participants would say things like, Oh, I like this piece much more than the other one because the tempo was faster or the band was more, uh, bonded and played [00:33:00] more together. Or the voice was, Uh, you know, better, better voice and so on. So people do come up with these very good explanations of, of, of one condition, uh, over the other, even though the actual stimulus itself is exactly the same.

So I think this is a very interesting phenomenon to think about beauty and music in that coming back to what we said before a little bit is that. I think the object itself is only part of the, of the overall story to understand why music moves us or why we find music beautiful. There is all of these other factors that we need to think about, like the context.

Uh, how prestigious we think the artist is, what is the aesthetic value of that Of that situation and and so on

James Robinson: right? Yeah, I think Intuitively it it makes sense and it’s probably something that has been thought about a lot in [00:34:00] Other contexts like certainly with art, you know people People will look at the price tag first And that gives them an idea of how You know how good it is.

Um, but yeah, this is just, so it is such a wonderful and yeah, quite amusing study in some ways, basically like the fact that you used an Elvis, um, track, um, and, and then it was live and it was actually Elvis playing. Right. So this, none of, none of the recordings were from impersonators all the same. And like you say, yeah, a live track.

So lots of things going on in the background that should give, you know, clues that, um, These are uncannily close. Um, and yeah, I just love the fact that, that people, people kind of judge the, just these descriptions of the career of the impersonator has such a [00:35:00] strong influence for so many people over their appreciation is just completely mind blowing.

Um, and yet in other ways, it really does fit in well with what has been seen in other areas of just, just how, how much we kind of. Price anchor ourselves or base area, you know, use various heuristics to, um, to make judgments.

Manuel Anglada Tort: Yeah. And, um, there is a, these very interesting, uh, parallelism, as you say, with other aesthetic domains. So people have done very similar studies with wine, for example, and in, in, in the case it’s interesting because. Music and judge, aesthetic judgments of music and wine in a way are all affected by the same psychological mechanism.

So there are like this famous paper, for example, that put people in an fMRI in a brain scanner machine. Um, and then they taste a bit of wine and the only thing that [00:36:00] changes is the price tag of, of how much is this wine. Yeah. Not only people, of course, like more the same wine when they think is more expensive, but also there are different areas in the brain that are being activated.

So it, so there’s difference in the price. It creates such a different firing of like different parts of your brain as well. So of course there is extremely. Relevant for marketing, you know, people of advertising and all of these, because, um, we know that this contextual framing effects, uh, can create such strong behavioral responses.

James Robinson: Yeah. I mean, and it does suggest like that’s particularly those fMRI experiments that it’s not just that people are sort of lying about their experiences to try to make themselves. you know, feel more knowledgeable about wine or music, right? And not kind of make a mistake, which, you know, there is an extent to which that that may be a confounding factor.

But the fact that different, you [00:37:00] know, the fMRI, if one equates regions with the brain with actual experiences, um, then, you know, it is quite strong evidence that people are actually having a different experience rather than just reporting the same experience differently. So, yeah, where my mind goes with this is I just wish there could be some sort of like.

You know, buffer around everything I do so that it, it gets presented to me in the most positive light. So, you know, when my, when my groceries arrive from Tesco or something, uh, actually they come in a waitrose van and in waitrose bags or something. And I’m convinced that it’s like from a, from a better quality supermarket.

Um, because yeah, I’m sure this stuff is, is real. And, um, You know, you can improve your experience somehow for free, if only you can sort of double blind yourself or blind yourself to, um, the actual origins of things. [00:38:00] Yeah.

Manuel Anglada Tort: Yeah. This is a very positive way of, of looking at this, this is kind of a study because in the past I thought more of the negative consequences, right?

So, um, what we did in this study, for example, as well, is that we compared, we studied participants with no musical expertise. Uh, versus, uh, a group of very expert musicians, which might have an average of 10 years of formal training of playing instruments. So we were interested in seeing whether this effect is reduced in the expert musicians.

And what we find is that expertise has, it doesn’t protect you from these. So expert musicians with 10 years of experience are equally susceptible to this than non musicians. And then the negative interpretation Of these is that, of course, this has lots of consequences for any serious objective examination procedure.

Like when you want to make it into like music school or win [00:39:00] like a music competition, which is all assessed by human raters with lots of expertise, but nevertheless, you know, music is like human. So this. This kind of judgment is affected by all of these sorts of contextual factors, being that the prestige, but also the, your visual appearance, your gender probably, and all of these different things that are not musical and should not matter.

James Robinson: Yeah. Yeah. I think this, this plays into a kind of more general question, which, that I have, which is, you know, to what extent does, um, empirical aesthetics and these sort of studies like affect how,

to what extent are they kind of deflationary, right? In terms of sort of reducing that sense of, of, of wonder in things, um, versus actually in some cases, maybe they [00:40:00] can kind of be inspiring and like two non musical, um, thoughts that come to mind that, that kind of push in those different directions. Um, one is, um, prospect refuge theory, you know, this theory that kind of purports to explain why we like the views from top of mountains and, and, and from, I dunno, from a glade in the woods out upon a lake, which is that evolution, evolutionarily, um, you know, there are reasons to think that we, we like to have a large view like a prospect so we can detect predators, but we also like to have it from a kind of.

