Clifford's Consequentialism

Abstract

It is morally negligent or reckless to believe without sufficient evidence. The foregoing proposition follows from a rule that is a modified expression of W. K. Clifford's ethics of belief. Clifford attempted to prove that it is always wrong to believe without sufficient evidence by advancing a doxastic counterpart to an act utilitarian argument. Contrary to various commentators, his argument is neither purely nor primarily epistemic, he is not a non-consequentialist, and he does not use stoicism to make his case. Clifford's conclusion is a universal generalisation that is in a precarious position because of potential counterexamples. But the counterexamples do not preclude a rule against going beyond the available evidence and it is worthwhile making a moral case for such a rule.

Introduction

W. K. Clifford's ethics of belief (1879a) is not merely a Victorian curiosity. In a world where pseudoscience, conspiracy theories, and post-truth politics are commonplace, it is still – or especially – worthwhile to make a moral case against letting our beliefs outstrip the available support for them. Clifford's argument for the proposition that it is wrong to believe without sufficient evidence can best be described as a doxastic counterpart to an act utilitarian argument. This interpretation is supported by both the background to and an analysis of the argument. Contrary to some commentators, none of whom successfully rebuts the evidence for his utilitarianism, Clifford did not advance a purely or primarily epistemic argument; did not take a non-consequentialist position; and did not propose an argument with a stoic basis. But Clifford's conclusion, which is a universal generalisation, is undercut by potential counterexamples. Nevertheless, there is still a case for a rule against believing without sufficient evidence, where ‘evidence’ is interpreted liberally to encompass anything that can be used to help to establish the truth of a proposition, including observation, testimony, empirical evidence, logical proofs, and so on. The case for the rule is based on prospectivism (Zimmerman 2014) or subjectivism (Andric 2011), the view that we should do what is best given the evidence we possess. The term ‘prospectivism’ will be used here.

Clifford qua act utilitarian

The aim of this section is to justify characterising Clifford's argument for the ethics of belief as a doxastic counterpart to an act utilitarian argument.

Clifford's meta-ethics was evolutionary. He thought that conscience, which tells us what is right and wrong, is an evolutionary phenomenon that is developed by communities to promote their own welfare. Different communities develop different moral codes or traditions: the Mishna in the case of the Jewish people (Clifford 1879b: 132), stoicism in the case of the Romans (Clifford 1879b: 133), and utilitarianism in the case of the British (Clifford 1879b: 173), as will be confirmed below. He assumed the truth of group selection and believed that the moral code adopted by a group makes a difference to its fitness. ‘Those tribes have on the whole survived in which conscience approved such actions as tended to the improvement of men's characters as citizens and therefore to the survival of the tribe’ (Clifford 1879c: 119). This view comes from Darwin, who wrote that ‘an advancement in the standard of morality and an increase in the number of well-endowed men will certainly give an immense advantage to one tribe over another’ (1871: 166).

Clifford, like Darwin and others at the time, was wrong about group selection. For one thing, he was apparently unaware of its dark side. He maintained that there was more than one sort of community, writing that ‘we have a sense of family duty, of municipal duty, of national duty, and of duties towards all mankind’ (Clifford 1879b: 170), and maintained that there could be selection for morality at all levels. But while it ‘does provide a setting in which helping behavior directed at one's own group can evolve; however, it equally provides a context in which hurting members in other groups can be selectively advantageous’ (Sober and Wilson 1998: 9). Clifford did not show that group selection could create universal norms instead of resulting in ‘ethics’ for insiders and realpolitik for outsiders. Furthermore, group selection is no longer taken seriously as a potential selective process. With a few exceptions like Sober and Wilson (1998), contemporary evolutionary biologists regard as plausible selective processes only individual selection, kin selection, and reciprocal altruism. The prevailing view is that group selection is possible but so improbable as to have had an insignificant effect (Wade 1978).

