Dubious Maxim 6: We learn from our mistakes

“And now go,” Neil Gaiman urged his audience at the end of his commencement speech at the University of the Arts, in Philadelphia, in 2015. ”Make interesting mistakes, make amazing mistakes, make glorious and fantastic mistakes.” The idea was that we learn how to do a thing well by first doing it badly.

The speech, posted online, has been watched “many millions of times, and it is also available as a small book,” says Gaiman, and is included in his essay collection The View from the Cheap Seats. The idea that one of the best-known writers of our time has sent millions of people out to fantastically fail is a bit disconcerting, especially since its wisdom depends on the belief that making mistakes is how we learn. And there is no good evidence that that’s the case. Making mistakes may simply be the way we learn how to make bigger and more fantastic mistakes.

The notion that we learn by failing appeared in 1897, in A Dictionary of Thoughts, by Tryon Edwards, an American theologian. “Some of the best lessons we ever learn we learn from our mistakes and failures,” he wrote. “The error of the past is the wisdom and success of the future.” Edwards was the minister of the Second Congregational Church in New London, Connecticut, until he retired in 1857 to write books. Most of the books he published were collections of sermons by other people. For example, he gathered sixteen sermons his grandfather had delivered on a single verse from the New Testament: 1 Corinthians 13:13, “And now abideth faith, hope, charity, these three; but the greatest of these is charity.”

It’s hard not to be cynical about a biblical tribute to charity, repeated at church fundraisers for the past two thousand years, that is based on a mistranslation. The Greek word agape does not mean “charity,” it means “love.” And not just any love, but the kind of transcendent, beatific love that God is believed by many to extend to his children: unconditional and possibly unrequited love. Charity can be viewed that way, as it is for example in Lawrence Scanlan’s Year of Living Generously, but all too often it’s the kind of love that earns one a more comfortable pew and a tax receipt.

Edwards’s assertion that we learn from our mistakes has been repeated by many sages, Winston Churchill, Albert Einstein, and Oscar Wilde among then, none of whom were teachers. William Saroyan, an optimist, remarked that “we get very little wisdom from success.” But is it true that trial and error is better than logic and reason when it comes to solving problems? If we want to find our way out of a dark room, is it better to stagger around in the dark than to turn on a light? Recent neurological evidence (and most of history) tells us that, dwell on them though we will, we learn very little from past mistakes.

History might have have taught us this much earlier. In 1807, for example, France and Napoleon Bonaparte were enjoying a period of relative calm after a decade of wars against most of the rest of Europe. All Napoleon had to do was not invade anyone for a while, and a lasting peace might have allowed French culture and economy to recover from the previous years of turmoil. So, what did Napoleon do? He invaded Spain. Ostensibly, he was trying to prevent Portugal from trading with Britain, but during the ensuing Peninsular War, Napoleon became snarled in the question of the succession to the Spanish throne: he ended up placing his brother, Joseph Napoleon, on the throne, thus alienating himself and France even further. In 1812, with the war in Spain coming to an end, all Napoleon had to do was not invade anyone else for a while. So, what did he do? He invaded Russia. Ostensibly, he was trying to prevent Russia from trading with Britain.

As the title of a Japanese manga has it, “We Never Learn.”

A while ago, before the pandemic placed us all in splendid isolation, Merilyn and I isolated ourselves at 221B Baker Street, an “escape room” in which people are shut into a room, or a series of rooms, with all the tools and clues their little grey cells need to figure out how to escape. There’s usually a story involved. In our case, Dr. Watson had been kidnapped by the nefarious Professor Moriarty, and we had to find him before he was used as bait to lure Holmes to his death. Personally, I’ve never much liked Watson. He never seemed to have learned from his thirty years’ association with the great detective; too often he looked at Holmes in amazement and asked, “How the devil did you work that out?” He didn’t learn from his mistakes, and he didn’t learn from Holmes’s successes. But I went along with the caper anyway, for Holmes’s sake.