Protected place, a kind of refuge. So top of the mountain is really good because, uh, you know, things got to come up to get you and you can see for miles and miles around. So you’ve got loads of warning, um, for predators. Um, and that’s kind of deflationary because, you know, all these beautiful vistas and things, if you just think of them in those terms, it’s like, Oh, well.[00:41:00]

Um, Would I really find this beautiful if it weren’t for the fact that I’m really at root, just concerned with not being eaten. Um, but on the other hand, there are, so the other study that kind of points in the other way is, um, there’s a physicist. Oh, I’ve forgotten his name. Let me just, I’ll, I’ll look it up.

Uh, gosh, where I put it., Richard Taylor. That’s it., who’s looked at fractal patterns in art. And in nature, it’s got this this beautifully titled paper, which is something like, uh, the resonance between art, art and nature. Um, and. He looks at the kind of the fractal dimension that you find you know, commonly occurring in nature.

Obviously, you get different fractal dimensions depending on what you’re looking at. Um, but he kind of suggests that that in Jackson Pollock’s art, which is also quite fractal. So [00:42:00] if you look at those kind of Paint splashes at different levels. There’s a kind of scale free nature to them. Um, and so yeah, that they have so you can characterize them with the fractal dimension.

And that, you know, the suggestion is that the we kind of have this appreciation of art because of kind of fractals in art, because they kind of chime in with with fractals in nature. And that for me just seems like a wonderful fact that actually boosts my experience of Jackson Pollock. And one might make similar arguments about, I don’t know, the golden ratio, um, and how, you know, that’s something appears a lot of nature and maybe it’s, you know, You know, and again, actually you could have two takes on why the golden ratio is, is, is, um, kind of visible so much like the kind of deflationary tape would just be all the Greeks loved it and they just put it everywhere, but the more kind of interesting and kind of inducing of wonder, um, version would be, Oh, actually, yeah, this is something that really does occur very commonly in nature.

And when we see it [00:43:00] in a painting, it kind of reminds us of all these beautiful natural phenomena. Um, and yeah, I, uh, yeah, I’m curious as to. Yeah. Just whether, um, like how the work that you’ve done kind of affects your appreciation of things in general. Um, do you find this sense of wonder or, or, or does it, or do you think, oh no, well, even if, if music experts can be duped as much as anyone, um, then what’s the point of all this, this study?

Manuel Anglada Tort: Yeah, this, this is a wonderful question, I think, because, um, I guess it does affect me because I think. Okay. So for example, I study a lot music from a very physical point of view, like what is music in terms of this sound wave and what is speech, which is just this frequencies of low and high and how this changes [00:44:00] over time.

So as a researcher in music, you decompose music in these very artificial ways. And then if you think about it, if you, this kind of thing, you can think about music as just this, sound that is organized in different ways. Um, and the building blocks are finite. You have like notes or frequencies and, um, scales or melodies that can, you can put them in different ways.

Um, but the possible combinations of all of these building blocks is completely endless. Now that the, the, if you think about it in this way, it’s very fascinating because the question is like, how is it possible that some combinations. Move us so much that make us cry or laugh or dance all night long.

Whereas other combinations completely, we hate them or they don’t tell [00:45:00] any, like they’re completely, we’re completely neutral to them. And I think that relates to the example that you were putting with, with the same is true with visual art. And I think that. When you do research on these aspects, um, there is some kind of beauty at this, um, paradox of like when you decompose it, it kind of, it makes no sense.

It’s how, how, how is it possible? But then actually when you put it together, some of these combinations work out for different reasons very well and evoke these very strong sensations. And there is something very beautiful, I think, in this process. from the building blocks to the emotion. Yeah,

James Robinson: yeah, yeah.

That makes sense. And I think certainly, I mean, just, just understanding all of the things that influence one’s appreciation of music, in a sense, there is something wondrous about that, in knowing just how complex, like, our own feelings towards, you know, Music are like, [00:46:00] um, and yeah, I think again, moving away from this thing, moving away from the idea that beauty is just something that’s objective and, um, not based on the circumstances doesn’t mean that we have to abandon the idea of beauty.

Beauty being beautiful and wondrous. Um, it’s just much more complicated in a, in a sense. Um, so I do want to talk about, and we’ve talked a lot about a lot of interesting things, but I’m really excited to talk about your work on, on music evolution. Um, we’ve talked previously on this podcast about.

Language evolution and how, um, out of a kind of structureless soup researchers have shown that, that, that grammars evolve, um, through really cunning set of experiments and, and, um, I’ll kind of refer listeners to, to, to previous episode with Simon Covey, if they’ve not come across it, but, [00:47:00] um, yeah, you’ve worked on very similar things with, with music and, and showing how.

the structure in music seems to evolve over generations. This is probably sounding very abstract to people who’ve not listened to the previous episode. So maybe you can, um, yeah, give a bit more, uh, detail on, on, on how these experiments work and what, um,

Manuel Anglada Tort: Yes. So I guess like a place to start is with this question of how do, um, how do complex cultural systems evolve?

