Effects of Background Context for Objects in Photographic Targets on Remote Viewing Performance

Katz, D. L., Lane, J. D., & Freed-Bulgatz, M. (2021). Effects of Background Context for Objects in Photographic Targets on Remote Viewing Performance. Journal of Scientific Exploration, 35(4), 752–787. https://doi.org/10.31275/20212273

AI Assessment

An exploratory experiment whose two scoring methods returned significant background effects in opposite directions. Twelve experienced remote viewers completed 360 open-response trials describing photographed objects set on white, normal, or unusual backgrounds. Sum-of-ranks matching favored white backgrounds, viewers’ own per-item hit rates favored normal and unusual backgrounds, and independent judges’ hit rates showed no significant condition effect. The paper reports the conflict openly and consistently; it does not report a preregistration registry entry, and it describes its own randomization as only partial.

Provenance

DOI. 10.31275/20212273 · Open access, Creative Commons License CC-BY-NC.

Authors. Debra Lynne Katz (International Remote Viewing Association), James D. Lane (Rhine Research Center), and Michelle Freed-Bulgatz. The article’s byline prints the third author as “Michelle Freed-Bulgatz”; the running heads and the journal’s own citation metadata list her as “Michelle Bulgatz.”

Study type. Exploratory open-response remote viewing experiment, described by the authors as triple-blind, comparing performance on photographic object targets across three background conditions.

Funding. The project was a recipient of the IRVA/IRIS Warcollier Proposal Award.

Data availability. The paper contains no data-availability statement. The authors state the target pool “can be made available to other serious researchers upon request.”

Source basis. Figures confirmed against the primary article (publisher PDF, Journal of Scientific Exploration, 35(4), 752–787).

What the paper reports

Twelve experienced remote viewers each completed 30 open-response trials, one per assigned target number, describing the photographic image they would receive by email a few days later. The 360 transcripts contained 8,460 written descriptors and 1,472 sketches.¹ Each of ten object themes appeared once in each of three background conditions: white (devoid of information), normal (a setting where the object is typically found), and unusual (a setting where it is not). In the Phase II sum-of-ranks matching analysis, background condition made a significant difference, F(2,22) = 5.58, p = .01: the mean sum of ranks for white backgrounds (M = 21.2, SE = 0.97) was significantly lower, meaning stronger, than for normal backgrounds (M = 25.5, SE = 1.0; difference 4.3, SE = 1.4, p = .005) and unusual backgrounds (M = 24.7, SE = 1.0; difference 3.5, SE = 1.4, p = .02). In the Phase I per-item analysis, viewers’ self-rated combined hit rates went the other way, F(2,22) = 4.53, p = .02, with white backgrounds (0.53) scoring below normal (0.61) and unusual (0.62) backgrounds; the independent judges’ hit rates showed no significant condition effect, F(2,22) = 1.76, p = .19. Hits pertained to the main object far more often than to the background (average proportions 0.64 vs 0.09), rejecting the authors’ third hypothesis.

The two scoring methods produced significant but opposite background effects, and the analysis the result depends on is the one a reader chooses: matching favored white backgrounds, viewers’ per-item hit rates favored normal and unusual backgrounds, and independent judges’ hit rates showed no significant difference at all.

How it was run

• Twelve remote viewers (five men, seven women) each completed all 30 targets in the same order at the same biweekly schedule, for 360 sessions; six viewers had more than 10 years of remote viewing experience, and nine were trained in Controlled Remote Viewing methodology or a derivative.
• The target pool, built by the first author, held 30 photographs, each containing one prominent, immediately identifiable “main object”; photographs were mostly collected from royalty-free internet sites, arranged as ten object themes appearing once per background condition.
• Viewers worked alone, unmonitored, from their homes, and emailed their transcript and data score sheet to Researcher #2 before the feedback photo for that trial was released.
• An outside volunteer (“the Randomizer”) generated the target order with an online target reference number generator and was instructed to keep the same object theme from repeating in succession; the authors state that “in this respect the randomization was only partial.” Only the Randomizer and a backup key holder, the Director of the Rhine Research Center, ever held the list linking target numbers to photos before feedback.
• Phase I judging scored every descriptor and sketch as correct or incorrect: each viewer self-judged after feedback, and six independent judges, blind to the viewers’ self-scores, rated the same items (each judge covered two viewers, 60 transcripts).
• Phase II judging, which viewers were not told about, had twelve judges rank each transcript against a randomly ordered set of four same-background photos (the target plus three unrepeated decoys), best match to worst, using the sum of ranks method of Solfvin et al. (1978).²

Results, as reported

The paper reports means, standard errors, F ratios, t tests, and p values; it reports no standardized effect sizes and no confidence intervals for the background comparisons, so none are stated here.

Eleven-dimension audit

Pre-registration

The paper does not report a preregistration registry entry. The three hypotheses and both judging phases are stated in the paper itself, and the authors mention that advisors reviewed the project proposal for the IRVA/IRIS Warcollier award, asking for a standard matching task to be added. The study is internally pre-specified and labeled exploratory by its own authors, not independently registered.

