Since there is no curative therapy for Parkinson's disease, researches are trying to identify the best methods and the most effective tools to help these patients. This project addresses this question by investigating the efficacy of a new rehabilitative approach. Starting with a systematic review, three innovative approaches were identified as effective treatments for Parkinson's symptoms: Action Observation Learning (based on the theory of mirror neuron system), the non-invasive anodal neurostimulation on the primary motor area and the use of music. However, while recognizing the effectiveness of these three promising approaches, they have never been combined. The main objective of the present project is to assess the role of neurostimulation and multimodality (namely the presentation of a visual model together with a synchronized musical track) in motor learning, by considering both healthy adults and Parkinsonian patients. In order to achieve this goal, three experimental and longitudinal studies were conducted, using different tools such as tDCS, biofeedback and KinectTM. This work provides new suggestions for innovative and effective treatments in the motor learning field.