COGNITIKA
EN
Get in touch

Part of Evidence of Educative Game Design. Effect sizes are explained in How large is the effect, really?

Four principles that make a game teach

Design quality is what separates a game that teaches from one that does not, and design quality is not a single decision. It is four of them, each with its own evidence, and each within a designer's control.

By Ilija Bojović, Founder / Lead game designer · 16 August 2026

The learning is the progression mechanic

The most consequential decision is what a learner must do in order to advance. Where the subject matter is tested between the enjoyable parts, the game is not teaching it. The test is teaching it, and the game is wrapping. Where exercising the skill is how the learner gets through, the game teaches continuously, and the score is a by-product rather than the point.

Clark and colleagues found that effects varied sharply with design quality rather than clustering near the average, which is why this decision carries the weight it does. Sailer and colleagues tested the negative case directly: gamification is not effective on its own. Points and badges attached to unchanged material do not teach that material. The mechanic has to do the work, not decorate it.

The design term for this is intrinsic integration: the game's structure and its content are the same system, not two systems glued together. Habgood and Ainsworth demonstrated the effect with a mathematics game called Zombie Division, tested with children aged 7 to 11. That is a child study, and the age should be stated plainly rather than quietly extended to adults. What holds regardless of age is the mechanism Clark and Sailer already establish for adult-relevant populations: content bolted onto a game teaches less than content built into one.

Feedback carries the loop

Every game runs a cycle: the learner acts, the rules respond, and the cycle repeats. The response is where learning happens.

Wisniewski, Zierer and Hattie reviewed 435 studies of feedback and found a moderate average effect, close to 0.5 — roughly two out of three learners did better with feedback than without it. The important finding was not the size. It was the condition attached to it: the content of the feedback decides whether it works. Feedback that says what went wrong and what to try differently substantially outperforms feedback that only marks an answer right or wrong, a distinction Hattie and Timperley's earlier framework accounts for directly.

A game has a structural advantage here that a classroom does not. Specific, immediate, consequence-bearing feedback is a normal part of play. Producing the same thing by hand, for every learner, every attempt, is expensive with a human assessor and rare in a lecture. This is also where the principle connects to what happens after the game: a debrief is the same feedback loop, run once, at a larger scale.

Difficulty must be desirable, and matched to the learner

Bjork's work on desirable difficulties established that conditions which slow acquisition often improve long-term retention. The qualifier in that sentence is not decorative: difficulty that defeats the learner is not desirable difficulty. It is simply a barrier.

Chernikova's review supplies the calibration. Scaffolding has to match prior knowledge, and support that helps a beginner can get in the way of someone more advanced. This is where working adults need real attention, and it is worth being precise about why. Lei and colleagues, reviewing game-based learning in school science, found the effect was stronger for primary pupils than for undergraduates. That is a finding about school and university populations. The useful reading of it is not that games fail with adults; it is that adults arrive with substantial prior knowledge, so material pitched for a beginner lands in the wrong place. The scaffolding has to move with the learner, not stay fixed at one level. Choosing the right format also affects how difficulty is delivered, which is the subject of the next article.

Practice must be distributed

Dunlosky and colleagues reviewed the study techniques people commonly use and rated only two as reliably effective across conditions: testing yourself, and spreading practice over time rather than massing it into one session. Cepeda and colleagues measured the spacing effect directly, and Roediger and Karpicke showed that attempting to recall something beats simply rereading it.

Games are unusually well suited to both, and suited to them without extra design effort. Repeated attempts are a normal part of play rather than an imposition on it, and a game a learner returns to across several weeks distributes practice without anyone having to schedule it.

What this adds up to

None of these four principles is exotic. What is unusual is treating them as requirements rather than as nice-to-haves layered on afterwards. A game that gets the progression mechanic right, delivers specific feedback, matches difficulty to the learner, and is returned to more than once, is doing everything the evidence says produces learning. A game missing all four is, at best, entertainment with a lesson attached.

Ilija Bojović is Cognitika's founder and lead game designer, with nearly five years designing and leading commercial mobile games. More about the team.

References

Back to articles