◆ SpookStack

Approved For Release 2000/08/08 : CIA-RDP96-00789R003100140001-2 to permit evaluation of this hypothesis of random similarity, and I] must now turn to that aspect. Results To evaluate the chance hypothesis, the researchers obtained judgments of similarity between the dream content and the actual target for the night, and at the same time obtained judgments of similarity between the dream content and each of the other potential targets in the pool from which the target had been selected at random. The person judging, of course, had no information about which picture had been randomly selected as target; the entire pool (in du- plicate) was presented together, with no clue as to which picture had been the target and which ones had not. That is, in the experimental condition a picture was randomly selected from a pool and concentrated on by the agent, and in the control condition a picture was left behind in the pool. Any consistent difference between target and nontarget in similarity to dream content, exceeding what could reasonably be. ascribed to chance, was considered an apparent anomaly. The data available for the largest number of ses- sions came from judgments made by Judges who had no contact with the experiment except to receive (by mail, generally) the material necessary for judging (transcripts of dreams and interview and a copy of the target pool). For many sessions, judgments were also available from the dreamer, he or she, of course, made judgments only after completing participation in the experiment as dreamer (except in some series where a separate target pool was used for each night and the dreamer’s judgments could be made at the end of the session). For many sessions, judgments were made for the dream transcripts alone and for the total transcript including the morning interview; for con- sistency I have used the latter, because it involved judges who had more nearly the same information as the subjects. The only form in which the data are available for all series of sessions is a count of hits and misses. If the actual target was ranked in the upper half of the target pool, for similarity to the dreams and in- terview, the outcome was considered a hit. If the actual target was ranked in the lower half of the pool, the outcome was considered a miss. The hit-or-miss score is presented separately in Table | for judges and for subjects in the first two data columns. Where infor- mation is not supplied for one or the other, the reason is generally that it was impossible for the researchers to obtain it, and for a similar reason the number of cases sometimes varies. ! * Of course, usable judgments could not be obtained from the subject in precognitive sessions, because at the time of judging he or she would already know what the target had been. For Line F. the single subject was unable to give the extra time required for Judging, and for Line O ome of ihe four subjects failed to make Each data row in Table J refers to one segment of the research, and segments for the most part are labeled as they were in the table of Uliman et al. (1973, Pp. 275-277). Segments that followed the general procedure I described—all-night sessions, with an agent concentrating on the target during each of the percipient’s REM periods—are gathered together in the first eight lines, A through H (in five of these seg- ments, all but A, C, and H, a single percipient con- tinued throughout a series, and in four of these the Percipient was a psychologist). Other types of segments are presented in the rest of the table. Lines I, J, and K summarize precognitive sessions; here the target was not selected until after the dreaming and interview had been completed. The target consisted of a set of Stimuli to be presented directly to the percipient after it had been selected in the morning. Lines L and M Tepresent GESP sessions in which the percipient's dreams were monitored and recorded throughout the night, but the agent was attempting to transmit only before the percipient went to sleep or just after, or sporadically. Line N refers to a few clairvoyance ses- sions; these were like the standard GESP sessions ex- cept that there was no agent (no one knew the identity of the target). Finally, Line O reports on some GESP sessions in which each dream was considered sepa- rately, these formed a single experiment with four percipients, comparing nights involving a different target for each REM period with nights involving re- peated use ofa single target. Regardless of the type of session (considering the five types I have described), each session fell into one of two categories: (a) pilot sessions, in which cither a new dreamer or a new procedure was being tried out; these appear in lines H, K, and N, or (b) sessions in an experimental series, planned in advance as one or more sessions for each of two or more subjects, or as a number of sessions with the same dreamer through- out. Most of the researchers’ publications were de- voted to the results obtained in the experimental se- ries, but the results of the pilot sessions have also been briefly reported. A glance at the score columns for judges and for subjects is sufficient to indicate a strong tendency for an excess of hits over misses. If we average the outcome for judges and for subjects, we find that hits exceed misses on every one of the 15 independent lines on which outcome for hits and misses differs. (On Line E. hits and misses occur with equal frequency.) By a simple sign-test, this outcome would be significant beyond the 0.0001 level. I would not stress the exact value here, for several reasons. There was no advance judgments. In a few of the pilot sessions (Lines H, K, and N) only the subject's judgment was sought, and in some sessions only that of onc or more judges; in a few the mean judges’ rating was neither a hit nor a miss but exacuy at the middie. i rte November 1985 + American Psychologist 122) Approved For Release 2000/08/08 : CIA-RDP96-00789R003100140001-2