Doxastic Deontology and Cognitive Competence

Abstract

The paper challenges William Alston’s argument against doxastic deontology, the view that we have epistemic duties concerning our beliefs. The core of the argument is that doxastic deontology requires voluntary control over our beliefs, which we do not have. The idea that doxastic deontology requires voluntary control is supposed to follow from the principle that ought implies can (OIC). The paper argues that this is wrong: in the OIC principle which regulates our doxastic duties the “can” does not stand for the ability to shape our beliefs voluntarily. As an examination of everyday examples shows, it stands for cognitive competence, the reliable ability to acquire beliefs in compliance with the epistemic norms. The doxastic OIC principle asserts, in brief, that one is only obliged to believe something if one’s cognitive capacities are sufficiently strong. It is also explained why the doxastic duties do not require voluntary control as opposed to moral duties. This understanding of doxastic duties saves our everyday doxastic deontic judgments from Alston’s argument, but does not help the deontological conception of justification, which understands justification as not violating one’s epistemic duties. It actually provides another argument against the deontological conception: if the OIC regulating our doxastic duties is construed as suggested, the deontological conception of justification implies that one’s doxastic duties and, consequently, whether one’s belief is justified depend on one’s cognitive competence. Since cognitive competence varies from person to person, justification will not matter to truth and knowledge in the way epistemic justification is supposed to do.

1 Introduction

Prior to philosophical reflection we tend to take it for granted that we are responsible for our beliefs in roughly the same way as we are responsible for our actions. Just as we have moral duties prescribing or forbidding certain types of actions in various situations, we also have epistemic duties prescribing what we should or should not believe under various conditions. Moreover, just as we can be blamed for failing to fulfill our moral duties and praised for fulfilling them, we can be blamed and praised for our beliefs. Doxastic deontology is the view that this analogy is right: beliefs are subject to a kind of deontic evaluation which is very similar to the deontic evaluation of actions, so there are true doxastic deontic statements.

Doxastic deontology has received much attention lately and an important thread in the discussion was initiated by William Alston (1985, 1988), who pointed out that actions and beliefs are quite unlike in an important respect: beliefs are not under voluntary control. Whereas a basic action, like raising one’s arm, can be carried out simply by willing it, the will is completely inefficacious in bringing about and sustaining belief. Wanting to believe a proposition is never sufficient for believing it, and one may believe a proposition even if one does not want to believe it in the least. Alston further contended that deontology requires voluntary control: it cannot be your duty to φ and you cannot be blamed for not φ-ing if wanting to φ has no effect on whether you φ or do not. These two claims together entail that doxastic deontology is unfounded.

Alston’s attack on doxastic deontology was not aimed primarily at undermining the common doxastic deontic practice involving things like condemning someone who does not believe an eyewitness who is known to be trustworthy, chiding someone for failing to draw an obvious conclusion from propositions he accepts, or teaching our children that they should not believe everything they hear. He was after a bigger game, the deontological conception of justification, which tacitly informed several normative epistemological theories especially up until the advent of externalism. On this view the concept of epistemic justification is to be spelled out in terms of epistemic duties; one’s belief that p is supposed to be justified if and only if believing that p does not involve violation of one’s epistemic duties; more simply, justified belief is blameless belief.¹

In this paper I am going to argue for two claims. First, common doxastic deontic practice does not work the way Alston and some of his critics suppose. According to Alston, the limits of our duties are set by voluntary control, i.e. our duties do not extend beyond what is under voluntary control. In contrast, I contend that the limits of our epistemic duties are determined by our cognitive competence, i.e. if believing that p requires abilities and skills you do not possess, it is not your duty to believe that p and you cannot blamed for not believing it. The issues whether belief is voluntary and if it is, in what sense, are interesting and important, but the fate of doxastic deontology does not hinge on how they are resolved. If this is right, Alston’s attack on doxastic deontology does not succeed.

Second, this defense of doxastic deontology leaves little hope for the deontological conception of justification. The reason is that cognitive competence, as I will construe it, varies from person to person. If epistemic duties are regulated by cognitive competence, and epistemic justification is analyzed in terms of epistemic duties, then epistemic justification gets tied to personal qualities. The resulting conception of epistemic justification is thus person-dependent, which does not only yield highly counter-intuitive judgments in certain situations, but also renders the concept of justification incapable of doing the sort of work the concept is supposed to do in epistemology.

In the following section I will explain doxastic deontology more precisely and lay out Alston’s argument (Sect. 2). Then I will examine two cases and variations on them in order to find out how doxastic deontic practice works (Sect. 3). Relying on this examination, I will argue that epistemic duties are regulated by cognitive competence and that Alston’s attack is, therefore, unsuccessful (Sect. 4). Finally, I will inject the suggested understanding of doxastic deontology into the deontological conception of justification and conclude that the conception of justification we get is untenable (Sect. 5).

2 Alston’s Attack on Doxastic Deontology

The concept of deontology is best explained by contrasting deontic claims with evaluative claims (cf. Chuard and Southwood 2009, p. 601, Peels 2014, pp. 4–6). Evaluative claims are concerned with assessment and appraisal and express how well the item evaluated fares with respect to the appropriate norm, rule or standard of appraisal. Any subject can be evaluated provided that there exists an appropriate norm; thus “The weather is nice” or “This is a shameless lie” are evaluative claims. Deontic claims, on the other hand, are prescriptive or regulative: they tell us how to proceed; thus “You should keep your premises” or “You mustn’t lie” are deontic claims. The benchmark they appeal to is not just a norm but a duty, an obligation or a requirement. Since only reasonable agents can be subjects to duties, obligations and requirements—you cannot demand of the weather to be nice—, deontic claims can only be made about such agents.² These agents bear responsibility for what they do and are, therefore, subjects to the kinds of emotional responses Strawson (2008) called “reactive attitudes”, such as praise and blame, which can only be occasioned by the activities of responsible agents. This limitation of the sphere of application of deontic claims is one reason why there is not always mutual implication between an evaluative and a corresponding deontic claim; the deontic claim always implies the corresponding evaluative claim—“You should keep your promises” implies “Keeping your promises is a good thing”—but not the other way around. Another reason is that not all actions evaluated as good can be prescribed as a duty; donating 80% of one’s income to charity is certainly noble, but it does not follow that one ought to do that. Demanding the latter would violate a generally though not universally endorsed constraint on duties, the principle that ought implies can (hereafter OIC). Someone with no wealth and a subsistence level income cannot give up 80% of his salary, so this is not something he ought to do.

Deontic claims may also be characterized in terms of the vocabulary in which they are couched. In addition to the terms for the benchmarks—like “duty”, “obligation”, “prohibition”, “permission”—, there are adjectives expressing how the thing judged stands with vis-a-vis these benchmarks—“obligatory”, “forbidden”, “permissible”. The deontic status can also be expressed by modals—like “ought to”, “should not” and “can”— and terms expressing what reactive attitude is appropriate—“blameworthy, “praiseworthy”, “blameless”. Even though this vocabulary as a whole gives a good idea of what the deontological perspective is like, most of these terms have non-deontological uses as well, so we do not have simple syntactic criteria for the identification of deontic claims.³ For instance, if someone says that “Teachers ought to be better paid”, this can be construed both as an evaluative claim (the present situation is unjust and/or undesirable) and as a deontic claim (whoever is responsible for the teachers’ salaries has the duty to raise them). We cannot, therefore, determine if a statement is deontic simply by looking at the terms employed; we must also consider the context and the intentions of the speaker.

