Near-Death Experience: What the Studies Found

In an intensive care unit in Michigan, a ventilator is switched off. The patient's brain has no prospect of recovery after cardiac arrest. The electrodes on his scalp keep running. What they record over the next few minutes was published in 2023 — and it is the reason two camps that have argued about near-death experiences for four decades now point at the same curve.

Ann Arbor: the last EEG

The neurological ICU at the University of Michigan has recorded continuous EEG in comatose patients since 2014. That is routine, not research: the point is to catch seizures that are invisible from the outside. For four of those patients the recording became something else. They died with the electrodes still attached — three after cardiac arrest with severe oxygen deprivation, one after an extensive brain hemorrhage. All four were deeply comatose and unresponsive. In all four cases the family had decided to withdraw life support.

In one patient the ventilator was switched off and the external pacemaker 273 seconds later. The team around neurologist Jimo Borjigin, which analyzed the traces afterward, broke the dying process into segments that show when the heart lost its rhythm and when the brain fell silent. In between lies the finding: in two of the four patients, activity in the gamma band — the fast brain waves above 25 hertz associated in waking people with attention, memory and conscious perception — climbed steeply at the moment of oxygen loss. In individual regions the increase ran from twofold to 391-fold over baseline. And it was not scattered randomly across the head. It concentrated at the back, at the junction of the temporal, parietal and occipital lobes, in a zone consciousness research has for some years called the hot zone, because in healthy people activity there tracks experienced content most closely, awake and dreaming alike.

The authors wrote this up in 2023 in the Proceedings of the National Academy of Sciences. They wrote something else that did not make the headlines: none of the four patients survived. Nobody could say whether any of them experienced anything during those minutes. For that reason, the authors said in as many words, it remains possible that the surge is merely an epiphenomenon, or pathological.

That sentence is the hinge of this whole article. The same curve has been cited from both sides ever since. One side says: there, the dying brain fires one last time, and that is where the light, the tunnel and the life review come from. The other says: there, the brain is capable of organized activity while dying — so what if it is receiving something rather than producing it? Both have read the curve. Neither can claim it. To see why, you have to go back forty years.

What "cardiac arrest" means for the brain

Cardiac arrest is the only condition in which a near-death experience can be studied cleanly. In a car crash, a hemorrhaging birth or a near-drowning, nobody knows exactly how badly supplied the brain was. In cardiac arrest you do know: blood flow to the brain collapses immediately. Dutch cardiologist Pim van Lommel, who returns below, summarizes the older measurements this way — in cardiac arrests recorded during surgery or while testing an implanted defibrillator, the EEG went flat after about fifteen seconds on average and stayed flat, even under chest compressions, until the rhythm came back with a shock. Resuscitation from outside keeps tissue alive for a while, but it does not restore measurable brain function. That is what the textbooks said for decades.

That is the ground the whole dispute stands on. If the brain falls silent after fifteen seconds, then nothing can be experienced and nothing remembered in the minutes that follow. And yet people pulled back from exactly that state have for decades reported something they say they lived through.

Intensive care physician Sam Parnia, whose studies form the core of this article, pushed through a second shift you need to know about: death is a process, not a moment. In a 2022 consensus statement in the Annals of the New York Academy of Sciences, written with a group of researchers from neurology, emergency medicine and psychology, the point is made that scientific study of dying only became possible once it was recognized that brain cells are more resistant to oxygen deprivation than assumed — they are irreversibly damaged over hours to days after cardiac arrest, not seconds. When a resuscitation attempt succeeds, we call the state cardiac arrest. When it fails, we call the same state death. The difference is not in the body. It is in the outcome.

Why this is a large American question

The American Heart Association puts out-of-hospital cardiac arrests in the United States at well over 300,000 a year, with in-hospital arrests adding six figures more. Survival to discharge from out-of-hospital arrest runs in the single digits to low teens depending on region and bystander CPR rates. Even at the low end, that leaves tens of thousands of American survivors every year — which is why the question is not academic here. Someone in your zip code came back last month.

The vocabulary is American too. Raymond Moody, a philosopher then in Virginia, coined the term near-death experience in his 1975 book and gave the phenomenon a name that stuck. Psychiatrist Bruce Greyson built the measuring instrument at the University of Virginia, where a research unit has studied these reports since the 1960s. The International Association for Near-Death Studies, founded in 1981 and long headquartered in Durham, North Carolina, is where most American experiencers first find other people who describe the same thing.

