Earworms

Why does a fragment of a song repeat in your head for hours when you never chose to play it?

You did not decide to hear it. Some fragment — eight bars, a hook, the two words before the chorus — started up while you were rinsing a plate, and it has been going round since. Notice what is strange about that. Most of memory has to be summoned; this arrived unbidden, and it will not leave when asked. Stranger still, it is almost never the whole song. It is a fragment, and it plays to the same place each time and starts again.

So what would a mind have to be doing for that to happen — and why a fragment rather than the piece?

Begin with what earworms are made of. Surveys and diary studies agree on the shape: most people report them at least weekly, they last minutes to hours, and they are typically short excerpts — a chorus or hook rather than a verse-to-verse whole. They are also overwhelmingly music you have heard recently, or heard a great many times. That last point is dull but it is the strongest predictor there is, and any account has to start with it.

Now ask what kind of tune gets caught. If exposure were the whole story, the songs looping in people's heads would simply be the songs on the radio in proportion to airplay. They are not quite: some tunes are named far more often than their exposure would predict. When researchers have compared frequently named earworm tunes against matched songs of similar popularity, the recurring findings are faster tempo, a common and easily sung melodic contour — and then some small local oddity, an unexpected interval leap or a rhythm that does not sit where you expect. Familiar enough to reproduce without effort; irregular enough that reproducing it is not quite trivial.

Does that combination suggest a mechanism? Consider what your mind does with a short sound sequence it wants to hold: it rehearses it, in the same inner-voice loop you use to keep a phone number alive for the ten seconds before you dial. That loop is built to repeat. It has no natural stopping rule — it stops when something takes it over, or when the material it is holding gets discharged.

Which raises the interesting possibility: what if the fragment does not get discharged? A tune has closure built into it — a cadence, a return to the tonic, an ending your ear can feel. If what is looping is a piece of a tune, it never reaches that resolution, so it arrives back at the start still open. This is the Zeigarnik-style account, borrowed from the older finding that interrupted work is remembered better than finished work — the same effect this collection treats under unfinished tasks. It has real support here: people who hear part of a song loop it more than people who hear all of it, and deliberately playing the tune through to its end is one of the few things that reliably helps.

There is a matching prediction on the other side. If the loop is the inner voice, then occupying that voice should evict the tune — and it does. Chewing gum, in a controlled experiment, measurably reduced earworm episodes, presumably by engaging the same articulatory machinery. Not a cure; a competing tenant.

Last piece: when do they surface? In low-load moments — walking, showering, driving a familiar route — and after a cue: the word, the place, the mood attached to the song. That fits a mind that is always partly wandering, and a fragment that is easiest to notice when nothing else is asking for the same channel.

Three refinements hold the steps together.

First, exposure and structure are not rival explanations; they multiply. Recent or repeated hearing decides which small set of tunes is even available to loop, and the tune's shape decides which of those actually catch. This is why the same handful of songs recur across a population — shared exposure, common features.

Second, "unfinished" is doing a lot of work and should be read narrowly. It does not mean your mind is troubled by an unresolved thing; it means a fragment lacks the ending that would let rehearsal discharge it. That is a claim about the material, not about your state of mind.

Third, and least comfortable: this literature is largely survey, diary, and self-report. Earworms are involuntary by definition, which makes them hard to induce and impossible to observe directly — what gets measured is a person's report of an inner experience, often recalled later. There are good experiments inside it, but much of the field is correlational, sample sizes are modest, and results about which musical features predict earworms rest on relatively few studies. Treat the mechanism above as the best available account rather than a settled finding. The one part that is not in doubt is the exposure effect, and that is the least interesting part.

• Why the phenomenon is so much rarer in people who report little or no voluntary musical imagery.
• Whether the same open-loop account explains intrusive phrases and slogans, which are not musical at all.

Every term the collection defines is gathered in the glossary.