Doxastic deontic claims in the broadest sense are deontic claims about doxastic attitudes, i.e. belief, disbelief and suspension of belief. (For ease of expression I will be talking mainly about belief, but what I say is intended to apply to the other two doxastic attitudes as well.) I will, however, use the term in a narrower sense. First, one may think that there are moral and prudential obligations concerning belief, like “You should not believe that your mom has committed a heinous crime” and “You ought to believe that you will overcome the disease”, but I am going to reserve the term for claims which refer to epistemic obligations, meeting which is likely to get us closer to achieving some worthwhile epistemic objective like truth, knowledge or justified belief.

Second, we certainly have epistemic obligations to create favorable conditions for belief formation: we should gather evidence, hone our cognitive skills, educate ourselves in the relevant field, engage in serious reflection, etc. I will not call these doxastic obligations either, because they are concerned with what we should do to improve the chance of arriving at justified or true beliefs rather than what we should believe. So blaming someone who failed to obtain some highly important and easily available piece of evidence, but formed his belief in accordance with the evidence he actually possessed will not be taken here as a doxastic deontic judgment.

Alston’s attack on doxastic deontology is directed against the existence of doxastic obligations. If there are no doxastic obligations, all judgments declaring a belief or some class of beliefs obligatory or forbidden will be false, and all judgments declaring a belief or some class of beliefs permissible will be vacuously true. You can believe anything, you will never violate your doxastic duties, for there is none. The argument runs as follows.

• (1) If there are doxastic obligations, we have voluntary control over our beliefs.

• (2) We do not have voluntary control over our beliefs.

• (3) Therefore: there are not any doxastic obligations.

Alston seeks to derive premise (1) from OIC, which he puts as “one can only be obligated to do A if one has an effective choice whether to do A” (1988, p. 259). This somewhat unusual formulation is probably motivated by the consideration that moral and prudential obligations apply to intentional actions, which are by definition voluntary. Premise (2) adds that when it comes to beliefs, we lack effective choice: our wanting to have or not to have a particular belief does not result in our having it or not having it. He takes it as a brute psychological fact that the will does not act directly on belief,⁴ and that we only have “indirect voluntary influence” on belief, meaning that we can, by performing intentional actions like gathering evidence and developing our cognitive skills, change our situation in a way that influences what beliefs we come to have.⁵

The argument has been criticized from various directions, the most important of which are perhaps the following. First, (1) is unfounded, because OIC does not extend to doxastic obligations (Feldman 2001, 2008; Chrisman 2008). Second, (1) is unfounded, because the “can” of OIC does not imply voluntary control (Chuard and Southwood 2009). Third, (2) is false, because there are cases in which we do exercise the sort of voluntary control which Alston believes we never have (Frankish 2007; Ginet 2001; Weatherson 2008). Fourth, doxastic deontology should be spelled out in terms of acceptance rather than belief, and then the argument fails, because acceptance satisfies the demand for voluntary control (Bondy 2015). The critics of the argument called doxastic compatibilists, who draw on ideas originally developed by compatibilists in the free will debate, do not form a separate group.⁶

Some of them reject (1) saying that the sort of control doxastic obligations presuppose is not voluntary control (McHugh 2013, 2017; Owens 2000)—they belong to the second group. Others deny (2) and maintain that we do exercise the same sort of voluntary control over our beliefs as we exercise over our actions (Heller 2000; Russell 2001; Ryan 2003; Steup 2000, 2008, 2012, 2017)—they belong to the third group.⁷

The objection to be raised below resembles the second line of criticism in denying that OIC lends support to (1), but instead of directly showing that when OIC is applied to doxastic obligations the “can” does not imply voluntary control, I will propose an alternative construal. I will first motivate this construal by examples (Sect. 3) then explain it and argue for its superiority (Sect. 4).

3 Two Stories and Variations on Them

I will begin by telling two stories the characters in which end up passing what seem to be true doxastic deontic judgments on their prior beliefs.⁸ The stories are not intended as counterexamples directly refuting Alston’s argument; rather, they serve to bring out certain features of our doxastic deontic practice, i.e. the way we normally make doxastic deontic judgments.

Apartment

Jack and Jill are a highly intelligent middle-aged couple who are fairly experienced in practical decision-making concerning apartments. They both bought and sold, leased and leased out apartments, they know very well how to weigh financial considerations including long-term market value and convenience. Jack does not own an apartment, and the layout of Jill’s apartment is not ideal for the two of them, so when they moved in together they decided to lease out Jill’s apartment and rent a more commodious one in the same neighborhood; this way they only have to pay the difference between the two leases, which they can very easily afford. The only problem is that Jill’s mother lives in a small town several hours’ drive away, and visiting her takes up a complete weekend. Loving daughter as she is, Jill would like to see her mother more often. After carefully reviewing all possibilities, considering the pros and cons and doing the necessary calculations, they come to agree that the best option is to terminate the lease of Jill’s apartment and move Jill’s mother there. They will not get paid rent, but they can still afford paying for the lease. However, they know that rents soared in the last couple of months, and all predictions say that they will keep on rising for a while. If the rent they pay is raised to the market level, they will not be able to afford it or to find an equally convenient apartment for an affordable price. According to their lease agreement, the rent can be raised only once a year, so they ask their landlord about his intentions. He says he is going to adjust the rent only to inflation, not to the market level. They have every reason to believe him, because he has always been honest and reliable, he hates the hassle that comes with finding a new tenant, and also appreciates that Jack and Jill are ideal tenants, who always pay on time, take good care of the apartment etc. So they decide to go ahead. They talk to Jill’s mother and are just about to give notice to Jill’s tenant, when the landlord calls saying that he has changed his mind and that he will want a 15% raise (which is still below the market price). As soon as he tells that, Jack and Jill realize what a mistake they have made. Of course, they were wrong about the landlord, but the mistake they realize now is that they failed to take into account all relevant possibilities. They only took into consideration one scenario in which they would have to leave the apartment they lease, namely that landlord gets tempted by the rising rents and wants to make more money off the apartment. But there are other scenarios as well. The landlord’s family conditions might change so that he might decide to move into the apartment himself. Or he might get into financial trouble which forces him to sell the apartment. They rightly blame themselves: they ought to have seen that their plan carries risks which are way too high; they should not have believed that it is safe.

Theorem

Emily is an outstanding mathematician, one of the greatest authorities in her field, whose lifelong ambition has been to prove the Grand Theorem, fulfilling which would write her name in the history of mathematics. Recently there has been significant progress in the field, partly due to Emily’s work, and mathematicians are hoping that the proof is already within reach. After carefully reviewing the various ways which might lead to the proof, Emily opts for proving Lemma X, then proceeding from there. She is nearly finished with the proof of Lemma X, when one day she has a discussion with one of her PhD students working on a different topic, and she recommends that he use Theorem Y to solve one of his problems. That theorem was proven by Emily herself in her doctoral dissertation, and it turned out to be highly important in a neighboring branch of mathematics, but as a road towards the Grand Theorem it was soon judged as a dead end, and Emily had not thought about it for a long time until it came up in the discussion with her student. However, the moment she mentions Theorem Y, it dawns own her that together with some fresh results it offers a straightforward route to the Grand Theorem, which is superior to the one via Lemma X. When the student leaves, she rushes to her computer to start working on her new idea and finds an e-mail from a journal asking to referee a paper which promises to prove the Grand Theorem from Theorem Y. She feels like banging her head against the wall. The idea is so obvious that she cannot understand why she has not thought of it earlier. She should not have believed that Lemma X offers the best route. She ought to have seen that Theorem Y is the key.