1988 to 2001: the Dutch count

Until the late 1980s, near-death research consisted of collected stories. People contacted researchers years after the event because they had experienced something. Anyone who did not come forward appeared in no statistic. The only way out of that trap was to count the other way round: not gather the experiences, but gather the cardiac arrests — and then ask every survivor whether they remembered anything.

That is what began in 1988 in ten Dutch hospitals. Cardiologist Pim van Lommel and colleagues enrolled all consecutive patients who survived a cardiac arrest, 344 people, and interviewed them within days. The result appeared in December 2001 in The Lancet: 282 patients, 82 percent, remembered nothing. 62 patients, 18 percent, reported a memory from the time of the arrest. 41 of those, 12 percent, described what the researchers called a core experience: knowing they were dead, a feeling of peace, leaving the body, the tunnel, the light, the dead, a life review, a border.

The real finding, though, was in what did not correlate. The researchers expected the experience to become more frequent the longer the brain went without oxygen. It did not. Neither the duration of the arrest nor the duration of unconsciousness, neither medication nor prior fear of death had any influence on whether someone reported a near-death experience. The last sentence of the abstract is therefore an open challenge: if a purely physiological explanation such as oxygen deprivation were correct, most patients who had been clinically dead ought to report such an experience. Only a few did.

The reply was in the same issue. British psychologist Christopher French published a commentary asking whether these were visions of a dying brain or false memories. His objection remains the strongest the other side has: nobody knows when a memory formed. It could come from the seconds before the arrest, from the minutes after the heartbeat returned, or from the days afterward, in which the brain assembles fragments into a story. A memory of the time of arrest is not evidence that it was formed during it.

Both van Lommel and French worked with the same instrument: a scale psychiatrist Bruce Greyson developed in 1983 at the University of Virginia. From 80 features that recurred in reports, Greyson filtered 16 that can be asked reliably — from whether time seemed to speed up or slow down, to whether the person came to a border they were not allowed to cross. Each answer scores zero, one or two points; 32 is the maximum; above a threshold a report counts as a near-death experience. The scale has been the field's yardstick for forty years. It does not say what the experience is. It says only that two researchers are talking about the same thing.

2008 to 2014: a thousand shelves

Van Lommel's study counted. The next one wanted to test. If people report looking down at their own bodies from the ceiling, that ought to be testable — with something visible only from the ceiling.

In 2008 Sam Parnia, then at the University of Southampton, began the AWARE study — AWAreness during REsuscitation. Fifteen hospitals in the United Kingdom, the United States and Austria took part. In each, between fifty and a hundred shelves were mounted high on the walls of emergency departments and intensive care units, where cardiac arrests were most likely. On each shelf lay a picture face up: symbols, people, animals, newspaper headlines. Nothing was visible from the floor. Anyone who had genuinely floated near the ceiling would be able to name the image.

Over four years the study recorded 2,060 cardiac arrests. 330 people survived. 140 were well enough to be interviewed. 55 of them said they remembered something from the time of unconsciousness. When the researchers reported in Resuscitation in 2014, the distribution looked like this: 46 percent of those with memories described things that did not fit the near-death pattern at all — fear, animals and plants, a bright light, being pursued, violence, déjà vu, family, and events that had obviously happened after resuscitation. Nine percent had an experience that scored as a near-death experience on the Greyson scale. Two percent, two people, described a near-death experience in which they claimed to have seen or heard actual events in the room.

One of the two was a 57-year-old man. He described hearing an automated voice say "shock the patient" twice, then looking down from above at himself, a nurse, and a heavyset man in a blue cap with no hair. The medical record confirmed that an automated defibrillator with voice prompts was in use, and the man in the blue cap existed. The researchers calculated that the perception described must have occurred up to three minutes after the heart stopped — in a window when, by the textbook, no brain is working.

And then the number the whole arrangement broke on: 78 percent of the cardiac arrests had happened in rooms with no shelf. The two men with the detailed descriptions were in such rooms. A thousand pictures, four years, and not one patient who could have seen one. Parnia said on publication that it had not been possible to prove the reality of these perceptions — but equally impossible to disprove them.

