Ethical Evidentialism

I have tried in a number of different papers to defend the position of W. K. Clifford that “it is wrong, always, everywhere, and for anyone, to believe anything without sufficient evidence.” I believe I have been the first person in well over a century to argue that he was right.

You should not believe that adding arsenic to the curry will just make it zestier. It is imprudent if you are going to be the only one eating it. It is immoral if you are going to serve it to others.

Critics point out that there are lots of beliefs that are not as harmful. If you believe that mint sauce would improve your curry, you are not going to kill anyone.

The Cliffordian reply, which the critics have largely ignored, is that a particular belief itself might not have immediate harmful effects but accepting it can open the floodgates to other beliefs on analogous grounds and it is irresponsible to simply assume that all of the others will be equally harmless.

The situation has got mixed up because of William James’s attack on a straw man. James’s assumed that what mattered was truth or falsity, full stop, not the effects on others. In other words, he assumed that Clifford’s position was a form of epistemic evidentialism, not ethical evidentialism. With epistemic evidentialism, the Nazis’ false beliefs about Jews are no worse than false beliefs about the number of black holes in the universe. Ethically, James’s position is indefensible ivory-tower-ism.

Recently, I have worked on a new argument for ethical evidentialism using prospectivism (subjective consequentialism) and the moral encroachment thesis as my starting points. The result is a form of ethical evidentialism that is distinguishable from Clifford’s. This work has been accepted for publication but there is an unofficial version below.

This version of the article has been accepted for publication, after peer review but is not the Version of Record and does not reflect post-acceptance improvements, or any corrections. The Version of Record is available online at: Zamulinski, B. A New Argument for Ethical Evidentialism. Philosophia (2024). https://doi.org/10.1007/s11406-024-00769-7. Use of this Accepted Version is subject to the publisher’s Accepted Manuscript terms of use https://www.springernature.com/gp/open-research/policies/acceptedmanuscript-terms

A New Argument for Ethical Evidentialism

Abstract

This paper contains a new argument for evidentialism as an ethical rather than an epistemic doctrine. The argument relies on new developments in consequentialist thinking. The insights of the proponents of the moral encroachment thesis are used to show that we need higher standards of evidence, and to develop the concept of ethically sufficient evidence. It is demonstrated that prospectivism (subjective consequentialism) supports the contentions that we should not believe without ethically sufficient evidence, that we are permitted to believe when we have it, and that we are obligated to investigate to acquire it. It is argued that we do not have any obligations to believe. The ethical evidentialism argued for here is distinguishable from Clifford’s ethics of belief. Developing the doctrine clarifies some of our moral obligations, makes a positive contribution to prospectivism, and shows that evidentialism is better construed as an ethical doctrine.

Introduction

Consequentialist arguments for ethical evidentialism appear to depend on the philosophical work predominant at the time. According to Zamulinski (2022), Clifford, in ‘The Ethics of Belief’ (1879), relied heavily on Sidgwick’s Methods of Ethics (1874). Hence, it may be that developments in consequentialism can, and should, result in new arguments. The aim here is to provide a new argument for ethical evidentialism by combining the insights of prospectivism, which is also known as subjective consequentialism, and the moral encroachment thesis.

Prospectivism is roughly the stance that we are obligated to do, not what is objectively best, but what is best in light of the evidence. The moral encroachment thesis is the contention that some epistemically justified beliefs are, for moral reasons, not in fact justified. And the ethical evidentialism to be argued for here is the position that it is morally forbidden to believe anything without ethically sufficient evidence, morally permissible with it, and morally obligatory to investigate on occasion. For the time being, it can be said that ethically sufficient evidence is evidence that contrasts with epistemically sufficient evidence and that would satisfy proponents of the moral encroachment thesis when they are critical of epistemic standards. If anyone suspects circularity, I promise to define it later in a way that will eliminate the suspicion. As used in this paper, ‘evidence’ means whatever supports the contention that a proposition is true. It includes observation, testimony, mathematical proofs, and so on, and not just empirical evidence.

There are three reasons for developing and discussing a new version of ethical evidentialism. First, it clarifies and justifies some of our moral obligations. Second, it contributes to the development of prospectivism. And, third, it shows that evidentialism is better construed as an ethical doctrine than as an epistemic one.

