Characterising Hume’s response to scepticism in THN 1.4.7 is a notoriously fraught endeavour.[1] In this paper, I offer a novel interpretation of THN 1.4.7, which sees his sceptical problem and solution in THN 1.4.7 as taking a broadly deontological structure. Briefly, I read the ‘Dangerous Dilemma’ (THN–7) as embodying a false dichotomy between two deontological extremes concerning reflection, that is, thinking carefully about our mental states and faculties.[2] The two horns of the Dangerous Dilemma are as follows: either embracing an absolute duty to constantly and incessantly reflect (leading to excessive scepticism); or maintaining that it is not the case that we have any duty to reflect to any degree (leading to credulity). Hume thus seeks to straddle these two horns and find a deontological middle path. The resolution to this dilemma turns on Hume’s realising that we have a duty to reflect only up to a point. Beyond this threshold, there is a level of reflection that is not required of us, but which is nevertheless good; in other words, such reflection is supererogatory. However, this seems to render excessive scepticism supererogatory. This unintuitive outcome can be avoided by appealing to a suitable account of value beyond the deontological threshold that is founded on usefulness and agreeableness. In the end, Hume manages to tread a path between scepticism and credulity, while nevertheless rejecting superstition and endorsing science and philosophy.

The Supererogatory and Hume’s Ethics

Before I examine Hume’s epistemology, I will first briefly argue that he recognises a class of actions which are supererogatory in his ethics. Given this, a reading of THN 1.4.7 that turns on the supererogatory is not implausible.

How to characterise supererogation, and if such a category even exists, are difficult issues.[3] This paper is not a paper on ethics (nor even primarily on Hume’s ethics!), and I cannot delve deeply into the issue here. For the purposes of my paper, I follow Urmson’s (1958) seminal work, which challenges the traditional trichotomy of the obligatory, the permitted, and the prohibited. The key claim is that ‘there are many kinds of action that involve going beyond duty proper, saintly and heroic actions being conspicuous examples of such kinds of action’ (Urmson 1958: 215). While Urmson does not use the term ‘supererogation’ nor its cognates, we may put his point as follows: there is a class of supererogatory actions such that they are not are not required of us as a duty, but are nevertheless good to do.

It might strike some readers as peculiar to attribute the notion of supererogation to Hume, as his moral theory is nowadays typically seen as taking on a virtue-ethical cast, free from deontological trappings. For instance, Taylor states that

Hume’s moral philosophy may plausibly be construed as a version of virtue ethics. Among the central concepts of his theory are character, virtue and vice, rather than rules, duty, and obligation. (Taylor 2006: 276)[4]

If Hume does not recognise duties, there would trivially be a great deal of supererogatory actions in his ethics: all good actions would be good without being required of us as duties. I will briefly argue that Hume does recognise a non-trivial class of supererogatory actions in his ethics. In doing so, I only make the limited claim that even if his moral framework by and large takes the form of a virtue ethics, it nevertheless leaves room for duties.[5]

First, Hume clearly recognises a class of actions which are required as obligations. See the following passage from ‘Of the obligation of promises’:

All morality depends on our sentiments, and when any action, or quality of the mind, pleases us after a certain manner, we say it is virtuous; and when the neglect, or non-performance of it, displeases us after a like manner, we say that we lie under an obligation to perform it. (THN

The above passage indicates that there are obligations or duties, which rest on sentiments (in keeping with Hume’s sentimentalist frameworks).[6]

Of course, this does not yet establish that there are supererogatory actions, since it might be that all good actions are required of us as duties. In this respect, the following passage is significant:

No action can be requir’d of us as our duty, unless there be implanted in human nature some actuating passion or motive, capable of producing the action. (THN

Hume above claims that only actions such that human nature in general contains a motive or passion capable of producing it can be required of us as duties. From here, it is a short walk to the claim that he recognises a class of actions as supererogatory.

In his discussion on the indirect passions, Hume clearly states that we can only be proud or humble of qualities that are relatively peculiar to us (THN The idea is that it is difficult to be particularly proud or humble of a very common trait, and so some degree of rarity is required to stir these passions. Of course, pride and humility, when concerning our mental qualities, are intimately related to moral approbation and blame (THN For Hume, actions are laudable or blameable on the basis of the character traits they reveal (THN; THN Thus, any action that indicates a quality of the mind that is useful or agreeable as well as rare will be an action that we can take pride in, as well as one that receives moral approbation.[7] However, there is no actuating motive capable of producing these actions present in the generality of mankind, given the rarity of these underlying traits.[8] In short, an action that indicates a useful or agreeable and also rare trait will be supererogatory.

Moreover, it should be noted that Hume’s characterisation of duties seems to leave ample room for the supererogatory. An action is virtuous if it ‘pleases us after a certain manner’, and it is obligatory if its neglect ‘displeases us after a like manner’ (THN By lights of this account, an action would be virtuous without being required as a duty if it pleases us in the relevant manner, without its neglect causing us displeasure. Hume’s account of duties thus clearly leaves open room for the supererogatory in his ethics.

The Dangerous Dilemma

The nexus of Hume’s consternation in THN 1.4.7 is undoubtedly the ‘dangerous dilemma’.[9] The first horn of the dilemma (henceforth ‘the credulity horn’) is as follows:

For if we assent to every trivial suggestion of the fancy; beside that these suggestions are often contrary to each other; they lead us into such errors, absurdities, and obscurities, that we must at last become asham’d of our credulity. Nothing is more dangerous to reason than the flights of the imagination, and nothing has been the occasion of more mistakes among philosophers. (THN

Essentially, Hume here cautions us against trusting ‘every trivial suggestion of the fancy’, since such a course of action would undermine reason and lead us to ‘credulity’. The other horn of the dilemma (henceforth ‘the sceptical horn’) is no better:

But on the other hand, if the consideration of these instances makes us take a resolution to reject all the trivial suggestions of the fancy, and adhere to the understanding, that is, to the general and more establish’d properties of the imagination; even this resolution, if steadily executed, wou’d be dangerous, and attended with the most fatal consequences. For I have already shewn, that the understanding, when it acts alone, and according to its most general principles, entirely subverts itself, and leaves not the lowest degree of evidence in any proposition, either in philosophy or common life. We save ourselves from this total scepticism only by means of that singular and seemingly trivial property of the fancy, by which we enter with difficulty into remote views of things, and are not able to accompany them with so sensible an impression, as we do those, which are more easy and natural. (THN

The sceptical horn has its roots in Hume’s ‘Of scepticism with regard to reason’ in THN 1.4.1. Briefly, Hume’s worry stems from the fact that we seem to be required to continually make higher-order judgments on the reliability of our judgments. Each iteration gradually undermines our beliefs until nothing remains. Hume thinks we only escape this fate because of the ‘trivial property of the fancy’ (THN that is the psychological difficulty we face in engaging with overly complex reflection and reasoning.

In this section, I will interpret the Dangerous Dilemma as embodying a false dichotomy between two deontological extremes: either it is not the case that we have any duty to reflect whatsoever (engendering the credulity horn), or we have an absolute duty to constantly and incessantly reflect (engendering the sceptical horn).[10] I proceed to argue that Hume straddles the horns of the Dangerous Dilemma by recognising that there is a level of reflection that is supererogatory.

There is some prima facie textual evidence for thinking that the Dangerous Dilemma might have a deontic nature in THN 1.4.7. Hume characterises the central problem of the Dangerous Dilemma as the question of ‘how far ought we to yield to these illusions’:

… the question is, how far we ought to yield to these illusions. This question is very difficult, and reduces us to a very dangerous dilemma, which-ever way we answer it. (THN

And upon initially failing to find a satisfactory resolution to it, he remarks, ‘For my part, I know not what ought to be done in the present case’ (THN

However, the above is relatively thin substantiation for reading the Dangerous Dilemma as taking deontic shape; putting that aside, the deontic nature of the Dangerous Dilemma would still be woefully underdetermined. Thus, we will need to more closely examine both horns of the Dangerous Dilemma—the sceptical horn (which has its roots in THN 1.4.1), and the credulity horn (which I will argue has its roots in Hume’s discussion of general rules in THN 1.3.13). However, before examining the horns themselves, I will first examine Hume on the ‘second influence of general rules’ (THN in order to lay the groundwork for these interpretive claims. This might surprise readers, since it is not immediately obvious that the Dangerous Dilemma turns on this discussion. Nevertheless, I motivate that there is such a link, and this will prove vital in showing that the Dangerous Dilemma takes on a deontological form, concerning duties of reflection.

