Chapter 15

Dreams, Hallucinations, and Altered States

Tell an NDE experiencer that what happened to her was just a dream. Watch the face change. Not because you have offended her — though you might have. Because she has already considered the possibility. She has had vivid dreams. She has had hallucinations. Whatever happened on the operating table or the pavement was nothing like either one.

One man in a published interview was emphatic. He looked his interviewer in the eye and said, “It was real — as real as me sitting across from you and talking to you now. Nothing could ever convince me otherwise.”¹

That is the strange, stubborn fact about near-death experiences. Years pass. Memories of the surrounding events fade. The NDE doesn’t fade. If anything, it sharpens. People who had their NDE forty years ago describe it like it happened last week. They make life decisions around it. They lose friends over it. The one thing they almost never say is, “I had a really vivid dream once.”

But what if they are wrong? What if their certainty is itself a trick of biology — a feature, not a bug, of a brain temporarily playing the wrong tune? That, in essence, is Michael Marsh’s view. It is, in a more aggressive form, the view of neurologist Kevin Nelson. The dream theory is the most popular naturalistic explanation of NDEs in print. So let’s give it a careful answer.

Marsh’s Sleep Argument

In Chapter 7 of his book, Marsh leaves the world of out-of-body sensations and moves to the world of sleep. He titles the chapter “Falling Asleep, Perchance to Dream — Thence to Reawaken,” and he is upfront about his strategy. “I continue my neurological theme relating to the realms of conscious-awareness and its disturbances, and their relevance to ECEs,” he writes (p. 128).² The case unfolds in three layers.

First, brain-scanning shows that certain regions of the brain shut down while we dream. The dorsolateral prefrontal cortex — the part most tied to wakeful self-monitoring — goes dark. Yet the dreaming brain still cooks up rich, vivid mental events. If the dreaming brain can produce those experiences while parts of itself are offline, Marsh argues, why couldn’t a recovering brain — one that has just been through cardiac arrest — do the same? The functional disconnection looks comparable. The output may be too.³

Second, Marsh points to hypnagogic states (the strange perceptions that happen as you drift off to sleep) and hypnopompic states (the same kind of thing as you wake up). He calls these states “closely reminiscent of the ECE phenomenologies” (p. 128).⁴ Both involve floating sensations, sudden lights, voices, vivid scenes that feel meaningful. If the brain can manufacture all that as it idles at the edge of sleep, why not while resuscitating?

Third, Marsh turns to sleep paralysis. This is the condition in which the body cannot move but the mind is alert. Researchers have classified sleep paralysis into three types: intrusive (sensed presence, often a hooded figure), incubus (chest pressure, suffocation, choking), and vestibulo-motor — floating, flying, the sense of leaving the body.⁵ It is that third type that does the work for Marsh. Sleep paralysis can produce out-of-body sensations. Marsh concludes flatly: such episodes happen “in the mind, and, most certainly, not out-of-body” (p. 142).⁶ He thinks the same applies to the OBE that begins many NDEs.

To Marsh’s credit, he sees the obvious objection. Experiencers insist their NDEs are real, not dreams. Marsh nods to that on page 129. But he says the response misses the point: dreams during a dream feel real too. Your NDE feels real to you the way my dream of falling off a building feels real to me — until I wake up.⁷

Kevin Nelson, a neurologist at the University of Kentucky, presses the same case harder. In The Spiritual Doorway in the Brain, Nelson argues that NDEs occur when “part of the dreaming brain erupts in a brain already awake. And blending REM with waking consciousness creates experiences that are realistic and memorable.”⁸ This is the REM-intrusion hypothesis. The brainstem fires REM patterns into waking awareness, and the brain interprets the result as reality. Fischer and Mitchell-Yellin commend Nelson’s general approach as a fine example of multi-factor naturalistic explanation.⁹

Put simply: Marsh sees a family resemblance between NDEs and the strange events of the sleep cycle. Nelson goes further — NDEs are a kind of dream, a misfired one. Either way, no soul leaves the body. There is only the brain, doing what brains sometimes do at the edge.

