Musical Predictions, Redux

In response to my recent post on the neuroscience of musical predictions, Alex Rehding, the Fanny Peabody Professor of Music at Harvard, wrote in to offer a musical theorist perspective. He makes several excellent points, and complicates the neuroscience in useful ways, so I thought I'd reproduce the relevant parts of his email below:

The point you raise in your latest posting -- about expectation and prediction -- is one that has fascinated music theorists pretty much for the last 200 years. I realize that yours is a science blog, and I'll try my best to resist the urge to add too many traditional music-theoretical talking points.

There is one interesting problem with this model, though, that has generated some interesting discussion in the music-theoretical world over the last twenty or so years. If we derive pleasure from anticipating potential connections - and especially being surprised by thwarted expectations - then it becomes difficult to explain why we would want to listen to a piece more than once: the novelty factor wears off, the uncertainty factor becomes less pronounced. In principle, the piece should get less interesting each time we hear it. Experience, however, shows that this is not the case: we greatly enjoy re-hearing familiar pieces. The whole recording industry makes a lot of money on the basis of this phenomenon.

Personally, I'd frame the discussion slightly differently. (And please take my views with a huge grain of salt. I'm a lay listener, not an expert.) I think our inexhaustible need for new music - we want the latest Rihanna radio hit - demonstrates that, once we memorize a piece of music, it grows a little stale. The essential surprise is drained out of the notes, so that there are no subtle patterns left to learn. And that's when our attention begins to wander, and we buy the current pop phenom on iTunes. While there are certain songs I will be listening to forever - most of Blonde on Blonde, late 70s Bruce, Astral Weeks, Otis R., a few Pavement songs, Wilco, a little Bright Eyes, etc. - I'm always struck by the short half-life of most of my music. The stimulus goes from intoxicating and enthralling to tired and tedious in a few short listens. And so we keep on consuming, searching for another shot of acoustic excitement. I think the recording industry makes a lot of money on this phenomenon.

We can now see the neural anatomy that makes this cultural learning possible. The auditory cortex, like all our sensory areas, is deeply plastic. Neuroscience, stealing vocabulary from music, has named these malleable cells "the corticofugal network," after the fugal form Bach made famous. These contrapuntal neurons feed back onto the very substrate of our hearing, altering the specific frequencies, amplitudes and timing patterns that our sensory cells actually respond to. The brain, in other words, tunes its own sense of sound, just like violinists tune the strings of their instrument. One of the central functions of the corticofugal network is what neuroscience calls "egocentric" selection. When a pattern of noises is heard repeatedly, the brain memorizes that pattern. Feedback from higher-up brain regions reorganizes the auditory cortex, which makes it easier for us to hear that same pattern in the future. So when we get sick of the latest top 40 jingle playing on the radio, these are the cells to blame. Their infinite capacity to learn means that we quickly get bored.

This, of course, raises the larger question of why certain pieces of music don't go stale. Why are we still listening to Bach's fugues, or Beethoven's symphonies, or Kind of Blue? What is it about these particular soundwaves that allows them to evade the corticofugal boredom? I'd suggest that their place in the canon is inseparable from their ambiguity - their ability to encourage a multiplicity of interpretations - so that new listens reveal new elements to listen for. In other words, we are continually surprised by their sounds, by the capacity of the music to subvert our expectations. Frank Kermode famously argued that literature worked the same way: What makes a novel or poem immortal is its complex indeterminacy, the way every reader discovers in the same words a different story. The same book manages to inspire two completely different conclusions. But there is no right interpretation. If there were - if there was only one way to read Hamlet - then the words would be far less interesting. The art that endures is the art that never loses its capacity to surprise.

I think Professor Rehding makes another essential point in his discussion of rhythm:

The other point that I think is quite important in this respect -- and that's one that most music theory traditionally ignores -- is the importance of rhythm and pulse. We don't simply yearn for the eventual return of the tonic, but we want it to fall in one particular place (or a small range of places) within the temporal order of the piece of music. Composers generate a lot of tension by carefully manipulating the temporal features of their music: a musical climax is not merely about writing loud music but also about the careful gradual build-up and the decline afterwards.

