Many studies document how people willfully choose ignorance to avoid decision environments in which the pursuit of individual interests transparently conflict with the pursuit of socially favorable outcomes. However, quite a few results, many from the same set of studies, show much lower rates of ignorance in seemingly similar situations. Why does ignorance prevail in some situations but not others? What conditions must hold for it to be useful as an exculpatory tool? Looking at both existing and new data, I argue that, while models of rational signaling with standard inference can account for a wide set of results, other, typically unmodeled features of the environment clearly are important and warrant more attention from both theorists and experimentalists.