Chapter 11

Oxygen Deprivation, Carbon Dioxide,
and Chemical Explanations

For decades, medical schools ran a teaching demonstration that has since gone out of fashion. Students would breathe in and out of a sealed bag with a carbon dioxide absorber inside. The CO₂ they exhaled was scrubbed away, so their blood chemistry stayed comfortable. They did not gasp. They did not panic. The oxygen in the bag, however, kept dropping with every breath. Slowly, quietly, their brains were starving. The students were given simple tasks — count, name colors, draw shapes — and watched their performance fall apart before they finally lost consciousness.

Thousands of students. Thousands of times. Across decades.¹

Not one ever reported a near-death experience.

That single fact is worth pausing over. If a lack of oxygen is what produces NDEs — lucid, structured, transformative journeys remembered in detail for years — then this experiment should have produced them by the thousand. It did not produce one. Something is wrong with the theory before we even start unpacking it.

This chapter is about the most popular “chemical” explanations for NDEs: that they are produced by oxygen deprivation, by rising carbon dioxide, by endorphins flooding the brain, or by some combination of all three. These are not strawmen. Michael Marsh devotes a whole chapter to them.² They are also, in the popular skeptical mind, the standard answer. So they deserve a careful look.

A. The Critic's Argument

Marsh covers three main families of chemical explanation.

First, endorphins. Endorphins are the body's own painkillers — close cousins to morphine, made inside the brain. Marsh follows the work of Daniel Carr, who proposed that the stress of dying triggers a flood of endorphins, which then activate the limbic system and the temporal lobe.³ The result, Carr argued, would be the bliss, the analgesia, and the felt sense of peace that NDErs report. Marsh thinks endorphins probably play at most a supporting role. But he does not dismiss them. They could, he suggests, be the trigger that activates a deeper temporal lobe disturbance, which would in turn account for the more elaborate features of the NDE.⁴

Second, hypoxia and hypercapnia — too little oxygen and too much carbon dioxide. During cardiac arrest, severe hemorrhage, or low blood pressure, the brain undergoes (in Marsh's words) “variable degrees of hypoxic insult.”⁵ Marsh acknowledges Michael Sabom's well-known counterexample of a heart-attack patient whose blood oxygen was high during his NDE. He responds by suggesting that peripheral arterial gas measurements may not reflect what's happening locally inside brain tissue.⁶ For carbon dioxide, Marsh draws on the older work of L. J. Meduna, who exposed psychiatric patients to a 30 percent CO₂ / 70 percent O₂ mixture and reported a range of mystical-sounding experiences.⁷

Third, the tunnel and light. Following Susan Blackmore's “dying brain” model, Marsh argues that the tunnel-and-light feature reflects cortical recovery. As blood returns to a starved retina and visual cortex, peripheral vision comes back first, then central vision. The result, he says, is the impression of moving through a dark passage toward a brightening point of light.⁸ Marsh tweaks Blackmore's account — he says the integrated image must come from association cortex, not the primary visual area — but the basic story is the same. Hypoxia, then recovery, then the illusion of a tunnel.

Marsh's overall claim is more modest in tone than in implication. None of these mechanisms, he allows, fully explains NDEs by itself. But taken together, he suggests, they sketch a picture of the dying brain sufficient to account for the phenomenology — without supposing that consciousness has somehow left the body.

The argument has surface appeal. The brain is starving for oxygen. Endorphins are flooding the system. Carbon dioxide is rising. Of course we'd expect strange experiences. The dying body is a chemistry experiment going off the rails. Why look for anything more exotic?

It's a clean story. It just doesn't fit the data.

B. Identifying Weaknesses

Three problems hit Marsh's chemical case before we even reach the positive evidence.

First is the selectivity problem. Every patient who survives cardiac arrest has been hypoxic. Every one. That is what cardiac arrest is — a sudden stop in the supply of oxygenated blood to the brain. Within seconds the brain is starving. Within minutes, damage begins. If hypoxia produced NDEs, then every survivor of cardiac arrest should report one. The actual rate is between 10 and 20 percent, depending on the study.⁹ The other 80 to 90 percent have all the same brain chemistry and report nothing. Pim van Lommel, the Dutch cardiologist who led the largest prospective study of cardiac-arrest patients, makes the point bluntly: if the cause were physiologic, all patients should have the experience. They don't.¹⁰

Second is the lucidity problem. Hypoxia is not just any drug. It is one of the best-studied conditions in medicine. We know what it does to a conscious mind. R. A. McFarland's classic 1930s study of mountaineers found mental laziness, irritability, confusion, slowness in reasoning, and trouble remembering — the standard high-altitude profile.¹¹ Pilots in centrifuge studies report tunnel vision, then a few seconds of confused dreamlets, then nothing. The medical-school spirometer experiments mentioned at the start of this chapter all yielded the same trajectory: progressive disorientation, declining performance, loss of consciousness. Climbers near the summit of Everest describe a fog of muddled thought that has cost lives. Patients in failing cardiac care decline into delirium, not enlightenment. NDEs are the opposite of all of this. Bruce Greyson, drawing from hundreds of cases, finds that experiencers describe their thoughts during the NDE as more vivid, clearer, and faster than ordinary waking thought.¹² A failing brain, in Peter Fenwick's clinical phrase, “produces experiences which are limited, confused and disorganized. The very opposite is true of the NDE.”¹³

