Abstract

Sequential Card Search (SCS) is a variant of a forced-choice ESP task, where subjects search for one card symbol (target) from among a finite set in an array of concealed cards. The subjects are asked to continue the search until the symbol is found, or remains undiscovered on the last card. The task is appealing and well accepted even by subjects who otherwise feel discouraged by experimental tasks with an overt ESP component. Also, the experiment provides a rich data structure as the probability of a chance hit varies with the number of alternatives. A simple test of the null hypothesis is based on the expected distribution of `average search run length' (ASRL), i. e. the number of sub-trials needed, on the average, to find the target.

The first SCS-study (Copenhagen 2000) was conducted with 26 participants recruited from two sub-populations: `clairvoyants', subjects practising ESP in their community sessions (group C, n = 11) and `controls', subjects not claiming any ESP abilities (group N, n = 15). A software implementation of the SCS task (program jcards) on a portable computer was used to run the experiments at participants' homes. Five classic Zener symbols were used. Each participant completed one run consisting of 100 complete trials.

The ASRL statistics showed no significant departure from the mean chance expectation (MCE) in either group C or N; the difference between the groups was also not significant. Hit rates were analysed separately by successive sub-trials, as the hit probability due to chance varies with the number of alternatives from which to choose, and numbers of sub-trials naturally decrease with decreasing number of alternatives. The deviations of hit rates from hit probabilities by chance show similar profiles between groups C and N, but the deviations, taken as a whole, are not significant.

The most remarkable feature of the data structure is an asymmetry observed at extreme ends of ASRL distributions in both groups C and N: extremely low values (i. e. in the expected direction) deviated farther from MCE than the extreme values on the opposite wing of the distributions. Although this observation has to be taken with extreme care, the joint probability of these departures, given the null hypothesis, is rather low (P z .0 12). This observation is strongly suggestive of an inhomogeneity in the source population; we might hypothesise that the experimental samples were statistical mixtures of a major portion scoring on the chance level and a minor fraction utilising an unknown, extra-chance cognitive process.