Integrating Hume's Accounts of Belief and Justification
Integrating Hume's Accounts of Belief and Justification
Abstract and Keywords
In Treatise I.iii.5–10, Hume's claim that association by the relation of cause and effect produces belief is often intertwined – though without his remarking on this fact – with the claim that belief based on causal inference is justified. To explain this, I offer the hypothesis that, in Hume's view, stability plays a double role: whether belief is justified depends upon considerations of stability, and fixity, a species of stability is also essential to belief itself. Hume identifies belief with steadiness, an infixed disposition, rather than (as tradition has it) vivacity or a lively idea. To establish that a state is a belief is thus to establish that it is stable, other things being equal. In one version of this stability‐based interpretation, the justification of a belief is a matter of its stability in the belief system of a fully reflective person; in a second version (which I favor), justification is to be assessed in terms of the degree to which the person who holds the belief is reflective.
III.1. A Puzzle in Regard to I.iii
In this chapter, I argue that a stability‐based interpretation of Hume's theory of justification resolves a puzzle in regard to Treatise I.iii. To begin, I present a significant textual phenomenon in Part iii: Hume's claim that association by the relation of cause and effect produces belief is often intertwined—though without his remarking on this fact—with the claim that belief based on causal inference is justified. To explain this, I offer the hypothesis that, in Hume's view, stability plays a double role. Whether belief is justified depends upon considerations of stability, and (a species of) stability is also essential to belief itself. In § III.2, I show that, for Hume, any belief is stable, in that it is steady or infixed. To establish that a state is a belief is thus to establish that it is stable, other things being equal. In § III.3, I factor in Hume's treatment of education. In § III.4, I observe that a belief, though steady in that it is infixed, might nevertheless be unstable in its influence on thought, the will, and action, owing to the presence of other beliefs with which it conflicts. I argue that the point of Hume's distinction between justified and unjustified belief is to call attention to circumstances in which a belief, though steady, is unstable in its influence, all things considered. I then show that this perspective is useful in understanding Hume's readiness at I.iv.7 to reject all belief, including belief based on causal inference, as unjustified. In § III.5, I distinguish two versions of my stability‐based interpretation. The issue is whether Hume takes the justification of a belief to be a matter of stability within the belief system of the person who holds the belief or to depend on the belief's stability within the belief system of a suitably reflective person. I address this and related issues in §§ III.6–7.
The titles of sections 5–10 of Part iii of Book I include “Of the nature of the idea or belief,” “Of the causes of belief,” and “Of the influence of belief.” In any interpretation, these sections have for their subject matter the nature, causes, and effects of belief. These topics remain central at sections 11–12, where Hume (p.61) extends his treatment to the probability of chances and the probability of causes, beliefs that fall short of proofs (§ II.1). There are also numerous passages in I.iii.5–10 and neighboring sections that register Hume's epistemic approval of beliefs based on causal inference (§§ II.1, 3). In I.iii.6, Hume allows that causal inference is “just” (T 89). In I.iii.9, he attributes causal inference to “the judgment” (T 108) and introduces a general contrast between the understanding and the imagination, making official his willingness to apply ‘reason’ and its cognates (‘reasoning’, ‘reasonable’) to association by the relation of cause and effect. In I.iii.11, and again in I.iii.13, he distinguishes degrees of probabilistic evidence (T 124, 153–54). In Part iii, the claim that causal inference is justified thus arises in tandem with the claim that causal inference results in belief. Yet, Hume does not give due recognition to the fact that these claims are different.
This phenomenon emerges in Hume's early statements of the causal theory of assurance (§ II.1). According to Hume's first statement of the theory in the Treatise, causal inference is the only relation that provides “assurance” (T 73; cf. EHU IV, 23) of the existence of objects that one does not perceive or remember. Similarly, causal inference “carries [the mind's] view” (T 82; cf. EHU IV, 22) beyond the objects of perception and memory. These formulations straddle a distinction between versions of the causal theory that advance a claim about the source of belief in the unobserved and versions that advance a claim about the source of knowledge of the unobserved. Since Hume is discussing the “nature” (T 94) and “causes” (T 98) of belief at pages 94–106, we might expect that, at this stage of the Treatise, the causal theory is advanced as a theory of the sources of belief in the unobserved. Yet, the formulation with respect to “assurance” is immediately preceded by the claim that relations other than cause and effect cannot “discover” (T 73) the existence of unobserved objects; this implies that the relation of causation can discover such existence. Similarly, cause and effect is the only relation that “informs us of existences and objects, which we do not see or feel” (T 74) and that can “lead us beyond the immediate impressions of our memory and senses” (T 89). These formulations connote success, what we call “knowledge,” not mere belief.1 So there are passages that favor the stronger version of the causal theory, an impression that is reinforced by the other evidence of Hume's endorsement of causal inference. It is surprising that the formulations that suggest knowledge emerge in a context where the official topic is (mere) belief and that Hume does not call attention to or signal this discrepancy.
Hume's discussion of the two systems of realities in I.iii.9, “Of the effects of other relations and other habits,” provides a striking instance of Hume's unacknowledged intermingling of claims about belief and claims about knowledge. The section discusses the relations of resemblance and contiguity. Hume grants (p.62) that these relations can enliven an idea and intensify an existing belief while insisting that they do not produce belief (cf. T 107–10). In the second paragraph of the section, Hume summarizes some previous results:
In the continuation of the paragraph, Hume raises a difficulty:
I have often observ'd, that, beside cause and effect, the two relations of resemblance and contiguity, are to be consider'd as associating principles of thought, and as capable of conveying the imagination from one idea to another. I have also observ'd, that when of two objects connected together by any of these relations, one is immediately present to the memory or senses, not only the mind is convey'd to its co‐relative by means of the associating principle; but like‐wise conceives it with an additional force and vigour, by the united operation of that principle, and of the present impression. And this I have observ'd, in order to confirm by analogy, my explication of our judgments concerning cause and effect. (T 107)
The third paragraph begins: “This is the objection; let us now consider its solution” (T 107).
