The Glitch That Isn't

The sensation arrives without warning. A conversation in a kitchen, a stranger’s gesture at a bus stop, the angle of light against a wall — and suddenly the present moment seems to be playing a second time. The words about to be spoken feel pre-spoken. The room, never visited, feels revisited. For perhaps a second, sometimes two, the architecture of time appears to fold. Then the seam closes, the feeling evaporates, and life resumes its forward march.

Somewhere between sixty and eighty percent of people report having experienced this at least once.¹ It is among the most democratic of cognitive oddities — common across cultures, ages, and personality types — and one of the most stubbornly difficult to study. It cannot be summoned on demand. It refuses to sit still for inspection. By the time a subject reports it, the phenomenon is already gone.

For most of the modern era, déjà vu sat in a strange limbo: too universal to ignore, too elusive to investigate. It belonged to poets and parapsychologists more than to neurologists. Only in the last three decades has it begun to yield its secrets, and the answer that has emerged is stranger than the mystical readings it replaced. Déjà vu, it now appears, is not a malfunction at all. It may be the sound of the brain catching a mistake in itself.

A word for a feeling that had no name

The term was coined in 1876 by a French philosopher named Émile Boirac, writing in a letter to the Revue Philosophique. He called it la sensation du déjà vu — the sensation of the already-seen — and his framing has held for nearly 150 years.² Before Boirac, the experience drifted under various names: paramnesia, promnesia, the fausse reconnaissance of nineteenth-century French alienists. Romantic writers were drawn to it. Dickens described it in David Copperfield; Tolstoy turned it into a recurring motif. Walter Scott called it “the mysterious feeling which we all have experienced, of our consciousness recognising objects to which it cannot have been previously accustomed.”

Early theorists treated it as something more than a quirk of cognition. Spiritualists offered reincarnation: the soul remembered what the body had forgotten. Others suggested prophetic dreams — the present moment had been rehearsed in sleep and was now being recognized in waking life. Sigmund Freud, predictably, located it in the unconscious. In a 1901 essay he argued that déjà vu arose when a current experience resonated with a repressed memory or fantasy; the resemblance produced familiarity while repression blocked recall.³ It was an elegant theory and, like many of Freud’s, unfalsifiable.

For most of the twentieth century, déjà vu remained the province of philosophy and parapsychology. Mainstream neuroscience kept its distance. The phenomenon was too brief, too private, too unreliable. There were no neat behavioral measures, no animal models, no obvious clinical stakes. To study déjà vu was to chase smoke.

What eventually pulled it into the laboratory was an unexpected door: epilepsy.

The temporal lobe opens a window

Neurologists treating patients with temporal lobe epilepsy began noticing a pattern in the early aura preceding seizures. Just before a fit, patients sometimes reported an overwhelming sense of familiarity — the conviction that the present scene had been lived before. The same patients, between seizures, also reported far more spontaneous déjà vu than the general population.⁴ Something in the seizing tissue, it seemed, was producing the feeling.

The temporal lobes are the brain’s memory district. They sit roughly behind the ears, and folded into their inner surfaces are the hippocampi — twin seahorse-shaped structures essential to encoding new experiences — and, adjacent to them, the perirhinal and entorhinal cortices, sometimes grouped under the name rhinal cortex. The rhinal cortex is, among other things, the brain’s familiarity detector: the system that decides whether something has been encountered before, distinct from the system that remembers the encounter itself.

In the 1990s, the Montreal neurologist Pierre Gloor exploited a clinical procedure to test the connection directly. Patients being evaluated for epilepsy surgery sometimes had electrodes implanted deep in their temporal lobes; while they were awake, surgeons could deliver gentle electrical pulses and ask what they felt. Gloor and his colleagues found that stimulating the rhinal cortex reliably produced experiential phenomena: sudden, vivid floods of familiarity, dreamlike scenes, the conviction that an unfamiliar face was an old acquaintance.⁵ When the hippocampus alone was stimulated, patients sometimes reported memory-like imagery. When the rhinal cortex was the target, they reported the feeling that the present moment had been lived before.

