Varvoglis and Bancel note that the cumulative micro-PK database spans thousands of experimental sessions, yet individual studies are typically powered to detect only moderate-to-large effects. If the true effect size is in the r = 0.01–0.02 range, consistent with the broader RNG meta-analytic literature, a single study of 100 or even 500 trials will have very low power to detect it. This means that null results from individual experiments cannot be taken as strong evidence against the hypothesis; only adequately powered or meta-analytically aggregated designs can adjudicate the question.^[3] The proposed mechanism, that conscious intention modulates quantum-level randomness, remains contested and is treated by Varvoglis as a working hypothesis rather than an established explanation.^[7]

The Varvoglis and McCarthy (1986) study used an electronic RNG whose output was displayed as a feedback task. Participants were assigned to conditions varying in task-orientation and awareness of the psi goal. The key manipulation addressed demand characteristics and response-bias artifacts: by varying whether participants knew they were in a psi study versus a cognitive task, the design partially tested whether anomalous deviations were driven by expectation effects rather than genuine PK. The study found that conditions involving conscious, purposive engagement with the RNG task produced larger deviations from chance than passive or distracted conditions. Effect sizes and exact sample sizes are not fully recoverable from the pool entry, but the study is cited as foundational for the “conscious-purposive” framing of micro-PK that Varvoglis developed in subsequent theoretical work.^[1] A parallel theoretical paper evaluated the conformance-behavior model and observational theories of PK, situating the empirical findings within competing theoretical accounts of goal-directed versus observer-dependent PK.^[8]

The Schechter, Barker, and Varvoglis (1983) preliminary study used a PK game format in which participants were either focused on the RNG task or given a secondary cognitive task designed to divert attention. The study was explicitly preliminary in scope, with sample size and stopping rule not disclosed in the pool entry. The distraction manipulation was intended to test whether attentional resources mediate PK output, a non-psi alternative explanation being that any observed deviations reflect response-bias or demand characteristics that are reduced when attention is divided. Results were preliminary and the study is cited primarily as a methodological precursor to the 1986 Varvoglis-McCarthy design.^[9]

The Honorton, Berger, Varvoglis et al. (1990) autoganzfeld report covered 11 experimental series with 243 volunteer participants (102 men, 141 women; age range 17–74, mean 37.2 years; mean education 15.6 years). Participants were recruited as volunteers and were predominantly psi-believers (mean 6.21 on a 7-point belief scale; over 90% reported personal psi experiences). The automated system addressed sensory leakage by physically isolating sender and receiver, with computer-controlled target presentation eliminating direct experimenter cueing during target selection. The file-drawer artifact was addressed by including every session completed since the system’s 1983 inauguration, including pilot and ongoing studies. The primary hypothesis, that participants would identify the actual target above chance, was tested via exact binomial probability (p = .25, α = .05, one-tailed). Results: 123 direct hits from approximately 358 sessions = 34% hit rate vs 25% chance baseline (highly significant, p < .001; exact figures vary slightly across Honorton 1990 vs Bem & Honorton 1994 reporting). Dynamic targets (video clips) were predicted to outperform static targets (photographs), and friend/acquaintance senders were predicted to outperform stranger senders, both secondary hypotheses were prespecified. Remaining analyses were exploratory. The proposed mechanism, that the ganzfeld state of sensory attenuation facilitates psi signal reception, remains the researcher’s preferred interpretation; alternative interpretations include experimenter-expectancy effects during target loading (partially addressed but not fully eliminated) and sampling bias toward psi-believing participants.^[2][10]

Varvoglis and Bancel (2015/2016) reviewed the micro-PK literature to assess whether effect distributions were consistent with a “gifted individual” model (in which a small subset of participants drives aggregate results) or a “universal small effect” model (in which all participants contribute equally small deviations). The review is theoretical and meta-analytic in character rather than reporting new primary data. The key methodological question it raises is whether participant pre-selection on putative PK ability improves effect sizes, a moderator variable that has been inconsistently tested across the literature. The proposed mechanism, that micro-PK reflects a universal property of conscious systems interacting with quantum-level randomness, is explicitly framed as the researcher’s preferred interpretation; competing non-psi explanations include publication bias (studies showing null results in unselected populations are less likely to be submitted) and multiple-comparison inflation in studies that test many participant subgroups. The review concludes that the distribution of effects across studies is more consistent with the exceptional-performer model than with the universal small-effect model. Re-analysis of the PEAR database showed that two operators contributed nearly a quarter of the total trials and that excluding them substantially reduced the cumulative effect, supporting the view that micro-PK is concentrated in a minority of gifted participants rather than distributed broadly.^[3][7]

In a 2013 chapter, Varvoglis explored how micro-PK research, along with other psi phenomena, challenges standard neuroscientific and physicalist accounts of the self as bounded by the skin. He considered four candidate frameworks: soul-based, neuroscientific, particle-physics, and “mind-at-large” accounts. The chapter is theoretical rather than empirical and does not report new experimental data. The argument is that if micro-PK effects are genuine, they imply that intentional states can have causal effects on physical systems outside the body, a claim that requires either revision of standard physical theory or a new account of the causal efficacy of mental states. Varvoglis frames this as an open epistemological question rather than a settled conclusion.^[11]

The Sharefield protocol, introduced by Varvoglis, Bancel, Si Ahmed, Bailly, and Béguian (2013), is a forced-choice GESP (General Extrasensory Perception) methodology in which multiple participants simultaneously attempt to identify a shared target from a constrained set of alternatives. The design addresses two specific artifacts: (1) response bias, by requiring participants to choose from a shared field of alternatives rather than generating free responses, individual idiosyncratic biases are partially controlled; and (2) multiple-comparison inflation, by specifying a single primary hypothesis (above-chance identification of the shared target) rather than testing multiple target-response combinations. The study is exploratory in character and the pool entry does not provide full sample sizes or effect sizes. The proposed mechanism, that a shared target field facilitates GESP by providing a common referent for multiple percipients, is the researcher’s preferred interpretation; a competing non-psi explanation is that shared target fields introduce demand characteristics when participants are aware of each other’s choices.^[4]

The Bancel, Varvoglis, Boban, and Bensahra (2025) study used the internet-based Psi@Home platform to run a preregistered forced-choice precognition experiment with two cohorts: experienced meditators and a general-interest cohort. Each session comprised 20 forced-choice trials. The preregistered study collected 80 sessions per cohort (160 total) from 47 participants. Three hypotheses were preregistered: (H1, confirmatory) increased variance of session hit rates relative to mean chance expectation (MCE); (H2, exploratory) increased total hit rate relative to MCE; (H3, exploratory) higher session-hit-rate variance for the meditator cohort than the general cohort. Results: only H3 was confirmed (p = .03, exploratory). H1 and H2 were not confirmed in the formal study. However, 90 tryout sessions, collected from the same participants using the same protocol, with collection specified in the preregistration, showed a markedly strong variance increase (p ≈ .00003). The researchers’ preferred interpretation is that uncontrolled psychological variables (participant attitudes differing between tryout and formal phases) account for the discrepancy, consistent with a mixture of psi-hitting and psi-missing across sessions. A competing non-psi explanation is that the tryout result reflects optional stopping or participant self-selection during the informal phase, even though the preregistration specified its collection. The platform is available to outside researchers. Blinding was achieved by computer-controlled target selection with physical separation of participants and experimenters; randomization used computer-controlled randomization (target selected post-response); stopping rule was fixed-N for the formal study.^[5][5]