Putting aside Alston’s doubts concerning doxastic deontology for a while, I submit that the “should not” and “ought to” statements at the end of the stories are true doxastic deontic claims.

Let us see first why they are deontic rather than merely evaluative. The characters in the story violate the epistemic rule that all relevant possibilities must be taken into consideration:⁹ Jack and Jill forget about certain possibilities in which they would have to leave the apartment they lease, and Emily fails to take into account Theorem Y as a possible starting point. If they simply noted that they had violated the rule, their judgments would be evaluative. But they do not just note that: they are upset, and what they are upset about is not just that their belief turned out be erroneous but that they committed an error. They take responsibility for their errors and blame themselves for believing what they did. Blame and responsibility are marks of a deontic judgment.

Why are then these claims doxastic? Doxastic judgments, as I use the term, concern beliefs themselves and not the actions or omissions influencing the conditions under which beliefs are formed. To use Alston’s expression, they are not about the proper or improper way of exercising “indirect voluntary influence” on belief. There are two sorts of cases in which neglecting a relevant possibility is a failure of action rather than of belief. First, there are situations in which the relevant possibilities are already given and one only has to decide which ones to consider. Suppose your computer is acting up, and you look up a couple of forums discussing the problem. One of the posts is rather long and technical, so you decide to ignore it and, as a result, come to a false belief as to which solutions are worth trying. In this situation your conscious decision puts you in a position which is unsuitable for arriving at the truth, so you fail to exercise the right sort of voluntary influence on your belief. However, this case is very different from what happens in the stories. Jack and Jill do not consider the possibilities that the landlord may decide to sell the apartment or move in himself, and then consciously decide to ignore them; rather, these possibilities do not cross their minds. Likewise, Emily does not contemplate Theorem Y then decide to drop it. In fact, as soon as it comes to her mind, she embraces it enthusiastically.

The other sort of case is the one in which one does not explore the possibilities in sufficient depth. Your computer is acting up once again, but this time look up only one forum and consider only the solutions described there, and arrive at a false belief. You do not consciously ignore any possibilities, but do not search for relevant possibilities tenaciously enough. The decision which puts you in the wrong position to find the truth concerns how extensive search you should conduct. But this case is not similar to the stories either. The extensiveness of the search can be understood in terms of time and in terms of the number of the items to be reviewed. The characters in the stories think on the issue at hand long and hard and thus cannot be faulted for failing to devote sufficient time to the search. Did they then fail to review sufficient number of possibilities? No, because this sense of extensiveness is inapplicable to what they do. In the computer case the possible solutions are readily available, the browser has already found them, and you decide at a point that you have reviewed a sufficient number of them. Jack and Jill and Emily do not have the possibilities available in this way: they have to think them up themselves. They concentrate on the problem and trust that all the relevant possibilities will eventually occur to them. The emergence of possibilities is not a voluntary action. The only voluntary action through which they may influence the outcome of their thinking process concerns how long they should keep thinking about the problem, and they keep thinking about it long enough. Hence they can only blame themselves for their beliefs.

The doxastic character of the claims at the end of the stories is further supported by analogous cases cited in discussions of the epistemic condition of moral responsibility. Even though ignorance excuses from moral responsibility, culpable ignorance does not: if you do something wrong, because you did not know something you should have known, you are still responsible. Since the idea that culpable ignorance is always rooted in a prior blameworthy act or omission threatens with regress (Zimmerman 1997, also explained in Wieland 2017), some have argued that one may be directly blamed for one’s doxastic state. One example is that of a mother who leaves her dog in the hot car thinking that it will take only a couple of minutes to pick up her children, but ends up leaving the dog in the car for hours, because there are problems at the school to be sorted out. (Sher 2009, p. 24) In another example a professor driving home forgets about his wife’s request to buy milk, because he is engrossed in thinking about the paper he is writing. (Clarke 2014, p. 164) The deontic judgments that the mother should have known that the dog was left in the hot car and that the professor should have known that he had to buy milk are taken to apply here to doxastic states rather than to prior actions.¹⁰

If the claims at the end of the stories are both doxastic and deontic, the final issue is whether they are true. One may dispute that by saying that the characters’ self-reproach may be inappropriate. It is a psychological fact that we are sometimes prone to ill-founded self-reproach when what we did has grievous consequences. If a child runs in front of a car within braking distance and gets killed, the driver will most likely reproach himself even if he kept the brakes in good condition and hit them as soon as he got sight of the child. Jack and Jill’s failure nearly landed them in big trouble, and Emily’s failure has thwarted her lifelong ambition, so it is understandable that they blame themselves. However, so the objection goes, their self-reproach is as inappropriate as that of the blameless driver, for they did everything they could to take all relevant possibilities in consideration. Since OIC instructs us not to demand more than what one can do, it was not their duty to take all relevant possibilities into consideration.

I grant that self-reproach may be misplaced, and I also grant that the characters did everything they could in so far as “doing everything one can” refers to the voluntary actions. In fact, I have just argued that they cannot be faulted for not exercising proper indirect voluntary influence on their beliefs. What I take issue with is that the characters could not have arrived at the correct belief. There are two details which are important in this respect. First, our characters are highly adept at the sort of task they are facing. Jack and Jill are smart and have a good deal of experience in making such decisions. Emily is an outstanding mathematician, which involves the capability to pick the right starting points for proofs. They are the sort of people we would expect not to make such errors. Second, they realize their errors right away. As soon as their landlord tells Jack and Jill that he is going to raise the rent, they understand that their plan could have failed in other ways as well, so they come to grasp the possibilities they ignored even though these possibilities are not actualized. As for Emily, as soon as Theorem Y comes to her mind, she recognizes at once that it is an excellent starting point. Indeed, she finds the idea so obvious that she cannot understand why she has not thought of it earlier. This shows that the possibility which escaped their attention was easily accessible to them; in other words, they have the ability to notice what they did not.

What we witness here is a simple error, which is anything but unusual. Have you never made a mistake in addition? Have you never put down an ungrammatical sentence? Have you never made a mistake in grading undergraduate papers? People occasionally fail at tasks which they normally handle without difficulty. Of course, such errors typically happen when one is tired or distracted, but they also occur even if one has no such excuse. Occasional performance errors do not raise doubts about one’s abilities. Jack and Jill’s and Emily’s tasks are more complex than the ones I mentioned, yet they are the sort of tasks which they are very well equipped to handle.

Let us dig deeper. I suggested that the reason why we judge the characters capable of noticing the possibilities they did not happen to notice is that these possibilities are easily accessible to them. Let us now have a look at some variations in which this element of easy accessibility is missing.

Apartment: the young couple

Replace Jack and Jill with a young couple who are neither particularly smart nor have any experience in making such decisions. They are appalled by the landlord’s change of mind but do not realize that they ignored the risk that the landlord might have to terminate their lease for some other reason.

We certainly blame the young couple for not seeking advice from someone more experienced, but we do not blame them for the erroneous assessment of the risks involved. What we blame them for is their failure to create conditions favorable to true belief, i.e. failure to exercise proper indirect voluntary influence on their beliefs, which is not doxastic blame in the narrow sense in which I use the term. We all know well that people with no experience in such matters are liable to make mistakes, and this is exactly why they should have consulted with someone more astute.