It was the first hard evidence that familiarity and recall are dissociable — that they live in different tissue and can be switched on independently. And it suggested a mechanical account of déjà vu. If the familiarity system fires while the recall system stays quiet, the result is a memory signal without a memory: the sense of having been here, with no “here” to attach it to.

Two systems, one experience

Cognitive psychologists had been edging toward the same conclusion from a different direction. By the 1980s, the field had largely accepted a dual-process model of recognition memory: a fast, automatic familiarity signal and a slower, more effortful recollection of context.⁶ You recognize a face on the street as familiar in a fraction of a second; placing the face — coworker’s brother, met at a party last March — takes longer and engages different machinery. The two processes usually run together, and we rarely notice the seam.

Déjà vu, on this account, is what happens when the seam tears. The familiarity detector flags the present moment as known. The recall system, asked to confirm, finds nothing. The mismatch registers as that distinct, vertiginous shiver: the certainty that this has happened before, paired with the equally firm certainty that it hasn’t.

This fits why déjà vu so often arises in mundane settings — a sentence at dinner, a stretch of pavement — rather than the high-stakes moments memory normally privileges. Familiarity signals can be noisy. They can be triggered by a partial overlap between the current scene and any number of fragments scattered through experience: a similar room, a similar tone of voice, a similar arrangement of objects. Most of the time, the brain resolves these near-matches without consciousness ever getting involved. Occasionally, a scene assembles just enough partial overlaps to push the familiarity signal across threshold without producing any conscious memory to pin it to. The recall system, finding no record, sounds an alarm.

The question was whether this story could be tested outside the operating room.

Manufacturing the feeling

In the mid-2010s, a cognitive neuroscientist at the University of St Andrews named Akira O’Connor decided to try. The challenge was straightforward and difficult: produce déjà vu in healthy volunteers, reliably enough to study, without relying on surgery or seizure. O’Connor turned to a tool psychologists had been using for decades to study false memory — the Deese–Roediger–McDermott paradigm, or DRM.⁷

The DRM works like this. A subject is read a list of words clustered around a theme they are never told: bed, pillow, dream, night, tired, blanket, snooze, wake, doze, rest. The theme word itself — sleep — never appears. Asked later whether sleep was on the list, most people confidently say yes. The brain has stitched together the gist of the cluster and produced a false memory of the central word, complete with the warm confidence of having heard it.

O’Connor’s modification was a clever piece of misdirection. After hearing the cluster, volunteers were first asked whether any of the words they had just heard began with the letter S. They had not heard any. So when sleep was offered to them moments later, they were caught in a strange position: the word felt unmistakably familiar — the gist machinery had already done its work — but they had also just confirmed, accurately, that they had heard no S-words. The result, in a substantial fraction of subjects, was a reported sensation closely matching déjà vu: familiarity without traceable memory, accompanied by an active sense that something was off.⁸

O’Connor then ran the procedure inside a functional MRI scanner. The prediction, drawn from the temporal lobe literature, was that the hippocampus and rhinal cortex would light up — that déjà vu would prove to be a memory event. The data said otherwise. The regions that activated during induced déjà vu were not in the memory system at all. They were in the frontal lobes: the dorsolateral prefrontal cortex, the anterior cingulate, the brain’s conflict-monitoring and decision-checking apparatus.⁸

This was a quiet reversal of the standard reading. Déjà vu, the scans suggested, was not the brain mistakenly producing a memory. It was the brain flagging a mistake — running a quality-control check, noticing that the familiarity signal and the recall record were out of sync, and registering the conflict as a feeling.