2017 to 2023: AWARE-II, with the electrodes running

The second study drew the obvious conclusion. Instead of hanging pictures on walls and hoping the arrest happened underneath them, the test would come to the patient: a research team was paged on every alarm and brought a tablet that displayed images facing upward during resuscitation, plus headphones through which words were played at intervals for survivors to be asked about later. Added to that was the thing that separates this study from everything before it: electrodes on the forehead recording EEG and cerebral oxygen saturation continuously during chest compressions. For the first time, researchers were not only asking what someone experienced but simultaneously measuring what their brain was doing at that moment.

Between May 2017 and March 2020, 25 hospitals in the United States and the United Kingdom recorded 567 cardiac arrests. The numbers published in Resuscitation in 2023 show first of all how brutal the baseline is: 53 patients survived, 9.3 percent. 28 could be interviewed. Eleven of them reported memories or perceptions suggesting awareness during the arrest. The researchers sorted them into four groups: emergence from coma during the chest compressions themselves, in two patients; emergence in the period afterward, also two; dreamlike experiences, three; and finally what the team now calls a recalled experience of death, in six of 28 interviewed — more than one in five. A parallel survey of 126 survivors outside the hospital setting added a fifth group: delusional states in which medical events were misinterpreted.

The picture test failed again, this time on mortality: nobody identified the image on the tablet. One of 28 named the sound from the headphones. With 53 survivors, you cannot say whether that was chance.

The measurement, on the other hand, produced something nobody had expected. At a mean cerebral oxygen saturation of 43 percent, far below normal, wave patterns normally associated with consciousness — delta, theta, alpha — appeared in some patients during the pauses in chest compressions, and did so up to 35 to 60 minutes after resuscitation began. The authors interpret this cautiously: the emergence of these patterns could mean that networks in the brain are reorganizing to a level at which mental processes are possible, and it could be a biomarker of lucidity and of the recalled experience of death. It is a could in both sentences. But it is the first measurement that calls the textbook flat line after fifteen seconds into question, at least for the resuscitation phase.

The gamma wave: rats, an 87-year-old, and the four from Michigan

Running alongside Parnia's interviews was a second line of work coming from the other direction — not from survivors, but from the dying. In 2013 Jimo Borjigin and her team at the University of Michigan published a study in the Proceedings of the National Academy of Sciences on rats in which cardiac arrest was induced under anesthesia. The textbook expected a brain that falls silent. Instead the EEG showed, within the first thirty seconds after arrest, a brief highly synchronous wave in the gamma band that spread across the whole brain and whose strength exceeded that of the waking state. The mammalian brain, the authors wrote, can paradoxically generate the neural correlates of heightened conscious processing while dying.

Whether that holds for humans first showed up by accident. An 87-year-old patient in Canada who had suffered a brain hemorrhage after a fall was being monitored on EEG for seizures when his heart stopped. The recording kept running. In 2022 Raul Vicente and colleagues analyzed the 900 seconds around the arrest in Frontiers in Aging Neuroscience and found what Borjigin had seen in rats: a rise in the gamma band, coupled to slower rhythms, continuing after blood flow ended. The researchers noted explicitly that injury and swelling of the brain could have influenced the data — it was a single patient, recorded in an emergency, not a study.

Then came the four from Ann Arbor this article began with. Borjigin's team now had humans, high-resolution recordings and a clear time point: the withdrawal of ventilation. In two of four patients, the gamma surge, the coupling to slow waves, the connectivity between hemispheres, the hot zone. In the other two, none of it. The authors checked whether hidden seizures could explain it and found no evidence, though they could not rule it out for deep brain regions. And they wrote the sentence that lands on both camps at once: activation of the hot zone suggests heightened conscious processing but does not prove it — because none of the patients survived and nobody can say whether they experienced anything.

Why the same curve serves both camps

Now the puzzle can be stated precisely. There are two kinds of data, and they come from two groups of people who do not overlap.

One group survived. From them we have the reports: van Lommel's 62 of 344, Parnia's nine percent in AWARE, the six of 28 in AWARE-II. For these people there is almost no measurement from the moment itself — in AWARE-II, for the first time, a forehead EEG during chest compressions, but with so few survivors that curve and report could hardly ever be matched to each other.

