Story shape

Why do stories from unconnected cultures fall into such similar shapes?

It is a striking claim, and it is repeated everywhere: myths from societies that never met turn out to tell the same story. A hero leaves home, faces trials, is transformed, returns changed. If that is genuinely true it demands an explanation — something about human minds, or about what stories are for. But before we explain a pattern, there is a duller question to settle first: how much of the pattern is in the stories, and how much is in the way we went looking?

Begin with the strongest evidence rather than the most famous. Vladimir Propp, working in the 1920s, took a body of Russian wonder tales and found that beneath wildly different casts and settings they were built from a limited stock of narrative functions — an interdiction issued and violated, a villain's deception, a departure, a test, a struggle, a return — appearing in a stable order, though never all at once. That is a real finding, and it is precise enough to be checked. Notice its scope, though: one genre, one tradition. Propp was describing a corpus, not humanity.

Then there is the catalogue tradition — the Aarne-Thompson-Uther index — which records recurring tale types across Europe and Asia, the same plot turning up in dozens of languages. Recurrence there is well documented. But ask what it demonstrates. Neighbouring peoples trade, marry, conquer and translate. A story that appears from Ireland to India may be showing us contact, not independent invention — and inheritance from a common ancestor tradition, not a common mind. The premise of "unconnected cultures" is doing a great deal of work in our question, and it is often simply false.

Now the famous case. Joseph Campbell's monomyth gathered material from everywhere into a single departure-initiation-return arc. It is exhilarating to read, and folklorists have been unkind to it for reasons worth taking seriously: the examples are selected rather than sampled, counter-cases are not sought, and the template is loose enough that a determined reader can fit almost anything into it. Two Campbell readers rarely agree on which stage a given scene occupies — a symptom of a pattern with a lot of give in it. There is also a feedback loop people forget. Campbell was absorbed into screenwriting manuals in the 1980s and taught to a generation of writers, so contemporary films really do follow the monomyth — because their authors were handed it as a blueprint. That is a pattern created by the theory, not evidence for it.

Kurt Vonnegut proposed something narrower and testable: stories trace simple curves of fortune over time, a man falling in a hole and climbing out, a boy getting, losing and regaining a girl. Later work applied sentiment analysis to a large corpus of English fiction and reported a small number of recurring emotional arcs. That is genuine measurement — but the corpus was one literary tradition, and an emotional arc is not a plot. Two stories can share a curve and share nothing else.

So what survives all this deflation? Something, and it is not nothing. Comprehension research keeps finding that people encode and recall narrative as a causally connected chain built around an agent with a goal, and that unfamiliar stories get regularised toward the listener's own expectations in retelling — Bartlett's experiments with an unfamiliar folk tale showed this a century ago. A story with no agent, no obstacle and no change of state is hard to remember and hard to pass on. That is a filter on what circulates, and it is the same filter everywhere.

The disagreement here is often mistaken for a fight over whether patterns exist. It is really a fight about which claim is being made, and the claims differ enormously in strength.

The weak claim — narratives generally involve an agent, an obstacle and a change of state — is well supported and unsurprising, since it follows from what makes a sequence of events memorable and retellable at all. The strong claim — that one specific template with named stages underlies the world's myths — is contested, and the criticism is methodological rather than aesthetic: selected evidence, elastic categories, and no procedure that could have falsified it.

Three explanations also compete for whatever similarity is real, and they are not mutually exclusive. Shared cognition, which would produce independent convergence. Diffusion and common descent, which would produce similarity with no cognitive claim at all. And observer effects — survival bias in what got recorded, translators smoothing the unfamiliar toward the familiar, and modern authorship following the template deliberately. Separating these requires sampling that most sweeping claims have never done.

One more caution: flattening is a cost. Reading very different traditions as three renderings of one arc discards precisely what each does differently — cyclical rather than resolving structures, communal rather than individual protagonists, endings that refuse transformation. The universal reading is not neutral; it foregrounds the traditions the template was built from.

• How Bartlett's retelling experiments show a story being reshaped toward the listener's own expectations.
• What a properly sampled cross-cultural test of narrative structure would have to look like.

Every term the collection defines is gathered in the glossary.