Precis of prejudice: a study in non-ideal epistemology

ABSTRACTThis article provides an overview of ideas and arguments developed in my book Prejudice: A Study in Non-Ideal (Oxford University Press, 2021; paperback edition, 2022).KEYWORDS: Epistemologysocial epistemologynon-ideal epistemoloyepistemic rationalityprejudice Disclosure statementNo potential conflict of interest was reported by the author(s).Notes1 To be clear, this is not meant to tell us anything about whether particular prejudiced individuals are justified in believing as they do, whether particular kinds prejudiced beliefs are typically justified, and so on. Indeed, part of the point of the book is to discourage these sorts of questions: for virtually any class of beliefs, the question of whether a subject is justified in having such beliefs can only be answered after careful examination of their evidential situation and their cognitive processing. Given the proper methodology, we should not expect broad generalizations at this level. More about this in my response to Thomas Kelly below.2 Moreover, reflection suggests that even true beliefs can be prejudicially deployed in context, as I argue in Begby (Citation2021, 10–11).3 Indeed, on my view, stereotype-reasoning just is the manifestation of categorization in the domain of social cognition. To argue that our cognition should only ever track individuals qua individuals is basically to give up on the idea of social cognition altogether.4 Recall Fricker’s stipulation that ‘[t]he idea of a prejudice is […] most naturally interpreted […] as a judgement made or maintained without proper regard to the evidence’ (Fricker Citation2007, 32–33, my emphasis.)5 The fact that so-called ‘looping effects’ (cf. Haslanger Citation2011, 196–198, drawing on Hacking Citation1999) might be in play here, does nothing to change the fact that these observations might nonetheless constitute evidence.6 More about this in my response to Giulia Napolitano below.7 Note the echoes of Gordon Allport’s view (Citation1954, 9) that ‘[p]rejudgments become prejudices only if they are not reversible when exposed to new knowledge. A prejudice, unlike a simple misconception, is actively resistant to all evidence that would unseat it’.8 In particular, it helps shed light on the phenomenon of ‘echo chamber construction’, as argued in Nguyen (Citation2020).9 The mechanics of evidential preemption is further discussed in my response to Thomas Kelly below.10 These issues receive further discussion in my response to Renee Jorgensen below.11 Correspondingly, when we talk about ‘algorithmic bias’, it seems natural to assume that the phenomenon arises specifically from the deployment of algorithms in decision making (perhaps computerized algorithms, in particular).12 On this, see Begby (Citation2018). In the book (Begby Citation2021, 161–163), I explore and cautiously endorse the possibility that the considerations levied by moral encroachment are good ones: the only problem is that they don’t, strictly speaking, bear on belief, but rather on a distinct psychological function called ‘acceptance’. I leave that out of the reckoning here, but hope to develop the idea further in future work.13 For more on this, see Begby (Citation2021, 165–170).14 For an illustration of how these different dimensions of responsibility might be codified, the book offers a look at the diverging conceptions of the grounds of responsibility that provide the foundations of criminal law and tort law, respectively: in brief, criminal law holds people responsible only if they satisfy a minimum ‘mens rea’ (‘guilty mind’) criterion. By contrast, tort law can hold people liable for damages even after acknowledging that they were in no position to know that what they did was wrong.15 Many thanks to Holly K. Andersen for valuable feedback, to Giulia, Renee, and Tom for agreeing to serve as critics, and to Paul Silva Jr. and the Cologne Concept centre for hosting the original symposium, which can still be viewed at https://youtu.be/C-1j8j4OSpY?si=e1q8bO9KEXpgEMIl.