Hearing movement, seeing sound: multimodal predictive coding in pianist-dancer interaction.

Live piano accompaniment for dance poses a "zero-latency paradox": performers achieve near-simultaneous audiovisual alignment despite sensory and integration delays that should make purely reactive control too slow. This review argues that pianist-dancer coordination can be usefully framed as bidirectional inference under cross-modal predictive coding, supported by converging behavioral, kinematic, and neurophysiological evidence. Motion-capture and time-series studies suggest that pianists can use dancers' preparatory kinematics, such as trajectory shifts and acceleration changes, to shorten the prediction window for timing and dynamics, while neuroimaging and EEG findings indicate action-perception coupling consistent with internal simulation of action-sound mappings ("seeing sound"). Sensorimotor synchronization paradigms show that micro-timing perturbations in auditory beats elicit rapid, asymmetric phase correction in stepping and tapping, consistent with predictive control in dancers ("hearing movement"), and autonomic measures further suggest that musical tension can modulate arousal before overt movement changes. Integrating coupled-oscillator modeling and EEG hyperscanning, we highlight quantifiable bidirectional adaptation and converging evidence that coordination is dynamically co-regulated rather than purely reactive or unidirectional. Taken together, the reviewed literature supports a neurally informed account of closed-loop dyadic coordination while also underscoring the need for more direct evidence from pianist-dancer interaction itself.