Perceiving musical boundaries

Welcome to Science Sessions, the podcast of the proceedings of the National Academy of Sciences, where we connect you with Academy members, researchers, and policymakers. Join us as we explore the stories behind the science. I'm Paul Gabrielson. Music has boundaries, signals in the music that separate movements in a symphony and verses and choruses in pop songs. Studying how our brains anticipate and process those boundaries can show how we process and parse complex information. In a recent PNAS article, Petri Toyvianin, an ebiburinat of the University of Uvascular in Finland, and Daniel Levitan of McGill University in Canada, described how people processed boundaries in three very different pieces of music based on a small study. The study included 18 musicians and 18 non-musicians. The results show how musical training affects musical perception and suggest possible similarities between musical perception and other cognitive tasks, like language processing. Petri, let's start with some context. What is a musical boundary? How do we hear them in the music we listen to every day? A musical boundary is a natural stopping point or shift in music. It's like punctuation in a sentence. Usually we hear boundaries through some kind of changes in the music, such as poses or shifts in melody. So the melody can jump from low pitch to high pitch or vice versa, for instance, changes in rhythmic patterns or harmonic structure or even dynamics. E.B. There is strong evidence that we instinctively perceive boundaries at points where we encounter unpredictable or surprising events. We believe that anticipating these events requires more mental effort. And so we create a boundary as a way to cope with the increased cognitive load that we can't anticipate what comes next. Daniel, when you look at a visual scene, we're taking an auditory scene. How is it that you differentiate objects from one another? How do you know where one object begins and another object ends? In the auditory world, we have this undifferentiated stream of information, all of which arise from the eardrums simply wiggling in and out. And so different auditory objects and events need to be extracted from that dense signal. Tell us about your experimental setup. What pieces of music did you play for participants and why? We used three different pieces with a duration of roughly seven, eight minutes each. So first we had some 20th century concert music composed by Igor Stravinsky and some progressive rock by the group Dream Theater and Argentinian tango composed by Astor Piazzola. And so we wanted to have this kind of wide range of different music. So to be sure that our findings were not specific to a certain musical genre. So then we presented our participants with these pieces of music while they were laying inside a functional magnetic resonance imaging scanner. And their task was just to listen freely to the music. So it was unlike many other similar studies where the participants are asked to maybe rate the emotional content of the music or some other aspect of music. What we wanted to do is to simulate a real music listening situation as well as possible. And then we used a separate listening task outside the scanner where listeners marked boundary transitions in real time. This is what allowed us to identify the transitions or boundary points in the music, which then we used with the brain signal to do our analysis. Why did you include musicians and non-musicians in your study? I think all of us are musical. But what we don't have all of us is systematic training in creating music. And because musicians are actively involved in that, they have a more highly developed statistical expectation map for what's likely to come next. What did you find in general? How does the brain perceive musical boundaries? So right before a boundary, the brain prepares for the change and anticipates the end of the phrase by engaging a small focus network of mainly posterior auditory areas. These are at the back of your head behind your ears. During and after the transition, the brain processes the new information, the new phrase by shifting towards middle and anterior auditory regions, so more towards the front of the brain. And at the same time, there is a massive deactivation of several areas in the prefrontal cortex, temporal and parietal regions. And this pattern that we observe is very similar to how information flows during sentence comprehension in language. We're talking about separate networks that are anticipating where a boundary might occur, other networks that recognize when our prediction was met or when it was violated in the moment. And then after the fact, consolidating networks that kind of review and short-term memory, what's just happened. And then a fourth network that's indexing probably the hippocampus and memories of what music has come before. And then in contrast, there's this transition network that's active during and after the transition. And that's more forward in the brain. And it includes an area that we identified as Rolandic Operkula. What were the differences in how musicians and non-musicians perceived boundaries? We found that musicians showed more intense engagement in brain regions related to auditory and motor functions, while non-musicians exhibited broad engagement of areas related to cognitive control and attention. And I think that's quite interesting. So it's just musicians use some specialized neural networks for musical processing, while non-musicians rely on more general strategies. And one possible explanation for that is that musicians, by listening to music, they kind of mentally simulate the sound production actions, because they know how the sound is produced with instruments. Whereas non-musicians may be lack this ability and then they result to different kind of neuroprocessing. If musicianship is a technical/athletic virtuosity, we've got the wrong definition of musicianship. So playing an instrument is not necessarily the best index of musicality. But it does factor into any study we do because musicians tend to have many years of systematic training. So you're talking about with that level of training and learning and memorization, fundamental changes to brain circuitry, because that's what learning is. What are the caveats or limitations of the study? Although we tried to simulate a real musical listening situation, of course that's not possible with an ephemeral scanner. When the scanner is in function, it makes quite loud noise. And we did our best to insulate the noise so that the music was audible. Another limitation was that the music selection was limited to Western genres. We could expect that processing music from other cultures, for instance, might give us a different kind of rare responses. Third limitation is that music is a very complex stimulus. So there are many elements changing dynamically and simultaneously. Like melody harmony, rhythm dynamics and so on. Complex stimulus gives rise to complex neural responses. How to filter out the responses that are particularly epoch- why music is challenging. We may wrongly assume that a specific brain area is related to auditory processing, simply because it's showing up in musicians and it's in a sound processing region of the brain without considering that it might also play a role in other cognitive processes. Another thing we didn't consider in our design is other people were sub vocalizing or maybe they were internally singing along why isn't it to the music in the scanner. Because this could of course have influenced the results, especially in the motor-related areas of the brain like the Rolandico perculum, which includes representations of the larynx. Then another drawback is the limited number of participants, which on the other hand is common in neuroimagine studies because of the high cost. What do these results add to the field of music neuroscience? This whole idea of event segmentation music is understudied and event segmentation is a fundamental process in preparation for memory and for parsing of any dent stream of information. In order to pay attention to something you have to know what it is you're paying attention to and what it starts and what it stops. Given the overlap between language and music we could explore how integrating musical elements into language-phrase transitions perhaps may improve comprehension for people with speech difficulties. For instance we can make sentences more music-like by embedding them in melody and rhythm so that could be a promising aspect to study with a tangible impact in speech therapies. Thanks for tuning into science sessions. You can subscribe to science sessions on iTunes, Spotify or wherever you get your podcasts. If you liked this episode please consider leaving a review and helping us spread the word. (upbeat music)