The ganzfeld experiments

What it is

The most carefully designed experimental paradigm in psi research, and the one on which the case for telepathy chiefly rests.

The design:

A receiver sits in a state of mild sensory homogeneity — halved ping-pong balls taped over the eyes under red light, white noise through headphones — producing the undifferentiated perceptual field the Germans called the Ganzfeld, whole field. This reliably generates drifting imagery, as described under scrying, and the reasoning is that a weak signal might be easier to notice against a quiet background.

A sender in another room views a randomly selected target — usually a short video clip.

The receiver speaks their impressions aloud for around half an hour. Afterwards they are shown four clips, one of which is the target, and asked to rank them.

Chance is 25 per cent.

Where it came from

Charles Honorton began ganzfeld psi experiments in 1974, and by the early 1980s a body of results had accumulated reporting hit rates above chance.

Then something unusual happened, and it is the reason this article exists.

The sceptic Ray Hyman and the proponent Charles Honorton conducted a public methodological exchange in the mid-1980s. Hyman analysed the existing database and identified specific flaws: inadequate randomisation, possible sensory leakage — including the risk that a target clip handled by the sender might be identifiable by handling cues — multiple statistical analyses without correction, and selective reporting.

Rather than trading accusations indefinitely, in 1986 they published a Joint Communiqué. They agreed that the existing database was not sufficient to establish psi, and — crucially — they set out in advance the methodological standards that future experiments would have to meet: automated randomisation, duplicate targets so handling cues could not operate, thorough documentation, and preregistered analysis.

That is adversarial collaboration, decades before the term was current, and it is a better model for settling a dispute than anything else described on this site. The house systems argument, the yarrow-versus-coins question, the Advaita-Dvaita disagreement — none of them has ever produced an agreement about what would count.

How it is used

Honorton then ran the autoganzfeld studies (1983–89), automated to meet the agreed standards. They reported a hit rate of around 34 per cent against 25 per cent expected.

What we can and cannot say

We can say the subsequent record is genuinely mixed, and this is where honesty is required in both directions.

Milton and Wiseman (1999) meta-analysed thirty studies conducted after the Joint Communiqué and found no significant effect — the post-agreement database, run to the agreed standards, did not replicate Honorton's result.

Storm and colleagues (2010) meta-analysed a larger and differently selected body of studies and found a significant effect.

Subsequent meta-analyses have gone both ways depending on inclusion criteria, and the disagreement is now largely about which studies belong in the pool — which is the characteristic terminal state of a literature with a small, unstable effect.

We can say the aggregate picture is a hit rate that hovers a few points above chance and does not replicate reliably in independent hands. That is consistent with a very small real effect. It is also consistent with residual methodological artefact plus publication bias, and the two cannot be distinguished by more meta-analysis.

We can say the sensory-leakage concerns were legitimate and were addressed, which is to the field's credit. Autoganzfeld's automation removed the handling-cue problem that Hyman had identified.

We can say the effect has not behaved as a real effect usually behaves. Genuine phenomena tend to get clearer as methods improve and as investigators learn to produce them. This one has stayed the same size for forty years while the arguments have moved to the meta-analytic level, which is not the trajectory of a discovery.

We cannot close the question, and the honest position is that a small anomaly sits in this literature that nobody has been able to make either disappear or grow.

What can be said unambiguously is that Hyman and Honorton did something almost nobody in this site's territory has done: they agreed, in writing, in advance, what would count as evidence — and then let the results come.

Further reading

• Hyman and Honorton, "A Joint Communiqué: The Psi Ganzfeld Controversy" (Journal of Parapsychology, 1986).
• Milton and Wiseman (1999) and Storm et al. (2010), read as a pair.
• Ray Hyman's later assessments, which are courteous and unyielding.