Listen to the following music clip.
Last week on Casual Friday, we asked our readers to explain what it's about, in concrete terms. Did you get it right?
Chances are, you did not. It's a selection from Claude Debussy's La Mer, from the movement intended to represent the wind and the sea. Only 36 of 357 respondents answered correctly. Even when I gave half-credit for mentioning either the wind, or a storm, or waves, or a boat, only an additional 90 got it. Most respondents -- over 200, in fact, got it completely wrong.
I picked seven different clips like this, from seven different works that were all intended by their composers to represent specific things, not just emotions or adjectives. I tried to pick pieces that seemed relatively obvious, based on the composer's initial intentions. I scored each response on a scale of 0 to 2, with 2 being perfect, and 1 meaning some portion of the response was correct. The average score was a mere 0.38, and 72 percent of the time people got the answer completely wrong.
I was quite generous with my scoring. Take this selection, from Camille Saint-Saëns' Carnival of the Animals:
The movement is called "Elephants," but I also accepted bears, hippos, cows, or any large animal as completely correct. Saying the work was about fat people counted as one point. Still, the average score for this piece was just 0.65. Despite the fact that the scores on this selection were the highest for any piece, most people still got it completely wrong.
But most people who took the test didn't have much musical training. Surely people with more training performed better, right?
Actually, they did score better -- just not much better. Take a look at this graph:
Music majors had an average score of 0.61, compared to non-majors' 0.36 -- meaning they still got it mostly wrong. Professional musicians fared even worse (though still significantly better than non-pros). Interestingly, just having attended a classical music performance within the last month was enough to predict a slightly higher score on the test.
These groups all also said they were more familiar with the works:
That said, when you consider the fact that this scale went from 1 to 5, on balance most people weren't very familiar with the works. Indeed, familiarity with the works didn't correlate significantly with accuracy in identifying what the works were about.
Overall, for our readers, it seems that just listening to a musical work isn't enough to reveal what it's about. While there is ample evidence that people can recognize some emotions in music, the same doesn't appear to be true for more concrete subjects of songs -- even when the composer is overtly trying to tell a story.
But maybe the problem was how open-ended our initial study was. What if we gave you a multiple choice test? Would you do any better? Here are the other five clips from the study. How many can you get right?
Clip 1:
Clip 2:
Clip 3:
Clip 4:
Clip 5:
Bonus points in the comments if you can identify the composer and title of each of these works (but don't peek before you answer the poll!).
Update: Looks like even with a multiple-choice test, we're still not very good at this. Here are the answers:
1. Ravel, Gaspard de la Nuit: Water nymph
2. Sibelius, Tapiola: Forest
3. Strauss, Don Juan: Duel and death
4. Grofe, Grand Canyon Suite: Sunset at Grand Canyon
5. Schoenberg, Pelleas und Melisande: Parting lovers
As I write this, respondents chose the correct response most often on just two of the clips: 3 and 5. And with clip 5, only 28 percent picked the correct answer.
Finally, a response to some of the commenters who suggested that "music can mean anything." I agree that there are a variety of possible interpretations of any musical work, but these clips were picked specifically because they came from tone poems, which are constructed with a specific story or text in mind. The question isn't what the music objectively "means," but whether by listening to the music, we can figure out what the textual basis for the work was. For the most part, we can't.
- Log in to post comments
Interesting! I'm looking forward to the follow-up.
This is not science. Like most "Cognitive Science" It's just woo.
I'd like to suggest what I'd like to call "Mr_G's law": any field of endeavor that calls itself "_ science" in its self description, is not science. This would apply to "political science", "social science", "Neuro Science", "Cognitive Science", etc.
You people are frauds. That you are allowed to promulgate this crap on ScienceBlogs makes me think the law should apply to blogs as well.
I wonder if the whole movement would make it more accurate. These clips are too short, in my opinion, to get the imagery of the whole piece.