This idea is an even more urgent given current musical trends. Over the last few decades, popular music has been transformed by its rhythms, so that some rap songs consist of nothing but words propelled by a pulse. Why is this so exciting, at least for the auditory cortex of people under 25? Does rhythm also take advantage of our musical prediction machinery? Or is it a kind of acoustic scaffolding, making it easier for us to follow the subtle patterns in the rest of the song?

I won't even venture a bad guess to those questions. But I want to thank Professor Rehding for his feedback and comments.

More like this

The "fug-" root means to move away from, as in "centrifuge". In "fugue state" it seems to mean to fly away, as representing flight from normal consciousness and memory. When I first hear it, I thought it was meant to suggest the automatic, repetitive movement of a fugal theme. This is poetic, but inaccurate.

I prefer your derivation, but that moves away from the actual historical roots.

By Roger Bigod (not verified) on 01 Feb 2010 #permalink

Interesting post. I don't think this phenomenon is limited to music. It applies to drinks as well. It explains the continuing popularity of Coke and Dr. Pepper (with its 23 flavors) vs. an orange soda which is far less complex. Typically things that are complex take more work upfront to appreciate but then give back for longer.

The trick for pop artists is perhaps to find the right balance between (or blend of) simplicity or complexity.

The stimulus goes from intoxicating and enthralling to tired and tedious in a few short listens. And so we keep on consuming, searching for another shot of acoustic excitement. I think the recording industry makes a lot of money on this phenomenon.

I agree that this is closer to how the recording industry operates than the description Prof. Rehding gave. Most popular music is crud. It has always been thus and always will be thus. One of the reasons people remember fondly the music of their preferred era, whether it's the Renaissance, the 80s, or anything in between, is that the mediocre stuff has largely fallen down the memory hole. Of the music that has been produced in the last 20 years or so, we don't know yet what will be remembered and what will be deservedly forgotten. Classic example: "You Light Up My Life". At the time (late 1970s) it was the biggest thing in popular music. Today, few people will admit to having liked it. There are lots of other former #1 hits that are now generally hated.

There is also a phenomenon of liking the music you were exposed to at a certain age, which is the phenomenon Prof. Rehding referred to. He is entirely correct that the recording industry makes lots of money from this phenomenon, largely by repackaging old favorites in new formats. But for this phenomenon, there would be few or no recordings available of most classical pieces or popular pieces more than a few years old.

By Eric Lund (not verified) on 01 Feb 2010 #permalink

Has anyone looked into the musicology and neuroscience of the "hook"? Sometimes it's an unexpected modulation, but the match of lyrics with rhythm can do it.

There's also the phenomenon of the recurring phrase that someone "can't get out of his head". Surely a candidate for an fMRI study, especially the really sanity-threatening ones that won't go away. It would be interesting to know the recipe, and whether we can have that recipe again.

By Roger Bigod (not verified) on 01 Feb 2010 #permalink

What interests me the most in this area is the change in what you hear as you listen to a song the first few times. To me, a song is very different on the 5th listen than on the first. Another slant on this is that there are songs you love the first time you hear them and progressively grow tired of, and then there are the songs that you are somewhat indiferrent to on the first listen and then grow progressively more fond of. I am not a big fan of fMRI studies (for reasons of validity), but it would be interesting to look at response patterns across repeated listenings divided across categories of initial and eventual appeal.

By Rob Crooks (not verified) on 01 Feb 2010 #permalink

A few years ago when iTunes arrived, I re-discovered some progressive rock music (Yes) I hadn't heard for twenty five years. I was amazed to discover I could hum, sing and whistle my way through every single note and was moved to tears in the same passages I had been more than two decades before. Where had that been stored in my head for all those years?

Yet now that I've listened to those same pieces dozens of the times over the past couple of years, they've mostly lost their emotional power over me. Is it like the joy of running into an old friend for whom time has erased all but the best memories about them? Then after a few hours of discussion we're reminded of why we drifted apart?

By Dimitri Cherny (not verified) on 01 Feb 2010 #permalink

and if music that we love escapes our pattern recognition, then why doesn't everyone love ornette coleman, or anyone else who moves far from normal key and time signatures? to me it would seem like there are two needs to create a piece of music that people will love: 1 that there are always new patterns to be discovered, but also 2 that the brain perceives there are patterns to grasp in the first place.