But this very argument may, perhaps, be turn'd against me. . . . For it may be said, that if all the parts of that hypothesis be true, viz. that these three species of relation are deriv'd from the same principles; that their effects in inforcing and inlivening our ideas are the same; and that belief is nothing but a more forcible and vivid conception of an idea; it shou'd follow, that that action of the mind may not only be deriv'd from the relation of cause and effect, but also from those of contiguity and resemblance. But . . . we find by experience, that belief arises only from causation. (T 107)
The second paragraph sets the problem initially under discussion in I.iii.9. To respond, Hume needs to explain why causal inference, unlike resemblance and contiguity, produces belief. This explanation occupies the third through seventh paragraphs. Paragraphs three and four discuss belief based on the senses, memory, and the relation of cause or effect. Paragraphs five and six discuss resemblance and contiguity. Paragraph seven compares the relation of cause and effect to these other relations.
In the third and fourth paragraphs, Hume introduces the two systems of “realities.” I have provided a preliminary discussion of aspects of this material in §§ II.1, 3. Recall that the first system is based on the senses and memory alone; the second is based on custom or the relation of cause and effect and extends or supplements the first. In the discussion of the two systems, Hume seems to be saying not only that the relation of cause and effect, unlike resemblance and contiguity, produces belief, but that the relation of cause and effect produces justified belief.
The evidence for this is considerable. Hume writes that the mind “dignifies” the second system, as well as the first, “with the title of realities” (T 108). He writes, more directly, that the second of the two systems is “the object . . . of the judgment” (T 108). He also contrasts the ideas belonging to the second system with those “which are merely the offspring of the imagination” (T 108). This (p.63) discussion is naturally read to suggest that Hume approves epistemically of beliefs based on the relation of cause and effect. It is difficult not to be sympathetic with Passmore's comment on the passage: “The fact is that ‘reality’ . . . has a[n] honorific sense.”2 To reply to the objection at hand, Hume only needs to show that the relation of causation produces belief. Yet, Hume claims that causal inference produces justified belief. Hume seems to change the subject and to do so without notice.
One could go a certain distance toward avoiding this reading by taking the two‐systems passages in relative isolation. It might be thought that in saying that the mind dignifies the objects comprised in the second system “with the title of realities,” Hume simply means that we believe those objects exist. And it might be thought that when Hume writes of the second system as “the object . . . of the judgment,” he simply means that they are the objects of belief. There is substantial evidence against the view that Hume thus confines his claims to belief.
In the first place, there is a parallel development in the fifth and sixth paragraphs, where Hume explains why resemblance does not produce belief. (Here and elsewhere, I use ‘resemblance’ to stand for the relations of resemblance and contiguity.) Hume grants that resemblance produces a state that has affinities with belief, in which we “feign” the existence of an object. In the sixth paragraph, he offers epistemic claims: “such a fiction is founded on so little reason, that nothing but pure caprice can determine the mind to form it” (T 109), and “we . . . form a general rule against the reposing any assurance in those momentary glimpses of light, which arise in the imagination from a feign'd resemblance and contiguity” (T 110). These epistemic assessments seem curiously out of place. In the context of the objection he is considering, Hume needs to show that resemblance does not produce belief. In one compressed discussion, he also claims that the states that resemblance does produce are not justified.3 In paragraphs five and six, the states that fall short of belief are unjustified, much as in paragraphs three and four, where the states that constitute belief are justified.
In the second place, there is recalcitrant material internal to the discussion of the two systems of realities: “'Tis this latter principle [judgment], which peoples the world, and brings us acquainted with such existences, as by their removal in time and place, lie beyond the reach of the senses and memory” (T 108). “Brings us acquainted” implies success, implies that judgment leads to knowledge, not mere belief, about objects that lie beyond the reach of the senses and memory. This is of a piece with the handful of formulations cited above, which suggest that Hume subscribes to a causal theory of knowledge.
In the third place, in the footnote at pages 117–18, which closes I.iii.9, Hume reinforces his contrast between beliefs based on causal inference and those “which are merely the offspring of the imagination.” This is the note in which Hume contrasts “probable reasonings”—attributed to “reason” (T 118 (p.64) n. 1) or the understanding—with “those whimsies and prejudices, which are rejected under the opprobrious character of being the offspring of the imagination” (T 117 n. 1). The note repeats language in the two‐systems passage contrasting states that “are merely the offspring of the imagination” (T 108) with the products of perception, memory, and causal inference. The footnote is evidence that Hume intends the two‐systems passage to have a place in a sustained contrast between justified and unjustified belief.
In the fourth place, Hume's approval of causal inference in the two‐systems passage is continuous with his favorable discussion of causal inference throughout Part iii and well into Part iv of Book I. I have documented the evidence for this in § I.1. In the second paragraph of the present section, I reviewed evidence located in the immediate vicinity of Hume's discussion of the nature and causes of belief, including probabilistic belief. In addition, there is the I.iii.15 discussion of “rules” that are “proper” (T 173) for ascertaining causes and effects and that constitute a “LOGIC” (T 175). Also, Hume reiterates that causal inference is justified in I.iv.2 (T 216) and I.iv.4 (T 225).
