2.1 Worries about the value monism of JAB 1 and JAB 2

Zagzebski considers both JAB 1 and 2 to be appropriate (though competing) ways to specify who counts as an epistemic authority in the Preemption Thesis. Both version set up a strong standard and thus limit the scope of that thesis. And both are based on what Zagzebski considers the natural foundations of our human reasoning processes. JAB 1 is based on her idea that we have a natural desire for truth, and that, to be active epistemic agents, we must also believe that we can achieve true belief (Zagzebski Reference Zagzebski2012: 35). JAB 1 then recommends treating someone as an authority when we have good reason to believe that doing so will lead us to the truth. But the desire for truth is not the only natural foundation Zagzebski discusses; she also bases her conception of epistemic rationality on the process of resolving dissonance (between beliefs and between beliefs and desires) that she thinks we naturally engage in. Rational processes are those that achieve the goal of resolving dissonance well, and these are characterized as conscientious self-reflection. JAB 2 recognized the value of this type of rationality, and advises us to view as authorities those whose beliefs, when adopted by us, are most likely to meet this standard of rationality.

JAB 1 and JAB 2 are each formulated to prioritize one of the two natural foundations of our cognitive lives. However, the value monism embedded in each version may problematically encourage us to take as an authority someone we have good reason to think will fail to provide the other desirable feature. JAB 1, with its exclusive focus on the value of truth and avoiding error, may lead us to accept as an authority and Opaque Oracle known to hold only true beliefs, but whose beliefs are so obscure that one cannot integrate them into one's overall body of beliefs. It may be true that “If you cross the river, a great empire will be destroyed,” and so in believing as the Oracle does one comes to have a true belief. But incorporating this belief within one's existing body of beliefs may be difficult or potentially misleading. Thus one's interpretation of this oracular pronouncement may not survive conscientious self-reflection. On the other hand JAB 2 asks us to focus on what survives conscientious self-reflection, without a concern for truth; it seems then to endorse accepting an Obsequious Advisor as an authority. Such an advisor, aiming to please, may only advise (or even, to fit the requirement of the preemption thesis, only believe) those things that you would accept on your own. Perhaps he is even psychologically so canny that he can suggest only beliefs that will perfectly survive your self-reflection. But as an advisor he is unconcerned with the truth, only with satisfying your self-reflection. And so he is useless in achieving the goal of reaching the truth and avoiding error, and may even lead you further from the truth.

The problematic outcome of each formulation follows directly from the monism that each encodes; JAB 1 and JAB 2 are offered as alternatives, not as supplements to each other. As a result, taken alone, each focuses on one value at the expense of another. But if we are pluralists about epistemic value, including truth and surviving self-reflection (perhaps along with other values like understanding and wisdom) in our considerations, we should only accept as authorities those who can help us achieve a balance of our epistemic values.Footnote ⁴

The monism of Zagzebski's two versions of the normal justification thesis is also notable because this monism is not found in Raz's original.

Note that in Raz's formulation there is no restriction on the sorts of reasons under consideration. In so far as it is plausible that we have reasons of many types, any application of the normal justification thesis would involve balancing reason types against each other.

In the realm of epistemic authority, we may well want to constrain the reasons considered in a formulation of the normal justification thesis to only epistemic reasons; in the epistemic context we often want to exclude practical and moral reasons for belief, focusing only on epistemic reasons for belief. But JAB 1 and JAB 2 both go too far with such a constraint, by limiting reasons not only to epistemic reasons but further to only a specific kind of epistemic reason. If we recognize a number of kinds of epistemic reasons (following from the recognition of a number of epistemic values), then each of these kinds should be permitted to play a role in the selection of an epistemic superior.

