Discovery Lab Answers

Experiments using psi games

The term “psi games” in parapsychology research refers to computer-based software applications designed to test for ESP or psychokinesis effects under controlled laboratory conditions, typically by having participants interact with a random number generator (RNG) while receiving audiovisual feedback. The research retrieved here spans from a 1986 preliminary review through a 2024 VR-based study.

The PsiLab II platform and early psi game research

The foundational work in this area came out of the Psychophysical Research Laboratories (PRL) in Princeton. Berger, Schechter, and Honorton conducted a preliminary review of performance across three computer psi games ^[5] — including titles called Psi Invaders and PsiBall — examining RNG-based performance patterns across participants.

A key finding from that review concerns the distinction between two types of participants the researchers called “MEPs” and “nMEPs.” The sources retrieved indicate that silent-RNG variability was significantly smaller than feedback-RNG variability in Psi Invaders data for the MEP group, and nonsignificantly smaller for the other group ^[5]. The extraversion/introversion scale (E/I) showed a predicted negative relationship with feedback-RNG scores across all three experiments, though none of the three correlations reached statistical significance ^[5]. These differential performance patterns across task types informed the subsequent design of automated protocols.

Psi effects without real-time feedback

Rick E. Berger’s 1988 paper ^[3] is the most detailed primary source retrieved here on psi game methodology. Using two PsiLab II computer psi games across four experiments at the Science Unlimited Research Foundation (SURF), Berger structured each session so that a subject-initiated button press generated two yoked RNG data samples simultaneously — one tied to real-time audiovisual feedback, and one “silent” sample the subject never observed.

The logic was direct: if psi effects require real-time feedback to occur, they should appear only in the feedback condition. Instead, Berger found that directional effects appeared in the silent data as well ^[3]. The ESP-Nexus page on this work summarizes the outcome: across four experiments using two different PsiLab II psi games, significant departures from the expected run-score distribution appeared consistently in the silent condition, while feedback and simulation data remained at chance — with an exact binomial probability of four out of four silent outcomes exceeding the significance threshold.

The simulation (computer-generated baseline) data showed an anomalous peak at the mean run score, structurally similar to baseline data collected elsewhere under subject observation — though Berger’s paper treats this as most parsimoniously a chance departure rather than a psi-related signal ^[3].

Berger explicitly frames “control” conditions in psi experiments as potential “baseline” conditions rather than inert controls, given these results ^[3] — a methodological implication that reaches well beyond psi game research specifically.

VR video games as sender stimuli (2024)

Lieb (2024) introduced a novel variation: rather than having the receiver interact with a psi game, the sender played a VR video game during the ganzfeld stimulus phase ^[1]. The rationale was that VR games are engaging enough to induce a flow state in the sender, potentially enhancing whatever mechanism might underlie psi communication. This appears to be, based on the authors’ own statement, the first study to use VR video games in this role in a ganzfeld experiment ^[1]. The paper describes itself as a proof-of-principle study and acknowledges its limitations; the authors discuss design implications for future work rather than asserting a settled positive result ^[1].

Casino gambling as a naturalistic psi game analogue

Radin (1998) ^[2] explored a different application: treating casino slot machine outcomes as a large-scale naturalistic test environment. The paper examined correlations between daily payout percentages and seven environmental variables (including lunar phase, barometric pressure, geomagnetic activity, and others), using a 7-day moving average to smooth random daily variation ^[2]. This framing treats gambling outcomes as a proxy for psi-related human-machine interaction at scale — the “dream” scenario Radin describes of thousands of motivated participants testing in real-world conditions around the clock. Radin notes explicitly that any detected correlation would still mean losing “a little slower than usual,” since empirical payout percentages remain below 100% ^[2].

Skeptical critiques

What critics argue. A central methodological concern with RNG-based psi game research is the possibility of undisclosed multiple testing and post-hoc selection of conditions. When both feedback and silent samples are generated simultaneously, the choice of which condition to treat as primary — and whether that choice was pre-specified — affects the interpretation of any significant result. Berger’s 1988 paper ^[3] acknowledges that the silent-condition effect sizes in this literature carry a Type II vulnerability (the ESP-Nexus page flags this explicitly): silent-condition effects may be systematically underpowered relative to feedback conditions, meaning failures to replicate them could reflect insufficient power rather than absence of effect.

What the experimental data show. The 1986 review ^[5] found that correlations between extraversion/introversion and RNG performance were in the predicted direction across all three games but did not reach significance in any of them — a pattern consistent with a small effect requiring larger samples, but also consistent with no effect and chance variation in the direction of correlation. The 1988 study ^[3] found that all four silent outcomes exceeded the significance threshold, which Berger treats as the primary support for psi-without-feedback; however, the subject pool for those experiments was self-selected and included parapsychological researchers as participants.

Analysis. The psi game literature raises a question that the retrieved sources leave open: whether the silent-condition effects in Berger’s four experiments have been independently replicated outside the original SURF laboratory setting by experimenters not themselves serving as subjects. The 1986 review ^[5] was preliminary and explicitly framed as such. The 2024 VR study ^[1] shifts the design substantially enough that it does not directly address the feedback/silent distinction. Independent replication of the specific silent-condition finding, using pre-registered protocols and experimenters not participating as subjects, has not been documented in the sources retrieved here.

For a fuller account of the theoretical framework and experimental design behind the psi-without-feedback paradigm, the ESP-Nexus spoke page Psi effects without real-time feedback and the Rick E. Berger hub carry the most relevant synthesis.