Chapter 5

What Happens to the Brain at Death

Cardiac Arrest, Anoxia, and the Flat EEG — The Medical Baseline Against Which Veridical NDEs Must Be Read

The Morning Frank Collapsed

It is just after seven on a Tuesday morning, and a fifty-eight-year-old man named Frank—that is not his real name—has come to the emergency department complaining of a tight, heavy feeling in his chest that started while he was eating breakfast. He is talking to the triage nurse one moment. The next, his eyes roll back. He goes gray. His knees buckle under him. The nurse catches him as he falls.

What follows unfolds in seconds.

The cardiac monitor, slapped onto Frank's chest at lightning speed, shows ventricular fibrillation—a chaotic electrical storm in the heart that means the chambers are quivering instead of pumping.¹ No pulse. No blood pressure. No blood is moving. None.

Within the first ten seconds, Frank is unconscious. His brain, deprived of the steady river of oxygenated blood it needs to do anything at all, is going offline. By twenty seconds, the organized electrical signals at the surface of his cortex—the signals that, on every standard scientific model, generate his thoughts, his perceptions, his sense of self—have ceased. By thirty seconds, the EEG, if anyone had time to attach one, would be effectively flat. A line. Or close to it.²

The team is on him. Compressions begin. Two milligrams of epinephrine. Bag-valve-mask ventilation. Defibrillator pads. "Charging—clear—shock!" Frank's body jolts. Compressions resume. Another shock. Another round of epi.

Three minutes. Four. Six.

At about seven minutes, a rhythm returns. They get a pulse. Frank is in the ICU on a ventilator within the half-hour. Twenty hours later, he is awake and extubated and answering questions correctly. Forty hours later, he is sitting up in bed eating Jell-O and watching, of all things, a basketball game on the wall-mounted television.

And the next day—when his attending physician, a thoughtful woman who has read a little of the NDE research and is willing to ask, gently asks him, "Frank, do you remember anything from when you collapsed?"—Frank looks at her steadily, sets his spoon down, and says: "I remember everything. I watched you all working on me. I was up by the ceiling. I want to tell you what I saw."

I am going to ask you, in this chapter, to take that last paragraph seriously.

Not because we know yet whether Frank's report—what he says he saw—will hold up under careful checking. Maybe it will. Maybe pieces will not. Maybe there will be details he could not have known and details he might have heard from the team afterward and details that fall in some gray area. The question of how we sift veridical reports from non-veridical ones is the work of later chapters. We will get there.³

I am asking you to take the paragraph seriously because of what was happening to Frank's brain during those seven minutes.

Because here is the thing the cardiologists and the neurologists and the resuscitation researchers all agree on, and it is not even a slightly contested claim. During the period when Frank was in ventricular fibrillation with no measurable cardiac output, his brain was not producing organized conscious experience. It could not have been. The electrical activity that, on the standard physicalist account, is his conscious experience—that activity had ceased. The brain was offline. There is nothing to argue about there. Every textbook of clinical neurology says so. Every cardiac-arrest specialist says so. Every emergency physician you ask says so.⁴

And yet Frank says he remembers everything.

That is what this chapter has to set up. The medical baseline. The neuroscience of brain shutdown during cardiac arrest. The window during which, on every standard model, no conscious experience should be possible. Because if we cannot establish that baseline carefully and honestly, the rest of the book's argument has nowhere to stand. And if we can—if the brain really does go dark when the heart stops, in the way the medical literature says it does—then every veridical NDE that comes from that window of darkness becomes a problem. A serious one.

Stay with me. We are going to walk into the brain.

Two Stories at the Boundary

Two stories sit side by side at the boundary of death, and they appear to contradict each other.

The first story is the one neuroscience tells. It goes like this: the brain is a beautiful, hungry, fragile organ. It runs on a continuous stream of oxygen and glucose delivered by a continuous flow of blood. When that flow stops—for whatever reason: cardiac arrest, exsanguination, profound shock—the brain has only seconds before it can no longer do its work. Within ten to twenty seconds, organized electrical activity at the cortical surface ceases. Within roughly thirty seconds, the EEG goes effectively flat. After about four to six minutes without restored circulation, neurons begin to die in numbers that cannot be undone. The story is sober and well-documented and grim. The brain, on this story, requires its biological substrate to do anything at all. No fuel, no firing. No firing, no thinking. No thinking, no experience. Period.

The second story is the one a substantial number of cardiac-arrest survivors tell. It goes like this: I went into arrest, and then—somehow, somewhere—I was conscious. I saw the team. I heard what was said. I drifted to the ceiling and looked down. I went somewhere else. I came back. Sometimes the second story includes details that, when checked against the medical record, against the team's later reports, against what the patient could not have known any other way, turn out to be accurate. Specific. Real.

These two stories do not fit together easily. If the first story is right and tells the whole story, the second one cannot be true as reported. There must be some other explanation: confabulation, memory misattribution, brief windows of preserved activity, REM intrusion, anesthetic side-effects. If the second story is right as reported, the first story cannot tell the whole story. Something must be conscious during the brain's shutdown that the standard model is not capturing.

That is the central problem this chapter sets up. Not by deciding it. The deciding comes later, when we look at the actual cases—the verified, well-documented ones—and weigh them. This chapter does the prior, prerequisite work of establishing the first story carefully. What does the brain actually do during cardiac arrest? What does the EEG actually show? When does conscious experience become biologically impossible according to the standard physicalist model?

I want to do this work fairly. I want the strict physicalist model to get its strongest hearing in this chapter. Because if the model is sloppy here, every later move is sloppy. If the medical baseline is honestly built, then when we lay the NDE evidence next to it later in the book, we will know what that evidence is up against.

