Cognitive Daily

Just by listening to music, we can learn a lot about its structures and conventions. For example, even you have no musical training, you can tell that something is wrong with this scale (it’s followed by a proper C-major scale):

But we learn a lot more than just standard scales when we listen to music. When you’re exposed to a particular type of music for many years, you learn much more. Consider the following sequence of chords:

Anyone who’s been raised listening to Western music should recognize this sequence as an appropriate musical phrase (if you don’t read music, don’t worry — I’ll play it for you in a moment).

But change just one of the chords, and now something seems “off.” You might not be able to identify the third chord in this sequence as a Neapolitan, but you’ll know that doesn’t sound right when you hear it:

Even worse is when the final chord is changed, as in this sequence:

Now listen to this clip where all three sequences in a row are played:

See?

But surely we aren’t born with such sophisticated musical knowledge. When, exactly, is it acquired? A team led by Stefan Koelsh had adults listen to dozens chord sequences like these and monitored their electric brain responses. They found a different response when Neapolitan chords were played compared to the expected, in-key chord.

In a new study, they tested 5- and 9-year-olds using the same apparatus. The children were told to press a button when the chords were played by instruments other than a piano (this occurred about 8 percent of the time). The researchers weren’t interested in the other instruments; this was just to make sure the kids were listening. They analyzed the rest of the data and found the following results:

This graph shows the electrical response of a 5-year-old while the final chord was played. The dotted line is the response when the chord was in key, and the solid line is the response to a Neapolitan chord. The two arrows show where the response is different, starting at around 200 milliseconds after the onset of the chord. Both 5- and 9-year-olds show responses similar to those of adults. However, if the third chord was changed, as in the second sequence played above, the results were different:

Now there is no difference between the in-key chord and the Neapolitan. This differed from the previous results with adults, who, even with no musical training, showed a similar (though smaller) response to a Neapolitan in the third position. Even 9-year-olds show the same results as 5-year-olds. So though children as young as 5 clearly have a sophisticated cognitive mechanism for understanding music, as late as 9, they still haven’t developed a complete adult understanding of music.

Koelsch et al. note that language development follows a similar pattern: up to the age of 9, children can learn languages rapidly and recover quickly from disorders such as aphasia. After 9 years, both processes become more difficult. The team argues that this is evidence for a common system for processing both language and music.

Update: Since we’re seeing different accounts about whether people can hear what’s “wrong” with the clip, I created a poll to find out what portion of our readers can hear the Neapolitans. The poll refers to the second clip above (the chords, not the scales).

Koelsch, S., Grossmann, T., Gunter, T.C., Hahne, A., Schröger, & Friederici, A.D. (2003). Children processing music: Electric brain responses reveal musical competence and gender differences. Journal of Cognitive Neuroscience, 15(5), 683-693.

Comments

#1 JohnJB
August 31, 2006

I couldn’t hear anything “wrong” in either clip. How unusual is this for an adult who has been listening to music all his life?
#2 Joe Shelby
August 31, 2006

I too, I suppose, am an odd one in that nothing sounded “wrong” (though certainly the last one was “unresolved”, but that’s a personal problem).

Then again, at age three I had a father cranking out Close to the Edge, Dark Side of the Moon, the first ELP album, and Tubular Bells, all on a regular basis, and by 5 I’d seen Fantasia and had no problems dealing with the Stravinsky or the Bach segments at all.
#3 Dave Munger
August 31, 2006

John, it’s possible that you have tone deafness (amusia). Check out this NPR article for a test.

This BBC article also has some good information, claiming about a 10 percent rate of tone deafness.
#4 Dave Munger
August 31, 2006

Joe, I think the “unresolved” sensation you’re getting for the last one is what I’m talking about. Western music has some standard patterns and resolutions, and the Neapolitan ending the final series is an example of a violation of those rules.
#5 Joe Shelby
August 31, 2006

so the question is at what point does the recognition of “unresolved” figure into it?

(its not a sensation – i used that term specifically because i know the resolution back to the tonic that it’s “supposed” to do, and from a composition standpoint ending themes in such a way gives way to further development, extending the work)

and its an assertion that maybe can be done on averages but says nothing for the individual, since people are exposed to different musics through different means.

its no longer the case where the only music one once heard at home was that which the parents could play and sing (or if they were rich enough, could afford to have played for them).

the most this type of thing can show is that there are cultural differences at play. I really don’t think there’s an inherent “resolved” sense of music, only over-exposure to (pop) music that quickly resolves itself, which in turn I believe reduces the chance that the child may develop the patience to accept longer musical forms.

i’d like to see more research on this, but the skeptic in me feels this is really mostly a nurture thing, particularly when one considers eastern music and how much of what they sing and perform has been “unresolved” by western ears (at least, until western music arrived through colonials and affected things).
#6 Dave Munger
August 31, 2006

Joe,

I think your hunches are right — this is definitely something that is learned, in much the same way that language is learned. However, that doesn’t mean that there isn’t a physiological component. Humans are equipped to learn language better at certain times during their lives, and research such as this suggests that the same thing may be true about acquisition of knowledge about music.

