Hume's Second Thoughts About Belief

1. A Revision in Response to a Counterexample

The Appendix to Hume’s Treatise of Human Understanding contains thirty-five paragraphs: an introduction, twelve paragraphs of second thoughts on personal identity, a paragraph that confesses ‘two other errors of less importance’ (TApp.¶22), eight paragraphs near the beginning that give new arguments for a slightly revised theory of belief, and thirteen paragraphs to be inserted in various places in Book 1. The text can seem like an attic filled with random leftover material with the mysterious figure of the second thoughts on personal identity lurking in the corner.

I want to show that the material on belief in the Appendix fits together more snugly than appears on first reading. Hume gives a counterexample to his theory of belief; he then modifies a premise in his argument for that theory in light of that counterexample. Because this doctrine is a crucial premise in his original argument for his account of belief, he needs new arguments for his positive account. The material at the beginning of the Appendix provides new arguments to justify the revised account, an account in which the concept of belief is distinguished from the brightness of mental imagery. The revisions in the Appendix have a secret coherence where each of the parts illuminates and supports the others. They turn what had been an idiosyncratic account of the vivacity of mental imagery into a more attractive theory of how experiences determine credence in unobserved matters of fact.

These revisions are all set in motion by a counterexample. In the main part of the Treatise, Hume argues that beliefs are phenomenologically vivid ideas. In the Appendix, he comes up with a counterexample to that hypothesis. The ideas produced by poetic fictions are more vivid, in their way, than the ideas connected to boring matters of fact. But we do not believe in the poetic fictions, and we do believe in the boring matters of fact. So, belief cannot be just a matter of having relatively vivid ideas.

In the Appendix, Hume directs his readers to insert three paragraphs after his Book 1 discussion of how agitated animal spirits cause the insane to be more credulous. The first paragraph to be inserted begins, ‘we may observe the same effect of poetry in a lesser degree’ (T1.3.10.10). Because there already was a paragraph in that place beginning with those words, David and Mary Norton omit the earlier paragraph as having been made ‘redundant’ (T650), which seems like a reasonable editorial judgment.

We can compare the old paragraph to the new one and see why Hume abandons the view that perceptions of the same object only vary along a single dimension of vivacity. The original paragraph from 1739 presents vivid poetry and faint belief within a uniform scale of vivacity, with poetry always being fainter than belief. In that original paragraph, the difference between the insane person and the listener to poetry is ‘that the least reflection dissipates the illusions of poetry, and places the objects in their proper light. ‘Tis however certain, that in the warmth of a poetical enthusiasm, a poet has a counterfeit belief, and even a kind of vision of his objects’ (T650). Here, in accordance with the principle that perceptions with the same object only differ along one dimension, Hume is presenting steps in a ladder of vivacity. Well-wrought poetical ideas are less vivid than beliefs, but, at the height of enthusiasm, poetry approaches belief.

In the new paragraph, Hume emphasizes that the feelings evoked by poetry are different in character from the feelings associated with belief and ‘the mind can easily distinguish betwixt the one and the other’ (T1.3.10.10). All the same passions can be evoked in life and in poetry, but the ‘feelings of the passions are very different when excited by poetical fictions, from what they are when they arise from belief and reality’ (T1.3.10.10). Thus, ‘A passion, which is disagreeable in real life, may afford the highest entertainment in a tragedy, or epic poem’ (T1.3.10.10).

In accordance with this distinction, Hume creates two tracks of vivacity, one for belief and one for poetry. On the one hand, ‘A poetical description may have a more sensible effect on the fancy, than an historical narration. It may collect more of those circumstances, that form a compleat image or picture. It may seem to set the object before us in more lively colours’ (T1.3.10.10). So, for example, a gripping novel may cause us to form more vivid mental imagery than a list of purchases from a land registry. Similarly, our imagining of Dracula biting Lucy will probably have more phenomenological vivacity than the belief that Dover is the capital of Delaware. We can call the ‘more lively colors’ associated with poetry ‘phenomenological vivacity.’¹

Hume revises his account of belief in light of the counterexample by appealing to a second kind of vivacity. He writes, ‘Where the vivacity arises from a customary conjunction with a present impression; tho’ the imagination may not, in appearance, be so much mov’d; yet there is always something more forcible and real in its actions, than in the fervors of poetry and eloquence’ (T1.3.10.10). That is to say, when our ideas are vivified by the usual method of causal inference, even if we do not form vivid mental pictures, the resulting ideas still have a kind of forcefulness that motivates us in a way that poetical fictions do not.