Unaware of the problems with group selection, Clifford regarded act utilitarianism as the normative theory that encapsulated the dictates of conscience as they had been received by his own community. He described the theory – where ‘the end of right action is defined to be the greatest happiness of the greatest number’ (Clifford 1879b: 173) – as the ‘moral system that has deservedly found favour with the great mass of our countrymen’ (Clifford 1879b: 173). The phrase ‘deservedly found favour’ indicates that Clifford approved of act utilitarianism and the phrase ‘with the great mass of our countrymen’ indicates that he believed that act utilitarianism predominated in the United Kingdom at the time. The essay in which Clifford expressed these opinions immediately precedes ‘The Ethics of Belief’ in his Lectures and Essays and its final paragraphs are a segue to the latter.

Clifford believed that act utilitarianism was a good fit with his evolutionary view that the community developed a normative morality for its own benefit, writing that ‘the reason and the ample justification of the success of [act utilitarianism] is that it explicitly sets forth the community as the object of moral allegiance’ (1879b: 173). In Clifford's opinion, the requirement of act utilitarianism that we pursue ‘the greatest happiness of the greatest number’ is an affirmation that our duty is to our community. However, to make utilitarianism fit with his evolutionary views, Clifford denied that happiness is the goal, maintaining instead that we are to serve the community, that the goal of morality is to make us better servants, and that the happiness of others and ourselves is a by-product of our service. He may have been inspired by the paradox of hedonism. Henry Sidgwick – whom ‘the scholarly consensus does for the most part classify … as an act utilitarian’ (Schultz 2020) – was the first to point out the paradox, and it seems likely that Clifford learned of it from him and adapted the insight to his own purpose. Clifford certainly approved of Sidgwick's Methods of Ethics (1874), describing it as ‘an admirable book’ (Clifford 1879b: 161). He did not laud any other work in ethics in the same manner.

Given his evolutionary views, his belief that they were a good fit with utilitarianism, and his belief that utilitarianism was the normative theory of his audience and his own community, Clifford probably felt bound to be an act utilitarian. The internal evidence in ‘The Ethics of Belief’ confirms that he took an act utilitarian approach, except, of course, that he was dealing with the consequences of beliefs instead of acts. His belief in group selection does not matter, because his argument is logically independent of evolutionary selective processes. We can also ignore his notion that our happiness is a by-product of service to the community. In his argument, he actually assumed only the orthodox view that the interests of everyone affected have to be considered.

Clifford's conclusion is that it is always wrong to believe anything without sufficient evidence. He started making his case for it by giving two examples. In the first example, he described a shipowner who overcame doubts about the seaworthiness of his vessel and sent it forth full of emigrants, who drowned when it sank. In the second, he described a group agitating against people whom they believed, without justification, to have used underhanded methods to indoctrinate children. He emphasised that it is the belief and not just the action that follows the belief that must be evaluated. Clifford explained that ‘the reason of this judgment is not far to seek: it is that in both these cases the belief held by one man was of great importance to other men’ (1879a: 182). The reason the belief of one person was important to other people is obviously that the belief was acted upon by the believer and thereby had adverse effects on the others.

Clifford maintained that the shipowner and the agitators would have done wrong even if the ship had completed its voyage without incident and even if the accusations had been true. In these two cases, if the over-believers had not actually harmed anyone, they still failed to take any steps to ensure that the risk of harm was minimised. A duty to maximise happiness necessitates taking steps to minimise the risk of harm to others, everything else being equal; it is morally negligent or reckless to fail to take steps to minimise it. Since Clifford asserted that the shipowner and the agitators would have done wrong even if their beliefs had turned out to be true, the cases do seem to be examples of negligently or recklessly over-believing.

The shipowner seems reckless rather than merely negligent; ‘inasmuch as he had knowingly and willingly worked himself into that frame of mind, he must be held responsible for it’ (Clifford 1879a: 178). But Clifford condemned all over-belief and not just those instances that involve the knowing and willing rationalisation of self-interested expediency. The special characteristics of the shipowner case do not mean that Clifford's generalisation is too broad, however, because other examples could replace it without altering the overall structure or thrust of the argument. We could refer instead to the desperate parents who believe the hype about the industrial bleach that is marketed as a ‘Miracle Mineral Supplement’ (Porter 2018) and who want to benefit their autistic children by making them drink it. In the suggested replacement example, the objectionable element is the parents’ belief, full stop, their desire to benefit their children being laudable.