The other group died. From them we have the curves: Borjigin's rats, the 87-year-old, the two of four in Michigan. For these people there is no report and never will be.

If you hold that the near-death experience is a product of the dying brain, you read the gamma wave as the cause: the brain fires while it suffocates, and that firing is the light. That is a coherent reading. It has one problem, which van Lommel stated back in 2001 — if oxygen deprivation is the cause, why do only a few report it rather than most, and why does the frequency not depend on the duration of the deprivation? Borjigin's data offer no answer; they sharpen the question. Two of four showed the wave, two did not. Why?

If you believe instead that consciousness is not produced by the brain but only mediated through it — van Lommel himself holds this position and calls it nonlocal consciousness — you read the same wave as a receiving signal: the brain is capable of organized activity while dying, so it could take something in at that moment. That is also a reading, and it also has a problem. It is compatible with the data, but the data do not require it. Nothing about a gamma wave distinguishes transmitting from receiving.

And over both of them lies French's objection from 2001, which no measurement has made smaller: even if we knew the brain was active during the arrest, we would not know the memory formed in that moment. Someone interviewed three days later has had three days to build a story out of fragments.

That is where things stand. It is unsatisfying, and it is honest. The one person who would unite both data sets — someone whose brain was recorded in high resolution during arrest, who can be asked about a concealed image, and who names that image — has not existed in 35 years of research. Not because nobody looked. Because nine out of ten people you could measure it on do not come back.

A molecule as a model: what DMT research contributes, and what it does not

There is a detour the field has been taking for some years, and it runs through pharmacology. Since the 1990s it has been discussed whether the endogenous tryptamine N,N-dimethyltryptamine, DMT, might be involved in near-death experiences — a suspicion resting mainly on similarities between reports. In 2018 a group at Imperial College London around Christopher Timmermann and Robin Carhart-Harris measured that similarity for the first time: thirteen healthy volunteers received DMT under clinical supervision in a placebo-controlled design and afterward filled in the Greyson scale, the same instrument used for near-death experiences. Scores after DMT were markedly above placebo, and when compared with a group of actual near-death experiencers, nearly all features overlapped.

What that shows is narrow and specific: the state DMT produces in a laboratory resembles the near-death experience closely enough to be studied as a model for it. What it does not show: that DMT causes the near-death experience in the dying brain. For that you would have to demonstrate DMT in the human brain at the moment of death in relevant quantity, and that has not been done. The overlap in reports could equally mean that two entirely different triggers open the same door in the brain. Timmermann's study is a research tool, not an explanation — and certainly not a route anyone should take themselves to look death in the face.

The experiment it would take

Having read this far, you can say where the problem lies, and that is more than most articles on the subject manage. Three things would have to come together that never have.

First, a test stimulus that is actually there. AWARE tried that with a thousand shelves and failed on geography; AWARE-II brought the tablet to the patient and failed on mortality. The next step is obvious: images and sounds that run automatically during every resuscitation in every room, not only where a research team arrives in time.

Second, a measurement that gets close enough to the brain. The forehead EEG in AWARE-II sees a fraction of what the whole-head recordings in Michigan saw. A device that does both — electrodes applied in seconds, high-resolution, artifact-free under chest compressions — does not exist. That would be the actual invention.

Third, numbers. With a survival rate under ten percent and a rate of roughly twenty percent recalled experiences of death among those interviewed, you need thousands of cardiac arrests to find a few dozen people for whom curve, stimulus and report are all present. Parnia's 2022 consensus statement mapped exactly this path: common terms, common measurements, many hospitals. Whether it gets walked depends on money and on hospitals' willingness to permit research in the worst minute of their patients' lives.

And even then French's objection stands. A patient who names the concealed image would have shown that they perceived during a time when they should not have been able to. They would not have shown what consciousness is. But they would have changed the textbooks. Measured against forty years of argument, that is not nothing.

