Introduction: In this study, we examined the cognitive processes and physiological responses involved in learning a flute piece by the composer Charles Koechlin among musicians of different expertise levels. Participants performed the piece four times consecutively, with a 2-min practice interval between the first and the second trial. Methods: Using data obtained from an eye tracker, respiratory sensors, and an audio recorder we assessed short-term improvement and the effect of musical structure and practice on key variables identified through a multivariate approach: eye-hand span (EHS), time index of EHS, thoracic and abdominal amplitude (breathing patterns) and pupil dilation. Results: The analysis revealed two main dimensions: one associated with EHS; and the other with embodied responses to music, closely linked to breathing patterns and pupil dilation. We found an effect of musical structure on all the variables studied, while the EHS improved with practice. Expert musicians demonstrated enhanced EHS and adapted their breathing patterns more effectively to the music's structure. Discussion: These insights support the hypothesis of a coupling between anticipation and breathing, emphasizing the role of perceptual and embodied components in music reading and learning.
This study examined how regressive fixations relate to temporal anticipation and unfolding temporal dynamics during music sight-reading. Eye-movement, performance, and respiratory data were recorded from 28 flutists performing three scores differing in musical and metrical structure; performance recordings were MIDI-aligned with eye-movement and respiratory measures. Regressive fixations were defined as fixations following regressions toward previously viewed score notes. A participant-grouped random forest model explained 60% of the out-of-fold variance in regressive-fixation count, highlighting fixation count, mean fixation duration, and time-based eye–hand span as leading predictors. Mixed-effects analyses showed that time-based eye–hand span was the most consistent predictor, while performance duration and mean fixation duration provided additional information. Rhythmic accuracy moderated the relationship between temporal anticipation and regressive-fixation count, with a stronger positive association in more accurate performances (β = 1.062, 95% CI [0.282, 1.842], p = 0.008). This interaction remained significant when regressive-fixation count was modeled relative to performance duration and after controlling for mean fixation duration. An additional analysis found no evidence that shorter fixation durations attenuated the association between more regressions and longer relative performance duration. These findings suggest that regressive fixations are associated with temporal anticipation during music sight-reading and may inform cross-domain research on the role of regressive eye movements in reading comprehension.