The fact-checker speaks

In the years since, O’Connor and others have refined the picture. A complementary 2018 study from the same group found that older adults, who report less déjà vu, showed weaker engagement of these prefrontal monitoring regions during the DRM task — even though they were just as susceptible to the underlying false-familiarity signal.⁹ In other words, the elderly brain was not necessarily generating fewer faulty familiarity signals; it was simply less inclined to catch them. The familiarity went through, untagged.

This inverts the intuitive reading of déjà vu as a glitch. A glitch, by definition, is a system failing. What the imaging suggests is closer to the opposite: déjà vu is a system succeeding — succeeding loudly enough that consciousness gets pulled in. The conflict between two memory subsystems is detected, escalated, and presented to awareness as a strange feeling worth attending to. The eerie quality is not a bug. It is the alarm.

The pattern fits other features of the phenomenon. Déjà vu is most common in adolescents and young adults, and declines steadily through middle age.¹⁰ The prefrontal cortex, the seat of the monitoring machinery, peaks in functional maturity in late adolescence and early adulthood and slowly loses efficiency in later decades. Sleep deprivation, stress, and travel — all known triggers — are also known to perturb both the familiarity system, by exposing the brain to novel stimuli that partially overlap with stored experience, and the prefrontal system, which becomes less reliable under fatigue. New environments are exactly the conditions under which a finely tuned familiarity detector would be most likely to generate noise, and under which a vigilant fact-checker would be most likely to catch it.

This also helps explain a more pathological cousin of the phenomenon. In rare cases — most often after temporal lobe damage, severe psychosis, or some dementias — patients develop persistent déjà vu, a chronic sense that everything is being relived.¹¹ One famous case, described by neuropsychologist Chris Moulin in the 2000s, involved a man with dementia who refused to watch television because he insisted he had already seen every program, and who declined to visit his doctor because he was certain the appointment had already happened. Here the alarm has either become stuck in the on position or, more plausibly, has stopped distinguishing genuine familiarity from spurious familiarity. The monitoring system is not catching errors; it is generating them, or accepting them all. The result is not a fleeting shiver but a continuous, exhausting hallucination of repetition.

In its healthy form, déjà vu does the opposite job. It is brief precisely because the check completes. The frontal lobes notice the mismatch, flag it, resolve it — no, you have not been here before — and the feeling dissolves. The whole episode takes a second or two because that is roughly how long a competent error-correction process takes to run.

What it means to feel a process

There is a philosophical undertow to all of this that the experimental literature tends to leave alone. Most of what the brain does, it does silently. Conflict-monitoring circuits run constantly, comparing predictions against evidence, flagging discrepancies, adjusting models. We are rarely aware of them. We do not consciously feel the moment our visual system corrects for the blind spot, or the moment our motor system rebalances after a stumble, or the moment our auditory cortex decides which of two overlapping voices to attend to.

Déjà vu is one of the rare moments when one of these low-level processes pushes through the floor of consciousness and is felt directly. It is, in a sense, a glimpse of the brain’s housekeeping — a thin slice of cognition that is normally invisible, made briefly available because the discrepancy it is resolving happens to involve a high-salience signal (familiarity) and a high-salience absence (the missing memory).

That may be why the experience is so unsettling. The feeling is not, strictly speaking, of the past. It is the feeling of two systems disagreeing about whether there is a past to feel. Consciousness, accustomed to receiving the polished output of memory rather than its draft, is briefly handed the unfinished work. The result has the structure of a paradox — I remember this, but I do not remember it — and the texture of vertigo.

It also, in a small way, complicates the popular metaphor of memory as recording. A recording either contains a moment or it does not. The brain’s memory systems are more like overlapping committees, each casting a vote about whether a given scene has been seen before, and each capable of being wrong on its own. Most of the time the votes align and we feel nothing about the process. When they diverge, we feel the divergence, briefly, before it is reconciled.

The shiver, reconsidered

It is tempting, having traced the circuitry, to feel that the mystery has been disenchanted. The opposite seems closer to the truth. Déjà vu is not less interesting for being a conflict-monitoring artifact; it is more interesting. The experience that for centuries was attributed to past lives, prophetic dreams, and repressed unconscious material turns out to be a small, lawful, decipherable event in a working brain — a moment when the machinery of self-correction becomes audible to its owner.