Hume distinguishes between ‘philosophical’ and ‘unphilosophical’ probability; the former kinds of probability (which encompass the probabilities of chances and causes) are ‘allow’d to be reasonable foundations of belief and opinion’, while the latter kinds of probability ‘have not had the good fortune to obtain the same sanction’ (THN We might characterise unphilosophical probabilities as those whereby our degree of belief is affected by factors which are epistemically irrelevant, such as recentness of observation (THN, or length and complexity of argument (THN One important species of unphilosophical probability turns on the first influence of general rules (henceforth ‘first influence’), which are general rules for judgments that, in generalising our probable judgments, conflate the relevant and the irrelevant circumstances:

When an object appears, that resembles any cause in very considerable circumstances, the imagination naturally carries us to a lively conception of the usual effect, tho’ the object be different in the most material and most efficacious circumstances from that cause. Here is the first influence of general rules. (THN

This is in turn corrected by the second influence of general rules (henceforth ‘second influence’):

But when we take a review of this act of the mind, and compare it with the more general and authentic operations of the understanding, we find it to be of an irregular nature, and destructive of all the most establish’d principles of reasonings; which is the cause of our rejecting it. This is a second influence of general rules, and implies the condemnation of the former. (THN

The second influence ensures that we adequately discern the relevant from the irrelevant circumstances:

We shall afterwards take notice of some general rules, by which we ought to regulate our judgment concerning causes and effects; and these rules are form’d on the nature of our understanding, and on our experience of its operations in the judgments we form concerning objects. By them we learn to distinguish the accidental circumstances from the efficacious causes; and when we find that an effect can be produc’d without the concurrence of any particular circumstance, we conclude that that circumstance makes not a part of the efficacious cause, however frequently conjoin’d with it. (THN

The instances of general rules referred to above are simply the ‘Rules by which to judge of causes and effects’, explained in THN 1.3.15, to which Hume refers his readers in a footnote.

Importantly, note that the second influence is intimately related to reflection.[11] When acting on judgments caused by the first influence, the ‘second influence’ involves taking ‘a review of this act of the mind’, comparing it ‘with the more general and authentic operations of the understanding’, consequently dismissing it (THN This ‘review’ is clearly an act of reflection; see Hume’s claim that ‘a like reflection on general rules keeps us from augmenting our belief upon every encrease of the force and vivacity of our ideas’ (THN Hume also notes that reflection can correct for the first influence of general rules:

We may correct this propensity by a reflection on the nature of those circumstances; but ’tis still certain, that custom takes the start, and gives a biass to the imagination. (THN

Hume also emphasises that the second influence involves forming higher-order judgments which are ‘deriv’d from the nature of the understanding’ rather than ‘from the nature of the object’—in short, the second influence involves introspective reflection on our mental states and faculties. With this in mind, we can now see why the Dangerous Dilemma embodies a false dichotomy between embracing an absolute duty to exhaustively reflect, and a denial of any duty to reflect whatsoever.

Let us begin by examining the sceptical horn. It is widely recognised in the secondary literature that this horn of the Dangerous Dilemma stems from THN 1.4.1.[12] However, what has been little noted is the sceptical argument contained in this section arises from distinctly deontological considerations.[13] The core of Hume’s problem here obviously stems from the seeming requirement that we always make higher-order judgments regarding the reliability of our judgments and faculties. Indeed, without such a duty, it is difficult to see why Hume’s doomsday scenario is compelling at all—so what if higher-order reflection undermines our beliefs, if it is not the case that we ought to engage in such higher-order reasoning in the first place? Without any genuine requirement to engage in higher-order reflection, the sceptical horn remains little more than an idle curiosity.

In line with this, it turns out that Hume’s discussion of this issue is indeed couched in distinctly deontological terms. In arguing that knowledge degenerates into probability, Hume claims that we ‘must’ reflect on the fallibility of our understanding:

We must, therefore, in every reasoning form a new judgment, as a check or controul on our first judgment or belief …. (THN

And with respect to judgments of probability, Hume says that ‘we ought always’ to perform higher-order reflection on the reliability of the lower-order judgments:[14]

In every judgment, which we can form concerning probability, as well as concerning knowledge, we ought always to correct the first judgment, deriv’d from the nature of the object, by another judgment, deriv’d from the nature of the understanding. (THN

Moreover, Hume also describes such a course as required by ‘all the rules of logic’:

… all the rules of logic require a continual diminution, and at last a total extinction of belief and evidence. (THN

Thus, a requirement to incessantly engage in higher-order reflection engenders the sceptical horn.[15]

An example of such obligations to reflect might be Hume’s own ‘Rules by which to judge of causes and effects’ in THN 1.3.15, which are an importance instance of the second influence of general rules. These seem like a salient candidate for the ‘rules of logic’ that engender ‘a total extinction of belief and evidence’ (THN he says of these rules, ‘here is all the Logic I think proper to employ in my reasoning …’ (THN[16] Thus, the sceptical horn of the Dangerous Dilemma seems tied to the second influence of general rules.

That the problematic higher-order judgments described in THN 1.4.1 have their roots in the second influence can also be verified by Hume’s characterisation of this reflection: these higher-order judgments are formed ‘as a check’ (THN on our beliefs, in the same way the second influence checks the first influence, as detailed in THN Moreover, unlike the first judgment, these higher-order judgments are ‘deriv’d from the nature of the understanding’, rather than ‘from the nature of the object’ (THN This seems to correspond to Hume’s description of his ‘Rules by which to judge of causes and effects’ as being ‘form’d on the nature of our understanding, and on our experience of its operations in the judgments we form concerning objects’ (THN

Putting this together, on Hume’s view, if we embrace the view that we have an absolute duty to continually and incessantly reflect in accordance with the second influence (as per THN 1.4.1), this directly leads to the extinction of our beliefs, that is, the sceptical horn.

On the other hand, we can see the credulity horn as embodying the rejection of any duty to reflect to any degree. This is a natural reading, both by contrast with the sceptical horn, and also more directly, from a straightforward examination of Hume’s discussion of credulous belief in THN 1.3.13. Hume undoubtedly sees a lack of reflection as engendering ‘errors, absurdities, and obscurities’ and ‘credulity’ (THN, since without the second influence, we have nothing with which to correct our unphilosophical probabilities. It is through his rules by which to judge of causes and effects that ‘we learn to distinguish the accidental circumstances from the efficacious causes’ (THN[17] More strongly,

The following of general rules is a very unphilosophical species of probability; and yet ’tis only by following them that we can correct this, and all other unphilosophical probabilities. (THN

In short, if we reject the requirement that we reflect in accordance with the second influence, we are left with no means by which to check the credulous beliefs arising from the first influence.[18]

Thus, a lack of reflection clearly leads to ‘errors, absurdities, and obscurities’ (THN Indeed, Hume’s discussion of the credulity horn has clear correlates with his discussion of the first influence. Examine:

Nothing is more dangerous to reason than the flights of the imagination, and nothing has been the occasion of more mistakes among philosophers. (THN

In discussing the first influence, Hume remarks that ‘the imagination naturally carries us to a lively conception of the usual effect, tho’ the object be different in the most material and most efficacious circumstances from that cause’ (THN This parallels his talk of the ‘flights of the imagination’ above.[19] Moreover, he takes these flights of the imagination to be ‘dangerous to reason’, which clearly corresponds to his claim that the first influence is ‘destructive of all the most establish’d principles of reasonings’ (THN In short, a lack of reflection (via the second influence) allows for the ‘trivial suggestions of the fancy’ (THN, which, being ‘of an irregular nature’ (THN, inevitably lead to credulity. The credulity horn is a direct result of rejecting that we have any duty to reflect whatsoever. Without such a duty, human weakness is such that we will succumb to systematic doxastic error.

Thus, Hume’s Dangerous Dilemma can be seen as a dilemma between two deontological extremes: should we embrace an absolute duty to always reflect, or should we reject that we have any duty to reflect whatsoever? The former leads to excessive scepticism, while the latter leads to credulity.[20] Hume’s resolution to this dilemma will be to pave a deontological middle way.

The Deontological Threshold

So Hume is in quite a pickle at this point. It is instructive to consider what doxastic aims he would like to achieve come the end of THN 1.4.7. He would certainly want to reject the two horns of the Dangerous Dilemma—that is, excessive scepticism and credulity. Additionally, he would want to endorse our common-life beliefs, and also science and philosophy.[21] And he would ideally like to reject superstition as well. I see Hume’s strategy for achieving these aims as taking a distinctly deontological structure. More specifically, I see it as involving the recognition of duties of reflection up to a point (in order to reject credulity and preserve the beliefs of common life), and an account of the supererogatory beyond it (in order to endorse science and philosophy in favour of superstition and excessive scepticism).