I will start by outlining the argument for the moral encroachment thesis, examining the thesis critically, and arguing that its insight justifies higher standards generally and not just in morally fraught instances. With those higher standards, people who adhere to ethical evidentialism can achieve what proponents of the moral encroachment thesis want us to achieve. I will continue by outlining prospectivism. I will accept that prospectivism is true in view of the thought experiments presented on its behalf. I will show how it supports the contention that we should not believe without ethically sufficient evidence. Following that, I will argue that it is permissible to believe when we have ethically sufficient evidence and that an obligation to investigate follows both directly from prospectivism and indirectly via the other propositions of ethical evidentialism. I will argue that we lack obligations to believe—even when ethically sufficient evidence is available. Finally, I will distinguish ethical evidentialism from the consequentialist interpretation of Clifford’s ethics of belief.

Moral Encroachment and Standards of Evidence

The moral encroachment thesis is that the same type of evidence with the same degree of strength can justify belief in one situation but not another because of moral considerations. Rima Basu (2019a) starts arguing for what she calls radical moral encroachment by discussing a situation, first discussed by Tamar Gendler (2011), in which there is about a 90% probability that a black man at a venue is an employee and not the guest of honour at a function being held there, in which some individuals infer that the guest of honour is an employee, and consequently ask him for help with their coats. In the situation outlined, it is purportedly epistemically rational to infer that the guest of honour is an employee in the same way that a 90% probability of rain makes it epistemically rational to believe that it will rain. It is also racist. While the purported rationality of the inference licenses the belief that the black man is an employee, its racism means that it is forbidden. Basu concludes that rationality needs to be modified by morality, and, specifically, that, in cases like this, we need more evidence than we do ordinarily.

Using similar reasoning, James Fritz (2021, 101-102) discusses the case of a woman asked to send a package where, given the evidence provided by her memory of the instructions she was given, she is 95% confident that it is not necessary that she dispatch it immediately and 5% confident that it is. There is a long queue and she will be late for an important social engagement if she posts it immediately rather than waiting until the next day. Objectively, she can wait. According to Fritz, however, the seriousness of the situation means that she ought to post the package immediately despite the long lineup and despite her subsequent late arrival at the social event. According to Fritz, it is also true that the evidence permits her to believe that it is not necessary to mail the package immediately.

But there are at least three problems with this kind of example. First, there are sometimes alternatives to adhering to higher evidential standards. In the case of Basu’s guest of honour, one option for the potential wrong doers at the venue is to stand near the coat check room in the expectation that, if the black man is an employee, he will come over to help them or ask another employee to help. Another is to make an approach to the man that does not presuppose that he is an employee. For instance, they could say, ‘Excuse me, sir, do you know where we can find an employee?’ This does not insult the black man, no matter whether he is an employee or the guest of honour. As for the Fritz example, whether there is an alternative to requiring more evidence depends on things like whether someone else can be persuaded to take responsibility for the package.

There is also the problem of how much evidence we need to justify believing that a particular situation is one in which we should require more than the standard amount of evidence. Arguably, we should require less to ensure that we err on the side of caution. If not, we are entitled to ask why not, and there is no good answer. So, we would end up in a situation in which we should sometimes demand less with respect to one belief because we should demand more with respect to another. And, we would always have to ask whether a situation is one in which we should demand less. The moral encroachment approach would require us to monitor our circumstances constantly.

Finally, it is arguable that it is not rational to believe either that the black man is an employee or that it is not necessary to post the package. As mentioned, Basu accepts that evidence that it is 90% probable that it will rain is sufficient to justify believing that it will rain. But, unless this is the near-tautology that evidence that it is 90% probable that it will rain is sufficient to justify believing that it is 90% probable that it will rain, it could be argued that it means that evidence that it is 90% probable is sufficient to justify believing that it is more than 90% probable. In light of this, it could also be argued that the inference is not rational at all. Fritz likewise gives no weight to the difference between the specific degree of confidence justified by the available evidence and the potentially greater degree of confidence with which the proposition in question is believed. Given that they propose higher standards in other cases, both Basu and Fritz need to defend the rationality of believing that it will rain when there is just a 90% probability of rain. The need for them to do so is underscored by the fact that there is an argument against their position, which will be discussed below.

Having indicated three problems, it must be added that there are cases in which none of them occur. Suppose that someone is concerned because their child could have either a minor condition that will clear up without intervention or a serious condition that requires urgent medical care. In the circumstances, it could be that they are not justified in relying on the judgment of a general practitioner but should, instead, get an opinion from a specialist. Obviously, the opinion of a specialist is better evidence than the opinion of a general practitioner. In circumstances in which the stakes were lower, it would be permissible for them to rely on the opinion of the general practitioner, but the more serious possibility requires better evidence.