A note on the audience. If you are a pastor or a chaplain or a hospice worker, the medical detail in the next pages may be more granular than you expected. Stay with me. You will be amazed how often the question "what is happening in her brain right now?" arises at the deathbeds you visit. Family members ask it. Doctors ask it. The dying person, in some lucid moments, asks it. Knowing the actual answer, in language you can convey gently, is part of the equipment of caring well at the bedside. If you are a physician or a researcher, the level of detail will be familiar. Skim the parts you know cold. The book is not trying to make a neurologist out of you; it is trying to put us all on the same page about the medical baseline so the later evidential work has a common foundation.

We are now going to walk through what neuroscience says about brain shutdown—first, the brain's astonishing hunger; then the cascade that happens when the food stops coming; then the timeline of electrical silence; then the major studies. After that, we will look at what the cardiac-arrest NDE literature has actually found. By the end of the chapter, we will have a map of what, on the standard model, the brain can and cannot do during the seconds and minutes after cardiac arrest—and the place where the cardiac-arrest NDE evidence collides with that map.

Alright. Let us start with the brain's appetite.

The Brain's Hunger

Pick up your hand. Make a fist. Look at it. That is roughly the size of your brain. Open your hand. That is, roughly, the volume.

That fist-sized object weighs about three pounds—somewhere between 1,300 and 1,500 grams in most adults. It is somewhere around two percent of your body weight. By every measure of mass, it is a small organ. Smaller than your liver. Smaller than your lungs combined. Roughly the same weight as your heart.

And yet, at rest, doing nothing in particular—not running, not solving math problems, not even listening hard to a conversation—your brain consumes about twenty percent of the oxygen your body uses, and roughly twenty percent of the glucose your bloodstream delivers. Two percent of body weight. Twenty percent of resting fuel consumption.⁵ The brain is, by metabolic intensity, the hungriest organ you have.

Why so hungry?

Because the brain is the most electrically active object in your body. Inside your skull, roughly eighty-six billion neurons signal to one another across roughly one hundred trillion synaptic connections, in a kind of staggering ongoing electrical conversation that does not stop while you sleep, does not stop when you are bored, does not stop when you think you are doing nothing. To make any one of those eighty-six billion neurons fire requires energy—specifically, the energy stored in a molecule called ATP, adenosine triphosphate, which is the fuel currency of the cell. A neuron that wants to fire must first set up an electrical gradient across its outer membrane: lots of sodium outside, lots of potassium inside, calcium very tightly held. Setting up and maintaining that gradient takes ATP. Every firing of every neuron uses ATP. Every reset takes ATP. Every neurotransmitter packet shipped across a synapse and recycled takes ATP. The brain is, in essence, a colossal ATP-burning electrical machine.

Where does the ATP come from? In the brain, almost entirely from oxygen and glucose, delivered together by cerebral blood flow.

This is the key fact. Other organs can store fuel. Your liver keeps glycogen on hand. Your muscles can run on stored glycogen and on a little bit of anaerobic burning when oxygen is short. The brain has almost no reserve. It cannot run on anaerobic metabolism for any usable length of time. It does not store glucose to speak of. It depends—continuously, second by second—on the steady delivery of fresh oxygen and fresh glucose by the bloodstream. Cerebral blood flow is, for the brain, the umbilical cord. Cut it, and the clock starts running.⁶

How fast is the clock?

Faster than you might guess.

The Ischemic Cascade

Imagine the moment when blood flow to the brain stops. What happens in the next few seconds, minute by minute, second by second?

The first thing to fail is ATP production. The brain has only enough on hand to keep its electrical machinery running for a matter of seconds. As the supply runs out, the cells can no longer pay the bill for their electrical gradients. The pumps that had been keeping sodium outside and potassium inside slow and then stop. Sodium leaks in. Potassium leaks out. Calcium, normally held outside or in tight intracellular stores, floods in.⁷

This is not benign. Once calcium floods the cell, a cascade of damage starts. Enzymes that should have been quietly resting are activated. They begin chewing up cell structures. The cells start releasing massive amounts of glutamate, the brain's main excitatory neurotransmitter, into the spaces between them. Other cells, whose receptors are flooded with this glutamate, are over-excited—driven into a state called excitotoxicity, where the very signals that normally let cells communicate are, at this concentration, lethal. Free radicals are generated. Mitochondria—the little cellular furnaces—begin to fail. Cell membranes start to lose integrity. Inflammation kicks in.⁸

Doctors call this the ischemic cascade. It is what happens to brain tissue when the blood stops flowing. It is, in slow motion, what causes a stroke. It is, in faster motion, what is happening throughout the brain during a cardiac arrest.

The cascade is not a dimmer switch. It is a cliff.

When the brain has fuel, it functions. When the fuel is cut off, function does not gracefully diminish. Within seconds, organized electrical activity collapses. The neurons cannot pay for their own firing. The cortical surface goes quiet. The lights go out.

And here is where it gets interesting for our purposes. Because if the standard physicalist model is right—if conscious experience just is the organized electrical activity of the cortex, in some yet-to-be-fully-specified way—then the moment the cascade hits, conscious experience must end. There is no fuel for the lamp; the lamp must go dark.

The question is: how fast?

The Timeline of Electrical Silence

The clearest data on this question come from a small set of remarkable studies in which human beings were temporarily put into cardiac arrest under controlled conditions and their brains' electrical activity was monitored throughout. The conditions were not punitive. They were therapeutic. In the late 1980s and through the 1990s, cardiologists implanting devices called ICDs—implantable cardioverter-defibrillators, used in patients prone to dangerous arrhythmias—needed to test that the device could detect and shock a patient out of ventricular fibrillation. The standard test involved deliberately inducing the arrhythmia in the operating room while the device was already implanted, confirming the device worked, and shocking the patient back. For patients in danger of sudden cardiac death without the device, this brief, controlled test was a small price for a life-saving instrument.