One of your possible answers for clip #4 gave me a hint as to what the piece was, even though I'm not familiar with it. I'm going to take a wild guess that that was from Grofe's Grand Canyon Suite. I still don't have a clue on the others.
I think I did pretty well..100%, but then I suck at math. I might be blind too. I can see where I scored in comparison with others, but I couldn't identify with certainty the works or their composers. Is that info listed?
I think Mr G needs to get-in on some of what's going around this merry month of may. Even the peasants are out doing it in their fields.
Really, Mr_G? You don't believe that psychology, sociology, and anthropology are "science"?
Funny, I thought science was defined by using the scientific method to answer questions about the world. I would love to know what YOUR definition is.
To my ear (mind?) they all sound like soundtracks to Disney cartoons.
Mr_G: citation needed; or at least an argument why cognitive science isn't science -- instead of the "Proof by Assertion" technique you employed above.
Science has something to do with objective reproducible results.
You don't find that in ""_ sciences".
Particularly in crap like we find here.
For one, this is a blog post, not a real journal article. I think it's obvious to everyone that there are serious methodological problems in relying on online polling as a sole source of data. But if you're claiming that *no* cognitive psychology experiments have *ever* led to reproducible results, you're either seriously misinformed, deluded, or just trolling for the hell of it.
Mr_G, neuroscience isn't science? (Or is "Neuro Science", as you put it, something different than neuroscience?).
Damn, now I feel annoyed that I did all those experiments and data analysis and such for all that time. Oh well.
Neuroscience isn't science? Really? Determining what neurotransmitter is responsible for the sensation of taste in taste receptor cells sounds pretty concrete to me, not that that is what makes it science.
Guesses:
1. Winter
2. Spring/insects
3. Life & then Death
4. Overlooking a magnificent valley.
5. Soaring/flying.
How confident am I? Zero! I am actually
confident I got none of these right. I am
not surprised--I have never thought that
instrumental music on its own could convey
narrative well other than in fringe cases
in which an instrument is carefully imitating
an animal call. And isn't that to be expected?
Nature and daily life is not accompanied by
orchestral symphonies.
I wonder if the whole movement would make it more accurate. These clips are too short, in my opinion, to get the imagery of the whole piece.
Mr_G: citation needed; or at least an argument why cognitive science isn't science -- instead of the "Proof by Assertion" technique you employed above.
Neuro"science", BTW, is a latter day version of phrenology. Good luck with that.
Yes, truly an editorial sound bite is the ultimate proof of the fallacy of cognitive science.
I am constantly amused when people define science by its topics rather than methodologies. Consider astrophysics; it's given us gems such as 'dark matter' (unobservable by definition) and 'string theory.' But because it bears the name of physics and involves massively complex mathematical equations, its status as an upright science is apparently untouchable -- never mind that it infers the existence of unobserved phenomena to a much greater extent than any modern cognitive theory.
Out of interest, did you define in advance what percentage of participants would have to have correctly identified the themes for you to conclude that we do 'get' what a song is about? I think the open-endedness of the study adds to its accuracy and interest, because (as with the poster above about the Grand Canyon Suite) multiple choice answers are themselves suggestive. The scoring showed how many responses were correct/not correct, but for each clip were there some wrong answers that were more common than others? For example, after listening to La Mer,was there a large group of respondents who recognised it as birds flying, buffalo migrating or something completely different? Or was each one of the wrong answers bizarre in its own individual way? Would another conclusion to this test be: 'Music does possess innate meaning. It is possible to 'get' what a piece is about without being told, and many people do, but most don't.'? Finally, am I too interested? The title is 'Casual' Friday, after all.
Mr_G: so research on (for example) Parkinson's, dementia, autism, education/learning, neural networks, addiction, depression, dyslexia, and decision theory... you believe that all these things are useless?
research on (for example) Parkinson's, dementia, autism, education/learning, neural networks, addiction, depression, dyslexia, and decision theory... you believe that all these things are useless?