In sum, though Hume's ostensible purpose in I.iii.9 is to explain why the relation of cause and effect produces belief, there is substantial evidence, within I.iii.9 and the surrounding sections, that Hume takes the relation of cause and effect to produce knowledge. Other commentators have observed that Hume seems to run together a theory of belief and a theory of justified belief. Passmore writes: “What set out to be a theory of belief, in something like the ordinary sense of the word, has become, with no explicit acknowledgment of that fact, a theory of what it is ‘rational’ to believe.”4 In this view, what is initially a theory of belief—with belief resulting only from association by the relation of cause and effect—is ratcheted up to a theory of justified belief. The textual phenomena that bother Passmore are real enough. In reading Part iii, one easily gains the impression that Hume's claim that belief arising from the relation of cause and effect is justified is somehow yoked to his claim that the relation of cause and effect produces belief. The question is how we are to explain this puzzling feature of Part iii.
Passmore maintains that Hume “vacillates” among different theories of belief. I believe there is substantial textual evidence against this, but there is room for a different worry. According to Passmore, Hume begins with a liberal definition, in which belief is a lively idea. Hume tries to explain why some beliefs so characterized, those arising from the relation of cause and effect, are more rational than others, and he concludes that no adequate explanation is forthcoming. Why does Hume then redefine ‘belief’, as a lively idea arising from the relation of cause and effect? In Passmore's account, finding himself unable to (p.65) distinguish between justified belief arising from causation and unjustified belief, Hume privileges lively ideas arising from causation by another means, declaring that they are the only genuine beliefs. This maneuver, were it Hume's, patently dodges the alleged difficulty. This is reason to reexamine the claim that Hume takes the problem of explaining why some beliefs are more rational than others as “unanswerable” (§§ I.3, 5, IV.5, V.3).5
The key to the puzzle is to suppose not that Hume identifies belief with justified belief but that, in Hume's view, establishing that the states produced by a psychological mechanism are beliefs is to establish that they are justified, other things being equal.6 On what grounds could Hume maintain this? It is easy to see, at least schematically, what is required. We need to locate a property that is necessary for the states produced by a psychological mechanism to constitute beliefs, such that to establish that the states are beliefs and thus have this property is also sufficient to establish that the beliefs are justified, other things being equal. In other words, there must be a property that plays a twofold role. The presence of the property must constitute a necessary condition for belief. In addition, establishing that the beliefs produced by a psychological mechanism have that property must constitute a sufficient condition for establishing justification, other things being equal. My claim is that stability is the property that plays this dual role, one within Hume's theory of belief, the other within Hume's theory of justification.7 In light of this, I proceed in two stages. In §§ III.2 and III.3, I show that, in Hume's view, stability is a necessary condition for belief. In § III.4, I address the nature of Hume's interest in stability in the context of his theory of justification.
III.2. Steadiness and Infixing in Hume's Theory of BeliefIII. Integrating Belief and Justification III.2. Steadiness and Infixing
I begin with a sketch of the role of stability in Hume's theory of belief. Tradition in Hume interpretation has it that beliefs are lively ideas. In my interpretation, beliefs are steady dispositions. It is a commonplace that Hume uses a cluster of closely related terms—‘vivacity’, ‘vividness’, ‘intensity’, and ‘liveliness’—to characterize belief. This vivacity cluster, however, is prima facie distinct from a second cluster of terms—‘firmness’, ‘solidity,’ and ‘steadiness’, together with ‘fast’, ‘firm’, ‘settled’, ‘solid’, and ‘steady’ (T 97, 105, 106, 108, 116, 121, 624, 625, 626, (p.66) 627, 629, 631)—that also has a prominent role in Hume's discussion of belief.8 For ease of exposition, I use ‘steadiness’ (and its cognates) to stand indifferently for all of the terms in this steadiness cluster. (A number of other terms—‘force’, ‘strength’, and ‘vigor’—at least in some of their occurrences, perhaps belong in this cluster as well.) Although Hume suggests that the terms in the two clusters may be used interchangeably (T 629), we need to disentangle them in order to attribute to Hume a coherent theory of belief. Hume contrasts steady ideas with ideas that are “momentary” (T 110), “floating” (T 116), and “loose” (T 97, 106, 116, 123, 595, 624, 625; cf. 110). The terms in the steadiness cluster often refer to a kind of staying power. I suggest that steadiness plays a more fundamental role than vivacity in Hume's theory of the nature of belief.9
As I do not wish to deny that the vivacity cluster is entrenched in Hume's texts, some stage setting might be helpful. I am sympathetic with Stroud's suggestion that Hume “takes for granted the theory of ideas,” “a theory that he adopts . . . from his predecessors,” though “he never gives any arguments in support of it.” It is a central component of this theory that “thinking and mental activity generally consists in the presence before the mind of perceptions,” so that they are identified with occurrent mental states. As Stroud notes, the theory of ideas exerts considerable pressure on the direction in which Hume develops his theory of belief. What is more, as I view matters, Hume seeks to integrate his associationist ambitions (§ I.6) into the tradition's theory of ideas, with the result that he takes associationism to involve the transmission of an occurrent property (vivacity or liveliness).10 Hume's most self‐conscious thinking about the nature and causes of belief proceeds within an associationism grafted onto the theory of ideas. This framework determines the character of much of the text—not only Hume's formulaic summaries of his theory of belief but also the more extended applications of that theory and the related development of his associationist psychology. (Such textual phenomena explain the allure of the traditional (p.67) interpretation of Hume's theory of belief.) At the same time, Hume develops a position in which beliefs are steady dispositions.11 The theory of ideas‐cum‐associationism, however, so permeates his thinking as to impede Hume's appreciation of the role he assigns to steadiness. The psychological hold of the inherited theory of ideas explains the centrality of vivacity in Hume's discussions even as he advances a view that breaks out of this mold.12
In providing an interpretation of Hume's position, I will first show that beliefs are steady and then show that they are dispositions. Hume's conception of steadiness is closely connected to his discussions of the ways in which ideas are infixed. There is ample evidence that when an idea is infixed, the result is a belief. At Treatise 86, 109, and 225 (cf. T 99, 121), ideas are infixed with force or vigor, or enlivened, and Hume's formula—though it proves inadequate—is that belief is a lively or forceful idea. Furthermore, Hume writes that in belief the mind “fixes and reposes itself” (T 624) on its conceptions, or “fixes and reposes itself in one settled conclusion and belief” (T 625). I suggest that infixing is a process that produces steadiness and hence belief.
