The Autoganzfeld Series

Honorton's laboratory in Princeton rebuilt the ganzfeld as a machine. A computer selected the target from the pool using a hardware source rather than a person or a table of numbers, controlled the playback to the sender, timed the session, presented the four candidate images to the receiver in a randomised order at the end, and recorded the judgement, so that no human being involved in a session knew the target while the session was running. The two rooms were separately shielded and separated, with the sender's room isolated so that no sound could pass. Each of these features answers a specific line in the 1986 document.

The reported outcome across eleven series and 240 participants was a hit rate of roughly thirty-two percent where twenty-five was expected, which is a small effect by the standards of a laboratory demonstration and a large one by the standards of this literature. It was presented to mainstream psychology in Psychological Bulletin in 1994, in a paper co-authored with Daryl Bem, in language deliberately pitched at readers with no prior interest in the subject. One detail of the automation is easy to miss and mattered a great deal to the argument. Because the computer selected and presented the four candidate images at the judging stage, the order in which they appeared was randomised too, which closed a route by which an experimenter's unconscious preference could favour the target.

Three things kept the dispute open. The first is that automation closes the loopholes it was designed to close and no others; critics raised the acoustic isolation of the two rooms, and specifically whether a video target with a soundtrack could be faintly audible, which the 1986 list had not anticipated in that form. The second is that every series came from one laboratory, run by the person whose hypothesis was being tested, and the field's own literature on experimenter effects makes that hard to wave away. The third and most consequential is what happened next. Milton and Wiseman pooled thirty ganzfeld studies conducted after this series and found no significant overall effect, which is the outcome one expects if the original result was specific to its laboratory. Storm, Tressoldi and Di Risio later pooled free-response studies from 1992 to 2008 and found a significant effect, using a wider inclusion rule. Neither team accepts the other's pool. The laboratory itself closed in 1989 when funding ran out, and Honorton died in 1992, so nobody could simply run the series again under his supervision and see. The result stands as reported, unreplicated by an independent group, and unexplained.