Okay. And by cultural systems here, we Can think about language and technology, um, which have been studied a lot. Um, but also music or art at the end of the day, music is this system that has been transmitted over generations of humans for many generations over time. And this has been changing in [00:48:00] systematic ways and evolving and evolving over time.

So the music that we listen to, that we listen to today, but also the music in different cultures. It can only be understood as this product of cumulative cultural, uh, transmission. And this is a very interesting way to think about music because, uh, I guess it puts the focus on this transmission process.

So if we want to understand why music is the way it is, and why do we like the music? Why, why does this song evokes like this very strong pleasure in me and not that one to address this kind of questions. we need to think about this, uh, underlying evolutionary process. Um, so I got, I’ve become very interested in these questions and in part, again, that’s from my postdoc with Nori Jacobi, who was one of the first researchers to take this, uh, pioneering work from Simon Kirby in language [00:49:00] evolution and think, can we apply the same methods, uh, to study the evolution of So this kind of work, it all started with this, okay, language and music are similar, but also they are different.

Would kind of the same process also shape the evolution of music and can this kind of experimental paradigms help us understand how music evolves. So the idea of all of this work is to use these, uh, simulated cultural evolution experiments in the lab where people create music and you pass this music from person to person, like in the telephone game, um, to kind of simulate these processes of evolution with, with music, with like rhythms or with, with melodies and people seeing these melodies and passing it to each other.

And then we can study how. Different, uh, features and different structures emerge for the simple act of transmission, like hearing a melody and singing, singing this melody and pass it to the next, to the next person. [00:50:00]

James Robinson: So, um, yeah, if I can kind of, uh, read that back and make sure that, um, I’ve got it and listeners have got it.

So, um, for example, with the, with the rhythm example, you would kind of generate, you start out with a lot of kind of almost random or random rhythm rhythms indeed, um, like a few taps, uh, each and. You, you have, you play one of those random rhythms to a listener and they have to tap it back. You, um, record their response and kind of synthesize it.

So I guess it sounds, you know, it’s the same sort of taps as previously, but, but now the rhythm itself is matched to how they tapped. And then you take, uh, that rhythm, you play it to a second person who didn’t hear the very first random one that you had. And they, they listened to it. [00:51:00] They tap it out, record it, play it on.

And, uh, you do this many times and it, and it goes down generations. And as people listen to it, they make mistakes, um, when they’re tapping it out. And those mistakes, well, some of them get passed on. Some of them don’t. Um, and. You not only do this, uh, sort of in one chain, but you, you have many, many different chains.

Um, so you can, you know, really drill into what happens for a large different set of, of random rhythms. Do they converge like over generations on to something similar or do they stay more or less the same? Um, uh, So that’s the kind of idea of the telephone game, I guess, where yes. You, you ring someone up that you, you whisper something to them.

So we also call it Chinese whispers, I guess. Um, and then they ring someone else up. They, they say that they, they pass the message on and over time. Those things tend to, to, to get [00:52:00] distorted. Hmm. Um, yeah. So I guess is, is that the bones of, of, you know, functionally how these things work?

Manuel Anglada Tort: Yes. So this, um, that’s a, an experimental paradigm, I guess, uh, but also a framework in that this kind of artificial process that you describe, it turns out to be very similar, even though very si a simplistic, uh, a simplistic version of it, but very similar to how cool culture evolve.

So, um, you are born today and you are exposed to many cultural artifacts and products that have been generated by your previous generation. And then you are exposed to these and then you modify these in different ways. You create new things and all of these new creations will be passed on to the next generation.

So of course, this happens in large Uh, timescales, but we can simulate this with these behavioral experiments in the lab. [00:53:00] Um, and it works quite efficient, efficiently, like, uh, the work of Simon Kirby with language, but people have done it across all sorts of domains with drawings, uh, narratives. Um, and it also works, of course, in the case, in the case of music and critical to this is what you mentioned, which is that cultural transmission.

So this transmission of information from person to person. Is not perfect, right? So people always introduce systematic errors. Um, so, okay. Like naively someone could think like, okay, we are just machines who copy and pass this to the next person and we are very good at it. So in this scenario, you would find no cultural evolution or no, or no cultural change because we would be transmitting exactly the same things.

But in reality, of course, we are humans. We have cognitive limitations. Production limitations. So we introduce all sorts of systematic errors [00:54:00] in our productions to the next generations. And that’s one, one case, but also there’s the case of aesthetic creativity. Some people want to actually create something new as well.

Uh, so sometimes this systematic variation is accidental. Other times it’s actually voluntary. Someone wants to create a new mega hit. Um, but no matter which type of. Of, of, of thing is, is always, there is this addition of variation that is systematic, um, this guided variation, which over time, um, we introduced these errors that are systematic.

And over time, what happens is that these transmitted, um, products in this case, musical language, end ups converging towards our, uh, individual cognitive biases. So limits in our memory or in our perception will shape. From person to person, from generation to generation. the kind of [00:55:00] cultural systems that can evolve and can be transmitted.