There are certainly cross-cultural differences in music, but differences within a culture (between genders, or between children and adults) can also be revealing — especially when you consider that these individuals have had the same exposure to music.
#7 Joe Shelby
August 31, 2006

especially when you consider that these individuals have had the same exposure to music

But my (again, skeptical but based on my experience) question is: have they? really?

except within traditional churches with full organ rigs, its unlikely most kids ever heard Bach by that time (and such a number is relatively shrinking as churchgoers are more and more the evangelical variety with modern (and lousy) “pop” music as their main style).

i was a kid of that age in the 70s to a father that kept “pop” and “disco” out of the house (on pain of death). the main focus was rock and progressive rock, with some classical and movie soundtracks. in the car to American Top 40 I was exposed to a rather wide realm of the “pop” music of the time.

contrast that to radio today that’s increasingly pigeon-holed and parents that never change stations. i think its harder to actually get the variety and easier to stay stuck on a particular style (often where there’s little to no development – if i’ve heard 10 modern “country” songs or 20, have i heard more music?).

i think it more likely that kids are exposed to less and less music (and decreasingly sophisticated music) as the decades have progressed. 10 hours of modern country or rap is not 10 hours of Beethoven or even Pink Floyd.
#8 Joe Shelby
August 31, 2006

another field of study on this is certainly the background music factor. how much music do we not consciously hear at that age because its background for the tv we’re watching and the talking we’re hearing.

we learn lots of hidden clues to musical intent through the association of emotions between a chord or chord progression and what emotional connection we have with what’s on the screen. is an Em chord really “sad” in and of itself, or is it “sad” because we always hear it with “sad” things to go with it. and if its the association, then who first associated it as “sad”? Bach (its in the St. Matthew’s Passion at a rather grim moment)? or did Bach learn it from something before?

hard part of that is getting a control, a person who had never heard an Em chord in their life before. not likely, of course.
#9 Joe Shelby
August 31, 2006

(responding to the new poll) i don’t consider the 2nd sequence to be “wrong”, merely different – i marked it down as “both” because i did at least recognize the intent of difference.

its “a-tonal” but so is a lot of western music since Mahler and Stravinsky. so would a child (like me) who was exposed to a-tonal music at an early age consider it “wrong”?

wrong is such a loaded term, implying a sense of musical rightness to which the wars over that were fought over 100 years ago between the Wagnerites and the Brahmsians, and in the end the Russians won.

(that last bit was an intentional musical-history joke, as the Russian works stayed on the sidelines and then exploded in popularity around 1910 just as the “war” was coming to an end with the deaths of many of the composers of the German traditions).
#10 Joe Shelby
August 31, 2006

i hate when sb.com’s comments system eats my urls.

http://en.wikipedia.org/wiki/War_of_the_Romantics
#11 JohnJB
August 31, 2006

I couldn’t hear anything “wrong” in either clip.

John, it’s possible that you have tone deafness (amusia). Check out this NPR article for a test.

I always suspected I was tone deaf. I was surprised when I learned that the first six notes of the chorus to “Jingle Bells” were the same tones – I think I hear rhythm as tone (not sure of my terms. here)…
#12 JohnJB
August 31, 2006

I should add that I did sorta catch the “unresolved” chord sequence. But at first couldn’t call it “wrong.” Being teased or left hungry can be pleasurable in small doses.
#13 Laura
August 31, 2006

While I realize you most likely recognize this – the poll isn’t really valid because it begs the question by asking if someone hears something wrong. It would make more sense to simply ask questions about what people are hearing, but I guess you couldn’t wrap that up in a simple multiple choice question. I’m with Joe on this issue.
#14 Stephen Downes
August 31, 2006

I could very clearly hear that they were different, however, since I was not familiar with the piece of music, it wasn’t clear to me that the other two were ‘wrong’ (though the sequence and the text told me to perceive the first as right and the others as wrong).