In the course of the Treatise, Hume specifies particular mechanisms as ones that infix belief. In I.iii.5, Hume maintains that “belief or assent . . . always attends the memory and senses” (T 86). He writes of “custom and habit having . . . the same influence on the mind as nature, and infixing the idea with equal force and vigour” (T 86; cf. 225). In context, “nature” refers to the senses and memory, so that custom, in addition to the senses and memory, produces steady ideas and hence infixes belief. In a similar vein, Hume writes of “the eternal establish'd persuasions founded on memory and custom” (T 632).13
Hume identifies “CUSTOM” (T 102) or “habit” (T 105) with “every thing . . . which proceeds from a past repetition, without any new reasoning or conclusion” (T 102). In the case of inference based on the relation of cause and effect, (p.68) the repetition consists in the frequent observation of resembling pairs of objects (T 109). Custom also includes the repetition of a mere idea (§ III.3). Repetition produces and thus explains steadiness. Hume writes that a “principle [that] has establish'd itself by a sufficient custom” also “bestows an evidence and firmness on any opinion, to which it can be apply'd” (T 105, emphasis added). In the Appendix, Hume observes that “it must be allow'd, that the mind has a firmer hold, or more steady conception of what it takes to be matter of fact, than of fictions” (T 626). He adds in the next paragraph that by the association of “frequently conjoin'd” objects “we can explain the causes of the firm conception” and that these causes “exhaust the whole subject” (T 626). Hume appeals to a higher‐order habit (T 104–5; cf. 131, 173–74) to defend this hypothesis against the objection that it is not compatible with our forming a belief based on a single experience (§ I.5). Repetition or frequent conjunction—as well as the senses and memory—give rise to firmness or steadiness in belief.
My claim that beliefs are steady may be strengthened on the basis of I.iii.7, “Of the nature of the idea or belief.” Hume writes: “We may mingle, and unite, and separate, and confound, and vary our ideas in a hundred different ways; but ‘till there appears some principle, which fixes one of these different situations, we have in reality no opinion” (T 96). We have not achieved opinion unless we have fixed on a particular idea; the infixing of an idea is essential to belief. Since infixing is a process that results in steadiness, steadiness must itself be essential to belief.
There is confirmation of the role of steadiness in belief in II.iii.10, where Hume offers observations about the relationship between uncertainty or doubt and certainty or belief (§ IV.5). These include claims about the “nature” of doubt and belief. Hume writes: “'Tis the nature of doubt to cause a variation in the thought,” so that doubt involves “instability and inconstancy” (T 453). By contrast, belief or certainty involves “fixing one particular idea in the mind, and keeping it from wavering in the choice of its objects” (T 453). In this passage, Hume treats belief and certainty as the opposites of doubt and uncertainty. If it is of the nature of doubt to cause “variation,” “instability,” and “inconstancy” in thought, then the property of “fixing one particular idea in the mind, and keeping it from wavering” must be of the nature of belief. It is in the nature of belief, unlike doubt, to be infixed and steady.
To this point, I have been writing as if steadiness is a property of an idea. If ideas are conscious or occurrent states, however, they are obviously unsteady—they come and go, or change, abruptly. At the same time, there is substantial evidence that steadiness is crucial to Hume's account of belief, evidence that will be reinforced as this section proceeds. What is needed is a framework that can accommodate steadiness as a property of belief. Steadiness, I suggest, is best construed not as a property of an idea but as a property of a disposition.
Hume often invokes mental dispositions or propensities that cannot plausibly be identified with occurrent states. John Bricke observes that in the context of Hume's moral theory a person's character is a “durable” (T 575) quality, which may exist in the absence of its typical effects (T 411, 584, 585). In the course of this study, we encounter the mind's propensity to attribute identity to (p.69) related objects (§§ V.1–2), to add new relations to related objects (§§ V.3–5), and to continue in a train of thought (§§ VI.1–3). Indeed, as Robert Paul Wolff has argued, the association of ideas by the relation of cause and effect is a second‐order disposition, a disposition to develop under specified conditions a first‐order disposition (for example, to expect fire to be accompanied by smoke).14 Hume's I.i.7 development of Berkeley's position on abstract ideas relies on a “power” or “readiness” (T 20) to call up any of a number of related ideas.