To further illustrate the problem with epistemic monism, we might consider how implausible a similar monism would be in the case of political authority. We might recognize security as a legitimate political value, and so look for an authority who can provide maximal security. But such an authority would likely fail to provide other relevant goods such as free speech. Rather than selecting an authority who can maximize only security or maximize only free speech, we instead look for an authority who, by offering authoritative directives, can help us achieve a balance of these political values. This authority is the one who allows us to better comply with all the reasons that apply to us in a particular circumstance. Thus, an extension of Raz's normal justification thesis would be better captured by a more inclusive version of Zagzebski's JAB 1 or JAB 2, a justification thesis which recognizes that there are a range of epistemic values which we might look to an authority to help us realize.

2.2 Epistemic preemption: examining arguments from Raz

In his chapter laying out the preemption thesis for action, Raz gives two arguments in favor of allowing the directives of an authority preempt our reasons for action. The first we might call the “double- counting” argument. Raz illustrates the problem by considering a simple social rule such as “when being with one person and meeting another, one should introduce them to each other” (Raz Reference Raz1986: 57). Such a rule might be justified by noting that following it facilitates social contact. When considering whether you should introduce two friends, one might note that both (a) it will facilitate your present social circumstance and (b) doing so is in keeping with a well-motivated social rule. So, it might seem that you have two reasons to introduce your friends to each other. But this looks like double counting, and Raz points out that, “the reasons for the rule cannot be added to the rule itself” (Raz Reference Raz1986: 58). The rule here could be the content of an authoritative directive. But then, “either the directive or the reasons for holding it to be binding should be counted, but not both” (Raz Reference Raz1986: 58). So, since the authority has already considered my reasons (wanting my own social facility) while considering the reasons for her directive (wanting general social facility) to count my reasons again would be double-counting. So the reasons that such a directive gives me should replace my original reasons.

The double-counting argument gives adequate grounds for preemption when I know that the authority has already taken my reasons into account. But how can I know that my reasons have been counted? This is ensured in Raz's account by the addition of the following requirement:

This thesis provides a justification for preemption; and authoritative directive should already have counted my reasons once. Counting them (along with the reasons of all the others to whom the directive is addressed) is required for a legitimate authoritative directive.

Zagzebski, in moving the focus to epistemic authorities, want to consider a wider array of potential authorities, including those who give me no directives. In fact, they may not communicate with me, or even know that I exist. (This is why testimony is considered in her book as only one special case of belief on authority.) Rather than focusing on directives, which can only be given by those who are considering my situation, Zagzebski is treating the beliefs of a recognized epistemic superior as authoritative. There is an advantage to this broadening the class of authorities, in that it allows one to treat as authoritative someone from the far past or someone who is not interested in directing the beliefs of anyone else. But the cost of this broadening is that it no longer dictates the reasons that the authority must consider; in particular, my reasons may not have been considered at all. This is why Zagzebski updates Raz's thesis to the following form:

While this formulation allows us to recognize a wider array of authorities, it also drops out any mention of the reasons of the person who is responding to the authoritative belief. Thus, while for Raz our own reasons must be replaced by the reasons from the authority in order to avoid double-counting, the same “double-counting” argument for preemption does not work within Zagzebski's framework.

Raz's second argument for preemption does more easily extend to cover Zagzebski's epistemic cases. This might be called the “accuracy” argument. It holds that I can best improve the accuracy of my performance by following the directives of the authority; to add my own reasons to the reasons given by the authoritative directive would be to water down the accuracy of the authority.

Zagzebski clearly wants to make use of this argument from Raz and she cites his example in full.

This argument for preemption, by focusing on the accuracy of the putative authority, applies equally to action and to belief. In the case of action, the putative authority is picked out by the fact that following its directives is a more efficient means to comply with one own reasons (Raz Reference Raz1986: 53). In the epistemic case, following the beliefs of the putative authority is an efficient means to forming more true beliefs and fewer false ones (JAB 1) or a means to forming rational belief (JAB 2). In both kinds of cases, the goal in following the belief or directive of the putative authority is to improve accuracy. The argument above attempts to show that this goal is undercut if one does not allow the putatively authoritative reasons to preempt one's own reasons, and instead simply adds those reasons to one's own.