For neuroscientists, however, these procedures opened a window. Patients in induced cardiac arrest could be hooked to EEG. Their brain activity could be watched in real time, second by second, as their hearts stopped. We could see, as no observation in ordinary medicine had let us see, what the brain actually does in the seconds after circulation ceases.

Michael Aminoff and his colleagues at the University of California, San Francisco, published the most important of these studies in 1988, in the Annals of Internal Medicine.⁹ They monitored EEG during induced ventricular fibrillation in patients receiving ICDs. They documented the timing precisely. Within roughly six to seven seconds of the onset of induced ventricular fibrillation, the EEG began to slow dramatically. Within ten to twenty seconds, organized cortical activity had effectively ceased. The line was approaching flat.

Other studies replicated and extended these findings. Clute and Levy, working in similar territory, found similar timings. The literature on the natural history of EEG during cardiac arrest converged on a stable picture: ten to twenty seconds for the disappearance of organized cortical activity; by thirty seconds, an essentially isoelectric EEG.¹⁰

What does that mean?

It means that within roughly half a minute of cardiac arrest, the brain's surface electrical activity—the activity that, on every mainstream model, supports conscious experience—has gone silent.

Now, two important caveats need stating up front, because the strict physicalist will press both, and they deserve a fair hearing.

Caveat one: deep brain activity. A standard surface EEG, the kind used in clinical practice, picks up electrical signals from the cortex—the outer layer of the brain. It does not directly measure activity from deeper structures (the brainstem, the thalamus, the hippocampus). So when we say "the EEG is flat," we mean cortical surface activity has ceased. Deeper structures may have residual activity for a brief additional period before the ischemic cascade reaches them.¹¹ Whether this residual deep-structure activity could support conscious experience is hotly contested. The mainstream view in cognitive neuroscience is that conscious experience requires integrated activity across cortical regions; the deeper structures are necessary for arousal and modulation but not, by themselves, sufficient for the rich content of conscious thought and perception. But the physicalist who wants to find a wedge for residual consciousness during the flatline period will press here. We will return to that in the section on physicalist alternatives.

Caveat two: the limits of EEG. Surface EEG is an imperfect instrument. It picks up the summed electrical fields of large populations of neurons firing in synchrony at the cortical surface. It can miss small populations firing asynchronously. It can miss activity in deep structures, as just noted. It can be confounded by electrical noise (CPR motion artifact is a well-known problem in cardiac-arrest EEG). So when we say "isoelectric," we mean "no organized signal detectable at the cortical surface above the noise floor of the recording instrument." That is not exactly the same thing as "no neural activity at all anywhere in the brain."¹²

The careful physicalist who wants to leave room for some residual brain activity during cardiac arrest will note these caveats and lean on them. That is fair. But notice what the caveats do not establish. They do not establish that organized, integrated, conscious experience is possible during a flat EEG. They establish that there might be some neural noise we are not measuring. The claim that this noise is consciousness, or that it could support the kind of detailed, accurate, perceptual reports cardiac-arrest survivors give us, is a claim that needs to be defended on its own. It cannot simply be assumed.

For now, hold the timeline. Within ten to twenty seconds of cardiac arrest, organized cortical activity has ceased. Within roughly thirty seconds, surface EEG is effectively flat. Within four to six minutes without restored circulation, neurons begin dying in irreversible numbers. After about ten minutes, full neurological recovery becomes increasingly unlikely.¹³ This is the window in which cardiac-arrest NDEs report happening: a window during which, by every standard medical measure, the cortex has gone silent.

Clinical Death, Biological Death, and the Boundary Between

One more piece of medical scaffolding belongs in this chapter before we look at the major studies. It will pay off later. We need to understand what doctors mean when they say someone is "dead."

It turns out they mean different things at different points in the dying process, and the differences matter. Consider three terms.

Clinical death traditionally meant cardiac arrest with absent breathing and absent reflexes. The heart has stopped; the lungs have stopped; the patient is unresponsive. Before the development of effective resuscitation, clinical death meant death, full stop—there was no further movement of the patient back toward life. Once resuscitation became possible, clinical death became a reversible category. A person can be clinically dead (no pulse, no breath) and then, with effective CPR, return.

Biological death is irreversible. It refers to the cessation of all vital biological processes—cellular as well as systemic—to the point that no resuscitation is possible. A person whose neurons have died, whose myocardium has been without perfusion long enough to be unsalvageable, is biologically dead. The line between clinical and biological death is not a sharp clock; it is a gradient. With every minute of cardiac arrest, the chance of full recovery decreases and the proximity to biological death increases.

Brain death is a third concept, developed in the modern era largely for organ-donation purposes. It refers to the irreversible cessation of all brain function, including the brainstem. A brain-dead patient may be on a ventilator, with a beating heart, but is legally and medically dead. The concept emerged in the 1968 Harvard ad hoc committee report on the criteria for brain death and was codified into U.S. law by the Uniform Determination of Death Act in 1981.¹⁴

Why does this taxonomy matter for our chapter?

Because the cardiac-arrest patient who survives an arrest and reports an NDE was, during the arrest itself, in clinical death. They were not biologically dead. They were not brain dead. Their brain was alive in the sense that some neurons were still salvageable, that with restored circulation it could come back. But during the arrest—during the window when the EEG would have been flat—they were in a state that, before modern resuscitation, would have been understood simply as "dead." Their physicalist counterpart would say: they were in a state in which, on any standard model, conscious experience should have been impossible, even though their brain tissue was not yet permanently destroyed.