Clifford generalised from his two examples. He added that no belief ‘is actually insignificant or without its effect on the fate of mankind, [so] we have no choice but to extend our judgment to all cases of belief whatever’ (Clifford 1879a: 182). The idea that literally every belief has an effect on the fate of humanity indicates again that Clifford's argument is similar to those advanced by act utilitarians. The subsequent paragraphs are intended to support the claim that no belief is ‘without [adverse] effect’. In every case, it is consequences that are mentioned. In every case, it is the consequences of believing in particular instances. Specifically, Clifford claimed that ‘every time we let ourselves believe for unworthy reasons, we weaken our powers of self-control, of doubting, of judicially and fairly weighing evidence’ (1879a: 185). In other words, if nothing else, people who over-believe render themselves less capable of responsibly taking steps to avoid over-belief when acquiring other beliefs thereafter. Clifford contended that if society became generally credulous, it would ‘lose the habit of testing things and inquiring into them; … [and] sink back into savagery’ (1879a: 185–86).

A plausible mechanism for what we can call Cliffordian credulity is that we tend to need what we take to be evidence for what we believe, that over-believing necessitates altering our standards of evidence to accommodate the over-believed proposition, and that our altered standards permit us to acquire other over-beliefs willy-nilly (Zamulinski 2004). In other words, over-believing tends literally to ‘weaken our powers … of judicially and fairly weighing evidence’ by tending to alter our standards of evidence for the worse. It is bad enough that some false beliefs cause people to do wrong but this kind of credulity is worse because it opens the doxastic floodgates. Consequently, the effects of the original over-belief are sometimes augmented by the effects of other over-beliefs that slip through altered standards of evidence. The former are the primary adverse effects; the latter are secondary.

Clifford seems to have assumed that there is no way for us to do anything to prevent the adverse consequences besides avoiding over-believing. There will be an argument for the contention in connection with the case for a Cliffordian rule, but Clifford himself did not argue for it, apparently taking it to be obvious. In view of the assumption and of the potential consequences, both the primary ones connected to a particular over-belief and the secondary ones that result from beliefs permitted by altered standards of evidence, it does appear to be morally negligent or reckless to over-believe. It follows that there are utilitarian reasons for Clifford's universal generalisation: ‘To sum up: it is wrong always, everywhere, and for anyone, to believe anything upon insufficient evidence’ (1879a: 186).

Clifford devoted the two subsequent sections of ‘The Ethics of Belief’ to arguing that we can accept the testimony of others and that we can go beyond our immediate experience. According to Clifford, we can do the former when we have evidence that the individuals on whose testimony we rely are neither dishonest nor deluded, that they have the expertise or experience necessary to discover what they claim to have discovered, and that they have used their expertise appropriately or interpreted their experience reasonably (1879a: 189). And, he allowed that we can make inferences that presuppose the uniformity of nature (1879a: 206). In other words, he responded to the anticipated objections that his doctrine is unreasonably restrictive because it prevents us from sharing our knowledge and from understanding our world scientifically. His responses indicate again that he was concerned with consequences. At the end of the third section, Clifford reiterated that ‘it is wrong in all cases to believe on insufficient evidence’ (1879a: 211).

While it is reasonable to regard the argument in ‘The Ethics of Belief’ as a doxastic counterpart to an act utilitarian argument, it has recently been claimed that Clifford's conclusion is merely a ‘rhetorical flourish’ (Nottelmann and Fessenbecker 2019). It is an unconvincing contention because, asserting that Clifford was really discussing blameworthiness, its proponents rely not on an interpretation of his arguments but on his supporting examples, concentrate on a subset of the examples without justifying the selection, and, despite having selected the examples for discussion, still have to discount some of them to save their hypothesis. But, even if their position were plausible, it would still be possible to construe Clifford's essay as a utilitarian argument for the conclusion that it is wrong to believe propositions without possessing sufficient evidence for them. I will now underscore the act utilitarian interpretation by examining, and rejecting, some non-utilitarian interpretations.