This argument shows that failing to preempt some of our reasons may undercut our accuracy. To see exactly which reasons are at issue, it is easiest to consider the different cases of fit between our beliefs and those of our recognized epistemic superior.

Epistemic superior believes P; I believe P:

The concern about accuracy gives me no reason to preempt my own reasons in this case. My reasons and the reasons that I get from the belief of my recognized superior are in agreement. Thus, to archive the level of accuracy of my epistemic superior, it is sufficient for me to add the reason from their belief to my own reasons for belief.

As we saw above, the argument from double counting also does not apply here, even though we are in agreement. Since my theoretical (third person) reasons will often not be included in the reasoning of my epistemic superior, and since my deliberative (first person) reasons cannot be included in their reasoning, respecting my own reasons will always add evidence to the reasons for belief I can get from the belief of my superior. So, by avoiding preemption, not only will my resultant belief be accurate, it will also be better supported by evidence.

Epistemic Superior believes P; I believe not-P:

In this case, concerns about accuracy might provide an argument for preempting my own reasons for believing not-P. If I fail to replace my own reasons with the reason I get from the belief of my superior, but rather simply add this new reason to my set of reasons, my overall conclusion might be either P or not-P depending on the weight of the evidence. When I do not follow the belief of my epistemic superior I am at greater risk of error and I undermine my goal of accuracy. This case then is the central target of the accuracy argument. One's reasons for believing not-P are exactly the ones that the accuracy argument aims to make me preempt.Footnote ⁵

Epistemic Superior believes P; I have a degree of belief in P:Footnote ⁶

The other cases only consider full belief. But, of course, we might often find ourselves with a more nuanced degree of belief towards P, one that neither counts as full belief in P nor as full believe in not-P. Does the accuracy argument show that we should preempt in these cases?

In her argument for preemption, Zagzebski gives an example that might be particularly helpful in this context. She considers the behavior of animals and humans with respect to probabilistic distributions, noting that animals tend to settle on a simple maximizing strategy, while humans tend to try to match their predictions to the probabilities. Her example is a light that flashes green 75% of the time and red 25% of the time. Rats will settle on a behavior that predicts green all of the time. Humans, on the other hand, tend to predict green in a large proportion of the trains but also to predict red sometimes, since they know that red does appear as well. But this attempt to match probabilities makes human predictions less accurate; they get 25% of their green predictions wrong (just like the rats) but they also get 75% of their red predictions wrong (unlike the rats). So the rat's strategy results in a higher level of accuracy in prediction. Humans could get this same level of accuracy by allowing the directives of an authority (“Always choose green”) to preempt their own reasons for choice. Zagzebski concludes that “we are better off preempting” (Zagzebski Reference Zagzebski2012: 115).

This argument clearly applies to the case of action and prediction. But the fact that preemption in this case leads to the most beneficial pattern of behaviors does not automatically give us any reason to preempt with respect to belief. Even though the best outcomes follow from predicting green every time, it does not seem rational to believe the generalization, “All of the trials will result in green” or even to believe of each trial “This one will definitely be green.” Rather, it seems rational to have a high but limited degree of belief that each trial will result in a green light. Cases where one has a degree of belief also illustrate the earlier point that action and belief can diverge; accuracy in one neither ensures nor is required for accuracy in the other. This cleavage shows why the accuracy argument cannot extend to many cases of belief; even if preemption is required on the grounds that it makes action more accurate, similar considerations may not apply to belief.

Epistemic Superior believes P; I have no belief concerning P:

This is a case where no preemption is required. It also reflects the most clear cut cases where we are willing to believe on the basis of someone else's authority. We often seek out the opinions of an expert when we are lacking any expertise in the area in question. For example, one might ask their doctor which blood test is most accurate to diagnose hypothyroidism. Whichever test the doctor then recommends, it is reasonable to come to the belief that this is the most accurate test. But coming to that belief (which is formed on the basis of accepting the belief of one's recognized superior) doesn't make use of the preemption allowed to an authority. If one has no reasons relevant to the question at hand there are no reasons to be preempted. Even if one has some limited and weak reasons, preemption is still not required. One might remember hearing about the named blood test and remember that a neighbor had that test, either to good or poor effect. But this reason for belief is so weak that it is in no danger of leading one to reject the doctor's proposed test. So long as we take the beliefs of those experts we consult to provide us with strong reasons for belief, and so long as we have only weak reasons for belief (if any), there is no need for preemption. The accuracy of the expert's belief is directly reflected in our own beliefs, even without preemption.