The line between "the brain is silent but salvageable" and "the brain is silent and gone forever" is, for our purposes, less important than the line between "the brain is producing organized conscious experience" and "the brain is not." The cardiac-arrest patient in flatline EEG is on the wrong side of the second line. Whether or not she is permanently dead, she is, during that window, not producing the kind of activity the standard model identifies with conscious experience.

That is the baseline this chapter is building. Now let us see what the major studies have found in cardiac-arrest patients who survive that window.

The Major Studies

Aminoff and the EEG Window

The Aminoff study has already been mentioned, but it deserves a moment more. Aminoff and his colleagues did not set out to study NDEs. They were neurologists interested in syncope—the medical term for fainting—and they wanted to characterize what happened to the EEG during the kind of brief cardiac arrhythmias that produce loss of consciousness. The ICD-implantation patients gave them the rare chance to monitor brain activity during a controlled, witnessed, fully-instrumented cardiac arrest. They published their findings in Annals of Internal Medicine in 1988 under the title "Electrocerebral Accompaniments of Syncope Associated with Malignant Ventricular Arrhythmias."¹⁵

The findings have been confirmed many times since. EEG slowing within six to seven seconds. Loss of organized cortical activity within ten to twenty. Effective isoelectric EEG within thirty. The Aminoff data are the bedrock for every later discussion of consciousness during cardiac arrest. They establish the baseline: when the heart stops, the surface of the brain stops shortly after, and stays stopped until the heart starts again.

What Aminoff and his colleagues did not study was what happened, if anything, to the patients' subjective experience during the flatline window. Their patients were quickly defibrillated and revived; they were not systematically asked about subjective experience afterward. The question of whether anything was happening for the patient during the flat-EEG window was not their question. But it was the next obvious question, and the work of asking it began roughly contemporaneously, in a different part of the medical world.

Sabom and the Atlanta Study

Michael Sabom is a cardiologist, originally at Emory University in Atlanta. In the late 1970s, he was a young physician with a healthy skepticism about the new fad of "near-death experience" research that Raymond Moody had launched with Life After Life in 1975. Sabom set out, by his own account, to debunk the NDE accounts by careful interviewing of his cardiac patients. Instead, what he found turned him into one of the most rigorous early NDE researchers.¹⁶

Sabom's Recollections of Death (1982) reported on a careful prospective study of cardiac patients who had survived cardiac arrest. He compared their reports of their resuscitations to a control group of cardiac patients who had not had NDEs. The control group, asked to imagine their own resuscitations, produced descriptions full of medical errors—wrong sequences of events, wrong equipment, wrong drug names, wrong placements. The NDE group, asked the same questions, produced descriptions that were strikingly accurate—correct sequences, correct equipment, correct drug names, correct placements—in a way that the control group simply was not.¹⁷

This was a methodologically careful finding. Sabom did not assume the NDE patients had supernatural perception. He compared what they reported to what other cardiac patients—equally exposed to medical television, equally familiar with hospital routine—could come up with by guessing. The NDE group beat the control group dramatically. Sabom's later work (Light and Death, 1998) extended these findings and engaged the famous Pam Reynolds case, which is owned by Chapter 12 and will be developed at length there.

The Sabom data are not the proof of veridical NDE phenomena. They are one piece of a converging picture. But they were among the first to systematically demonstrate that cardiac-arrest survivors with NDE reports describe their resuscitations more accurately than cardiac-arrest survivors without NDEs and substantially more accurately than other cardiac patients asked to imagine such events.

Van Lommel and the Lancet

If any single study marked the medical legitimization of NDE research, it was the Pim van Lommel Lancet paper in December 2001. Van Lommel, a Dutch cardiologist, led a prospective study of 344 consecutive cardiac-arrest survivors at ten Dutch hospitals over a roughly four-year period. The patients were interviewed within a few days of resuscitation about whether they remembered anything from the period of unconsciousness. Eighteen percent reported what could be classified as NDEs. The study was rigorous: it controlled for medication, type of cardiac event, duration of arrest, depth of unconsciousness, and a host of other factors. It documented that NDEs occurred in patients with very brief arrests and in patients with prolonged arrests; in patients on no medication and in patients on every kind of cardiac drug; in religious patients and in skeptical ones.¹⁸

The Lancet, one of the most prestigious medical journals in the world, ran the paper. Van Lommel's study became the most-cited NDE paper in the medical literature. The paper itself argued, carefully, that the standard physicalist explanations of NDE—cerebral anoxia, drug effects, expectation, hallucination—could not account for the data. NDEs occurred in arrests as short as a few minutes, where anoxic damage was minimal. They occurred in patients on no psychotropic medication. They occurred whether the patient had any prior religious framework or not. And the patients with NDEs frequently reported coherent, detailed experience from a period during which their EEG, had it been monitored, would have been flat.¹⁹

Van Lommel did not stop with the 2001 paper. His later book, Consciousness Beyond Life (2010), expanded the argument. He argued that the cardiac-arrest NDE data were inconsistent with the view that consciousness is identical to brain function and were consistent with what he called a "non-local" view of consciousness, in which the brain is more like a receiver than a producer.²⁰ Whether one agrees with van Lommel's full metaphysical conclusion or not, the empirical core of his work is hard to set aside. Cardiac-arrest patients, in a substantial percentage of cases, report coherent conscious experience from periods when, on the standard model, no such experience should be possible.