The epistemic interpretation

Clifford argued that over-believing is wrong because it leads to harm or the risk of harm, which necessitate that we make efforts to prevent them, and to Cliffordian credulity, a propensity to over-believe in other cases that can thereby amplify the harm and risk of harm. Since these are moral reasons, his conclusion is that it is morally wrong to believe without sufficient evidence. In contrast, the epistemic interpretation is that the ethics of belief is about acquiring true beliefs and avoiding false ones, which means that we should avoid believing without sufficient evidence because it results in our believing falsely, full stop. Proponents of the purely epistemic interpretation do not add the Cliffordian proposition that believing falsely is morally wrong because it sometimes leads to harmful actions. Characterising Clifford's ethics of belief as an epistemic doctrine poisons the well because errors, like ignorance, often appear tolerable when they are considered merely intellectually. Some are even amusing. Switching from moral to intellectual issues suggests that there is something unreasonable about Clifford's position, which reduces the probability that it will be taken seriously.

William James (1896b) purported to address Clifford's position in the course of setting out his own. Near the end of his essay, James declares that ‘we have the right to believe at our own risk any hypothesis that is live enough to tempt our will’ (1896b: 33; italics added). But while James made efforts to show that over-believing can sometimes result in true beliefs, he did nothing to demonstrate that we can ever do so at our own risk, that is, without putting others at risk. Since Clifford was concerned with the consequences for others, James failed to come to grips with his position. The best explanation for James's failure is that he mistook Clifford's position for an epistemic one. The fact that his criticisms of Clifford are epistemic in nature supports the inferred explanation. James certainly tolerated errors and treated Clifford's views as unreasonable: ‘Clifford's exhortation has to my ears a thoroughly fantastic sound. … Our errors are surely not such awfully solemn things’ (1896b: 26).

James's own default position was epistemic. He declared that ‘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). He adds: ‘Wherever the option between losing truth and gaining it is not momentous, we can throw the chance of gaining truth away, and at any rate save ourselves from any chance of believing falsehood by not making up our minds at all till objective evidence has come’ (1896b: 26–27; italics in original). But, as is well known, James went beyond the epistemic view when he contended that, in hopes of acquiring true beliefs, we may proceed pragmatically and over-believe when certain conditions are met. He argued that the gamble is permissible when the option of believing is a genuine option (James 1896b: 21), that is, when it is live, forced, and momentous (James 1896b: 15–16). An option is live when some ‘hypothesis makes some appeal, however small, to your belief’ (James 1896b: 15); forced when suspending belief is tantamount to disbelieving; and momentous when it is important enough to the prospective believer.

But, to imagine Clifford's response, consider Clifford's shipowner again. The option of believing his ship to be seaworthy would certainly have been live, because it concerned a real possibility for him. It would have been forced, because doubts about the seaworthiness of his ship would have kept it from sailing just like the belief that it was not seaworthy. ‘Either sail as scheduled or delay the voyage’ is parallel to James's own example of a forced option, ‘Either accept this truth or go without it’ (1896b: 16). The two are both forced because there is no third option that enables us to avoid doing one or the other, which is the feature that James regarded as characteristic of forced options. Finally, it could have been momentous if, say, the survival of his shipping business depended on the revenue from the voyage proceeding as scheduled. Obviously, satisfying James's criteria would not obviate the risk to passengers or crew. Clifford would have still condemned the shipowner. The fact that some over-beliefs turn out to be true, which possibility Clifford acknowledged in connection with the shipowner example, is not enough to justify overturning Clifford's prohibition of over-believing. It would still be morally negligent or reckless even if we were sometimes lucky.

Some interpret Clifford's position as primarily rather than purely epistemic. In his entry on the ethics of belief in the Stanford Encyclopedia of Philosophy, which recapitulates an earlier argument (Chignell and Dole 2005), Andrew Chignell contends that ‘Clifford … claim[s] that the issue of whether we have done our doxastic best is an epistemic one and also (given a few further premises) a moral one’ (Chignell 2018; italics in original). But Clifford did not first argue that we must seek truth and avoid error, and only afterward add that it was morally wrong to do otherwise. His argument involves a reference to the epistemic concept of sufficient evidence but it is a moral argument throughout.