This case also might be used to show why one need not preempt even in cases where one is in direct conflict with the recognized superior; so long as one is willing to give the belief of a recognized superior a high weight and one's own reasons a low weight, one's beliefs in cases of conflict will still generally follow those of one's superior. So long as one's beliefs follow those of the superior, one will have the same level of accuracy as that superior.

By working through the different cases to which we might apply the accuracy argument from Raz, we now see that the accuracy argument for political preemption will not generally require us to preempt our reasons for belief, except in a limited range of cases – those where one is in disagreement with a recognized epistemic superior. In the other cases, retaining one's personal reasons either gives the belief further support or provides one with reasons to retain a more nuanced degree of belief. Since preemption is often not called for, the preemption thesis, which is a general claim, is false.Footnote ⁷

3.1 Preemption applied to one's own beliefs leads to close-mindedness

Zagzebski's general argument for the possibility of our trust in other's beliefs starts with the claim that we should treat like cases alike; if we trust ourselves we should also trust others we take to be similarly situated (Zagzebski Reference Zagzebski2012: 55). If we apply this principle to the Preemption Thesis we can consider the case where, on the same evidential basis, we take ourselves to be the epistemic authority on a subject. For example, by a principle of parity, someone should be able to become their own authority by meeting the requirements of JAB 1. The self-applied version of this principle for determining authorities would look like this:

By a principle of parity, anyone who has been determined to meet the requirements for authority should likewise have their beliefs granted a preemptive authority.

But these two principles, taken together, seem to provide me with a good reason to be perfectly close-minded towards any reason that someone else might give me. This closed-mindedness reflects the strength of the original preemption thesis which recommends replacing all your original reasons relevant to believing p. It should be clear from the construction above that the closed-mindedness is a result of preemption. The assignment of epistemic superiority in Egocentric JAB 1 is simply a relative judgment about the epistemic position of oneself and of those one interacts with. Although the conditions that it describes are probably not often the case, when they are, it would be appropriate for one to judge that they are epistemically superior to those around them.

It is only when I take my epistemic superiority to license me as the epistemic authority, and so conclude that my belief preempts the reasons I can gain from others that I become closed-minded. Note that this is a case where I am closed-minded, not on a whim, but rather as a response to evidence about relative epistemic superiority; but even when I take this attitude in response to evidence, the attitude of being closed-minded still seems to be properly characterized as a vice.

While it is easier to see how being closed-minded to all reasons from others is an intellectual vice, those same considerations should equally apply to being closed-minded to one's own previous reasons and evidence. A self-directed vice is still a vice. It may often be more difficult to observe when we are treating ourselves viciously, since there is (often) no overt behavior for us to consider. But no virtue theory should drop the requirement that we also treat ourselves virtuously. We might also notice this requirement less often since we tend to want to treat ourselves generously and benevolently. Even though cowardice is often more tempting in situations calling for courage, Aristotle reminds us that its opposite, rashness, is also a vice.Footnote ⁸ Closed-mindedness to the self (even if not often tempting) is still an epistemic vice.

However, at this point the defender of preemption might object that dropping the preemption thesis would result in producing the vice of intellectual pride; we ought to instead display the virtue of intellectual humility. The proponent of preemption could then argue the preemption is the proper way to capture (at least some of) this virtue; the intellectually humble person must recognize others who are better epistemically situated as her epistemic authorities.

