The following article was submitted by my student Tom Page. It’s a great article and offers some very interesting concepts regarding practice and learning. Great job Tom, thanks!
The chunking theory of learning is based on the concepts that:
• Performance consists of known patterns (chunks) inherent in the task you are performing
• Practice consists of acquiring the necessary patterns (chunks) that you build out of tasks already mastered.
High levels of performance are made possible by the magic of chunking. The time required to process a larger chunk is shorter than the sum of the times to process all the component chunks that comprise it. Hence, acquiring skill consists of building up increasingly larger-scale chunks, such that tasks of increasing complexity can be performed much more rapidly and fluidly than all of the underlying component skills required would imply.
Psychologists have known the power law of practice since the 1920s. It states that the time it takes to perform a task decreases as a power-law function of the number of times the task has been performed [Snoddy, G.S., “Learning and Stability” Journal of Applied Psychology, 1926]. The basic model of practice is to acquire the skill to perform a task correctly (albeit slowly), and then to repeat that task slowly but perfectly (practice) so that the time it takes to perform it correctly improves. The power-law of practice has been shown to hold across a very wide range of human skill acquisition endeavors. A sample graph of a power-law distribution is shown below (on a linear scale).
In the above graph consider the vertical axis to represent the time it takes to perform a given task correctly (say tying your shoe). The horizontal axis represents the number of times a task has been practiced. As you can see from the curve, performance improves rapidly with practice (initially), but the marginal return on practice eventually levels off (decreasing returns). This matches our experience of watching a 5-year old tie her shoes; it takes a long time. But by the time she is 12, she can tie it about as fast as she ever will, achieving only minor improvements from then on.
The Learning Curve on Guitar
This curve also matches what most of us have experienced with respect to acquiring skill on the guitar. Improvement at first is rapid and gratifying. However, as we progress towards intermediate levels of ability, each new increment of improvement takes increasingly more input of practice. This decreasing return on practice investment can be discouraging, even causing some people to jump to other areas of skill acquisition (such as taking up golf), where they are at an earlier (and therefore more gratifying) spot on the improvement curve. However, as their skills eventually plateau there as well, they are tempted to jump to yet another arena (“maybe I’d have more fun learning to paint”). Thus, the average person either reaches his level of frustration or his level of sufficient satisfaction, and is therefore not sufficiently motivated to continue putting in the increasing additional practice input required to move further out the performance curve.
The vertical line shown past midpoint on the X-axis represents the edge of the proverbial goal of “world-class” or “virtuoso” performance. It has been proposed that to achieve world-class ability in any significant skill endeavor requires 10,000 hours of intense, focused practice. This idea seems to apply well to skills varying widely from music, to athletic performance, to chess. The message behind the 10,000 hours theory is that experts do not possess some innate talent that the rest of us lack, but rather they are extreme outliers on this power-law of practice curve. They managed to, firstly, have the quality of instruction to teach them how to perform a task correctly, and secondly, have the opportunity and drive to put in enough practice that they can perform that task correctly in a fast, effortless, and reliable way. Malcolm Gladwell’s popular book Outliers: The Story of Success is a very readable presentation of the 10,000-hours hypothesis.
How Power-Law Practice and Chunking Work Together
One model that explains the power-law of practice is the chunking theory of expertise. Various researchers have shown that experts do not have any inherently superior cognitive abilities over the rest of us. For example, chess masters who can reconstruct entire games from memory, or bridge masters with total recall of hands they played days before, do not possess measurably better memories than normal. However, studies in which the experts are requested to think aloud while completing representative tasks in their domains have revealed that experts encode information in larger “chunk sizes” than do less trained people. Experts do not simply know more; rather, they encode what they know in a way that makes domain- relevant information rapidly and reliably retrievable. A famous study in this area showed that expert chess players did no better than the average person at remembering the configuration of chess pieces when they were placed incorrectly on the board. But when shown a position from an actual game, they could reliably reconstruct the configuration from memory at far superior levels of success compared to non-chess experts.