Immediately upon posing the Dangerous Dilemma, Hume initially seems to back down from the challenge of overcoming it:

For my part, I know not what ought to be done in the present case. I can only observe what is commonly done; which is, that this difficulty is seldom or never thought of; and even where it has once been present to the mind, is quickly forgot, and leaves but a small impression behind it. Very refin’d reflections have little or no influence upon us; and yet we do not, and cannot establish it for a rule, that they ought not to have any influence; which implies a manifest contradiction. (THN

However, Hume quickly realises that refusing to engage with the Dangerous Dilemma is untenable; in THN he states that the position endorsed above that ‘reflections very refin’d and metaphysical have little or no influence upon us’ is an ‘opinion I can scarce forbear retracting, and condemning from my present feeling and experience’. In short, retreat is not a viable option.[22]

The obvious solution to the Dangerous Dilemma is of course to understand that it embodies a false dichotomy: we can recognise a duty to reflect only up to a point, hence avoiding both horns of the Dangerous Dilemma. This ‘Goldilocks’ path mandates sufficient reflection such that we avoid credulity, but not so much that we will fall into excessive scepticism.

He initially provides a doomed attempt in this direction immediately upon raising the Dangerous Dilemma:

Shall we, then, establish it for a general maxim, that no refin’d or elaborate reasoning is ever to be receiv’d? (THN

Here Hume suggests forbidding refined and elaborate reflection (presumably also accepting a corresponding duty to reflect up to a certain point, in order to avoid reducing this new option to the credulity horn).[23] This would indeed constitute a deontological middle way between the two horns: it requires some reflection and so avoids the credulity horn, and yet disallows the extreme level of reflection that engenders the sceptical horn.

Unfortunately, this principle turns out to be a non-starter. First, it cuts off ‘entirely all science and philosophy’ (THN Second, it fails to provide a principled reason as to why we should accept only certain qualities of the imagination, since ‘by parity of reason’ we ‘must embrace all of them’ (THN Contrast this with the normative distinction that Hume draws between the ‘permanent, irresistible, and universal’ principles of the imagination and the blameable ‘changeable, weak and irregular ones’ (THN Such a distinction is principled, because the latter, unlike the former, are ‘opposite to the other principles of custom and reasoning’ (THN; there is a clear and natural divide between the principles aligned with custom and reason on the one hand, and the principles that are not on the other. However, with regard to Hume’s proposed principle above, we cannot appeal to any such considerations in order to justify the normative distinction between the trivial propensities of the imagination that stave off excessive reflection (and hence scepticism) and the ones that lead to credulity, since science and philosophy are forbidden and yet are very much consonant with the principles of custom and reasoning. Third, the principle contains an element of hypocrisy in forbidding refined reasoning and reflection despite being founded on such ‘refin’d and metaphysical’ reflection (THN

Hume ultimately realises that there is another middle way that avoids the above problems, one which turns on recognising the epistemically supererogatory.[24] We have a duty to reflect only up to a certain point; further reflection beyond this deontological threshold is not required of us as a duty, but might nevertheless be epistemically good to do.[25] Such reflection should not be required (which avoids necessitating the sceptical horn); nevertheless, it should not be forbidden (which is where Hume’s initial attempt above went wrong).[26] This avoids the problems that plagued Hume’s first attempt at a deontological middle way. First, it allows for science and philosophy. Secondly, since science and philosophy are not forbidden, we can maintain a principled epistemic distinction. Within the deontological threshold, there is a clear and natural distinction to be made—accept the principles that are aligned with custom and reason, and reject those that are not.[27] Thirdly, there is no hypocrisy involved, since the refined reasoning that founds this principle is not forbidden.

Support for Hume’s recognition that a certain rarefied level of reflection is supererogatory largely derives from his repeated acknowledgments that philosophy is optional, coupled with his recognition of the value of such reflective pursuits; jointly, these reveal his understanding of such a level of reflection as supererogatory.

Hume’s resolution to scepticism in THN 1.4.7 comes with a sustained emphasis on the optional nature of philosophy. This point is also made by Ainslie:

philosophy is entirely optional, dependent on the propensities of the individual and her or his inclinations at the time. Because we are fundamentally carried along by reason and the senses, no one must take up the reflective posture whereby we observe their operations. (2015: 239)

Schafer (2014: 12) likewise sees philosophy as merely elective.[28]

Upon beginning to shake off his sceptical funk, Hume says, ‘Nay if we are philosophers, it ought only to be upon sceptical principles, and from an inclination, which we feel to the employing ourselves after that manner’ (THN Certainly, philosophy is by no means mandated of us. If we have an inclination to it, then we should feel free to proceed to engage with it (upon sceptical principles). Philosophy may be a fine pursuit, but it is not a required one.

Correspondingly, Hume’s discussion of the ‘honest gentlemen’ of England is also pertinent:

… there are in England, in particular, many honest gentlemen, who being always employ’d in their domestic affairs, or amusing themselves in common recreations, have carried their thoughts very little beyond those objects, which are every day expos’d to their senses. And indeed, of such as these I pretend not to make philosophers, nor do I expect them either to be associates in these researches or auditors of these discoveries. They do well to keep themselves in their present situation …. (THN

Although this passage might be seen as somewhat sardonic, it nevertheless seems clear that irrespective of any jibes at the expense of the English, Hume here recognises that we have a duty to reflect only up to a point.[29] The honest gentlemen of England are evidently not a particularly reflective lot, having ‘carried their thoughts very little’ beyond ‘their domestic affairs’. Hume does not demand that they expand their epistemic horizons; he does not seek to make them philosophers, or even auditors of philosophy. They ‘do well to keep themselves in their present situation’.[30] In short, these gentlemen are under no requirement to reflect more than they do. This is not to say that they lack any duty to reflect whatsoever. As we have seen, this would lead to credulity, which is unambiguously culpable, even at the level of common-life; after all, to successfully function in daily life, we need to ensure that we form generally reliable beliefs. But having reflected to this minimal point, thus avoiding credulous beliefs such as prejudice and unphilosophical probabilities, the common gentlemen are within their rights to remain content with this limited level of reflection. In short, we only have a duty to reflect up to the point called for by ‘domestic affairs’, or common life; any reflection beyond this point is not required of us.[31]

Hume repeats the point when preparing to continue his sojourn into Book 2:

If the reader finds himself in the same easy disposition, let him follow me in my future speculations. If not, let him follow his inclination, and wait the returns of application and good humour. (THN

Not all readers are required to follow Hume into his abstruse reasonings. Those who are inclined to do so should follow Hume’s lead, but those who are not are perfectly entitled to follow their inclinations instead.

In coming to see a certain degree of reflection as optional in THN 1.4.7, Hume changes his mind regarding the putative duty to reflect without limits on all our beliefs. Grappling with the Dangerous Dilemma and seeing no other way out, Hume decides to renounce his commitment to this overly demanding unrestricted duty in favour of something more circumspect in order to stave off the counter-intuitive extreme scepticism that results from the former.

Nevertheless, despite not being mandated, science and philosophy are clearly laudatory pursuits. This can be seen from the fact that Hume rejects the maxim of forbidding all refined reasoning precisely because this would cut off all science and philosophy in THN Elsewhere, he makes the point that philosophy is such that its ‘sovereign authority ought every where to be acknowledg’d’ (THN[32] Thus, although we are not required to pursue science and philosophy, it is laudable to do so: these pursuits are supererogatory.[33]

To sum up, there is a minimal level of reflection corresponding to the sphere of common life that we are obliged to undertake.[34] These baseline obligations to reflect are those involved in the prosaic contemplation of everyday life; that is, reflection that is ‘employ’d’ in our ‘domestic affairs’ (THN We have duties to perform adequate minimal reflection that ensures that our causal reasoning about daily life is not unduly defective. For instance, we should avoid unphilosophical probabilities such as ‘Prejudice’: concluding that ‘An Irishman cannot have wit, and a Frenchman cannot have solidity’ (THN would violate these baseline duties of reflection. Similarly with falling prey to recency bias in weighing more heavily experiments or observations that have freshly occurred (THN We are obliged to sufficiently reflect to avoid these cognitive pitfalls that typically infect our reasoning about daily life.

