Some think that delusions are belief states (e.g., Bortolotti, 2010; DSM-5, 2015), and others think that they are not (e.g., Berrios, 1991; Campbell, 2001; Currie, 2000). Those of the latter conviction contend that delusions do not behave sufficiently like beliefs to qualify as such, partly because they tend to be less inferentially integrated with subjects? other beliefs. Responses to this contention (e.g., Bayne & Pacherie, 2005) have appealed to independently motivated treatments of belief systems as *fragmented*: as comprised of compartmentalized subsystems which need not be mutually consistent, and which need not be simultaneously accessible to the subject (Cherniak, 1986; Stalnaker, 1984).
This talk investigates the extent to which delusions can still be accommodated as beliefs within Yalcin (2016)?s analysis of belief as both fragmented and *relativized to questions*, understood as partitions of logical space. It is also an investigation of this framework itself, aiming to take some first steps towards formulating satisfactory constraints on *rationality* of fragmented, question-sensitive beliefs. Those who want to classify delusions as beliefs typically do not want to classify them as just any type of beliefs, but minimally as *irrational* beliefs (Bortolotti, 2010). Without added constraints on rationality, Yalcin?s model runs the risk of classifying too many beliefs as rational; including, I will argue, delusions (if understood as beliefs).
Presented at the Fifth PLM Workshop on Delusion in Language and Mind, in Amsterdam, October 23---24, 2020