Theorem allows more variations.

Theorem: the mediocre mathematician

Instead of Emily take a second rate mathematician lacking the kind of intuition by which brilliant mathematicians like Emily are able to spot the possibility for a significant novel result. When talking to his PhD student, it does not occur to him that Theorem Y is the best starting point. He finds the idea of a proof from Theorem Y surprising and understands it only after reading the paper very carefully.

Theorem: the stroke of genius

Let us keep Emily, but imagine that the proof is truly brilliant. It is based on the sort of insight even the best mathematicians rarely come by. After mentioning Theorem Y to her doctoral student Emily still does not think that it might be the right starting-point, and resumes her works on Lemma X. Looking at the abstract of the paper, she first thinks that the proof is surely fallacious and the paper must be rejected, but when she reads it, she understands the idea quickly and cannot but admire the author’s ingenuity.

In these scenarios the idea that Theorem Y is a possible starting point becomes less available for different reasons: in the mediocre mathematician scenario the protagonist is made less capable (like in the young couple variation of Apartment), in the stroke of genius scenario it is the idea which is made extraordinary. Yet in both cases we find that we are no longer willing to endorse the same doxastic deontic judgment we regarded as true at the end of the original story. Should the mediocre mathematician have believed that Theorem Y is the best point to start from? There is a sense in which he should have, since Theorem Y is indeed the best starting point, but this is the evaluative rather than the deontic sense of “should”: requiring the mediocre mathematician to believe the truth would be saddling him with a duty he cannot fulfill. We certainly require more from the brilliant Emily. However, being brilliant means that one regularly comes up with ideas which the less capable hit upon rarely and only in their best moments, and that in her best moments one comes up with quite extraordinary ideas which only the most capable can conceive and only in their best moments; it does not mean that one can always accomplish what the most capable can only accomplish in their best moments. Emily is brilliant, still, we do not demand of her a stroke of genius every time a stroke of genius is called for.

What do these examples teach us about the way we make doxastic deontic judgments? We have seen that our willingness to blame a subject for a failure depends on how we evaluate his abilities, skills and experience, his competence, for short. Since blaming is appropriate exactly in those cases in which someone fails to fulfill one’s duty, it follows that subjects’ duties depend on their competence. Generalizing from the examples yields the idea that it depends on the subject’s competence whether he has a duty to form a certain doxastic attitude regarding a particular proposition. More competent subjects have duties which the less competent do not have.¹¹

However, in order to draw this conclusion, the argument needs to be tightened. The argument applies the method of difference and consists of the following steps.

• (4) In the original stories the characters are doxastically blameworthy.

• (5) In the variations the characters are not doxastically blameworthy.

• (6) The only relevant difference between the original stories and the variations is that it is only in the original stories that the characters possess the required level of competence.

• (7) Therefore: possession of the required level of competence is a necessary condition of doxastic blameworthiness.

I have only worked on (4) and (5), so one may still wonder if (6) is true. Surely, there are further differences between the two sets of cases; for instance, in Apartment Jack and Jill are older than the young couple in the variation. We also find further differences if we fill out the details in a realistic fashion; for example, Emily is likely to work at a more prestigious department and to be better paid than her mediocre counterpart. I do not think such differences pose a threat. Some of them are irrelevant, because they have no effect on our willingness to blame. For instance, the fact that Jack and Jill are older matters only because if they were not, it would not be realistic that they are experienced in making such decisions. But age itself is irrelevant: we would not blame someone who is old but has no relevant experience, whereas we would blame a young real estate agent. Emily’s position and salary, on the other hand, do not seem irrelevant, since we certainly expect more from a well-paid professor at a prestigious department. However, these and similar explanatorily relevant differences are not independent variables, because they owe their explanatory power to the fact that they are indicative of competence. In real life we rarely have direct knowledge of one’s level of competence and can only infer to it from fallible signs like experience, track record, and recognition.¹² Position and salary belong to these signs, and they only influence our willingness to blame in virtue of their correlation with competence. If the correlation were suspended—suppose hiring at universities were highly dysfunctional, and it were based on family connections and social skills, not on academic excellence—we would not be more willing to blame those who have a more prestigious position and a higher salary; on the other hand, suspending the correlation would not affect our willingness to blame the competent but not the incompetent. Putting the point in terms used in probabilistic accounts of explanation, competence screens off academic position and salary as a predictor of willingness to blame.

Granting that the stories and their variations are unassailable, one may still wonder if (6) can be generalized. Cannot one construct contrasting cases like the original stories and the variations in which our willingness to blame does not depend on competence? I can think only of one kind of example which may appear to be of this type.

Sentence

A judge, who is very experienced and highly respected, is to decide about a case concerning a particularly heinous crime: during a bank robbery all the people in the bank were ruthlessly murdered. The prosecution charges that the defendant participated in the robbery and was standing in the door to turn away customers who wanted enter the branch, while his unidentified accomplice committed the murders. The judge carefully reviews the evidence and decides that the testimony of the professional criminal who named the defendant in exchange for mitigation of punishment and the reward put up by the police is reliable, and so are the testimonies of the customers who identified the defendant as the person who turned them away at the entrance. On the other hand, he discards the defendant’s persistent denial of participation even though the defendant also confesses to crimes the police was not aware of and which carry severe punishment. He also discards the evidence that the defendant’s mobile phone was in use at the time of the robbery in another location. As a result, he finds the defendant guilty and sentences him for life.

Couple of years later the murder weapon is found in possession of another criminal, and it turns out that the defendant had nothing to do with the crime. The judge, now as much criticized as he was celebrated at the time of the trial, resigns from a prestigious national board of lawyers.

It seems that the judge can be rightfully blamed for believing the defendant guilty beyond a reasonable doubt in virtue of misjudging the reliability and the relevance of the evidence presented. Of course, much depends on further details, but let us assume that he is indeed blameworthy. What is interesting is what happens later.

Sentence: the revelation

Interested in how such a miscarriage of justice could occur, an investigative journalist scrutinizes the judge’s career and finds that his fame and recognition is unfounded. It is due to a small number of cases which attracted much public attention, but his record is not impressive at all: his judgments are repeatedly overturned by higher courts, far more often than those of other judges. The journalist concludes that the judge was incompetent.

Would we still blame the judge for his belief that the defendant was guilty beyond a reasonable doubt? We probably would, and this seems to create a problem for my argument. The original story fits the pattern: the judge is competent and is blameworthy. In the variation (i.e. the revelation) it comes to light that the fallible signs of competence were, in fact, misleading: the judge was incompetent. If competence is indeed a necessary condition of doxastic blameworthiness, we should revise our doxastic deontic judgment, but we do not.

I do not think this example shows that (6) cannot be generalized because the blame which persists after the judge’s incompetence is revealed is not the sort of blame we are interested in. Let me start with the observation that even though we still blame the judge after we learned about his incompetence, we blame him less. We feel that the responsibility for what happened should be shared between him and those whose decisions eventually put him into a position in which he faced a task he could not cope with. We do not regard him as blameworthy as he would be if he were competent.