However, beyond this relatively undemanding point, reflection can be supererogatory: any reflection that is esoteric to intellectual pursuits (rather than endemic to common life) might be good, but not required. How does this framework allow Hume to achieve his doxastic goals? It clearly dismisses the credulity horn that stemmed from rejecting any duty to reflect. Since there is a level of reflection required of us as a duty, we can dismiss most epistemically bad trivial propensities, and thus banish most errors and obscurities. It also goes some way to avoiding the sceptical horn. Because we do not have an absolute duty to always reflect, we are not bound by duty to the epistemic apocalypse described in THN 1.4.1.

At this point, my interpretation might rouse some worries. Although we are no longer bound by duty to the destruction of all our beliefs, we still want to claim that such an outcome is bad. But if higher levels of reflection are supererogatory, as Hume must recognise in order to preserve the claim that philosophy and science are laudatory pursuits, he seems nevertheless to be forced to admit that epistemic annihilation, while not required per se, is yet laudable.[35] And this is deeply counterintuitive.[36]

Moreover, there is still the problem of superstition. Superstition typically involves postulating supernatural entities as explanations for phenomena—the earthquake struck because God was unhappy, the seizure happened because of possession by evil spirits, and so forth. These posits proliferate as the result of insufficient reflection.[37] As Hume describes the genesis of religion in the Natural History of Religion, polytheism (the first form of religion) arose from our tendency to explain phenomena as effects of intelligent entities that are human-like in various ways (NHR 3.2). This is clearly an instance of the first influence of general rules: in postulating such a causal relation between intelligent agents and natural phenomena, we fail to distinguish the ‘essential’ from the ‘superfluous’ (THN circumstances involving causation by intelligent beings, and thus mistakenly attribute natural phenomena to the actions of unseen gods and spirits.

As Hume notes, such beliefs belong to the ‘vulgar’, and would be easily corrected by reason (NHR 1.8). Thus, this form of superstition is the result of the first influence when uncorrected by the reflective second influence, which is to say, it arises from insufficient reflection. One does not need to be a philosopher or a scientist to recognise that such beliefs are problematic, as a simple application of the second instance of general rules would suffice. In virtue of the fact that such primitive superstition would not survive even relatively cursory reflection, such superstition is readily dismissed by Hume’s framework, given that it results from a failure to fulfil a minimal duty to reflect.

However, some superstition involves a great deal of reflection, for instance reasoning regarding the nature of God or the origin of the universe. This is likely the form of superstition that Hume has in mind in THN 1.4.7:

But even suppose this curiosity and ambition shou’d not transport me into speculations without the sphere of common life, it wou’d necessarily happen, that from my very weakness I must be led into such enquiries. ’Tis certain, that superstition … opens a world of its own, and presents us with scenes, and beings, and objects, which are altogether new. (THN

Superstition in this sense goes far beyond the bounds of common life, being highly reflective in nature. Is this to say that such superstition is laudatory as well?[38] This is surely an unwelcome result! More needs to be said.

Beyond the Deontological Threshold

Hume has managed to avoid one horn of the Dangerous Dilemma, and has gone some way towards addressing the second. As mentioned above, there is nevertheless still the problem of distinguishing good levels of reflection from bad beyond the deontological threshold. How can Hume castigate excessive scepticism and superstition, while nevertheless lauding science and philosophy? To resolve this issue, we have to examine THN 1.4.7 for a theory of what makes a certain level of reflection good beyond the deontological threshold of common life.

First, a quick note. If one recognises the supererogatory, one is committed to a theory of the good beyond the bounds of duty. One needs some basis, apart from deontological considerations, on which to claim that a supererogatory action is in fact laudable. Of course, we would not want to claim that goodness beyond the bounds of a duty to X is simply a matter of ‘more X is better’. Take the example of duties we have to ourselves. A person has duties to improve themselves to a degree; we should develop and nurture our natural talents. Some people pursue excellence in certain fields diligently, such as the concert violinist, and we might say that such pursuits are supererogatory. But one who dedicated one’s life to violin-playing at the expense of friends, family, and ethical considerations would not be laudable; this would be a tremendously selfish way to live. In short, a theory of the good beyond the deontological threshold should not simply consist of ‘more’ of the duty. A different account is needed.

The most natural basis on which to evaluate actions beyond the deontological threshold seems to be in terms of their results. This is perhaps why Urmson ends his paper by suggesting that his proposed moral account of the supererogatory might ‘be called a version of utilitarianism’ (1958: 216), and moreover notes that ‘utilitarianism can best accommodate’ the supererogatory (1958: 215).[39] This fear of diluting the deontological is likely why some deontologists have been so reluctant to straightforwardly recognise the supererogatory, either accounting for supererogatory phenomena in terms of weaker duties (such as imperfect duties), or rejecting them tout court.[40] Thus, if I am right that Hume recognises philosophy and science as supererogatory, he is required to give a distinctly non-deontological account of what makes a certain level of reflection good beyond the bounds of duty, and such a framework will likely evaluate levels of reflection in terms of their results.[41]

It is no great originality on my part to note that there is a strong theme of usefulness and agreeableness in THN 1.4.7.[42] This, I suggest, is how Hume distinguishes good levels of reflection from bad ones when beyond the bounds of duty. Beyond the deontological threshold, good levels of reflection are those that produce useful or agreeable beliefs, while bad levels of reflection are those which produce disagreeable or detrimental beliefs. This allows Hume to dismiss superstition and excessive scepticism while nevertheless endorsing science and philosophy. Since our limited duty to reflect has already ruled out credulity, Hume’s epistemic framework in THN 1.4.7 licenses him to accomplish all his doxastic goals.[43]

Hume clearly takes science and philosophy to be agreeable:[44]

These sentiments spring up naturally in my present disposition; and shou’d I endeavour to banish them, by attaching myself to any other business or diversion, I feel I shou’d be a loser in point of pleasure; and this is the origin of my philosophy. (THN

A true sceptic will be diffident of his philosophical doubts, as well as of his philosophical conviction; and will never refuse any innocent satisfaction, which offers itself, upon account of either of them. (THN

Moreover, Hume is led to return to philosophy by his passions of curiosity and ambition (THN As often noted in the secondary literature, curiosity is stirred only when the discovery of truth is agreeable due to the ‘genius and capacity’ required to discern it (THN, or when the truth concerned is useful (THN[45] Moreover, Hume’s ambition is ‘of contributing to the instruction of mankind’, thus ‘acquiring a name’ by his ‘inventions and discoveries’ (THN; contributing to the instruction of mankind is clearly useful to society, while the acquisition of fame is agreeable to Hume. Although science and philosophy involve a level of reflection beyond the bounds of duty, Hume nevertheless endorses science and philosophy on the basis that such a level of reflection is useful and agreeable.

By contrast, the level of reflection that engenders excessive scepticism is neither useful nor agreeable.[46] Such scepticism is certainly detrimental to oneself: Hume takes it to be ‘dangerous, and attended with the most fatal consequences’ (THN It is also detrimental to others; Hume points out that to ‘torture’ one’s brain does not ‘serve either for the service of mankind, or for my own private interest’, and is an ‘abuse of time’ (THN It is also disagreeable to oneself: Hume’s somewhat hyperbolic description of the ‘melancholy and delirium’ (THN engendered by sceptical worries makes that clear enough. Moreover, it is disagreeable to others: when beset by scepticism, Hume notes that ‘Every one keeps at a distance, and dreads that storm, which beats upon me from every side’ (THN Since the extreme level of reflection that leads to excessive scepticism is not required as a duty, nor it is good beyond the bounds of duty, we have every reason to reject it.

Similarly, Hume dismisses superstition precisely on the grounds of the danger it poses. He notes that he prefers philosophy to superstition on the basis that it is ‘safest and most agreeable’; he also remarks that ‘the errors in religion are dangerous; those in philosophy only ridiculous’ (THN Hume’s discussion of the monkish virtues likewise emphasises that superstition is moreover disagreeable to the self and others; such qualities ‘neither qualify him for the entertainment of company, nor encrease his power of self-enjoyment’ (EPM 9.3). The monkish virtues also lack usefulness, since they do not render the possessor ‘a more valuable member of society’ (EPM 9.3). Again, given that superstition involves a level of reflection beyond ‘that narrow circle of objects, which are the subject of daily conversation and action’ (THN, it is not required of us as a duty; given further that such a level of reflection it is neither useful nor agreeable, we have no reason to endorse it.[47]

Such an interpretation thus delivers all of Hume’s doxastic aims in THN 1.4.7. It dismisses credulity, superstition, and excessive scepticism, while requiring a minimal level of reflection in common life as well as endorsing science and philosophy. Hume can now continue his voyage into Book 2 with a clear conscience and easy heart.