Third is the timing problem with endorphins. Endorphins surge quickly, but they decay slowly — over several hours. If endorphins were producing the bliss of an NDE, the bliss should fade gradually as endorphin levels drop. NDErs do not report this. They report a clean, sharp boundary: peace during the experience, pain returning the moment they're back. One of the patients quoted in the literature put it plainly: “I could feel my being rising out of the body. . . . Then I started floating back down to my body. As soon as I got down to my body, the pain came back.”¹⁴ That is not pharmacology. Pharmacology has half-lives. The pain returning the instant consciousness returns to the body is the wrong shape for any chemical explanation.

Marsh tries to soften these problems by appealing to local brain conditions we cannot measure, or by chaining together multiple mechanisms — endorphins triggering the temporal lobe, which then produces NDEs. But this just relocates the problem. The chemical-explanation story requires a chemical condition that produces lucid, coherent, structured, transformative experiences with veridical content. Nobody has identified one. That is not what dying brain chemistry does.

C. The Pro-NDE Response

Now to the positive case. The chemical explanations for NDEs do not just fall short. They fail at exactly the points where the evidence is most carefully controlled. Three lines of research show this clearly.

1. Whinnery's G-LOC studies

The single most-cited skeptical analogy for NDEs comes from fighter pilots. Pull enough Gs in a tight turn — five or six — and blood is forced down out of your head. The brain starves for oxygen. The pilot loses consciousness. The Air Force, alarmed by lost crews and lost aircraft, built a centrifuge to study this. James Whinnery, a Navy physician, has now collected data on more than a thousand episodes of this so-called G-LOC, short for “G-force induced loss of consciousness.”¹⁵

Skeptics point to Whinnery's research because some G-LOC subjects do report features that look NDE-like at a glance: tunnel vision, a sense of floating, occasional mild euphoria, even rare out-of-body sensations. A few researchers have called this a “brain-state” model of the NDE.

But look at what Whinnery actually found. The experiences are short — a few seconds. He calls them “dreamlets” and is careful to distinguish them from NDEs. They are not lucid. They are not coherent. Their content, in his own description, shows “emotional intensity, detailed sensory imagery, illogical content and organization, uncritical acceptance, and difficulty in remembering once it is over.”¹⁶ That is the description of an ordinary dream, not an NDE. Subjects report bumper cars at an amusement park, fishing, drowning, lying on a beach — random scraps stitched together.

Compare those to a typical NDE: the experiencer leaves the body, watches the resuscitation team, sees a tunnel, meets a being of light, encounters deceased relatives they recognize, undergoes a panoramic life review, makes a conscious decision to return, and remembers all of it years later with crystal clarity. Whinnery's pilots got drowsy bumper cars. NDErs got transcendent journeys remembered for life.

Whinnery himself notes that no G-LOC subject has ever reported a panoramic life review. None has met deceased relatives or a being of light. None has had the experience end with a deliberate decision to return. None has reported the lasting personality changes that follow nearly every full NDE.¹⁷ The few overlap points — floating, brief tunnel vision — are exactly what we'd expect when blood pressure drops and oxygen leaves the eyes. The disanalogies are everywhere else.

2. Sabom's blood-gas case and the Parnia study

Marsh allows that Michael Sabom documented a cardiac-arrest patient whose arterial oxygen was high — well above normal — during his NDE. The numbers were striking: pO₂ = 138, pCO₂ = 28.¹⁸ That is the opposite of what hypoxia/hypercapnia theories predict. The patient was getting plenty of oxygen, his CO₂ was actually low, and yet he had a vivid NDE during which he correctly described a femoral arterial blood draw he had no ordinary way to see. He even noticed it was “a shot in the groin” — an apt lay description of the procedure. The blood drawn for those very gas measurements was the blood drawn during the experience he described.

Marsh waves this off by suggesting that peripheral blood gases may not reflect cerebral conditions.¹⁹ That is technically true. But it cuts both ways. If we cannot trust blood gases to tell us what's happening inside the skull, we cannot trust them to confirm hypoxia either. The skeptic cannot have it both ways.

The Sabom case is not alone. In 2001, a team of British physicians led by Sam Parnia published the first prospective study of NDEs in cardiac-arrest patients in the journal Resuscitation. Sixty-three survivors were interviewed. Four had full-blown NDEs. The researchers had access to the patients' arterial blood gases drawn during the resuscitation. The result? Patients who reported NDEs had higher mean blood oxygen levels than those who did not.²⁰ Higher, not lower. The sample is too small for statistical significance, but the direction is exactly wrong for the hypoxia theory. If hypoxia caused NDEs, the NDErs should have been more oxygen-deprived than the controls. They were less.

