Learning by Adventure

Ask an adult what they remember from school and you'll get a strange list. Almost never the definitions. Almost always the stories — the experiment that went wrong, the teacher's digression about a shipwreck, the one lesson that was structured like a mystery.

This isn't a failure of attention. It's how memory works, and the research on it is old and unusually clean.

Narrative is a memory technology

The classic demonstration is Bower and Clark's: participants asked to memorise lists of unrelated words recalled six to seven times more when they linked the words into a narrative chain than matched controls did. Same words, same study time. The only difference was that one group had a story and the other had a list.

Later work extends it. Mar and colleagues compared comprehension and recall of stories against essays and found stories easier to understand and to remember. Research on children's recall found that highly scripted narrative content was recalled better, and — the part that matters — didn't decay across later sessions the way weakly structured content did.

The mechanism isn't mysterious. A story supplies causality. Each element is held in place by what caused it and what it caused, so recalling one pulls up the next. A list has nothing holding it together but your effort.

Games add the part stories are missing

Narrative gives you retention. It doesn't by itself give you application — knowing a story about photosynthesis is not the same as being able to use it.

That's what game structure adds, and the evidence there has also firmed up considerably. A 2024 meta-analysis of digital game-based learning in schools found a medium effect on overall learning (g = .54) and a larger one on cognitive outcomes (g = .67). A three-level meta-analysis published in 2026 put game-based learning at d = 0.871, meaningfully above gamification at d = 0.754 — the distinction being between a real game and points bolted onto a worksheet.

The finding we keep coming back to is from a meta-analysis of gamified learning across 37 trials: of all the combinations of game elements studied, the highest impact came from rules and goals, plus challenge, plus mystery.

That is a description of an adventure game. It is also, almost exactly, a description of what Conch is built to author.

What an adventure does that a quiz doesn't

Consider teaching the water cycle.

The worksheet version asks a student to label a diagram. The gamified version gives them a badge for labelling it faster. The adventure version is this: you're trapped in a valley where the river has stopped. To get out, you need to work out where the water went — and the mountain, the cloud cover, and the farmer who remembers last summer each hold part of the answer. You can't leave until you understand the system well enough to explain what broke.

The difference isn't motivation. It's that the second version makes understanding load-bearing. You don't demonstrate knowledge after the fact; you can't proceed without it. Getting the concept wrong doesn't cost you points, it costs you the exit.

This is also why the authored part matters. If a student could simply declare "I fix the river," there is no learning event — just a sentence. Conch adventures have real world state: an object you don't have is an object you don't have, and a door opens because you worked out how, not because you asserted it. That constraint is the pedagogy.

Multiple routes, and why they're the point

Good adventure design gives every obstacle more than one solution. So does good teaching.

A locked door might open three ways: find the key, convince the guard, or notice that the hinges are on your side. Three different kinds of thinking — search, persuasion, observation — and a student who is strong in one gets in. They don't see the routes they didn't take. They just experience a world that had a way through for them.

For a teacher, this is differentiation that doesn't require writing three worksheets. You author the obstacle once with several valid solutions, and the class self-sorts.

Where the audio matters

Conch adventures are played by speaking. That does several things for a classroom:

No devices required. A single speaker serves a group. In settings where one-to-one hardware is not realistic — and that's most of the world — this is the difference between usable and not.

Reading fluency isn't the gate. A child who understands a great deal and reads slowly is not held back by the interface. Neither is a language learner, whose listening runs ahead of their reading.

Blind and low-vision students get the same activity. Not a described version, not an aide reading a screen aloud. The identical adventure, played the same way. That is rare enough to be worth stating plainly.

Speaking is practice. In language learning especially, the constraint is rarely comprehension — it's the willingness to produce speech in front of other people. A world that requires you to say something to proceed, and never judges the accent, is a lot of low-stakes repetition.

Building one

The reason this hasn't happened before is that authoring an adventure is game design, and most teachers are not game designers — nor should they have to become one to make a lesson.

Conch Studio is the part of the product built around that problem. A teacher describes the world, the obstacle, and what a student needs to understand to get past it. The tools scaffold the rest: the branching structure, the multiple solutions, the state that tracks what a learner has and hasn't figured out. No coding. No engine to learn.

Then it's shareable — to a class, or to the wider library, where another teacher can play it or set it for their own students. A good lesson usually dies in the room it was taught in. This one doesn't have to.

The honest part

Two things we don't yet know.

We don't have outcome data for Conch specifically. The research above is about narrative and about game-based learning in general, and it is substantial — but nobody has yet measured what a student retains after a Conch adventure versus a lesson. That study is one we want to run with a partner rather than claim in advance.

And effect sizes in this literature vary a lot by subject, duration, and design quality. Games are not automatically better than teaching. A badly designed adventure is worse than a good lecture. What the research supports is that well-designed game-based learning works — which puts the weight back on the authoring, which is exactly where we think it belongs.

If you teach, and you want to try building one, the studio is open.