In this paper it is argued that studies involving subjects who attempt to verbally predict the outcome of experiments concerning unexpected and surprising psychological processes do not constitute a critical test of simulation theories. The failure of ‘observer-subjects’ to predict the outcome of such experiments does not demonstrate that their anticipatory abilities are ‘cognitively penetrable’ and thus based upon theory, as Nichols, Stich, Leslie and Klein (1996) maintain. This is because such studies are based upon the doubtful assumption that our ability to verbally predict another’s behaviour is a reliable measure of our ability to anticipate it. It is suggested that studies involving nonverbal measures of our anticipatory capacities would constitute a more decisive test of simulation theory.