Poster F114, Tuesday, March 27, 8:00-10:00 am, Exhibit Hall C
Learning to control unstable dynamics via movement sonification improves generalization
Dobromir Dotov1; 1McMaster University
As per motor learning theory, unpredictability of training improves generalization. Predictive processing suggests that interactive synchronization facilitates learning of complex patterns. Movement sonification in relation to designated movement trajectories has proven useful in refining error-driven motor learning, due to multimodal integration by associative cortical areas and novel feedback pathways. To test the roles of unpredictability and mutual synchronization in a sonification paradigm, we designed a task of controlling an unstable (chaotic) system by synchronizing with it. Hand movement was sonified in the left channel and the unstable system was sonified in the right. In an unstable but interactive condition of training (U-I) the stimulus was weakly driven by the participant’s movement, making it possible to achieve mutual synchronization if the participant learned to predict the short-term course of the chaotic system. In an unpredictable non-interactive condition (U-NI) the same system was delivered in decoupled mode. In a predictable non-interactive (P-NI) condition a sine wave pattern was used. Learning was evaluated in a design with pre-test, training, and immediate post-test. Tests comprised of non-interactive harmonic and unstable stimuli. Transfer entropy, cross-correlation, and average error showed that performance during training improved in P-NI (predictable non-interactive) and U-I (unstable interactive) but not in U-NI. We found transfer to the non-trained stimuli in the U-I group, some transfer but to fewer stimuli in P-NI, and no transfer in U-NI. Predictive processing can explain why unstable (unpredictable) stimuli hamper learning but a synergistic effect is produced if interaction is added to unstable stimuli.
Topic Area: PERCEPTION & ACTION: Motor control