The non-consequentialist interpretation

While it certainly looks as though Clifford's argument is like an act utilitarian one, some have characterised it as non-consequentialist (Vorstenbach 1999). Some even maintain that Clifford switches from consequentialism to non-consequentialism (Madigan 1997). Proponents of the non-consequentialist interpretation take as indicating that stance the statements that ‘when an action is once done, it is right or wrong for ever; no accidental failure of its good or evil fruits can possibly alter that’ (Clifford 1879a: 178) and that ‘a bad action is always bad at the time it is done; no matter what happens afterwards’ (Clifford 1879a: 185).

But Clifford's two statements do not entail that he is not an act utilitarian. Since Clifford spoke of ‘accidental failures’ and used the phrase ‘no matter what happens afterwards’, Clifford apparently had in mind a world in which outcomes were matters of probability. In such a world, an act utilitarian could say that we always have a duty to perform the act that is most likely, to the best of our knowledge at the time of its performance, to have the best consequences – even if it does not actually produce those consequences. In other words, an act utilitarian could be a prospectivist, maintaining that it is the rationally anticipated consequences of acts that are the sole determinants of our duty. For instance, a physician would have a duty to prescribe the course of treatment that was most likely to result in the patient's recovery, given the evidence. It would still be the right thing to do even if the patient did not recover – and patients would not always recover. It follows that the moral status of the physician's actions would be established at the time they performed them, no matter what happened afterward. Hence, since Clifford's statements could be true in a probabilistic world in which a prospectivist version of act utilitarianism was true, they do not entail the truth of a non-consequentialist interpretation.

The stoic interpretation

Scott Aikin (2014) contends that Clifford had a stoic view of duty. The stoic view, according to Aikin, is that people have roles and that they have duties in virtue of having the roles. It would have been a type of view familiar to Victorians, who attributed duties to people in connection with their stations in life. Aikin maintains that Clifford held that people have a role that encompasses a duty to avoid believing without sufficient evidence. Clifford did laud stoicism, declaring that ‘one of the most important expressions of the moral sense for all time is that of the Stoic philosophy, especially after its reception by the Romans’ (1879c: 133). But there is no evidence that Clifford imported its doctrines into ‘The Ethics of Belief’. There is no indication in that essay itself that he appealed to the notion of duties attached to roles to make his case. When he did refer to roles, he merely denied that the duty to avoid over-believing was limited to specific role-holders: ‘It is not only the leader of men, statesman, philosopher, or poet, that owes this bounden duty to mankind’ (1879a: 183). Aikin points out that Clifford's first two examples resemble examples from Cicero (Aikin 2014: 45–46), but the context of discovery – or inspiration – is not the context of justification.

Of course, with Clifford's emphasis on service to the community, everyone can be seen as occupying the role of servant of the community. The philosophical difficulty is that it is impossible to derive the duties that come with such a role from the mere fact that the role exists. In order to justify the contention that a servant of the community had a particular duty to the community, it would be necessary to show that fulfilling the duty would actually benefit the community. Therefore, even granting that Clifford's ultimate goal was to establish that we have a role of the sort spoken of by the Stoics, it could still be the case that his goal in ‘The Ethics of Belief’ was just the proximate one of proving the existence of a duty not to over-believe, a proximate goal that could be achieved by applying an act utilitarian approach to beliefs. Obviously, having established a duty not to over-believe by adverting to the consequences, Clifford could have added that a good servant of the community would fulfil the duty – but the addition would do no philosophical work. Aikin acknowledges that ‘stoic ethics integrates elements of consequentialist reasoning to identify duties’ (2014: 46). But consequentialist reasoning does not have to be associated with stoicism, and Clifford did not require anything else.

Counterexamples

There are a number of putative counterexamples in the literature. There is the person who improves their chances of recovery by over-believing that they will get better (Code 1987); the person who preserves their relationship with another individual by over-believing charitably about the latter's activities (Meiland 1980); the person who is able to achieve an aim under uncertainty by over-believing that they can succeed (James 1896a; Marusic 2011); and the person who is made happy by the over-belief that the sports team they support is better than any other (Ward 1878).

