[ Music ] >> Data skeptic features interviews with experts on topics related to data science all through the eye of scientific skepticism. [ Music ] >> Samuel Mayer is a graduate student at Harvard studying the origins and functions of music. His methods leverage fields like cognitive science and evolutionary psychology. He's been involved in a number of empirical studies about the effects of music on people, particularly at an early age. Sam, welcome to Data skeptic. >> Thanks for having me, Karl. >> So I thought maybe we could start out with a very general question, maybe one that would take hours to answer properly, but just to get a directional sense, why is it that people like music? >> That's a very difficult question. There is quite a bit of research going on across a whole bunch of fields in the cognitive sciences and at large in academia, also in anthropology and ethnomusicology. I'm just trying to answer it. And I really don't have an answer at this point. The driving question behind my research program and the research programs have a lot of people who study music is essentially that. Why are we musical? Where does music come from? Why is music the way it is and how does it work? >> I think it's safe to say though that music is quite important to human beings and has been for a long time. I'm not really a history podcast, but I'm sure a good history podcast could go into music being there perhaps from our earliest days. Can you talk a bit about the evolutionary need or maybe what music fills in for us socially? >> There's been quite a bit of debate over the last 150 years and actually a little bit farther back than that about the evolutionary origin of music. This was a question for Darwin in his book, The Descentive Man. The philosopher Herbert Spencer also wrote a great deal about his ideas about where music came from. It's generally assumed that music is a human universal. There are many, many, many cultures out there that use music, including most small-scale societies that are known in the present day. Interestingly, there's not really definitive evidence yet that it is a human universal, but it is the case that it's very hard to come across a culture that doesn't have music. It's also the case that the types of music that leave fossils, instrumental music, instrumental music is very old and we know that because of bone flutes that archaeologists have dug up and that are usually dated to about 40,000 years of age, which is pretty old. But of course, you can produce music with the human auditory system and vocal tract. Singing, that is, singing doesn't leave any fossils at all, so given the age of the vocal tract and the age of the auditory system, but also the fact that the vocal tract shares a lot of features with other non-human primates and also mammals in general. It's a pretty safe assumption that singing was around for much longer than those instruments. And that's actually an assumption that Helmholtz and Darwin had 150 years ago. That's pretty much a safe assumption. It is the case that it's probably universal and that it does quite a bit for people of all ages. Infants are captivated by music, toddlers are captivated by music, adults all over the world listen to music. It's certainly a hallmark of human behavior. - It sounds like perhaps you hesitate to say strongly that it's universal. Is that out of being a proper scientist and not wanting to overstate the evidence or is there some particular room for doubt rather than just having not catalogued every culture and check the box that says yes, they're musical? - Well, it's mostly out of scientific jurisprudence, I guess. I'm pretty sure that it's gonna turn up in every culture, but the questions that's sort of more interesting about whether or not a culture has any kind of music is what particular types of music those cultures have. And that starts to bear on evolutionary questions. So when you study the evolutionary origins of a particular behavior, one of the necessary but not sufficient conditions for something having adaptive value in evolutionary history, namely that it increases reproductive fitness in some fashion is that it is universal. But the way universality is measured is a bit tricky because all human cultures are not currently present on Earth. There's many, many, many cultures that have died out or are now modern societies or have become more modern with the advent of technologies. So it's difficult to actually assess what particular features of music you would want to test for in studies of universality of music. So those studies haven't really been done yet and actually a couple of collaborators of mine and Harvard and I are working on a project that's attempting to do that. Essentially asking not just is your music universal, but what particular features of music are universal? The case that some types of songs are found in pretty much every culture we can find, but others aren't. For instance, I suspect, I don't know this yet because we haven't finished up our project and we haven't done all the analyses yet. But one hypothesis that I have is that instrumental music is less likely to be a universal feature of music than vocal music. And that's simply because, my hypothesis is simply because there requires some technology to build those instruments. And that's just by virtue of what technology is less likely to be found in every single culture, whereas singing an infant with no training at all or a toddler with no training at all can start singing. All you need is the physiology of the human body there in order to produce vocal song. - Interesting, and then I guess maybe we'd say that there's a technological parallel here, not that everyone would invent a trumpet, but a trumpet-like instrument requires metalworking, for example. - There are definitely