Ultimately, we want kids to believe they can get good at skills and talents if they practice and try hard. We want them to be persistent in the face of early difficulty. The work by psychologist Carol Dweck and others suggests that this adaptive mindset is dramatically a function of the praise kids hear.

If you’ve heard this research, you know the new rules: praise the process, not the person. Avoiding suggesting that success is due to innate qualities. Instead, steer the child’s attention to strategies they can do again to repeat their success.

A question I often get is “Does this mean I can never tell my kid she’s smart?” We’re not perfect, we’re enthusiastic, and the old “you’re so smart!” just flies off the tongue. Where's the line? Is there a margin for error here?

In everyday life, kids hear a wide mix of praise types from parents, teachers, and other children. Even a kid who gets praised correctly by his parents (“you studied really hard, so you did well on the test”) will hear innate-style praise at school (“You’re so smart in math”). Or they’ll hear “you’re a great gymnast” and “you’re a great drawer.”

So it’s a legitimate question: what’s the impact of a real-life (less-than-perfect) mix of praise?

Well, an interesting experiment last year by two scholars provided some insight. Shannon Zentall and Bradley Morris did a series of drawing games with 135 kindergartners. Each kid heard a little story and then was asked to draw a certain object that had been in the story—an apple, a bus, a cat, a tree, etc. Each child worked through six stories and drawings.The first four drawings were praised, but with different mixes of praise. One group received only process praise, another mostly received process praise, with some person praise thrown in. Then there was a 50-50 group, and so on.

After the last two drawings, the teacher noted something imperfect about the illustration (“the cat is missing an ear” or “the bus is missing a wheel”). This was meant to challenge their hot streak of success. Did they crumble after being criticized?

To find out, the kids were asked four questions to measure their persistence in the face of this negative feedback. For instance, they were asked if tomorrow they’d like to draw again, or instead do something else. And they were asked what objects they’d draw next time. If they were willing to draw the objects they’d made mistakes on—trying to get better—that was scored as persistent. If they only wanted to draw objects they’d been praised for, that was scored as lacking persistence.

Just as Dweck’s work would predict, the type of praise the children heard really affected their persistence. The most persistent kids were ones who never heard person praise and only heard process praise. As the mix of praise migrated from that end of the spectrum to the other, persistence fell. While most of the kids who received process praise were willing to draw tomorrow, only half of the kids who received person praise did.

But the downslide in persistence was not an even slope. Most children who heard 75 percent process praise (thus, 25 percent person praise) were almost as persistent as the top group. There was very little fall-off in persistence for this group. That fall-off was apparent when the praise mix was 50-50.

While parents and teachers certainly aren’t perfect in their praise, this experiment provides helpful guidance: get the praise right 75 percent of the time or more, and a persistent mindset should take root.

This is a picture of a Quick-Cap, which measures electrical activity on the surface of the scalp. While it looks like something out of a futuristic movie about thought control, it’s actually quite comfortable and unobtrusive. While it’s not nearly as precise as a fMRI, electroencephalography (EEG) is much easier to use and drastically cheaper; the cap does a decent job of registering which regions of the brain are firing moment to moment.

Carol Dweck and Jennifer Mangels had Columbia undergraduates wear the cap while taking a computerized trivia test. The students worked through over 200 questions, covering topics from geography, religion, world and US history, math and science, literature, and art history. These questions were chosen because they’d ring a bell of familiarity – students felt like they should know the answer, but often weren’t quite able to recall it. Example: In what country is Kathmandu? The computer adjusted the level of difficulty just slightly, so that most students felt a little frustrated, getting about 60% of the answers wrong. If they were below that level, the computer might feed them an easier question – “Who was the Confederate General at the Battle of Gettysburg?” – but if they were above the level, they got “Who was the Union General at the Battle of Gettysburg?”

Students typed in their answer. A couple seconds later, they saw a green light or a red light to indicate if they were accurate. A few seconds after that, they were shown the correct answer (i.e., General Robert E Lee, or General Meade, respectively). This gave them a chance to learn from their mistakes.

But whether they wanted to learn was an entirely different question.

That was the purpose of the scalp cap. It measured activity in the brain’s attention systems – just how vigilantly were the students paying attention when they found out of their answer was correct or not? And were the students bothering to pay attention a moment later, when they saw the correct answer? Fundamentally, this study separated students who just wanted to do well on the test from students who really wanted to learn new material.

Before the testing, the students had filled out a questionnaire to assess their motivation personality. Based on their responses, Dweck and Mangels split them into two groups. One group was concerned, primarily, with being better than others. They agreed with statements like, “You have a certain amount of intelligence and you can’t do much to change it,” or “It’s important to me to be smarter than other students.” The other group disagreed with those statements, and instead agreed to comments like, “It’s very important to me that my coursework offer real challenges.” This latter group wasn’t into comparing themselves. Let’s call these two groups the Grade-Hungry and the Knowledge-Hungry.

These attitudes had a significant impact on the EEG readings – the students’ motivation personality was visible in the brain waves, on a millisecond to millisecond graph.

The Grade-Hungry students paid extremely close attention to the moment of green light/red light – they really were obsessed with whether they got the answer right or wrong. But immediately after, their attention systems took a break. They checked out, and they weren’t really paying much attention when the correct answer flashed by.

Respectively, the Knowledge-Hungry paid attention (but not quite as obsessively) to whether they were right or wrong, and they paid significantly more attention to the correct answers. They took advantage of the chance to learn. This contrast was most dramatic when each group got an answer wrong. The Knowledge-Hungry activated deep memory regions, indicating they were storing these new facts away for later. Such activity was not nearly as deep in the Grade-Hungry, suggesting a far more cursory interest; instead, their brains seemed to feel threatened by learning they’d gotten an answer wrong. Their brains indicated a far more emotional, fearful response. They clearly did not like being wrong, and they didn't care that Katmandu is in Nepal.

Not surprisingly, when the students were later surprised by a retest, consisting only of the questions they’d gotten wrong the first time around, the Knowledge-Hungry kids did far better.