But why do we still blame him if he is incompetent? What distinguishes the incompetent judge from the young couple in Apartment or the mediocre mathematician in Theorem? It is the fact that the judge has a duty which does not have an analogue in the earlier cases. It is the duty of a judge to acquit anyone whose guilt has not been established beyond a reasonable doubt. This is a duty all judges share no matter what their level of competence is; in contrast, the characters in Apartment do not have a duty to make the right decision and the characters in Theorem do not have a duty to prove the Grand Theorem. The reason we find the judge’s belief blameworthy is because it lead to a blameworthy decision. This kind of blameworthiness is derivative: it is not based on the fact that the judge violated the epistemic rules relevant in the situation but on the fact that he did not discharge his duty as a judge. Now recall that I construed doxastic deontic judgments as concerned with fulfilling one’s epistemic duty. The duty the judge has with respect to his belief, which is derivative and is independent of his competence, is not an epistemic duty, but a by-product of his professional duties. So blaming him for his belief is not a doxastic deontic judgment in the sense in which the term is used here. If this is right, Sentence and its sequel are ineffective against (6), which is concerned with differences relevant to doxastic deontic judgment.

That this is indeed right can be supported by two considerations. The first one consists in seeing what happens to our willingness to blame if the obligation to acquit those whose guilt was not proven beyond a reasonable doubt is cancelled. Let us suppose, therefore, that a retired judge also attends the proceedings out of pure interest. Since he can neither convict nor acquit, he does not have the duty the judge trying the case has. Suppose further that the case unfolds as in Sentence and its sequel. Initially we think that the retired judge is competent and we are inclined to blame him his belief that defendant is guilty. However, if we learn that he is incompetent (e.g. we notice signs that he is going senile), we revise our judgment of blameworthiness. Our judgment follows the same pattern as in the original stories and their variations.

The second consideration relies on analogous non-epistemic cases. For instance, suppose that someone takes out a loan from a bank, but then a series of most unfortunate things happen: a severe economic decline sets in, she and her spouse both lose their jobs, their savings and real estate lose much of their value, his husband falls sick, and she receives huge medical bills. As a result, she cannot pay the instalments. Does she have a duty to pay? Yes and no. Yes, because by taking out a loan she accepted a contractual obligation which is not dependent upon her financial situation. No, because she cannot do so, and that is not her fault. Lots of similar cases can be constructed, and they suggest there are two kinds of duties. There are duties which derive from contractual obligations, from the office one holds, from the job one does, generally speaking, from social arrangements in which one voluntarily participates. Such duties do not fall under OIC: as long as one is a judge, a mayor, a teacher etc. there are things one must do even if they are beyond one’s talent.¹³ On the other hand, there are duties independent of social arrangements, which apply to anyone, not only to judges, to mayors, to teachers etc. These latter duties are subject to OIC, and inability to fulfill them warrants exemption. The reason why these non-epistemic cases are relevant is that they show that the distinction drawn between the epistemic and the non-epistemic duties of the judge fits a general pattern, so it is not arbitrary or ad hoc.

It is easy to see how the idea that competence is a necessary condition of doxastic duties can be brought to bear on Alston’s argument: our doxastic duties are indeed subject to an OIC principle, but the “can” of the principle signifies competence and does not imply voluntary control. Before developing this point in more detail, it is worth noting that it is quite common to evaluate someone’s performance in a cognitive task on the basis of a benchmark adjusted to his level of competence. Suppose that two high school students, a weak one and a top one, do equally well on a math test; both solve all the problems of the kind they practiced in class but fail at a problem demanding some degree of ingenuity. In spite of giving them the same grade, the teacher would praise the weak student but would find the top student’s performance somewhat disappointing. Or take a soccer player who has to choose the right moment to pass the ball to a forward moving fast up the field to an excellent position, but he waits too long and passes only when the forward is already in offside position. If he is a great football player, an agitated commentator supporting the team would pour scorn him; if he is an inexperienced player, the commentator would remark that he was not up to the task and blame the coach for selecting him to play. Or if we consider a very rare disease, we do not demand of a primary care physician to diagnose it; we only demand that he should recognize that the case lies behind his competence and refer the patient to a specialist. The specialist, on the other hand, is required to make the correct diagnosis.

4 The Doxastic OIC

The idea is then that doxastic duties, similarly to moral duties, are subject to an OIC principle, but the “can” of the principle that applies to doxastic attitudes—in contrast with the “can” of the OIC principle that applies to actions—does not imply voluntary control.¹⁴ Consequently, Alston’s argument for premise (1), saying that the existence of doxastic obligations would imply voluntary control over belief, fails. In this section I will explain the doxastic OIC principle in more detail and complete the argument against Alston.

The “can” of the ethical OIC principle is a “can” of ability rather than of possibility. If you are put in William Tell’s situation, it is not your duty to shoot the apple off your child’s head, because you cannot do that in the sense of ability, even though you can do that in the sense of possibility, because you may get lucky. The “can” of the OIC principle regulating our doxastic duties is also a “can” of ability. I will call the ability this “can” stands for cognitive competence and now I need to explain it in some detail.

Competence in general stands for a reliable disposition to perform well.¹⁵ Each disposition is associated with a triggering condition and a manifestation. In the case of competence the triggering condition is a situation which calls for accomplishing some task and the manifestation consists in accomplishing it. To be competent one does not have to succeed every time. First, because good performance depends not only on the possession of competence but on other conditions as well. A competent archer may not manifest his competence by hitting the target if his internal condition or the external circumstances are unfavorable, e.g. if he is drunk or exhausted, or it is dark or the wind is too strong. Second, because the disposition does not have to be perfectly reliable. A competent archer may occasionally miss the target even if he is in excellent shape and the external conditions are ideal. Of course, if a competent agent fails to manifest his competence, our first reaction is to look for adverse conditions, but sometimes we will not find any.

In the case of what I call cognitive competence good performance consists in forming and maintaining doxastic attitudes—believing, disbelieving or suspending belief—in keeping with the relevant epistemic norms.¹⁶ Cognitive competence is not single and undivided but a bundle of different competencies. One may be cognitively competent to assess the risks associated with decisions concerning apartments without being cognitively competent to assess the potential of different routes to proving a mathematical theorem. So the application of the term always involves tacit reference to the sort of cognitive task the agent is facing. The term as I understand it behaves more like “smart enough” than “smart”: the former invites questions like “to figure out what”, the latter does not.¹⁷

As is the case with competence in general, being cognitively competent (in a certain area) does not imply that one never fails. The reasons are the same: a reliable disposition to succeed implies only a high success rate under favorable conditions rather than a perfect score under all conditions.

The information we possess—what we already believe and what we are aware of by our senses—has a crucial role in shaping our doxastic attitudes, so one may inquire whether it should be counted among the conditions affecting the manifestation of cognitive competence. The answer is no: cognitive competence is manifested in forming and sustaining doxastic attitudes in accordance with the relevant epistemic norms, and one may respect the epistemic norms even if the information one has is insufficient or incorrect. Low quality information is unfavorable for finding truth or obtaining knowledge but not for following the epistemic rules.¹⁸

Relying on these ideas, we may now turn to the OIC principle which applies to doxastic attitudes. One may formulate the principle abstractly by saying that one has a duty to respect the epistemic norms in so far as one is cognitively competent to do so. This formulation, however, offers little help in deciding whether one has a duty to believe a given proposition under the given circumstances. Suppose someone has excellent logical skills, and it is a general epistemic duty that one ought believe the (not too remote) logical consequences of one’s beliefs. Does he have the duty to believe a given proposition p? It depends on whether p follows from his other beliefs. This suggests that if we want to formulate the principle with respect to individual propositions we need to relativize duties to the information possessed.

