Piano Knowledge Representation and Generalization

It might seem obvious that multiple learning systems interact while learning the piano, however, only in the past decade or so has research come to grapple with this idea. One way we can divide up the types of learning is explicit and implicit learning. Explicit learning refers to conscious learning. Implicit learning refers to learning that is below our consciousness. Abstractly in motor learning, explicit learning narrows down the solution space of a motor sequence or program for the correct action. Or in other words, it figures out what the goal of the action should be and generally what that movement will look like. The implicit learning system will take this information and fine tune the parameters of the muscles for that movement.

I have been getting more into skiing, so what this looks like for me is that I now know when I am carving I need to initiate turns by rolling on the balls of my feet, starting from one toe and ending at the opposite toe while bending my legs in the direction I am rolling my feet. Even though I know this explicitly, I need to fine tune the movements through implicit learning in order to become a more proficient carver. These two systems have different learning speeds. The explicit learning portion is marked by a fast increase in learning speed (initial learning), but once that part has been learned, the implicit learning (later learning) portion is much slower and takes a long time to fine tune. This is why we see motor learning following a curve that generally will look like the figure below, where a quicker reaction time (lower on the graph) indicates superior learning.

It is important that we know which learning system is responsible for the current gap in knowledge. For example, I would still benefit from explicit feedback about my form while I am carving, because maybe I am not doing a good job of keeping my upper body independent of my lower body while turning. Knowing this explicitly can help me recalculate the goal of my movements and then fine tune the implicit learning. This will cause my improvements to be much greater and faster than if I had to figure this out on my own or if I kept fine tuning my implicit learning. This same idea can be applied to piano learning. For example, if one does not know that a finger tuck is necessary to have a consistent speed for playing scales, getting that feedback would greatly increase their ability to play a scale at consistent speed and allow them to optimize for the correct goal.

However, too much feedback can inhibit learning. Feedback taps into the explicit learning system and there is evidence that explicit and implicit systems may share resources or inhibit each other. Thus, if the explicit system keeps getting pinged, this will inhibit learning in the implicit system. This is not desirable if one has already learned the goal with the explicit system and now all of the learning needs to be done with the implicit system.

When learning is more difficult with a high cognitive load, the selected action is learned slowly and feedback helps the learner figure out what the optimal action to take is. Thus, frequent feedback can enhance the system that is driving learning and performance gains instead of taking away from it. Knowing this will dictate how feedback is given. If it is the case that a learner is slowly accelerating, like learning the C-major scale, we can infer that the task is being driven by implicit learning and that maybe it is a low load task. This means we do not need to give that much feedback. In fact, we could even increase the difficulty of the task, like by increasing the BPM that the scale is played at to enhance learning. In this way, we can appropriately scaffold learning rather than giving learners crutches.

Learning any skill is a complex process that requires us to break the problem down and understand what systems in our brain are driving the learning at that point. We can wield this information to enhance learning by changing feedback or task difficulty. Here at MIDIScale, we are excited to bring a new experience to piano learning that will help learners reach their learning goals faster. We will be launching the Alpha version at the end of this month!