Sartori and the British ICU

Penny Sartori is a British intensive-care nurse who completed a doctorate at the University of Wales on the NDEs of intensive-care patients. Her prospective study, conducted in a single Welsh hospital's ICU over a five-year period, was smaller than van Lommel's but allowed for far richer in-person data collection. Sartori was at the bedside; she was the interviewing nurse; she had access to the medical records and the bedside team. Her resulting study, published as The Near-Death Experiences of Hospitalized Intensive Care Patients (2008), extended van Lommel's findings and added a number of striking veridical cases.²¹

Sartori's most discussed case involved a patient who, during cardiac arrest, reported leaving his body and observing his resuscitation. He reported specific details about the resuscitation team's actions. When Sartori subsequently asked the team to imagine what their actions had been (they were not the patient's regular team), they got many of the details wrong. The patient's report matched the actual sequence in a way the imagined controls did not. This is the same methodological strategy Sabom had used. The pattern repeated.²²

What makes Sartori's work distinctive is the ICU setting. Her patients were, in many cases, profoundly compromised. They were mechanically ventilated, sedated, on multiple drips, sometimes paralyzed. They were as close to a continuous medically-monitored state as patients get. The pattern of NDE reports—and of veridical content in those reports—persisted in this setting. The standard physicalist explanations had to work against an even more challenging set of clinical conditions.

Parnia and the AWARE Studies

Sam Parnia is the figure who, more than any other, has tried to design a definitive study of cardiac-arrest NDEs. Parnia is an intensive-care physician at NYU Langone, and he has spent more than fifteen years organizing multi-center prospective trials with built-in tests for veridical perception.

The AWARE I study, published in Resuscitation in 2014, was the first major attempt at this kind of study at scale. AWARE I enrolled 2,060 cardiac-arrest patients across fifteen hospitals in the United States, the United Kingdom, and Austria. The study placed hidden visual targets on shelves above the heads of patients in the bays where cardiac arrests were most likely to occur. The targets were images that could be seen only from above—facing the ceiling. The hypothesis: if cardiac-arrest patients really do have an out-of-body perspective from near the ceiling during their arrest, some of them should be able to identify the hidden images. If none of them can, that is evidence against the strong veridical-perception claim.²³

The shelf experiment, by itself, did not produce a positive identification. No patient identified the hidden image (although only a small minority of arrests occurred in bays where shelves had been placed, and only a small minority of those produced an NDE; the actual sample size for a positive shelf-target hit was very small). But AWARE I produced something else of substantial interest. Of 140 cardiac-arrest survivors who were interviewed about their experience, nine had NDEs by standard scoring. Two reported explicit awareness during the arrest. One of those two reported, with remarkable accuracy, the actions of the resuscitation team during the period when his EEG would have been flat. He named a sound—an automated voice from a defibrillator saying "shock the patient now"—that occurred at a specific point in the resuscitation. He described seeing himself from a position in the corner of the room. The medical record corroborated the timing. He could not have heard the voice while unconscious, on the standard model. He could not have seen himself from outside himself, on any model. And yet his report matched the documented sequence of events.²⁴

AWARE II, published in Resuscitation in 2023, extended the study with new methodology. The team measured EEG during cardiac arrest itself in a subset of patients, looking for any neural correlates of the reported experiences. They found, intriguingly, that some patients showed transient gamma-band activity during CPR—activity that is, in normal contexts, associated with conscious cognition. This complicates the simple "flat EEG, then NDE, then return" picture. It does not eliminate it. The transient gamma was brief, intermittent, and not in the same patients who reported the most vivid experiences. AWARE II also documented additional veridical reports, including patients who described the contents of their resuscitations with high accuracy.²⁵

The AWARE studies will be engaged at greater length in Chapter 12. For our purposes here, the central point is the convergence with the earlier work. Parnia's prospective methodology produces the same finding as van Lommel's, as Sabom's, as Sartori's: a substantial fraction of cardiac-arrest survivors report conscious experience from a period when, on the standard model, none should be possible, and a meaningful subset of those reports include verifiable content.

The Dissertation Data

The author's doctoral dissertation, submitted in 2025 to Trinity College of the Bible and Trinity Theological Seminary, attempted a systematic mixed-method analysis of the NDE evidence at scale. The dataset comprised 5,278 cases drawn from two pools—832 from peer-reviewed scholarly sources, including the studies just discussed, and 4,446 from the publicly accessible online databases maintained by NDERF and IANDS. Each case was scored on a three-dimensional rubric: medical-context score (severity and verifiability of the medical crisis), veridical-quality score (clarity, detail, and inaccessibility of perceived information through normal means), and corroboration score (quality of independent witness testimony and documentation).²⁶

Of the 5,278 cases, 30.7 percent met the dissertation's "Exceptional" or "Strong" thresholds for combined evidential weight. Within the cardiac-arrest subset of the database—a smaller, more medically uniform sample—a particular finding stands out for our purposes: 89.96 percent of cardiac-arrest patients with documented NDEs reported their conscious experiences during the period when there was no measurable brain activity. That is, in 89.96 percent of cardiac-arrest NDE cases for which timing could be reliably reconstructed from the medical record, the patient's reported subjective experience corresponded to a window of documented EEG flatline (or, where EEG was not directly recorded, to a window in which standard medical baselines would predict EEG flatline based on the duration and severity of arrest).²⁷

That number deserves a moment of reflection. Almost nine in ten cardiac-arrest NDEs occur, by the patient's own timing report and by the documented medical timeline, during the very window the standard physicalist model identifies as the window of cortical silence. This is not what the model predicts. The model predicts a vanishingly small percentage, ideally zero, of conscious reports from that window. The data are roughly the inverse of the prediction.