interesting questions to be asked as well, so let's say I'm wrong and maybe instrumental music is totally universal in every culture that we've got ethnography from, we see evidence of instruments. Then the interesting question is gonna be, okay, so there's some universal motivation to produce sounds with objects. What are those objects like? What kinds of sounds are they? And are those sounds themselves reflected in potentially the acoustic properties of the human voice? There's a lot of interesting work from Dale Purvis's lab and from a few others, Josh McDermott as well, who are looking at a couple of theories for where human preferences of consonants come from. So consonant sounds are like the sounds that are found in a lot of music around the world. They're basically pitches that sound nice together to us. There's a couple of interesting theories about why we have preferences for those sounds as opposed to other sounds to elaborate that sort of to the extreme. Pretty much nobody around the world would find the sound of like nails on a chalkboard to be pleasant, right? Whereas, you know, a piano's middle C is kind of nice, it sounds fine, but so I'm not, the consonant's dissonant stuff is not going that far, but it's talking about sounds from music that tend to be used in a lot of cultures and that fall into this sort of under the umbrella of consonant sounds. And the theory is basically goes something like the fact that the human vocal tract approximates a closed pipe and a closed pipe has a particular set of acoustic properties that when you extrapolate them, you end up with a group of tones that roughly sound consonant with one another. That's a pretty big simplification of that theory, but it's the general idea. The idea there is basically like if, for instance, that is the origin of our preferences for consonant tones because we're hearing all of these sounds produced by closed pipes from other humans in our own voices. It could be that when we build instruments, we try to approximate those sounds. So we build instruments that are able to produce the same sets of tones more or less as the human voice does. - When I think of evolution, naturally my brain goes first to biology, although I've been really enjoying getting into this sort of evolutionary music literature over the last few months, but on the biological side, I've heard it stated that the proof of evolution from the genetics alone is just there, set aside the fossil record and everything. We can see a clear path from A to B and we can explain how different groups of people move from here to there and trace their origins. Do you think the same possibility is there in music that as different techniques or chord structures, maybe the use of sevens and things like that, or just rhythm techniques is as people migrated about and developed new evolutionary ways of looking at music, could we trace people's origins from their music? - So it's certainly the case that there's the potential to study or test for one particular evolutionary origin of music or the other, but there's a couple of separate questions to tease apart from one another. One of them is where this behavior came from in the first place, to take an analogy with language. There's one sort of group of people that's looking at evolutionary origins for language, gua language. So the specific features that are included in the behaviors that are language, so including speech and grammatical structure and that kind of thing, that's one set of questions. But then a totally separate set of questions that's related, but it's not the same set of questions, has to do with how language changes as people move about the world and different groups, maybe with one another and you get many, many languages that maybe share structural properties of their grammars, for instance, that are actually sound quite different from one another. Those questions have clear analogs in music. So there's one question which is a very difficult question which is where does music come from in the beginning? At some point, there was a human-like species that didn't have music and at some point after that, there was a human-like species that did have music, somewhere along the line music came about. Now, whether it came about as a result of an evolutionary adaptation that led to the motivation to produce sounds that were sort of proto music and the motivation to listen to those and positive responses to them and so on, whether that was an evolutionary adaptation or whether that was a byproduct of other adaptations like adaptations potentially for language, that's a very hard question. It's a testable question, but it's a hard question. And that's the question that's been the source of quite a bit of debate among evolutionary psychologists, evolutionary biologists and psychologists for many years. The related set of questions that are also very interesting of how once you have music, how it changes across cultures and how it affects people at different ages and lifespan, that's another set of questions that doesn't necessarily give an answer to the first question, but it's also very interesting. In my work, we're doing studies addressing both of those questions. So in one set of studies, we're testing one particular evolutionary theory that's looking at a potential role for music in the interactions between parents and offspring as a signal from parents to offspring of their presence and their attention. That evolutionary theory gets tested against, you know, a potential other evolutionary theory, like maybe music's just a byproduct of language faculties, for instance. And in order to test those theories against one another, we design experiments that can rule one in or one out. In this case, we use studies of people who have rare genetic conditions, like Proder Willy