There's also the phenomenon of the recurring phrase that someone "can't get out of his head". Surely a candidate for an fMRI study, especially the really sanity-threatening ones that won't go away. It would be interesting to know the recipe, and whether we can have that recipe again

The parallels to literature are very applicable. Some of the most timeless music takes us somewhere, a place we want to return to. The song doesn't let us stay there long. This is similar to tension with the tonic, but is different. There's a difference between yearning to go back to a place of joy and quenching thirst.

This idea of music taking us somewher was explored in the Romantic Era, almost literally. The music was story-based. (Bohemian Rhapsody might be a more modern parallel.)

But the most powerful form of the concept of "Music taking us somewhere", music that traverses time and space, is the gradual build-up and climax that Rehding mentions.

In my opinion, no form of music exemplifies this build-up better than trance (and other closely related forms). It's also multi-dimensional, so there's always new things to listen to within the same song. Thus, these songs tend to stick around. (How old is Sandstorm?)

Most of pop music tends to have a short lifecycle because its really not that great of music. It's a catchy melody that quickly goes stale. iTunes loves the short-term desire to hear a song. I'd agree with the argument of indeterminacy in great works. We want to discover new things and gain new feelings.

One of the elements of music that I find can really draw me into a piece is Counterpoint - where you have competing/complimentary melodies. I would guess that the draw is my brain trying to resolve these two ideas that are similar, and yet distinct.
In agreement with your point regarding our enjoyment dealing with missed predictions, my personal favourite drummers excel with this - most often their accents skirt the beat, leaving you with an unsettled feeling.

I would like to suggest that there are likely many different elements that attract listeners to particular pieces of music. So many different elements of the brain are utilized when listening to music, it's only likely that the competing Logic and Emotion elements of one's brain are both getting excited, and quite possibly for different reasons. How a song builds with regard to dynamics (soft/loud - like a Philip Glass piece) probably invokes the emotional parts of your brain more. Rhythm and tempo (and when chord changes occur) likely excite the logical/pattern prediction elements of your brain more. The best pieces of music are likely simple enough that the first listens can engage both logic and emotion, but I think its the emotional response that really will predict how long a piece will remain popular. Those songs that can get you to both emotional lows and highs will remain exciting for a long time.

I once saw a documentary on animals and song. In this documentary(i forget the name)they posited the idea that the human propensity for music and dancing may be rooted in bygone mating behavior similar to that of birds and other animals that use particular sound and movement to attract mating partners. If there is any truth to their theory, and I think logically it is plausible, then it strikes me that perhaps it is not left over behavior but that it still serves the same purpose. Blue whales many miles apart will sing long complex songs that are almost identical except for a few changes. After a bit of time a new "trendy" song emerges and they all sing that one with their own personal flair. A new trendy song comes out and you will hear a guy blasting it out of his car speakers for everyone to hear. Maybe he is just trying to look cool, maybe "trying to look cool" is another way of saying he is singing a mating song. Perhaps his song is an attempt to show his masculine dominance or intellectual dominance. We definitely read people by their musical taste and it seems that perhaps the way we enjoy, dislike, and remember music is just a matter of evolutionary biology. Maybe we process music the way we do because our brains want to attract mates, and finding something appealing is an attempt to stay ahead in the mating game.

By Timothy Schuchard (not verified) on 03 Feb 2010 #permalink

A bit off topic, but I don't believe I saw timbre mentioned in this post or the first musical predictions post. (Sorry if someone already commented on this.) Both posts focus on melody, and this one briefly touches on rhythm, but I believe timbre may be an important aspect of grabbing a listener's attention, and keeping hold. This is especially true with current popular music, which has a wealth of electronic tools with which to produce unique timbres that have never been heard before.

I, myself, find unexpected new timbres to be particularly appealing when it comes to current music. The West Coast "glitch hop" scene, including artists like Nosaj Thing and Flying Lotus, is one blossoming area in which, I believe, timbre plays an important role as a "hook" and as an important way to create patterns of tension and release.