The upshot is that the difference between contemplation and belief lies not in phenomenological vivacity, but rather in a distinctive feeling that constitutes the essence of belief. That is the position taken in another paragraph in the Appendix Hume tells us should be added to the discussion of belief:

I conclude, by an induction which seems to me very evident, that an opinion or belief is nothing but an idea, that is different from a fiction, not in the nature, or the order of its parts, but in the manner of its being conceiv’d. But when I wou’d explain this manner, I scarce find any word that fully answers the case, but am oblig’d to have recourse to every one’s feeling, in order to give him a perfect notion of this operation of the mind. An idea assented to feels different from a fictitious idea, that the fancy alone presents to us: And this different feeling I endeavour to explain by calling it a superior force, or vivacity, or solidity, or firmness, or steadiness. This variety of terms, which may seem so unphilosophical, is intended only to express that act of the mind, which renders realities more present to us than fictions, causes them to weigh more in the thought, and gives them a superior influence on the passions and imagination. (T1.3.7.7)

Here, he argues from introspection that beliefs are ideas with a sui generis feeling.² We can call this feeling ‘doxastic vivacity.’

In another passage from the Appendix that Hume instructs us to insert in Book 1, he considers the case where one person out of a pair has forgotten some common experience. The person who remembers it might describe the event and the forgetful one might understand the description without recalling the event. At a certain point, the forgetful person might recall the event, and, in Hume’s words, ‘the very same ideas now appear in a new light, and have, in a manner, a different feeling from what they had before. Without any other alteration, beside that of the feeling, they become immediately ideas of the memory, and are assented to’ (T1.3.5.4). Hume gives the presence of a certain feeling as necessary and sufficient for belief, and he does not mention vivacity.

In the original version of Book 1, Hume says that merely asserting that believers conceive of objects ‘in a different manner’ is insufficient, ‘not because it contains any falshood, but because it discovers not all of the truth’ (T1.3.7.4). He then proceeds to identify the relevant manner with phenomenological vivacity. When he writes the Appendix, Hume concludes that the gain in specificity does not outweigh the loss of plausibility. He rejects the identification of belief with a phenomenologically vivid idea, and he retreats to the view that belief is an idea considered in a particular manner.

Hume never asserts that belief is entirely independent of phenomenological vivacity. In ‘Of Tragedy,’ published in 1757, he endorses some of Bernard de Fontenelle’s remarks on tragedy as a partial solution to the paradox of why spectators enjoy watching tragic drama. According to Fontenelle, mild sorrows are pleasurable and conceiving of a situation as fictional diminishes the pains associated with it (Fontenelle, ‘Reflexions’ 163–64, quoted and translated in ‘Tragedy,’ 218–19). I do not want to lean too heavily on Hume’s endorsement of Fontenelle’s explanation, but the account suggests that every belief has more phenomenological vivacity compared to the mere contemplation of the same situation. Even if Hume does think that the passions associated with a believed circumstance are always more vivid than the passions associated with the mere contemplation of the same circumstances, it does not follow that every belief is phenomenologically more vivid than every contemplation of a fictional situation. The latter claim is what he denies in the new paragraph on poetic vivacity from the Appendix.

Though Hume endorses Fontenelle’s analysis as a partial solution to the Paradox of Tragedy, he also offers his own supplemental solution to the problem, and the absence of belief does not figure in his original contribution. He illustrates his solution with an example from Cicero’s prosecutorial oration against Verres, an example in which sadness over unjust executions is not ‘softened by fiction’ (‘Tragedy’ 219).³

‘If belief is a vivacious idea,’ Louis Loeb (2002: 70) asks, ‘how can the lively products of poetic enthusiasm fail to count as beliefs?’ The answer is that they would not. By the time he writes the Appendix, Hume no longer believes that perceptions of the same object only vary with respect to one sort of vivacity. This allows him to make belief into a sui generis feeling, and allows him to say that poetic enthusiasms can be more vivid in some respects without counting as a belief.

According to Trudy Govier, Hume’s point in his discussion of poetical fictions is to distinguish between force and vivacity (1972: 46–7). This does not seem to be well put. After all, as Anthony Nguyen observes (2017: 73), in Treatise 1.3.7.7, Hume makes it clear that he is using ‘vivacity’ and ‘force’ as variant terms that apply to the relevant feeling in this context. But what Govier means by vivacity is being painted in more distinct colors, which is what I have called phenomenological vivacity, and what she means by force is assent, which is what I have called doxastic vivacity. That is the distinction that Hume is drawing, and his point that these are distinct, independent feelings that can inhere in a conception and that each allows of stronger and weaker degrees.

Hume’s second thoughts strike me as being his better thoughts. It is not plausible to suppose that every belief is phenomenologically more vivid than every fiction. Insofar as the first mental state strikes us more forcefully than the second, it is only because it makes us believe in the state of affairs which it represents.

