Aug. 26, 2009 — -- There is a biological reason why we learn more from our triumphs than from our failures, according to new research at the Massachusetts Institute of Technology.
The fact that we learn more from our successes than from our failures is not new. Even family pets respond more to reward than punishment, and so do children, according to numerous psychology studies.
But what wasn't known until now is how feedback from the environment guides the learning, according to Earl K. Miller, professor of neuroscience at MIT and senior author of a study in the July 30 issue of the journal Neuron.
Miller and co-authors Mark Histed, now a postdoctoral fellow at Harvard Medical School, and Anitha Pasupathy, now an assistant professor at the University of Washington, reached their conclusions after studying neurological activity on the cellular level in monkeys, which, as we all know, have brains that function very much like ours.
"The main thing that's new here is there have been a lot of studies about how the brain learns, but there is very little understanding about how feedback from the environment guides learning, and that's critical because that's the way we learn," Miller said in a telephone interview. "We learn from the consequences of our actions."
The scientists used monkeys in the research because it isn't possible to monitor individual brain cells, or neurons, without implanting an electrode in the brain, and of course that's out of the question with human subjects. But monkeys can learn things, and then relearn things, fairly quickly, although they are not as good at it as humans.
Miller said the research led to two findings.
"We found that neurons in the prefrontal cortex and the striatum, two brain areas known to be involved in learning, keep track of recent successes and failures for many seconds, long enough for it to play a role in guiding the learning the next time an opportunity to learn comes up," Miller said. "The second thing we found was that the neural processing in the brain improves after a recent success and doesn't improve much after a recent failure. That's quite new. No one has really shown that before."
Neurons Retain Memory of Correct Action
Histed, lead author of the study, said via e-mail that the research explored the neural circuitry "at the single neutron level."
"We found that the memory of past actions could be maintained by the activity of the neurons," Histed said. "Some neurons started to fire after correct responses, and continued to fire for several seconds until the next action was to be made."
In other words, the neurons themselves retained a memory of the correct action, paving the way for continued success.
The researchers found that monkeys that were rewarded for the right response to a cue learned quickly how to respond the next time they saw the cue, but monkeys that responded incorrectly weren't any better equipped to deal with the same cue the next time they saw it.
"If the monkey just got a correct answer, a signal lingered in its brain that said 'you did the right thing.' Right after a correct answer, neurons processed information more sharply and effectively, and the monkey was more likely to get the next answer correct as well," Miller said. "But after an error there was no improvement. In other words, only after successes, not failures, did brain processing and the monkeys' behavior improve."
There is apparently an evolutionary basis for why the brain learns that way, Miller said.
"There's a practical reason for it," he said. "Successes are more informative than failures."
If you fail at something you probably know why. You got fired because you showed up late most of the time. Your spouse left because you showed up too many times at the wrong place, with the wrong person. You already know the reason you failed.
"If you succeed, everything has gone right, so there's a lot more information in successes than failures," Miller said. "The brain probably evolved to take advantage of successes because there's more information there."
Many Questions Still Unanswered
The research may be an important contribution to our understanding of how we learn, but in this rapidly evolving field, there are still many, many unanswered questions.
Histed noted that scientists still don't understand how the brain works.
"How many neurons store a given memory?" he asked rhetorically. "How does the code work? Do patterns of activity across hundreds or thousands of neurons matter, or can we just study single neurons one by one? I would say most of the questions are unknown."
At the very least, the MIT research challenges an old bromide. Maybe we really don't learn from our mistakes.
And maybe that's why we repeat them, over and over again.