In this book I study and empirically test whether belief attributions are context-sensitive, i.e., whether the truth value of a belief report of the form 'A believes that S' is sensitive to contextual parameters. More specifically, I examine whether the Referential Knowledge that the Audience possesses (that is, whether the hearer of a report is familiar with the name employed in it) and/or the Stakes for the agent at the time of attribution, affect patterns of attribution in a way such that the variation in them causes that a single report be correctly made in one context but not in other while nothing in the mental state of the agent has changed. To this end, I designed original experimental material and tested it on several samples of undergraduates at the Universitat de Barcelona and Universitat Autònoma de Barcelona. I here make an analysis of methodological approaches in empiric testing of some philosophical matters and offer considerations about experimentation on belief ascriptions in particular.