x
Learning in layers of muscle memory — LessWrong