James Robinson: Yeah. It’s interesting. What strikes me as a slight difference with, with, with the, the kind of explanation for how this works in, in language evolution versus musical evolution is, I think you’re right in, in, in the, your musical evolution experiments, um, the errors seem to be systematic rather than random, um, because the sort of rhythms are so small that you, that you have, they’re just like three, three taps, uh, I think.

Um, so there’s, it would take a long time if the errors were purely random and the, um, let me put it like this. On the language evolution experiments, like the, the kind of alternative explanation [00:56:00] is something like, well, that the errors that people introduce when they’re copying something are, um, fairly random, but the, they will, the next generation will be more likely to, to pass on the features that sort of map onto a kind of emerging grammar, if you like.

So if they, if there are two, um, if there are two words that get passed on that describe similar things, and it just so happens that by, by random, you know, mishearing, uh, one person at one generation gives those two words a kind of similar sound. Um, well, the next generation is kind of more likely to pass on that, um, inherited, uh, error, if you like, uh, because the two words are kind of similar.

And so the fact that they’re saying similar kind of makes sense. Um, but I can’t really think of a way of [00:57:00] explaining the way that it works in, in, in your, uh, kind of rhythm experiments, for example, except in terms of, Oh, actually people. When they hear something, they kind of want to make it move towards, I mean, maybe we should make it more concrete.

When they hear one of these random rhythms, they kind of end up evolving towards having integer ratio, um, into, into ratio durations, if you like. So there might be like, you know, two beats or two, I don’t know. One second between two notes and then half a second till the next one. Um, something like that. Um, and yeah, I can’t really think of a way of explaining that other than, uh, people really kind of want to move them towards those kind of notches.

Um, yeah.

Manuel Anglada Tort: Yeah. I guess the exact alignment between these language evolution, music evolutions is a bit [00:58:00] complicated because it depends on, you know, what Some experimental features and decisions, I guess, but I think there is, um, some extent of overlap in that, for example, many of these, um, results that we find, for example, are constrained by our, uh, learning biases and one learning bias that is very important is memory.

James Robinson: Yeah.

Manuel Anglada Tort: So the reason that people over time in language will tend to combine chunks of information, for example, It’s to some extent related caused by memory, right? Because you have limited memory. So it’s good to use comp comp, uh, compositionality or things like this, because it’s like efficient in terms of memory.

The same also happens with, with music. So I think that the realm of, of musical melodies, which is. Um, yeah, like rhythms is something I’ve been working, [00:59:00] um, most recently is that melodies are these long sequences of tones. So when you start with a random sequence of tones, it’s very difficult to remember it and to process it and represent it and then sing it.

So we also use chunking or different things that help our memory to process this, this, this. Melodies. So then we can sing them back. So one way of doing this is by using rhythm, right? Rhythm is just like an efficient way of putting tones together. So we can kind of cluster it and, uh, you know, remember it more efficiently.

Another way in which we do this is by using pitch intervals, like the difference between two notes using pitch intervals that are familiar or that sound very good to us. So, uh, an octave, what a perfect thief. We can remember this very well because it gives us a sense of stability and consonance. And again, this is, we will see the emergence of this kind of intervals [01:00:00] because we can remember them very well and they help us organize melodies around them.

So here we just kind of the root of this systematic variations is always potentially some sort of limits on memory. Right. Um, but we see these different kinds of effects. Um, but probably all kind of explain to some extent for, for memory, memory biases. And I, I like to think about, um, this world, for example, imagine like a machine or like a human that had like three times more memory capacity than we do.

In this scenario, musical systems and language systems would be very different than the, than the music and the language that we, we see. Yeah,

James Robinson: I think that’s a fascinating point. And in some ways, like it’s a feature and not a bug that we are finite. Like, otherwise we [01:01:00] might not have evolved this kind of common system for, um, you know, these, these incredible symbolic systems, which need to be very efficient in that they have sort of.

A limited number of symbols, but can those symbols can be combined in an indefinite number of ways. Um, and yeah, one of the kind of, yeah, most striking things from language experiments and from this is just, yeah, that, that is so related to the fact. I mean, it just really depends on the fact that we, um, We need to kind of compress the information down somehow.

Um, but if, if, if we didn’t have that constraint, it’s, it’s not clear, um, that there would be any kind of convergence on a, uh, a system with a grammar and which are the most kind of expressive systems. Um, yeah,

Manuel Anglada Tort: there is something very interesting now that you mentioned these. And I think that’s, [01:02:00] I don’t, I’m not sure we know yet an answer, but, um, there is like some important differences between language evolution and music evolution.

So one of them is the way in the, the kind of evolutionary goal, right? So language has to be efficient to communicate linguistic information. Like information that could depend on your survival, right? Um, so in a way you want to be very efficient you want to be able to communicate effectively Using the minimum number possible of combinations in music, the goal is different.