It is doubtful that any of the putative counterexamples succeeds in falsifying Clifford's universal generalisation. As mentioned earlier, there are not only the effects of a specific over-belief – the primary adverse effects – but also the consequences of Cliffordian credulity, changes in our standards of evidence that result in our willy-nilly acquiring other over-beliefs that can have secondary adverse effects. Clifford believed that there could be secondary as well as primary adverse effects and that we must consider both. So, a good counterexample would be one in which someone believed a proposition, in which believing it had neutral or beneficial effects, and in which there were neither primary nor secondary adverse effects or in which the adverse effects were outweighed by the beneficial ones. The problem with the counterexamples mentioned in the previous paragraph is that they do not clearly show that there are no adverse effects; and if it is obvious that primary adverse effects are improbable, it is not equally obvious that the same is true of secondary ones.

The following is the kind of counterexample that would definitely falsify Clifford's universal generalisation. Consider someone who lacks evidence for a true belief about what constitutes good evidence. Specifically, consider a situation in which someone comes to believe that an informal fallacy is fallacious not by having it explained to them but by having it drilled into them. Subsequently, as a result, they recognise and reject instances of the fallacy despite not understanding why it is a fallacy. They would therefore be less likely to believe certain things on the basis of insufficient evidence, which would be good, but they would lack sufficient evidence for their conviction that the fallacy was a fallacy. Furthermore, there would be no secondary adverse effects, particularly if, in reaction to the drill, they also acquired an aversion to having other things drilled into them.

If we discovered an actual instance of the foregoing type of counterexample, we would have to reject Clifford's universal generalisation as false. Even in the absence of an actual example, however, it is plausible that there have been such cases and it is plausible that we would find them if we searched. So, even though the example in the preceding paragraph is fictional, it shows that Clifford's universal generalisation is in a precarious position. Clifford would have done better to argue that there is a justification for a rule against over-believing.

Justifying a Cliffordian rule

If prospectivism is true, then it is possible to base an argument for Cliffordianism on it. As mentioned, prospectivism is the view that we should do what is best given the evidence we possess. It is more plausible than objectivism. Consider a medical case (Jackson 1991) in which, as a matter of fact, the first course of treatment would be a complete cure, the second would be somewhat beneficial, prolonging life of a reasonable quality, and doing nothing at all would result in a lingering death. But, in the same case, the evidence obtained shows that the first course of treatment would be either a complete cure or immediately fatal, that the second would be somewhat beneficial, and that doing nothing would result in a lingering death. In these circumstances, even though the first course of treatment is objectively best, it is the second that we should choose. The thought experiment is convincing. The remainder of this section shows how it is possible to argue for Cliffordianism on this basis.

If we have an obligation to do what is best in light of the evidence we possess, we have an obligation not to believe without sufficient evidence anything that is relevant at present or that would be relevant in the future. The claim in the consequent is supported by the claim in the antecedent. If we believed without sufficient evidence anything that is relevant at present or that would be relevant in the future, then we would contaminate the evidence we possess at present or would possess in the future. If we did not fail to do what is best, it would only be by chance that we succeeded and lucking out is not the same as acting responsibly.

If we have an obligation not to believe without sufficient evidence anything that is relevant at present or would be relevant in the future, then, when we believe a proposition, we need either sufficient evidence for the proposition itself or sufficient evidence that it will never be relevant. If we could eliminate the second disjunct in the consequent, we would be left with Cliffordianism.

We can eliminate the second disjunct. We lack the ability to acquire sufficient evidence that a proposition will never be relevant. Checking the consequences of believing a proposition would usually require the examination of an indeterminate number of possible futures, each for an indeterminate length of time into the future, and we could not do that even if we wanted to. In the alternative, if it were possible, it would still impose such a burden as to make it impractical to over-believe in an ethical way.

It gets worse. While it might appear that there are cases in which we can readily establish that there are no relevant adverse effects, it appears that there are none only if we limit ourselves to a consideration of the primary effects of a particular over-belief. The trouble is that, as indicated earlier, over-believing can sometimes alter our standards of evidence (because we need it to appear that we have evidence for the original over-belief), the altered standards would sometimes permit additional over-beliefs, and sometimes the additional beliefs would have secondary adverse effects. If over-believing is sometimes accompanied by altered standards of evidence, it is impossible to predict the other beliefs that they would permit us to acquire, let alone their effects. It is therefore impossible to prove that an over-belief would never have relevant secondary effects – unless we have the ability to believe purely at will.