(DOIC) If it is one’s duty to believe (disbelieve, suspend belief about) p given the information he has, one has the cognitive competence to believe (disbelieve, suspend belief about) p given the information he has.

DOIC accounts for our different doxastic deontic judgments about Apartment, Theorem and their variations in the following way. The characters in each story are supposed to possess all the necessary information. Moreover, they all fix their beliefs in a way that violates the epistemic rule that all relevant possibilities must be taken into account. Since it is our duty to respect the epistemic rules, forming beliefs in a way that violates the epistemic rules prima facie amounts to failing to carry out one’s duties. It is at this point that DOIC enters the picture, limiting one’s duties to what lies within one’s cognitive competence. The characters in the variations—the young couple, the mediocre mathematician and Emily in the stroke of genius case—lack the cognitive competence to believe in the given situation what the epistemic rules tell us to believe. As a result, they are granted exemption from the duty to believe that. This is why we do not think it is right to blame them for their beliefs and to pass the doxastic deontic judgment that they should not have believed what they did and should have believed something else. The characters in the original stories, on the other hand, are presumed to be cognitively competent enough to shape their beliefs in keeping with the epistemic rules, so DOIC does not exempt them from their duty. That is why we find it justified to blame them.

The OIC principle Alston advocates (1988, p. 259) is different. Keeping the linguistic form of DOIC, it says:

(AOIC) If it is one’s duty to believe (disbelieve, suspend belief about) p, one has an effective choice whether to believe (disbelieve, suspend belief about) p.

The difference is that AOIC requires “effective choice”, i.e. voluntary control. Building voluntary control into OIC is harmless as long as we talk about moral obligations, which apply to actions, because voluntary control and possession of the necessary abilities mutually imply one another: if you voluntarily control doing something, you can do it; and if you can do it, you have voluntary control over doing it. When it comes to doxastic attitudes, however, possession of the necessary abilities—i.e. competence—does not imply voluntary control, and I suggest that doxastic duties and responsibility require competence only. Why not voluntary control? If we have a choice between voluntary control and competence, why do we opt for the latter in the doxastic sphere? This question needs to be addressed, because it is deeply ingrained in our theories about duties and responsibility that they require voluntary control.

The answer in brief is that where performance errors are possible, we always settle for competence, even when we are talking about voluntary actions. Here are two stories to illustrate this.

Aria

Diana is a magnificent soprano, who is particularly famous for her rendition of Mozart. The Queen of the Night is one her celebrated roles, which she has performed in several productions all over the world. Tonight she sings it again. She is in perfect shape, she has rehearsed carefully, and everything goes well up until scene 3 of act 2, but when she starts the great aria “Der Hölle Rache”, she feels that it is not quite right. She is not singing out of tune, she is way too accomplished for that, but that special quality which makes her rendition so expressive is missing. When she is finished, she looks down to the orchestra pit and sees the disappointment in the conductor’s eyes. After the performance she says to him: “I’m sorry, I know, I should have done better.”

Penalty

Cristiano is a great professional footballer, a starting member of the national team, who has won several cups with his club. He is excellent at free kicks, and it is usually he who kicks the penalties, because he is so good at it. Today his team is awarded penalty, and he is to kick it. He has trained carefully, he has also practiced penalty kicks, he is neither tired nor tense, there is nothing wrong with the grass or the ball. His plan is to place the ball in the top right corner, but as all accomplished penalty shooters, he is prepared to place it elsewhere if the goalkeeper moves into that direction. When he raises his foot for the kick, he sees that the goalkeeper is already moving to the right. This actually makes his job easier: he does not have to place the ball very precisely or apply great force. A solid kick to the middle or to the left would do, since the goalkeeper cannot change his course in time. To his dismay, he hits the left post. He falls to his knees, shakes his head and mutters to himself in utter despair: “How stupid. I should have scored.”

These stories are very similar to Apartment and Theorem except for the task the characters undertake. Diana and Cristiano are highly competent. The conditions are ideal for manifesting their competence, partly in virtue of their efforts to assure that they are in perfect condition. Yet they fail and reproach themselves for doing so. Just as in Apartment and Theorem, changing the stories in a way which makes the characters less competent would make the deontic should-statement at the end false. If an average soprano had to drop in, because Diana had an accident right before the performance, she could not be blamed if she did not sing as magnificently as Diana usually does. If Cristiano were replaced by a novice defender with no experience with penalties or free kicks, we would not blame him.¹⁹

What the formation of doxastic attitudes, singing an aria and shooting a penalty have in common is that they are all susceptible to performance errors. By performance error I mean mistakes committed by competent subjects when both their internal condition and the external circumstances are favorable, so the subjects have no excuses. In the moral sphere, however, there is no room for performance errors. If the internal and the external conditions are right, our morality—which includes such things as sensitivity to pick out morally relevant aspects of the situation, ability to determine what is morally right, tendency to be motivated by moral judgment—cannot fail to be manifested.

Let me elaborate. Imagine that someone does something which we would normally rule morally wrong. Suppose that it is not a piece of involuntary behavior which is not subject to moral evaluation, like when the bus suddenly stops and, grabbing for something to hold on, you unintentionally hit someone. Suppose that neither the subject’s internal condition nor the external conditions are such that they could prevent him from acting in the way morality requires—he is not coerced, is not non-culpably ignorant, and is able to act in the required way under the given conditions. In these circumstances we would conclude that there is something wrong with him from the moral point of view, he must suffer from some moral defect. We wouldn’t judge that he is fully moral, but his morality has somehow failed him. If one is a fully moral person, one cannot fail to act morally unless he has excuses. In other words, we do not allow that one can simply have a “bad day” from the moral point of view.

I take this to be a trivial point about the way we think about morality, but here is an example.²⁰ A child is drowning, and there is an adult standing by, who sees that and could easily plunge in the water and save the child, yet he does not do that. If he were to say that even though he is a thoroughly moral person, it just did not occur to him that he should save the child, we would not accept that. We would point out that the very fact that this did not occur to him shows that he is lacking in moral sensitivity to detect that he has a moral duty in this case. We would do similarly if had an explanation which does not qualify as an excuse, e.g. he wanted to save the child, but then noticed that the child is strikingly similar to someone who bullied him as a kid, which somehow prevented him from acting on his intention. We would say that the fact that his instinctive aversion to the lookalike of the childhood bully kept him from fulfilling his duty shows that his motivational system is morally skewed.

These considerations explain the difference between the moral and the doxastic sphere in the following way. Whenever there is room for performance errors, there is no full voluntary control. Singing an aria or shooting a penalty are voluntary actions, yet one does not have full voluntary control over how well one performs. Diana and Cristiano want to perform as well as they normally do, yet they fail. If we hold them responsible for meeting our legitimate expectations, we cannot tie responsibility to voluntary control; we have to maintain that responsibility in these domains does not require more than the competence necessary for meeting those expectations. Doxastic attitudes are not voluntary, so we do not have full voluntary control.²¹ Hence, if we consider ourselves accountable for shaping our doxastic attitude in keeping with the epistemic rules as much as we are able to, we must settle for cognitive competence as a requirement of responsibility.

