A Closer Look at Visual Life Reviews and Age in Adulthood During Near-Death Experiences

King, R. A. (2022). A Closer Look at Visual Life Reviews and Age in Adulthood During Near-Death Experiences. Journal of the Society for Psychical Research, 86(4), 209–225.

AI Assessment

An observational study of 3,394 near-death-experience accounts from the public NDERF archive, asking whether the visual life review (seeing scenes of one’s past) becomes less common with older age at the time of the experience. The author read and classified every account himself and found a small but statistically significant age gradient: the visual life review was most common in adolescents and emerging adults (16.5% at ages 10 to 25) and least common in older adults (7.9% above age 50), chi-square p < .001, Cramer’s V = .09. The strengths are a large sample, a tightly specified definition, and a candid, exploratory tone. The central limitation is that a single, non-blinded rater classified subjective narratives with no independent inter-rater check, on a self-selected retrospective online sample. This audit reports what the study found and how; it takes no position on the nature of near-death experiences.

Provenance

Publication. Journal of the Society for Psychical Research, 2022, volume 86, number 4, pages 209 to 225. This audit page draws on the article as printed in that issue.

Study type. A quantitative observational (correlational) analysis of archived self-report accounts. It is described by its author as an exploration rather than a confirmatory test.

Author. Robert A. King, an independent researcher who has published a series of papers specifically on visual life reviews.

Data source. Self-reported near-death-experience accounts posted on the Near-Death Experience Research Foundation (NDERF) website from 1999 to the end of 2021. Age-at-occurrence spreadsheets were supplied by the NDERF webmaster (Jody Long, personal communication).

Source basis. Every figure below is taken from the article’s own Abstract, Methods, and Results (including Table 1). A cataloguing note, surfaced during this audit: in the reference database the underlying file is a full-issue scan whose stored title and author were mislabeled as a different article in the same issue (Radin’s plasma paper); the correct article, author, and page range for this audit are King (2022), pages 209 to 225, and every figure here is grounded in King’s own text.

What the paper reports

The visual life review experience (VLRE), as the author defines it, is an involuntary revival of multiple self-inclusive visual scenes from one’s own life, appearing rapidly or uncontrollably, of the kind sometimes reported during near-death experiences.¹ Earlier studies had disagreed about whether life reviews are related to age: some found no association (Greyson, 1983; Sabom, 1982), while others hinted that younger people report them more often (Noyes & Kletti, 1977; Katz et al., 2017). Using a much larger sample, this study tests specifically whether the visual life review becomes less frequent as the age at the time of the experience increases.

The results of that exploration and analysis found that a statistically significant negative association does exist. Although the implications of this finding are uncertain, the paper points out a number of different possibilities for preliminary consideration.

How it was run

• Sample. The author read and examined 3,394 self-reported NDEs posted on NDERF from 1999 to the end of 2021, restricted to accounts specifically categorized as NDEs (not “possible,” “probable,” shared, or second-hand) and to experients aged 3 or older. Duplicate re-submissions of the same incident and content-empty accounts were removed.
• Classification. Each account was classified as VLRE present, absent, or uncertain based solely on the narrative and the respondent’s own written answers, not on the questionnaire’s forced-choice life-review item (which the author argues can produce false positives). Careful exclusion rules were applied: non-visual memory revivals, discussions with NDE figures without visual scenes, past-life or future imagery, and non-self-inclusive imagery were all excluded.
• Age coding. Only cases with a specific or reasonably approximate age at the time of the NDE were included; ranges were resolved by a stated rule; children under 3 were excluded on childhood-amnesia grounds (and none reporting a VLRE were found under 3 in any case).
• Analysis. A Cochran-Armitage trend test across individual ages, then a Pearson chi-square across four age groups (children 3 to 9; adolescents/emerging adults 10 to 25; adults 26 to 49; older adults above 50), with Cramer’s V as the effect size and Fisher’s exact tests for pairwise group comparisons.

Results, as reported

Values are reproduced from the article’s Results section and Table 1. The direction of the effect is a decline in visual life reviews with older age at occurrence, peaking in adolescence and early adulthood; the effect is statistically clear but small in magnitude (Cramer’s V = 0.09).