syndrome or Angelman syndrome, as sort of a natural experiment for what happens when different genetic effects are pitted against one another. And that's one set of questions. On the other hand, when we're looking at, you know, how music affects people in different parts of the lifespan, those are not evolutionary questions. Those are questions about sort of psychological functions, how things work at different ages. There, we're doing studies of infants who hear songs from new people. They know the songs, but they've never seen these people before. And the question is, what do they think about these new people once they've got this musical information from these potential new social partners? That's one line of work. But the main distinction is between those evolutionary questions, on the one hand, and the functional questions on the other. One thing I've admired about your work and your research is there seems to be really strong reverence for the scientific method in trying to make sure that you're controlling for all the variables that could be having a secondary or hidden effect. Just if you could dwell into the research you mentioned where you were studying people that had certain syndromes that ended up creating natural experiments, curious to know what the particular study was there. So, I mean, in general, before I address that question, in general, I think there's a little bit of a funny thing that happens when it comes to complex and difficult behaviors like music and the arts. When psychologists and biologists try to explain these behaviors, sometimes there ends up being kind of a fuzzy, magic feel to like, oh, this is such an incredible, weird behavior, how are we going to explain it? And I feel like people have lost sight in some cases of just basic science when you're studying these things. And a core assumption of my research program and of the research programs that I have going up with my collaborators is basically that music is no different than any other human behavior in that it has to be explicable through the basic principles of biology and psychology. In that vein, we try and design studies that we could substitute music for any other human behavior that has already been well-explained. Music is just one of these behaviors that hasn't yet been very well-explained. The subject is basically like music shouldn't get a free pass. It should be subject to just as much experimental rigor is any other area. So for instance, in the study of people with rare genetic conditions, there, we're looking at a couple of genomic imprinting disorders, specifically Prada-really syndrome and Angelman syndrome, because they're really interesting illustrations of a phenomenon I'm describing by Robert Trivers in the early '70s called parent offspring conflict. The interesting issue with parent offspring conflict is that because of the biological facts of relatedness, essentially that a parent is only-- a gene that's an apparent has an only 50% likelihood of being in a particular child. And that particular child, a gene in that child, has only a 50% likelihood of being in that child's sibling. In cases where the parent's reproductive fitness is at odds with the infant's reproductive fitness. For instance, a parent may not want to give a particular infant breast milk for ad infinitum. They want to stop breastfeeding that infant in a particular time so that they can give birth to another child and feed that infant. The infant may have a bit of a disagreement with that parent about when that time should be. And that's sort of the prototypical example of parent offspring conflict. What's interesting in these genomic handwriting disorders is that by virtue of essentially a chance error in the genome of people who have this disorder or one of these disorders, essentially the resulting behaviors turn off behaviors that are in the maternal interest or in the paternal interest. In Prada Willy syndrome, for instance, you see behaviors that really favor maternal fitness at a cost to paternal fitness. You see things where in early infancy, they basically don't nurse very much and they're pretty lethargic and behave in a way that is very conducive towards maternal behaviors that will lead to an additional infant from the mother as opposed to behaviors that favor that first infant and not which it was. So it's a very interesting set of behaviors. And that's Prada Willy syndrome. In Angelman syndrome, you see the reverse. You see behaviors that favor paternal fitness. So that increase costs on the mother, but not on the father who can go out and impregnate other females in the population. So what we're asking in these studies with these genomic imprinting disorders is basically whether music plays a role in these parent-infant interactions because we're testing a theory that one potential origin of music is in the dynamics of parent offspring conflict, that one form of investment that a parent can provide to an infant is essentially a tension. And one way to signal that attention is to provide music to them. One straightforward test of that there is, if it is the case that music has some adaptive value, some evolutionary origin in the interactions of parents and infants, we might see differential responses to music in these two syndromes, in Prada Willy syndrome and in Angelman syndrome. So that's what we're testing. - Of course, that's a very extra convoluted theory and a tricky thing to go out and test. And not least of our problems is that these are very rare disorders. So we've been very slowly and methodically finding people who have these disorders and working with them to see how they respond to music, to survey their parents on how they responded to music as infants and in a variety of other measures with these populations. - There was some great discussion in