Resistant Beliefs, Responsive Believers∗ Carolina Flores April 2021 Abstract Beliefs can be resistant to evidence. Nonetheless, the traditional view analyzes beliefs as evidence-responsive attitudes. I address this tension by developing an account of belief that does justice to its epistemic role without limiting beliefs to idealized epistemic agents. In doing so, I argue for a capacities-first account of belief: belief requires the capacity for evidence-responsiveness. More precisely, if a subject believes that 푝, then they have the capacity to rationally respond to evidence bearing on 푝. Because capacities for evidence-responsiveness are fallible and may be masked, beliefs can be held in the face of strong counter-evidence. Indeed, I will argue that our best science of belief supports the claim that evidence- resistant beliefs result from masks on evidence-responsiveness capacities. This account of belief not only allows for resistance to evidence, but provides us with a framework for describing and explaining actual cases of evidence-resistance. Word count: 9,135 1 Introduction Belief plays a central role in both epistemology and cognitive science, yet the roles it plays in these two areas are in tension. Belief’s role as the central object of epistemic assessment motivates the orthodox view in epistemology that belief is constitutively evidence-responsive. This is hard to square with the cognitive science of belief revi- sion, according to which resistance to evidence is the norm. In the face of this tension, one may be tempted to throw away the orthodox con- ception of belief and start from scratch. But, in line with recent work by Helton (forthcoming) and Ganapini (forthcoming), I think the orthodox view—if articulated carefully enough—has much to offer. As traditionally held, it furthers our understand- ing of the epistemic role of belief. Perhaps more surprisingly, I will argue that it also helps us systematize findings in the cognitive science of belief. ∗Acknowledgements: Thanks to Frankie Egan, Danny Forman, Daniel Hoek, Dee Payton, Jake Quilty- Dunn, and Ezra Rubenstein for comments on past drafts. Thanks also to Austin Baker, Eric Bayruns Garcia, D’i Black, Laura Callahan, Andy Egan, Megan Feeney, Adam Gibbons, Samia Hesni, Tyler John, Jackson Kernion, Kelsey Laity D’Agostino, Eric Mandelbaum, Ting-An Lin, Jorge Morales, Albert Newen, and Jill North for helpful discussion. Special thanks to Elisabeth Camp, Verónica Gómez Sánchez, Susanna Schel- lenberg, Ernest Sosa, Christopher Willard-Kyle, and Elise Woodard for detailed feedback on multiple drafts. 1 To argue for these claims, I articulate the orthodox claim that belief is constitu- tively evidence-responsive by developing the following view: The Evidence-Responsiveness Capacities View: Necessarily, if 푆 believes that 푝, then 푆 has the capacity to rationally respond to evidence bearing on 푝. On my view, belief’s evidence-responsiveness is not a matter of always, frequently, or reliably changing one’s mind in accordance with the evidence we have. Instead, it is about rationally responding when conditions are right for the exercise of our rational capacities. Such conditions include ones that are internal to our cognitive systems, which, needless to say, include much beyond rational capacities. But these conditions can and do fail to be in place for many of our beliefs, leaving space for plenty of evidence-resistance. This view vindicates the orthodox view of belief in the face of cognitive scienceby showing that we can account for evidence-resistant beliefs while holding that belief is constitutively evidence-responsive. At the same time, my way of articulating the orthodox view is robust enough to do justice to the epistemic role of belief. It provides a naturalistic way of spelling out the claim that belief aims at truth. And it correctly excludes attitudes that are not subject to epistemic assessment based on their respon- siveness to evidence—such as imaginings and acceptances—from the belief category. To argue for this view, I will proceed as follows. In §2, I motivate the claim that belief is constitutively evidence-responsive, and argue that other proponents of that claim leave it under-specified. This makes the claim susceptible to two challenges which I outline in §3. In §4, I put forward the Evidence-Responsiveness Capacities View, providing an account of such capacities. I, §5, I articulate empirical findings on evidence-resistance in the light of the discussion of capacities in §4 showing that this view addresses the two challenges outlined in §3. In §6, I wrap up by showing how, while accommodating evidence-resistance, the view is still robust enough to satisfy epistemic desiderata on a notion of belief. The result is a conception of belief that unifies work in epistemology and cognitive science. 