What I found more interesting was that the different sequences resulted in different expectations. This was clearest between 1 and 3 (both of which, again, were equally valid). 1 sounded like it was finishing, while 3 sounded like it should go on a few more notes. But this is cultural, isn’t it? The expectation that songs finish on certain notes?
#15 Kris Shaffer
August 31, 2006

Regarding the article:
This isn’t so much interesting because kids don’t have a “complete adult understanding of music” or at what age one might acquire such an understanding; this is interesting because of what pattern did produce a cognitive change and which did not. The pattern which did not elicit a different response from the “correct” pattern in kids is one in which the “right” chord and the “wrong chord” have the same tonal function (the F major and D-flat major–Neopolitan–both fill Riemann’s Subdominant function in the key of C). Where the two progressions elicited different responses is where the two chords fulfill different functions (D-flat major is Subdominant, C major Tonic). So, in this experiment, the kids seem to have a relatively accurate conceptualization of the stereotypical progression in terms of tonal function, but not a precise one in terms of scale degrees or chord roots.

Funny, my undergraduate students have a lot of the same problems in ear training. Since the chords which share function share one or two common notes and come in all the same places and work the same way, it’s harder to hear the difference.

By the way, an interesting correlary to the poll you have now is to convert all of these progressions to C Minor. The reason the Neopolitan sounds “wrong” in the second progression is not that F Major is expected exclusively here, it’s that Neopolitan is expected to occur in minor keys far more often than major. In minor, there is only one note difference between the Neopolitan and the Subdominant chords, and the voice-leading is much smoother. Hence, the Neopolitan adds color without jarring the listener. In major, there are more discontinuities, and the voice-leading is a nightmare.

On the other hand, the third progression would sound just as wrong in minor as in major, because the tonal functions differ and there are voice-leading problems in both keys.
For those who found the third progression to sound “unresolved,” it is probably because the progression ended on the subdominant function (also called dominant-preparation) which is an intermediate function, rather than the tonic function which typically begins and ends tonal pieces and major sections of tonal pieces.

Regarding the above comments:
A lot of research has been done into the emotional aspect of music. David Huron’s new book Sweet Anticipation: Music and the Psychology of Expectation asserts that it is psychological expectation derived from the statistical properties of music developed over time which prompts most emotional responses to music. This book builds on Leonard Meyer’s Emotion and Meaning in Music. Some of the studies cited by Huron suggest that our minds are pretty good at separating genres/styles from each other, so that chord progressions common from Rock and Blues which are uncommon in classical music sound just as out of place to our modern ears in a Bach chorale or a Haydn sonata as they always did, and vice versa.
As for background music and enculturation, check out Robert Fink’s Repeating Ourselves: American Minimal Music as Cultural Practice.

Random:
A Neopolitan chord in C has a D-flat, not a C-sharp (as notated above). Though they both sound the same on a synth, the D-flat looks “wrong” to someone with a “sophisticated . . . understanding of music” (that was tongue-in-cheek, by the way); and to 19th-century composers, there is a huge difference between the two in terms of their function, voice-leading, and even tuning.
#16 Scott Spiegelberg
August 31, 2006

The second progression is not wrong, or even atypical for Bach, Mozart, Brahms, etc. While Neapolitan chords are slightly more associated with the minor mode, they are found in major keys quite often. And the second progression uses the Neapolitan chord correctly (though it is misspelled in the score) both in harmonic function as a predominant and with voice leading. The third progression does have an atypical cadence, though still not atonal as a previous commenter suggested. This type of deceptive cadence would be found in very late Romantic music, as an alteration of the more typical V bVI cadence. (I wrote this before reading Kris Shaffer’s comment above. That answer is more eloquent and complete than mine.)

But to continue, this study shows the problem with some music cognition research. Cognitive scientists assume they know enough about music from taking piano lessons as a kid, to make decisions about right and wrong chords. Just as I need to consult with psychologists when I design experiments, these neuroscientists and psychologists need to consult with music theorists on appropriate stimuli and interpretation of data.
#17 Joe Shelby
August 31, 2006

Kris: bookmarked for future reference. thanks.
#18 Joe Shelby
August 31, 2006

Scott: I realize its not specifically “atonal” by modern composition definitions (where something is atonal because its missing the concept of the tonic entirely). the progression is in C and the sound is still in the key under certain rules. the notes would have to really be off to fit a true atonal definition, particularly given the chords on either side of it.

as for the experiment, well, using the words “right” and “wrong” is certainly a potentially error-prone approach as we’ve seen. given that, then what should be measured in such experiments? recognition of dissonance? (is the Neapolitan chord even technically a dissonant chord, or is it merely its place in the sequence that’s troublesome?). reaction without provocation? if the dissonant chord is really there, i think having the kids “push a button” to make sure they “pay attention” should have been irrelevant because they hear it regardless of how deep they’re listening, and if there’s nothing else to do, they’re going to use at least some of their attention to it. maybe that should have been a control case first – does attention and listening to the music make a difference in the reaction? (or has that already been addressed in previous studies on this age group? i haven’t looked at the reference list yet).