To avoid the problems while retaining the insight of the proponents of the moral encroachment thesis, we could approach the issue from the opposite direction. Instead of accepting that it is rational to believe that it will rain when it is 90% probable and therefore rational to believe that the black man is an employee when it is 90% probable, we could deny that it is rational to believe that the black man is an employee and infer that it is not rational to believe that it will rain in the first place. In other words, we could contend that, if there is insufficient evidence that the black man is an employee, there is also insufficient evidence that it will rain.

Not only could we do so, but also we should do so, because beliefs that appear harmless in isolation may be harmful nonetheless. They may be harmful because our standards of evidence have to be adjusted to accommodate them and our adjusted standards then permit us to acquire an unknown number of other, presently unknown, beliefs too. Some of the unknown beliefs could be harmful. Indeed, the point is illustrated by the examples provided by Basu and Fritz. The standards of evidence that permit someone to believe that it will rain when it is 90% probable permit them to believe objectionable propositions too. If we adhered to higher standards of evidence in all cases, we would not have to consider alternatives to adhering to them, we would not have to monitor our circumstances constantly, and we would avoid problematic beliefs automatically. Thus, adhering to higher standards would solve all three problems with the moral encroachment thesis that were identified earlier.

The foregoing is like a move that Clifford made. Clifford acknowledged that beliefs like ‘It will rain’ do not themselves always have morally adverse consequences. However, he argued that believing in such cases makes us susceptible to accepting other beliefs that do: ‘every time we let ourselves believe for unworthy reasons, we weaken our powers of self-control, of doubting, of judicially and fairly weighing evidence’ (1879, 185). Someone who starts out believing that it will rain when it is merely 90% probable, which has no morally adverse consequences per se, risks ending up believing that the black man is an employee when it is merely 90% probable, which can have adverse consequences. While proponents of the moral encroachment thesis would say that rationally sufficient evidence must be supplemented by moral considerations, Clifford would say that what their examples really show is that what has been accepted as sufficient evidence for epistemic reasons is not actually sufficient at all—he recommended that we adopt higher standards across the board.

Clifford’s recommendation is a reasonable one from a moral perspective. Epistemic evidentialism concerns what promotes the acquisition of true beliefs. As William James put it when purportedly criticizing Clifford, ‘We must know the truth; and we must avoid error—these are our first and great commandments as would-be knowers ...’ (1896b, 25; italics in original). In contrast, ethical evidentialism has to do with what promotes or impedes the acquisition of beliefs because they affect what we do, particularly when there are consequences for others, especially adverse consequences. Epistemically, one case of believing without sufficient evidence is much like any other; ethically, one case can be much more serious than another. An unwarranted belief that members of one ethnic group are inferior to members of another is no worse epistemically than an unwarranted belief about the number of moons around Jupiter, but it is obviously far more serious from the moral point of view, because the potential for harm is far greater—and this is far from being the only example of the phenomenon. It is obviously more important to ensure that we act righteously than it is to ensure that our minds merely have the right kind of contents. Therefore, epistemic evidentialism is by no means a credible alternative to ethical evidentialism.

If evidentialism is supported by a moral doctrine like prospectivism, as it will be shown to be in the next section, and if epistemically sufficient evidence is not good enough, sufficient evidence must mean ethically sufficient evidence, where ethically sufficient evidence contrasts with epistemically sufficient evidence. As for what constitutes ethically sufficient evidence, if a 90% probability of rain were sufficient evidence for ‘It will rain’, then sufficient evidence would be compatible with a 10% probability of error. With the foregoing in mind, it is reasonable to define ethically sufficient evidence as the amount and quality of evidence that objectively minimizes the probability of error when we believe. This definition is clearly not circular. It is ‘ethically sufficient evidence’ because of the contrast with epistemically sufficient evidence and because it is ethically significant cases that compel us to notice the need for higher standards of evidence, not because its definition involves a reference to ethical considerations.

