The Reward Trap

When I was in elementary school, Pizza Hut ran a program called Book It. Read a certain number of books, get a free personal pan pizza. I loved it. I still remember the little pin you’d get, the excitement of walking into Pizza Hut with my certificate, and the little pan pepperoni pizza I’d order. I remember feeling like I really did something.

What I don’t remember is a single book that I read. Even then I probably couldn’t have told you what any of them were about. It was a quantity task. I wasn’t reading to understand or to discover or to fall in love with stories. I was reading to accumulate.

And the moment the program ended, so did my motivation. It wasn’t until college I truly read a book for the sake of it again.

This is a story about a system working exactly as designed. And it’s still largely in place in classrooms today. And that's a problem.

Let’s name these loveable cattle.

Sticker charts. ClassDojo points. PBIS token economies. Accelerated Reader. Pizza parties for attendance. Honor roll assemblies. Clip charts that publicly rank children’s behavior on a spectrum from green to red. These systems are everywhere. PBIS alone operates in over 28,000 U.S. schools. ClassDojo claims to reach 95% of pre-K through 8th grade classrooms. Accelerated Reader is in roughly 60,000 schools.

They share a common logic: if we reward the behavior we want, we’ll get more of it. In the short term, they’re right: kids do read more books when there are points attached, they do sit quietly when there’s a clip chart watching. The behavioral research on PBIS, for instance, is genuinely strong. Students in PBIS schools are about a third less likely to receive an office discipline referral.

But these are compliance measurements. They don’t take into account anything that’s happening inside the minds of those compliant (and non-compliant) students. This narrative doesn’t share about what happens to the motivation underneath the behavior because that’s where the story gets uncomfortable.

In 1973, researchers at Stanford took a group of preschoolers who already loved drawing and split them into three groups. One group was promised a reward for drawing. Another received the same reward as a surprise afterward. The third just drew, no rewards.

Two weeks later, the kids who had been promised the reward spent about half as much time drawing as everyone else. The kids who got a surprise reward? Unaffected. The ones who just drew? Unaffected. Only the children who expected the reward lost interest.

This is called the overjustification effect. When you give someone an external reason for doing something they already enjoy, you replace their internal reason with your external one. They think, “If this task were really fun or really worth doing, why does this person feel the need to reward me for it? It must not be fun or not be something likeable.” Then, if we pull the reward away, the reason to engage disappears with it.

Deci, Koestner, and Ryan’s 1999 meta-analysis confirmed this across 128 experiments. Tangible, expected rewards reduced intrinsic motivation with an effect size of d = -0.40. And here’s the finding that concerns us educators the most: children were more susceptible to this effect than adults. It makes sense, they trust us to know!

The undermining was strongest when rewards were given simply for doing a task (engagement-contingent) and nearly as strong when given for completing it. Even performance-contingent rewards, the kind defenders of incentive systems most often champion, showed a significant negative effect. The one exception is positive verbal feedback, which actually enhanced motivation in adults, though the effects were weaker for children.

Here’s where I need to complicate the narrative slightly, because accuracy matters. It’s more nuanced than just “rewards = bad.”

The overjustification effect requires pre-existing intrinsic motivation to undermine.

For the student who already loves reading, tokens tank that love. But for the student who has never experienced success with a book, who has been defeated by reading so many times that they won’t pick one up voluntarily, the tokens can get them started, get the ball rolling (hopefully) long enough for them to start to enjoy reading for themselves. You can’t erode motivation that doesn’t exist.

Self-determination theory maps a continuum from complete amotivation through external compliance all the way to genuine intrinsic engagement. The research shows that students can move along this continuum, from “I do this for the points” toward “I do this because I see its value” and eventually “I do this because it matters to me.” But that movement requires specific conditions: a meaningful rationale for why the task matters, acknowledgment of the student’s feelings (including “I know this feels boring”), and genuine, not performative, choice.

For administrators, this means the question isn’t “should we ever use extrinsic supports?” The answer to that is yes, sometimes, temporarily, for students with no initial engagement. The question is whether your systems are designed as scaffolding that gets removed when no longer assisting or if they are established, permanent architecture that students never outgrow.

Accelerated Reader is probably the clearest case study. A meta-analysis of 44 studies found a marginal achievement effect of g = 0.278 in controlled designs. But when researchers tracked over 1,500 seventh graders, students who used AR in elementary school showed no difference in reading habits compared to those who never used it. In some districts, non-AR students were reading more. The program’s point structure incentivizes choosing long, easy books over short, challenging ones. Teachers restrict students to AR-quizzed titles because only those “count.” Parents report children refusing to read books that don’t have quizzes. The system doesn’t build readers. It builds point collectors.

ClassDojo is used by 3 million teachers and has essentially no rigorous, independent evidence that it helps students. What research exists describes it as a surveillance system that gamifies compliance, with documented concerns about disproportionate negative points for Black boys. Unlike a sticker chart that eventually goes in the trash, ClassDojo creates persistent digital behavior profiles.

Behavior clip charts publicly sort children into good and bad categories in front of their peers. Shame research is unambiguous: children who experience public shame don’t improve, they get worse. One study found that children whose names were posted at the bottom of a public ranking actually behaved worse afterward. Nothing destroys psychological safety quite like public shaming.

Even attendance awards backfire. Harvard research across 15,000 students found that students who received attendance recognition had worse attendance in subsequent weeks.

I’m not saying the people who implemented these systems have bad intentions, most are trying to solve real problems. But the road between “this produced a short-term behavioral change” and “this built lasting motivation” is longer than most realize, and the research suggests many of these systems take us in the wrong direction.

The alternative isn’t removing all structure and hoping for the best. It’s designing environments that build intrinsic motivation through autonomy, competence, and relatedness, the three needs that Self-Determination Theory identifies as foundational.

Training teachers in autonomy-supportive practices, where students get meaningful choices, clear rationales, and informational feedback, produces an effect size of d = 0.63. That’s larger than most educational interventions. Concept-Oriented Reading Instruction, which integrates reading with content knowledge and emphasizes choice and relevance without any point system, produces effect sizes reaching 1.94 for curiosity. No data on compliance.

A few principles that administrators can act on:

• Audit your reward systems for the overjustification trap. If you’re rewarding activities students already enjoy (or could enjoy), you may be teaching them that those activities require compensation. Reading is the most obvious target, but there are many in the schoolhouse.

• Design extrinsic supports as temporary scaffolding with a planned fade. If a student needs initial external motivation, pair it with rationale, choice, and competence-building. Then reduce the external support as engagement grows. Just like any scaffold, if your system has no fade plan, it’s not scaffolding. It’s building dependency.

• Replace public behavior tracking with private, informational feedback. The research on shame is clear. Clip charts and public ClassDojo scores don’t motivate students, they humiliate them. Move to systems where this corrective feedback is between the teacher and the student.

• Redirect resources from incentive programs toward autonomy-supportive instruction. The training that produces d = 0.63 effects costs less than most schools spend on behavior management technology and reading incentive programs combined.

I still remember that little Book It! pizza. I remember feeling proud. What I can’t tell you is how any of those books changed how I thought because I wasn’t thinking about the books. I was thinking about the pizza.

Fifty years of research says that’s exactly what these systems produce. The question is whether short-term compliance is worth the long-term cost to the thing we actually care about: students who want to learn when nobody’s counting.