Citation

Here we reconsider expectation bias in so-called presentiment experiments, with focus on how to handle it. In such experiments, presentiment is usually thought to be demonstrated by showing that significant physiological differences precede stimuli presumed to give rise to different arousal levels. Often these differences suggest that physiological arousal is more likely to precede arousing rather than calming stimuli. Conceivably, however, such reactions can be explained as resulting from expectation bias of the gambler’s fallacy type. This bias is based on the (false) notion that the likelihood of an arousing stimulus being presented grows as the number of consecutive calming stimuli increases. Different ways of controlling or avoiding the bias are discussed. Our resulting recommendation is to use analysis of variance (ANOVA) to separate the effect of the bias from the hypothetical presentiment effect, preferably at the trial-by-trial level. We also recommend applying ANOVA to each participant separately and using a “counting” method to test for possible presentiment effects at the group level. Application of ANOVA is illustrated using a simulated example. We anticipate ANOVA can handle not only the gambler’s fallacy bias but also similar biases, in presentiment

experiments as well as in some conscious precognition experiments.