Outcomes, with their limits attached

The result: In 4,354 NDERF records, four questionnaire answers track depth scores in both the older and newer halves of the archive: saying you gained information about God, saying you gained special knowledge, reporting MORE consciousness than in normal life during the event, and still believing today the experience was real. And one clean near-zero: people who had heard of near-death experiences before their own score exactly as deep as people who never had (correlation +0.002 across 3,239 records). All of this is exploratory, not confirmed: the adversarial review found an ambiguity in one statistical control and a rule fix made after results were open, which caps every label until a repaired rerun on fresh records.

What it means: What people say they brought back from the experience tracks how deep it scored, and the cultural-script worry took a real hit: knowing the NDE story beforehand gives exactly zero extra depth. If people were retelling a script they had learned, the ones who knew it should score deeper, and they do not.

What it is NOT: Not confirmed findings, because the review caught an ambiguous control rule and an after-the-fact label fix, which by this lab's own rules caps everything at exploratory until a clean rerun. Not proof the experiences are real or accurate, because every measure is what one person says about themselves in one sitting. Not four independent discoveries, because two of the four answers partly overlap what the depth score already asks about.

Next test that could settle it: A repaired rerun with the control formula frozen exactly and decision rules tested on fake data first, judged on NDERF records submitted after 2026-08-15 that no one has seen yet

Open question: Do the four exploratory depth-linked answers survive a repaired frozen rerun judged on NDERF records submitted after 2026-08-15?

The result: A reproducible slow-wave correlate appeared but is most consistent with arousal or sleep depth.

What it means: A brain-wave correlate of dream reporting does reproduce, but it behaves like ordinary arousal or sleep depth rather than a signature of the dream itself. A promising signal, most likely explained by the boring variable.

What it is NOT: Not a dead end, because absolute-power and source-localized replication with stage controls could still separate dreaming from depth. Not a confirmed dream marker, because the boring explanation currently fits.

Next test that could settle it: Absolute power and source-localized replication with stage/time controls.

Open question: Is the dream-report EEG correlate more than arousal or sleep depth?

The result: NDE and dream-report corpora show high visual-light/altered-state proxies while UAP report corpora show visual-light without the same internal-state framing; this suggests a separable altered-state imagery mechanism rather than a single generic light-reporting effect.

What it means: Bright-light imagery shows up in two different ways: near-death and dream stories wrap the light in language about their own inner state, while UFO sighting reports mention light without that inner framing. The light in a vision and the light in a sighting may be two different kinds of report.

What it is NOT: Not a confirmed mechanism, because the labels came from automated text matching that has not been independently validated. Not a claim about what anyone actually saw, because these are word patterns in stories.

Next test that could settle it: Re-run with source-locked human/LLM labels and matched report-purpose controls.

Open question: What survives when the exploratory light separation pattern is tested on validated held-out labels?

The result: Across NDE and dream-report corpora, sound, motion, and physiological proxies suggest a bounded sensory-reduction transition pattern: internally generated perception becomes more salient when ordinary sensory anchoring is reduced. Ganzfeld/protocol-experiment material is excluded from canonical support and retained only as experimental follow-up pending EP-PROTO controls.

What it means: Across near-death and dream stories, sound, motion, and body signals point the same direction: when ordinary sensory input drops away, internally generated perception takes over and becomes the story. This fits sleep onset, sensory deprivation, and near-death settings alike.

What it is NOT: Not proof of one shared mechanism, because the pattern comes from word proxies that have not passed a matched control test across settings. Not a statement about experiment data, because protocol studies were deliberately excluded here.

Next test that could settle it: Run the matched Ganzfeld/hypnosis/dream/sleep-paralysis control test.

Open question: Does a sensory-reduction pattern survive matched Ganzfeld, hypnosis, dream, and sleep-paralysis controls?

The result: Full-registry rows suggest NDE claims cluster death/afterlife, memory, and peace/love semantics differently from dream rows, implying that transition meaning rather than imagery alone may separate NDE-like material from ordinary dreams.

What it means: What separates near-death stories from ordinary dreams may not be the pictures but the meaning: death, afterlife, memory, and peace-and-love themes cluster differently in near-death material than in dream archives.

What it is NOT: Not a validated separator yet, because the meaning labels came from automated matching and the locked comparison test has not run. Not a claim that the experiences differ in nature, because this measures how they are told.

Next test that could settle it: Run the locked NDE/DreamBank meaning-communication contrast.

Open question: Do NDE and DreamBank meaning/communication contrasts survive source, length, and ontology controls?

The result: G-LOC and acute-hypoxia narratives suggest a recoverable transition subtype: sensory/motor fragments and dreamlets can occur without the full canonical NDE package, especially without prominent life review or deceased-relative encounter.

What it means: When fighter pilots black out from G-forces, they report dreamlets and sensory fragments but usually not the full near-death package with life review and deceased relatives. Brief oxygen loss alone seems to produce a lighter version, which makes the full package more interesting.

What it is NOT: Not a controlled comparison, because the blackout accounts come from scattered sources without matched denominators. Not proof the full package needs something beyond oxygen loss, because duration and context differ wildly between a centrifuge and a cardiac arrest.

Next test that could settle it: Acquire raw/prospective G-LOC or syncope reports with controls.

Open question: Do G-LOC, syncope, coma, anesthesia, sleep-paralysis, and NDE data preserve trigger-specific body-boundary patterns?