Such an objection would seem to bring the conversation to an impasse, with each side accusing the other of epistemic vice. Worse, if we accept the claims of both sides we seem to have generated an epistemic dilemma. If one follows the preemption thesis, one lacks open-mindedness towards oneself; if one refuses to preempt one's own reasons for believing, one lacks intellectual humility. How could such a dilemma be resolved?

We can resolve this apparent dilemma by noting that the virtues in seeming conflict do not both involve preemption. We have already noted that being closed-minded (to oneself or another) can follow from an application of the preemption thesis. The preemption thesis is a claim about how we should treat the reasons for belief possessed by ourselves and by others; it tells us which reasons we should take into account when we reason about a particular proposition. But intellectual humility is generally taken to be properly implemented, not at the level of reasons regarding a particular proposition, but rather at the level of generally evaluating ourselves and others and comparing our epistemic abilities. A willingness to recognize someone else's epistemic superiority (in the correct situations) is sufficient for intellectual humility. This focus on the level of general evaluation is a feature of many recent accounts of the virtue of humility. Nancy Snow argues that, “To be a humble person is to recognize your limitations, to take them seriously, and thereby to foster a realism in attitudes and behavior regarding self and others” (Snow Reference Snow1995: 210). It can also, “force us to realize that others are better than we in some respects, and thereby enable us to give due recognition to their talents and abilities, as well as to ask their help when we need it” (Snow Reference Snow1995: 211). Robert C. Roberts and W. J. Wood, in their recent exploration of the many intellectual virtues, suggest that intellectual humility “is an unusually low dispositional concern for the kind of self-importance that accrues to persons who are viewed by their intellectual communities as talented, accomplished and skilled” (Roberts and Woods Reference Roberts and Wood2007: 250). These accounts of humility, along with others offered in the recent literature, focus not on the particular beliefs or on the specific belief-forming practices of the humble person, but rather on their general attitudes towards their own abilities and either a lack of concern or a properly limited concern for those abilities and how they are perceived by others.Footnote ⁹ Thus the person who is willing to evaluate others as their epistemic superiors should be counted as having intellectual humility.

How one responds to the perceived superiority of others is an independent matter, and this is where the virtue of open-mindedness enters. We display intellectual humility when we follow the evidence and judge that someone else is our epistemic superior. We display open-mindedness to ourselves by not preempting our own reasons for belief in response to that judgment of superiority. The open-minded person thus remains open to her own reasons for belief, just as she would to reasons offered by others.

3.2 Why preemption is epistemically vicious

In the previous section I have argued that accepting the preemption thesis leads to the epistemic vice of closed-mindedness. Open-mindedness is plausibly a candidate epistemic virtue on any list of those virtues; closed-mindedness is a plausible candidate for any list of epistemic vices.Footnote ¹⁰ But such an argument against preemption need not depend on any particular canonical list of the epistemic virtues. There may be competing and overlapping lists of the epistemic virtues, and preemption might exhibit different epistemic vies with respect to different lists. The important issue is not which particular vice preemption exhibits, but rather why it should always be characterized as a vice. To explain this we need to look beyond lists and explore the nature of virtues and vices.

It is certainly beyond the scope of this paper to give a general account of (epistemic) virtue and vice; I believe we can show the problematic nature of preemption within either of two ancient accounts of virtue: the Aristotelian or the Stoic. These two theories are particularly relevant because Zagzebski offers a neo-Aristotelian virtue epistemology while I favor a neo-Stoic approach.

According to Aristotle the moral virtues (which Zagzebski uses as her model for neo-Aristotelian epistemic virtues) are each the mean between two extremes.Footnote ¹¹ As an original and simple argument we might note that preempting one's own beliefs does not appear to be a mean; it is an extreme position. It goes far further than epistemic universalism which gives some prima facie weight to anyone's belief. It goes further than a moderate position which recognizes the epistemic superiority of others by giving their position more weight than one's own. It asks us to entirely replace our own reasons for belief with the reason from the authoritative belief.