2. The Significance of Minor Concession

Given the counterexample of poetic vivacity, something has gone wrong with Hume’s original argument for his theory of belief. That argument runs as follows:

When you wou’d any way vary the idea of a particular object, you can only encrease or diminish its force and vivacity. If you make any other change on it, it represents a different object or impression. The case is the same as in colours. A particular shade of any colour may acquire a new degree of liveliness or brightness without any other variation. But when you produce any other variation, ’tis no longer the same shade or colour. So that as belief does nothing but vary the manner, in which we conceive any object, it can only bestow on our ideas an additional force and vivacity. (T1.3.7.5)

Hume supposes that if you have the idea of a particular shade, then modifying any features besides its brightness (e.g., hue or saturation) will change the color it represents. The example is not entirely persuasive, since one might think changes in brightness entail changes in color, as Justin Broackes (2002: 192) and Daniel Flage (1990: 169) have argued. Still, Hume’s argument is clear: belief and contemplation are ideas. Belief differs in some respect from contemplation but can have the same object; the only dimension along which two perceptions of the same object can vary is vivacity, so beliefs differ from contemplation with respect to vivacity (Broackes 2002: 189).

Hume decides that the premise that has gone wrong is that perceptions only vary along a single dimension of vivacity. He quotes from the passage from Treatise 1.3.7 and recants: ‘I believe there are other differences among ideas, which cannot properly be comprehended under these terms. Had I said, that two ideas of the same object can only be different by their different feeling, I shou’d have been nearer the truth’ (TApp.¶22). So, instead of saying that complex ideas of the same object can only vary by phenomenological vivacity, the new view is that they can vary by feeling, where feeling includes more possibilities than vivacity.

David Owen rightly conjectures that the problem that motivates the minor concession ‘seems to be, in part, that ideas of poetry and fiction, though not similar in feeling to beliefs, are aptly characterized in terms of vivacity’ (1999: 173). Along the same lines, Broackes writes, ‘it may well have been this objection that actually prompted Hume in the Appendix to the Treatise to renounce the view that the only way in which ideas may differ while still being ideas of the same object is in respect of force and vivacity . . . thus renouncing his first view of belief’ (2002: 193–94). Hume admits that ‘poetic enthusiasm’ produces vivid ideas but does not produce belief. So, Broackes concludes, ‘vivacity can’t constitute belief’ (2002: 194).

In the Appendix, Hume suggests that the mistake does not matter much. One can just say there is an idiosyncratic kind of vivacity associated with belief. He says that the mistake is of ‘less importance’ than the problem that he raises with personal identity (TApp.¶22). How much one thinks this depreciates the importance of the correction depends on how serious one thinks the difficulty concerning personal identity is; but, no matter what, Hume makes the concession share a paragraph with a point about the perceptibility of angles. Rhetorically, he may think that he made a dramatic confession of error in his discussion of personal identity, and that, if he makes too many of those, people will start to lose confidence in him as a metaphysician and a philosopher of mind. Substantively, he may have thought that his new view is not far from his old view. He may have to marshal new arguments in defense of a slightly different theory of belief because of the counterexample, but what he is saying in the Appendix is not radically different from what he said in the original version of Book I.

Hume’s belief that his new and old views are not that far apart may help to explain a new passage in which he complains of the subtlety and difficulty of his topic. He begins one of the inserted paragraphs that we have already discussed by describing his new conception of belief with a declaration of originality: ‘This operation of the mind, which forms the belief of any matter of fact, seems hitherto to have been one of the greatest mysteries of philosophy; tho’ no one has so much as suspected, that there was any difficulty in explaining it’ (T1.3.7.7). It is not true that Hume is the first to consider the question of how the mind forms beliefs about matters of fact. As Lewis Powell has observed, Locke begins the Essay Concerning Human Understanding by saying that his project is ‘to enquire into the Original, Certainty, and Extent of humane Knowledge: together with the Grounds and Degrees of Belief, Opinion, and Assent’ (Essay1.1.2).⁴ But the idiosyncracy of Hume’s theory of belief leads him to new problems, which no one else has considered.⁵ It is easy to sympathize with his situation. He is trying to draw fine distinctions between the ways that the vivacity of ideas may constitute the nature of belief, and no one in the tradition is giving him any help at all.