Where the Argument Breaks Down

Both Marsh and Nelson have done their neurology homework, and that deserves real respect. But the dream theory has three problems serious enough to make a careful reader pause.

The first problem is that the analogy doesn’t fit. Dreams are messy. They jump around. They forget themselves halfway through. They are full of impossible spatial transitions, faces that change mid-sentence, plotlines that vanish into other plotlines. Anyone who has kept a dream journal for a week knows what dreams are like. NDEs are the opposite. They have a clear arc — beginning, middle, closure. They contain consistent elements across thousands of independently collected cases: the out-of-body view of one’s own body, the dark passage, the light, the encounter, the life review, the boundary, the return.¹⁰ Jeffrey Long’s analysis of more than 600 NDEs notes that the elements unfold in the same logical order across age groups and continents, while dreams “often skip around and lack a logical flow.”¹¹ Marsh’s analogy compares apples to a roller coaster and calls it close enough.

The second problem is biological. REM intrusion requires REM. REM requires a functioning brainstem with characteristic patterns of cholinergic activity, monoamine fluctuations, and detectable cortical processing. None of that happens during cardiac arrest. Within roughly ten to twenty seconds of cardiac arrest, EEG activity in the cortex flatlines, and brainstem reflexes are lost.¹² The brain is not running REM. The brain is not running anything our instruments can pick up. Nelson’s REM-intrusion model needs a window of dream-capable brain activity — but cardiac-arrest NDEs occur precisely where that window is closed. Chapter 17 takes up the timing question in detail. For now, the point is simple. You can’t have a REM-intrusion episode in a brain that isn’t producing measurable electrical activity.

The third problem is the most decisive — and Marsh, oddly, brushes past it. Dreams do not produce verifiable perception of real events outside the dreamer’s body. Hypnagogic states don’t either. Neither do hypnopompic states. Neither does sleep paralysis. The veridical core of NDE evidence — the cases where the experiencer accurately reports a conversation, a piece of medical equipment, the position of a colleague, the contents of a hallway, an event in another room — has no parallel anywhere in the sleep literature. Marsh’s own sleep-paralysis subjects (p. 141) sometimes realize they are dreaming because the room is wrong: an alarm clock has no LED, or the dreamer notices she is wearing pajamas she has never owned.¹³ Their OBEs are obviously internal. NDE OBEs are obviously not. We will see in a moment why this matters more than anything else.

I’ll note in passing a methodology problem in Nelson’s own data. Nelson’s much-publicized study used 55 subjects recruited through Jeffrey Long’s NDERF site — a self-selected pool already deeply interested in NDEs.¹⁴ The survey did not distinguish paranormal experiences before the NDE from those after the NDE, even though research shows people often report new ongoing experiences after an NDE.¹⁵ A careful evaluator of the REM-intrusion hypothesis is not going to bet the house on those 55 surveys, especially when Long’s broader database of more than 3,000 NDE accounts tells a more textured story than Nelson’s narrow frame can hold.

Why NDEs Are Not Dreams

The dream hypothesis would work only if NDEs behaved like dreams in their feel, their shape, their biology, and their relationship to the outside world. They don’t behave like dreams in any of those four ways. Take them one at a time.

Real, more real, realer than real

The first problem for the dream theory is that the people who would know say it is wrong. NDErs say their experiences were not dreams. They are explicit about this. They are also persistent — they keep saying it, even when researchers push back. And many have vivid dreams, lucid dreams, and drug hallucinations to measure against.