We have music is mostly used for, uh, expressive communication or like social situations that are not so much about communication, specifically communicating an idea specifically, we use it for socially bonding. We want, you know, some units of rhythm that allows us to coordinate. At the same time we use it for parent to infant communication [01:03:00] to kind of play and communicate with their kids.

So you need some sort of like pitch contours to be effective in this, um, communication. And I guess what I’m saying here is that these different goals for evolution, I think also are very important to understand why there are differences in these two communication systems. So with music, the goal is not always to be super efficient in terms of information units, but sometimes it’s actually being expressive, which is a bit the opposite.

It’s like, how can I be very creative with this building blocks and this information units in a way that people can still understand me, but potentially it’s not always very efficient, right? If you think about the art, like the art examples that you mentioned, or like some complex music, um, You know, it doesn’t obey the same rules, uh, than language, for example.

And that’s, I think what makes this comparisons, uh, very interesting. [01:04:00] Um, and, and, and to finish about this is that if you think about vocal transmission, like speech and language and language, this is really interesting because. Both systems come from the same physical system or a vocal system. This coordinated use of this apparatus creates speech and creates song.

And even though everything comes from the same physical system, these two systems, speech for language and song for music are very different in terms of their structures. So it’s very interesting how everything from the same instrument can actually produce these very different communication systems.

potentially due to the goal in which they make them evolve, uh, I guess.

James Robinson: Yeah, I’m going to throw out one of my favorite quotes here, which is, um, like poetry and integral, uh, lower limit speech, upper limit music, um, [01:05:00] which is Louie Zukofsky. And I, there is this kind of space in the middle where they, they meet, which according to Zukofsky is, is poetry, but I think even, you know, you could, you could say it’s rap as well.

It’s, you know, there’s a real musicality to, or can be a real musicality to the way that that language is. Is used. Um, yeah, so I, yeah, it’s, it’s a wonderful thought. Um, maybe another kind of intriguing difference between the language evolution and, and the music evolution is, um, language evolution seems to end up producing lots of different organisms or languages as it were.

Um, and, and that’s just obvious because, you know, you know, natively we speak different languages. Um, and, uh, there’s. Hundreds of different languages spoken around the world, but even in the labs, like different iterations of the very same experiment, which start with the same stimuli, you know, different chains of recipients [01:06:00] will end up converging on different languages and grammars.

Even if some of those kind of structural features are similar, like they will just sound different. They will look different. Um, and to an extent that’s true with music in that we, you know, there are different kind of musical preferences and, um, this is something that you’ve looked at. In your work, but in another way, there’s a lot of similarity.

Um, in terms of, again, for rhythms, there is a kind of preference for, um, integer ratios. The exact ratios really intrigued, intriguingly do change between, um, different cultures. Uh, but it, it, it just strikes me that there’s a bit more kind of commonality, um, there maybe. Uh, but perhaps that’s, I mean, I don’t know.

I suppose it’s quite hard to measure because obviously there’s a lot of commonality between languages, even if they sound different, you know, we all use [01:07:00] verbs and adjectives and so on. And, um, so maybe I’m, I’m drawing more of a disanalogy than that than actually exists. It’s very hard to compare the two very different systems.

Manuel Anglada Tort: Yeah. I wouldn’t dare to make that comparison, whether there is more, um, similar cross cultural similarities or differences in language versus, versus music, because it’s a very I mean, it’s a fascinating question, but a kind of a complex one, but about music, which is what they know, um, you’re, you’re right.

That there is a lot of cross cultural similarities or people call them universals, right? So all musical systems across cultures use repetition and use different scales that consists of consists of stable pitches and isochronous rhythms and so on. However, Actually, I think there is also like an incredible amount of diversity and availability in musical systems.

So, um, even though everything [01:08:00] is based with the same kind of universal rules or building blocks, the combinations that make up different musical traditions are so huge that if you were to listen to some traditional North Indian music, for example, we would struggle to understand it. Like it would be a little bit like listening to Chinese.

Um, and you really need this exposure to understand the rules and the musical grammars. to really make sense of that music. And you don’t even have to go across a different musical culture. You can just think about musical styles. Like, many people have experienced probably the situation of trying to listen to classical music and being frustrated because you don’t like it.

And I think classical music, in a way, is like learning a language. You need to do this effort of like listening to it, and someone has to explain you a little bit of the structures. And then over time, you start understanding different beats, and potentially at some [01:09:00] point you will really understand it and really enjoy it.

Um, but it does require this, uh, yeah, getting familiar with the rules and the, and this, the vocabulary and so on.

James Robinson: Yeah, that’s a that’s a beautiful thought. Um, yeah, it was so one thing I really enjoyed from your your rhythm, um, experiment, um, was just how clear some of the kind of variations were in terms of if you run one of these kind of, um, telephone game type experiments, people converge on quite Yeah, I think that’s you know, there are similarities.

So there’s this kind of integer ratio thing, which actually I do find the integer ratio really fascinating. Just the fact that it’s an integer is not obvious why it should be that we chunk things that way. Like we don’t, you know, going back to the golden ratio, right. We don’t. That, that is not an inter ratio, uh, inter ratio that’s like, uh, irrational number, right?