Seeing the connection between the parts of the Appendix that have to do with belief will save us from interpretive error. First, the way the material about belief in the Appendix hangs together shows that there is indeed a change in Hume’s treatment of belief. Emily Kress concedes that Hume says in the inserted passage on poetic enthusiasm that the vivacity bestowed on our ideas by poetry ‘never has the same feeling, which arises in the mind, when we reason’ (T1.3.10.10) and grants that this suggests ‘a distinction between two kinds of vivacity: the vivacity that characterizes belief and the vivacity that characterizes counterfeit belief,’ that is, poetic enthusiasm (Kress 2017: 66). Against this straightforward reading, she argues, ‘that it appeals to a distinction that plays no role in Hume’s official definition of belief, which defined belief only as an idea with force and vivacity. This definition did not present belief as an idea having a particular kind of force or vivacity’ (2017: 66). That is true of Hume’s original 1739 definition of belief in Treatise 1.3.7.5, but it is not true of the revised account of belief that Hume gives in the beginning of the Appendix (‘belief is nothing but a peculiar feeling, different from the simple conception’ TApp.¶3).

In a footnote, Kress points to the beginning of the Appendix, where Hume writes that he is mostly correcting poorly chosen expressions as a reason for thinking that he did not change his mind on the topic of belief (2017: 85n29). Her discussion neglects Hume’s confession of error, where, as we have seen, he says that he was mistaken in saying that two ideas with the same object can only differ with respect to vivacity and that he should have said they can only differ with respect to feeling (TApp.¶22). The rejected doctrine is a crucial premise in Hume’s argument that beliefs are vivid ideas, and the proposed replacement makes room for the doctrine that there are two sorts of vivacity at work in beliefs and poetical fictions, which is what a plain reading of the inserted passage on the topic would suggest.⁶ It is true that Hume says that this revision is less important than the problem concerning personal identity, and it is also true that he tries to downplay the importance of the revision. The concession of error is there nonetheless.

In addition, the way the material about belief in the Appendix hangs together shows what sparked those changes. According to Martin Bell, the revision in the minor concession arises out of an objection that Hume considers in the original version of Book 1, after he lets it percolate for a year. Bell (2002: 179) writes, ‘The problem is that, according to the maxim [scil., that vivacity spreads across associated perceptions] ideas can be enlivened by association with the data of sense and memory by any of the three principles of association. Yet “we find by experience that belief arises only from causation” (T1.3.9.2).’

I do not think that it is likely that the minor concession in the Appendix is a reply to this objection. In the original version of Book I, Hume replies to it by first explaining why contiguity and resemblance do not have as great an influence on belief as causation; he then attempts to turn the objection into ‘a proof of the present doctrine’ by showing ‘Contiguity and resemblance have an effect much inferior to causation; but still have some effect, and augment the conviction of any opinion, and the vivacity of any conception’ (T1.3.9.8). In effect, Hume does not present the possibility that belief might spread across principles of association other than cause and effect as a straightforward objection to his theory. Rather, he presents it as a test of the theory. Upon considering the test, he argues that various phenomena confirm his view that vivacity flows across all the principles of association. One of them runs as follows:

To begin with contiguity; it has been remark’d among the Mahometans as well as Christians, that those pilgrims, who have seen Mecca or the Holy Land are ever after more faithful and zealous believers, than those who have not had that advantage. A man, whose memory presents him with a lively image of the Red-Sea, and the Desert, and Jerusalem, and Galilee, can never doubt of any miraculous events, which are related either by Moses or the Evangelists. The lively idea of the places passes by an easy transition to the facts, which are suppos’d to have been related to them by contiguity, and encreases the belief by encreasing the vivacity of the conception. (T1.3.9.9)

Hume’s various examples of the transfer of vivacity work in various ways, and some of them, as Bell might emphasize (cf. 2002: 180), would not be rightly described as generating a new belief. In this case, however, Hume supposes that pilgrimages strengthen belief in miracles by having vivacity flow across the associative principle of contiguity. We associate Calvary with the resurrection of Jesus by the principle of contiguity; when pilgrims are at Calvary and receive impressions of sensation from the place, the vivacity spreads to associated belief in the resurrection and confirms that belief. There is a similar example of how ‘the ceremonies of the Roman Catholic religion’ are used by Catholics ‘in inlivening their devotion, and quickening their fervour, which otherwise wou’d decay away, if directed entirely to distant and immaterial objects’ (T1.3.8.8) and that text is repeated in the first Enquiry (EHU5.16), so Hume did not change his mind about it. In the Treatise, Hume thinks that there is one sort of vivacity, and it spreads from perception to associated perception. In the Appendix and the first Enquiry, Hume thinks there is more than one feeling that comes in different degrees of vivacity, and these vivacities spread from perception to perception, each in its own kind.