Not at all but it's worth noting that the only effective treatments for autism are behavioral. Additionally:
Neural networks help us understand how behavior is produced, and, unlike classical AI provide a non-rule-based model of what's going on. As Skinner said: "Psychology is premature Neurology".
Mr_G is a troll. Can the owners of this blog do anything?
Anyway, I thought the poll was fun, but error bars are always welcome, ya know. Now that you taught us how to read them, shouldn't you use them? :)
zia: Mr_G is not a troll and it's not clear that the owners of this blog can do anything but post gee-whiz meaningless crap. I guess you prefer that. I'd suggest the Sunday Entertainment Section in you local newspaper if that's what you're after.
Mr_G, this B.F. Skinner shtick you're doing is a gas! This is better than Jack Van Impe Presents! Keep it up!
Mr. G. isn't exactly a troll, but his psychology is out of date. Skinner served a useful purpose in promoting scientific psychology, but his persistent advocacy of the strong form of behaviorism in the face of overwhelming evidence to the contrary has made him in some ways irrelevant today.
Your appeals to Skinner remind one of quoting scripture -- an equally convincing technique. But if that's the language you speak, here's another sound bite for you.
Mr. Munger, perhaps you could present us some of the "overwhelming evidence to the contrary". I've studied this since I worked at Skinner's Learning Research and Development Center almost 50 years ago and have yet to see any. If you could provide some it would be most appreciated.
Jake, the Bowers quote is the strawest of men. We are not trying to explain the behavior of apples here.
What is most amusing about Mr_G is that his "law" has been around for at least 20 years. That's when I first heard it. If you're going to be nasty, it is best not to be claiming very old jokes as your own inventions.
Oh, and by the way, it would apply to "behavioral science" as well, my dear Skinner fan.
Cranky: Ah, yes. Watson's classic "Psychology as the Behavioral Scientist Views It" and Skinner's "About Behavioral Science". Guess you've got me there.
Anything of substance to say?
To all. Written in verve, hang tight:
Wrong. And for illustration, the music for Appalachian Spring was written first, the name applied later because it had none, a dance applied later for there was none. All the interpretations are imagined by the listener, and are mysterious to the writer Aaron Copland, which he stated himself with some amusement.
The music means what it means to you.
A poem (with words obviously) also may also be seen as meaning what it means to you.
This meaning of poems is a famous "debate" in literature over many decades of the twentieth century. Famous? 'If it's so famous who come I never heard of it,' said my five year old.
[In that famous debate I like A. E. Housman's stance in 'The Name and Nature of Poetry.']
So if a poem is seen as meaning what it means to you, what about a speech? I'm afraid it too means what it means to you.
And in all these communications, what the listener hears can be different, even opposite, from what the speaker or writer intends to convey. So, the burden of the writer or speaker is heavy, so that his or her intentions may be understood by few, many, or none.
We carry meaning within ourselves, and the minds of some are like concrete.
The 'famous' debate on 'meaning' drifted, derived, into semantics naturally, and the name of Hayakawa's 60's magazine provides a conclusion for my terse verve.
You got them all right.
Etc.
I thought this was a really interesting poll. I am pleased to say that I got the one about elephants right; although I, like other commenters, viewed everything through the lens of a cartoon soundtrack.
I think that a music work about the sea is not supposed to imitate the sea; it is supposed to give a hint of the subjective universe that the author experiences when he/she sees the sea; in this way, depending on the author, the same musical excerpt might illustrate knights duelling, mountains crashing, driving without a licence and bad conscience after having one too many biscuits.
Interesting and relaxing break. But what is the purpose of this? Are the results used to some therapuetic reason or something?
Got sucked into this on false pretenses - these aren't "songs," which are "sung," these are "tunes" or "pieces" which are played - I still got sucked into the exercise though, and took my best guesses on the five pieces.
Now I'd like to know please where I can "peek" at the correct interpretations. Or is that another malapropism? Did you mean "a peek at the poll results" rather than "a peek at the correct answers" which is what your phrasing leads one to believe?