one of your papers. I think it was music in the home, new evidence for an intergenerational link. And I'll be sure to link to that paper in the show notes for anyone who wants to follow up. In a lot of these music studies, I've always sort of asked myself the questions like, well, how do we know that income is this isn't just parallel with income, that a richer family has more access to musical instruments and things like that. And that's why we see more music in the home. I think you guys did a pretty good exploration and set up good controls. Could you talk through a bit about the concerns you'd have along those lines and how you tried to structure a test that really truly tests for one control. And that's the musicality of the home. - So in that paper, we were using survey data from a couple of our randomized trials to just try and get a descriptive picture of what kind of music is happening at home with young kids. These were three and four year olds, or maybe just four year olds. And we basically surveyed parents on a whole bunch of things. How much do you sing with your kid? How much do you do visual arts with your kid? How much do you play recorded music? And we got basically just a whole bunch of measures for what parents were doing with their kids. So one question that falls out of that is, are there differences in the amount of musical behaviors across different demographics? So is it the case, like you mentioned, our families who have a lot more money are just more likely to do musical things? I mean, one of the interesting findings from that paper, which actually replicated a previous study that showed the same thing, was basically that it doesn't really matter how much money a family has. They tend to sing a lot. We didn't see many differences across different income levels in how much music was being provided to kids. Another interesting question that came out of that data set was, okay, so we've got some families who sing a lot, we've got some families who sing a little bit less. How do we predict whether a new parent, a new parent is gonna be in one of those groups or another? And it turns out that a really nice predictor of that is what kind of musical experiences that parent had when they were a child. So parents who report being sung to quite a bit as children from their parents or from a grandparent, tend to be frequent singers themselves with their own infants. But we don't see similar relations with other forms of artistic exposure. So parents who did a lot of arts and crafts with their parents when they were kids don't tend to do more arts and crafts with their kids. What we see is a specific effect there for music. And that raises a couple of interesting questions about why that is, there's one simple explanation is just that people who sing a lot are more likely to have a particular suite of genes that leads to singing a lot and they pass a lot of those genes onto their kids and see the same result when their parents. That's one possibility. Another possibility is when parents make decisions about what to do with their infants, they think back to what they had as kids and try to emulate what their parents did. We don't know which is which. The all of the results from the behavioral genetics literature would suggest that the first explanation is maybe a good one. But we actually have some twin studies going on right now at the Boston University Twin Project, which is directed by Kim Sandino that are starting to attempt to tease those possibilities apart. - I also had a slight tangent related to that paper. I noticed you guys use the Wilcox Mann Whitney test to test your contingency table. To my read, I think that was a correct choice. That's the statistical choice I think I would have made as well. But it's also not one that someone who has just a quick intro to stats would even necessarily know about. Curious to hear about your process and how you and your fellow researchers arrived at that choice. - It's sort of interesting. I think there's sort of often an assumption in a lot of survey analyses where you've got ordinal variables being used as opposed to interval scales that is just sort of fine to throw whatever analysis you like at it. And in fact, when we were deciding what analysis to use for that paper, actually, I think after we had already submitted the paper with that analysis, one of the reviewers actually pointed out, like it's funny that you use that. Why not use just a t-test or something? And it turns out that when you have a large effect size, really any of those measures are gonna give the same result. But when you take a stats class that teaches you how these models are built, you kind of want to use the appropriate tests no matter what, even though might get the same result that way. And there is sort of an interesting sort of small backstory, which is just that my background in statistics comes from the School of Education here because I'm a student in the School of Education, despite the fact that most of my research is in the Department of Psychology here. And the School of Education is a very different group of people who are teaching statistics than a psychology department does, which is I think typical of most universities actually, typically the people who are teaching statistics in an education school are economics and econometrics people, whereas people who are teaching statistics in a psychology department are site people who know about stats. But I think there are tend to be different approaches from those groups of people, which can lead to some interesting differences in opinion about what analysis to use in a given paper. - Yeah, yeah, often are. So I know there's a fair amount of pseudoscience around on the topic of just generally music makes you smarter. And maybe there's some worthwhile points