It is possible to have ethically sufficient evidence. Here are some examples. I have ethically sufficient evidence that a black cat is lying on the desk near me as I type this sentence because I can see her there and because I can reach out and stroke her. I have ethically sufficient evidence that a trust will be established after my death because I have made the requisite legal arrangements. We have ethically sufficient evidence that the theory of evolution is true even though it is possible in principle that it is false. A physician has ethically sufficient evidence that their diagnosis is correct if they treat the patient accordingly and the patient recovers—even though there will always be some probability that it was the natural healing process and not the physician’s treatment that caused the recovery. We would have ethically sufficient evidence that a drug is effective if randomized, controlled clinical trial has been a successfully replicated and there has been a history of successful use when prescribing it was indicated. In the final example, the probability of error has been minimized as much as practically possible; we could get closer and closer to certainty but never actually achieve it.

What counts as minimizing the probability of error will probably vary from field to field, and will have to be determined for specific fields by experts in the fields. It may be possible give examples but not to come up with a universally applicable definition, particularly because we can develop new ways of minimizing error—the first randomized, controlled trials were carried out in the 1940s. It is like recommending that we take measures to minimize accidental deaths and injuries. What counts as a reasonable thing to do to achieve that aim will depend on things like whether electrical wiring or heavy equipment is involved. Moreover, there is no reason to believe that there will be much overlap with respect to the recommendations in the two specified situations, let alone across the full range of situations in which safety recommendations could be made. New technologies will probably require novel recommendations as well. We need the generalization that we should promote safety to justify specific measures. Likewise, we need the generalization that we should minimize error to justify particular ways of doing so.

High standards of evidence do not impede action because we can act on probabilities or assumptions instead of beliefs. In the specialist/general practitioner kind of situation, the parents can have an obligation to act as though the specialist is right even if they do not yet believe them. In fact, they can be both obligated to follow the specialist’s advice and obligated to wait at least until after they have seen the results of what has been recommended before they believe or disbelieve. This is not an unusual situation. We frequently act on probabilities or assumptions instead of beliefs, and often must. For example, when it comes to the interpretation of symptoms, physicians are sometimes confronted with a range of possible causes, from the common to the rare (Apple 1996). They are not always able to determine which of the causes is the actual one without assuming one of them and treating the patient on that assumption, there being no diagnostic tests available. In such cases, physicians typically assume that the most common cause is the most probable one and prescribe a treatment on that basis. By doing so, they acquire evidence about which possible cause is the actual one. If the first course of treatment works, its success confirms that the most common cause is the actual cause. If it does not, its failure shows that the cause is one of the less common ones. In sum, not only can act we on probabilities or assumptions that we do not yet believe, but also we acquire evidence by acting on them, which is another point that Clifford (1879, 189) made. The difference between beliefs and probabilities or assumptions is that, when we believe, propositions are no longer on probation, so to speak. When we believe propositions, we usually consider the issue of their truth value to be settled. Believing prematurely can have morally undesirable consequences, as I will point out in the next section.

From Prospectivism to the First Proposition of Ethical Evidentialism

As stated in the introduction, prospectivism (Zimmerman 2014)—or subjective consequentialism (Andric 2011)—is the doctrine that we have an obligation do what is best in light of the evidence we have instead of what is objectively best. It originates with thought experiments by Frank Jackson (1991). In one of his thought experiments, the first treatment for a medical condition would result in some improvement, the second would be a complete cure, and the third would cause the death of the patient. However, the evidence shows that the first would result in some improvement, that the second would either cure or kill, and that the third would also cure or kill. Given the evidence, it is the first treatment that should be applied even though, objectively, the second treatment would be better. In such cases, the evidence justifies performing actions even though they are not objectively the best.

There are examples that make more realistic assumptions that lead us to the same conclusion. For instance, a physician may have to choose between two treatments, where the first is supported by the results of a well-designed randomized, controlled trial that has been replicated while the second has not yet been tested but would be shown to be even more effective than the former if it were tested. Although the second treatment is objectively superior, and even if they believe it to be superior, the physician should apply the first treatment. It is obvious that this is right; the occasional correct guess would be outweighed by the vastly more probable wrong ones. Moreover, it must be emphasized that it is not just that the physician may apply the first treatment; it is that they must. The physician is morally obligated, not merely morally permitted, to act on the assumption that the first treatment is the one that should be given to the patient. This result appears to be true no matter which moral theory is true and if it conflicted with a theory, it would be so much the worse for the theory.

Ignoring some special cases, Zimmerman (2015, 59-60) asserts that

the prospectively best option [is] that which maximizes projected value—that is, that option which constitutes one’s best bet in light of the probabilities of the various possible outcomes of one’s options and of the various possible values of these outcomes.