The Epistemic and the Ethical

In sum, there is a powerful argument to be made that there is a level of description of THN 1.4.7 that takes on a deontic structure, as described above. This makes sense not only of the texts within THN 1.4.7, but also without, particularly the distinctly deontic genesis of the sceptical horn in THN 1.4.1, as well as the basis for the credulity horn in Hume’s discussion of the first and second influence of general rules in THN 1.3.13. Generally, discussions of Hume’s response to scepticism tend to focus on Book 1 Part 4, and can consequently neglect to take account of Hume’s epistemology in Book 1 Part 3. I take it that one of the primary strengths of the interpretation defended in this paper over many of its competitors is its ability to integrate Hume’s nascent epistemology in Book 1 Part 3 (in particular, the discussion of general rules) with the sceptical problem (and its eventual solution) in Book 1 Part 4.[48]

Nevertheless, such an account leaves it very much open what Hume’s epistemology looks like at the fundamental level, and I refrain from committing to any particular account in this paper.[49] In short, one can accept that Hume’s framework in THN 1.4.7 involves the calibration of a deontic threshold and an account of the epistemically supererogatory, and still legitimately enquire regarding the normative grounds for these duties to reflect. I think that any answer at this point will have a speculative element to it, but at this point there are a wide variety of options left open.[50] One could take these duties to reflect to be normatively basic. Alternatively, one might relate these duties to Hume’s claim regarding the moral sense acquiring new force when reflecting on itself (THN, as well as his claims about a mind being able to bear its own survey (THN[51] Perhaps one could account for these duties on externalist grounds, such as stability (e.g., Loeb 2002) or reliability (Schmitt 2014); another externalist option would be to offer a ‘proper function’-type account (in line with Craig 1987 and Wolterstorff 1996) and ground these duties in something like the proper functioning of our faculty of reason. Or one could alternatively ground these duties on the basis of truth or probable truth, in line with Garrett’s (2015) general project.[52] One might offer a virtue-theoretic account (in line with Schafer 2014, although as noted earlier, Schafer’s framework also has strong connotations of a usefulness and agreeableness account), and perhaps take there to be duties to reflect because such reflection is what a wise person would do. Or perhaps one could base our duties to reflect on usefulness and agreeableness; perhaps the framework of THN 1.4.7 involves usefulness and agreeableness all the way down. In what remains of this paper, I will briefly discuss the last option.

One prima facie appeal to grounding our duties to reflect on usefulness and agreeableness is that it provides a pleasing unity to Hume’s epistemic framework in THN 1.4.7. However, I believe that this unity is not without cost.[53] As I argue (Qu 2014), the usefulness and agreeableness interpretation collapses moral and epistemic normativity, since moral normativity also turns on usefulness and agreeableness (EPM 9.1).[54] Of course, this is a distinction that Hume insists on, as seen in the following passage:

Actions may be laudable or blameable; but they cannot be reasonable or unreasonable: Laudable or blameable, therefore, are not the same with reasonable or unreasonable. (THN

If, on the other hand, one refrains from grounding our duties to reflect on usefulness and agreeableness, then one seems able to avoid this problem. For what determines epistemic justification is not merely usefulness and agreeableness, but also our distinctly epistemic duties to reflect. Thus, while moral normativity is grounded on usefulness and agreeableness all the way down, the same would not be true of our epistemic duties.

Moreover, it seems that even Hume’s account of the epistemically supererogatory (which is founded on usefulness and agreeableness) can avoid being merely practical.[55] In general, supererogatory actions seem to have the same kind of normative worth as the duties they surpass;[56] for instance, my going beyond the bounds of a moral duty to care for my parents is also morally laudable. Thus, if keeping to a duty to reflect is epistemically laudable, then going beyond such a duty (in a laudable manner) seem as though it should be epistemically laudable as well. If this is the case, then there remains a clear distinction between moral and epistemic normativity both within and without the deontological threshold, and so there is little threat of collapsing the two.

It might be objected that even if this is granted, this would nevertheless be to divorce Hume’s epistemology from considerations from truth, since false beliefs might be useful and agreeable, while true beliefs might be detrimental and disagreeable. In response, an account that evaluates beliefs or belief-forming dispositions can nevertheless have a correlation to truth, as I argue (Qu 2014: 518–519). Both passions and beliefs are required for action. In evaluating the usefulness and agreeableness of passions (or passion-forming dispositions), we hold fixed that we have ‘good’ beliefs: benevolence is taken to be useful and agreeable, because given true beliefs, it leads to useful and agreeable actions. If one instead had very poor beliefs about what actions helped others, a passion of benevolence could turn out very badly indeed. Conversely, in evaluating the usefulness and agreeableness of beliefs (or belief-forming dispositions), we should hold fixed that we have ‘good’ passions. And assuming we have ‘good’ passions, the beliefs which tend to be useful and agreeable will be true ones. Deviancy is possible, but Hume has some leeway to accommodate such deviancy, in line with his claim that ‘virtue in rags is still virtue’ (THN Thus, usefulness and agreeableness might be a moral consideration with respect to passions, but an epistemic one, relating to truth, with respect to beliefs. Admittedly, this is not as tight a connection as we typically take there to be between truth and epistemic justification, but it is a meaningful connection nonetheless.

In any case, my paper does not need to take a stand on this complex and deep issue; it merely attempts to provide an epistemic framework founded on duties to reflect, independently of the normative grounding for such a framework. This is all that I think is needed to explain—and perhaps defend—Hume’s infamously convoluted treatment of scepticism in THN 1.4.7.


For detailed and very helpful comments on the paper, thanks are owed to Donald Baxter, Jonathan Cottrell, Don Garrett, Ryan Hickerson, David MacArthur, Peter Millican, and Karl Schafer. I am also grateful to Miren Boehm and Jennifer Marusic for very useful discussion on these issues. Thanks also to an audience at the Hume Society Conference 2016 at the University of Sydney for a very stimulating Q&A session. A debt of gratitude is likewise owed to the area editor and two anonymous referees for this journal for some excellent comments.