The extremity of this position can be further illustrated by noting that it makes a belief on authority, once adopted, difficult to update. Zagzebski does allow that we can update our beliefs away from the belief of the authority, but we only do this through finding evidence that the putative authority is not in fact an epistemic superior and so should not be treated as an authority (Zagzebski Reference Zagzebski2012: 116). The counter evidence here would take the form of an undercutting defeater; it would attach the connection between “S believes the P” and P by bringing S's epistemic standing into question. But note that while Zagzebski makes room for undercutting defeaters of P, she does not have a mechanism to allow consideration of rebutting defeaters for P. The evidence against P in a rebutting defeater would need to be handled as a direct reason for one to believe not-P. But if one is truly replacing all one's reasons relevant to believing P, then any rebutting defeater would be lost to preemption.

The lack of any mechanism for accommodating rebutting defeaters constitutes an extreme position; it certainly does not look like an Aristotelian mean. This extreme position would also be rejected as vicious on a neo-Stoic account. The Stoics famously object to all instances of emotion; this seemingly implausible claim is explained when one recognizes that the Stoics define emotions as mistaken judgments about the value of things. To have an emotion, say anger, is to think that one's pride is good; in the sense of being something of final value; such a judgment, in addition to being false, is dangerous. It will lead someone to seek revenge in a way that is unresponsive to reason. The problem with emotions (as the Stoics define them) is that they put one in a state where one is out of control, much like the state of running:

The objection to a particular kind of impulse (which we might more naturally characterize as a decision) is that, being unresponsive to reason, it is out of control. Such an impulse would certainly not be had by a sage, and it cannot be part of an overall good life. The way that a person is out of control when they are in the grips of epistemic preemption is similarly problematic. While treating the reasons of another as preemptive does not sound like an emotion,Footnote ¹³ they both share the problematic quality of being unresponsive to reason. Even if preemption only makes one unresponsive to rebutting reasons, this insensitivity to reason would still count as problematic on the Stoic view.

How does someone who is following Zagzebski's account of epistemic authority get themselves into a position in which they are out of control and unresponsive to reason? I think the problem lies both in the strength of the pre-emption thesis and in the formulation of JAB 1 and 2. As demonstrated above, the preemption thesis requires that we replace our own reasons with those provided by the belief of the epistemic authority, forcing us to become closed-minded to ourselves. This closed-mindedness leaves us unresponsive to a wide range of reasons; this lack of responsiveness to reason is the hallmark of a vice.

In addition to preemption, which makes us insensitive to all our own epistemic reasons, value monism makes us insensitive to a particular excluded value. Both JAB 1 and 2 motivate the acceptance of authority as a means to an end that looks laudable; for JAB 1 the end is truth and for JAB 2 the end is rationality. But in either case, a single-minded focus on a recognized epistemic value can lead us to a state where we are out of control; a state where we will sacrifice open-mindedness towards ourselves in order to obtain that single epistemic value.Footnote ¹⁴ Such excess is motivated by value monism but it might also be motivated by a pluralism with too narrow a focus – for example melding the values of JAB 1 and 2 together into a single standard for justifying the authority of belief. So long as there are epistemic values that are excluded from consideration, we will be unresponsive to the reasons those values provide; again, this lack of responsiveness to reason is a sign of a vice.

It is interesting to notice that Zagzebski has argued against value monism about the value of truth, as part of her own diagnoses of the limitations of reliabilism (Zagzebski Reference Zagzebski and Axtell2000). The problems with value monism (both about truth and about rationality) suggest that a complete virtue epistemology must be grounded in all our epistemic values and capture the way that our search for them is balanced in our development of a virtuous character. Such an approach is exemplarized in Zagzebski's earlier (Reference Zagzebski1996) treatment of virtue epistemology. When we focus on developing our total epistemic character, this should lead us to deny that any epistemic value should be completely or systematically sacrificed for any other; rather we need to consider how they are to be balanced within an overall epistemic good life – the epistemic eudemon life.Footnote ¹⁵