Consciousness and the double-slit interference pattern: Six experiments

Radin, D., Michel, L., Galdamez, K., Wendland, P., Rickenbach, R., & Delorme, A. (2012). Consciousness and the double-slit interference pattern: Six experiments. Physics Essays, 25(2), 157–171. https://doi.org/10.4006/0836-1398-25.2.157

AI Assessment

The figures on this page match the published article. This study drills down to the primary data behind Dean Radin‘s double-slit work. The paper pools six experiments (137 participants, 250 sessions) and reports that a photonic interference measure shifted in the predicted direction when observers focused attention toward a double slit, at a combined 4.4 standard deviations, while 250 no-observer control sessions showed nothing. ESP-Nexus audits whether this page faithfully reports what the article states; it takes no position on whether the effect is genuine, and the record below lays out the honest weaknesses (post hoc analyses, the author’s own later null, and the absence of independent replication).

Provenance

DOI. 10.4006/0836-1398-25.2.157 · Published in Physics Essays, a peer-reviewed journal; the full text is held in the ESP-Nexus library.

Study type. Empirical series of six experiments (not a meta-analysis), each testing a consciousness-related interpretation of the quantum measurement problem on an optical double-slit system.

Funding. Bial Foundation; the Fetzer-Franklin Fund of the John E. Fetzer Memorial Trust; the Mental Insight Foundation; the Federico and Elvia Faggin Foundation; Dick and Connie Adams; and the HESA Institute. Disclosed in the article.

Data availability. Per-experiment summary statistics are given in the article’s Tables I to VI; combined results appear in Table V and the figures. No per-session data table is published in the text.

Source basis. Every figure below was confirmed against the primary article (abstract p. 157; Methods and Results pp. 157–168; Discussion pp. 168–170; Tables I to VI).

What the paper reports

The authors define a spectral ratio R, the double-slit interference power divided by the single-slit diffraction power, and predict that R declines when a person focuses attention toward the double slit versus away from it. Across six experiments, data contributed by 137 people in 250 test sessions showed that on average R decreased as predicted, a combined z of -4.36 (p = 6 x 10^-6), which the authors describe as a 4.4 standard deviation effect in the predicted direction.¹ A separate set of 250 control sessions run with no observer present, to test the hardware, software, and analysis for artifacts, showed no effect (z = 0.43, p = 0.67). Participants trained in meditation produced stronger effects than non-meditators, and several consciousness-related measures (electrocortical markers of focused attention, openness, and absorption) correlated with R in the predicted directions.

The authors state that “more research will be required to confirm, systematically replicate, and extend these findings,” and flag several of the supporting analyses as post hoc.

How it was run

• Six experiments run at the Institute of Noetic Sciences, with one off-site meditator subset, using a laser, a physical double slit, and a line-scan camera that recorded the interference pattern.
• Each session comprised roughly 40 counterbalanced epochs, alternating “attention toward the slits” and “attention away,” each epoch lasting 15 to 30 seconds.
• The outcome was the differential spectral ratio DR, the difference in mean R between attention-toward and attention-away epochs, converted to a z score by a 5000-iteration circular-shift permutation test per session; sessions were combined by weighted Stouffer z.
• Effect size was reported as es = z / sqrt(N), where N is the number of sessions.
• 137 participants across 250 experimental sessions, with 250 matched no-observer control sessions.
• Controls and artifact tests included the no-observer sessions, four thermocouples for radiant-heat (Experiment 3), headphone-delivered instructions to remove spoken-sound vibration (Experiments 5 and 6), a shielded chamber with EMI filtering, and detrending analyses for signal drift.

Results, as reported

All p values are one-tailed, matching the paper’s directional hypothesis. Effect sizes are the paper’s own z/sqrt(N) metric. The article does not report confidence intervals for the combined effect; the “-2 s lag” variants and per-cell peak and trough measures reported in the source are omitted here for readability and are available in the article’s Tables I to VI.

Eleven-dimension audit

Pre-registration

Mixed. The principal hypothesis (R declines with attention toward the slits, predicting a negative z) and the meditator-versus-non-meditator prediction were stated in advance, and the primary sessions were preplanned. Several supporting analyses, including the alpha-ERD correlation and one meditation-years correlation, are explicitly labelled post hoc by the authors, and 29 of the later sessions were outside the preplanned design and included “for completeness.” There is no public preregistration record.

