To make learning stick, go the extra mile

This is a cross post from the npj Science of Learning Community blog.

I’ve been teaching myself the guitar lately. Like many people, I start with a song that I like, and once I’m semi-proficient at it, I get bored and immediately move on to the next one. And so it goes, day after day, rinse and repeat. But here's the rub: even with diligent practice, none of songs seem to stick with me.

As it happens, I’ve been going at it all wrong. Musicians often practice well past the point of mastery, playing a song over and over until they nail every note. Rehearsing even when you’re no longer improving—what scientists call “overlearning”—may seem like a total waste of time. But according to a study published in Nature Neuroscience, overlearning rapidly alters the neurochemical balance in the brain, “locking in” your hard-earned skills in place.

When we learn, our synapses physically change to encode the memory. But this change isn’t necessary permanent. For memories to stick around, the brain goes through a process called “consolidation”, in which it strengthens the connections that will likely be useful in the future. Consolidation transforms short-term memories into long-term ones, and often happens over the course of hours after learning (often while we sleep).

Overlearning, however, is consolidation’s curious sibling: similar on the surface, but different at its core. To demonstrate overlearning’s powers, researchers asked two groups of volunteers to look for a striped pattern hidden in one of two flashing images on a computer screen. To make things harder, the images were distorted with static-like noise. Generally it takes about 20 minutes of practice for people to become good at the task.

At this point, the overlearning group practiced for another 20 minutes. Both groups then took a 30-minute break and moved on to a competitor task in which the stripes were oriented at a slightly different angle.

The next day, the volunteers were retested to see if they could still identify the striped patterns. Remarkably, the groups’ performances completely diverged.

The first group of learners performed well on the second task, but seemed to have completely forgotten the first one—as if they had never trained at all. This is expected, the researchers say. Learning seems to trigger a “plastic period” in the brain, when the circuits are primed and easily changed. Because both tasks were so similar, they likely used overlapping neural circuits. Thus, by learning the second task, the volunteers had unwittingly overwritten the first task in their malleable synapses—an effect dubbed “retrograde interference” in the field.

But remarkably, this effect was completely reversed in the overlearning group. These volunteers mastered the first task, but performed only half as good on the second. In other words, overlearning made the first skill “resilient”; so resilient, in fact, that the second task was pretty much blocked out.

To dig into why this happens, the team turned to Magnetic Resonance Spectroscopy (MRS), a type of brain scan that tracks protons in neurotransmitters and other metabolic chemicals in the brain. The team focused on the visual regions in the brain, which were previously shown to be involved in learning the stripe task.

The researchers found that immediately after learning, the brain generally showed a spike in glutamate, a type of excitatory neurotransmitter that’s crucial for plasticity at the synapse. At the same time, the levels of an inhibitory transmitter GABA decreased. The result? The brain regions involved in learning were hyper-excitable, primed and ready for new learning—unfortunately, at the expense of the previous skill.

Overlearning, on the other hand, completely flips the pattern by increasing GABA and lowering glutamate. In other words, overlearning cools a hot brain down: it rapidly changes the brain from plastic to stable, without undergoing the usual consolidation process.

What this means is that overlearning is likely to be useful when you’re trying to really nail down a skill when you know you’ve got a similar task coming in the near future: playing two songs with similar chord progressions on the guitar, for example. Overlearning the first will stop the second song from interfering.

Scientists haven’t yet rigorously tested the effect of overlearning on fact-based memories, though the team speculates that the trick should also work, especially if the memories normally interfere with each other.

That’s not to say overlearning is all-powerful. For one, studies suggest that you need to at least double up the duration of your learning to reap the benefits of overlearning on memory retention. So if it takes you three hours to play a song fluidly—practice it for six (yikes!). For another, overlearning stabilizes the first task at the expense of a second, which means that it might not be a good idea to immediately jump to a similar task—you may end up spinning your wheels.

And finally, don’t forget to reap the benefits of good old classical memory consolidation. This is an easy one: just go get some quality shuteye.