2 The Orthodox View On the orthodox view, belief is constitutively evidence-responsive.¹ This means that evidence-responsiveness is one of the key features that makes an attitude a belief (as opposed to an imagining, acceptance, desire, or another distinct attitude-type).² In particular, it implies that evidence-responsiveness is necessary for belief: no attitude 1. Its status as the orthodox view is made clear by its proponents. A recent sample includes Currie and Ravenscroft (2002), Egan (2008), Gendler (2008), Helton (forthcoming), Levy (2015), Mandelbaum (2016), Shah (2003), Shah and Velleman (2005), Smith (2003), Leeuwen (2014), and Velleman (2000), with notable earlier proponents including Davisdson (see e.g. Davidson 1985), Dennett (see e.g. Dennett 1981), and McDowell (see e.g. McDowell 1998. 2. See Burge 2010, 57–59 for more on constitutive conditions in philosophy of mind. 2 counts as a belief unless it is evidence-responsive.³ The wide appeal of this claim is best explained by the way in which it accounts for the epistemic role of belief. On an epistemology-first view, what is distinctive of belief (as compared to, for example, imaginings, acceptances, or desires) is the fact that it is subject to epistemic standards. Proponents then move from normative constraints on belief to descriptive claims about the nature of belief through the argument: con- stitutively, belief is subject to epistemic standards (truth, knowledge, etc.); epistemic standards only apply to attitudes that are evidence-responsive; therefore, belief is con- stitutively evidence-responsive.⁴ This is a popular line of argument. For instance, Velleman (2000) moves from argu- ing that belief aims at truth (which he takes to entail that truth is belief’s constitutive standard of correctness) to claiming that belief is regulated for truth. The resulting view, as characterized in a later paper, is that it is a “conceptual truth” that “the belief that 푝 tends to be… reinforced by additional evidence of it, and to be extinguished by evidence against it” (Shah and Velleman 2005, 500).⁵ Similarly, Tamar Gendler moves from the claim that apportioning beliefs to the evidence is a normative constraint on belief to excluding attitudes from the belief cat- egory because they are not in fact evidence-responsive (Gendler 2008, 565–566). The resulting view is that “beliefs are, roughly speaking, evidentially sensitive commit- ments to content that are quickly revisable in the face of novel information” (Gendler 2012, 763). More explicitly articulating the argument above, Grace Helton has recently ap- pealed to an epistemic ‘ought-implies-can’ principle to argue from the epistemic obli- gation to revise one’s beliefs in response to relevant counter-evidence to the view that subjects have the ability to revise their beliefs in response to such evidence (Helton forthcoming). Despite their shared motivation to hold that belief is constitutively evidence- responsive—namely, doing justice to the epistemic role of belief—philosophers differ in what they take evidence-responsiveness to require, and often neglect to specify what they mean. Given that it is clearly false that beliefs are always perfectly apportioned to the ev- idence, statements of the view tend to leave room for failures to respond to evidence. Proponents usually grant something to the effect of “[given that] belief can be influ- enced by evidentially irrelevant processes such as wishful thinking, responsiveness to evidence must be weak enough to leave room for such additional influences” (Shah 3. Evidence-responsiveness is not sufficient for belief. Plausibly, there are other necessary conditions on belief, such inferential promiscuity (i.e. figuring in inferences that lead to other beliefs) and a distinctive role in action production (i.e. interacting with desires and other motivational states in ways that leads to action). 4. Many (e.g. Shah and Velleman (2005), Engel (2007), Burge (2010)), hold that, constitutively, epistemic standards determine the correctness of beliefs, not just that they constitutively apply to belief. One can hold that epistemic standards apply to belief while holding that what one ought to believe is a function of moral or practical factors, thus rejecting the claim that epistemic standards determine the correctness of beliefs. See Maguire and Woods 2020 for a compelling articulation of such a view. 5. See also Shah 2003 for further discussion, especially fn. 45, where Shah explicitly sketches this line of argument. 3 and Velleman 2005, 500). But they do not usually specify what this “weak enough” evidence-responsiveness amounts to. They claim merely that beliefs “tend to” Shah and Velleman 2005 evidence-