My first experience as an educator was teaching middle-school mathematics and science in Baltimore. As a teacher, I was interested in the content and had a deep desire to impart my knowledge and enthusiasm to my students. I also worked hard to keep my students focused and tended to run my classroom like a tight ship. Day after day, I often would say the wrong thing, or say something in the wrong way.

The truth was that I tended to push too hard on students in ways that they didn’t respond to, and I often lost control of the class. Classroom management didn’t come naturally to me, and I often felt unsuccessful. And like many teachers, I had wrongly assumed that if I persisted with my strictness and focused teaching style, I would eventually gain control of the class.

To be clear, persistence is an indispensable quality that can have a dramatic impact on success. Trying something repeatedly can be productive under the right conditions. Often our moments of greatest frustration come just before the moment of breakthrough, where the scales fall from our eyes and we finally understand the solution.

But despite what we often tell our students, persistence can actually be counterproductive at times. If students are unprepared to take on a particular type of problem because they didn’t understand the prerequisite concepts, persistence isn’t helpful. It can even be harmful when routinely misapplied. We may make the same mistake over and over again until it becomes a habit. And students who continually find their efforts don’t lead to improvements may become less likely to persist over time.

The goal is not to remove the obstacles from learning. Rather, we want to make sure the right kinds of obstacles are in place.

Educators must be able to distinguish between moments of “productive persistence” and moments where students are simply spinning their wheels. Believe it or not, “wheel-spinning” is now a technical term and has become a topic of interest to researchers who study how children learn. According to researchers Joseph Beck and Yue Gong, wheel-spinning is generally defined as “students who do not succeed in mastering a skill in a timely manner.”

You can imagine how helpful it would be if teachers and computer systems could detect which students are likely to spin-their-wheels. As it turns out, wheel-spinning can be detected “in an early stage with high accuracy.” One study showed that a “wheel-spinning detector” could identify a high percentage of the students who had spinning wheels. (The tool relied on publically released data from Carnegie Learning.)

In another recent paper, published this April, researchers led by Anthony Botelho created a wheel-spinning detector using ASSISTments, a free online learning platform that my team at the Worcester Polytechnic Institute developed for math education. It uses “mastery-based” techniques for assignments, meaning students don’t need to complete long problem sets once they demonstrate they understand a concept.

In that study, mastery was defined as correctly completing three problems in a row, and researchers defined “wheel-spinning” as failing to gain mastery even after trying on 10 different problems (and getting feedback on each one). The researchers looked at more than 123,000 homework assignments given to students around the country in the 2016-2017 school year. They found that approximately 5 percent of the assignments led to wheel spinning, and another 5 percent of students exhibited “stop out,” meaning they gave up on the assignment before reaching the 10-problem threshold.

The researchers found that statistical models could be developed to predict wheel-spinning and stop-out behavior, because students’ previous performance can predict wheel spinning in the future. Although more research needs to be done, these models and tools can be enormously useful to teachers, curriculum developers and education technology developers.

Future education technology tools in the nascent field of learning engineering may be able to provide teachers with early warnings about which students need extra guidance, or, perhaps, what concepts the class as a whole may struggle with. Similar feedback may help curriculum developers pinpoint where students typically struggle with certain concepts.

The goal is not to remove the obstacles from learning. Rather, we want to make sure the right kinds of obstacles are in place—neither obstacles that discourage learning because they wrongly presume the child has already absorbed preliminary concepts, nor obstacles that pose no challenge.