Survey results vary widely from my choices, and with these things it's generally better to see the right answer while the movement is still firmly fixed in one's memory. Do we get the answer key please?
I think you're missing something and overinterpreting the creative process.
If a composer (and this applies to all composers including popular song) attributes a theme to a melody they are probably doing it for reasons unrelated to the theme itself.
In fact in many cases they pull half-finished melodies and themes off the shelf (ie. out of their notebook) and whack 'em in. The attributation that "this represents elephants" is after the fact not beforehand.
This practice is especially visible in jazz where whacky names are given to themes that someone may have been working on for a while before pulling it together in a tune.
In other words there may be actually only the vaguest link between music and the attributation of its meaning which is assigned for completely different reasons:-
* commercial (common in classical and opera where the melody is hijacked for the larger theatrical purpose)
* humor (jazz, where the attributation is just the name of the tune and has nothing to do with the theme itself)
* and so on
It is much, much rarer for a composer to think "oh I need something to represent elephants here" and then create something from scratch. Even in film music - perhaps the only area where that is the motivation - composers still pull a lot of stuff out of their notebooks and repurpose it.
"The amazing thing about the dancing bear is not how gracefully it waltzes, but that it waltzes at all". Given the open-endedness of the original poll (giving us the option to sift all of human experience for a reply), I'd say that getting even numbers as high as you do suggests that something interesting is going on here. There's definitely at least some shared vocabulary arising in terms of musical themes, and I think it's a fairly new invention (last few centuries). However, I also agree with onefinemorning, who inquired about polysemy... Is there clustering in the results that suggests that a lot of people get a consistent grouping of wrong answers?
I think there's a lot of cultural priming going on here. I may not be familiar with the Debussy piece from the poll, but I've seen a huge number of storm scenes in movies, ballet, etc. And while Debussy well pre-dates film, he's probably also drawing on the same large musical literature for his tropes as everyone else is, so we have a shared vocabulary, even if we don't know it. In the case of those who get it wrong, is it that they just don't get it at all, or is it that they're primed for a different interpretation of that piece, because they saw some film with dying butterflies or dueling Jedi set to a similar piece?
It would be very interesting to present this to people without any of that musical priming (probably some tribe in outer Mongolia, given the global penetration of western cinema and music) and see what they made of it. But far out of scope for a Casual Friday, I fear.
Sorry, for some inexplicable reason I forgot the most famous example of what I was talking about at #34 (attributation after writing not before):
Yesterday, Paul McCartney
McCartney spent several weeks on this tune and called it "Ham and Eggs" during that time. He had no particular theme or topic in mind until he had finished it and started on the lyrics. At that point the tune got a small tweak at the point where it re-enters the verse (".... Suddenly.") but was otherwise complete.
You can't treat music as if it were text. It isn't and its emotional effects are different.
I think that a music work about the sea is not supposed to imitate the sea; it is supposed to give a hint of the subjective universe that the author experiences when he/she sees the sea; in this way, depending on the author, the same musical excerpt might illustrate knights duelling, mountains crashing, driving without a licence and bad conscience after having one too many biscuits.
Sorry, for some inexplicable reason I forgot the most famous example of what I was talking about at #34 (attributation after writing not before):
Yesterday, Paul McCartney
McCartney spent several weeks on this tune and called it "Ham and Eggs" during that time. He had no particular theme or topic in mind until he had finished it and started on the lyrics. At that point the tune got a small tweak at the point where it re-enters the verse (".... Suddenly.") but was otherwise complete.
So how would you score an answer to your poll? Are "breakfast", "morning", or "salt-of-the-earth pragmatism" equally valid answers to "recent lost love"? I think you'd have to say they were on your thesis that the title or topic determines the intent of the music.
You can't treat music as if it were text. It isn't and its emotional sources are different.
I think that a music work about the sea is not supposed to imitate the sea; it is supposed to give a hint of the subjective universe that the author experiences when he/she sees the sea; in this way, depending on the author, the same musical excerpt might illustrate knights duelling, mountains crashing, driving without a licence and bad conscience after having one too many biscuits.
