Story by: Prof. Friedman Jason (Tel Aviv University) & Prof. Ritter Helge (Bielefeld University)

Learning to play the piano is a process that typically takes many years and requires a huge amount of practice. Our goal in this project was to devise methods to speed up the process of motor learning in this complex task. We aimed to learn general techniques that will also be useful in other places where we perform motor learning, including in children as they develop, in adults performing other new tasks, and in motor rehabilitation, as we re-learn how to perform lost skills.

Our approach was based on the notion of scaffolding – simplifying the learning process while the learner is in the learning stage, and then gradually reducing the assistance (i.e., the scaffolding). Rather than having a piano teacher or other expert define the scaffolding process, we instead trained automatic systems to select appropriate scaffolds. We used a variety of artificial intelligence and machine learning techniques in order to achieve these goals.

Along the way, we developed technologies necessary to achieve these goals, including techniques for automatically evaluating the level of playing, generating piano pieces at different levels of difficulty, and comparing and evaluating different types of practice.

This project was performed collaboratively between researchers in Germany and Israel. Both sides benefited from the complementary backgrounds of the researchers and the different perspectives brought by the two sides. Completing this research program brought up many new interesting questions, some of which we plan to tackle in our continuing collaboration. Recent advances in artificial intelligence have made the topic of optimizing the process of motor learning even more important and relevant.