To return to the case at hand, Hume's tendency to treat belief as a disposition, as well as an occurrence, is well known. For example, a passage in the Appendix begins on a tack that invites taking belief to be an occurrent or conscious state: “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” (T 629). The passage continues: “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” (T 629; cf. EHU V, 40–41, 46, and A 654). Here, belief is characterized with reference to its effects. This suggests a dispositional account. There are a number of similar passages: belief “gives [ideas] more force and influence; makes them appear of greater importance; infixes them in the mind; and renders them the governing principles of all our actions” (T 629), and the mind “is more actuated and mov'd by” beliefs than by states that fall short of belief (T 624).15
Such notions as “steadiness” apply more naturally to dispositions than to occurrent states (§ I.1). Much as a calm emotion is a “settled principle of action” (T 419), indeed such that “repeated custom and its own force have made every thing yield to it” (T 419), we can take the claim that it is essential to belief that it is infixed, and hence steady, to apply to dispositional beliefs. This fits nicely with Hume's explanation of steadiness, as the result of repetition or conditioning. Dispositions that are not infixed do not qualify as beliefs (cf. T 453, 629). A dispositional belief is a steady disposition to characteristic manifestations or typical effects on thought (hence including occurrent or conscious states), the passions, and action (verbal and nonverbal).16 Such steadiness is a species of stability, the stability of a disposition that is infixed.
(p.70) Construing beliefs as dispositions is not incompatible with Hume's associationist ambitions. Hume invokes association among dispositions at pages 204–5 of I.iv.2 (§ V.2). Fogelin notes that though “ideas are natural candidates to be the subject of associational laws,” “Laws of Association do not have to be applied to ideas—they can also be applied to dispositions.”17 Association might serve in the first instance to strengthen dispositions and precipitate their characteristic effects, so that the occurrence of enlivened or intensified occurrent states is a by‐product of associative mechanisms operating on underlying dispositions.
Hume can also admit occurrent beliefs, in the sense of conscious manifestations of a dispositional belief. These occurrent manifestations include lively or vivacious ideas. Though these lively ideas are not steady, steadiness is located at the level of the underlying dispositions that they manifest. In this account, liveliness or vivacity is one of the characteristic manifestations of belief. (One, but only one. My interpretation does not imply that a steady disposition merely to have an idea, which thus recurs, constitutes belief; there must be a disposition to the full range of effects on thought, the passions, and action characteristic of belief.) It is here, in application to occurrent beliefs, that terms in the vivacity cluster have their proper home. An occurrent belief, strictly speaking, is an occurrent manifestation of a dispositional belief.
H.H. Price takes note of an apparent contradiction in Hume's statements of his theory of belief. Hume holds that the liveliness in belief derives from a relation to a present impression, but also that an idea can be lively or forceful when it is not related to a present impression at all.18 Specifically, Hume holds that poetical enthusiasm does not result in belief, even though it can produce ideas that are as lively and vivacious as occurrent beliefs:
If belief is a vivacious idea, how can the lively products of poetical enthusiasm fail to count as beliefs? The answer is that the verbal and nonverbal behaviors and internal episodes that manifest dispositional belief can also arise from other sources, thereby mimicking belief. In these cases, we have pseudobeliefs, what Hume calls “counterfeit belief” (T 123) or “the mere phantom of belief or persuasion” (T 630). Hume writes that “the least reflection dissipates the illusions of poetry” (T 123).19 This is a point about the steadiness of the underlying (p.71) dispositions. Lively ideas, no matter how intense, constitute occurrent beliefs only when they manifest a steady disposition. This dissolves Price's apparent contradiction.
How great soever the pitch may be, to which this vivacity rises, . . . 'tis still the mere phantom of belief or persuasion. . . . A poetical description may have a more sensible effect on the fancy, than an historical narration. . . . It may seem to set the object before us in more lively colours. (T 630–31)
My interpretation does not preclude Hume's use of terms in the steadiness cluster to refer to occurrent states. We have seen that Hume uses ‘calm’ and ‘violent’ both to refer to the properties of being settled and volatile, respectively, and to refer to the occurrent awareness produced by those properties, felt calmness and felt violence (§§ I.1–2). Hume's theory of belief follows a similar pattern. In his discussion of the influence of belief (T 118–19), ‘unsteadiness’ refers to a dispositional property, staying power. Often, however, terms in the steadiness cluster are used to refer to an occurrent experience or feeling. In this usage, perhaps most explicit at Treatise 629 (“An idea assented to feels different from a fictitious idea”), steadiness is a phenomenal property. These two usages are connected in an obvious way. Many of the terms in the steadiness cluster do double duty, at times referring to a dispositional property of an idea and other times referring to the way that property feels when manifested in consciousness. Indeed, Hume's tendency to suppose that we are introspectively aware of mental dispositions can obscure the role of dispositions in his analysis of belief.20
Hume is not himself attentive to his tendency to treat belief both as an occurrence and as a disposition. He often runs together terms in the steadiness and vivacity clusters (for example, at T 97, 106, 629), taking terms from the different clusters to refer to “the same quality,” whatever we “call it” (T 106), so that they are “intended only to express” the same “act of the mind” (T 629). This is not surprising. An occurrent belief is itself a manifestation of a dispositional one. Furthermore, such locutions as “something felt by the mind” (T 629) invite an identification of belief with the feeling that manifests the disposition. These locutions do not, however, require the identification; the occurrent feeling can be taken to be the awareness of a dispositional belief. Finally, there is room to confuse distinct conscious states: vivacity or liveliness, on the one hand, and steadiness, on the other. The confusion is perhaps sustained by Hume's use of additional terms—‘force’, ‘strength’, ‘vigor’, ‘intensity’—that are sufficiently ambiguous to be construed either in terms of steadiness or vivacity.21 Beyond the general influence of the theory of ideas, these more specific factors are obstructions to greater clarity on Hume's part about his position.