In the first of the three rooms we found several enigmatic items: Dr. Watson’s tweed coat with a colour-coded card in a pocket; a wooden box containing a magnet; a short piece of iron bar. We used the card to open a box and retrieved a magnet and tried sticking the magnet to the lock on the door, but nothing happened. We used the magnet to pick up the iron bar. Again nothing. When we slid the magnet along a wall, however, we heard something move on the other side. We kept sliding the magnet until, eventually, a key dropped into a drop box, and we used the iron bar to weigh down a lever that exposed a lock and able to open the door into the second room, where there were similar challenges. Eventually, by trial and error, we figured out the significance of each object, and after about an hour, we emerged triumphant from the third room.

Neurologically speaking, making mistakes teaches us to pay attention to new information that might allow us to avoid making more mistakes. A study conducted at the University of Exeter in 2007 suggested that surprise increases our propensity to learn, especially if the surprise is finding out we made a mistake. During the experiment, subjects were asked to imagine they were working for a medical referral service, and that their job was to detect Jominy fever in a series of blood samples, based on the presence of six differently shaped cells in the samples. When the subjects were told they had predicted correctly, their brains seemed to relax. But when told that their original prediction was incorrect, and were given new information to make new predictions, the length of time their brains spent checking a second range of cell types and making new predictions increased; this time, they spent longer studying the cell samples, and made more accurate predictions. Their brains seemed to have learned from their previous errors.

The researchers concluded that “a stimulus whose consequence is well predicted is processed to a lesser extent than a stimulus that has recently been followed by surprising or unexpected  events.” In slightly plainer words, if we’re given certain information and asked to make a prediction based on that information, and we predict correctly, we stop thinking about it. But if we don’t know if our prediction will be right or wrong (because the last time our prediction was wrong), our brain pays more attention to new information when making new predictions. And we assimilate the new information very quickly (in about 0.12 seconds).

“It’s a bit of cliché to say that we learn more from our mistakes than from our successes,” commented the study’s principle author, Andrew Willis, “but for the first time we’ve established just how quickly the brain works to help us avoid repeating mistakes.”

A more recent experiment, however, suggests that thinking we learn from our mistakes is itself a mistake. In this study, conducted at MIT, monkeys were shown two alternating images on a computer screen. When picture A was shown, a monkey was rewarded when its eyes shifted to the right; when picture B appeared, the reward came when the monkey’s eyes shifted to the left. Electrodes attached to the monkey’s prefrontal cortex measured how long neurons fired as the monkeys figured out which eye movement matched which picture. “If the monkey got a correct answer,” said Earl Miller, who ran the study, “a signal lingered in the brain that said, ‘You did the right thing.’ After a right answer, neurons processed information more sharply and effectively, and the monkey was more likely to get the next answer correct as well.” The monkeys learned faster after a correct answer, because the neurons involved in learning kept firing until the monkey moved on to the next picture. After an error, on the other hand, “there was no improvement. This explains on a neural level why we seem to learn more from our successes than from our failures.”

Tryon Edwards and William Saroyan were both wrong.

As we worked our way through the three rooms in 221B Baker Street, we weren’t learning from our mistakes. When we stuck the magnet to the lock and nothing happened, we did not learn how to open the lock: we simply learned to try something else. This is learning of a sort, but in life there may be millions of other things to try, and “learning” becomes a process of elimination. Failing doesn’t teach us what else to try, it just gives us the negative message to fail better next time. In the rooms at 221B Baker Street, we simply went around trying and erring, sticking the magnet here and there, until we inadvertently stuck it to the key hanging on the other side of the wall. Only then did the neurons in our prefrontal cortexes begin firing. But with the next clue, we were back at square one: having success with the magnet did not lead to success with the next challenge. When we finally emerged from the third room, there was Dr. Watson, safe and sound, sitting in his customary chair by the fire, He looked up at us in amazement, and said, “How the devil did you work that out?”