I may be strange, but I'm sure I'm not the only one who does this - in addition to having a need for novelty and new music to listen to, I also have a need to listen to my very favorite recordings (which ranges from classical to jazz to rock, not so much top 40 pop, thank you) a zillion times over until I have memorized and can sing along with practically every note from every instrument. I'm not sure how that fits into the model Prof. Rehding proposes. Maybe that is the analytical part of my mind taking over.

Another observation: Although I am not a music producer, so my insights may be limited, I have noticed that almost all modern music has a lot of compression applied to it, which helps cover up bad tone, but which also destroys dynamic ranges in the music. The ear actually gets tired listening to it for too long. So I wonder if part of our desire to replace our top 40 tunes periodically stems from the fact that there is a key element of change missing.

On first thought it seems as if the rhythm of a given piece of music, once learned and programmed into the listener's auditory cortex, might possibly trigger the brain into a state of enhanced apprehension, allowing the listener to scrutinize and experience the music or lyrics more intensely, or in greater resolution and possibly with an increased span of impression. The familiar tonic, or a ring tone worthy hip hop beat say, could evoke this attention by rewarding the listener for their anticipation, imbuing the sense of confidence and achievement which in an ideal, stable mental condition would lead to greater levels of those neurotransmitters which lend such great depth and intensity to our experience. A distinctive melody acquired in such a way by the listener serves to reinforce a that semiconscious sense of accomplishment, a mindset of quasi-self-admiration in which credence is lent to the remaining multiplicity of facets exhibited by the particular musical piece regardless of whether we would have any interest in them were they not backed by the familiarized pattern. This would explain why songs we at first grow fond of but later begin to grow an aversion for leave us confused about what could have changed in the piece to cause the flop. Perhaps the process of inducting a new melody is the most exciting part and when mastered they simply begin to go stale. Yet also maybe in order to induce any sound as a musical pattern, we by default must leave all the more subtle nuances out until we have mastered the general image of the auditory pattern. Only after a clear impression has been made in the cortex does the rest of the art come into focus, and it is during this phase of our judgment that the piece becomes most vulnerable to our rejection. Whether or not we like the rhythm of a song or are impressed by the intricate configuration of a beat, it is the underlying emotion and perspective conveyed by the musician amidst the music that ultimately has the potential to establish itself in our more definite favor. If what you have written about prions and memory is true, then the increase in neurotransmitter levels resultant from such esteem for a piece of music would logically work to enhance our ability to apprehend it, and importantly to define more precisely and elaborately how we feel about it. This would also imply that during certain states of mind we might be more accepting to a particular piece of music which we might otherwise loathe when in a negative mindset. Or conversely, in times of despair the music we have always loved can seem to have been an illusion, that how we felt when we enjoyed it was only a fabrication of our excessive favor at the time. Hence in theory, we should be able to appreciate all music, but in reality the time constraints, world matters and the like which stress the mood and may cause neurochemical dysfunction work to curtail the open-minded exploration of alien tunes in favor of reverting to a more conservative exploitation of the sonic sequences which were engraved in times of lesser duress, the good times or better days. More precisely, my theory is that a higher level of neurotransmitters in the cerebrospinal fluid should serve the function of activating more of these prions and exposing the cortex to a more elaborate (and profound) interpretation of the senses, liberating the flow of thought and devoting additional mental capital to the neural integration of musical patterns along with whatever perceptual associations are derived from the empathy conferred by the musician. As anecdotal reinforcement to this theory, an experience that took this phenomenon to the highest degree of profundity possible (in my lifetime at least) was listening to Bob Marley after ingesting some quality psilocybic mushrooms. Whether or not psilocyn effects prions and perception in the same manner as serotonin and the other neurotransmitters, the experience left me with a revamped reverence for that profound emotive power that music can evoke in the soul.

Well now if the theory of music being a mechanism for mate decision is true then it seems playing a song, or maybe just actually singing it, for someone increases one's value as a potential mate. I don't know if one sex is supposed to be the singer/perceiver or both but if this is the case I wonder if a chemical that increases a person's appreciation for a song would increase the mating value of the song player/singer.

By Timothy Schuchard (not verified) on 10 Feb 2010 #permalink