Having explained why there is a difference between the moral and the doxastic sphere, I may now complete the argument against Alston. DOIC and AOIC often judge differently whether one has a duty to believe something under the given conditions; for instance, in Apartment DOIC suggests that it is Jack and Jill’s duty to believe that their plan is too risky, because their cognitive competence is sufficient, whereas AOIC suggests that they have no such duty, for they lack voluntary control. Nevertheless, this alone does not make DOIC and AOIC incompatible: there could be more than one OIC principles operative in the same domain, which impose different restrictions on our duties and supplement one another. DOIC and AOIC, however, cannot work in tandem, because AOIC renders DOIC inoperative. Granting Alston’s premise (2), we never have effective choice concerning belief, which, together with AOIC, entails that we have no doxastic duties, in which case DOIC cannot do any work, since it cannot ever exempt anyone from his doxastic obligations. In other words, if AOIC were operative, DOIC would be vacuous. I have argued that DOIC is the principle that regulates our doxastic duties. Since AOIC and DOIC cannot both regulate our doxastic duties, it follows then that AOIC is inoperative; therefore, it must abandoned, which leaves Alston’s premise (1) without support.

One might worry that this argument begs the question for the following reason. I argued for DOIC and against AOIC from our doxastic deontic practice, i.e. from facts concerning what doxastic deontic judgments we regard as true or false. Could not one reverse the argument and claim that what the incompatibility between our doxastic deontic practice and AOIC really shows is that our doxastic deontic practice is misguided? It seems that we need to show first by an independent argument that doxastic deontic practice is in good order, and only then can we use its incompatibility with AOIC to argue that the latter is wrong.

This worry is unwarranted, because it confuses the characteristics of doxastic deontic practice with its soundness. The argument for DOIC did not rest on the assumption that doxastic deontic practice is sound, i.e. that it does not involve false principles and does not systematically lead to false judgments. It was based, as it were, on anthropological facts concerning what doxastic deontic judgments we accept under various circumstances. Describing what we do does not commit one to the view that what we do is right. You might think of an anthropologist describing a group’s practices concerning witchcraft, the methods they use to identify witches, the measures they adopt to ward off curses and the like. The practices of the group depend for their soundness on the existence of witches, but the anthropologist’s characterization of these practices does not commit him to the existence of witches. It is one thing to describe the principles underlying a practice, and it is another thing to accept them.²²

One might respond that this reply merely exposes the real shortcoming of the argument. A purely descriptive analysis of a practice, which remains neutral on the issue whether the practice is sound, may identify the principles underlying the practice, but cannot show that those principles are true. Observing that the witch-believers shun the descendants of supposed witches, the anthropologist may conclude that being a witch is taken to be an heritable property, but his conclusion does not establish that witches exist. Likewise, one might protest that my argument establishes merely that our doxastic deontic practice rests on DOIC, but does not show that it is true. So doxastic deontic practice may be just as misguided as the practices concerning witchcraft, and DOIC may be just as false as the view that witches exist. So the incompatibility between DOIC and AOIC does not imply that the latter is false.

This objection does have point, but it gets the dialectic of the situation wrong. Alston’s only reason for premise (1), that doxastic deontology implies doxastic voluntarism, is that all duties are limited by effective choice. He makes no distinction between moral and doxastic duties, and takes it for granted that they are both regulated by the same OIC principle. I have shown that this cannot be granted, because the moral sphere is special in the respect that it rules out performance errors. Since there is no other reason to linking doxastic duties to effective choice than the idea that all duties are in the same boat, and that idea has been refuted, we have no reason to accept premise (1). What is right about the objection is that my reasoning could be defeated by showing that it is wrong to adopt a special regulation with respect to doxastic duties and other non-moral duties, and that all duties should fall under the same OIC principle as moral duties. Such an argument would show that even if our practice employs DOIC, DOIC is false and should be replaced by AOIC. I do not think it is easy to give such an argument.

5 Doxastic Deontology and the Deontological Conception of Justification

Alston’s argument is not primarily directed against the common deontic evaluations of belief we saw in the stories in Sect. 2, but against the deontological conception of justification (hereafter DCJ). So one might hope that the defense of doxastic deontology just offered goes at least some way towards rehabilitating DCJ. In fact, the opposite is true: the defense of doxastic deontology provides one more reason to reject DCJ—a reason similar to Alston’s arguments directed specifically against DCJ, which leave common doxastic deontic judgments unaffected.

DCJ holds that a belief is justified if and only if it does not involve violations of our epistemic duties, including our doxastic duties. Our epistemic duties derive from epistemic rules, but whether a particular person has the duty to adhere to the epistemic rules under the given circumstances is governed by an OIC principle, which was identified as DOIC. According to DOIC, it depends on one’s cognitive competence whether one has a duty to believe (disbelieve, suspend belief about) a proposition. Cognitive competence varies from person to person. Therefore, different people may not have the same doxastic duties even if they share the same information. It may then happen that two people who share the same information and believe the same things equally disobey an epistemic rule, yet only one them violates his duty by doing so, because the other one does not have the duty to obey that rule. Under these conditions DCJ implies that one of them has a justified belief, whereas the other one does not. So injecting the understanding of doxastic duties outlined earlier into DCJ makes justification dependent on one’s cognitive competence.

Here is an example. The issue is what to make of proposition p on the basis of the information at hand. Suppose that we have a good deal of information, and it is not easy to determine how it bears on p. On the face of it, it provides far more reasons to think that p is true than to think that it is false but, in fact, that evidence for p and against it is of equal weight. Enter two people, Sharp and Blunt, who have to decide whether to believe that p. Sharp, as his name suggests, has the necessary cognitive competence to evaluate the evidence properly, whereas Blunt does not. Both take into consideration all the information available, think long and hard, and eventually conclude that the evidence for p outweighs the evidence against it and come to believe that p.²³ The relevant epistemic rule says that if the reasons for and against a proposition are equally strong, one should suspend belief. They both violate this rule. But is it their duty to comply with it? Blunt does not have the cognitive competence which the proper evaluation of the evidence requires, so he is exempted from the duty to suspend belief about p. As he does not violate his duties, his belief is justified. Sharp, however, does not get exemption, so he violates his duties, and his beliefs are unjustified.

This conclusion is strange. Sharp and Blunt agree in everything we take to matter for justification, yet Blunt’s belief is justified and Sharp’s is not. They only differ in their cognitive competence, but that is sufficient to make a difference with respect to the justification of their beliefs. Of course, there is a trivial way in which justification depends on cognitive competence. Cognitive competence determines how good someone is at obtaining information, judging its relevance and seeing how it bears on a proposition, and someone who is more competent may acquire justified belief in cases in which the less competent cannot. In these trivial instances the standard of what counts as justified belief is kept constant, and cognitive competence determines one’s chance of meeting the standard. The Sharp-Blunt case works the other way round: the standard of justification is lowered for those with a lower level of cognitive competence, for we exempt them from the duty to comply with the epistemic rules they are unable to obey.

One might try to explain away this conclusion in this way. We know, after all, that Blunt makes a mistake in the evaluation of the evidence, so his belief is not justified either. But this will not do, since according to DCJ, a belief is justified if it does not involve a violation of one’s epistemic duties, and Blunt does not violate his epistemic duties. So to say that his belief is unjustified is to abandon DCJ.

