Contagious yawning

Why does watching a stranger yawn set you yawning when the yawn does nothing for you?

Someone across the train carriage yawns. You have never met them, you owe them nothing, and within seconds your own jaw is going. Nothing has been communicated and nothing has been gained.

There are really two questions tangled together here, and almost every popular account answers the second by helping itself to an answer to the first. What is a yawn for? And why should seeing one produce one? It is worth taking them apart, because the honest position on the first is much weaker than the confident stories in circulation suggest.

Start by separating spontaneous yawning from contagious yawning, because they are not the same phenomenon. Spontaneous yawning is very old and very widespread across vertebrates, and human fetuses yawn in the womb long before birth. Whatever it is, it is not learned, not social in origin, and not about company. Contagious yawning is far narrower: it is well documented in humans and chimpanzees, reported in some other primates and in dogs, and absent or unconvincing in plenty of species that yawn perfectly well. So contagion is an add-on to something older.

Now, what is the older thing for? The honest answer is that nobody knows. The intuitive oxygen story — that you yawn because you are short of air — was tested directly, by having people breathe air enriched in carbon dioxide or in oxygen, and yawn rates did not move the way the story requires. That is a clean falsification and it settled the matter. The currently famous replacement, that yawning cools the brain, is on much shakier ground than its popularity implies: the supporting work is largely correlational, leans on ambient and seasonal temperature, uses small samples, and has not held up consistently. Treat it as a live hypothesis with thin evidence, not a finding.

With the function unknown, look at the contagion on its own terms. The best-supported description is unglamorous: perceiving an action lowers the threshold for producing that same action. This is the same family as automatic imitation, where watching a movement makes the matching movement easier and the mismatching one harder.

Two observations make that account testable rather than decorative. First, the trigger is specific. A mouth opening for other reasons does not work nearly as well as a yawn, which is what rules out simple arousal or generic facial movement — if any gaping face served, the resonance account would be finished. Second, the trigger survives extraordinary degradation. A cropped mouth alone sets people yawning. So does the written word. So does reading this paragraph, quite possibly. A cue that thin cannot be carrying anyone's inner state to you, which is a problem for accounts that require you to appreciate the other's condition.

And that brings us to the claim everyone has heard: that catching yawns measures empathy. This has repeatedly failed to replicate. Well-powered studies find no stable relation between susceptibility and validated empathy measures; in one large sample the strongest individual predictor was simply age, with contagion declining as people got older, while empathy scores predicted essentially nothing. The companion claim, that autistic people catch yawns less, turned out to depend heavily on where participants were looking — when fixation on the face is controlled, the group difference shrinks or disappears. The eyes were doing the work, not the empathy.

So the threshold framing is what survives. A yawn is not summoned; it is released. There is some slowly building drive — we do not know of what — and the sight of another yawn does not add to that drive but lowers the bar it has to clear. That predicts the pattern people actually report: contagion bites hardest when you were already tired, and hardly at all when you are alert. It also explains why resisting fails. Instructing people to suppress a yawn increases the reported urge rather than removing it, which is what you expect if the underlying state is untouched and only the expression is being blocked.

The refinement that connects everything above is that nothing crosses. This distinguishes yawning sharply from the things we usually call contagion, where something real propagates — a behaviour spreading through a social network, or ripening fruit releasing ethylene that reaches the fruit beside it. Your yawn is manufactured entirely by you. The other person's yawn only changed the height of your threshold, and a written word can do the same job, which no transmissible agent could.

Two honest caveats. First, the reliability of the phenomenon and the reliability of its explanations are very different things: that people yawn more after yawn stimuli than after control stimuli replicates well, while nearly every account of why is contested. Second, some datasets do report more contagion between familiar people than strangers, which would be a social modulation worth having — but this too has mixed support, and it is not strong enough to rebuild the empathy story on.

• Why some yawning species show contagion and others do not, and whether that maps onto anything social.
• What actually builds up before a spontaneous yawn, which remains the open question underneath all of this.

Every term the collection defines is gathered in the glossary.