Discovery Lab Answers

Robin K. Taylor, PhD is a parapsychologist trained at the Koestler Parapsychology Unit (KPU), University of Edinburgh, where he completed his PhD under Robert L. Morris. His ESP-Nexus profile is at Robin K. Taylor.

Before drawing conclusions from the quantitative findings below, note the corpus coverage qualification: the studies currently in the ESP-Nexus library represent what the library holds on Taylor’s work, and the library’s share of the published literature on this question has not been measured. Treat what follows as a summary of the available holdings, not a settled account of the field.

Experiments

Imagery training and psychokinesis — two experiments with athletes (Taylor, 1996)

Taylor ran two experiments testing whether imagery training could improve performance on random number generator (RNG)-based psychokinesis (PK) tasks, using athlete populations at the University of Edinburgh and the University of the South Pacific. The first experiment used gymnasts; the second used shooters. Participants were assigned to more-imagery or less-imagery conditions and played an RNG-PK game designed around sport-relevant feedback animations. The study was published in the European Journal of Parapsychology and is also cited in Taylor’s own later cross-cultural research paper.

Remote mental influence of electrodermal activity — pilot replication (Radin, Taylor, and Braud, 1995)

Taylor co-authored a pilot replication of earlier DMILS (direct mental interaction with living systems) experiments, conducted at the University of Edinburgh. The study tested whether a sender’s directed mental intention could influence a remote participant’s electrodermal activity (EDA) across randomly assigned calm and activate epochs. Co-authors were Dean I. Radin and William G. Braud.

Edinburgh automated ganzfeld security audit (Dalton, Morris, Delanoy, Radin, Taylor, and Wiseman, 1996)

Taylor was a co-author on a security review of the Edinburgh automated ganzfeld system, published in the Journal of Parapsychology. The paper examined artifact classes — sensory leakage, subject bias, experimenter bias — and proposed countermeasures and recommendations for future automated ganzfeld settings (the paper is described across the Deborah L. Delanoy automated ganzfeld page).

Cross-cultural parapsychological research in Fiji (Taylor, 2000)

Taylor published a methodological and reflective paper on conducting parapsychology research across ethnic cultures, drawing on fieldwork at the University of the South Pacific in Suva, Fiji. This paper is conceptual and ethnographic rather than a controlled experiment.

Book review (Taylor, 2018)

Taylor published a review of K. Ramakrishna Rao’s Elements of Parapsychology in the Journal of Scientific Exploration, engaging with theoretical and methodological issues in the field, including the experimenter effect and replication problems.

Methodology

The two athlete experiments used an RNG-PK game engine — a well-documented format in micro-PK research — adapted with sport-specific animated feedback so that participants engaged with the task through imagery relevant to their actual athletic training. Key design features:

• Participants were divided into more-imagery and less-imagery groups; the primary analysis was a between-groups comparison of PK scores, with the direction of prediction (imagery group scoring higher) specified beforehand.
• The Movement Imagery Questionnaire (MIQ) was administered before and after training to measure self-reported visual and kinesthetic imagery ability, allowing a test of whether imagery change correlated with PK change.
• Scoring was designed separately from the feedback scoring (FINA regulations were not used for statistical analysis).
• The two experiments differed in a crucial timing variable: in the gymnast study, early PK sessions preceded any strategy instruction; in the shooting study, the timing of imagery training relative to the PK game was different. Taylor notes this as a meaningful design difference between the two studies.
• The gymnast study included a geomagnetic field (GMF) data collection component, treating local GMF intensity as an auxiliary measure.

For the DMILS replication, the protocol assigned 30-second epochs as either “calm” or “activate” in a randomized sequence within a 32-minute session, with EDA of the remote participant recorded continuously. The primary analysis used a percentage score index (PSI statistic) to allow direct comparison with previously published DMILS work. A 5-second window size for data scoring was selected before examining the effect of different window sizes, a pre-specification the authors note explicitly.

Data

The evidence base in the ESP-Nexus library for this question covers three papers and four result rows. Because the rows report different, non-comparable metrics (Cohen’s d, Pearson r, and a calibration null), they cannot be averaged or pooled into a single overall effect size. The table below presents each row on its own terms.

What the pattern shows:

In the gymnast experiment, the imagery group scored higher than the control group on the RNG-PK task, with a Cohen’s d of 0.68 — a moderate effect by conventional benchmarks — but the result just missed the p =.05 threshold (p =.06). The df-weighted average across both experiments yields a Cohen’s d of 0.49, pulled down from the gymnast result by the shooting study, which produced a near-zero effect. The DMILS pilot replication by Radin, Taylor, and Braud produced an effect size of r = 0.27 across 16 sessions, described by the authors as comparing favorably to a prior meta-analytic estimate from the broader DMILS literature.

The Dalton (1996) calibration row reports a null result across 53,000 trials, but this is a pre-series randomness check on the target-generating system itself, not a test of psi — its null outcome is the expected and desired finding, confirming the RNG was operating at chance before experimental sessions began.

Unsettled signals the evidence carries: the reported directions are not uniform across rows — one explicit null appears (the calibration), and the two athlete experiments diverged substantially in effect size. The overall picture is positive in direction but with meaningful heterogeneity across experiments, and the headline gymnast result did not reach conventional significance.

A publication-year trend chart is not rendered here because the rows do not carry at least four results on a single comparable standardized metric with publication years — the metrics mix Cohen’s d and Pearson r, which cannot be plotted on a common axis.

Skeptical critiques

The sources retrieved for this question do not include published criticism directed specifically at Taylor’s experiments. What the record does carry is relevant context:

Taylor (1996) himself notes two interpretive complications worth surfacing:

What the data raise internally. The shooting study produced a near-zero effect, contrasting with the gymnast study’s moderate result. Taylor attributes this in part to attendance problems — real-world scheduling demands for the shooting team reduced session participation — and to the different timing of imagery training relative to PK sessions. A post-hoc paired t-test comparing early versus late PK session scores was not significant (p =.10), though in the predicted direction. Taylor explicitly flags that motivation differences between athlete groups, rather than imagery skill per se, may explain the divergence, partially replicating earlier work by George (1982) and Braud (1983) while leaving open which variable was causally active.

What the experimental data show. The correlation between changes in self-reported visual imagery ability and PK scores in the shooting study was positive and described as a partial replication of Braud (1983), but the primary between-groups test in the gymnast study did not reach p =.05. The DMILS result was a single-lab pilot with 16 sessions — the authors describe it as a replication attempt, not a definitive demonstration.

Analysis. The primary methodological questions the sources raise are internal ones: the confounding of motivation and imagery training in athlete populations, differential attendance between experiments, and the timing of training relative to task. Independent replication of the imagery-PK athlete paradigm outside Taylor’s own work has not been documented in the sources retrieved for this question. The DMILS pilot was explicitly positioned as a replication of prior Braud and Schlitz work rather than a novel claim, and the authors note its pilot status.