The dissertation also documented a finding that, on physicalist grounds, is even harder to explain. There is, in the cardiac-arrest subset, a positive correlation between medical severity and clarity of veridical content. Patients with longer arrests, with deeper unconsciousness, with more documented neurological compromise during the arrest, tended to report clearer and more verifiable experiences than patients with briefer or less severe arrests. The physicalist prediction is the opposite: more severe brain compromise should produce more confused, less coherent, less verifiable reports. The dissertation's data point in the wrong direction for the physicalist account.²⁸

The Strict Physicalist Prediction and What the Data Show

Now we are in a position to state the empirical situation cleanly.

The strict physicalist model of consciousness holds that conscious experience is identical to—or at least supervenient on, in a tight, lawlike way—a particular kind of organized brain activity. The exact form of that activity is debated (global workspace theory, integrated information theory, higher-order theories, recurrent processing theories, and so on), but every credible candidate requires some organized large-scale activity at the cortical surface. Without that activity, the model does not produce conscious experience. Period.

From this, two predictions follow about cardiac arrest.

Prediction one: cardiac-arrest patients should not report conscious experience from the window of flat EEG. Whatever they may have experienced before the arrest or after recovery, the arrest itself should be a gap. They should report what someone under deep general anesthesia reports—nothing.

Prediction two: when cardiac-arrest patients do report something (perhaps because their experiences happened just before or just after the flatline window, in periods of brief or partial activity), the reports should be confused, fragmentary, lacking specific veridical content, and bearing the marks of the kind of confabulation that compromised brains generate.

What the data show:

Eleven to eighteen percent of cardiac-arrest survivors report some form of NDE.²⁹ The reports are not confused or fragmentary; they are, in many cases, more coherent and more specifically detailed than the recollections of equivalent patients of mundane events. The reports are not lacking in veridical content; a meaningful subset includes verifiable specifics. And the reports—to the extent that the timing can be reconstructed—cluster heavily within, not outside, the window of documented brain shutdown.

This is not what the model predicts. It is roughly what the model predicts the inverse of.

The strict physicalist has several lines of response available, and it is right to run through them honestly.

Physicalist Counter-Explanations

The "Before/After" Timing Hypothesis

The first and oldest line of physicalist response is the timing hypothesis. The idea: maybe NDEs happen not during the flat-EEG window, but rather in the brief moments before EEG flattens (the seconds during which the brain is still processing but consciousness is fading) or in the period of recovery just after circulation is restored (when the brain is producing chaotic, confused activity). On this view, the NDE is real, but its actual neural substrate is in one of these "edge" windows, and the patient simply misremembers the timing as having occurred during the arrest itself.³⁰

This is a serious proposal, and parts of it are right. We know that the brief moments before consciousness is lost can include vivid content; we know that the moments of recovery can include confusion. The physicalist who invokes the timing hypothesis is pointing to real biological windows of activity.

What the timing hypothesis cannot do, however, is account for the specific veridical content. The cardiac-arrest survivor who describes the specific actions of his resuscitation team—actions that occurred minutes after he lost consciousness and during his flat-EEG window—cannot have those memories from before the arrest (the actions had not happened yet) or from the recovery window (he was not present at his own resuscitation when he became conscious again, several minutes after the team had finished). The information had to enter his memory somehow, and the timing hypothesis does not specify a mechanism that would let him acquire information about events that happened during the only window when he could have been a witness—the arrest itself.

The hypothesis works, in other words, for the subjective content of NDEs (the lights, tunnels, and reviews). It does not work for the objective veridical content (specific accurate observations). And it is the veridical content that drives the book's evidential argument.

The Borjigin "Surge" Hypothesis

The most discussed physicalist counter-evidence in recent years is the work of Jimo Borjigin and her colleagues at the University of Michigan. In a 2013 paper in PNAS, Borjigin's team reported a brief "surge" of high-frequency gamma-band activity in the brains of dying rats, lasting roughly thirty seconds after cardiac arrest. The team interpreted this surge as a possible neural correlate of consciousness during the dying process and suggested it could explain the heightened, vivid content of human NDEs.³¹

A 2023 follow-up paper, also in PNAS, reported similar findings in a small sample of dying human patients in whom EEG had been continued through cardiac arrest as part of an end-of-life protocol. Two of four patients showed transient gamma-band increases following the withdrawal of life support.³²

The Borjigin findings are real and interesting and will be engaged at full length in Chapter 15. For our purposes here, three points need stating.

First, the gamma surge is brief. Roughly thirty seconds in the rat model; comparable timescales in the human. It does not span the full window during which cardiac-arrest NDEs are reported, which can extend over minutes and sometimes longer.

Second, gamma activity by itself is not consciousness. Gamma is a frequency band associated with conscious cognition in normal contexts, but its mere presence does not constitute conscious experience. A seizure produces wild, high-frequency activity, and the patient is not experiencing rich coherent cognition; quite the opposite. The Borjigin team's interpretation that the surge represents conscious experience is one possible reading; alternative readings (an unmasking of subcortical activity due to loss of cortical inhibition; a final discharge of synchronized firing as cells lose their resting potentials) are equally consistent with the data.

Third, and most importantly for the present chapter, the Borjigin surge does not explain the content of NDEs. Even granting that it might be a neural correlate of some kind of conscious experience, it does not generate the specific, accurate, externally-verifiable perceptual content that veridical NDEs include. A surge of gamma activity in a dying brain might produce confused mental experience. It does not produce a correct identification of who said what at the foot of the bed during a resuscitation, or an accurate description of medical equipment the patient could not have seen.