So like [01:10:00] phi, um, or, or like another common shape that we see in nature is, uh, you know, circles. And again, that’s, you know, pie, which defines, um, you know. Important properties of circles. That’s not an integer ratio either. Um, so sort of visually a lot of aesthetics is kind of explained in terms of irrational numbers.

So I do find it just amazing actually that that musically, um, there is this kind of appearance of integers. Um, but yeah, by, um, going back to my original thread, um, yeah, so for example, like K Dobe musicians from Uruguay have quite distinctive rhythms and anyone who’s listened to K Dobe will kind of recognize, uh, that and when they play this, this, um, repeated, uh, game quite quickly, you know, I think it’s like just maybe seven or so iterations, they tend to converge on.

[01:11:00] particular, uh, Candombe rhythms. So, they might converge on a much simpler rhythm as well, but, um, they’re much more likely to converge on a Candombe rhythm than, than, say, a, um, North American, or, or even a Uruguayan, so, Candombe is big in Uruguay, um, um, Uruguayan student who has lots of international exposure, which, yeah, speaks to just how complex, um, cultural studies are because you can’t just pick someone from Uruguay and expect them to, uh, like candombe or, or, or, or kind of really, you know, have an affinity to those rhythms.

Manuel Anglada Tort: Yeah. Yeah. So I think you are referring to this recent paper, uh, led by Nori Jacobi, who, um, um, is who designed, I think, this iterated learning experiment with tapping. And, um, I think this, this paper that I’ve been involved, um, in is, is, is a very beautiful, um, [01:12:00] insane kind of massive piece of research in that Nori put together this team of, I think, over 30 researchers from around the world, and they are all given the same setup to collect data using these telephone games with rhythms.

And this is really the only way. Um, to tackle, you know, these questions of nature and nurture that we started our interview with, because this allows, well, what, what, what he finds in this study is that, yeah, there is this to, there is this to, um, extremes, like on the one hand, you, you see this cross cultural similarities, like rhythm, people, people produce rhythms that are strongly biased towards this integral ratios, uh, simple rhythmic categories, um, And this we see cross culturally.

Um, however, we also see at the same time, a lot of cross cultural variation that depends on what you’ve been [01:13:00] exposed to. Um, I think the Uruguayan, uh, Uruguayan drummers, uh, is like a great example, but also we also see it very clearly in Europe where you have like people in France and then very close by in the Balkanic, uh, area with Greece or, or, to, um, Turkish, uh, traditional music.

Which in a way is very similar, but actually it uses more complex, uh, ratios and you can already see people that are very close by, but exhibit these rhythmic variation. And of course this is due to lifetime differences in this exposure, um, to, to rhythm. So there is this, we come back again to these, like these always to understand the complexity of musical behavior is this, uh, interaction between biology and exposure and cultural exposure.

James Robinson: Yeah. Yeah. And then of course, yeah, you, you, you’ve done something, uh, you’ve led a similar project, um, [01:14:00] but for singing, which I guess involves, um, you know, it looks more at the melodic features where there’s even more things in some ways going on than, than rhythm. Um, and. And again, I mean, to, to summarize very briefly, a really fascinating study.

You do the, you see these convergences that there are sort of, um, common ratios between, um, the frequencies or common, um, sort of minimum frequencies, I suppose. Um, but again, there, there is a, there is a very, there is a difference. Like there are kind of. Um, smaller, um, variations in, in Indian music, if I’m correct.

Um, but then there seems to be some kind of universal preferences for, um, kind of arcs, uh, if I, or like, uh, where there is a kind of rising and falling in, in the sets of, uh, notes. Um, one thing [01:15:00] that A couple of features that really stood out for me from, from, from, from that study, one was that things changed quite a lot where instead of getting people to sing, um, in, after they sing back the notes that they’d heard and using that as the, the means of transition.

If you got them to kind of use a slider, um, it, it really, yeah, it, it changed the features. So using the slider, they were less reliant on the, or sort of less drawn to a smaller vocal range, I guess. Um, and so the intervals between notes tended to be larger. But people also tended to be a little bit less accurate, um, which is kind of interesting because you’d think maybe that actually it should be easier to, you know, have larger intervals between things if we weren’t constrained by, um, vocal cords, but perhaps our kind of physiology is so attuned to [01:16:00] the, the sort of, um, the vocal range that we, we typically use that maybe we’re just really, really sensitive there.

Um, and the other thing, and this is similar to. Again, the, the work in linguistic evolution is just that you don’t see this emergence of, um, uh, let me say this differently. If, if, if instead of having a cultural transmission chain of notes being passed from person to person, someone is just, over and over, like passing something to themselves and kind of playing back the notes, then you don’t have this, um, same emergence of a, or convergence on particular, uh, notes.