My distinction between the two clusters of terminology regiments distinctions required by Hume's views. (A term that is typically associated with the steadiness cluster might, on a particular occasion of use, function as if it belonged to the vivacity cluster, and vice versa.) To provide a consistent interpretation, we must distinguish the concepts represented by the two clusters, and we must pay more than lip service to dispositional strands in Hume's theory. We must systematically identify beliefs with steady dispositions and distinguish (p.72) between dispositional beliefs and their manifestations, occurrent and otherwise. Absent such a regimentation of Hume's texts, there is no prospect of making sense of his discussions of poetical enthusiasm, where vivacity is held to be insufficient for belief. Also, there is no prospect of plausibly explaining how fixity and steadiness are essential to belief.22
Hume's I.iii.9 discussion of the effects of the relations of resemblance and contiguity fits nicely with the interpretation in which beliefs are steady dispositions. At paragraph six, Hume begins an examination of resemblance when it operates “single” (T 109), on its own. Resemblance produces only “momentary glimpses of light” (T 110), and “there is no manner of necessity for the mind to feign any resembling and contiguous object; and if it feigns such, there is as little necessity for it always to confine itself to the same [idea], without any difference or variation” (T 109). Hume sums up: “That principle being fluctuating and uncertain, ‘tis impossible it can ever operate with any considerable degree of force and constancy” (T 109). Resemblance produces an underlying disposition that is fluctuating or unsteady and hence fails to constitute belief.23
Hume takes care to contrast the effects of association by the relation of resemblance with those of association by the relation of cause and effect. He has already noted in the discussion of the second system of realities that “by their force and settled order, [ideas] arising from custom and the relation of cause and effect . . . distinguish themselves from . . . other ideas” (T 108). This observation immediately precedes the explanation at paragraphs five and six of why resemblance does not produce belief. Hume returns to the contrast at paragraph seven:
Again, notions related to steadiness are prominent. Whereas the effects of resemblance are “momentary,” “fluctuating,” given to “variation,” and lacking in “constancy,” the effects of the relation of cause and effect are “fixt,” “settled,” and “invariable.”24 I have previously reviewed evidence that the repetition that underpins (p.73) causal inference has the advantage of giving rise to steadiness. Hume's comparison of the relation of resemblance to that of cause and effect confirms that steadiness is essential to belief.
The relation of cause and effect has all the opposite advantages. The objects it presents are fixt and unalterable. The impressions of the memory never change in any considerable degree; and each impression draws along with it a precise idea, which takes its place in the imagination, as something solid and real, certain and invariable. (T 110)
Our original puzzle is to explain why the claim that causal inference results in belief and the claim that causal inference is justified frequently find themselves in close conjunction with each other in Part iii, though Hume does not explain the connection between them (§ III.1). What begins as a causal theory of assurance is reformulated as a causal theory of knowledge, though without Hume calling attention to the difference. An explanation of why causal inference, unlike resemblance, produces belief brings with it the claim that causal inference, unlike resemblance, produces states that are justified. Again, Hume does not call attention to the difference in these claims. To make sense of these phenomena, we must locate a property that Hume ascribes to belief and that Hume might take to be germane to justification.
The only salient candidate is steadiness. Hume extensively discusses the steadiness of belief in I.iii.5–10 (T 84–123) and associated material in the Appendix. Within these sections, Hume claims that infixing is essential to belief, that custom infixes belief, and that custom or repetition gives rise to steadiness. The steadiness that results from the senses, memory, or repetition is necessary for belief. This interpretation is required to make sense of Hume's repeated statements that beliefs result from a process of infixing and that they are fast, firm, settled, solid, and steady. It is also required to explain how the results of poetical enthusiasm can exceed belief in vivacity and yet amount only to pseudobelief. These grounds for taking Hume to hold the view that stability, in the form of steadiness, is essential to belief are independent of his theory of justification.
In the second of its roles, stability is connected to justification. The causal theory of belief is thus also formulated as a causal theory of knowledge. The claim that causal inference is justified—that it “brings us acquainted” with objects we have not perceived and is due to “judgment”—accompanies the claim that causal inference is due to custom or repetition. This discussion is followed by the explanation at paragraphs five and six of why resemblance does not produce belief, an explanation that stresses the unsteady character of dispositions arising from this relation and hence confirms the role Hume assigns to steadiness. In the course of this discussion, Hume writes that there is “little reason” (T 109) to feign objects based on this relation and that we form a “general rule” (T 110) against doing so.
When Hume writes in paragraph seven of I.iii.9 that, in comparison to the relation of resemblance, “the relation of cause and effect has all the opposite advantages,” the “advantages” he has in view are twofold, reflecting stability's interrelated roles in his theories of belief and justification. Justified beliefs result from mechanisms that tend to produce stability in belief, and beliefs are steady dispositions. This account serves to explain the intimate connection between claims about belief and claims about justification in Part iii. The explanation is that to establish that the states produced by a psychological mechanism are beliefs, and hence stable in the sense that they are steady or infixed, is sufficient to (p.74) establish that they are justified, other things being equal (§ III.1). My claim that whether a belief is justified depends upon considerations of stability thus integrates Hume's theory of justification with his theory of belief. In §§ III.4–6, I consider in more detail the role of stability in Hume's theory of justification and its relationship to the steadiness of belief.
III.3. The Bearing of Hume's Treatment of EducationIII. Integrating Belief and Justification III.3. The Bearing of Education
Section I.iii.9 is titled “Of the effects of other relations and other habits.” Hume discusses the effects of other relations, relations other than cause and effect—resemblance and contiguity—at pages 106–15 (§ III.2). He discusses the effects of other habits, habits other than those based on the frequent observation of conjunctions, at pages 115–17, in the final four paragraphs of the section.