in there, but some others that are a little iffy, good science and bad science under claims of that nature. Why do you think people find music makes you smarter type claims so compelling? - Well, I think that there's a bit of history to this claim that leads people to take it at face value and without really taking it apart and thinking clearly about it. The first big issue for me about that idea is that it doesn't really have any theoretical basis. So one of the things that's nice about studying human behavior from an evolutionary perspective is pretty much everything you're studying has to come from some theoretical basis because we can't travel back 100,000 years and run an experiment and then wait 100,000 years and see what turns out. So you have to think from a theoretical perspective. And I don't know anyone, even people who really believe strongly that music makes kids smarter. I don't know that anyone's presented a theory that predicts a priori why you might see that effect. That's a first problem. It's not a crucial problem. There's plenty of research and psychology that is exploratory and that just is trying to demonstrate phenomena in the world without having a theoretical basis. But that's a little bit of a red flag at first blush. The second problem is that that theory hasn't really been tested very rigorously in the same sense as say, you know, the effects of large-scale educational interventions or the effects of, you know, medical, like drugs, medicines. The gold standard for testing for cause-effect relations in those areas are randomized trials. And one of the first things I did when I got to Harvard was run a couple of randomized trials with Elizabeth Spelke testing the effects of exposure to music in a sort of instructional context on kids. And it turned out, we found out as we were running these trials, we were sort of surprised to realize that actually very, very few studies had done this. Five, in fact, had five randomized trials had been conducted. And the results of those trials were all basically at odds with one another. So there wasn't any sort of clear conclusion from that literature. So we set out to do some more studies and see what we found. When we found basically not much of an effect, there was quite a bit of fuss made, I think partly because Harvard has a fancy name and people are likely to listen when there's a press release from Harvard, which is not a very scientific reason for that fuss to be made, but it is what it is. But also because, not just because of our own studies, but because in that paper, and this is what I actually think is the real value of the paper that we published, we essentially outlined very concisely and kind of coldly what studies had been done, what they had found, and what the overall picture from that literature was. And when we did that, it basically is a table in that plus paper that shows each of the five studies, exactly what was done, exactly how many people have been running the studies and exactly what the effect sizes were. And the bottom line was they're not very big effect sizes. If you're really strict about what you count as a randomized trial, namely, that's to have randomization, it can't be pseudo-random. It has to have a particular music training compared to something else. It turns out that there's really very little support in this thing in the first place. - Yeah, I'd be curious to hear some of the principles of a really properly well-defined, randomized control trial in this context. - The simplest thing to rule out is that there's a ton of studies where researchers will give a bunch of music tests to people, and also a bunch of cognitive tests of math skills maybe IQ as a general test or spatial reasoning and any of those sort of standard cognitive areas. And they'll show correlations between them. So people who have more music training or who do better on music tests tend to do better on those other tests. That's a very replicable finding that you can find a million citations for that. But that's not really sufficient for describing a cause effect relation between going to music classes and having some kind of increase in cognitive skills as a result of those classes, simply because correlations aren't causation and/or don't imply causation. And it's also quite easy to come up with alternative explanations for those relations. Concentration is used a lot in private music lessons. If you concentrate very well, you're likely to concentrate well on other things as well. So that's easy to throw out. The other type of study that's out there that we didn't include in that review are pseudo-random studies. So studies where you'll take people who have already started in music classes and people who have started in some other class, kids usually, and then test their cognitive skills after a period of time. The problem with that, of course, well, there's a couple problems. One problem is there's a huge selection bias. If it's the case that people who are inclined to be good at music also are inclined to be good at other things, that's gonna account for any effects that you show in a pseudo-random study. That's one problem. Another problem is you're likely to see differential effects of the types of coursework themselves. So if you're taking people who went to a music class or didn't go to anything at all, the experience of being in a music class might be similar to the experience of being in any class whatsoever as opposed to no class at all. That can also be a problem with randomized trials, but that's usually a problem with pseudo-random studies as well. So that's a big problem. So really what you need is like you have with a drug trial where you start a priori with a bunch of people who aren't doing any music classes or are doing some baseline level. And then randomly