  • Ainslie, Donald (2015). Hume’s True Scepticism. Oxford University Press.
  • Ardal, Pall (1976). Some Implications of the Virtue of Reasonableness in Hume’s Treatise. In Donald Livingston and James King (Eds.), Hume: A Re-Evaluation (91–106). Fordham University Press.
  • Baceski, Tina (2013). Hume on Art Critics, Wise Men, and the Virtues of Taste. Hume Studies, 39(2), 233–256.
  • Baier, Annette (1991). A Progress of Sentiments: Reflections on Hume’s Treatise.Harvard University Press.
  • Boehm, Miriam (2013). Hume’s Foundational Project in the Treatise. European Journal of Philosophy, 24(1), 55–77.
  • Bernstein, Mark (1986). Moral and Epistemic Saints. Metaphilosophy, 17(2&3),102–108.
  • Boyle, Deborah (2012). The Ways of the Wise: Hume’s Rules of Causal Reasoning. Hume Studies, 38(2), 157–182.
  • Boehm, Miren (2013). The Normativity of Experience and Causal Belief in Hume’s Treatise. Hume Studies, 39(2), 203–231.
  • Broughton, Janet (2004). The Inquiry in Hume's Treatise. The Philosophical Review, 113(4), 537–556.
  • Chisholm, Roderick (1963). Supererogation and Offence: A Conceptual Scheme for Ethics. Ratio, 5, 1–14.
  • Collier, Mark (2008). Two Puzzles in Hume’s Epistemology. History of Philosophy Quarterly, 25(4), 301–314.
  • Cummins, Phillip D. (1999). Hume’s Diffident Skepticism. Hume Studies, 35(1&2), 43–66.
  • Craig, Edward (1987). The Mind of God and the Works of Man. Clarendon Press.
  • Dancy, Jonathan (1988). Supererogation and Moral Realism. In Jonathan Dancy, J. Moravcsik, and C. Taylor (Eds.), Human Agency (170–188). Stanford University Press.
  • Darwall, Stephen (1993). Motive and Obligation in Hume’s Ethics. Noûs, 27(4), 415–448.
  • Durland, Karánn (2011). Extreme Skepticism and Commitment in the Treatise. Hume Studies, 37(1), 65–98.
  • Falk, W. D. (1976). Hume on Is and Ought. Canadian Journal of Philosophy, 6(3), 359–378.
  • Falkenstein, Lorne (1997). Naturalism, Normativity, and Scepticism in Hume’s Account of Belief. Hume Studies, 23(1), 29–72.
  • Feldman, Fred (1986). Doing the Best We Can. Reidel.
  • Garrett, Don (1997). Cognition and Commitment in Hume's Philosophy. Oxford University Press.
  • Garrett, Don (2015). Hume. Routledge.
  • Hearn, Thomas K. (1970). ‘General Rules’ in Hume’s Treatise. Journal of the History of Philosophy, 8(4), 405–422.
  • Heyd, David (1982). Supererogation: Its Status in Ethical Theory. Cambridge University Press.
  • Hedberg, Trevor (2014). Epistemic Supererogation and Its Implication. Synthese, 191(15), 3621–3637.
  • Hickerson, Ryan (2013). What the Wise Ought Believe: A Voluntarist Interpretation of Hume’s General Rules. British Journal for the History of Philosophy, 21(6), 1133–1153.
  • Hume, David (1975). A Treatise of Human Nature (2nd ed.). L. A. Selby-Bigge and P. H. Nidditch (Eds.). Clarendon Press.
  • Hume, David (1975). Enquiries Concerning Human Understanding and Concerning the Principles of Morals (3rd ed.). L. A. Selby-Bigge and P. H. Nidditch (Eds.). Clarendon Press.
  • Hume, David (2008). Dialogues Concerning Natural Religion, and The Natural History of Religion. J. C. A. Gaskin (Ed.). Oxford University Press.
  • Hursthouse, Rosalind (1999). Virtue Ethics and Human Nature. Hume Studies, 25(1–2), 67–82.
  • Ievers, Lisa (2012). Hume’s Conception of Proper Reflection. Hume Studies, 38(2), 129–155.
  • Kail, Peter (2005). Hume’s Ethical Conclusion. In Marina Frasca-Spada and Peter Kail (Eds.), Impressions of Hume (125–139). Oxford University Press.
  • Korsgaard, Christine (1996). The Sources of Normativity. Cambridge University Press.
  • Loeb, Louis E. (2002). Stability and Justification in Hume's Treatise. Oxford University Press.
  • McCormick, Miriam (2005). Why Should We Be Wise? Hume Studies, 31(1), 3–20.
  • Moore, G. E. (1948). Principia Ethica. Cambridge University Press.
  • Moore, Michael (1997). Placing Blame: A General Theory of the Criminal Law. Oxford University Press.
  • Morris, William (2000) Hume’s Conclusion. Philosophical Studies, 19(1), 89–110.
  • Morris, William (2006). Belief, Probability, Normativity. In Saul Traiger (Ed.), The Blackwell Guide to Hume’s Treatise (77–94). Blackwell.
  • New, Christopher (1974). Saints, Heroes and Utilitarians. Philosophy, 49(188), 179–189.
  • Norton, David (1982). David Hume. Princeton University Press.
  • Owen, David (1999). Hume's Reason. Oxford University Press.
  • Pybus, Elizabeth (1982). Saints and Heroes. Philosophy, 57(220), 193–199.
  • Qu, Hsueh (2014). Hume’s Practically Epistemic Conclusions? Philosophical Studies, 170(3), 501–524.
  • Qu, Hsueh (2017). Hume’s Doxastic Involuntarism. Mind, 126(501), 53–92.
  • Qu, Hsueh (2019). Laying Down Hume’s Law. Pacific Philosophical Quarterly, 100(1), 24–46.
  • Qu, Hsueh (2020). Hume's Epistemological Evolution. New York: Oxford University Press.
  • Rawls, John (1971). A Theory of Justice. Harvard University Press.
  • Raz, Joseph (1975). Permissions and Supererogation. American Philosophical Quarterly, 12(2), 161–168.
  • Richards (1971). A Theory of Reasons for Action. Clarendon Press.
  • Ridge, Michael (2003). Epistemology Moralized: David Hume’s Practical Epistemology. Hume Studies, 29(2), 164–204.
  • Sasser, Nathan (in press). Hume’s Purely Practical Response to Philosophical Skepticism. Hume Studies.
  • Schafer, Karl (2014). Curious Virtues in Hume’s Epistemology. Philosophers’ Imprint, 14(1), 1–20.
  • Schmitt, Frederick (2014). Hume’s Epistemology in the Treatise. Oxford University Press.
  • Schumaker, Millard (1972). Deontic Morality and the Problem of Supererogation. Philosophical Studies, 23(6), 427–428.
  • Sturgeon, Nicholas L. (2001). Moral Skepticism and Moral Naturalism in Hume’s Treatise. Hume Studies, 27(1), 3–84.
  • Swanton, Christine (2007). Can Hume Be Read as a Virtue Ethicist? Hume Studies, 33(1), 91–113.
  • Swanton, Christine (2016). Hume and Virtue Ethics. In Paul Russell (Ed.), The Oxford Handbook of Hume (470–488). Oxford University Press.
  • Taylor, Jacqueline (2006). Virtue and the Evaluation of Character. In Saul Traiger, S. (Ed.), The Blackwell Guide to Hume’s Treatise (276–295). Blackwell.
  • Tidman, Paul (1996). Critical Reflection: An Alleged Epistemic Duty. Analysis, 56(4), 268–276.
  • Traiger, Saul (2005). Reason Unhinged: Passion and Precipice from Montaigne to Hume. In Joyce Jenkins, Jennifer Whiting, and Christopher Williams (Eds.), Persons and Passions: Essays in Honour of Annette Baier (100–115). University of Notre Dame Press.
  • Urmson, J. O. (1958). Saints and Heroes. In A. Melden (Ed.), Essays in Moral Philosophy (198–216). University of Washington Press.
  • Vitz, Rico (2009). Doxastic Virtue in Hume’s Epistemology. Hume Studies, 35(1–2), 211–229.
  • Walker, Matthew (2013). Reconciling the Stoic and the Sceptic: Hume on Philosophy as a Way of Life and the Plurality of Happy Lives’. British Journal for the History of Philosophy, 21(5), 879–901.
  • Williams, Michael (2004). The Unity of Hume’s Philosophical Project. Hume Studies, 30(2), 265–296.
  • Wolterstorff, Nicholas (1996). John Locke and the Ethics of Belief. Cambridge University Press.
  • Zimmerman, Michael (1996). The Concept of Moral Obligation. Cambridge University Press.


    1. In the references to Hume’s texts throughout, ‘THN’ refers to the Treatise of Human Nature, ‘EHU’ to the Enquiry Concerning Human Understanding, ‘EPM’ to the Enquiry Concerning the Principles of Morals, ‘NHR’ to the Natural History of Religion, and ‘DNR’ to the Dialogues on Natural Religion. Arabic numerals refer to section and paragraph numbers (EHU, EPM, NHR, and DNR) or book, part, section, and paragraph numbers (THN).return to text

    2. As Ainslie (2015: 121) points out, Hume frequently uses the term ‘reflection’ (or ‘reflexion’) and its cognates to indicate mental introspection (e.g., THN; THN; THN, as well as using the word as part of a technical term with regard to his ‘impressions of reflexion’ (e.g., THN to text

    3. See Chisholm (1963), Heyd (1982), Schumaker (1972), and Zimmerman (1996) for some discussion of how to characterise supererogation. See Feldman (1986), G. E. Moore (1948), New (1974), and Pybus (1982) for some denials of supererogation.return to text

    4. Besides Taylor, see Swanton (2007) for a defense of the virtue-ethical reading of Hume, and Swanton (2016) for an excellent discussion on reading Hume as a virtue ethicist. Hursthouse (1999) also reads Hume through the lenses of virtue ethics, although she finds his account to differ from the Aristotelian and Epicurean tradition. With respect to Hume’s epistemology, see Schafer (2014) for a virtue-theoretic approach. Baceski (2013) explores a virtue epistemologist reading of Hume by drawing parallels between Hume’s notion of the art critic and the wise man. Other readings of Hume along virtue epistemological lines include Boyle (2012) and Vitz (2009).return to text

    5. This is consistent with Taylor’s claim above that obligation is not a ‘central concept’ to his moral theory.return to text

    6. Darwall (1993: 424–426) distinguishes three notions of obligation in Hume: ‘natural’, ‘moral’, and ‘rule-obligation’, which is specific to justice, and takes THN to concern the moral species.return to text

