Lack of standardization and unblinding threaten the research of mechanisms involved in expectancy effects on pain. We evaluated a computer-controlled virtual experimenter (VEx) to avoid these issues. Fifty-four subjects underwent a baseline-retest heat pain protocol. Between sessions, they received an expectancy manipulation (placebo or no-treatment) delivered by VEx or text-only control condition. The VEx provided standardized "social" interaction with the subjects. Pain ratings and psychological state/trait measures were recorded. We found an interaction of expectancy and delivery on pain improvement following the intervention. In the text conditions, placebo was followed by lower pain, whereas in the VEx conditions, placebo and no-treatment were followed by a comparable pain decrease. Secondary analyses indicated that this interaction was mirrored by decreases of negative mood and anxiety. Furthermore, changes in continuous pain were moderated by expectation of pain relief. However, retrospective pain ratings show an effect of expectancy but not of delivery. We conclude that we successfully applied an automated protocol for inducing expectancy effects on pain. The effect of the VEx regardless of treatment may be due to interactions of attention allocation and locus of control. This points to the diversity of expectancy mechanisms, and has implications for research and computer-based treatment applications.