assign some subset of them to get music classes, some subset to get some comparison that's like a comparable level of intensity, but is not music. So in our studies we used art classes, but other studies have done other things like dance or drama. And then ideally a third group that has nothing at all to see if you can test for effects of just being in a class versus nothing at all. There's only five studies. We're August seven if you count ours that have done that. Those are the five plus R2 flat results that I mentioned earlier that basically have results that are in conflict with one another. - I assume you're referring to the paper, two randomized trials provide no consistent evidence for non-musical cognitive benefits of brief preschool music enrichment. - Yeah, it's a mouthful. Yes, that's right. - Well, we'll be in the show notes as well. There's a subtle point there in your conclusion. And if I were to quote it's few randomized control trials have been employed to assess causal effects of music lessons on child cognition. And no clear pattern of results has emerged. Which is a fickle statement in a way because as you also say in your frontiers on psychology piece failure to reject the null hypothesis does not imply evidence in support of the null hypothesis. Yet that subtle point seem to have been lost a little bit. Can you talk through your experience in that related to this study? - Essentially we didn't make any claims in that paper that there are no effects of music lessons on kids. We know for sure that there are effects of music lessons on kids. The most obvious one is kids who go to music lessons get better at music. But what's still unknown based on our read of literature and our experiments is whether there's any transfer that happens consistently from going to music lessons to other non-musical domain. One area where there is transfer which I don't believe we talked about much in that paper is a couple of studies have shown increases in fine motor skills as a result of instrumental music lessons. So fine motor skills that aren't necessarily using musical instruments but just fine motor skills in general. And that's a pretty intuitive finding if you've spent a lot of time practicing the flute you've worked on fine motor skills on the flute and those fine motor skills might transfer. But basically because across these seven randomized trials we don't see five of the seven with a positive effect on one particular test or six of the seven with some positive result no matter what the test is. There's not any clear pattern to those results at all. Our conclusion was okay, for now we don't have evidence of this thing but maybe with more randomized trials or bigger studies or studies that follow kids for longer or have more intense lessons maybe they'll be shown in the future but for now we don't have any evidence to reject the null hypothesis here. That's not how our paper was interpreted in the popular press and that's not how people sort of thought of our paper when it came out. Most of the reports that came out about those randomized trials had a headline and an article along the lines of Harvard scientists say that music is bad for your kid which could not be farther from the truth. We sit out to study these things because we're pretty sure that music is great for your kids. What we're not sure about is whether music is great for kids in a domain other than music. What was sort of interesting for me is that this was my first major paper and I had no idea that the press was even gonna cover it to begin with. In fact none of us on the paper thought that this was gonna happen. We sent the report to the journal PLOS because PLOS is notoriously open to publishing negative results. We think it's very important and PLOS agrees that it's very important for negative findings to be published that they're in the literature as opposed to just thrown into the file drawer and no one ever hears about them. So we sent it to PLOS not so that we would make a lot of news but because we knew that they would be willing to consider a negative result. So this was a big surprise to us and when we saw all of this sort of misinformation that was out about our paper, Liz Spelke and I basically discussed what we should do about it and she and I felt that the best thing to do is basically to respond to the criticisms and respond to the inaccuracies just by responding to interviews and saying, look, it's not what we said. Stop writing about writing that we said that. We didn't say that. So we made a few efforts to do so. We did a Reddit AMA. We responded to interview requests. There were a few radio things. And I wrote an op-ed for The New York Times describing exactly the extent of our studies and I hope clearly stating why we don't consider these negative results as evidence that music is positive or negative for kids. I think it's just study music because music is great because they enjoy it because it's an important part of our cultural history because it teaches them about society and because it's fun. Yeah, so yeah, that was the following. And I documented a bunch of that followed in the frontiers paper that you mentioned which came out a few months ago. In my opinion, those points were clear the first time but I definitely appreciate all the follow up. So I think we'd be remiss to be on these topics and not comment on the so-called Mozart effect. For anyone who I guess maybe didn't take a standardized test could you define what that is and comment on what the literature says about it. The Mozart effect is a term that was coined to describe a body of research asking whether listening to music led usually adults to perform better on standardized tests originally of spatial cognition but it ended up being generalized to a few other areas. This effect turned out