Zimmerman distinguishes prospectivism from subjectivism as well as objectivism. I will not recapitulate his arguments, which I find convincing, but emphasize that thought experiments like Jackson’s are so persuasive and the absence of counterexamples so striking, that I am willing to accept prospectivism as not merely plausible but true. Those who have less confidence in the truth of prospectivism may prefer to regard this paper as containing an argument for the conditional that ethical evidentialism is true provided that prospectivism is true, rather than an argument for the truth of ethical evidentialism, full stop.

Prospectivism does support the first proposition of ethical evidentialism. As suggested in the previous section, when we believe, propositions are no longer on probation. If we believe without sufficient evidence, we end probation prematurely. This has the following consequences. First, if we were to believe anything without sufficient evidence when there was evidence available, then we would be liable sometimes to act on those beliefs and thus liable sometimes to act on something other than the best available evidence. Our actions might sometimes be the same as if we were to act in light of the best available evidence—but only by chance—and they would frequently be different. Second, believing without sufficient evidence in one case would make us susceptible to believing without it in other, analogous, cases, which would magnify the effect of the first instance. To avoid doing the wrong thing, we would have to luck out not just once but many times. And, finally, we are liable to stop investigating, which obviously tends to prevent us from acquiring available evidence. If we do not acquire the best available evidence, we cannot act on the best available evidence. No doubt, we can re-examine beliefs, but we need some reason to do so. In the absence of a reason to re-examine them, beliefs close off avenues of investigation, so believing prematurely tends to prevent us from continuing investigations that should not be concluded. For these reasons, believing on insufficient evidence is liable to get in the way of acting in the light of the best available evidence. Therefore, prospectivism precludes believing without sufficient evidence.

Michael Pace argues extensively that it is unreasonable to interpret as a moral claim Clifford’s conclusion that it is wrong to believe anything without sufficient evidence. But he fails to make a strong case against interpreting it that way. Since his objections do not damage Clifford’s ethics of belief interpreted as a moral doctrine, they will not damage the first proposition of ethical evidentialism. It is worthwhile showing that they will do no damage by taking a detour to rebut his objections.

Pace declares that ‘it is not difficult to dream up counterexamples … in which it is not only morally permissible to believe on insufficient evidence, but it even seems to be morally required’ (2011, 243). His example is a fictional one in which Austin Powers has a moral obligation to believe something without sufficient evidence in order to satisfy a demand by Dr. Evil, and thereby save the world. One problem with this is that stipulating that an implausible piece of fiction is a counterexample does not make it one. Furthermore, later in the same paper, Pace writes:

But when one recognizes that one’s evidence does not make a proposition more likely than not, to believe on the basis of pragmatic reasoning is tantamount to committing oneself to p’s being more likely than not based on reasons that the subject recognizes are irrelevant to whether p is more likely than not. Reasoning in such a way in full reflective awareness of what one is doing may be psychologically impossible.’ (2011, 252)

Since the Austin Powers case is such a case, Pace’s counterexample requires that we have the ability to do what Pace says may be psychologically impossible. And he is right about the psychological impossibility. There is no more evidence that we can believe at will when we know we lack sufficient evidence than there is evidence that we can overcome gravity and levitate just by willing to do so.

Additionally, Pace contends that ‘subjects are not guilty of doing something morally wrong when they form beliefs based on insufficient evidence that are the result of cognitive malfunction or a lack of intelligence, although such beliefs are arguably epistemically unjustified’ (2011, 243). The two cases Pace mentions are quite different. Obviously, beliefs that result from cognitive malfunction are excusable. But arguments that it is wrong to do something frequently ignore the possibility that there may be excuses for doing the wrong thing, so Pace is imposing an arbitrarily high standard here.

As for lack of intelligence, Pace ignores the possibility that people can avoid believing on insufficient evidence by suspending belief—people do not have moral permission to believe what they cannot understand. Pace also ignores the fact that every human being exhibits lack of intelligence with respect to most areas of human inquiry. We see this whenever intellectuals stray outside their area of expertise; in such situations, we see intelligent people acting unintelligently. As was observed in one case, ‘For a clever man, he is awfully stupid’ (King 1991). Since lack of intelligence is widespread, even if unevenly distributed, if it excused unwarranted beliefs, it would undermine epistemic evidentialism too.