The timeline

What the four data sets carry — and where interpretation begins

Claim	Source	Status
About one in five survivors of cardiac arrest reports a near-death or recalled death experience.	Van Lommel 2001 (18% memory, 12% core); Parnia 2014 (9%); Parnia 2023 (6 of 28)	Verified, 9–21% depending on definition
Frequency does not depend on the duration of arrest or unconsciousness.	Van Lommel 2001	Verified for that cohort
EEG patterns associated with consciousness are measurable up to an hour into chest compressions.	Parnia 2023 (AWARE-II)	Verified for a subset of patients
The dying brain can generate a wave of synchronous gamma activity.	Borjigin 2013 (rats); Vicente 2022 (1 patient); Xu 2023 (2 of 4)	Verified in humans in individual cases
One patient perceived events in the room up to three minutes after arrest.	Parnia 2014, the 57-year-old, time-confirmed via the defibrillator	Documented single case, not proof
A survivor has named a concealed image.	AWARE 2014, AWARE-II 2023	Not proven — nobody, in either study
The gamma wave produces the near-death experience.	Interpretation; Xu 2023 calls the surge possibly epiphenomenal or pathological	Not proven
Consciousness outlasts the death of the brain.	Interpretation (van Lommel, nonlocal consciousness)	Not proven

What this means for you — and what it does not

Maybe you are reading this because you experienced something yourself that you tell nobody about. Maybe because someone you love died, and you want to know whether anything was there. Then you should know what these studies can honestly give you, and what they cannot.

They give you this: you are not alone. Up to one in five people who come back from cardiac arrest brings something with them, and the reports resemble each other across countries, languages and decades closely enough that researchers can measure them with the same scale. What you experienced is not a sign that something is wrong with you. It is ordinary enough to show up in the statistics.

They also give you this: dying does not look from the inside the way it looks from the outside. The recalled experience of death, as Parnia's team worked it out of the reports, is not fear but distance — observing without pain or distress, and a look at one's own life that examines one's own actions and intentions toward others. Anyone sitting at a dying person's bedside is allowed to know that.

They do not give you an answer to whether anything continues. Nobody wrote this article to promise you that or to take it from you. And they do not give you a way to try it yourself — not with substances, not with techniques. What Timmermann's thirteen volunteers experienced under medical supervision was research on a model, not a recipe. If you want to read further in that direction, our pieces on astral projection for beginners and whether astral projection is dangerous take the same evidence-first approach to a related claim.

What remains is the one finding that recurs in every study and can be taken seriously without a cardiac arrest: the life review. People who come back report having seen their lives — not as a film, but as an examination of what they did to others and what they meant by it. Van Lommel's long-term follow-up at two and eight years found these people had changed, and that the change came not from the cardiac arrest but from the experience. You do not have to die to put that question to yourself. You can put it to yourself every evening.

What we do not promise

We are not telling you that consciousness survives death. No study has shown it, and none was built in a way that could have. We are not telling you the opposite either. Anyone selling you certainty in either direction has not read the studies.

We are not telling you that a near-death experience is a sign of anything about you — spiritual advancement, chosenness, or illness. And nothing here is medical advice. If you are living with the aftermath of a cardiac arrest, or with grief, the right support is a clinician or a counselor, not an article. Our piece on what to do with a home after a death is about the practical and ritual side of grief, and it makes no claims either.

Frequently asked questions

Sources and further reading

Pim van Lommel et al., "Near-death experience in survivors of cardiac arrest: a prospective study in the Netherlands," The Lancet 358 (2001) · Christopher French, "Dying to know the truth: visions of a dying brain, or false memories?", The Lancet 358 (2001) · Bruce Greyson, "The Near-Death Experience Scale: construction, reliability, and validity," Journal of Nervous and Mental Disease 171 (1983) · Sam Parnia et al., "AWARE — AWAreness during REsuscitation: a prospective study," Resuscitation 85 (2014) · Sam Parnia et al., "AWAreness during REsuscitation II," Resuscitation 191 (2023) · Sam Parnia et al., "Guidelines and standards for the study of death and recalled experiences of death," Annals of the New York Academy of Sciences (2022) · Jimo Borjigin et al., "Surge of neurophysiological coherence and connectivity in the dying brain," PNAS 110 (2013) · Gang Xu, Jimo Borjigin et al., "Surge of neurophysiological coupling and connectivity of gamma oscillations in the dying human brain," PNAS 120 (2023) · Raul Vicente et al., "Enhanced interplay of neuronal coherence and coupling in the dying human brain," Frontiers in Aging Neuroscience 14 (2022) · Christopher Timmermann et al., "DMT models the near-death experience," Frontiers in Psychology 9 (2018) · Raymond Moody, Life After Life (1975).