Hume effects the transition to this material as follows: “The present hypothesis will receive additional confirmation, if we examine the effects of other kinds of custom, as well as of other relations” (T 115). The present hypothesis consists in Hume's views regarding the “nature” (T 94) and “causes” (T 98) of belief, the respective topics of the preceding two sections, I.iii.7 and I.iii.8. Hume has defined “custom” as “every thing . . . which proceeds from a past repetition, without any new reasoning or conclusion” (T 102). Hume observes that “custom . . . may operate upon the mind . . . after two several ways” (T 115), so that there are “two kinds of custom” (T 116). One way in which custom operates on the mind is in producing belief based on causal inference, where “in all past experience we have found two objects to have been always conjoin'd together” (T 115–16).
Hume turns to “examine the effects of other kinds of custom” (T 115). Custom includes the repetition of a mere idea, as well as the repeated observation of resembling pairs of conjoined objects. Hume thus invites us to suppose that “a mere idea alone . . . shou'd frequently make its appearance in the mind” (T 116). Such an idea “must by degrees acquire a facility and force; and both by its firm hold and easy introduction distinguish itself from any new and unusual idea” (T 116). It will “distinguish itself” just as ideas based on the relation of cause and effect “distinguish themselves” (T 108), that is, it will be a belief. This confirms Hume's hypotheses about the nature and causes of belief. Both kinds of custom involve repetition, which produces a “firm hold” on its object.25
Hume has previously introduced an example of custom due to the repetition of a mere idea. In I.iii.5, he writes of liars, “who by the frequent repetition of their lies, come at last to believe and remember them, as realities; custom and habit having in this case, as in many others, the same influence on the mind as nature, and infixing the idea with equal force and vigour” (T 86). In the case of liars, frequent repetition infixes an idea and results in belief. In I.iii.9, “liars, by the frequent repetition of their lies, come at last to remember them” (T 117).
(p.75) Though Hume reminds the reader of the case of liars, the I.iii.9 discussion of other habits has for its focus the repetition of an idea in education. Here Hume has in mind the mere repetition of ideas, without the source of the ideas being an accredited or corroborated authority for the person who receives the education.26 Were the source accredited, we would have an instance of “testimony,” as described by Hume just a few pages earlier in I.iii.9 (T 113)—a species of causal inference, which can establish facts about human veracity (§ II.1). If education were an instance of testimony, it would be an example of cause‐and‐effect reasoning, rather than an example of “other habits” or “other kinds of custom.” The contrast is clear when Hume writes that the “influence” of education “on many occasions prevails over that which arises from the constant and inseparable union of causes and effects” (T 116). Hume's “education” should not be assimilated to testimony; it is best thought of as inculcation.
What is the influence of education? Here again, “the frequent repetition of any idea infixes it in the imagination” (T 116) and thus produces “belief” (T 116) and “opinions” (T 116, 117). These claims are also implicit in I.iii.10, where Hume writes that “education infixes . . . opinion” (T 121).27 In sum, Hume's general discussion of a second form of custom, repetition of a mere idea, confirms that fixity is essential to belief and that repetition explains the presence of fixity.
Even putting aside issues about whether belief is dispositional, standard formulations of Hume's theory of belief are not fully general. It is not apt to say that belief (apart from belief that arises in perception and memory) is associated with a present impression by the relation of cause and effect. This formulation leaves out the cases of liars and education. Hume holds that belief arises from any repetition. This is the maximally determinate or specific formulation that does not exclude cases of genuine belief. Hume provides this formulation, unsurprisingly, in I.iii.9: “belief is an act of the mind arising from custom” (T 114), that is, “which proceeds from a past repetition.” Cast in terms of the dispositional account, belief is a disposition, arising from any repetition, to characteristic manifestations, including lively ideas. Beliefs can arise from repetition directly—as with liars and in education—or indirectly, triggered by a present impression against the background of an observed repetition or constant conjunction.28 (As we shall see in § V.2, Hume relies in Book iv of Part I on a highly (p.76) elastic understanding of the sense in which “repetition” or “custom” must be implicated in producing belief.)
The theory that belief arises from any repetition is an extension of the theory that belief arises specifically from repeated observation of resembling pairs of conjoined objects. The generalization is important to Hume; the “other kinds of custom” provide “additional confirmation” of “the present hypothesis.” Consideration of education confirms Hume's account of the nature and causes of belief.29 The point of the discussion of other kinds of custom is to confirm that beliefs are steady and infixed by repetition. The repetition of a mere idea, like the repeated observation of resembling pairs of objects and unlike the relation of resemblance, produces belief.
If belief arises from any repetition, what are we to make of Hume's repeated assertion that all assurance about the unobserved arises specifically from causal inference (§ II.1)? There are a number of points to make here. In the first place, the I.iii.9 discussion of the effects of other habits provides an explicit generalization of the theory that belief arises from repetition in observing resembling pairs of objects. As we have seen, Hume writes that “belief arises only from causation” in the two‐systems passage at pages 107–8, just nine pages before the discussion of education in the same section. Education confirms the role of repetition, confirmation that is more dramatic insofar as it has not previously been discussed. This helps explain why in earlier material Hume by and large sticks to the causal theory of assurance and to the claim that belief must be related to a present impression.
In the second place, recall that the causal theory of assurance is introduced in a technical sense in the Treatise (§ II.2). The thesis, strictly speaking, is that of the three philosophical relations that do not depend solely upon ideas (identity, situations in time and place, and causation), only causation gives us assurance of the existence of unobserved objects.30 This is Hume's clear meaning at pages 73–74, one that carries over to a number of subsequent statements of the thesis, including, “the only connexion or relation of objects, which can lead us beyond the immediate impressions of our memory and senses, is that of cause and effect” (T 89), and “the relation of cause and effect . . . is the only one that can assure us of matter of fact” (T 193). Now, suppose an object is present to the senses. Suppose further that one believes on the basis of education, in the sense of inculcation, that another (kind of) object is its usual attendant. One might then believe that an instance of the other kind of object is also present. This belief, however, does not arise from one of the relevant philosophical relations. In (p.77) the case of sheer inculcation, the testimony bears no philosophical relation to the object present to the senses. This case thus falls outside the scope of the initial thesis about the relation of causation. This helps explain Hume's insensitivity to apparent inconsistencies in his text.