Pace contends that ‘some cases of belief on insufficient evidence seem entirely trivial, and not morally wrong even in a prima facie sense’ (2011, 243). He ignores the fact that, although, in many cases, believing without sufficient evidence does not have immediate adverse moral consequences, there is still a risk that believing in such cases would make people susceptible to believing less benign propositions on analogous evidence. Consequently, like so many other critics who have made the same claim (see Ward 1878, James 1896a, Meiland 1980, Code 1987, and Marusic 2011), he has inadequate support for his contention. In sum, there is no reason to take Pace’s objections seriously.

Turning to the issue of counterexamples more generally, there is a difference between the proposition that it is wrong to believe anything without sufficient evidence qua stand-alone principle and the same proposition qua corollary of prospectivism. It may be possible to devise counterexamples to the stand-alone principle that would not succeed against the corollary. Against the former, what is needed is a counterexample in which there are neither direct nor indirect adverse consequences. Against the latter, the only thing that would work is a counterexample that showed not only that there were no adverse consequences of either kind but also that someone could act responsibly and conscientiously while failing to act in light of the best available evidence. It would not be enough to show that they are occasionally lucky and accidentally do what they ought to do or avoid doing what they ought not to do. It is hard to see that there could be such a counterexample.

Completing the Justification of Ethical Evidentialism

While the moral encroachment thesis shows the need for more demanding standards of evidence, prospectivism justifies the permissibility of believing in some kinds of cases. If we must act on the best available evidence, there is no bar to believing when we have ethically sufficient evidence, which was defined earlier as evidence sufficient to objectively minimize the probability of error. An obligation to act on the best available evidence is still an obligation to act. Consequently, any prior investigation must come to an end either when we have run out of time or resources, or when we cannot do better qualitatively with respect to the evidence than we have already done. We can make quantitative additions to the evidence ad infinitum but we cannot improve it qualitatively when we have objectively minimized the probability of error. Consequently, there is no justification for denying that it is permissible to believe when we have ethically sufficient evidence.

A moral obligation to investigate can be supported either directly by prospectivism itself, or indirectly by the other propositions of ethical evidentialism. As for the former, if prospectivism were not accompanied by an obligation to investigate, it would devolve into an obligation to act on whatever we happen to believe at a particular time, which is no obligation at all. If we have an obligation to act on the best available evidence but lack ethically sufficient evidence, we must try to get more or better evidence, provided that we have the time and resources to do so. Moreover, the greater are the stakes, the more stringent is the requirement. Investigation is obligatory because it increases the probability that we will have the best available evidence on which to act when we do act. Since we have an obligation to investigate, the objectively best thing to do becomes an ideal towards which we strive.

Putting to one side prospectivism and the immediate derivation of a duty to investigate, a duty to investigate is an essential part of any form of evidentialism. On the one hand, permissions to believe and prohibitions on believing would both be pointless without a duty to investigate—people would not have an obligation to do the only thing they could do to try to ensure that their beliefs met the appropriate standards. On the other, a duty to investigate would be pointless in the absence of permissions and prohibitions. Investigation makes sense only as a means to determine what we are permitted to believe or to do. If investigation were an end rather than a means to an end, medical scientists could fulfil their obligation to investigate by carrying out replications of a clinical trial of a drug past the point at which it was necessary or reasonable to do so, no matter the results of the replications. Worse, if investigation were an end, we could fulfil our obligation to investigate by continuing to investigate after we had acquired a belief that did not conform to good standards of evidence, no matter whether the standards were ethical or epistemic. For instance, someone who believed that homeopathy was effective could continue to carry out clinical trials despite failure after failure. But, since the claims of homeopathy are false, they would not be doing anything morally important. A duty to investigate requires that there be permissions and prohibitions with respect to beliefs, and vice-versa.

Rejecting Obligations to Believe

The three propositions of ethical evidentialism can now be restated more accurately as follows: 1) it is always morally wrong to believe anything without ethically sufficient evidence, 2) it is permissible to believe when we have ethically sufficient evidence, and 3) we have an obligation to investigate before acting when we lack ethically sufficient evidence for relevant propositions, provided we have the time and resources to do so. The three are not accompanied by any obligations to believe. I reject the view that we have an obligation to believe when we have sufficient evidence; I reject both Pace’s opinion that we are sometimes morally required to believe and Fritz’s stance when he writes of people being ‘rationally required to form some belief’ (2021, 109).

