If you had to guess what a learning game is best at, what would you say?
Most people say motivation. The game makes the medicine go down. A child who groans at a worksheet will happily do the same sums if there is a dragon involved, and the real subject matter arrives smuggled inside the fun.
Three large research syntheses published in the last two years point somewhere else, and the picture they draw is close to the opposite of that one.
What the numbers say
Start with the biggest. A 2026 meta-analysis in Educational Psychology Review pulled together 86 studies covering 7,128 school-age children and sorted the results by what was measured [1]. Domain knowledge, meaning the facts and procedures, improved substantially (Hedges' g = 0.82). Higher-order thinking, meaning applying and analysing and solving problems the child had not seen before, improved slightly more (g = 0.88). Feelings about the subject improved least (g = 0.58).
Read that ordering again. The effect on thinking was the largest of the three. The effect on how children felt was the smallest.
A second synthesis, published the same year in the British Educational Research Journal, went further. It was a second-order meta-analysis, meaning it combined 20 earlier meta-analyses covering 688 individual studies of games in maths education [2]. Cognitive outcomes moved reliably (g = 0.45). Affective outcomes, the enjoyment and confidence part, moved in the right direction, but the effect was small enough that it could not be told apart from noise (g = 0.23).
For our age group, a 2024 meta-analysis in Frontiers in Psychology looked only at early childhood and found the same shape. Of the five outcomes it measured, cognitive development had the largest effect (g = 0.46), ahead of engagement (0.44), motivation (0.40), social development (0.38) and emotional development (0.35) [3].
Three research teams, three methods, one consistent ordering. The thinking moves first.
Why that is the interesting result
The motivation story has a comfortable logic to it. Children like games, liking something makes you pay attention, attention produces learning. Fun as a delivery mechanism.
If that were the whole story, the enjoyment effect should be the biggest one, because it sits at the front of the chain. It is the smallest one. Whatever games are doing, sweetening the subject is not the main lever.
Something else fits better. A game is a place where a child has to decide. Which path leads to the treasure. What happens if I try the bigger number. Every turn asks for a choice, the choice gets an answer within a second or two, and the child adjusts and goes again. A worksheet asks for the same answer twenty times and tells you how you did tomorrow.
That loop of decide, find out, adjust is close to what higher-order thinking is made of. Which would explain why it is the outcome that moves most.
The honest limits
Two things this research does not say.
It does not say any particular game works. These are averages across hundreds of studies with wildly different games, and the spread between them is wide. A meta-analysis tells you the category has real effects available in it. It tells you nothing about the app your child opened this morning.
And the early childhood evidence is thinner than the school-age evidence. The two 2026 syntheses cover K-12 broadly, which puts most of their weight above age eight. The early childhood meta-analysis points the same way with fewer studies behind it [3]. We would rather say that plainly than round it off.
What this changes at home
If a game's job were mainly to make a subject feel nice, a happy child would be proof it is working.
That test is too easy. A child can be delighted by a game that asks them for nothing but taps.
The better question is what the game made your child decide. Sit next to them for one round and watch for the moment they pause before choosing. That pause is the product. A game that never produces it is entertainment, which is fine, and it is a different thing from what this research measures.
We hold Wonderix to the same test. When we review a level, we ask whether a child had to think to finish it. Finishing it smiling does not count on its own, and levels that only manage the smiling get cut, including some that score well on how much children enjoy them.
Games are not a spoonful of sugar. They are a place to practise deciding, and deciding is the expensive part of learning.
References
[1] Cheng, M., Lai, X., Zeng, W., Bai, S., Duan, C., & Sun, D. (2026). Unlocking the Power of Gamification in K-12 STEM Education: A Meta-analysis of the Effects On Domain Knowledge, Higher-order Thinking, and Affective Outcomes. Educational Psychology Review. https://doi.org/10.1007/s10648-026-10148-1
[2] Yao, D. (2026). The impact of gamified learning in mathematics education on students' cognitive and affective outcomes: Insights from a second-order meta-analysis. British Educational Research Journal. https://doi.org/10.1002/berj.70144
[3] Alotaibi, M. S. (2024). Game-based learning in early childhood education: a systematic review and meta-analysis. Frontiers in Psychology, 15, 1307881. https://doi.org/10.3389/fpsyg.2024.1307881