The objection that Bell emphasizes is one that Hume presents as a test for his theory and as a test that it passes. We should not attribute concessions to philosophers that they do not actually make. There is a better paper trail for thinking that poetic vivacity is the spark for the revision. In the 1739 version of Treatise 1.3.10.10, poetic fictions are always less vivid than belief. In the material that we are directed to replace that passage with, poetic fictions are more vivid in one respect, but not in another. Hume offers it as a counterexample to a previously maintained view and puts it forward alongside a revised view and arguments for that revised view.

Finally, the way the material about belief in the Appendix hangs together makes the nature of that change clear. My reading is something like Kaveh Kamooneh’s, according to which Hume has a ‘double-aspect account of vivacity,’ and perceptions can be vivid either phenomenologically or with respect to a quality that determines action (2003: 42). He appeals to Hume’s new treatment of poetical vivacity in the Appendix to justify this reading (2003: 46, 49). For Kamooneh, however, the second track is a disposition rather than a distinct feeling, and this disposition explains the causal differences between boring belief and poetic rapture (2003: 42). I do not think that Hume thinks of beliefs as dispositions or of dispositions as forceful,⁷ but let us set those worries aside. His appeal to poetic fictions is part of a suite of texts that should guide our interpretations. Nothing in the Appendix suggests that the counterexample of poetic fictions is supposed to ground a distinction between a feeling and a disposition. Instead, Hume give us a diagnosis of error that turns on acknowledging that perceptions can be distinguished by feeling and not just by vivacity. That diagnosis does not fit with the counterexample on Kamooneh’s reading.

We should figure out what Hume’s revised theory is not just by looking at the concession of minor error, but at the other texts on belief in the Appendix. Owen takes Hume’s point in the concession of minor error to be ‘that it is certainly feeling that differentiates beliefs from ideas, but that . . . feeling is not adequately characterized as the very thing which distinguishes impressions from ideas, i.e. force and vivacity’ (1999: 173). He worries that on his interpretation of Hume’s second thoughts on belief, according to which the connection between belief and vivacity is entirely broken, we lose ‘the picture of a unified reality of memory, sense impressions, and belief, and the characterization of probable reasoning as a species of sensation’ (1999: 174). Likewise, Broackes maintains that even after Hume comes up with his counterexample to the simple vivacity model, he inconsistently slides back into it, because he needs it ‘for his mechanics of belief’ (2002: 209). Frances Dauer argues along the same lines that Hume’s minor concession entails the collapse of ‘the hydraulic model’ according to which vivacity is conveyed across associated perceptions, ‘and this would entail a major revision to the Treatise. Precisely because of this, it is understandable that in the Appendix, Hume did not renounce the hydraulic model and face up to the consequences of his revised theory of belief’ (1999: 94). If Hume stopped thinking that vivacity had anything to do with belief, he could not appeal to the principle that vivacity spreads across associated perceptions to explain how we come to believe in unobserved matters of fact.

If Hume thought that his concession meant abandoning his theory of the origins of causal beliefs, he would not have made it share a paragraph with a remark on the perception of angles. A quotation from Cicero that Hume directs us to insert shows that he is still committed to the principle of the transfer of vivacity across associated perceptions when he writes the Appendix. In this passage, a character says that being in a place (for example, the site of Plato’s Academy or the old Roman senate house) associated with worthy men moves us more than when we merely read or hear about their deeds (T1.3.8.5n21). In another passage that I have already quoted, Hume says belief is a distinctive kind of feeling and then says he tries to explain this feeling ‘by calling it a superior force, or vivacity, or solidity, or firmness, or steadiness’ (T1.3.7.7).

The characterization of beliefs as conceptions with more force or firmness runs throughout the added material in the Appendix. When we have an impression of an object that has been customarily associated with another object, ‘the idea of its usual attendant immediately strikes us, as something real and solid’ (TApp.¶9). This idea, Hume tells us, ‘approaches the impression, from which it is deriv’d, in its force and influence’ (TApp.¶9, cf. App.¶3), which tells us that the impression has the relevant kind of force and influence as well. That is to say, impressions have doxastic vivacity in addition to any phenomenological vivacity they may have. As Don Garrett observes, ‘that memories and beliefs are perceptions having a feature in some degree that impressions have in a greater degree is one of Hume’s most distinctive doctrines’ (2015: 43). The new feeling of doxastic vivacity comes in degrees of vivacity, and the old system can apply to the new feeling. Doxastic vivacity can spread across associated perceptions. Hume’s later distinction between kinds of vivacity is an improvement in his view and not anything that undermines the integrity of his system.