We do not have the ability to believe purely at will. It is like levitation or psychokinesis: it has never been observed or demonstrated, and it therefore cannot be taken seriously as a real possibility.

Furthermore, there is a twofold evolutionary explanation for why we cannot do it. The first part of the explanation starts with the observation that we have a naturally selected ability to acquire beliefs. We would not have such an ability unless the beliefs it enabled us to acquire tended to be advantageous. But, at the time we acquire them, it is often unknowable whether beliefs will be advantageous. Consequently, we need a proxy for being advantageous and the only plausible proxy is truth. In other words, the beliefs we acquire tend to be fitness-tracking because they tend to be truth-tracking. We also need a proxy for truth and the only plausible proxy is what we take to be sufficient evidence. In other words, acquired beliefs tend to be truth-tracking because they tend to be evidence-tracking. Acquiring advantageous beliefs is our ultimate goal, acquiring true beliefs is our penultimate goal, and acquiring beliefs supported by what we take to be sufficient evidence is our proximate goal (see Mayr 1961). To put it another way, since we have a naturally selected ability to acquire beliefs, we also have a naturally selected disposition to prefer truths, and, since we have a naturally selected disposition to prefer truths, we have a naturally selected disposition to seek what we take to be evidence. Since we have a naturally selected disposition to seek what we take to be evidence, if our will has an influence on our beliefs, its influence will often be accompanied by alterations to our standards of evidence to make it appear, at least to ourselves, that we have evidence for what we have willed to believe. As for the second part of the explanation, believing purely at will is incompatible with reliably achieving the ultimate, penultimate, or proximate goals. Consequently, it is liable to be seriously disadvantageous and ‘strongly inadaptive features hold little prospect for an evolutionary legacy because natural selection must soon eliminate them’ (Gould 2002: 1247; italics in original). In other words, the lineages of organisms with the ability to believe purely at will would die out while the lineages of organisms that did not have the ability would endure. The combination of a naturally selected disposition to seek what we take to be evidence and the disadvantageous nature of an ability to believe purely at will explains why believing purely at will is as implausible as levitation or psychokinesis.

Some maintain that ‘natural selection does not care about truth; it cares only about reproductive success’ (Stich 1990: 62). But it would not care about truth only if all beliefs were adaptations, and acquired beliefs are not adaptations. The position here, expressed in the same terminology, is that natural selection cares about truth because it cares about reproductive success. Natural selection has to care about truth because truth is the only plausible proxy for the property of being fitness-promoting when it comes to many beliefs. Like other biological features, however, the hypothesised arrangement for the acquisition of beliefs does not have to work perfectly. It only has to work well enough and often enough to be advantageous enough to be selected for. Consequently, the arrangement is compatible with a fair number of errors. The example Stich uses to make his point involves overgeneralising about plants being poisonous and therefore off the menu, but the resulting errors surround a core of truths. His example may show that some errors will not decrease fitness significantly in a particular environment; it does not show that truths will not increase fitness in the same environment, that they will not increase it in other environments, or that there is an alternative to relying on truth.

Since we do not have the ability to believe purely at will, we cannot show that over-beliefs will never be relevant. If we should always act prospectively, we therefore have no option but always to try to avoid over-believing. This is Cliffordianism. The only possible objection is that we would lose too much by adhering to a rule against over-believing. The objection fails. There are beneficial over-beliefs as well as harmful ones, but the gains from believing without sufficient evidence are outweighed by the losses. For instance, the benefits of the putative counterexamples mentioned in the previous section do not come close to counterbalancing the harm caused by inadequately supported beliefs like the convictions that COVID-19 is a hoax or that anthropogenic global heating is a myth. Therefore, since prospectivism appears to be true and since the potential losses from over-believing are far greater than the gains from indulging in it, we should follow a rule against believing without sufficient evidence.

Conclusion

Understanding Clifford's argument as a doxastic counterpart to an act utilitarian one is essential to understanding historically why, and philosophically how, he tried to justify his contention that it is wrong to believe without sufficient evidence. But, to reach Clifford's conclusion, his universal generalisation needs to be replaced by a rule against believing without sufficient evidence. In light of the rule, it is morally negligent or reckless to believe without sufficient evidence – no matter who, no matter where, no matter when.