The first problem with obligations to believe is that there are cases in which it is obviously neither epistemically nor ethically obligatory to believe something even though we undoubtedly have sufficient evidence for it. To be specific, if someone has sufficient evidence for P, they have sufficient evidence for an infinite number of disjunctions that include P as a disjunct, each of which they can validly derive using Addition. If we had an obligation to believe even one of them, we would be committed to believe all of them, there being no relevant difference that would enable us to justify choosing some while rejecting others. But it is neither epistemically nor ethically required to believe an infinite number of such disjunctions. Therefore, for any possible disjunct Q, it is neither epistemically nor ethically required to believe P v Q even if we have sufficient evidence that P. Indeed, if P were logically true and if someone knew it, they would have the strongest possible support for P v Q. Nonetheless, they would still not be required to believe it, no matter whether we are referring to epistemic or ethical requirements. If we are not required to believe something when it has the strongest possible support, we are not required to believe anything that has less.

Another problem with obligations to believe is that people can have the evidence but fail to grasp its significance. This happens in classrooms all the time: some students understand but some do not. Some will never be able to understand. In fact, it seems likely that everyone will encounter something that they cannot understand despite the evidence that supports it, provided they are exposed to a wide enough range of possibilities. This makes obligations to believe impractical.

Finally, since we can act on probabilities or assumptions instead of beliefs, it is not necessary for us to believe any particular propositions in order to perform specific actions, or to do our duty. It may even be counter-productive. For example, a bridge inspector may have a duty to report that the physical state of a bridge indicates that the bridge could soon collapse. But he might also believe that a religious ceremony has guaranteed the integrity of the bridge. He can fulfil his professional duty while believing that the bridge is safe. Indeed, requiring him to believe what he is going to report would impede his fulfilment of his duties. Since it is never necessary to believe, it would be very difficult to prove the existence of an obligation to believe, if it were possible at all.

Distinguishing the Doctrine of Ethical Evidentialism

The first proposition of ethical evidentialism obviously resembles Clifford’s stirring conclusion that ‘it is wrong, always, everywhere, and for anyone, to believe anything without sufficient evidence’ (1879, 186). But ethical evidentialism is not an augmented recapitulation of Cliffordianism—despite my having appealed to some of Clifford’s insights. Of course, the task is to distinguish ethical evidentialism from a consequentialist interpretation of Cliffordianism. There are non-consequentialist interpretations of Clifford (Aikin 2014) but ethical evidentialism is not going to be mistaken for Cliffordianism when it is interpreted in that way.

First, the propositions of ethical evidentialism are corollaries of prospectivism whereas Clifford’s conclusion is a stand-alone principle, ostensibly supported by the contention that every act of believing without sufficient evidence is wrong because of its consequences. Hence, the kind of support required is different in the two cases.

Second, although purported counterexamples to Clifford’s principle typically fail because they neglect the indirect effects of believing propositions without sufficient evidence, counterexamples appear to be possible in principle (Zamulinski 2022). In contrast, deriving the principle from prospectivism immunizes it far more strongly against counterexamples, as indicated earlier.

A third difference has to do with the obligation to investigate. Although Clifford says that we should investigate to acquire evidence if we are going to believe, he appears to allow that we can always avoid investigating by refraining from believing, stating that someone who does not have time to investigate ‘should have no time to believe’ (1879, 188). In other words, his position is that our having an obligation to investigate is contingent on our attempting to acquire beliefs, the attempt being a purely voluntary matter. Here, since ethical evidentialism is a corollary of prospectivism, it is clear that it is not always permissible to refrain from investigating. Instead, some of us sometimes have an obligation to investigate, full stop. We have an obligation to investigate because we have an obligation to do what we can in the time available to try to ensure that we do what we ought to do.

Finally, ethical evidentialism is more clearly distinguished from its epistemic counterpart than is the ethics of belief. Philosophers do sometimes dispute whether Clifford’s ethics of belief is an epistemic or an ethical position. We have seen that James assumed that it was epistemic, and I considered and rejected Pace’s arguments for the claim that it must be interpreted epistemically. Whereas the ethics of belief is ambiguous, at least at a first glance, ethical evidentialism is unambiguously ethical. The difference is explained by the fact that ethical evidentialism is a corollary of prospectivism, which is unquestionably an ethical doctrine, whereas the ethics of belief is a stand-alone principle.

