Exploring the Links: Creativity and Psi in the Ganzfeld

Dalton, K. (1997). Exploring the Links: Creativity and Psi in the Ganzfeld. Proceedings of the 40th Annual Convention of the Parapsychological Association, 119–134.

AI Assessment

A strongly positive single ganzfeld telepathy result; the confirmatory hit rate is robust, while the creativity link the study was named for is mixed. Across 128 ganzfeld sessions with creative participants, receivers identified the target video clip 47% of the time against a 25% chance baseline, a large and highly significant effect and one of the higher single-study ganzfeld hit rates on record. The creativity and personality correlations the study set out to test, however, were mostly non-significant overall, with only scattered subgroup effects that the author reports cautiously as an initial report.

Provenance

Source. Conference proceedings (no DOI): Proceedings of the 40th Annual Convention of the Parapsychological Association, pp. 119–134. An initial report of the author’s PhD research at the University of Edinburgh.

Study type. Sender-and-receiver telepathy in an automated ganzfeld, with four creative-population groups. Four-alternative forced choice, 25% chance baseline.

Funding. Society for Psychical Research; Institut fur Grenzgebiete der Psychologie und Psychohygiene; Parapsychology Foundation.

Data availability. None stated; as a 1997 conference paper there is no open-data repository. The automated ganzfeld system recorded all session data to disk.

Source basis. Figures confirmed against the scanned original (Koestler Unit copy, pp. 119–134) and cross-checked between the abstract and Table 1.

What the paper reports

Across 128 ganzfeld sessions, 32 in each of four creative groups (Artists, Musicians, Creative Writers, Actors), receivers achieved 60 direct hits, a 47% hit rate against the 25% mean chance expectation, significant at p = 7 × 10⁻⁸ with a large effect size (ES(h) = .46). All four groups scored above chance: Musicians 56% (p = .0001), Artists 50% (p = .002), and Writers and Actors each 41% (p = .037). All p values are one-tailed.¹

The headline confirmatory result is the overall direct-hit rate. The creativity and extroversion correlations the study was designed to test were non-significant overall (verbal-creativity rho = −.068; extroversion rho = .021), with only scattered, sometimes oppositely-signed effects within subgroups.

How it was run

• Telepathy design: each receiver brought their own sender. 128 participants (77 female, 51 male; age 18 to 70, mean 28), all ganzfeld novices, one session each, split across the four creative groups.
• The automated ganzfeld system of the Koestler Chair at Edinburgh recorded data automatically and was built to resist sensory cues and both subject and experimenter fraud; program security was independently verified by Robert Morris and a second person before the study.
• Targets were roughly 60-second dynamic video clips, drawn mainly from the PRL pool. In standard ganzfeld (halved ping-pong balls over the eyes, a red light, white noise, a 15-minute relaxation), the receiver gave impressions over a 30-minute period while the sender, in a locked room about 25 metres away, viewed the target clip.
• Audio was one-way: the sender could hear the receiver, but no channel allowed communication from the sender to the receiver or experimenter.
• Judging: the experimenter, blind to the target, reviewed the impressions; the receiver then ranked four candidate clips (1 to 4) and rated each (1 to 99). A direct hit was a first-place rank, giving a 25% chance baseline.
• The target was selected by the program at run time; a randomness-certification test on the generator was run before the study and re-checked afterward, with no consistent departures found.

Results, as reported

All p values are one-tailed, as the paper states. The direct-hit rate against chance (hypothesis 1) is the pre-specified primary outcome; the creativity and extroversion correlations (hypotheses 2 and 3) are pre-specified secondary tests.

Eleven-dimension audit

Pre-registration

Three hypotheses (the direct-hit rate would exceed chance; success would correlate with creativity; success would correlate with extroversion) were stated in advance, and the four-group, 128-session design was fixed before data collection. As a 1997 conference paper there is no formal registration repository; treat the hypotheses as a priori rather than independently registered.

Randomization

The target was selected by the system at run time. A randomness-certification test was run on the target generator before the study and the program was re-checked at the end, with no consistent departures from expectation. Randomization is adequate and explicitly verified.

Sensory leakage

Strong by design. The automated system was built to resist sensory cues and fraud; sender and receiver were about 25 metres apart with the sender in a locked room, and the audio was one-way, so the sender could hear the receiver but not the reverse. Program security was independently verified before the study.

Blinding

The experimenter who reviewed the impressions and ran the judging was blind to the target, and the receiver ranked four clips with target selection and data recording automated. The sender was a person known to the receiver, which is a telepathy-design choice rather than a blinding flaw, but it means the sender-receiver pairs were not independent strangers.

Optional stopping

The sample was fixed in advance at 128 sessions (32 in each of four groups), one per participant. There is no sign of data-dependent stopping.

Outcome measure

The pre-specified primary measure was the direct-hit rate against the 25% baseline (a first-place rank among four clips), chosen for comparability with other ganzfeld work. Stanford and Sargent z-scores on the ratings were used for the correlational hypotheses.

Effect size

Large for this field: ES(h) = .46 overall (47% against 25%), rising to .65 for the Musicians. These are among the higher single-study ganzfeld effect sizes reported.

Multiple comparisons

Beyond the single primary hit-rate test, the paper runs many secondary tests: a hit rate and two correlations (creativity and extroversion) within each of four subgroups, reported without correction. Several subgroup correlations reach nominal significance in inconsistent directions (Musicians positive on both; Artists significant toward introversion; Actors significant on verbal creativity; Writers negative). These should be read as exploratory.

Internal replication

All four independent subgroups scored above chance (41 to 56 percent), each individually significant, which is internally consistent. The study is a single series rather than a repeated experiment.

External replication

The result replicates earlier positive ganzfeld work, including the PRL autoganzfeld series and the Schlitz and Honorton (1992) study of artistically gifted students, and it sits within the contested wider ganzfeld literature.

Transparency

The methods, the independent security checks, and the randomness certification are described in detail. As a 1997 conference paper there is no open-data repository; the system recorded all data automatically to disk. The author flags this as an initial report, with further creativity analyses and a sender-receiver-relationship analysis to follow.

The adversarial record

• The headline 47% rests on the overall direct-hit rate; the creativity-and-psi relationship the study is named for was non-significant overall (verbal-creativity rho = −.068), with only scattered subgroup correlations, several in unexpected directions, which the author treats cautiously and labels an initial report.
• The wider ganzfeld base is contested: an independent meta-analysis published soon after (Milton and Wiseman, 1999) found no significant overall ganzfeld effect across studies, and the 1980s Hyman and Honorton exchange set the methodological standards this autoganzfeld design was built to meet.²
• Reading against the study: it is a conference-proceedings report rather than a peer-reviewed journal article; the p values are one-tailed; the senders were associates chosen by the receivers, a telepathy-design feature but not independent pairings; and the many uncorrected subgroup and correlational tests inflate the chance that some reach significance.