If you've ever worked on developing speed on guitar you've probably heard references to "muscle memory". But what is this mysterious facility? Is it something real or just a part of musicians' folklore? Can we do anything to make our learning more efficient, or should we be suspicious of "automatic" playing as less creative?

Procedural Memory

Psychologists tend to distinguish between two general kinds of memory: discursive and procedural. The former is the kind of memory you use when you learn a fact you can later repeat; an example would be learning the circle of fifths, the number or sharps or flats in a key signature or the chords that make up the riff to "Smoke on the Water". We're not so interested in this here.

The other kind -- procedural memory -- is about our ability to "just do" things without analytically recalling facts about what to do. Think about riding a bicycle, catching a ball, typing on a computer keyboard and perhaps even playing a video game you've played many times before. All these are things you once couldn't do and had to learn, but you don't have to remember facts when you do them. It's as if your ability to do them is somehow "burned into" your brain.

I'm pretty sure this is what musicians mean when they talk about "muscle memory". It's as if the muscles act without the conscious part of the brain having to direct them explicitly. A lot of our day-to-day competence seems to be like this, and clearly we would like at least some aspects of guitar-playing to be like that too.

How Procedural Memories Are Made

There are various models of how we acquire procedural memory; this isn't a psychology class, and I'm not a psychologist, so I'll just pick the one that seems easiest to apply to the experience of playing. It was originally proposed by Paul Fitts in the 1950s, and although it may now be rather dated I think we can still recognise its general contours and apply that to our real-life practice.

Fitts's description goes in three phases. The first is the "cognitive" phase, in which we have to consciously think about everything. Think back to the first chord you learned on guitar, and how you had to think about how to place each finger just behind the fret, which string it should go on, which fret it goes behind, how to hold your wrist, where to put your thumb, pressing down with the right pressure, how to keep the guitar balanced on your knee, keeping your posture straight and so on. There were dozens of things to hold in your head all at once, and the task of remembering and executing them all seemed impossible.

That's because it is impossible. Nobody can play guitar that way. Unfortunately the cognitive phase is the first step of the process and can't be bypassed, because you have to know what to do not in order to do it but in order to practice it until you don't have to think about it any more. This happens during the "associative phase".

Here multiple admonitions from the teacher start to get combined into aspects of a single, larger pattern: "how you play a chord" rather than all those fragmented instructions we started with. During this phase a kind of integrated picture of the task forms by repeating it many times. Things that seemed important but weren't drop away and the essentials come together into a single action. At least, that's the idea.

The cognitive phase is a frustrating process of learning that involves discursive memory. The student will often say things like, "Yes, I know I ought to be keeping my thumb behind the neck, I just seem to forget". It's tempting to give up at this point since the skill seems prohibitively hard: how can anyone hold all that information in their head, especially when playing something that seems infinitely more complex than what the student's attempting?

In contrast the associative phase is much more rewarding: the student feels as if they are "starting to get it", and sees real improvement. This encourages more practice, and more practice leads ever closer to the final, "autonomous" phase in which the action can be carried out more or less unconsciously.

Any guitarist beyond the raw beginner stage will have different skills at different phases. Maybe, for you, playing an open "E" chord is autonomous, playing a C Major Scale straight up and down is in the associative phase but playing it in broken fourths is still cognitive. Knowing you achieved the final stage with the "E" chord should encourage you to keep working at those broken fourths.

Where Things Go Wrong

The three-step model described suggests we acquire new guitar skills by first explicitly learning the details of what to do (cognitive), then repeating them until they become a single action (associative, leading to autonomous). But things can go wrong.

The best-known example is summed up in the well-worn phrase, "Practice doesn't make perfect; perfect practice makes perfect". If we rush the cognitive phase we may well fail to properly learn all the components of the skill before we start "drilling" it in the associative phase. If we persist we will surely form a procedural memory based on what we learned in the cognitive phase. Yet this may be faulty.

For example, if we're practising scale sequences we must learn at the cognitive stage that hand synchronisation is a crucial part of the process. That way we will continue to attend to it during the associative phase, getting it "dialled in" as we go. If we rush the (admittedly painful) cognitive phase and neglect hand synchronisation we will develop a procedural memory that is sloppy and prone to mistakes. Just being conscious of this danger makes it less likely to happen.

If you doubt the importance of perfect practice, see the conclusions of this dramatic study. Consciously noticing and correcting errors during the early stage was the only real factor associated with top-level performance.

Practice Techniques

What does this model imply for how we should practice? Unsurprisingly, it suggests a lot of things that are already part of the received wisdom among musicians. Having a clear, systematic picture, however, makes it easier to put all of this together.

When first learning a new skill, recognise you're in the cognitive phase; this can be a frustrating stage and shouldn't be rushed through. Work methodically and be ultra-conscious of everything you do. Look for details and attend to them. Is your pick movement as efficient as it could be? Are you muting successfully? Is your tone good? Are your hands in synch? Are you tense? Are there specific areas of difficulty where you frequently fail? Do you literally forget which notes or chords to play, or get confused?

At first work without a metronome, ultra-slow, observing and trying to commit all this to discursive memory. When you feel ready, set a metronome at a very slow tempo and begin trying to "do the whole thing" in slow motion. Do not rush this part: get it right. It's the basis of the "perfect practice" that will make the next stage a success.

Soon enough all these things will click and you'll want to increase metronome speed. This is a good sign that you're entering the associative phase. Be conscious of that and, at first, be extra-critical to make sure what you're doing is as perfect as it can be. Then work to build speed by repetition; as a by-product you'll also be building the "automaticity" that represents real competence.

One sign of this is that you're able to play while your attention is divided, for example while watching TV or -- at a more advanced level -- carrying out a conversation. This is a terrible thing to try in the cognitive or early associative phase, because the danger is that lack of concentration will introduce undesirable elements into your practice. Once your practice is 100% confident, though, this can be a good way to get in some extra repetition that will make the skill even more automatic.

Is Automatic Playing Robotic?

We all know of players who have amazing chops but are boring to listen to; is this the result of automaticity? Are they on "auto-pilot" when they should be being creative? It may look that way but I think this is, in fact, a different problem.

First off, the "autonomous" phase relates to our physical and mental ability to perform the task. It doesn't relate in any way to our choice about which task to perform. Being able to ride a bike without thinking about it doesn't have anything to do with where we choose to ride it, whether we're considerate and safe road-users and so on. Those are just different considerations from the acquisition of a physical skill.

Secondly, automaticity actually frees us up to think about the more artistic aspects of playing. Think again of the bike analogy: if I can't ride without thinking about pushing the pedals and keeping my balance, I have no business being on a busy road because I'm not going to be able to apply all my faculties to riding safely. It's the same with playing. We shouldn't be playing on "auto-pilot" while reading the paper, we should be attending closely to things like timing, phrasing, dynamics, tone, note-choice and so on. If we're still thinking about how to physically execute a particular line those things are likely to get pushed aside.

I hope this has been useful; I know it mostly confirms things you've heard elsewhere, but putting it into a simple (but moderately scientific) three-phase model may help you remember it and organise your technical practice more effectively.