Have you ever been driving a car and suddenly you’re not sure what to do? You’re traveling along in usual automatic fashion and there comes a moment when you need to engage a new subroutine — say, you need to give way at an intersection, or you stall at the traffic lights, or you stop the car — and suddenly, you don’t know what to do. There’s a flash of panic, even while you’re thinking, “This is stupid, I’ve done this a thousand times”, and then, maybe it’s all right, maybe you have to take a moment to get your head in the right space, and ... okay, you’re off again, control safely in the hands of the automatic pilot.
But you’re unsettled. There are lots of ways our memory fails us. Some of these are very common, so common we just accept them — noone (well, few of us) expect our memories to be 100% perfect all the time. But procedural memory — the memory that allows us to drive a car, ride a bike, type, play the piano, etc — is different from other types of memory. We don’t say “it’s like riding a bicycle” without reason. Once we’ve truly mastered a skill, we expect to have that, for ever. And, for the most part, we do.
The thing about procedural memory — the big difference between it and so-called declarative memory — is that it is not in conscious memory. That’s its huge advantage; we could never perform skills fast enough if they were under conscious control. As we acquire a skill, the declarative information we learn (‘use your little finger on the “a”; the “s” is next to the “a”; the “d” is next to the “s” ’ etc) is transformed into so-called “procedural rules”, which are completely internalized, beyond our conscious manipulation. This greatly reduces the involvement of working memory, and protects the skill from the types of interference that other types of memory are vulnerable to.
It also means that when we do have a failure, we really don’t know how to deal with it. A conscious mental search is not going to retrieve the needed information, because the information we want is not in our accessible database. So what usually happens is that we are forced to default to our backup — the declarative information we encoded during the original learning process. It is this that accounts for the lack of fluency in the subsequent actions; to regain fluency, you must engage the unconscious action sequence.
I don’t know of any research that has looked into these occasional glitches, but I presume that what happens is that the action sequence doesn’t immediately engage. As soon as it doesn’t, we pay attention — that makes it even more likely that the action sequence won’t be triggered, because conscious awareness is precisely what we don’t want.
One piece of research that is relevant to this is a recent study that looked at the phenomenon of “choking” — top athletes performing below par at crucial moments. It’s suggested that the problem lies in part in the athlete paying too much attention to what they’re doing. Skills are the one area of memory where too much attention is deleterious to performance!
I think the best way to deal with this very occasional glitch in performance is to relax, stop thinking about what you’re doing, go back a little in the action sequence to an obvious starting point (if you can’t or don’t need to physically re-do earlier steps, mimic the steps). Remember that skills are stored as sequences, and it’s hard to break in halfway through a sequence, you need to start at the beginning.
You might also be interested in a related (but separate) issue, that of action slips, which are a product of a lack of attention, not a surfeit.