6. Some handy learning concepts

Here are some music-relevant concepts from the study of learning.

• Declarative and procedural knowledge. Declarative is what you can say (recite your Social Security number), procedural what you can do (ride a bike). Much music pedagogy intertwines the two: say it and play it. Talking a song through, as advised by some of the reviewed authors, is a way to bind declarative and procedural knowledge, which since performance is procedural is an aid to retrieval.
• Episodic and semantic memory. Episodes are events or activities within a block of time; semantics are what things mean. Example from the late 40’s on chess grandmasters: they remembered tons of games (episodes); when they made errors they’d be ‘equivalent’ positions, that is, the same in meaning (semantics) for the game, but not literally the same.
• Savings. How deeply and thoroughly encoded a memory is. The melody to ‘Mary had a little lamb’ is likely high savings; an intermediate theme from the third-to-last movie you saw is probably low.
• Retrieval. Getting to the memory. The common heuristic of ‘what was I doing when I got home?’ is a retrieval strategy for remembering where you put the car keys. Memory palace techniques are about retrieval. Proust and his madeleines. Savings and retrieval are not the same. There are likely a lot of melodies you can’t call up by name but instantly recognize when heard. Multiple retrieval paths (coming at it from different directions) aids savings. (Or at least means retrieval of saved information is better.)
• Fluency. We can’t directly assess the strength of our memories, so we use fluency, which is how quickly and easily we can recall, as a proxy for savings. Fluency is subject to recency effects, cueing, all sorts of things. It’s not a reliable guide to savings. It’s crucial for playing music, though.
• Working memory. Called short-term memory when mentioned by our authors; working memory here. What you can hold mentally at hand for a task. Crucial both for playing and for listening to music. This is much more complex than our authors seem to believe. There are ‘working memories’ for (at least) visual, linguistic, numerical, object, aural/auditory, spatial, even olfactory content. And of course musical working memory, as distinct from auditory or verbal, e.g., this study. (Domain specificity is mildly controversial, likely because focus of attention is an executive task which may be in common across domains. Even this is controversial.) These different types rely on different networks in the brain and mind, can run concurrently, can instantly draw on savings, and are highly sensitive to chunking and intention (which with expertise allows selection of relevant factors and filtering of irrelevant factors, something experts do far more efficiently and effectively than novices). As well, not only the number of ‘things’ but the number of relationships, the functions, and the meanings of things are relevant to working memory. As well, working memory runs concurrently with filtering systems that monitor the environment for relevant stimuli (someone saying your name, for instance, or the smell of something burning will interrupt task-focused working memory). Working memory is always influenced by general cognitive load, even when you’re not conscious of it.
• Metacognition. Thinking about your own thinking (and feeling). Includes things like your knowledge about your knowledge (knowing whether you know your home street address, whether you know how to speak Slovenian), certainty (confidence in the degree to which you remember your credit card number), and so on. Metacognition is critical for self-assessment and developing effective personal learning strategies.
• Context. Everything that isn’t the thing in question but surrounds it. Music theory is deliberate context. Patterned places to play or practice are context. We use context to both aid memory construction and reinforce retrieval. You can think of context as the current complex network in which learning is to wriggle its way into.
• Chunking. Aggregating things you know well into coherent packages in memory, e.g., storing or recalling a familiar pattern rather than trying to remember all of the parts and their pairwise relationships. This tremendously expands what you can have in working memory at any given moment. Learning music involves (or should involve) chunking into semantic units that become larger/richer with increasing musical knowledge—a conductor’s semantic chunks are bigger and more complex than mine.
• Spacing. Studying some, then doing something else. Modest shots of content and some time for consolidation. Usually contrasted with massing: doing a lot of the thing in a temporally continuous block. Spacing is much more powerful for learning than most people know, perhaps because it often feels like the opposite. Chopin advocated spacing in practice. Even tiny breaks are critical for learning: ‘most of the improvement while learning a motor task comes not while actually practicing, but instead during the breaks between practice sessions.’ For declarative information, long breaks can be helpful.
• Incompleteness. We remember things that are not quite done more reliably than finished things.
• Forgetting, and remembering wrongly. Trying to bring it up when it’s vague or blurry, bringing something up even if it’s not right, seem to improve both savings and retrieval. Even unsuccessful retrieval actually strengthens (correct) memory. Counterintuitive.
• Judgments of learning. What we think we know. We are way worse at this than we believe ourselves to be. JOLs are a subtype of metacognition.
• Testing. Powerful. Self-tests, tests by others. Provides snapshots and feedback that are essential to learning. Important: testing immediately after learning is great for overconfidence but not effective for learning. There needs to be a significant gap.
• Interleaving. Ongoing compare/contrast, perhaps, but whatever the mechanism, mixing content improves discrimination learning. To learn to recognize Satie, mixing Satie with Ravel and Debussy and Faure is more effective than massing some Satie. However, there are limits on how far this goes as an effective strategy.
• Teaching others. There’s a cognitive bias called ‘the illusion of explanatory depth:’ we think we know more than we do. Teaching others is a sometimes-painful way to discover the limits and inconsistencies of one’s own knowledge, and as well improve integration of the various different factors of musical and performance knowledge.
• ‘Desirable difficulty’ and, with a strong family resemblance, ‘region of proximal learning:’ the easiest thing you can’t yet do. Whatever is just past the edge of current competence. This is where learning occurs.
• Transfer. Learning in one area improves (or doesn’t improve) performance in another.
• Consolidation. Letting it sink in, away from the task. Improvement while not doing the thing one wishes to improve upon. One of the first discoveries of experimental psychology, and long discounted because common sense said it couldn’t be right. It is, though.

Much less researched but relevant to to music, learning by exaggeration / intentional error. This (perhaps) turns up the contrast, so to speak, making correct performance more easily distinguishable. This is (surprisingly) not well established in the learning literature but we do it all the time with music: play far from tempo, experiment with exaggerated dynamics, switch rhythms or tone colors, and so on.