Holmstrom's (1982/99) career concerns model has become an important workhorse for the analysis of agency issues in many fields. The underlying signal jamming argument requires players to use information in a Bayesian way - which may or may not reasonably approximate real-life decision makers' behavior. Testing this theory with field data is difficult since typically little is known about the information that individuals base their decisions on, and this explains the dearth of empirical studies. We provide experimental evidence that the signal jamming mechanism works in a laboratory setting. Moreover, subjects' beliefs fit remarkably well requirements imposed by the Bayesian equilibrium concept.