One way in which ethical evidentialism is the same as the consequentialist interpretation of the ethics of belief is that both have to do with the consequences of belief. In neither case is believing something ever intrinsically wrong. For example, racist beliefs are morally consequential only if they influence our behaviour. A world in which racists acted indistinguishably from non-racists, so that those who would otherwise be adversely affected were completely unaffected, would be no worse than a world in which there were no racists. And such a world is a logical possibility, however improbable. If people were admonished to avoid performing any racist actions, the admonition could counteract any racist beliefs they might possess or acquire. And surely this could work. After all, there was a significant increase in manifestations of racism after Donald Trump effectively gave Americans and others permission to be racist. Many of his followers had probably been racist prior to his emergence on the electoral scene but had not previously permitted their racism to show because of social disapproval. If it is possible for people to hide their bigotry some of the time, then it is possible for them to conceal it all the time. Thus, it is logically possible that people could keep their racist beliefs from having any effect on their behaviour. So, what matters morally must be the actual or potential adverse consequences of the beliefs on people subjected to racism. It is likewise in every other case.[1]

Conclusion

Ethical evidentialism imposes on us a moral obligation always to try to bridge the gap between the actual and the ideal, the ideal being what we objectively ought to do. Therefore, while based on prospectivism, which requires us to act on the best available evidence, ethical evidentialism also augments prospectivism with a requirement to continually strive to make improvements for as long as we lack ethically sufficient evidence. As we have seen, ethical evidentialism trumps epistemic evidentialism as well. Evidentialism is better construed as an ethical doctrine than as an epistemic one. Truth is never an end but always the means to an end. The goal of evidentialism is ensuring that we act in righteous ways.[2]

References

Aikin, Scott F. 2014. Evidentialism and the Will to Believe. London: Bloomsbury.

Andric, Vuko. 2011. Objective consequentialism and the licensing dilemma. Philosophical Studies, 162: 547-566

Apple, Michael. 1996. The Penguin book of symptoms and early warning signs: a comprehensive new guide to self-diagnosis. Penguin: Godalming, Surrey.

Basu, Rima. 2019a. Radical moral encroachment: The moral stakes of racist beliefs. Philosophical Issues 29: 9–23

Basu, Rima. 2019b. The wrongs of racist beliefs. Philosophical Studies 179: 2497-2515.

Clifford, William Kingdon. 1879. The Ethics of Belief. In his Lectures and Essays, Vol 2 (London: MacMillan & Co), 177-211.

Code, Lorraine. 1987. Epistemic Responsibility (Hanover and London: University Press of New England).

Fritz, James. 2021. Akrasia and Epistemic Impurism. Journal of the American Philosophical Association 7(1): 98-116.

Gendler, T. S. 2011. On the epistemic costs of implicit bias. Philosophical Studies 156: 33–63.

Jackson, Frank. 1991. Decision-Theoretic Consequentialism and the Nearest and Dearest Objection. Ethics 101: 461-482.

James, William. 1896a. The Sentiment of Rationality. In his The Will to Believe and Other Essays in Popular Philosophy (Cambridge, MA and London, England: Harvard University Press, 1979), 57-89.

James, William. 1896b. The Will to Believe. In his The Will to Believe and Other Essays in Popular Philosophy (Cambridge, MA and London, England: Harvard University Press, 1979), 13-33.

King, Naomi. 1991. Personal Communication.

Marusic, Berislav. 2011. The Ethics of Belief. Philosophical Compass 6: 33-43.

Meiland, Jack W. 1980. What Ought We to Believe? Or The Ethics of Belief Revisited. American Philosophical Quarterly 17: 15-24.

Pace, Michael. 2011. The Epistemic Value of Moral Considerations: Justification, Moral Encroachment, and James’ ‘Will to Believe’. Nous 45:2; 239–268.

Sidgwick, Henry. 1874. Methods of Ethics (London: MacMillan and Co.).

Ward, W. G. 1878. The Reasonable Basis of Certitude. The Nineteenth Century 3(13): 531-547.

Zamulinski, Brian. 2022. Clifford’s Consequentialism. Utilitas 34(3): 289-299.

Zimmerman, Michael. 2014. Ignorance and Moral Obligation (Oxford: Oxford University Press).

[1] Basu (2019b) would disagree, but her argument is vitiated by the fact that her ‘rational racist’ actually commits the fallacy of division, inferring that his actual black customers are bad tippers because the average black person supposedly tips less than others.

[2] I am grateful to anonymous reviewers for their insightful comments.