as for “an adult understanding” of music, I’ve found that most adults i’ve met have NO understanding of music, merely “i know what i like”, often being what they’ve listened to for most of their lives. hence possibly how classic rock radio ranks high on the airwaves (and dominates commercial soundtracks) even though its been the same (now 35-45 year old) music for 20 years since the market started. there perhaps are (subconscious and bio-electric) adult reactions to music that maybe are different than in young children, but i wouldn’t use the term understanding in such a context.
#19 Matt Todd
September 1, 2006

I notice the differences, but I also don’t think they are too bad, nor unresolved (necessarily). It sounds like a very interesting beginning to a song (if a bit simple).

However, I listen to a great deal of experimental or indie music. I find myself attracted and enamored by interesting and explorative tunes.

I think, though, that there might be a bit of a shortcoming to this little test: the music we listen to tends to be much longer. I understand that the testers were children in this case, and attention span is pretty important, but perhaps longer samples (2-7 minutes long) would provide us with some more interesting if accurate results!

Sidenote: when I wrote “(if a bit simple)” above, I originally had written “dated”. I find it interesting how my mind associates simpler music with being dated. (To be sure, by dated I mean from this and the past centuries.)

M.T.
#20 Dave Munger
September 1, 2006

Great discussion going about what’s “right” and “wrong” in music. The point here, really, is about expectations. In the Western musical tradition, the first chord sequence represents standard expectations. The other two sequences violate those expectations, and so seem unresolved. Commenters are right when they say that this doesn’t mean they are wrong. I agree with you, and that’s why I put “wrong” in quotes.

However, I think the poll results make it clear that this is part of the typical Western adult’s understanding of music. That’s all I meant when I said “sophisticated” in the article. I didn’t mean to make a value judgement that preferred standard music to more innovative forms (as an aside, however, I’d point out that understanding the innovation requires an understanding of the standard).

A Neopolitan chord in C has a D-flat, not a C-sharp (as notated above). Though they both sound the same on a synth, the D-flat looks “wrong” to someone with a “sophisticated . . . understanding of music” (that was tongue-in-cheek, by the way); and to 19th-century composers, there is a huge difference between the two in terms of their function, voice-leading, and even tuning.

Good catch, Kris! I was lazy and just captured the notation that was automatically generated by Garageband. Sorry about that.
#21 Scott Spiegelberg
September 1, 2006

I have to repeat that the second progression does not violate standard expectations, unless those expectations are based purely on diatonic-only music. And very little music in our popular culture is diatonic-only other than some pop. Film music, TV jingles, much of rock and hip-hop, these genres are loaded with chromatic chords that include the Neapolitan.
#22 Dave Munger
September 1, 2006

Scott,

I’ll take your word for it on the second progression not violating standard expectations. I know that you’re much more knowledgeable about music than I am (and I love your blog, by the way). But how, then, would you explain the data?
#23 Kris Shaffer
September 1, 2006

But how, then, would you explain the data?

I’d like to offer a possibility, even though I know the question was addressed to Scott.

To use Gjerdingen’s terminology, the first progression is the prototypical instance of the schema for a tonal progression (I IV V I, or T S D T). The second progression (I flat-II V I) fits the schema (and is still T S D T), but is more on the fringe of typicality. This is what gives it its color. It is less expected than the prototypical progression, causing some sort of orienting response or limbic jolt, but upon retrospection, it fits the rules. Thus there is a chemical response in many listener’s brains to the second progression which is different from that of the first progression, but that doesn’t mean that it doesn’t fit the pattern.

The third progression is even less expected (V flat-II instead of V I), and does not allow for the same positive evaluation within the tonal system upon subsequent reflection. Hence this one is more jarring than the first two to most tonally encultured listeners. However, were the composer to take a second pass at the progression and finish it “correctly” the second time, then the unexpected Neopolitan can be accounted for in the schema of a “deceptive cadence,” which typically leads to a true closing cadence in the following phrase.
(This expectation-evaluation analysis is a simplification of Huron’s ITPRA theory from Sweet Anticipation. Also, Gjerdingen has a book on musical schemata coming out sometime in the next year or two. In the mean time, he has several articles on JSTOR.)