3. New Arguments that Belief is a Feeling

Once Hume abandons the doctrine that perceptions of the same object only vary with respect to vivacity, he owes us new arguments for the claim that belief is a way that ideas are felt. In the beginning of the Appendix, Hume provides these required new arguments for the conclusion that belief ‘is merely a peculiar feeling or sentiment’ (TApp.¶2). He proceeds by arguing against two alternative views. First, he argues against the possibility that belief is some idea ‘such as that of reality or existence’ which we annex to our idea of the object (TApp.¶2). Second, he argues against the possibility that a belief is a distinguishable impression attached to the idea of an object (TApp.¶¶4–7). Some of these arguments are inscrutable in an interesting way. Other arguments in this collection can be understood as arguing for his theory of belief on the basis of its explanatory power.

Hume gives two arguments against thinking that belief is a matter of annexing the idea of existence or something like it to the idea of an object. First, ‘we have no abstract idea of existence, distinguishable and separable from the idea of particular objects’ (TApp.¶2).⁸ Second, and more generically, belief cannot be an idea that we attach to other ideas since, if it were, we could believe anything we wanted: ‘The mind has the command over all its ideas, and can separate, unite, mix, and vary them, as it pleases; so that if belief consists merely in a new idea, annex’d to the conception, it wou’d be in a man’s power to believe what he pleas’d’ (TApp.¶2). It is not within our power to believe whatever we please, so belief is not an idea annexed to our idea of a situation (Hansen 1988: 291; Broackes 2002: 188).

Before considering Hume’s arguments in the Appendix against the possibility that belief is an annexed impression, it is useful to take a step back and look at Hume’s discussion of belief as it appears in the Abstract. The Abstract appears in the Spring of 1740, about a year after Books I and II of the Treatise and about six months before Book III and the Appendix. The Abstract is a kind of advertisement for the Treatise in which Hume pretends to be an admiring reviewer, and he summarizes his argument for his account of causal inference, including eight paragraphs on the nature of belief.

The account of belief in the Abstract is pretty much the same as what we get in the Appendix. Belief is an indefinable feeling ‘which every one must be conscious of in his own breast’ (Abstract¶22). Poetry does not motivate in the same way not because of a lack of vivacity, but because such conceptions ‘never feel in the same manner as those which command our belief and opinion’ (Abstract¶22). One difference between the two treatments is that Hume does not make concessions of error in the Abstract, since he is pretending to be someone else and trying to drum up business.

In the Abstract, Hume says that ‘there are only two hypotheses’ to account for the nature of belief (Abstract¶19). Either it consists in an annexed idea, or it consists in ‘a different manner of conceiving an object’ (Abstract¶21). Hume gives two arguments against the first possibility, including an argument that appeals to the involuntariness of belief, and concludes that the second option is the right account (Abstract¶¶19–20).

In the Appendix, Hume considers a second rival hypothesis, that belief ‘consists in some impression or feeling, distinguishable from the conception’ (TApp.¶4). He gives what I would count as five arguments against the hypothesis that belief consists in an annexed impression.⁹

It is a little difficult to distinguish between the view that Hume is attacking and the view that Hume is defending here (Broackes 2002: 208). Hume wants to show that the difference between a belief and a mere conception ‘consists merely in a certain feeling or sentiment’ (TApp.¶2). The view he is criticizing is

that belief, beside the simple conception, consists in some impression or feeling, distinguishable from the conception. It does not modify the conception, and render it more present and intense: It is only annex’d to it, after the same manner that will and desire are annex’d to particular conceptions of good and pleasure. (TApp.¶4)

Notice the contrast with will and desire, which Hume supposes are impressions annexed to an idea rather than feelings which render the idea more present and intense.

If Hume’s point were that the annexation is loose enough to be broken by an act of will, then he could offer the same argument against this possibility that he did against the hypothesis that belief consists in an annexed idea. On this account, belief should be voluntary, but belief is involuntary, so this account must be mistaken. Hume does not, however, argue in this way, so that does not seem to be his point.

The basic distinction is that, on the annexed impression view that Hume criticizes, there are two perceptions involved in belief: an idea that provides the object and an attached impression that turns the idea into a belief. On the manner of conception view that Hume defends, there is one perception involved: an idea that represents an object and which is also infused with a feeling of belief.

Broackes denies that the one-perception view is available to Hume on his principles, since ‘a feeling of belief would be distinguishable and different from the conception believed; it must therefore in Hume’s view be capable of existing on its own’ (2002: 209). However, Hume could appeal to his theory of distinctions of reason (T1.1.7.17–8) to get around the difficulty.¹⁰ Similarly, we cannot quickly dismiss Hume’s two-perception accounts of will and desire by saying that we do not will to will and we cannot choose our desires. The relevant kind of separability is separability in thought and not separability by volition. The northern and southern hemisphere of Earth are separable in the relevant sense. When Hume uses the expression ‘annexed,’ he means a merely contingent connection, as Elizabeth Radcliffe observes (2018: 75). The contingency of a connection, however, does not mean that the connection is up to us.

