Multimodal Brain Imaging of Motor Sequence Learning

Motor skill learning is characterized by rapid initial performance gains, then a period of slower, steady improvement, and finally asymptotic performance that entails task automaticity. The current study investigated functional brain reorganization during multiple-session learning of a motor sequence task. Functional near-infrared spectroscopy (fNIRS) was recorded to investigate hemodynamic changes in the execution phase of the task. In addition, electroencephalography (EEG) was measured to examine changes in event-related desynchronization (ERD) of the mu rhythm at the C3 and C4 electrodes during both planning and execution phases. Participants (N=14) learned a finger tapping task in seven sessions and brain responses were recorded in Sessions 1, 2, 4, and 7. In each trial, a cue indicated whether participants should prepare for the trained or a random sequence, and then participants completed visually-paced sequence execution. In addition, a dual-task paradigm with a secondary tone-counting task in addition to sequence execution was used to examine task automaticity. Accuracy and RT measures revealed typical motor sequence learning of the trained sequence, including an increase in automaticity from Session 4 to 7 in comparison with the random sequence. fNIRS showed bilateral M1 activation in early learning stages. A linear trend analysis of the fNIRS data revealed decreasing activation in right M1 across sessions. There was also an inverse-quadratic trend (inverted U-shaped) in SMA, right DLPFC, and left M1. Analyses of ERD revealed a quadratic trend in C4 during preparation and in C3 during execution. Hence, brain activation in primary and secondary motor regions initially increases with learning, but this is followed by widespread reduction with automaticity for both motor preparation and execution.