From Rhine to the ganzfeld

The teaching

As the experimental parapsychology of J. B. Rhine and his successors teaches it.

Psi and its kinds

Parapsychology studies the claim that the mind can know or act beyond the known senses and the known forces. J. B. Rhine grouped the knowing under the name extrasensory perception, ESP, in three kinds: telepathy, knowing another's thought; clairvoyance, knowing an object or event that no one knows; and precognition, knowing what has not yet happened. The acting he called psychokinesis. The field later took the Greek letter psi as a name for the whole, chosen because it assumes no theory of how any of it might work.

The cards

Rhine's tool was a pack of twenty-five cards, five each of five plain designs: a circle, a cross, three wavy lines, a square and a star. Simple shapes were chosen so that every guess could be scored right or wrong without argument. A subject calls each card in turn, a run of twenty-five calls is the unit, and since each call has one chance in five, chance alone gives an average of five hits a run.

The test conditions

Rhine's laboratory told the kinds of ESP apart by how a test was arranged. When an agent looks at each card as the subject calls it, telepathy and clairvoyance cannot be separated, and the result counts as general ESP. When no one looks at the cards until the run is over, only clairvoyance could explain a hit. Rhine's subjects did this in two ways: by calling the top card of a face-down pack, which was then lifted off unseen and laid face down on a second pile, or by calling the whole pack down through, from top to bottom, without anyone touching it until the end.

For pure telepathy there is no card at all: the agent thinks of a symbol by a private code and records it, so that no physical object exists for clairvoyance to reach. For precognition the subject calls the order the pack will have after it has been shuffled later, by hand or by machine.

Counting

The arithmetic is the heart of the method. A single run proves nothing: by chance alone the scores of runs of twenty-five scatter widely, most of them between about two and eight, and a nine or ten turns up now and then with no power at work. What matters is the total over many runs, compared with the spread that chance would produce. The excess over chance is divided by that spread, and the larger the ratio, over hundreds or thousands of runs, the less likely luck becomes as the explanation.

Psi-missing, decline, sheep and goats

The field's own work added doctrines of its own. Psi-missing is scoring well below chance, too far below to be luck, which parapsychologists read as psi working in reverse, the subject avoiding the target without knowing it. The decline effect is the falling off of scores over a run, a session or a subject's career, read as a sign of boredom and lost spontaneity. Gertrude Schmeidler's sheep-goat effect holds that those who accept that ESP is possible, the sheep, tend to score above chance, and those who reject it, the goats, below.

These doctrines matter to anyone weighing the field, because together they allow almost any score to be read as psi: high scores, low scores and falling scores alike. A test can settle a claim only if it is laid down beforehand which result would count against it.

The safeguards

A century of argument has taught what a sound test needs. The subject must have no sensory route to the target: no sight of the cards, front or back, no reflections, no glimpse of the agent's face, no whispering, and ideally separate rooms or buildings. The order of targets must be truly random, from tables of random numbers or a machine, not from shuffling by hand. The calls must be recorded before checking, and checked by two people independently, ideally people who cannot tell which calls were meant for which targets.

The number of trials must be fixed in advance, so that a run of luck cannot be stopped at its best moment, and every session must be reported, the dull with the bright. The experimenter's hopes are a source of error in themselves, and the best designs keep anyone who has hopes away from the scoring.

The ganzfeld

The free-response test came to replace the cards. In the ganzfeld the receiver reclines in mild sensory isolation, with halved table tennis balls over the eyes under red light and an even hiss in the ears, and after a period of relaxation speaks aloud every impression for about half an hour, while a sender elsewhere watches a target picture or film clip chosen at random from a pool. The idea behind it, Charles Honorton's, is that any psi signal is faint and drowned by the noise of the senses, and that quieting the senses lets it through.

Judging a free response

At the end of a session the receiver, or a judge who has not seen the target, ranks four candidates, the target and three decoys, against the record of what was said. A first place for the target is a hit, and by chance one session in four is a hit. The safeguards here are particular. The judging set must be fresh copies, never the picture the sender handled, so that no smudge or crease can point to it. The target must be chosen by machine, and the order of the candidates shuffled, so that no one's habits can leak in.

Counting many studies

When many small studies are pooled, a further danger appears: studies that found nothing tend to go unpublished and stay in the file drawer, so that the printed record looks stronger than the whole. Two remedies are now standard. Robert Rosenthal's fail-safe number asks how many unpublished studies that found nothing would be needed to cancel a pooled result. And registering each study, its design, its numbers and its analysis, before it is run makes the file drawer visible, since a registered study that is never reported is itself a result.

What the method asks

The method's last rule is the strictest. A finding is established when competent experimenters following a published protocol can obtain it whoever they are, believers or not. That standard, more than any single score, is what the field has sought for a century, and it is the test that every claim in this art must pass.