Though the basic outlines of the metaphysics of the issue may be clear enough, it is not clear what difference the question makes to phenomenology or to the experimental science of man. What is the empirical upshot of whether one perception or two are involved in belief? On Hume’s preferred view, the feeling of belief is internal to the relevant idea whereas in the view that he is rejecting, the feeling is merely inseparably annexed to the idea. The point cannot be that, on the view that Hume is criticizing, the impression of belief is spatially external to the relevant conception, as if it might be to the left or the right. But if that is not the difference, it is not clear what the difference is. Before March 15^th, Cassius had a will and a desire that Caesar be assassinated. After March 15^th, Cassius had the belief that Caesar had been assassinated. If Cassius had looked inward, could he have seen what Hume wants us to see, a difference between one-perception beliefs and two-perception desires?

Of the five arguments that Hume offers under the heading of replies to the rival view, only two of them directly target it. Those criticisms seem as if they apply to his own view as well. The first argument is that reasoning produces beliefs, but reasoning does not produce impressions: ‘nothing ever enters into our conclusions but ideas, or our fainter conceptions’ (TApp.¶4). But if reasoning never gives rise to a new impression, then how can it give rise to a new feeling? Hume’s second argument is that it is ‘the subject of plain experience’ that ‘no distinct impression attends every’ belief. (TApp.¶4). Hume grants that there is a pleasurable feeling of tranquility when we move from agitated doubt to a settled, satisfying conclusion (TApp.¶4). Even so, Hume argues, in more usual cases of belief formation, for example, where we see only legs but infer the existence of the rest of the body, we believe, for example, that the whole human exists, without any such pleasant feeling (TApp.¶4). I am sympathetic to the doctrine that there is no distinctive feeling that attends each and every belief. Nevertheless, that is the view to which Hume commits himself by asserting ‘when we are convinc’d of any matter of fact, we do nothing but conceive it, along with a certain feeling, different from what attends the mere reveries of the imagination’ (TApp.¶2).

The annexed impression view is similar enough to Hume’s own view that it makes it difficult for him to launch criticisms of that view that do not also strike his own. This similarity raises the question of why he bothers. Why not follow the argumentative practice of the Abstract: criticize the view that belief is an attached idea, say that the alternative is an idea with feeling, and omit the metaphysical hairsplitting?

I think the explanation is Hume’s pride, which also explains why he says that most of the remarks in the Appendix are aimed at fixing infelicities of expression, and why he makes his admission of minor error share a paragraph with a point about the perception of angles. Hume wants to get things right, which is why he is correcting his account of belief in the Appendix. At the same time, he is a young man who has just written a big ambitious book, and he does not want his reader to think he is incompetent. I think that this pride not only makes him want to present his errors as not being especially egregious, but it also makes him minimize the scale of the corrections themselves. His original view that beliefs are relatively vivid ideas is a one-perception theory. Though the one-perception and two-perception versions of the doctrine that beliefs are ideas with a sui generis feeling are not very different, they are different. I think that Hume’s justifiable pride in the Treatise pulls him to make his revisions as small as possible, even when he cannot back up his position with arguments. That leads him to make the revised view a one-perception view, even if he does not have any useful arguments against the two-perception view.

Although the next three arguments are presented as criticisms of the rival account that beliefs might be attached impressions, they make sense as independent arguments that stand on their own, and so we can consider their merits independently of delicate differences between distinctive feelings and annexed impressions. These three arguments appeal to the explanatory virtues of Hume’s account. One appeals to his success in describing the causes of belief:

We can explain the causes of the firm conception, but not those of any separate impression. And not only so, but the causes of the firm conception exhaust the whole subject, and nothing is left to produce any other effect. An inference concerning a matter of fact is nothing but the idea of an object, that is frequently conjoin’d, or is associated with a present impression. This is the whole of it. Every part is requisite to explain, from analogy, the more steady conception; and nothing remains capable of producing any distinct impression. (TApp.¶6)

On Hume’s principles, constant conjunction generates a principle of association, and vivacity is transferred across associated perceptions (T1.3.6.15). Given these principles, after we observe two sorts of events as constantly conjoined, having the impression of one will lead to the doxastically vivified idea of the other. These antecedents will not, according to Hume, explain the production of the separable impression.