to be discredited by Glenn Shellingberg's group who I believe was at Toronto at the time. Essentially, follow up experiments showed that people who did, who would do better after listening to Mozart would also do better after listening to other music. But most of all, when you measured their arousal, their mood and adjusted in a simple regression, their score is for their mood level. What was found is basically listening to something nice made you feel better and feeling better helped you do better on whatever test you were taking. So it turned out to just be sort of a restatement of what we already know, which is that listening to music is nice. What was problematic though is that the catchiness of this Mozart effect's term and the many studies that sort of followed it because it turned into this sort of silver bullet-y idea and eventually it got thrown at, "Oh, let's do it for kids also." And maybe it's not just music listening, maybe it's music training, which I think led to the music makes you smarter, kind of slogan. All those ideas were snowballed. And even though it had been clearly discredited in the Journal of Psychological Science, it still hangs today. I just reviewed a paper that was testing the Mozart effect. I had to state to the editor, "This doesn't make any sense." This effect has been completely discredited. I don't understand why the authors are still testing it. So it still sort of holds some weight in the public, in the eages of the public eye, I guess. - Yeah, it's funny how things like that stick around. So despite the commentary on them and the follow-ups and replications. - Yeah, it's very weird. There's a few kind of scientific legends like that the myth of left-brained people and right-blain people is one like that. In fact, there was a frontiers paper published, I think in the last year or so, that was a list of 50 scientific myths that people should really stop citing. And the Mozart effect was one of them. - Oh, I have to check that out. - The Mozart effect was one of them. The left-brained right-brained thing was one of them. There were a whole bunch. - Yeah, so I'm not a parent, but if I was, I presume I would want what was best for my child. There's a fair amount of popular press, pseudoscience floating around about music. What's the actual argument? Why should a child be involved in music and study and sing and play instruments and these sorts of things? - I mean, I think that kids should study and play music. If they're interested in studying and playing music, I think parents should send their kids to music lessons. If they think that's a cultural experience, they want their kids to have. Similarly, I'm not particularly good at painting or drawing. In fact, well, let's not sugarcoat it. I'm terrible at drawing. But I would send my kid to art lessons if my kid was interested in it and if my kid had a good time doing it. I think the visual arts and music and the performing arts in general are cultural products. They're part of what humans do. It's clear that we do them all over the world. The jury's still out on what the evolutionary origins, if any, are of those behaviors. And we're still studying a great deal of possibilities for what the experiences in those behaviors actually do to people, whether it has to do with transfer to other cognitive skills or more interestingly has to do with their understanding of other people who do these behaviors, potential social effects of music and art or psychological effects in other domains. Those studies are still ongoing. But the basic reason to do it is because it's great. And I'm not sure we need any more justification than that. - Yeah, I think that's the key point there. Music is its own joy, I guess. - Right, right. - While I've got you here, I'd love to get your thoughts on a particular area. I've been a little bit fascinated by, since I learned about the works of Alan Lomax. I'm guessing perhaps you've heard of Cantometrics, which was a study he kind of pioneered? - Yeah, so in the late '60s, Alan Lomax published the book, Folk Song, Style and Culture, which was a project that I believe was supported by the American Association for the Advancement of Science, published as the Journal of Science. It was this sort of radical idea in the '60s to systematically look at the musics of the world and compare them to one another. Which in modern times seems kind of quaint. It's like, oh, that's so nice that they would think of doing comparative work. What's kind of amazing is that Lomax, with only a couple of small exceptions, is pretty much the only person to ever have done this. And Cantometrics was a word that he made up to describe essentially a set of rating scales that he had a bunch of research assistants use to extract data from field recordings that he had collected as an ethnomusicologist. And not just things that he had collected, but that others had collected as well. The project is, I think, it's super interesting. I have the books right off my shelf here actually. And its heart was in the right place. It's well-intentioned. It ran into some difficulties that sort of, I think led to it not achieving the prominence that it should have. At least in current day, obviously I can't speak to how prominent it was in the '70s when I went around. But essentially, one huge problem with it is that when he was doing these analyses, Lomax was basically working with the very beginnings of a modern field of statistics. And as a result, the analyses are very difficult to get through. It's not really clear how much we should believe the results of those analyses. Armand L'Orois, who's a professor in the UK, recently did a paper a couple of years ago, I'm trying to replicate a lot of those analyses and trying to dig into the original Cantometrics project. And the results were basically