    7. Hume makes the similar point that we laud anger and hatred when they are present in lower degrees than usually found in humanity (THN to text

    8. Human nature in general might of course have weak motives to, say, sacrifice oneself for the greater good (which are overwhelmed by other motives, e.g., the desire for self-preservation), but these motives will not be ‘capable of producing the action’ (THN, as Hume requires. Thus, such actions cannot be required of us as a duty, which is an intuitive result. Thanks to Jonathan Cottrell and Don Garrett for discussion.return to text

    9. Hume does raise a number of sceptical considerations in THN–5, but the Dangerous Dilemma is what drives forward the dialectic and structures the section. And, as we shall see, a solution to the Dangerous Dilemma will also offer a resolution to the other sceptical worries: we should address them in the same manner that we address scepticism with regard to reason. This is not a unique take on THN 1.4.7 on my part, for instance, approaches based on the Title Principle, such as Garrett (1997; 2015), adopt a similar stance.return to text

    10. This reading of the Dangerous Dilemma as founded on reflection (and the lack thereof) is also defended in Ievers (2012: 146), although she does not frame the issue in the distinctly deontological way that I do.return to text

    11. The reflective nature of the second influence is noted in Baceski (2013: 240), Boehm (2013: 227), Falkenstein (1997: particularly 47 and 49), Garrett (1997: 205), Hearn (1970: particularly 410 and 412), Hickerson (2013: 1146), Morris (2006: 87–88), Norton (1982: 208–221), Owen (1999: 149, 213), and Traiger (2005: 110–112).return to text

    12. See for instance Durland (2011: 66), Garrett (2015: 226), Loeb (2002: 86), and Morris (2000: 22), among others. Note that, for the purposes of this paper, I do not pass judgment on whether Hume’s argument in THN 1.4.1 is sound. All I require is that, come THN 1.4.7, Hume is convinced of its correctness.return to text

    13. An exception is Schmitt (2014: 320–330), who takes the sceptical worry of THN 1.4.1 to stem from what he calls the ‘Norm of Reduction’, which is broadly deontic in nature.return to text

    14. I do not mean to claim that Hume always refers to duties in making statements involving ‘ought’. For instance, in my Qu (2019), I do not read Hume’s passage on the is/ought distinction (THN as being specifically about duties; this is in line with Falk (1976: 360) and Sturgeon (2001: 9). I believe that context makes clear that Hume’s talk of ‘ought’ in THN clearly indicates that he thinks duties are involved, given what else he says in this section (e.g., ‘all the rules of logic require …’ in THN Might it be objected that Hume’s use of deontic language more generally (including his talk of rules, requirements, and must) in THN 1.4.1 should not be taken at face value? Without doing so, it is difficult to make sense of why exactly the sceptical scenario threatens, as noted above in the main text. We are not naturally or psychologically compelled to form these higher-order judgments, and it seems the most natural explanation of why the sceptical scenario looms large is due to an ostensible requirement that we form these higher-order judgments.return to text

    15. A worry might be raised about whether a deontological framework can be imputed to Hume with regard to beliefs, given his doxastic involuntarism (THN App. 2; Abstract 20–21; EHU 5.10–11): if ought implies can with regard to beliefs, then if we cannot change our beliefs, it follows that we should not believe otherwise than we do. I have argued elsewhere that Hume rejects ought implies can with regard to our beliefs (Qu 2017), and also that Hume rejects ought implies can more generally (Qu 2019: 41). In any case, even if one subscribes to ought implies can, it should be noted that this paper reads Hume as concerned with duties of reflection rather than assent, and reflection is something that we do have direct voluntary control over, even if assent is not.return to text

    16. Note that these rules play a role both in the realm of common life and also science and philosophy. On the one hand, they are ‘all the logic’ that Hume employs in his philosophical reasoning, and are crucial to both ‘natural philosophy’ and ‘moral’. On the other, they ‘might have been supply’d by the natural principles of our understanding’, and the ‘scholastic head-pieces and logicians shew no such superiority above the mere vulgar’ in this respect (THN As obligations to reflect, they apply to all cognitive pursuits, whether humdrum or intellectual.return to text

    17. Collier (2008: 311) also highlights the importance of these rules for avoiding what he calls ‘the fallacy of substitution’.return to text

    18. Hume also points out that the second influence finds the first influence to be ‘of an irregular nature, and destructive of all the most establish’d principles of reasonings’, in contrast with the ‘more general and authentic operations of the understanding’ (THN Indeed, this first influence is described as ‘capricious and uncertain’, while the second influence is ‘extensive and constant’ (THN In this, Hume seems to be anticipating his later epistemic distinction between the laudable ‘permanent, irresistible, and universal’ principles of the imagination, from the blameable ‘changeable, weak and irregular ones’ (THN to text

    19. Hume also discusses the role of the imagination with regard to the first influence of general rules in THN to text

    20. This provides a novel way of reading Hume’s description of the Dangerous Dilemma as a choice ‘betwixt a false reason and none at all’ (THN I had, as is typical, understood ‘false reason’ as referring to credulity, while taking ‘no reason at all’ to refer to excessive scepticism (cf. Qu 2014: 503). But given the above, there is another way to understand this claim. Perhaps the ‘false reason’ horn refers to the deontological claim that we always have a duty to reflect, which provides a reason to reflect that is in fact false; on the other hand, the ‘no reason at all’ horn is simply the rejection of any duty to reflect, which leaves us with no reason to undertake any reflection. In any case, my interpretation is consonant with either reading.return to text

    21. Of course, sceptical interpretations will take Hume to reject excessive scepticism, and endorse philosophy and common-life beliefs, in only a very weak sense. For instance, Broughton argues that Hume can only continue with his philosophical project in a ‘detached’ manner (2004: 550).return to text

    22. It might be queried if the sense of ‘ought’ in THN above, and the sense in which philosophy is optional (as we will see shortly), should be read as epistemic, rather than, say, moral or pragmatic. This is an instance of a more general question: why should we see Hume as offering an epistemic response to scepticism? If Hume does not offer any such response, then he is an unmitigated epistemic sceptic: he thinks there is no epistemic reason to hold the beliefs we do. I cannot provide a substantive discussion of this position here, but, briefly, I take such epistemically sceptical readings of Hume’s response to scepticism to struggle with what Cummins (1999) refers to as the ‘integration problem’ of adequately explaining Hume’s return to philosophy; see also Collier (2008: 311), Ainslie (2015: 229), and Boehm (2013: 73–74) for some responses in this ballpark to epistemically sceptical readings.return to text

    23. The notion of ‘refinement’ in particular indicates that reflection is involved in such reasoning—to refine some reasoning, we have to reflect on it.return to text

    24. As Hedberg (2014: 3622) notes, there has been relatively little discussion of epistemic supererogation in comparison with moral supererogation. For some discussion of epistemic supererogation see Bernstein (1986), Hedberg (2014), and Tidman (1996).return to text

    25. As mentioned earlier, Ievers (2012) similarly sees the Dangerous Dilemma has embodying a choice between two extremes of reflection, but her solution rests on favouring indolence unless we have good reason to reflect, rather than on a recognition of the supererogatory.return to text

    26. Although this is a good start, it does not completely address the sceptical horn: although excessive scepticism is not necessitated, it seems on the face of it to be good, which is unintuitive. I will address this issue in the following section.return to text

    27. Beyond the deontological threshold, the relevant epistemic distinction will have to take another form, as I will explain later.return to text

    28. Ainslie, Schafer, and I differ, however, on how Hume arrives at the thesis that philosophy is optional. I see it as turning on the deontological considerations described above. Meanwhile, Ainslie argues that the source is Hume’s understanding of ‘the reflective interference’ caused by the introspective project that philosophers undertake (Ainslie 2015: 244). Schafer argues that the difference in the degree of reflection that is recommended to each individual by the Title Principle is partly a function of that person’s natural curiosity (Schafer 2014: 12). My interpretation is noncommittal regarding the ‘Title Principle’: ‘Where reason is lively, and mixes itself with some propensity, it ought to be assented to. Where it does not, it never can have any title to operate upon us’ (THN This principle is consistent with the account I offer—while I emphasise Hume’s norms of reflection, the Title Principle concerns Hume’s norms of assent. Once we reflect, it might well be that we have a duty to assent to the results of this reflection. However, I take the crucial question to be whether we should indeed reflect in the first place. For the purposes of my paper, I do not take a position on the Title Principle, nor do I need to.return to text