Bruce Greyson, the psychiatrist who developed the standard NDE Scale, has noted again and again that NDErs rule out dreams and hallucinations as candidate explanations. So has Michael Sabom, the cardiologist whose Recollections of Death was the first major prospective study in the field. Sabom puts it cleanly: the NDE is perceived as stark reality both during the experience and later in reflection, and the variability of dreams from person to person and night to night sits in pointed contrast to the consistency of NDE elements.¹⁶

Long’s NDERF data sharpens the point. Of more than 600 NDErs surveyed in detail, over 95 percent described their experience as “definitely real.” Virtually all of the rest described it as “probably real.” Not one said it was definitely not real.¹⁷ One NDEr told J. Steve Miller that his NDE was nothing like the hallucinations he had had on codeine in the hospital — nothing like them at all.¹⁸ Another said simply that it was more real than what we call reality.¹⁹

Marsh’s response — that dreams feel real too while you’re in them — does not engage this. The point is not that NDErs experienced their NDEs as real during the NDE. Hallucinators experience their hallucinations as real, too. The point is that NDErs who have also had vivid dreams, lucid dreams, and drug-induced hallucinations consistently distinguish the NDE as something else, and they keep distinguishing it for the rest of their lives. They never wake up the next morning and say, “What a strange dream that was.” They say, instead, “I died, and I came back, and I am not the same.”

The shape of the story

Dreams skip. They drop loose ends. They cut from a wedding to a parking lot to a bird that becomes your high school principal. Anyone who has tried to write a dream down in the morning has watched the story dissolve as soon as you tried to make it stay.

NDEs are the opposite. They are tightly structured. They unfold in a recognizable arc — initial awareness of being out of the body, a passage of some kind, a place of overwhelming light, an encounter, sometimes a life review, a sense of being told it is not yet time, and a return. Different cultures color the encounter differently, but the structure recurs. Greyson’s NDE Scale, applied across thousands of cases, shows striking consistency.²⁰

This consistency is exactly what we should not expect if the experiences are dreams. Dreams are profoundly idiosyncratic. Two people taking the same medication don’t have the same dream. Two siblings raised in the same culture don’t dream alike on the same Tuesday. Yet a Brazilian construction worker who flatlines in 2018 and a Welsh schoolteacher who flatlined in 1979 will report experiences that overlap in startling ways — even when neither person had read about NDEs beforehand.²¹ A theory that says these experiences are dreams owes us an account of that consistency. None has been forthcoming.

The body knows the difference

Now the biology. To have a vivid REM dream, you need a specific brain state: a working brainstem, certain neurochemistry, a particular electrical pattern in the cortex. None of that is available during cardiac arrest. Within ten to twenty seconds, the cortex shows a flat EEG. Brainstem reflexes are lost.²² The patient is not in REM. The patient is, by every electrical measure available to medicine, not in any state that should be capable of producing organized experience at all.

And yet patients in cardiac arrest produce highly organized experience. Pim van Lommel’s prospective Dutch study, published in The Lancet in 2001, found that cardiac-arrest survivors with NDEs were no different demographically from those without — same age range, same education, same religious background — but their reports were vivid, structured, and life-changing.²³ Sam Parnia’s AWARE studies have documented patients accurately recalling resuscitation events that occurred while their cortex should have been silent.²⁴

Wilder Penfield, who probably did more brain stimulation work than anyone else in the twentieth century, came to a striking conclusion at the end of his career. After a lifetime of trying to explain mental life by brain action alone, Penfield concluded that it is simpler and more logical to adopt the hypothesis that human beings consist of two fundamental elements rather than one.²⁵ The cardiac-arrest data tugs in that same direction. The brain is not generating these experiences. The brain is barely doing anything during the experiences that matter most.

The decisive test — real events

Now the argument that ends the dream debate, in my judgment, all by itself.

Dreams do not produce verifiable perception of real events outside the dreamer’s body. Hallucinations don’t either. NDEs do.

Chapter 4 covers the strongest cases in detail. There’s the man with the dentures, whose nurse hung the missing teeth on the cart and watched a comatose patient identify her by sight a week later, calmly, having “seen” her place them there. There’s Maria’s tennis shoe, perched on a third-floor ledge in a Seattle hospital, described accurately by a woman who could not have walked there to look. Chapter 5 takes Pam Reynolds apart in detail — a patient describing surgical instruments and conversations during a procedure that flatlined her brainstem and rendered her ears nonfunctional.