Anesthesia, Anoxia, and Drug Effects

Other physicalist proposals point to specific medical conditions that might generate NDE-like content. Anesthesia awareness is real, well-documented, and produces a particular kind of intra-operative consciousness in patients whose anesthesia has not adequately suppressed cortical activity. Cerebral anoxia—the partial loss of oxygen short of full arrest—produces hallucinations, paresthesias, and disordered thought. Ketamine, an anesthetic with dissociative effects, can produce out-of-body sensations. DMT, the endogenous psychedelic, has been proposed as a possible neurotransmitter mediator of NDE content.³³

Each of these proposals has merit as a partial explanation for some features of some NDEs, and each will be engaged in detail in Chapter 16. None of them, individually or collectively, accounts for the cardiac-arrest data. Cardiac-arrest patients are not under anesthesia. They are not in partial anoxia; they are in full circulatory arrest. They are not, typically, on the relevant pharmacological agents (although some are; the van Lommel data showed no correlation between drug exposure and NDE incidence). And no proposed pharmacological model accounts for the specific veridical content.

Confabulation and Memory Reconstruction

The final line of physicalist response is the confabulation hypothesis: the patient does not actually remember anything from the arrest itself; what they call a "memory" is a post-hoc reconstruction, built from fragments of medical knowledge, cultural NDE expectations, brief overheard conversations during recovery, and the brain's general tendency to fill in gaps with confabulated content. On this view, NDEs are a kind of false memory, generated after the fact by a recovering brain trying to make sense of an unconscious gap.³⁴

This is a serious hypothesis and worth taking seriously. The brain does confabulate; false memories are real; cardiac-arrest patients are, post-recovery, in a vulnerable cognitive state. Some NDE reports are probably significantly shaped by post-arrest reconstruction.

But the hypothesis cannot account for the strongest cardiac-arrest cases. A patient who reports the specific words spoken by a nurse at his foot during a moment of his arrest—words the nurse confirmed—cannot have confabulated those words from cultural expectations or from medical-television exposure. The words were specific to that arrest. They had to enter the patient's memory by some channel. Confabulation does not provide the channel.

More broadly, the methodological strategy of the major NDE researchers—Sabom, van Lommel, Sartori, Parnia—was specifically designed to control for confabulation. They interviewed patients quickly, before extensive opportunity for reconstruction. They cross-checked reports against medical records. They compared NDE reports against the imagined accounts of control patients who had not had NDEs. The veridical content, in study after study, exceeded the control accuracy. Confabulation, on its own, does not predict that pattern.

Assessment: What This Chapter Has Established

Let us step back and assess.

What has the chapter established?

It has established, with high confidence, that the brain at cardiac arrest goes into a state of organized cortical electrical silence within a small window of time—ten to twenty seconds for the loss of organized activity, roughly thirty seconds for an effectively isoelectric EEG. This window is well-documented across multiple studies, is consistent with what we know about the brain's energy economy, and is not seriously contested by any party in the relevant scientific literature. This much is medical baseline.

It has established that the strict physicalist model of consciousness predicts no rich, coherent, veridical conscious experience during this window. Whatever else the model can do, it cannot accommodate the kind of detailed perceptual reports that the cardiac-arrest NDE literature has documented.

It has established, by gesturing at the major studies—Sabom, van Lommel, Sartori, Parnia—that the empirical literature on cardiac-arrest NDEs has consistently found a substantial fraction of survivors reporting coherent conscious experience from this window, with a meaningful subset reporting verifiable details. The dissertation's analysis of 5,278 cases yields the striking figure that 89.96 percent of cardiac-arrest NDEs occur, by patient timing and documented medical timing, during the documented brain-shutdown window.

It has established that the standard physicalist counter-explanations—timing displacement, the Borjigin gamma surge, anesthesia and drug effects, confabulation—each have something to contribute as partial explanations of some features of some NDEs but do not, individually or collectively, account for the strongest cardiac-arrest cases with veridical content.

What the chapter has not established, and does not pretend to have established:

It has not proven substance dualism. The case for substance dualism is built across the entire book; the synthesis is in Chapter 23. This chapter only establishes one piece of the case—the medical baseline against which the empirical evidence in later chapters takes on its evidential weight.

It has not engaged the strongest specific cases. The Pam Reynolds case, the Maria-and-the-shoe case, the AWARE veridical-perception cases—these are owned by Chapter 12 and Chapter 10, where they will be developed at the length they deserve. This chapter has only sketched the medical setting.

It has not addressed the harder phenomenological questions—what cardiac-arrest survivors actually experience subjectively, what the shape of the experience is, what it reveals about the nature of postmortem consciousness. Those are the work of Chapters 18–22.

And it has not closed off the genuinely difficult questions of methodology. How tightly can we reconstruct timing from medical records? How well-controlled are the prospective studies? How much of the veridical content survives the most aggressive skeptical scrutiny? These are addressed in Chapter 8, where the methodology of veridicality is developed in full.

What the chapter has done, I hope, is the prerequisite work. It has put us all on the same page about the medical baseline. It has been honest about what the strict physicalist model predicts, honest about what the data show, and honest about the auxiliary hypotheses needed to keep the strict model afloat. It has not pretended that the case is closed; it has insisted only that the contradiction is real. The brain, at cardiac arrest, goes silent. People in cardiac arrest, in a substantial fraction of cases, report rich and sometimes verifiable conscious experience from the silent window. Whatever account we ultimately give of this contradiction, we cannot give it without acknowledging it.