What it is not is evidence of a fault. The people who report déjà vu most often are not the people with the worst memories. They are, broadly, the people with the most active monitoring systems: younger, more educated, more frequently exposed to novelty, more attentive to their own mental states.¹⁰ In several surveys, déjà vu correlates positively with travel, with higher income, with liberal political orientation — variables that share little except a tendency to put the brain into unfamiliar environments and ask it to make sense of them quickly. The experience is, in this sense, a marker of cognitive engagement rather than its decay.

This is also why the feeling fades so cleanly. A genuine pathology of memory does not resolve in a second and a half. A working brain catching a near-miss does. The shiver is the closing of the audit, not the opening of a wound.

Next time the room tilts — the half-second when the conversation seems to be replaying itself, when the angle of a doorway feels memorized — it may be worth holding the moment a beat longer than usual. Not because there is anything to remember. There isn’t. But because what is happening is, in its quiet way, remarkable: a system designed to keep its own records honest has noticed an inconsistency and surfaced it to the only part of itself that can be informed. The feeling is the report. By the time it has finished arriving, the discrepancy has already been resolved, and the brain has gone back to its silent work. For one strange second, it allowed itself to be overheard.

Sources

1. Brown, A.S., ‘A Review of the Déjà Vu Experience,’ Psychological Bulletin, 2003. — https://pubmed.ncbi.nlm.nih.gov/12848217/
2. Boirac, É., ‘Correspondance,’ Revue Philosophique de la France et de l’Étranger, 1876. — https://en.wikipedia.org/wiki/%C3%89mile_Boirac
3. Freud, S., ‘The Psychopathology of Everyday Life,’ 1901 (English edition Norton, 1966). — https://www.sas.upenn.edu/~cavitch/pdf-library/Freud_Psychopathology.pdf
4. Bancaud, J. et al., ‘Anatomical origin of déjà vu and vivid memories in human temporal lobe epilepsy,’ Brain, 1994. — https://pubmed.ncbi.nlm.nih.gov/8004991/
5. Gloor, P., ‘Experiential phenomena of temporal lobe epilepsy: Facts and hypotheses,’ Brain, 1990. — https://academic.oup.com/brain/article-abstract/113/6/1673/253994
6. Yonelinas, A.P., ‘The nature of recollection and familiarity: A review of 30 years of research,’ Journal of Memory and Language, 2002. — https://www.sciencedirect.com/science/article/abs/pii/S0749596X02928642
7. Roediger, H.L. & McDermott, K.B., ‘Creating false memories: Remembering words not presented in lists,’ Journal of Experimental Psychology, 1995. — https://psycnet.apa.org/record/1996-10428-001
8. Urquhart, J.A. & O’Connor, A.R., ‘The awareness of novelty for strangely familiar words: a laboratory analogue of the déjà vu experience,’ PeerJ, 2014; and O’Connor, A.R. presentation, International Conference on Memory, 2016. — https://peerj.com/articles/666/
9. Urquhart, J.A., Sivakumaran, M.H., Macfarlane, J.A. & O’Connor, A.R., ‘fMRI evidence supporting the role of memory conflict in the déjà vu experience,’ Memory, 2018. — https://www.tandfonline.com/doi/full/10.1080/09658211.2018.1524496
10. Adachi, N. et al., ‘Demographic and psychological features of déjà vu experiences in a nonclinical Japanese population,’ Journal of Nervous and Mental Disease, 2003. — https://pubmed.ncbi.nlm.nih.gov/12826918/
11. Moulin, C.J.A. et al., ‘Disordered memory awareness: Recollective confabulation in two cases of persistent déjà vecu,’ Neuropsychologia, 2005. — https://pubmed.ncbi.nlm.nih.gov/15949516/