Hume also argues that his account can explain the consequences of belief, in a way that an analysis of belief as a separable impression cannot:

The effects of belief, in influencing the passions and imagination, can all be explain’d from the firm conception; and there is no occasion to have recourse to any other principle. These arguments, with many others, enumerated in the foregoing volumes, sufficiently prove, that belief only modifies the idea or conception; and renders it different to the feeling, without producing any distinct impression. (TApp.¶7)

Most importantly for Hume, belief motivates us in a way that idle contemplation does not (T1.3.10.3). Other examples of his drawing consequences from beliefs are scattered throughout the Treatise. For instance, judgments about ‘our own worth and character . . . are always attended with passion’ (T2.1.11.9). ‘When good is certain or probable, it produces joy,’ and, Hume continues, ‘When evil is in the same situation there arises grief or sorrow’ (T2.3.9.5). We do not take pleasure in listening to the conversation of people we think are liars ‘because those ideas they present to us, not being attended with belief, make no impression upon the mind’ (T1.3.10.5). As we have seen, Hume thinks that he can explain the paradox that we enjoy watching tragedies that present suffering on the stage, since tragic events produce different passions when we believe that those events occurred than they do when presented to us in fiction (T1.3.10.10).

Hume appeals to Ockham’s razor in his third argument against the view that belief is an impression annexed to an idea. He moves from the premise that the mind has ‘a firmer hold, or more steady conception of what it takes to be a matter of fact, than of fictions’ to the conclusion that that is all that belief is, since we should not ‘multiply suppositions without necessity’ (TApp.¶5). This argument will not persuade anyone who does not already think that beliefs are all and only those conceptions that are more firmly held than those of fiction. And, even if one does think that this account is extensionally correct, one might still hold out for a better, explanatory account for one’s definition of belief.

Hume’s thought in these arguments is that the explanatory virtues of his account justify us in believing it. I do believe that his account of causal inference is a great achievement in the history of psychology. The psychological principles he offers give a plausible first description of the mechanisms underlying our unreflective inferences concerning unobserved matters of fact, and that is a project that had not been considered before. Generally speaking, the explanatory virtues of a theory confirm the parts of the theory.

We might worry that by moving from phenomenological vivacity to doxastic vivacity, Hume’s account of belief becomes trivial. A belief is an idea with a certain sort of vivacity, where the vivacity amounts to belief in the state of affairs represented by the idea. Someone might reply to the triviality objection that Hume’s view still carries difficulties with it. He is still committed, first, to the doctrine that belief depends on an image, second, to the doctrine that belief requires a feeling, and, third, to the doctrine that belief is an inner representation. There are reasonable objections that a philosopher might levy against all three of these views.

Alternatively, someone else might point to the substance of the wider theory in which Hume’s account of belief is embedded. In addition to maintaining that belief is a sui generis feeling, Hume also maintains that this feeling is spread across associated perceptions, and that the association of cause and effect between perceptions is established by experiences of constant conjunction. Even if his account of belief were trivial, the wider theory of cognitive associationism would not be.

Having said that, let me add two caveats. First, it is important to distinguish between the explanatory power of Hume’s whole system and the explanatory power of his particular thesis that belief is a particular sort of forceful idea. The whole system consists of a web of causal connections between impressions and associations, between associations and the spread of belief, and between belief and passions and volitions. That system has, I think, a lot of explanatory power, but what exactly goes in the nodes of the web does not matter so much. The thesis that the belief in forthcoming pleasure causes joy and the belief in forthcoming pain causes grief is not original to Hume. It can be found in Hobbes and Locke (Leviathan122; Essay2.20.7), and it is independent of any particular account of belief.

Second, though Hume’s theory has explanatory virtues relative to having no theory at all, it is not therefore superior to every rival theory. In particular, it is not superior to the two-perception theory that belief arises from an annexed impression. Hume’s simile of vivacity being carried across associated perceptions as if by pipes or canals is evocative (T1.3.10.7; T2.2.9.14), and it effectively clarifies his doctrine that vivacity spreads across associated perceptions.¹¹ Hume is imagining vivacity spreading from our impressions of the cause to our ideas of the effect, making the ideas relatively more vivid. As an alternative, however, we could imagine that the pipes carry vivacity to a new, annexed impression (Flage 1990: 179–80). Even if it were true that animal spirits were more agitated when we believe than when we merely conceive, it still would not follow that beliefs are a kind of forceful conception of an idea rather than an annexed impression. To get to that conclusion, we would need to bridge principles connecting animal spirits and perceptions, principles that Hume is not inclined to offer and, if he did offer them, would only be arbitrary inventions.

Acknowledgments

I got helpful comments on an early version of this material from Mike Dacey, Cameron Buckner, and Talia Morag. I gave this paper at the Hume Society Conference in Prague in 2022, with excellent comments from Jason Fisette and excellent questions from the audience. I received useful feedback, both in and out of that session, from David Owen. I thank my anonymous referees for their comments as well. This work was supported by a grant from the National Endowment for the Humanities.

Competing Interests

The author has no competing interests to declare.