An additional piece of honesty. The Borjigin findings are genuinely interesting, and the AWARE II evidence of transient gamma during CPR genuinely complicates the simple "flat EEG, then NDE, then return" picture. The scientific story is not as clean as it was in the 1990s. There is more residual activity than the early studies suggested. A careful researcher today has to grant the physicalist model more room to maneuver than would have been granted twenty years ago. I want to grant that room. The dualist case does not depend on the brain being absolutely electrically silent during arrest; it depends on the brain being unable to support the kind of integrated, coherent, veridical conscious experience that cardiac-arrest survivors actually report. Even with transient gamma, the brain in arrest is not in a state where standard models predict such experience. The contradiction stands. It just stands with sharper edges than in the simpler picture.³⁹

One more thing belongs in the assessment, because it is the kind of methodological honesty that pastoral readers will want and skeptical readers will demand. The 89.96 percent figure from the dissertation is, like every figure derived from a mixed-source database, dependent on the quality of the underlying timing reconstructions. Some cases in the database have exquisitely detailed medical records and patient timing reports that match cleanly. Others have looser data. The dissertation's scoring methodology weighted timing confidence as part of the corroboration score; the highest-confidence cases drive the strongest claims. The 89.96 percent is the headline figure, but the texture beneath the figure varies. The Exceptional and Strong cases—the 30.7 percent of the database that the dissertation's three-dimensional scoring identified as evidentially weighty—are the ones the book will rely on most heavily. The reader who wants the sharpest version of the data will find it in Chapter 14.

Why This Matters

I want to close where I started: at Frank's bedside, on the morning after his cardiac arrest, when the attending physician asked him what he remembered.

Whatever Frank tells her, he is going to tell her in a body that was, twenty-four hours earlier, biologically defunct in the way every cardiology textbook describes. His brain was, by every standard medical instrument, not in business. His cortex was offline. The voltage on the surface of his brain was, for the duration of the relevant window, indistinguishable from the noise floor of the EEG.

And yet there he is, eating Jell-O, watching basketball, ready to speak.

If the strict physicalist model is right, Frank should remember nothing. He should report what someone under deep general anesthesia reports—a clean, dreamless gap. The data say a substantial fraction of patients in his clinical situation do not report a clean gap. They report something. Sometimes they report something verifiable.

This is where the argument of this book begins to land—not on a doctrine, not on a theory, but on a person in a hospital bed who is about to say something. The medical baseline tells us what should be impossible for him to say. The clinical literature tells us what people in his situation routinely do say. The disagreement between those two facts is not a polemic against physicalism; it is the starting place of an honest theology of dying.

Why does it matter? Three reasons, in closing.

First, for the dying themselves. People who are facing death often want to know whether the standard medical picture is the whole picture. They are not asking out of idle curiosity. They are asking because they are, themselves, about to enter the territory the picture describes. To answer "yes, the picture is the whole picture; you will simply cease" when in fact the data point in another direction is to give a person facing death an answer that may not be true, on the strongest available evidence. To answer "the picture is part of the picture; what your body and brain do at death is real and accurately described by the standard medical account; but the evidence suggests there is more—something the brain's silence does not silence" is to give a person facing death an answer that is more honest, more careful, and more in keeping with what we have learned. This is the answer the historic Christian tradition has always offered, on theological grounds. The NDE evidence supplies an empirical convergence with that ancient answer.

Second, for those who companion the dying. Family members at a deathbed often ask, sometimes aloud and sometimes silently, "is she still in there?" The answer requires both pieces. The neurological account describes what is happening to the brain—the slow withdrawal of cortical function, the changes in breathing, the possible loss of awareness as the body shuts down. The soul-account, on the historic Christian view, says the person is more than what the brain is doing. The veridical NDE evidence gives us reason, beyond confessional testimony, to believe the soul-account is tracking something real. "Is she still in there?" The honest answer, on the best current evidence, is: yes—in a way the medical instruments cannot fully measure. That answer is not a pious fiction. It is what the data, taken together, support.³⁵

Third, for the larger argument of this book. The remainder of When Death Approaches rests on the foundation this chapter has tried to lay. When we get to the cases in Chapters 10–14—the cases of accurate distant observation, the cases of NDEs in the congenitally blind, the cardiac-arrest NDEs with documented EEG flatline, the encounters with deceased loved ones not known to be dead—the evidential weight of those cases rests on the medical baseline this chapter has built. If the brain at cardiac arrest could plausibly support coherent conscious experience, the cases would be merely interesting. Because the brain at cardiac arrest cannot, on the standard model, support such experience, the cases become probative. Each of the five evidential chapters is a needle. The medical baseline is the thread that lets each needle do its work. Without the baseline, the cases are floating anecdotes. With the baseline, they are pieces of evidence in a cumulative case for substance dualism, the conscious intermediate state, and the Christian theological account of dying.

I want to honor the brain in closing. The brain is a precious, intricate, biologically magnificent organ. Its death is a real loss. The historic Christian tradition has never taught that the brain doesn't matter; it has taught that the brain is the soul's instrument, not its generator. When the instrument falls silent, the musician does not vanish—but the music we have known is paused.³⁶ The body must be honored at death. The brain must be honored at death. The dying person must be honored at death.³⁸ And the evidence, gently and increasingly, suggests that the person whose brain has gone silent is still there, in a way the silenced brain cannot itself say.

The next chapter takes up the question that this one has been edging toward: when, exactly, are you actually dead? The medical-legal conventions for declaring death have shifted dramatically in the last sixty years. The moment of "death" is, in a sense, a definitional choice as much as a biological event. What that choice has implied for the patient—and what NDE evidence has begun to suggest about the boundary—is the work of Chapter 6.

For now, hold the picture. The brain at cardiac arrest goes silent. The data say something happens during the silence that the standard model cannot easily account for. The contradiction is real. The evidence is the next several chapters' work.⁴⁰ The hope—Christian hope, ancient hope, hope grounded in Christ's own descent and return—is what makes the work worth doing.