Deja vu

Why can a place you are certain you have never visited feel overwhelmingly familiar anyway?

You step into a street in a city you have never been to, and it is drenched in familiarity. You know, with complete confidence, that you have never stood here. You also feel, with complete confidence, that you have.

Notice which half is the puzzle. A false sense of familiarity is common and unremarkable — we mistake strangers for acquaintances constantly. What is odd about deja vu is that you know it is false while it is happening. Any account that explains only the feeling has explained the easy part. It has to explain the doubling.

Start with a fact that sounds technical and turns out to do all the work: recognition is not one process. It is at least two, and they can come apart.

The first is familiarity: a fast, graded signal that says "encountered before" and nothing else. It carries no content — no when, no where, no with whom. The second is recollection: the retrieval of the episode itself, its context and its associations. Normally they arrive together so smoothly that you would never guess they were separate systems.

But you have felt them separate. The stock case is meeting the butcher on a bus — a face soaked in familiarity while you cannot place it. Something told you that face was known before anything told you why. The dissociation also shows up in damage: hippocampal injury can badly impair the recollective side while leaving familiarity comparatively intact, and familiarity has been associated more with perirhinal and surrounding rhinal cortex than with the hippocampus proper.

Now put the street back in front of you. Suppose familiarity fires strongly — for whatever reason — and recollection, queried for a source, returns nothing at all. And suppose there is some monitoring process comparing the two. What should that feel like? Exactly what deja vu feels like: an overwhelming sense of having been here, an inability to say when, and a simultaneous certainty that the sense is wrong. The doubling is not a strange extra. It is the monitor doing its job.

That reframing immediately separates deja vu from something it is often lumped with. In a false memory, familiarity fires and recollection obligingly supplies a source — frequently the wrong one. No conflict is registered, so nothing feels odd, and you are confident and mistaken. Deja vu is the opposite configuration: the monitoring works, which is why the experience is merely eerie rather than dangerous.

So can the mechanism be produced deliberately? Two lines say yes. Deja vu occurs as an aura at the onset of temporal lobe seizures, and during pre-surgical electrode mapping, direct electrical stimulation of medial temporal structures — rhinal cortex prominently — has evoked the experience in patients. Stimulate the familiarity machinery, get the feeling. That is about as close to a causal demonstration as this subject has.

The second line is a laboratory analogue. Build a novel scene whose spatial configuration — the layout, not the objects — matches a scene the participant studied earlier and cannot now recall. Participants report more familiarity and more deja vu for those scenes than for unrelated novel ones, while remaining unable to name the source. That is the account assembled on purpose: a real basis for familiarity, a retrieval that fails, and the resulting conflict.

And it is falsifiable. If deja vu were simply misattributed arousal, or memory failing in a general way, then a configurally matched novel scene should produce no more of it than an unrelated one, and stimulation should evoke it as readily from arbitrary cortical sites as from rhinal cortex. Either observation would sink the familiarity-without-recollection account.

That last point is the misconception worth taking apart. The experience presses you toward "I must have been here and forgotten." But familiarity is not a lookup returning true or false; it is a strength, and a novel scene sharing enough structure with something in your past can drive it high with nothing having been forgotten. This is why the whole account can be true and there is still nothing in your history to find.

It is also worth marking the boundary against the other memory phenomena nearby. In prospective memory, a cue in the environment successfully pulls back an intention — the retrieval works, which is the point. With odour-triggered memory, a cue delivers a rich, content-laden recollection. Deja vu is the case where the cue delivers a signal with no content whatsoever, and its whole character comes from that emptiness.

Now the caveats, which here are substantial. Deja vu has many competing explanations and very little decisive evidence between them. Alongside the familiarity account there are timing accounts, in which one processing route lags another so a single perception is registered twice; attentional accounts, in which a brief inattentive glimpse is followed by full attention to a scene that really was seen a moment before; and accounts appealing to brief seizure-like temporal-lobe activity in healthy brains. These are not mutually exclusive and none is settled. It is also genuinely contested whether the laboratory analogue produces the same thing as the spontaneous experience. I am also deliberately not quoting prevalence figures: reported rates move around a great deal with how the question is worded.

One directional fact does discriminate a little. Deja vu is reported less often as people get older. That is awkward for any account treating it as simply memory failing, since memory failures go the other way — and it fits the reading that the experience requires a monitor healthy enough to notice a conflict.

• Jamais vu, the opposite failure, and whether it can be induced as reliably as it can be described.
• How recognition research separates familiarity from recollection experimentally, and where that split is disputed.

Every term the collection defines is gathered in the glossary.