Um, but people do kind of recall quite well, the, the, the precise notes they have, that they, they, they come up with their own kind of individual system that allows them to memorize, um, what the notes were. Um, but they don’t end up converging on, on kind of music as we know it, I suppose.[01:17:00]

Manuel Anglada Tort: Yes, um, this is, um, I think that the power of this kind of experiment in that, um, because we can simulate, recreate this cultural transmission process in the lab, um, we can then study for the first time the causal role of underlying mechanisms. So this is very powerful because normally these mechanisms, you know, the role of production Or the role of transmission.

We know that are important, but they’re impossible to study because they are hidden or, or just not accessible, right? In available data. So doing this in the lab allows us to manipulate different aspects and try to say, okay, how much does it matter production? If I sync versus if I produce music with an slider, um, or how much does it matter the transmission process?

If I transmit from person to person or I transmit. On [01:18:00] the music and myself, so we can run these studies. And that’s why what I did in this, in this paper is that we do different manipulations to tease apart the role of individual mechanisms. And we find that some of these mechanisms are very important.

So in the case of production, I was very surprised in that many of these, many of the features that we see in real music across the world, Can be just explained by, by production constraints, by the limits of our vocal system. So as I’m speaking to you, I’m breathing in air and I have limited air capacity in my lungs.

So at some point I don’t have more air and I have to start like going down in pitch or in, in, in contour because I don’t have, I don’t have enough energy. And this limit, um, physical limit from, from, from based on our vocal system, it does explain a lot of the features that we see in music. [01:19:00] So a very obvious one that we see in our experiments.

Um, how large the pitch intervals are. So of course I can sing an octave, which is 12 semitones, and I’m going to really have to force this, but I’m not going to be able to sing anything larger than an octave myself. So this will really limit the kind of music that we, that we can make. Right. So in this experiment, where we repeat this transmission experiments, but people match melodies with the sliders.

Suddenly we see that the musical systems that evolve are much different in this feature. So the average pitch interval is much larger

James Robinson: because

Manuel Anglada Tort: it happened because you can produce it. So this is just one, you know, teasing apart this role of, of production, um, for example.

James Robinson: Yeah. It comes back to that, that fascinating idea of the extent to which just our physical constraints, whether they’re memory or, um, you know, perceptual or, [01:20:00] um, the ability to reproduce certain things, just how much that defines the, the space, the aesthetic space, I guess.

Um, you know, we, we do tend to think of music, art, et cetera, as these otherworldly things that somehow detach from our physical, um, incarnate existence. But actually, you know, the, the formats that they take are so related. I mean, they’re completely dependent on, um, yeah, the contingent facts of our, of our material makeup, I suppose.

Yeah. And then there is

Manuel Anglada Tort: this, I guess, one of the most. Amazing questions that you can ask, but also one of the most difficult or impossible to address, which is these, um, what people call sometimes this music musicality, uh, co-evolution or like gene Gene’s [01:21:00] cultural evolution, which is that of course our, uh, musicality, our musical abilities.

And, and, and cogniti and cognitive, uh, abilities will shape the kind of music that we can make. So the music that exists today is really determined by, by, by our musicality. But at the same time, the music that we can make will shape, um, our biology or, or our ability over time and to understand. Why music is the way it is, we really need to think about this kind of co evolution of, of musical ability and, and, and exposure to actual, actual music in, in the world.

And I think this is really the way to kind of address the complexities of, of, of music evolution. But of course, it’s very hard, very hard to study empirically. But, um, there are very interesting things when, when you think about this, because for example, [01:22:00] some people think that, um, some of our aesthetic preferences for different music intervals, uh, might just come from the way in which, uh, different, uh, when, when, when you make music with an instrument, for example, you can just, there’s a certain combinations of music that you can make, um, and certain frequencies that will resonate.

And because that’s how you would do music at, you know, at the early, early, uh, first early humans, for example, would be singing or will be hitting different, like, uh, kind of drums and so on. It could be that this kind of physical, uh, resonances that you could make at that point. With these instruments could shape the kind of aesthetic preferences that we, that we have now.

So we have, we have this preference for like consonant intervals, um, based on the kind [01:23:00] of frequencies that these instruments could make years, years ago, years, years ago. And of course, one of these instruments is our own voice, which also resonates, um, on certain frequencies. So there is this very complicated co evolution, uh, system here that is very important to, to try to, to address these questions.

James Robinson: I guess the kind of corollary of that is that, uh, I dunno, the, the Moog synthesizer maybe in some thousands of generations time will have had some effect on the, the appreciation of, of music down the, the, the line. Totally.

Manuel Anglada Tort: Yeah.

James Robinson: And I guess one thing we should say is that, you know, even if one wants to be very deflationary about this and say, Oh, well, you know, the landscape of music is completely defined by, um, our physical nature and, you know, there’s a evolutionary aspect to that and so forth.[01:24:00]

It doesn’t mean that there’s not just kind of infinite ways that we can play in that landscape. So, you know, I, in some ways, like it just makes it. I think it’s, it’s a very interesting to understand what defines that landscape, but we’ll never answer the kind of eternal questions, I guess, of within that, like, what is it that makes one particular phrasing or one particular, um, piece of music, particularly beautiful.

