Entropy-Learning-Atonal

This project investigates the effect of formal learning about serial music on expectation and uncertainty in this style. BMus students will rate surprise (expectation/prediction), uncertainty (precision) and closure for 8 serial music phrases twice: before studying serial music as part of their degree programme and after studying serial music.

OSF project

You can find this project's pre-registration at https://osf.io/ab9dg.

Rationale

Predictive coding (PC) is a broad framework of brain function (Friston & Kiebel, 2011). In this framework, the brain processes prediction error caused by a mismatch between its predictive models and incoming real-world information. In the PC framework, two types of predictions are generated: the contents of the event (what, when) and the likelihood of this event happening. The former is referred to as the 'prediction' while the latter is called 'precision'. It is possible to be very certain about something and be completely wrong as much as it is possible to guess and be correct. In the past few years, PC has been applied to music perception (Furl et al., 2011; Kumar et al., 2011; Vuust et al., 2009), for example to the construct of groove (Vuust & Witek, 2014). Groove, or the desire to move to a beat, is particularly strong for syncopated rhythms, where strong musical beats can be replaced by silence. This silence on a strong beat violates our musical expectations; done repeatedly, the precision of these predictions also decreases. Koelsch, Vuust and Friston (2019) suggest that body movement is initiated in order to strengthen the sense of beat that is obscured by these regular violations of our expectations. The notion of predictability in music is older than PC and also applies to pitch, harmony, phrase boundary perception and melody extraction for example. The highly structured nature of music allows us to have constructed robust predictions about what will happen next, particularly in the Western classical idiom. A predictive model of the hierarchical structure of tonal music (i.e. dominant tones and chords resolve to tonic tones and chords; Krumhansl & Kessler, 1982) is developed as early as a few years of age (Cohen, 2000).

In this study, we will apply the PC framework to the perception of serial music, focusing on pitch content. As serial music lacks the hierarchical pitch structure of tonal music, it offers an interesting application of the PC framework. We will explore both prediction and precision when listening to monophonic phrases of serial music, with an emphasis on the idea of 'expected uncertainty'. Does knowing that serial music lacks hierarchical structure lead to the listener expecting surprise? Is there an empirical difference between prediction and precision when listening to serial music? Might explicitly learning about serial music in a university course change predictions and/or precision when listening to serial music?

Collaboration

Given the goal participant number is 34 and the pool of participants for the study is smaller on a yearly basis (20-25 in the posttonal music course), collaborators are most welcome. [Email me] (sarah.a.sauve@gmail.com) if you might be interested in collaborating. I am looking for help collecting data in particular. Collaborators will be consulted during manuscript preparation. Authorship will be offered to collaborators who participate in at least two of the following: data collection, data analysis, manuscript preparation.

STATUS The first round of data collection is currently under way.