Independence of Consciousness: Non-Local Mind Models and Reported Perceptual Anomalies

Scope and status

This draft summarizes a supplied corpus-scale aggregate on reported near-death, end-of-life, and related perceptual anomalies, and on how those reports are classified relative to non-local mind models versus neural-mechanism accounts. The dataset field arxiv_source_found is true for the corpus-aggregate description. That flag does not mean the independence of consciousness is an established finding.

The supplied data_type is meta_analysis_illustrative. The accompanying model note states that 88 abstracts were statistically classified by an LLM into four labels (supports_veridical, null_result, neuro_mechanism, inconclusive). The note describes a corpus-scale aggregate, not a single-paper effect size and not a hand-verified systematic review.

Boundary status in the dataset is reported as CORPUS_ANALYSIS_COMPLETE.

Reported corpus metrics

The dataset reports a classified total of 88 records, matching the stated corpus size of 88. Raw tallies as supplied: 3 supports_veridical, 0 null_result, 8 neuro_mechanism, and 77 inconclusive. Corresponding percentages as supplied: 3.4%, 0.0%, 9.1%, and 87.5%. The arXiv-context string in the dataset restates the same percentages for 88 deduplicated sources attributed to Semantic Scholar and PubMed.

Draft reading of the aggregate (attributive, not settled)

On the numbers as supplied, the large majority of the corpus (87.5%; 77 of 88) is labeled inconclusive. That distribution is consistent with a retrieval set in which most abstracts do not clearly support a veridical, non-local interpretation and also do not clearly isolate a single neural mechanism. It is not a basis for treating non-local mind models as shown.

A small slice (3.4%; 3 of 88) is labeled supports_veridical in the dataset. The three titles so labeled are: a 2026 case discussion of a distressing near-death experience with reported veridical perceptions during coma; a 2025 scoping review of end-of-life experiences and implications for the mind-brain relationship; and a 2013 case study of terminal lucidity (the death of Anna Katharina Ehmer). Those labels are LLM classifications of abstracts in the supplied list. This draft does not treat the labels as verified evidence that perception occurred independently of brain function.

Another slice (9.1%; 8 of 88) is labeled neuro_mechanism. Titles in that bin include work on brain “twilight zones,” associations of near-death experiences with REM-sleep intrusion and with migraine aura, a 2006 paper asking whether the arousal system contributes to near-death experience, a 2026 “dying-moment dream” hypothesis, a comparative analysis of a near-death experience and a subsequent 5-MeO-DMT experience, and a large-scale semantic-similarity study of neurochemical models of near-death reports. Those items are consistent with accounts that locate anomalous perception in known or candidate neural and neurochemical processes. They do not, on the supplied metadata alone, rule out other models, and they are not here described as conclusive.

The dataset reports 0.0% null_result (0 of 88). In an illustrative LLM-labeled corpus, a zero count of null results is consistent with retrieval, labeling, or category-boundary effects. It should not be read as an empirical demonstration that null findings are absent from the wider literature.

Competing interpretations therefore remain open in this draft:

• Non-local or independence models would expect at least some well-documented veridical perception when cortical function is severely reduced. The supplied corpus contains only three abstracts labeled in that direction, and those labels are unverified.
• Neural-substrate models in the labeled set include hypoxia-related transitional brain states, REM intrusion, migraine aura, arousal-system activity, a terminal dream hypothesis, and overlap with psychedelic phenomenology. The eight neuro_mechanism labels are also unverified as to the correctness of the bin.
• The skeptic critique in the dataset states that a pattern of reported near-death perceptions does not by itself distinguish those families of explanation.

A further caution, still without new numbers: many titles in the inventory appear only loosely related to independence of consciousness (for example, veterinary stunning, school-zone speed, clavicular anatomy, ovine coenurosis, or oncology case reports). That pattern is consistent with noisy keyword retrieval later marked inconclusive. Human review would need to decide which records belong in a target corpus at all.

Publication years are reproduced as listed, including multiple 2026 dates. Those dates are unverified and require fact-checking before any downstream use.

Source inventory

Access status is taken only from the supplied records: each item lists a PubMed URL and source: pubmed. This draft does not claim that full text was obtained or that link persistence was checked. Non-English or bracketed titles from the input are given in English.

Items for human fact-check before any later version

1. Re-label or exclude records after reading full abstracts or papers; do not rely on the LLM bins.
2. Verify every bibliographic year, title, and PubMed record, including future-dated (2026) entries.
3. Separate on-topic consciousness and near-death reports from incidental keyword hits.
4. Treat the 3.4% / 0.0% / 9.1% / 87.5% split as an illustrative corpus signal only, in line with the supplied skeptic critique.
5. Do not convert this draft into a claim that consciousness is independent of the brain, or that neural models are sufficient, until a protocol-driven, human-coded review exists.