Biomedical ethics require that research subjects be aware that the drugs they take or procedures they undergo are designed to fulfill the conditions of the experiment and not to benefit a subject's health. This apparently straightforward distinction between research and treatment is a source of much controversy and misunderstanding. Ethicists have labeled this problem the "therapeutic misconception." This misconception and, more broadly, informed consent have been studied extensively. Nonetheless, the therapeutic misconception persists among research subjects. This paper argues that one factor overlooked in the persistence of the therapeutic misconception is the effect of the theoretical paradigm that guides the practice and analysis of informed consent. The paradigm poses an idealized model of communication that ignores social context. This paper examines informed consent practices associated with a cancer research trial to demonstrate an alternative approach to studying informed consent to research. Through analysis of informed consent session transcripts, it demonstrates the importance of taking account of not only what is said, but how and by whom it is said.