Parnia and his colleagues went on to characterize the NDE memories themselves as “highly structured, narrative, easily recalled, and clear” — explicitly unlike the disorganized, confusional hallucinations that hypoxic brains produce.²¹

3. Endorphins, CO2, and the lucidity question

Endorphins. We've already noted that their pharmacology is wrong: onset takes time, decay takes hours. NDE pain returns instantly upon return to the body. No endorphin acts that way.

Add to that: endorphins are not hallucinogens. They are painkillers. They produce contentment, mild euphoria, an absence of pain. They do not produce structured visual narratives, life reviews, encounters with deceased relatives, or accurate visual perception of the operating room. To get from endorphins to a life review, you have to add something like the temporal lobe, the temporo-parietal junction, or some other neural mechanism — and we treat the temporal lobe argument fully in Chapter 12 and the temporo-parietal junction in Chapter 13.

Carbon dioxide narcosis is sometimes invoked from L. J. Meduna's mid-twentieth-century work. Meduna gave psychiatric patients a 30 percent CO₂ mixture and reported some out-of-body and mystical experiences. Marsh draws on this. But even Marsh notes the differences. Meduna's patients also experienced “illogical compulsions, frightening figures and shapes, and the perception of complex geometric shapes.”²² None of those are typical NDE features. Meduna's subjects also had violent muscle convulsions and saw in duplicate or triplicate — again, not features of NDEs. The thin overlap is dwarfed by loud disanalogies.

Fenwick adds a pointed clinical observation: in any modern hospital setting, allowing a patient's CO₂ to build up while resuscitation is in progress would be unacceptable. CO₂ rises do happen during cardiac arrest, but they are accompanied by oxygen drops that produce confusion and unconsciousness, not lucid OBEs. A failing brain cannot build the elaborate, coherent models that NDEs require.²³

4. Veridical perception: the showstopper

I have saved the strongest point for last. None of the chemical explanations — endorphin, hypoxia, hypercarbia, G-LOC — addresses the central evidential challenge of the NDE: veridical perception. (We treat the strongest cases in Chapter 4 and Chapter 5.) These are cases in which the experiencer accurately reports specific, otherwise unknowable details about the resuscitation, the operating room, conversations of staff, or events in distant rooms. Pam Reynolds described a bone saw she had never seen. The dentures man identified the nurse who removed his teeth. Maria saw a tennis shoe on a third-floor ledge that no one in her room could have known was there.²⁴

Here is the question for the chemical theorist: which neurotransmitter conveys accurate visual information from outside the patient's line of sight, encodes it in memory, and delivers it for verbal report after recovery? There isn't one. There never has been one. Endorphins do not generate accurate spatial perception. Hypoxia degrades it. CO₂ distorts it. None of these explains how a brain receiving no sensory input correctly reports what is happening in another room.

D. Counter-Objections

A serious skeptic might press two points.

First: “Just because we cannot currently identify a chemical mechanism doesn't mean one doesn't exist. Maybe local brain hypoxia in some specialized region produces a state we have not yet measured. Maybe a yet-unknown neurotransmitter is at work.”

That is fair as an in-principle hedge. Science is humble about what it doesn't yet know. But the burden falls on the skeptic. The scientist who claims a chemical cause owes us either the chemistry or a falsifiable model. To say “some chemistry, somewhere, somehow, produces what we see” is not science. It is a placeholder. Compare two positions. One says, “This evidence is best explained by consciousness operating outside the brain in this state, and here are dozens of veridical cases that support it.” The other says, “There must be a chemistry that does this, even though every chemistry we have studied either produces the wrong phenomenology or the opposite of it.” When the data lean this hard against the chemical hypothesis, “we'll find a chemistry someday” stops being scientific caution and starts being faith.

Second: “Maybe NDErs are confabulating their experiences during recovery, mixing fragments from various stages of consciousness into a coherent story after the fact.” This is a memory objection, and Chapter 18 owns the full response. Briefly: long-term studies, including the Dutch follow-ups led by van Lommel, show that NDE accounts remain stable for years and decades, often more stable than ordinary memories of the same period.²⁵ Confabulated memories drift. NDE memories don't.

The chemical theory, in its strongest form, is a confession of where the evidence stands. Skeptics know that something unusual is happening during cardiac arrest and other near-death events. Their best move is to look for an internal mechanism. But every mechanism so far proposed gets the phenomenology wrong, gets the timing wrong, or fails to explain the content. The lights flickering across the readout panel keep telling us the same thing: when the body breaks down, in some patients, consciousness functions at peak clarity. That is not what dying brains do. That is what something else does.

The remaining neurological objections — the temporal lobe, the temporo-parietal junction, the phantom-body argument, dreams and REM, ketamine, and the timing problem — each get their own chapter ahead. Together they form a kind of skeptical patchwork. Chapter 31 will return to ask whether the patchwork can hold together at all.