    29. Hume’s tone in this passage is difficult to place. Loeb (2002: 92) reads the passage as approving, while Williams (2004: 286–7) takes the passage to imply no commendation. My view is that Hume sees the honest gentlemen as nothing to write home about—they do what they are required to, but little more.return to text

    30. Of course, this is not to say that they could not do better by reflecting to a higher degree. Hume’s statement might be read as somewhat condescending here. For instance: ‘You’d do well to make half of what I do’, said the banker.return to text

    31. Walker (2013: 899) similarly (albeit from a virtue-ethical standpoint) suggests that this passage indicates that the honest gentlemen might ‘exemplify a decent character’, although they would lack ‘the most perfect character’.return to text

    32. Of course, to state that everyone ought to recognise the authority of philosophy is not to claim that everyone ought to pursue it. Compare: all subjects ought to recognise the authority of the sovereign, but clearly, not all subjects ought to attempt to become the sovereign.return to text

    33. This theme of deep intellectual pursuits as supererogatory is also present to a degree in EHU 1. Here, Hume distinguishes the ‘easy and obvious’ philosophy from the ‘accurate and abstruse’ (EHU 1.3). Although Hume recognises that there is no fault in pursuing the easy philosophy, which he recognises to be more useful and agreeable, he also defends the more difficult philosophy from those who would reject it wholesale (EHU 1.7).return to text

    34. Note that this minimal duty to reflect does not entail that we need to reflect on each and every belief. For instance, it is implausible that reflection is required for immediate causal beliefs, or obvious and simple arithmetical truths.return to text

    35. It should be noted that strictly speaking, excessive scepticism is philosophy; indeed, it is philosophy that is extremely reflective and rigorous. For exegetical convenience, when I refer to ‘philosophy’ from here on, it refers to philosophy that falls shy of excessive scepticism. In line with this, my claim in this paper that philosophy is supererogatory is, strictly speaking, the claim that philosophy is supererogatory up to a certain point; excessively sceptical philosophy is neither required nor praiseworthy.return to text

    36. Hume does note in EHU 12.25 that it might be good to have once gone through excessive doubt. Nevertheless, remaining within excessive scepticism is certainly not laudatory (cf. EHU 12.23). In any case, I think THN 1.4.7 and EHU 12 to profess different epistemic frameworks, although I lack the space to defend this claim here; see my Qu (2020) for a defence of this claim.return to text

    37. In Qu (2014: 506–507), I argue that superstition arises from a lack of exercise of reason, which corresponds to a lack of reflection.return to text

    38. For exegetical convenience I use ‘superstition’ to refer to the reflective sort from here on.return to text

    39. Although Urmson is careful to note that he ‘does not wish to support any particular view about the supreme good’ (1958: 215).return to text

    40. Examples of those who try to account for supererogation in deontological terms are Dancy (1988), Rawls (1971), Raz (1975), and Richards (1971). Those who dismiss supererogation as a category were listed in an earlier footnote.return to text

    41. This leaves open whether such an account will be ‘act’ or ‘rule’—are we evaluating levels of reflection in terms of the results they in fact produce, or in terms of the results they would generally produce? The ‘rule’ variant seems more likely, given Hume’s position that ‘virtue in rags is still virtue’ (THN to text

    42. See Ardal (1976), Kail (2005), McCormick (2005), Owen (1999), Ridge (2003), Qu (2014), and Schafer (2014), for instance.return to text

    43. There are some parallels here with Michael Moore’s (1997) ‘threshold deontology’, whereby duties are binding, unless the consequences are sufficiently dire. My view is subtly different, however. In Moore’s account, the threshold resides with the consequentialist aspect—the threshold has to do with how bad the consequences are. In my account, the threshold resides with the deontological aspect—the threshold has to do with the level of duty required. Within the common sphere, we would have an epistemic duty to minimally reflect even if such reflection caused dire consequences, although in such a case this epistemic duty might be trumped by moral or pragmatic considerations.return to text

    44. Hume uses the term ‘philosophy’ more broadly than we do today; it includes natural philosophy (i.e., what we today call science), psychology, and so forth. Thus, his usage of the term ‘philosophy’ in these passages should be understood as encompassing science, which is surely useful and agreeable as well. As was common in the early modern period, Hume uses the term ‘science’ as an honorific term for disciplines that have secured a privileged standing (hence his stated ambition of establishing a ‘science of man’ in the introduction to the Treatise).return to text

    45. See Owen (1999: 219), Ridge (2003: 182–183), Qu (2014: 506), and Schafer (2014: 11–20), for instance.return to text

    46. One worry is that appealing to such considerations begs the question against the excessive sceptic, who will deny that we can know any such thing. Ridge (2003) engages deeply with this worry, consequently emphasising agreeableness to oneself, which he takes to be immune to scepticism. In correspondence, Peter Millican raises the related objection that this account begs the question against the superstitious, who will maintain that the future life received by the faithful will far outweigh this one in terms of usefulness and agreeableness; the point is that Hume can only triumph in this regard if he presupposes that the beliefs he uses to evaluate usefulness and agreeableness are true.

      My account has resources for dealing with both these objections: Hume can maintain that the beliefs that we use to evaluate usefulness and agreeableness must source from common life, which clearly appraises philosophy as more useful and agreeable than superstition or scepticism. This would not beg the question against the superstitious; since they too have to rely on common-life beliefs, it only seems fair to use this common ground in deciding disputes. Similar considerations apply to the sceptic, who Hume points out cannot help but maintain her common-life beliefs despite the influences of any convoluted sceptical arguments (cf. THN and THN to text

    47. In the Dialogues, Philo offers a response to scepticism that also distinguishes between the basis for common-life beliefs and the basis for philosophical pursuits; the former rests on ‘the absolute necessity … of doing so’, while the latter rests on ‘a certain pleasure and satisfaction’ (DNR 1.9). Although this account differs somewhat from the one I defend in the Treatise in this paper (notably, the defense of common-life beliefs rests on a pragmatic account of the sort found in EHU 12.21 rather than distinctly deontological considerations), there are also interesting parallels between the two: both distinguish between the basis for common-life and philosophical beliefs, and both take the latter to rest on considerations of agreeableness and usefulness.return to text

    48. Externalist interpretations, which we will discuss shortly, also account well for Book 1 Part 3 (although the same is not necessarily true of the discussion of general rules in particular), but such accounts tend to cohere less well with Book 1 Part 4. By contrast, my interpretation integrates the epistemologies of both parts in a cohesive fashion.return to text

    49. Thanks to Don Garrett and Karl Schafer for pressing me on the following issue, and for very helpful discussion.return to text

    50. The following options are not exhaustive, nor are they necessarily incompatible.return to text

    51. This would be continuous with Baier (1991) and Korsgaard (1996), who treat normativity as reflexivity. Such an account would likely privilege introspective reflection.return to text

    52. One possibility would be for reflection to be justified because it encouraged true or probably true beliefs. A more modest account would be for reflection to be justified merely because it aims at truth or probable truth—in reflecting and judging that p, one is taking p to be more probable than its competitors.return to text

    53. Besides the following, one textual kerfuffle for the usefulness and agreeableness account is that while it seems to have a great deal of textual substantiation with respect to Hume’s narrative regarding enquiries beyond common life, the same cannot be said with respect to Hume’s treatment of enquiries within common life. Hume unambiguously disparages superstition and scepticism on the basis of their lack of usefulness and agreeableness, but he does not seem to dismiss the ‘credulity’ horn of the Dangerous Dilemma on such a basis. He notes only that credulity is ‘dangerous to reason’ (THN, which in itself does not entail that it is dangerous per se (compare: an open mind might be said to be dangerous to prejudice, but this does not mean that it is dangerous). Doubtless, credulity is in fact not terribly useful (although wishful thinking may be quite agreeable), but the point is that unlike with scepticism and superstition, Hume does not obviously appeal to considerations of usefulness or agreeableness to dismiss it in THN 1.4.7. In general, Hume does not in THN 1.4.7 seem to appeal to usefulness or agreeableness to get the right normative results within the realm of ‘common life’—i.e., dismissing credulity, and endorsing our everyday beliefs. This feature of THN 1.4.7 coheres perfectly with my reading.return to text

    54. Although see Sasser (in press) for a usefulness and agreeableness account that looks to circumvent this objection.return to text

    55. It should be noted that this point is not crucial to my purposes. So long as the relevant duties are distinctly epistemic, this suffices from ensuring that epistemic normativity does not collapse into moral for Hume.return to text

    56. Where there are indeed duties in the vicinity.return to text