We investigate polar questions in Farsi with three discourse particles: ke, dige, and mage. We demonstrate that such particles convey certain (not necessarily epistemic) attitudes that turn out to predict the correct settings for the original and contextual biases associated with the corresponding questions. Based on this observation, we argue that introducing attitude implications can significantly simplify the theory of question bias, as original and contextual biases need not be considered native to the question but rather fall out as its use conditions.