Another, slightly more sophisticated response would be to modify DCJ by building possession of sufficient cognitive competence into the definition of justification. One would then have a justified belief only if he did not violate his duties and had sufficient cognitive competence. Blunt’s belief would then also be unjustified, because his cognitive competence is insufficient. But why say that Blunt’s cognitive competence is insufficient? The only reason to say that is that he cannot obey the epistemic rules (which apply in the given case). However, if sufficient cognitive competence is defined in terms of epistemic rules, the modified version of DCJ identifies justification with satisfying the epistemic rules. But that is not a deontological conception, so this response saves DCJ by tacitly abandoning it in favor of a non-deontological conception.

Yet another way to get rid of the strange conclusion fails for the same reason. One may suggest that DCJ could identify justification with fulfilling rather than not violating epistemic duties, which means having those duties and discharging them. Since Blunt does not have a duty to suspend belief about p, even though he does not violate this duty, he does not fulfill it either, so his belief is unjustified. But why think that there is a duty here which Blunt does not have? The only good reason is that the epistemic rules pertaining to the evaluation of evidence yield the verdict that the evidence for p and against it are equally strong, and another epistemic rule prescribes that belief must be suspended under these conditions, so there is a duty to do so. The duties pertinent to the given situation are identified with the epistemic rules applying to the situation, so fulfilling one’s duties reduces to satisfying the epistemic rules. But then any talk of duties is just a verbal flourish, and the new conception of justification is no longer deontological.

Adjusting one’s duties to one’s level of cognitive competence also leads to another conclusion, which is not only strange but clearly unacceptable. DCJ allows that people who share the same information and believe the same may have equally justified beliefs, even though only one of them reasons correctly. Let’s take a second Sharp-Blunt case to see that. Once again they have to determine whether the evidence at hand supports a particular proposition, and the evidence is so complex and convoluted that only Sharp has the cognitive competence to sort it out. After reviewing the evidence carefully and giving lots of thought to the matter, they both come to believe the proposition. Sharp does that in virtue of his superior competence, by discharging all his duties and reasoning exactly as one should reason in this case. Blunt’s reasoning is erroneous, he violates some epistemic rules, but given his limited cognitive competence, complying with those rules does not belong among his duties. Both of them observe their duties, so DCJ would have us say that they are both justified. Not only does DCJ allow that erroneous reasoning may yield justification, it also says that the justification based on erroneous reasoning may be just as good as justification based on impeccable reasoning. One may try to repeat the defensive moves we have seen in the first case, but they will not succeed this time either.

This is sufficient to show that DCJ is wrong, but reflecting on why we find the conclusion unacceptable may deepen the criticism. Everyone agrees that epistemic justification has something to do with two other important epistemic goals, truth and knowledge. Epistemologists have very different views concerning the relative importance of these goals and how they are related, so I need to tread carefully so as to avoid drawing fire from one camp or another. I take it that it is not controversial that justification has instrumental value in the sense that it matters to truth and knowledge, meaning that if you aim at truth or knowledge, other things being equal, it is better to have justified belief. This is a very weak claim. It does not imply that justified belief has no intrinsic value or that it is not the primary goal of inquiry. It does not imply either that justification is conceptually connected to truth or knowledge; in particular, it does not imply that justification is to be analyzed in a broadly reliabilist way or that knowledge is to be analyzed as justified true belief. It only means that if a belief is justified, we have more reason to think that it is true and that it constitutes knowledge, even if that reason is not sufficient and might be very weak.

The problem with DCJ is that it does not warrant the instrumental value of justification. If justification consists in observing one’s epistemic duties, there are two ways to avoid violating any particular duty: having that duty and discharging it, or not having that duty in virtue of one’s cognitive incompetence, which gives one license to break the corresponding epistemic rule—let us call these high road and low road, respectively. High road justification matters to truth and knowledge. Of course, Sharp’s justified belief in the second case may well be false, and even if it is true, he may not have knowledge because of some Gettier-problem. Nevertheless, he is better off with high road justification than without it: it is somewhat more likely that in the second case his belief is true and constitutes knowledge than in the first case when his belief lacks high road justification.

Low road justification, in contrast, does not matter to truth and knowledge. Blunt, who enjoys low road justification in both cases, may obtain true beliefs, but low road justification gives us no reason to think that he does.²⁴ Given that his cognitive competence is insufficient to evaluate the evidence correctly, i.e. he does not regularly succeed in observing the epistemic rules pertinent to the situation, why should we believe that he succeeds this time?²⁵ He stands worse with respect to knowledge, because we do not merely lack reasons to think that he has knowledge, we actually have reasons to think that he does not. Most epistemologists agree that knowledge involves true belief and something in virtue of which believing the truth is non-accidental. Even if Blunt’s beliefs are true, it is more likely that the mechanism which secures the non-accidentality of true belief is missing. Shaping one’s belief in accordance with the epistemic rules is supposed to be part of that mechanism. In the first case, Blunt’s cognitive incompetence suggests that the mechanism which secures non-accidentality is not present. In the second case we know that he reasons fallaciously, and even though fallacious reasoning may lead to true belief, that may only happen by accident. So we have a pretty good reason to think that he lacks knowledge.

One might perhaps wish to argue that low road justification still matters to truth and knowledge, even if it matters much less than high road justification, because low road justification means observing a smaller number of epistemic duties, and it is still better to observe these duties than to violate them. It is indeed true that there is a sense in which it is better to observe a smaller set of duties than to violate them, but doing better in this sense does not matter to truth and knowledge. Consider a parallel case. I have genuine difficulties in staying in tune especially when it comes to high notes. So given my abilities as a singer, it would be a major accomplishment to sing the aria of the Queen of Night with missing the tune only two dozen times; but I am sure you would not want to hear that. Good rendition is not relativized to one’s abilities as a singer. Truth and knowledge are not relativized to one’s cognitive competence either. If you lack the necessary abilities, skills and experience, it just does not matter that you do as well as you can. So even if low road justification has value, it does not have the instrumental value justification is supposed to have.

We have finally found why DCJ is fundamentally flawed. High road deontological justification is instrumentally valuable, it does the sort of work epistemic justification is supposed to do in epistemology. But by relativizing duties to cognitive competence, DCJ lumps together high road justification and low road justification, which has no instrumental value. Moreover, it cannot avoid doing so, because, if my argument stands, it is exactly by relativizing duties to cognitive competence, i.e. by adopting DOIC, that doxastic deontology can be saved. To put it differently, saving doxastic deontology comes at a price, which is that DCJ must be abandoned.²⁶

6 Conclusion

I have argued that Alston’s attack on doxastic deontology fails, because his crucial premise claiming that doxastic deontology implies doxastic voluntarism is unfounded. Alston supports the premise with a version of OIC in which the “can” is construed as having effective choice. I have tried to show with a couple of examples that this is wrong, and that in the version of OIC which applies to doxastic attitudes the “can” stands for cognitive competence: one has a duty to believe so and so if he has the cognitive competence to do that. However, this defense of doxastic deontology does not help the deontological conception of justification but aggravates its difficulties. Cognitive competence varies from person to person, which makes doxastic duties individually variable, which in turns yields that the standard of justification also varies from person to person. I do not think there is anything intrinsically wrong with a concept of justification which allows individual variation, but this is not the concept that matters for epistemology, since this kind of justification does not matter to truth and knowledge. It is one thing to judge whether one’s beliefs meet our legitimate expectations adjusted to his cognitive competence, and it is another thing to judge whether they are likely to be true or to constitute knowledge. Doxastic deontology is concerned with the first question, whereas most epistemologists are interested in the latter.