Eleven-dimension audit

Pre-registration

Not preregistered, and the author explicitly frames the work as an exploration rather than a confirmatory test. The primary direction (a negative age-VLRE association) was stated in advance and motivated by prior findings, but the four age-group boundaries were chosen partly from where the percentages began to differ in these data, which is a data-dependent decision the reader should weigh.

Randomization

Not applicable in the experimental sense: this is an observational analysis of an existing archive with no random assignment. The sample is self-selected (people who chose to submit an account to NDERF), which is the appropriate thing to flag here rather than randomization of conditions.

Sensory leakage

Not applicable; this is not a test of anomalous perception, so the ganzfeld-style leakage concern does not arise. The analogous validity threat, classification bias, is treated under Blinding.

Blinding

This is the study’s central weakness and it is not concealed. A single rater, the author, classified all 3,394 narratives, and there is no independent, blinded second coder and no reported inter-rater reliability statistic. The author mitigates this with a tightly specified VLRE definition, explicit exclusion rules, and worked examples of included, excluded, and uncertain cases, but a reader cannot independently verify the coding, and the rater was not blind to the study’s hypothesis.

Optional stopping

Not a concern: the analysis used all qualifying NDERF accounts through a fixed calendar cutoff (end of 2021) rather than stopping when a result looked favourable.

Outcome measure

The outcome is a binary VLRE present/absent judgment against a clear operational definition (multiple self-inclusive visual scenes, involuntary and rapid), deliberately based on the free narrative rather than the questionnaire’s forced-choice item, which the author argues over-counts. The measure is well specified; its weakness is that it remains a subjective coding of self-reported text.

Effect size

Small: Cramer’s V = 0.09. The age gradient is real and orderly (16.5% at 10 to 25 down to 7.9% above 50), but it is a modest effect, and the author does not overstate it, describing the implications as “uncertain” and the work as preliminary.

Multiple comparisons

Several tests were run: a trend test, an omnibus chi-square, and multiple pairwise Fisher’s exact comparisons. The omnibus result is strong (p < 0.001), but at least one pairwise contrast is only marginally significant (p = 0.018), and the paper does not describe a formal correction for the pairwise family, so those secondary contrasts should be read cautiously.

Internal replication

There is no internal replication or hold-out sample; the finding rests on a single analysis of a single dataset. The orderly monotonic pattern across four independent age bands provides some internal coherence, but not replication.

External replication

The result sits in a mixed prior literature. It aligns with hints from Noyes and Kletti (1977) and with Katz et al. (2017), who reported a negative age association for life-review-type experiences, but it runs against earlier NDE studies (Greyson, 1983; Sabom, 1982) that found no such association. The author attributes the difference to a larger sample and a tighter, visual-specific definition.

Transparency

Reasonable: the data source is a public archive, the age spreadsheets and inclusion rules are described, worked examples of the coding are given, and an appendix reports the age-level detail. The transparency limits are inherent to the design, self-selected retrospective self-reports, sometimes-approximate ages, and, most importantly, a single unverified coder, all of which the paper acknowledges.

The adversarial record

• One coder, no reliability check. The strongest objection is that the key variable, whether a narrative describes a visual life review, was judged by one non-blinded rater with no inter-rater agreement statistic. A critic can reasonably ask how stable the 432 VLRE classifications would be under an independent coder.
• The sample. NDERF accounts are self-selected, retrospective, and unverified; ages at occurrence are self-reported and sometimes approximate. Any prevalence figure (including the 12.7% overall) is a property of this convenience archive, not of near-death experiences in general.
• Small and exploratory. With Cramer’s V = 0.09 and no preregistration or correction for the pairwise tests, this is a small exploratory effect. Its value is as a hypothesis-generating observation, which is how the author presents it.
• What the author does well. Rather than claim the finding proves a neurocognitive basis for the life review, the paper lists several competing explanations, foregrounds the uncertainty, and calls for further research. That restraint is appropriate to the strength of the evidence, and it is what keeps a modest observational result honest.