The forced-choice literature is particularly susceptible to Type-II error (false negatives) because the typical effect sizes are small (hit rates a few percentage points above mean chance expectation) and because motivational decline across long trial sequences can suppress effects that might appear in shorter, more engaging protocols. Varvoglis’s Selfield and Sharefield designs attempt to address this by building in participant agency over trial timing and target salience, reducing the monotony that may dilute effects in classical designs.² The 2025 Psi@Home study illustrates the problem concretely: formal preregistered sessions showed no significant hit-rate increase, while tryout sessions from the same participants, collected under conditions of higher novelty and engagement, showed a variance anomaly at p = .00003.³

The Sharefield protocol was designed to address two specific artifacts common in forced-choice GESP research: (1) response bias, in which participants develop idiosyncratic preferences for certain response options across trials, inflating or deflating apparent hit rates independently of any psi signal; and (2) the ecological artificiality of the standard one-sender/one-receiver dyad, which may not reflect the social contexts in which spontaneous GESP cases are reported. By having multiple participants respond to the same target simultaneously, the Sharefield design allows response-bias patterns to be identified and statistically separated from group-level convergence on the correct target. The proposed mechanism, that a shared target field might amplify or facilitate GESP, remains the researcher’s preferred interpretation and is contested; alternative interpretations include simple statistical aggregation effects and demand characteristics arising from the collaborative framing.¹

In the Selfield design, participants choose both the moment of their response and the domain of target material (e.g., nature imagery, human faces, abstract art) before the target is randomly selected from that domain. This self-selection procedure is intended to increase the emotional resonance between participant and target, a variable identified in the broader forced-choice and free-response literature as a moderator of psi performance. The specific non-psi alternative addressed by this design is motivational decline: by allowing participants to respond only when they feel ready and to choose target categories they find engaging, the protocol attempts to eliminate the boredom-induced response-set artifact that may account for the near-chance performance typical of long forced-choice series. Whether the self-selection procedure introduces its own confounds, such as demand characteristics or differential attention to certain target categories, is noted as unresolved in the 2019 paper.²

Varvoglis has situated the Selfield and Sharefield protocols within a broader theoretical framework that treats psi as a phenomenon embedded in consciousness and intentionality rather than as a simple signal-detection problem. In this framing, the participant’s psychological state, their degree of focused, purposive attention and their emotional engagement with the task, is not merely a nuisance variable to be controlled but a constitutive element of the effect itself.⁴ This interpretation remains the researcher’s preferred framing; alternative interpretations include the possibility that apparent state-dependent moderation reflects nothing more than regression to the mean in small samples, or that participant self-selection introduces sampling biases that inflate apparent hit rates.

The 2025 Bancel, Varvoglis, Boban, and Bensahra study (N = 47 participants, 160 formal sessions at 20 trials each, plus 90 tryout sessions) preregistered three hypotheses: (H1) increased variance of session hit rates relative to MCE for each cohort, designated confirmatory; (H2) increased total hit rate relative to MCE, designated exploratory; (H3) higher session-variance for the meditator cohort than the general-interest cohort, designated exploratory. H1 and H2 were not confirmed. H3 was confirmed (p = .03). The tryout data, 90 sessions collected prior to formal data collection, with the same participants and protocol, and whose analysis was specified in the preregistration, showed a highly significant variance anomaly (p = .00003). The authors note that differences in participant attitudes during tryout versus formal phases (novelty, lower evaluative pressure, higher intrinsic motivation) may account for the discrepancy, but this interpretation is post-hoc and unconfirmed. The specific non-psi alternative addressed for the tryout anomaly is optional stopping or selective reporting: the tryout analysis was prespecified, mitigating but not eliminating this concern, since the boundary between tryout and formal phases was defined by the researchers.³

The Psi@Home platform was developed at IMI to allow participants to contribute experimental sessions from home using custom software, addressing the practical barrier of laboratory-based recruitment that limits sample sizes in parapsychology research. The platform is described as available to outside researchers, positioning it as a shared infrastructure resource for the field rather than a proprietary IMI tool. Each session comprised 20 forced-choice trials with computer-controlled target selection occurring after the participant’s response, eliminating the possibility of sensory leakage from target-to-response ordering. The platform’s internet-based architecture introduces a new artifact not present in laboratory designs: uncontrolled environmental conditions during participant sessions, which the authors acknowledge as a limitation, partially addressed by the within-participant comparison between tryout and formal phases, but not eliminated.³