For another example of increasing chunk size, consider the way an expert football quarterback scans a defense in the seconds before the snap. An inexperienced person sees 11 opponents in various places on the field, and in various stances, and has to reason through what each might do.But a professional quarterback has seen all of the standard defensive configurations thousands of times. He is able to cut through all of the irrelevant detail and diagnose, for example, that the defense is in cover-two and thus he should audible to a quick underneath pass to a tight-end. Through practice repetitions, the expert quarterback has encoded, in appropriately large chunks, the information he needs to pattern match the defense immediately, and then he retrieves from memory the correct counter action.
The Answer to a Paradox
The chunking theory of skill acquisition consists of recognizing the patterns that make up a task to be accomplished, and then practicing to build up the ability to perform those patterns, based on patterns already mastered. This is the answer to a basic paradox, “How can you acquire the ability to do something new by repetition of something which is not that thing you want to be able to do?” For example, if I can’t play a difficult passage, simply trying over and over is unlikely to make me able to play it. Rather, we have to take advantage simultaneously of the laws of increasing chunk size and the power-law of practice. First ,we have to build the new skill that we cannot yet do, out of more primitive chunks that we can do. (A corollary of this is that the definition of something that is too hard for us at our present level of development is that we have not yet mastered its component skills.) Once we have recognized what are the component chunks that must be sequenced to carry out the new task, then we can begin to sequence them very slowly. Through applying repetition, we can link those component chunks into a higher-level chunk. For example, a series of independent finger motions such as P A M I (four chunks) can be practiced until they become a single gesture (one higher granularity chunk) which can then be called upon to produce a tremolo pattern (an even higher-level chunk).
The Myelin Connection
Recent advances in neuro-physiology appear to explain chunking and the power-law of practice. When we perform a task, the neurological circuit in our brain that encodes that task gets reinforced with an insulating layer of myelin. The production of the myelin is called myelination. The effect of myelination on a neural circuit is to increase the speed of propagation of signals along the fiber and to provide a path for regeneration of the fiber if it is damaged. The more a neurological circuit is fired, the more its myelin is reinforced, and hence the faster it can fire, and the longer it will last. Further, circuits can be made of other, more primitive, circuits. A good book explaining this phenomenon is The Talent Code: Greatness Isn’t Born. It’s Grown. Here’s How.” by Daniel Coyle. For a guitar example, the circuit to play a free stroke with the P finger can call on the circuit that flexes the knuckle of P. Then the circuit which plays tremolo can call on the circuit that plays P, followed by the circuit that plays A, followed by the circuits for M and I. By practicing tremolo slowly, the circuit for correctly sequencing P A M and I is further myelinated, thus improving its speed and reliability. Hence, a larger tremolo chunk is built out of already-mastered chunks, and taken along the power-law of practice curve by myelination through repetition.
Most of us have achieved expert-level performance at talking. When we have a thought to convey, we don’t have to think about what position to hold our tongue in to produce the first sound, or how much to tighten our larynx. We hardly even think about what words to say. Rather, we think in concepts and the sentences spill out, full of their meaning. This is the level of fluency of guitar playing we should aspire to. The beginning guitarist plays notes. The advancing guitarist links those notes together into lines and phrases. We move beyond where to put our fingers, and how hard to pluck, instead conceiving of whole musical phrases which our body just knows how to produce as easily as uttering a spoken sentence. Increasing the chunk size from motions of fingers, to notes, to chords, scales and arpeggios and finally to phrases and sections of pieces is what happens over the course of the 10,000 hours of focused practice. Eventually, the real skill that is practiced is the translation from a musical image that we conceive of in our brains to the production of a good approximation of that musical image, as fluidly as we speak in our native tongues.
1. Power law functions produce linear plots on log-log scale graphs.