A model of time-varying music engagement

Hello and welcome. It's September 9th, 2024. This is Active Inference Art Stream number 3.1 with Diana Omegi talking about curious sounds, prediction, and the human drive to engage with music. There will be a presentation and then we'll read some comments and questions from the live chat. So thank you, Diana, for joining and looking forward to your presentation. Thank you, Daniel, for the invitation and to everyone who's joined today. I'm a senior lecturer and researcher at Goldsmiths University of London. And today I'll be talking, giving a talk that's entitled Curious Sounds, Prediction, and the Human Drive to Engage with Music. And the reason I chose this title was because it recently came across this statistic. The International Federation of the Phonographic Industry do a report every year where they say how much the average person listens to music. And it turns out it's about 20.7 hours a week. So if we assume that people get about eight hours of sleep a night, then we're looking at about a fifth of our waking hours that the average person is listening to music. And it struck me that there's a really interesting question here, which is what is behind this drive to engage with music at this scale? And so what I'd like to do in today's talk, partly because I wasn't quite sure who would be attending, but I assumed it would be people with lots of different backgrounds. And yeah, what I decided to do was just present a little bit about what neuroscientists have been finding out about the neural substrates of music perception and pleasure. We might assume that has something to do with why we choose to listen to music. I then want to talk for a few minutes about what the predictive processing framework could explain to us or tell us about why it is that our attention towards stimuli might change over time. And finally, use what we've learned from neuroscience, from the predictive processing coding framework to present a model, basically, of the dynamics of extended engagement with music. So that's a piece of work that was published earlier this year and which I'm keen to share with interested parties. Let's think about music perception and the things that we do without even realizing how powerful these things are. So we can spot wrong notes in music that we're hearing for the first time. By wrong, I mean, yeah, notes that were clearly not intended by the performer. We can say whether two melodies are the same or different, again, hold pieces of sequences of pitches and rhythms in our memory and compare them. We can recognize familiar pieces, even if they've been played differently, different lyrics, different instruments and so on and so forth. And we can correctly predict how a piece of music we've never heard before might unfold. We might be able to sing along to it, even if we're hearing it for the first time. And naturally, people have come up with ideas as to what's going on, why this is all possible, how this is all possible. For example, in this model published in 2003, the idea was this acoustic input comes in undergoes some analysis, inevitably gets separated out into its sort of pitch components and temporal components that undergo further processing these interface with some store of musical phrases and patterns that we've experienced in the past and so on and so forth. But of course, this is a sort of box and arrow model of what makes sense as a series of processes. When we look at the brain, we're looking at a network of connections between subcortical structures, the auditory cortex, so-called belt and power belt areas around the auditory cortex. And we're looking beyond that, we're looking at sort of more frontal parts of the brain that we know are involved in some of the more complex functions that we're able to do as humans. And what's interesting is that when it comes to something like pitch perception, which is obviously a very important part of processing music in lots and lots of different cultures, we see…