Um, one problem with this study. Music is (typically) not *about* anything. Just because a composer calls a piece "the 1905 revolution" or whatever doesn't make it about the 1905 revolution. Even if the composer thinks its about the revolution, it might not be. So it is not surprising that we can't tell what music is about.
Considering that most classical music (by which I mean neither folk nor jazz nor popular) is culturally removed from us by centuries, that our understandings of our emotional selves have changed substantially, that the harmonic vocabulary has shifted drastically, etc, etc, I find this whole study just a wee bit on the lame side.
I imagine that you'd get much more interesting results were you to throw in Wynton Marsalis attempting to evoke a train, a metal song evoking a rape, or any anonymous funk song evoking sex with a pizza man.
The clouds and hills in a landscape painting don't look like real clouds and hills.
It's not just that they're in 2D and reality is in 3D - it's that the textures, shadows, surfaces etc. don't look like the real thing, even taking the missing dimension into account. And yet we can still comment on the artist on being "realistic" or "unrealistic", "capturing the scene" or "reproducing what they saw".
Why? Because there are, in our culture, conventions about how clouds etc. are represented in painting. These conventions change over time, which is why a landscape painting from 1700 doesn't look like one from 1900.
So it is in music. Soaring strings don't resemble "romance" in any way, high pitched stabs don't "sound like" fear, and trombone slides don't have anything in common with slapstick comedy. But if you watch a film, the incidental music tells you what emotion you're supposed to associate with what you see. If you remove the music you see how flat it generally falls, and if you watch a foreign film, sometimes the soundtrack is puzzling.
Smoky saxophones are sexy, kazoos are comical, arpeggios are dramatic, church organs are elegiac, sitars are mystical. We all know this vocabulary connecting sounds to moods and situations, and it's so familiar we rarely question it.
But vocabularies change over time, and are different in different cultures. That's one reason why a piece of music changes it's meaning - or loses it.
This is fascinating, and brings up great questions about reference in art and meaning in aesthetics. I think that the premise is a bit off however.
I agree that the composers were intending to reference a specific story or setting, but to insist that the setting or story is what the music is 'about' is misleading. There is simply no one-to-one identity between subject and referent when it comes to aesthetics - it is not that a certain piece of music maps specifically to a duel, but that they both evoke the same (or, more likely, similar) aesthetic responses. The problem here of course is that aesthetic judgments are notoriously fuzzy and that they have lots of overlap. If a composer is intending to evoke a nymph by a piece, we should EXPECT that people will potentially interpret it as invoking a butterfly (perhaps I'm biased, that was my answer) because the aesthetic judgments associated with a nymph are also closely related (if not, in some cases, identical) to those associated with butterflies, or rivers, or snow falling... 'Aboutness' needs to be carefully understood when talking about art, because while there is definitely intersubjective agreement at the coarse-grained levels of judgment, it becomes much trickier to pull apart meaning at smaller and smaller scales.
This isn't just an issue in music - poetry has the same difficulties (although I hesitate to think that this sort of phenomena is a difficulty - I think it's more of an asset). In high school and college English classes I was TERRIBLE at figuring out the meaning of poems (and none of my classmates were any better). Again this is an artifact of aesthetic judgments - some particularly evocative phrase might be MEANT to refer to the sea, but due to the (wonderful) ambiguity in our language, and the similarly ambiguous nature of our personal aesthetics (and I suppose those two are deeply connected), the sort of evocation that maps to the sea ALSO maps to the sky, the plains, 'seas' of sound or color, a sea of humanity, of space, or a more abstracted void. That is the beauty (and unique asset) of such artistic impression - it IS underdetermined. If a poet or a composer wanted us to know exactly what they're talking about, then they wouldn't use these formats - pop music is notoriously unambiguous, and prose is specifically for picking out a singular meaning.
