Mistakes: A key to learning

Attempting a new task almost always involves trial and error. We pay attention to those errors, a new study shows. Our brains store memories of past blunders. We then use those memories to improve how well we do in future attempts, a new study finds. 

David Herzfeld discovered this newly identified type of memory. As a biomedical engineer at Johns Hopkins University School of Medicine in Baltimore, Md., he combines engineering and technology to aid public health.

In the new study, Herzfeld recruited people to play a simple video game. Participants were asked to move a cursor across a screen by manipulating a robotic arm. Critically, the robotic arm and the person’s hand were shielded from the player’s view. Participants instead had to focus on a computer screen. There, they saw a dot and a target. Their goal was to move the dot to the target.

That sounds easy enough. But the researchers could impose some challenges along the way. For instance, in one trial, participants had to move the robotic arm straight forward. But in some cases, the cursor moved a little more than the arm did. Other times, the cursor moved a little less. When those errors occurred in the same direction each time, participants remembered them. With each new attempt, the test participants corrected their movements a bit. And this slowly improved their ability to hit the target. But when those errors kept switching direction — being a little too far, then not far enough — participants ignored them.

Herzfeld likens this to playing a game of darts. “The first dart you throw is a little too low,” he says. When you try to correct on your next throw, it lands high. “You’ve over-corrected,” he explains. By doing so, you have made a mistake in a different direction. If such mistakes happen over and over, your brain learns to ignore them. And it stops trying correct for them.

But later in the game, you might throw the dart too low again. Herzfeld wanted to know what would happen then — could the brain recall the earlier low throw and correct the error?

To find out whether our brains really do “save” errors — remembering them for another time when they might prove useful — Herzfeld carried out a second set of tests. Here, when players moved the robotic arm forward, it rotated 30 degrees (one-twelfth of a circle, or the distance from one hour’s numeral to the next on a clock). One group experienced a clockwise rotation. The others got a counterclockwise rotation. For both groups, this was followed by a second phase with no rotation — the arm moved straight forward. In the third phase of the experiment, the arm rotated counterclockwise for both groups.

Herzfeld found that both groups hit the target faster during the last set of counterclockwise rotations. He expected this for the first group. That’s because they had experienced the same rotation in phase one. And they had “saved” memories of their errors, now using them to make the proper corrections.

The surprise was that the other group — the one that had the opposite rotation in phase one — also improved. Herzfeld calls this an example of “meta-learning.” That happens when we remember errors from one situation and apply them to a different one.

In this case, the people playing the game remembered the clockwise rotation from phase one. When the robotic arm stopped rotating, they made new errors by trying to correct for a rotation that was no longer there. When the direction of rotation changed again, in phase three, participants applied what they had learned in the first two sets of tests. They quickly adjusted, using what they learned earlier about rotation to deal with the new direction.

That switch from phase one to phase two was an important part of the learning process, Herzfeld found. When participants experienced a long pause between those two phases, they did not show the same recall of past errors. During the delay, the participants forgot what they had learned. His team’s study was published August 14 in Science.

It’s an exciting finding, says Farrel Robinson. He’s a neuroscientist at the University of Washington in Seattle and not involved in the research. This study is the first to show that the history of mistakes is what matters in motor learning, he says. Motor learning is the type that has to do with body movements. The new data also show that history determines whether our brains will learn from past errors when building new skills. 

So the next time you set out to try a new activity, be bold. Make mistakes. They just may help you learn faster.

Power Words

biomedical engineering  Combining engineering and biology to aid human health. Professions in this field develop artificial limbs, use biotechnology to produce new drugs and develop models to understand how diseases work.

cursor    (in computing) The dot or marker on a computer screen that points to where the action is. It may depict the end of a word that was just keyed in. Or it can mark where a “pointer” is, so that clicking on a mouse or some other action at that point may trigger a hyperlink or URL to open a new screen.

engineering   The field of research that uses math and science to solve practical problems.

meta-learning    Applying what one has learned in one situation to a novel situation.

motor skills (or learning)     The ability to make controlled movements, especially with the hands or limbs — or to learn new, controlled movements.

neuroscience    Science that deals with the structure or function of the brain and other parts of the nervous system. Researchers in this field are known as neuroscientists.

technology  The application of scientific knowledge for practical purposes, especially in industry.