a huge mess. Because essentially, the project essentially happened before it really should have happened. If it had happened in sort of modern day computer-based analyses with modern statistics, I think it would have done a lot better. I mean, there's actually a couple of groups ourselves included who are doing sort of updated versions of it where the big picture goals of the project are preserved essentially to systematically document the similarities and differences of the musics of the world, but using updated methods. So better sampling and more detailed transcriptions of the songs, that instead of using this Cantometrics thing, which is pretty subjective and has kind of odd variables, like how well integrated a coral piece is, sort of an odd measure for something, for like a hunter-gatherer dance song, for instance, we're trying to use more objective measures, things like tempo, which I can have 30 people tap to the beat along to a song. And from this result, I can get a distribution of what they think the tempo is. And from there, say, okay, our best estimate for the tempo is 70 beats a minute or whatever. So we're doing a study like that now in Patrick Savage and his group, Patrick Seppes in Tokyo, is sort of a pioneer in this area of comparative musicology, which is like following up to Lomax in the modern world. So our project here is called The Natural History of Song. It's a pretty large project. It takes as data the ethnographies of 60 world cultures that are documented in a big database called the Human Relations Area Files. So essentially we've put together with a team of about 25 volunteers, about a half a million words of text that describe how songs are used in the small scale societies and the state assesses from 60 world cultures. And we're beginning to analyze those data to see if there are statistical regularities across these cultures about, you know, who's doing the singing? When does it happen? What's the context? What kind of person does the singing? What are the songs about? What are the lyrics, you know, those kinds of questions? And then we have a second part of that project which gathers systematically field recordings in different song domains from around the world. We're getting lullabies and dance songs and so on. And from those we're doing detailed musical transcriptions of the songs to get out what the musical features are of these songs. And that helps us to eventually get towards evolutionary questions. If it's the case that there was some evolutionary adaptation that led to music in a particular domain, it's possible that the musical characteristics of that domain are somewhat preserved in modern environments. So for instance, if dance songs are more similar to one another across all cultures, then are say healing songs. It's more likely that dance songs are the target of some adaptation than healing songs. Now it's not evidence for an adaptation, but it points us in that direction. So these big data sets are gonna, I think, start to move us towards a rigorous cross cultural science of music that Lomax was trying to get to. But I think it was a little bit before his time, at least in terms of the technologies and the statistics that were available at the time. So yeah, that's sort of what we're working on in the Lomax area. - Oh, that's incredibly exciting. I'm really looking forward to following up on that. I agree with your sentiments almost exactly. I applaud Lomax for his intentions and especially for the catalog he built. - It's a very cool, very cool catalog. In fact, one of the resources we've been using for the Natural History of Song project is the Harvard Archive of World Music. And one of the first things we checked out of the archive of world music was the Kenometrics catalog. And there's training manuals and there's all these field recordings on 45s. And it's an incredible resource. The thing that's a little bit sad, I think, thinking retrospectively and people who have been in the field much longer than me can disagree with me if they like. It seems like from my read of the literature and how the field in musicology, not in musicology, responded to Lomax. It's almost like his work wasn't, it wasn't given the credit it really deserved when it came out because there was sort of an anti-imperialism feeling in the musicology when it came out that led to these just scathing critiques of it. And once in a while, as we're searching around for songs on the Natural History of Song project, we've bumped into people from that generation who really feel that this kind of comparative work is not the right way to go. People whose careers have been dedicated to investigating one or two cultures, but not 60 cultures and not 600 cultures. And there's a little bit of sort of academic tribalism that comes about when that happens, which is too bad because all the people involved in this Lomax, when he was doing it and the younger people like God Savage and like my group who are doing it now, it's all well-intentioned. We're not trying to attack any part of musicology. It's just a different approach to the sort of standard approaches in another musicology. - Very cool. I'm extremely excited to follow what comes out of the Natural History of Song project. - Thanks, yeah, I'm excited to do it. - Yeah, it works. So we don't know what's going on. - Yes, it works. - Excellent. Well, Sam, thank you again for your time. This has been a really awesome conversation. I'm sure the listeners are gonna really enjoy it. - Yeah, I hope so. Thanks so much for inviting me, Kyle. - All right, and until next time, I want to remind everyone to keep thinking skeptically of and with data. - For more on this episode, visit datascapic.com. If you enjoyed the show, please give us a review on iTunes or Stitcher. (upbeat music) [MUSIC PLAYING]
