Discovery Lab Answers

Ganzfeld results by condition

Coverage first: the figures below come from the ESP-Nexus library’s holding of 57 of the 78 studies in the standard ganzfeld meta-analytic database, plus the abstract of one more. The 16 studies still missing skew toward null results, so any breakdown drawn from these holdings — including this one — runs somewhat more positive than the full published database. Treat this as a summary of what the library holds, not a settled account of the whole ganzfeld literature.

The audited moderator breakdown for this question comes almost entirely from three 2024 sources — a Tressoldi moderator meta-analysis, a Harris reanalysis, and a Lieb single experiment. These report different, non-comparable metrics (standardized effect sizes, Cohen’s h, and raw hit rates), so they cannot be pooled into one overall number. Below they are grouped by the condition each isolates.

Where a positive effect appears in these holdings

Participant selection (Tressoldi 2024). The clearest split is who the participants are. Selected participants show a positive effect with a confidence interval clear of zero; non-selected participants show a much smaller effect whose interval crosses zero.

Task type (Tressoldi 2024). Of the three task types, only telepathy has a confidence interval that stays above zero. Precognition and clairvoyance subgroups both have intervals that include zero.

Combined selected + telepathy (Tressoldi 2024, exploratory). The largest subgroup effect in the set is the exploratory combination of selected participants doing a telepathy task.

Harris (2024) reanalysis. Across several sensitivity cuts — omitting studies with statistical errors, excluding Sargent’s studies by study or by investigator — the Cohen’s h estimates stay positive and fairly stable.

Where it does not appear

Peer-review level (Tressoldi 2024). Proceedings-level and full-journal subgroups produce nearly identical effect sizes — the effect does not vanish or inflate with review stringency in this breakdown.

Publication year / decline (Tressoldi 2024). The meta-regression slope over year is essentially flat, with an interval spanning zero (dir logged as null in one row, positive in another for the same slope).

Lieb (2024) single experiment. Neither the “green” nor the “red” ganzfeld condition reached significance; both are logged as null. At the game/target level within that study, some targets ran above chance (Eagle Flight) while others ran below chance (psi-missing) — Superhot and Fushimi Inari both fell well under the 25% expectation.

Unsettled signals

The directions genuinely disagree across these holdings. Of the 60 result rows, 17 report a null or below-chance (psi-missing) effect — not just weak positives but results that ran under chance. Tressoldi (2024) reports positive subgroup effects for selection and telepathy; Lieb (2024), a single controlled experiment, reports null condition-level hit rates and below-chance hits on individual targets. A separate methodological line runs through source (Rabeyron 2020), which uses the ganzfeld as its worked example of the psi replication problem and notes Bierman’s (2016) reanalysis attributing part of the database effect to questionable research practices while still finding it significant, and (the Koestler Unit review), which cites the Milton & Wiseman (1999) meta-analysis — a well-known null in the ganzfeld literature. These are context in the retrieved sources rather than rows in the audited evidence table.

Bottom line for these holdings: within the library’s (positively-skewed) subset, a positive effect concentrates in selected participants and telepathy tasks and is robust across peer-review level and across Harris’s exclusion sensitivity checks; it does not clearly appear for non-selected participants, precognition, or clairvoyance subgroups, and a single controlled experiment (Lieb) produced null and below-chance condition results. No decline over publication year is evident in the Tressoldi slope.

For the site’s synthesis on the protocol itself, see the Phenomena section.