Since I haven’t read the actual study, I hope this helps in interpreting the data in light of current theory on tonal harmony. Let me know if I’m on the right track. And I’m sure Scott will have some good stuff to add. (Sorry I hijacked your question!)
#24 Joe Shelby
September 1, 2006

i still think familiarity may be a factor.

perhaps they should run the same sequence (the 2nd one, which has the “jolt” in it on that graph) several times (with the different instruments for kids attention; keep the other things constant) and see if the jolt is still there on the 3rd or 4th listen? at what point does it become familiar enough to not cause that stimulation?
#25 Kris Shaffer
September 1, 2006

Familiarity is the key factor. That’s how we learn musical schemas.

There definitely is a point at which the jolt is minimized. But there’s a difference between hearing the same piece of music enough times that nothing is unexpected in that piece and hearing enough tonal music that incorporates enough of the same devices that we change our schemata for tonal functions. In other words, we have expectations for a given work based on hearing it before, and we have expectations for the genre (Huron’s veridical familiarity and schematic predictability). Both come into play, and this is one of the reasons familiar music can still grip us emotionally. The tension between the work and the larger style is still present. Though this is hard to hear in a three-bar progression.
#26 Nicole Brockmann
September 1, 2006

I’m a colleague of Scott’s, who wandered here via a link from his blog — and I’m not principally a music theorist (I’m a violist, so watch out!), so I’m not sure I’m up to the level here. But the topic is interesting, so I hope you’ll bear with me if I make no sense.

To me, neither the 2nd nor the 3rd progression sounds “wrong.” I hear what you’re after, but I can’t call it “wrong,” unless the assignment was to write a I-I6-IV-V-I progression. From a performance perspective, I’d understand exactly how the tensions build and resolve in the 2nd phrase because of the chromatically altered predominant (here, a Neapolitan, but other chromatically altered chords appear in comparable positions in other situations). The 3rd phrase doesn’t sound wrong to me either — it’s just not finished! For it to be “wrong,” it would have to be so far removed from what I know of harmonic language that my ears couldn’t find a way to relate it to any of the tonal functions I know of. Introduce a tone cluster here, or something atonal? Yes, that would seem “wrong” to me — but these examples aren’t far enough away from the norm.

Something that I don’t think has been mentioned is the effect of the harmonic rhythm on the perception of rightness and wrongness. In the current arrangement, the predominant chord falls on what we are taught is a “strong” beat — therefore, we expect that it will have a certain tonal solidity. Maybe it’s this expectation that is thwarted with the Neapolitan…? How would our perceptions be different if you rewrote the progression in 3/4, with not all chords having the same rhythmic values? Would we perceive the Neapolitan as more of a passing chromaticism (and therefore less “wrong”) if it occurred on beat 3 and passed into a strong V chord (or better, a cadential dominant) that took up the whole next bar?
#27 Laura
September 1, 2006

Nope — nothing “wrong” with the scales nor the chord progressions. I still have a problem with the premise begging the question.
#28 Asger Petersen
September 9, 2006

Is the sound clip one that was generated for the purposes of this post, or was it (or an identical one) used for that actual experiment described? I ask because the sound and the notation don’t match completely. The chord progressions are the same, but the top voice differs in the two last notes – in the audio, it’s got G-G instead of F-E in the first two examples, and G-Ab instead of F-F in the last one.

In any event, one possibility for error in the experiment is that the top voice (which is more or less by default the ‘melody’ of the piece) differs between the three examples. I suspect that the difference between G-C-A-F-E (example one, notation) and G-C-Ab-F-E (example two, notation) is subjectively smaller than that between example one and G-C-A-F-F (example three, notation). And in the sound clip, the difference between G-G (the two first examples) and G-Ab is striking.

If one were to design another experiment, I would recommend using examples where the melody is either completely identical or at least only differs by chromatic changes. Otherwise, it’s impossible to know if people react to the harmonic difference or the melodic one.
#29 Kim Boone
September 13, 2006

WeLl, yes, I can hear that some chords are a bit “off”, but I wouldn’t call them “wrong”.
Could this have anything to do with my love for jazz and/or Tom Waits?..
#30 Kandi
November 27, 2006

I very easily detected the wrong notes in the second and third sequences. I’ve been musically trained for the past four years and I think it may have helped me. Even though I play in a completely different key than piano (the key signature is sait to be a C I believe, while with no sharps or flats, it would be a Bb scale on my trumpet) I could still tell by reading the music that it would be completely dissonant, after a small mental transposition.
#31 Kandi
November 27, 2006

I very easily detected the wrong notes in the second and third sequences. I’ve been musically trained for the past four years and I think it may have helped me. Even though I play in a completely different key than piano (the key signature is sait to be a C I believe, while with no sharps or flats, it would be a Bb scale on my trumpet) I could still tell by reading the music that it would be completely dissonant, after a small mental transposition.