Whence the virtue?
Virtue epistemologists have seen in situationism a threat to their framework.They have worried that if our dispositions are sensitive to features of context, then we do not have the kind of stability of character required for epistemic virtue.
Sometimes, virtue epistemologists have attempted to befriend the threat, suggesting that the right response was to enlist the context: in appropriately structured environments, an agent can display stable characters and (therefore) epistemic virtue (Alfano and Skorburg 2017). I agree that enlisting the context is the appropriate response, but I will argue that an appropriately structured environment does not bring agents to manifest stable dispositions. Rather, it will cue shifts between epistemic arrogance, humility and servility, as appropriate to the context.7What cues shift the dispositions of agents from context to context? We know quite a lot about the cues to which agents are sensitive when they defer and when they imitate. Many of the cues to which we're sensitive are cues about other agents and their dispositions. Starting in early childhood, and increasing with age, agents are more likely to accept the testimony of those who evince signs of benevolence and of competence (Mascaro and Sperber 2009; Sperber et al. 2010). Perhaps because political orientation is a proxy for benevolence toward us (Levy 2019), we are sensitive to cues that others share our political views (Nyhan and Reifler 2013). We are often critical recipients of testimony, but these cues tend to trigger epistemic deference in us.
Of course, message content matters. But our assessment of the message is heavily influenced by information about the testifier. Maoz, Ward, Katz and Ross (2002) found that whether a peace proposal was presented as having been put forward by Palestinians or Israeli Jews was a significant predictor of attitudes toward it among their Palestinian and Israeli Jewish subjects.
Indeed, information about the attitudes of other agents may trump message content. Cohen (2003) found that information about whether welfare policy proposals were supported by House Democrats or Republicans overwhelmed policy content when it came to support.These patterns of deference might be best understood in the context of the conformity bias: we defer to those we identify with, because doing so effectively allows us to distinguish signal from noise.In the absence of cues that other agents are like or unlike us, or in addition to them, there is evidence that numbers matter. Of course, that's unsurprising: multiple sources of independent testimony constitute evidence in favor of a hypothesis that is stronger than a single source (other things equal). It is also unsurprising that we may be influenced by sheer numbers inappropriately, when we fail to recognize that the sources are not properly independent of one another. The range of cues for widespread agreement to which we are sensitive is much more surprising. For instance, the familiarity of a claim is a cue for acceptance, probably because familiarity is usually the product of repeated previous exposure, and such exposure is a proxy for widespread agreement. But, of course, familiarity is not always a reliable proxy for widespread agreement. Repetition of a claim increases familiarity, whether the claim is repeated by many different voices or the same one repeatedly (Weaver, Garcia, Schwarz and Miller 2007). In fact, repetition of a claim explicitly in the service of debunking it increases familiarity and may backfire (Schwarz, Sanna, Skurnik and Yoon 2007).
Apparent consensuality, cues that the source shares our political values, and cues of competence (even in unrelated domains and even evidence that the person takes themselves, rightly or wrongly, to be competent; Martens and Tracy 2013) are all triggers for epistemic deference. What triggers epistemic arrogance? The myside bias is triggered when the hypothesis under consideration is unfavorable in some way to the person (Sharot and Garrett 2016).
A hypothesis might be unfavorable to someone in multiple ways: because its truth conflicts with their selfimage, because it would predict bad consequences for them, or simply because they have staked something (their reputation, say) on its being false. These kinds of circumstances set the stage for motivated reasoning, whereby evidence against the view we defend is less likely to come to mind at all, and is systematically discounted when it cannot be ignored.Importantly, the cues for acceptance — for servility — are epistemically adaptive. We should be sensitive to cues of consensuality.We should be sensitive to cues of prestige.We should care about the benevolence and competence of others: knaves may try to epistemically exploit us and fools may unwittingly deceive us. Raising and lowering our epistemic defenses in response to these cues is how we ought to behave, in the service of knowledge.
Once again, content matters too.8 We are more likely to accept claims that cohere with our own beliefs, for instance.We should not think that we exhibit only epistemic servility and arrogance: no doubt much of the time we behave in the kinds of ways those who value epistemic humility would applaud. I have no idea how to assess the proportion of the time we are epis- temically arrogant, servile or somewhere in-between (no doubt, it is very context dependent). However, given that our success as deliberators is and remains very heavily dependent on group deliberation at a time and cumulative culture over time, and that epistemic arrogance and servile deference are plausibly adaptations for cumulative culture and group deliberation (respectively), these dispositions are not marginal features of human cognition. They are central. Accordingly, we should not expect calls for us to be epistemically humble to be either easy to conform to or to provide a panacea to our epistemic problems. Epistemic humility is risky: it risks some of our most important epistemic practices.
40.4