What discovery-based learning is
Discovery-based learning is learning where the person works something out rather than being told it. The facilitator sets up a problem, the group attempts it, and the lesson comes out of what happened rather than off a slide. The term goes back to Jerome Bruner, who argued in 1961 that a learner who finds a principle for themselves keeps it longer and applies it more widely than one who receives it finished.
It belongs to a wider family of ideas: John Dewey's learning by doing, Jean Piaget's constructivism, and David Kolb's cycle of experience, reflection, conceptualisation and experimentation. What they share is the claim that understanding is built by the learner, not delivered to them.
One line from the research literature is worth keeping. Whatever the definition, the target information has to be discovered by the learner within the confines of the task and its material (Alfieri, Brooks, Aldrich and Tenenbaum, 2011). The confines are the point. Discovery without a task is a conversation.
What the research says: guided discovery works, pure discovery does not
Pure discovery, where people are left to explore with no structure and no feedback, has a poor record, and it has had one for a long time. Richard Mayer reviewed three decades of studies in 2004 and concluded that learning of this kind works through cognitive activity rather than behavioural activity, instructional guidance rather than pure discovery, and curricular focus rather than unstructured exploration. Kirschner, Sweller and Clark made the same case more bluntly in 2006: minimal guidance does not work for people new to a subject, because the search fills their working memory and nothing is left over for the learning.
The 2011 meta-analysis by Alfieri and colleagues, covering 164 studies, settles it for practical purposes. Unassisted discovery does not help learners. Discovery with feedback, worked examples, scaffolding and elicited explanation does, and in the studies compared it did better than the alternatives.
So the useful version is guided discovery: a bounded problem, an attempt, feedback the group can see, and a conversation that names what happened. Read that list again and it is a description of a well-run learning game.
Why a game is the practical form of it
A learning game supplies the four ingredients without the facilitator having to invent them on the day.
- A constraint that makes the behaviour visible. In Unlock, each person can describe their image but not show it, and the team has to agree an order. Nobody has to be told the session is about communication; the rule makes communication the whole task.
- An attempt the group owns. The first round is theirs: their assumptions, their process, their result. Nothing the facilitator says afterwards is as persuasive as that.
- Feedback nobody can argue with. In Ranking, everyone ranks alone and then together, and the synergy score shows whether the discussion added to its members or averaged them away. A number the group produced is harder to dismiss than an observation the facilitator offers.
- A second attempt. In Diamonds, teams dig for diamonds on a grid with too few moves and too little information, so every move is a hypothesis and the next result is the test. The round that follows the debrief is where the lesson becomes behaviour: one round is an experience, two rounds are a comparison.
How to run a discovery-based session
Six things, in order. None of them is complicated, and skipping any one of them is how the session turns into a good time with nothing to take back.
- Set the problem, not the lesson. If you tell the room the session is about listening, they will perform listening. Give them the task and let the behaviour show up on its own.
- Keep the rules short and the constraint tight. Two minutes of briefing, then play. The constraint does the teaching, so protect it: no showing the image, no talking during the individual ranking.
- Make the result visible. A score, a ranking, a time. Discovery needs something to discover, and the group's own number is the best thing there is.
- Debrief in three moves. What happened, in the group's own words. So what: what it says about how they work together. Now what: one thing they will do differently, named by them and not by you.
- Play again. Run a second round and put the two results side by side. The difference between them is the learning, made measurable.
- Connect it to Monday. End on the real situation the game stands for: the handover, the briefing, the decision taken with half the facts. That is the transfer Bruner was talking about.
Where it goes wrong
The failures are as predictable as the successes.
- No structure. "Explore this and see what you learn" produces anecdotes, not learning. The research above is unambiguous about it.
- No debrief. The debrief is at least a third of the session. Cut it to fit the agenda and you have run an energiser.
- The lesson stated first. Announce the moral and the game becomes a demonstration of it. Discovery only works if there is something left to discover.
- The game as the point. If the mechanic is not the message, choose a different game. Novelty wears off in a round; a constraint that mirrors real work does not.
- Teams too big. Above six people the quiet ones discover nothing, because they never get to act. Four to six per team keeps everyone in the attempt.
Does it work with adults, and does it work online?
Most of the published research is on students, which is worth knowing when somebody quotes it at you. Adults bring something students do not: years of experience to hang the discovery on. That is why the debrief lands faster with a working team, and why the last question in it should always be about their real work.
Online changes the logistics, not the mechanism. Constraint, attempt, feedback and reflection all survive a video call, provided the game runs on each person's own screen and the facilitator keeps hold of the debrief. What does not survive is a game played by one person sharing their screen while the rest watch.