Randomization

An outside Randomizer generated the target order using an online target reference number generator and, by instruction, rearranged it so the same object theme never appeared twice in succession; the authors state plainly that “in this respect the randomization was only partial.” The Phase II judging sets were shuffled into different positions for each judge by an online survey program. The paper reports no raw randomization audit trail for either step.

Sensory leakage

No one in contact with the viewers knew the targets: Researcher #2, the sole point of contact, was blind to the target pool and saw one feedback photo per trial only after every transcript was in, and the list linking target numbers to photos was held only by the Randomizer and a backup key holder with no other connection to the project. Each transcript was submitted before the feedback photo was released. The viewers worked at home, unmonitored, so compliance with the trial schedule and conditions rests on self-report, but the paper describes no ordinary sensory path by which a viewer could reach a target photo before submitting.

Blinding

The abstract describes the trials as triple-blind. As detailed in the paper, all viewers and both Researchers #1 and #2 were blind to target order; Researcher #1, who built the pool, did not communicate with viewers during data collection; Phase I independent judges worked from score sheets with the viewers’ self-scores hidden and locked behind a password; and Phase II judges were a separate team who worked only after the experimental phase ended. Viewer self-judging is by design done with the feedback photo in hand and is therefore not blind; the paper treats it as a separate rating category and reports it separately throughout.

Optional stopping

The design fixed 30 targets and twelve viewers in advance, every viewer completed all 30 trials, and judging for one trial was completed before the next trial began. The paper reports no stopping rule and the fixed structure leaves no room for outcome-dependent stopping. The object-category analysis covers 207 trials described as “all available trials,” and the paper does not state why the remaining trials were unavailable.

Outcome measure

The study ran two outcome procedures: the traditional sum-of-ranks matching task (Phase II) and an exploratory per-item rating of every descriptor and sketch (Phase I, the Poquiz scoring method), with hit rates for items and sketches averaged into a combined score. The two procedures disagreed in direction on the background effect, a conflict the authors report openly and describe themselves as “somewhat perplexed” by. The paper also states that the Phase I approach “provides no statistical means to determine whether scores differed from chance.” The two measures are related at the transcript level: hit rate fell monotonically as match rank worsened, F(3,33) = 6.20, p = .002 for viewer self-ratings and F(3,33) = 3.70, p = .02 for independent judges.

Effect size

The paper reports means with standard errors and p values for the condition comparisons but no standardized effect sizes for them. The only variance-explained figure reported is R² = 0.10, from the regression of viewer mean hit rates on judge mean hit rates (F(1,10) = 1.18, p = .30), which the authors use to show the two rater types were unrelated.

Multiple comparisons

The analysis spans two judging methods, two Phase I rater types, three pairwise condition contrasts, and 36 viewer-by-condition Monte Carlo tests plus twelve all-trials tests, of which 5 and 3 respectively reached p ≤ .05 one-tailed. The paper reports no correction for multiple comparisons across these tests.

Internal replication

The study is a single experiment with no internal replication series. Its two scoring methods function as analytic cross-checks on the same data and reached significant effects in opposite directions, while the independent judges’ hit-rate analysis, the one fully blind per-item test, found no significant background effect (p = .19). Per-viewer results varied widely: three of twelve viewers scored significantly above chance across their 30 trials, and Table 6 proportions for the main object range from 0.41 to 0.98 across viewers.

External replication

The study is a conceptual replication transposing the object-recognition findings of Barenholtz (2013), in which objects in familiar contexts were identified fastest and objects with no background slowest, into a remote viewing protocol.³ The per-item hit-rate results matched her direction and the matching-task results reversed it, as the authors note. The paper reports no direct replication of this design and proposes future projects, including testing whether telling viewers about matching-task judging changes results.

Transparency

The article is open access under CC-BY-NC, names each researcher’s role in a dedicated table, quotes the viewer instructions verbatim, reproduces sample transcript sketches in an appendix, and offers the target pool to other researchers on request. The paper reports an informal inter-rater check rather than a reliability statistic, and the authors state on that basis that comparing hit rates among viewers would not be appropriate. The viewer experience counts the paper lists (three viewers above 1,000 sessions, five between 500 and 1,000, three between 200 and 500, and two between 20 and 99) sum to thirteen entries for the stated twelve viewers.

The adversarial record

• The design grew out of the target-characteristics literature the paper reviews (Delanoy 1989; Watt 1989; Warcollier 1948) and out of Barenholtz (2013), a mainstream visual-cognition study with no psi component; the authors frame their hypothesis as extrasensory perception mirroring physical perception.
• The two-phase judging structure was built to mitigate displacement, the long-debated tendency of percipients to describe the wrong target, which the paper traces from Richet through Carrington, who coined the term, to Tart’s argument against matching tasks. The authors sampled the Phase II misses, found no close decoy matches, and concede they “can’t say for sure” the approach reduced displacement; they also present a trial-6 accordion sketch that strikingly matches the trial-16 feedback photo as possible temporal displacement.
• Reading against the study: there is no external preregistration; the randomization is, by the authors’ own description, only partial; viewers worked unmonitored at home; viewers rating their own transcripts scored significantly higher than independent judges rating the same items (M = 0.58 vs 0.43, p = .01); the headline background effect reverses or disappears depending on the scoring method; and the object-category analysis covers 207 of 360 trials without stating why the rest were unavailable.