Randomization

The attention-toward and attention-away epochs were counterbalanced within each session, and significance was assessed by a 5000-iteration circular-shift permutation test rather than a parametric assumption, which guards against certain baseline-structure artifacts. Assignment of the counterbalancing order is described as controlled by the acquisition software.

Sensory leakage

The dependent variable is an optical measure of a physical apparatus, not a human judgment, so classical sensory leakage between sender and receiver does not apply. The relevant physical-channel concerns (heat, vibration, drift) were addressed directly: Experiment 3 placed four thermocouples around the apparatus and found no significant temperature differences even in a strong-effect subset, and Experiments 5 and 6 replaced spoken instructions with headphones to remove announcement vibration.

Blinding

The apparatus and analysis were blind to condition in the sense that the permutation test was automated; the 250 no-observer control sessions and the Experiment 4 retrocausal design (data recorded unobserved, condition labels generated months later) are the paper’s strongest blinding-type controls. Participants necessarily knew their own attention condition, which is intrinsic to the manipulation.

Optional stopping

Primary session counts were preplanned per experiment. The inclusion of 29 additional exploratory sessions “for completeness” is disclosed and is a place where a reader should apply caution, though these are a minority of the 250-session database.

Outcome measure

The primary outcome is the differential spectral ratio DR, a well-defined transform of the recorded interference pattern (R = double-slit power divided by single-slit power). It is objective and computed identically for experimental and control data. Secondary peak-height and trough-height measures are exploratory corollaries.

Effect size

Reported as es = z / sqrt(N). The combined effect is small in absolute terms; the single strongest experiment (Experiment 5) reached es = -0.63, while the first four experiments combined were es = -0.20, and the meditator subgroups carried most of the signal. The paper reports z and p but not confidence intervals for the pooled effect.

Multiple comparisons

This is the paper’s main methodological exposure. Across six experiments the authors report many subgroup, lag-variant, correlation, and environmental-covariate tests. Some supporting results are significant only after a post hoc step (for example, the alpha-ERD correlation reaches p = 0.004 only after removing 19 noisy-EEG sessions, having been p = 0.09 on all 50). The headline pooled z, by contrast, is a single prespecified combination.

Internal replication

Six experiments within the one paper give partial internal replication: the predicted direction recurs across most experiments and is strongest for meditators, but individual all-sessions results range from clearly significant (Experiment 5, z = -4.48) to null (the off-site Zen-temple subset, z = -0.05; Experiment 6 all-sessions, z = -1.21). The consistency is in the meditator subgroups more than in the unselected samples.

External replication

Independent, non-author replication is absent. The same group’s later work is mixed: a 2016 distant-observation study reported a fringe-visibility effect,² while a 2015 single-photon double-slit series by the same authors found no systematic effect.⁴ An independent reanalysis by W. Baer prompted a published reassessment from the authors.³

Transparency

The article discloses its funding sources, reports per-experiment tables, states its artifact-control tests, and explicitly labels its post hoc analyses and its non-supporting subsets. On the Zen-temple subset in particular, the authors themselves note that it was conducted outside a controlled laboratory in a rural location with erratic line power, which they say may have introduced unforeseen variance, and they set it aside on those grounds; they likewise flag the null meditation-years correlation. The article does not publish a per-session data table, and the venue, Physics Essays, is a smaller journal than the mainstream physics literature.

The adversarial record

• Precursor and related work. The design follows a line of consciousness-collapse tests; the authors extended it in a 2016 distant double-slit study and a 2015 single-photon series, the latter of which did not find a systematic effect, a candid negative from the authors themselves.
• Contested literature. W. Baer published an independent verification and critique of the analysis; Radin, Michel, and Delorme responded in a 2015 reassessment arguing the central interference minimum should be evaluated with the standard error rather than the standard deviation. The exchange is on the record and cited below.
• Reading against the study. The strongest honest cautions are the reliance on post hoc steps for some supporting results, the concentration of the signal in self-selected meditator subgroups, the absence of independent replication, the smaller publication venue, and the author group’s own later single-photon null. The headline pooled result and the clean no-observer controls are the paper’s most robust features; the surrounding correlational and subgroup claims are where a skeptical reader should apply the most weight.