Purpose – A placebo effect is a (positive) change in health outcomes that is due to a (positive) change in beliefs about the value of a treatment. Placebo effects might be “behavioral,” in the sense that revised beliefs lead to behavioral changes or new actions that in turn yield changes in health outcomes. Placebo effects might also include a “physiological” component, which refers broadly to non-behavioral, brain-modulated mechanisms by which new beliefs cause changes in health outcomes. Nearly all formal economic models of human behavior are consistent with behavioral placebo effects, but strongly inconsistent with their physiological counterparts. The reason is that the latter effects can imply that expectations enter, rather than multiply, state-contingent preferences. It is therefore unfortunate that little evidence exists on physiological placebo effects. We report data from novel clinical experiments with caffeine that seek to provide such evidence. Methods – Subjects visit the clinic on multiple occasions. On each visit they ingest either a placebo or caffeine pill. Subjects only know the probability with which the pill includes caffeine. We obtain physiological measurements prior to ingestion and at 30, 60, and 90min after ingestion. Importantly, we constrain subjects to remain seated and read pre-selected magazines during the interval between treatment and outcome measurement. Findings – Our design provides particularly clean inference because it (i) eliminates the possibility of behavioral confounds; (ii) provides for measurements at the individual level; (iii) manipulates beliefs without deception; and (iv) uses salient rewards. We find evidence for the existence of physiological placebo effects mediated by expectations. Implications – Our results are consistent with the possibility that the prefrontal cortex provides external, top-down control that modulates physiological outcomes, and make a case for the importance of research geared toward developing appropriate and tractable frameworks that accommodate non-linear relationships between expectations and preferences.