The value of art is that it's meaning is underdetrmined, so that we may all agree that it is about SOMETHING, and competent judges (whatever that means) might even agree on what it is categorically NOT about, what it IS about will always be deeper than some simple referent in the world.
"The value of art is that it's meaning is underdetrmined, so that we may all agree that it is about SOMETHING"
This is why as an information worker I don't get art. A communication's value goes up the lower its information content is? We normally call that 'shallow', not 'deep' when we talk about people.
If we can't agree WHAT the 'something' a piece of art is allegedly about - and we agree that we have no way of electing competent judges - how can we actually be sure that it *is* about any of those things at all?
And if the value of art increases the less specific it is (or the less information content it conveys about its subject), wouldn't that mean that the best art is the vaguest / most general?
If so, then:
That means anything and everything, and nothing as well. I will give you no clues as to what it is 'about' - that's an archaic concept. It's about whatever you think. There are an infinity of possible meanings overlaid into this one finely-honed non-symbol.
That thought! There! Yes, that one too! Sir, you in the corner - you should be ashamed. And you too, madam. And that one. And that thought too.
It's the most densely-compressed artwork known to mankind.
I ought to win at least the Nobel Prize for Literature for this. Perhaps the Booker as well?
Sarcasm aside, I do realise that there is value in ambiguity... to a point. I think modern art however has made a fetish of ambiguity. Art must by definition communicate *something*. It has to restrict the selection space of possible moods, emotions, subjects, interpretations. And yes, it's useful to have art that can be repurposed in multiple contexts, and the more general it is the more useful it can be as emotional wallpaper for our lives.
(I guess that's why black goes with anything - it's the fashion equivalent of zero, or the empty set.)
But it seems like there must be *some* minimum information content that art must have in order to be interesting. And that means a maximum density of 'encoded meanings'.
The more meanings you layer into something, the more contradiction you're also adding. At the point you can legitimately read in both 'love is swell' and 'go kill your mother with an axe' you've probably put too much ambiguity in.
Is there a sweet spot between a completely null artwork and a highly specific one of interest only to a very small audience -- and if so, can we reason about that idea of 'balance'?
[post deleted]
Mr_G, if you can't refrain from ad hominem attacks, you will be banned from commenting on this site.
-- Dave
Cool! Trolls on Cognitive Daily! Dave, you've really gotten somewhere if your blog is troll-worthy. One should normally respond to trolls either by ignoring them or with pure snark, but in this case, I think it's worthwhile reminding Mr_G that this is "Casual Fridays." This was not supposed to represent a rigorous scientific study according to the exacting standards of the best cognitive scientist. It's supposed to a semi-humorous survey of the people who read the blog, about exactly as official as a conversation around the water cooler. That said, you're a troll. You live under bridges and try to frighten people into giving you attention. Yawn.
Instrumental music is not like an actual song, where the idea is elaborated in verse (why are pieces without singing referred to as songs nowadays?). Instrumental works reflect the imagination of the composer, and sometimes even a post production concept, as others here have pointed out.
A piece that directly quotes another work, like Mendelssohn's Reformation Symphony that quotes Beethoven and C. Wesley and employs extra effects like church bells, might be said to represent plainly something beyond itself. Otherwise the product of the composer's dream might be open to anyone's fancy, and that's a good enough thing too.
But a better thing, I believe, is to enjoy the program knowing in advance what the composer tried to imagine in sound; to share the dream. It's not a cipher job, it's only music.
Pedantic point regarding the title of this post.
The word "song" implies singing. Instrumental pieces that do not include singing are not "songs".
And while I'm at it, I think I would say that this is not really a scientific examination of the brains of the listeners, but rather a comparison of the ability of the artist to convey an intended meaning through his medium.
This point would be more obvious if we switched to painting or sculpture. The subject matter of Whistler's Mother is far more obvious than that of a typical work by Jackson Pollack.
I wouldn't go so far as to call this "woo". But it doesn't seem terribly interesting from a scientific standpoint. The nature of the data being tested is not being held to terribly rigorous standards.