We, we might get some clues, I guess, as to, to why. You know, it comes with the form of notes that it does and perhaps even why we like arched contours. Maybe there’s some kind of, um, easy way of understanding that, but, but I feel that music is always trying to, like, like all of art, it’s always trying to play with itself and kind of usurp its limits in a way and do something slightly unexpected within the format that it inherits.

Um, [01:25:00]

Manuel Anglada Tort: Yeah, and I guess that’s one of The reasons that music is such a special thing to, to study, um, and from a scientific point of view, I think is such a puzzling behavior, you know, it’s so universal and widespread. So people spend so much time in it and resources. So in theory, there should be some sort of obvious value for, for.

For it to survive and evolve. But on the other hand, it’s not clear at all. And people disagree of why music exists at all. What is the actual, the actual value. And I think the only thing that we can do as scientists is to kind of. Pick our battles. And then there’s like some levels of analysis in which I think we can definitely get very good answers.

So in terms of why is music pleasurable from a neuroscience point of view, I think we are very close to understand how all of these works, [01:26:00] but from a completely, from another level of analysis, which would be. evolutionary psychology, um, then it’s much harder story, right? Like why does music, what are the origins of music and why did music evolve in the first place?

This, uh, this kind of level of analysis, uh, some questions I don’t think we will ever be able to, to completely solve.

James Robinson: Yeah. It’s a super exciting time. And I mean, just looking back at the conversation, um, one gets a flavor of just all the possibilities for study out there. Um, you know, all these big data sets that are going around the, uh, the genomics information, just the sheer, um, Volume of, um, things to look at, uh, it must, I don’t know, does it seem daunting sometimes, or are you just excited about the

Manuel Anglada Tort: field?

No, I think it’s exciting. And I think right now it’s particularly exciting because [01:27:00] with the, you know, the advent of technology and mobile devices and all of these social media platforms and stuff combined with all the computational techniques that we have now, we have like. Algorithms that can very accurately, accurately represent music and generate music.

Um, and make sense of all of these complexities of these systems. I think that’s very exciting for us, uh, for, for, for the science of music and, and aesthetics to try to address this, this question with a lot of data, but also very powerful computational and behavioral techniques.

James Robinson: Brilliant. Yeah. Um, yeah, I think this is a great place.

We’re both here in the UK and, uh, uh, getting late. I don’t know if you have any final thoughts, but I, this has just been such a fascinating conversation.

Manuel Anglada Tort: Yeah, no, I think, I think that, yeah, it’s been great and, uh, lots of fun to, to, to talk about all of these trajectory of all of these different approaches on music, music research.

So I think I’m very happy with [01:28:00] this. Yeah.

James Robinson: Brilliant. Thanks so much Manuel.

Language Evolution & The Emergence of Structure — Simon Kirby

Language is the ultimate Lego.

With it, we can take simple elements and construct them into an edifice of meaning. Its power is not only in mapping signs to concepts but in that individual words can be composed into larger structures. 

How did this systematicity arise in language?

Simon Kirby is the head of Linguistics and English Language at The University of Edinburgh and one of the founders of the Centre for Langauge Evolution and Change. Over several decades he and his collaborators have run many elegant experiments that show that this property of language emerges inexorably as a system of communication is passed from generation to generation. 

Experiments with computer simulations, humans, and even baboons demonstrate that as a language is learned mistakes are made – much like the mutations in genes. Crucially, the mistakes that better match the language to the structure of the world (as conceived by the learner) are the ones that are most likely to be passed on.

Dynamic Message Animation



(00:00) Introduction

(2:45) What makes language special?

(5:30) Language extends our biological bounds

(7:55) Language makes culture, culture makes language

(9:30) John Searle: world to word and word to world

(13:30) Compositionality: the expressivity of language is based on its Lego-like combinations

(16:30) Could unique genes explain the fact of language compositionality?

(17:20) … Not fully, though they might make our brains able to support compositional language

(18:20) Using simulations to model language learning and search for the emergence of structure

(19:35) Compositionality emerges from the transmission of representations across generations

(20:18) The learners need to make mistakes, but not random mistakes

(21:35) Just like biological evolution, we need variation

(27:00) When, by chance, linguistic features echo the structure of the world these are more memorable

(33:45) Language experiments with humans (Hannah Cornish)

(36:32) Sign language experiments in the lab (Yasamin Motamedi)

(38:45) Spontaneous emergence of sign language in populations

(41:18) Communication is key to making language efficient, while transmission gives structure

(47:10) Without intentional design these processes produce optimized systems

(50:39) We need to perceive similarity in states of the world for linguistic structure to emerge

(57:05) Why isn’t language ubiquitous in nature …

(58:00) … why do only humans have cultural transmissions

(59:56) Over-imitation: Victoria Horner & Andrew Whiten, humans love to copy each other

(1:06:00) Is language a spandrel?

(1:07:10) How much of language is about information transfer? Partner-swapping conversations  (Gareth Roberts)

(1:08:49) Language learning  = play?

(1:12:25) Iterated learning experiments with baboons (& Tetris!)

(1:17:50) Endogenous rewards for copying

(1:20:30) Art as another angle on the same problems