But take a quieter case from Sabom’s work. Mrs. M., a 60-year-old homemaker, went into cardiac arrest in her hospital room one January morning. During her NDE she watched the resuscitation. Sabom, a cardiologist, later compared her account to the medical record. She had described the equipment correctly. She had described the chest compressions. She had described how her eyelids were lifted, how a blood sample was drawn from her hand — a detail recorded in the chart. She had heard a doctor say she was being transferred to the ICU, and she had watched a young nurse pack her belongings in bags for that transfer. The transfer never actually happened. But the nurse really had packed the bags.²⁶

Or consider an AWARE study case. A 57-year-old social worker collapsed in a Southampton hospital. He went into ventricular fibrillation. During his NDE he watched the resuscitation team try to defibrillate him twice. He could hear an automated voice saying, “Shock the patient, shock the patient.” The medical record confirmed that an Automated External Defibrillator had been used — twice. The AED’s voice prompts at intervals corresponding to roughly three minutes, meaning the patient’s consciousness was operating during a stretch of time when his cortex should have been silent.²⁷

These cases are not unicorns. The Self Does Not Die catalogs more than a hundred verified veridical NDEs.²⁸ Jan Holden’s analysis in the Handbook of Near-Death Experiences found that 92 percent of veridical OBE perceptions reported in the literature were entirely accurate, with another 6 percent containing some inaccuracy.²⁹ No dream theory accounts for that distribution.

So here is the question I would ask Marsh, gently. In the hypnagogic, hypnopompic, sleep-paralysis literature you draw on so carefully — where are the cases? Where is the sleep-paralysis subject who accurately described the conversation in the kitchen below her bedroom? Where is the lucid dreamer whose dream of the conference room matched the conference room?

There aren’t any. That’s the answer. The dream literature has nothing analogous to offer. The reason it has nothing to offer is that dreams are produced by, and confined to, the dreaming brain. NDEs aren’t.

Two Objections, Briefly Answered

A careful skeptic will press back on what I have said. Let me name two of the strongest pushbacks and answer each.

First: but what about lucid dreams? Lucid dreams are dreams in which the dreamer knows she is dreaming and can sometimes shape the content. They feel real. They have narrative structure. Doesn’t this prove that a dream-state can produce a vivid, structured, realer-than-real experience?

It would prove that — except for the veridical perception problem. Lucid dreams happen in the dreamer’s bedroom. They do not include accurate perception of real events outside the body. NDErs who have had lucid dreams are clear about this. The lucid dream is, quite obviously, a creation of the dreaming brain. The veridical NDE includes elements that the experiencer’s brain could not have produced without help from somewhere outside the body.

Second: maybe NDEs are an unknown special form of dream-state mentation that we just haven’t characterized yet. Maybe we shouldn’t rule it out, even if no current dream theory fits.

This is a perfectly fair scientific stance — until you notice what it costs. To call NDEs “an unknown special dream-state” is just to give the mystery a name. We have no neurological description of such a state. We have no electrical signature for it. We have no candidate mechanism by which a flatlined brain would generate accurate descriptions of events outside the body. Calling the phenomenon a special dream is rebranding, not explaining. And it leaves the veridical perception untouched. A cortex that is electrically silent does not generate dreams of any kind. A dream — special or otherwise — does not see the AED in the corner of the room and identify whose voice prompted the second shock.

The dream theory fails on four levels at once. Phenomenologically, NDErs say it isn’t a dream — and the people most qualified to judge are those who have had every variety of vivid dream and hallucination against which to measure. Structurally, NDEs have the consistent, logical narrative that dreams notoriously lack. Biologically, the cardiac-arrest brain is not equipped to generate dream-state experience. And evidentially, dreams cannot do what veridical NDEs do — perceive real events at a real time in a real place. Marsh’s neurology is good. The analogy is not. In the next chapter we will see whether his other major candidate — the ketamine experience — fares any better.