Research on effective frequency and duration of practice for improvement?

Question

I've been playing piano for a long time now, did my undergrad in music and just finished my Teacher's Ed. program.

As someone who's suffered an overuse injury (permanent nerve damage) from practicing piano (inefficient technique contributed to this) I began thinking about the limited literature that I've been able to dig up so far about pedagogical 'no-brainers' that I wonder have anything to substantiate said 'no-brainers' in a controlled environment.

The general consensus to improve at piano and reach a moderate level of proficiency seems to be 6-7x a week, at least an hour a day.

Are there any academic research papers done on what the effective frequency and duration of practice are for improvement?

Physiologically I would assume recovery would be no issue (unlike trying to barbell squat every day), but then there are people who suffer overuse injuries, tendonitis, nerve damage, etc.

What is the duration one would need to practice a specific movement pattern before it becomes solidified and reliably reproduced on demand? Is there any way to measure this?

For the average piano student who also wishes to learn other musicianship skills such as ear-training, sight-reading, improvisation, and theory and also still wants to have a life outside of music - how effective would practicing with less frequency be? What if they were to have 3-4 practice sessions a week and leaving a day of rest in between to work on other musicianship skills?

Is it really necessary for optimal growth to practice at such high a frequency?

I'm sure some of the performance purists would scoff at this suggestion and claim that such circumstances will never lead to success or attainment of high proficiency in the instrument, but I am really curious to know if there is anything to substantiate these common 'no-brainers' we have in music pedagogy.

Let's discuss. And if anyone has links to research, if you could please share that would be fantastic!

Answer 1

I spent the last 20 years (off and on) designing training software for musicians, including interacting with thousands of users on how to get the best results. The following points are drawn from that experience. (*)

• There are huge differences from one person to another. 10x learning speed differences are common, and even more so in cognitive tasks (e.g. learning scales logically, knowing where to play them, etc.) compared to physical tasks (e.g. learning to play scales fast)

• Progress in learning is never linear. You progress a bit, and then hit a block. Sometimes the block is easy to overcome, other times it holds you back for a long time, until you eventually find a way to crack it. Fast learners find their way around blocks faster than others. People with less determination hit a block and give up.

• Some challenges are somewhat common to everyone (e.g. playing with good time), but many are subjective. I believe that if you wanted to study it "scientifically" you would need to test a very large number of people (in the thousands for sure) to get meaningful results, and even then, those results would be all over the place, and those statistics would not be very useful in predicting what a new student will go through.

• An experienced teacher will often be able to understand the musical potential of a new student in just a few minutes. Perhaps one day AI will be able to do that, too, after putting the person through various tests, but we're nowhere near anything like that at present.

• The relationship between time spent practicing and results achieved is very weak. Everything depends on the training strategy. It's entirely possible to practice for weeks and months with very little improvement, because of a bad strategy. And conversely, it's possible to improve dramatically in a short time with the right strategy. Quick example: if you start to record yourself while practicing, and then always listen to yourself afterwards, within a few days you will start correcting errors and improving technically in ways that eluded you for months or even years.

• Regarding ideal frequency of training, my experience tells that it depends on the subject. For example, for purely technical skills -- play fast, precise, with finesse -- frequent practice is necessary. The old violinist's adage is true: "If I don't practice for a day, I notice. If I don't practice for two days, the other musicians notice. If I don't practice for three days, the public notices."

• On the other hand, for more cognitive learning and performing skills -- e.g. ear training, improvisation, composition -- I find it better to alternate periods of practice with periods of rest. In these areas, one week of rest will sometimes help you more than one week of practice. (Assuming that you were otherwise practicing properly). The reason for that, in my opinion, is that in certain areas, the mind needs time to absorb and integrate what you learn. If you keep pushing new stuff in, you actually prevent the mind from assimilating the material properly -- I think this is called "interference", i.e. when learning something new has a negative effect on something else that you learned only recently, but you haven't assimilated yet.

• Because of all of the above, striking the right mix of frequency, intensity, and training strategies, is still much more an art than a science. A person who is more aware of one's own states and condition will instinctively know better what to do and when, and will get much better results than a person who makes a similar amount of effort, but is not nearly as self-aware as the first. Similar considerations apply to teaching: a good teacher will be aware of the student's individual states and condition on the learning path, and will adjust their input accordingly, while a bad teacher will follow some formula without understanding how the students are in fact responding to every input.

• In summary, I believe that while in theory it may be possible to devise learning formulas that optimize our musical learning, in practice we are still far from that. Things are a bit easier when you focus on a very specific skill, such as in the software-based training courses that I designed over the years, such as learning all the notes on the guitar fretboard, but even with something that seems to simple and straight-forward as that, you'd be surprised at the variety and weirdness of people's response and progress curves.

TLDR: the situation is very complex, and both the learning and the teaching themselves are still more of an art than a science.

(*) You can see some examples of my old work at http://www.micrologus.com