There is a third point, apart from this textual detail. I have stressed that Hume's claim that all “assurance” about the unobserved arises from the relation of cause and effect is ambiguous between a causal theory of belief and a causal theory of knowledge (§ III.1). At the same time, a number of formulations—where the relation of cause and effect can “discover” (T 73), “lead us” (T 89) to, “informs us of” (T 74), and “brings us acquainted with” (T 108) objects that have not been observed, matters of fact that can be “proved” (A 654) by that relation—clearly imply a thesis about knowledge, not mere belief. This is what we should expect in a constructive response to Locke's account of the limitations of sensitive knowledge (§ II.1). If we read Hume's statements of the theory in terms of a claim about knowledge, rather than mere belief, there is no inconsistency when Hume allows that education produces belief. As we shall see shortly, Hume does not think inculcation produces knowledge. It is worth noting that in describing the products of the repetition of an idea, Hume writes of “belief” (T 116), “opinions” (T 116, 117), “assent to any opinion” (T 118), and infixing “opinion” (T 121), but not of “assurance.”
The fact that Hume's statements of the causal theory of assurance straddle claims about mere belief and about knowledge is one manifestation of the textual phenomenon that set the problem for this chapter. In Part iii, the claim that causal inference is justified seems interconnected with the claim that causal inference results in belief, though Hume does not take note of any gap between these claims. My hypothesis is that we should attribute to Hume the view that establishing that the states resulting from a psychological mechanism are beliefs in the sense that they are steady or infixed is to establish that they are justified, other things being equal (§ III.2). This hypothesis implies a significant asymmetry in the requirements for establishing that a belief is justified, other things being equal, and establishing that a belief is not justified, all things considered. Once it has been established that a state results from a mechanism that produces belief, no separate or additional argument is required to establish that the belief is justified, other things being equal. The explanatory power of my hypothesis derives from this implication, for what is puzzling is that Hume moves, without further argument, from the claim that causal inference produces belief to the claim that it produces justified belief.31 By contrast, establishing that a belief is not justified, all things considered, carries a heavier argumentative burden. In cases where Hume seeks to establish that a belief is not justified, all things considered, he will need to provide a separate or additional argument, beyond an argument for its status as the product of a belief‐forming mechanism.
(p.78) This is precisely Hume's procedure in the case of education. In the final sentence of I.iii.9, Hume offers a basis for distinguishing the epistemic status of belief due to cause and effect and belief due to education:
Similarly, Hume writes in the first sentence of I.iii.10 that education is “disclaim'd by philosophy, as a fallacious ground of assent to any opinion” (T 118). In these passages, Hume does distance himself from the claim that beliefs based on education are unjustified, attributing this view to “philosophers” and “philosophy.” This is the first instance of this device, which we also find at Treatise 140 and 225. I have offered an explanation of the ambivalence in these texts. Hume comes to conclude—in I.iv.7—that his favored epistemological theory cannot sustain his own pretheoretical epistemological distinctions. In anticipation of this later result, Hume sometimes distances himself from epistemological distinctions to which he himself inclines (§ I.4).
But as education is an artificial and not a natural cause, and as its maxims are frequently contrary to reason, and even to themselves in different times and places, it is never upon that account recogniz'd by philosophers; tho' in reality it be built almost on the same foundation of custom and repetition as our reasonings from causes and effects. (T 117)32
At the same time, Hume does set out a list of the grounds on which philosophers reject beliefs based on education, while recognizing beliefs based on cause and effect. Though his discussion is decidedly cryptic—confined to a single sentence—Hume treats the grounds he cites as if they might well suffice (were it not for the destructive results down the road in I.iv.7) to support the philosophers' distinction. For all his argument shows at this stage, he is himself in the company of the philosophers who reject education. Though Hume distances himself from the philosophers' views about education, he has provided no rationale for doing so. The final sentence of I.iii.9 sets out grounds for rejecting beliefs based on education (T 117–18 n. 1). It is in the footnote to this sentence that Hume consolidates his own normative distinction between the understanding, the approved associative mechanisms that underpin demonstrative and probable reasoning, and the imagination, associative mechanisms that Hume holds in disfavor (§ II.3). The constructive project of sustaining epistemic discriminations thus survives Hume's brief discussion of the grounds for rejecting beliefs based on education. Hume has in place a general account of belief, as arising from any repetition, together with a commitment to vindicating discriminations between justified and unjustified belief.
Hume argues at pages 115–16 that education produces belief and appends a separate argument at page 117 to explain why beliefs based on education are not justified. At a stage where Hume wants to establish that a class of beliefs is not justified, all things considered, he explicitly recognizes a distinction between belief and justified belief. The overall structure of Hume's treatment of (p.79) education thus provides new data in support of attributing to Hume the view that to establish that the states produced by a psychological mechanism are beliefs is to establish that they are justified, other things being equal. Hume's I.iii.9 discussion of kinds of custom other than that involved in causal inference confirms my interpretation of Hume's theory of justification and my account of his theory of belief.
III.4. The Natural Function of BeliefIII. Integrating Belief and Justification III.4. The Natural Function of Belief
It remains to consider in more detail the role of stability in Hume's theory of justification and its relationship to the steadiness of belief. Why does stability, other things being equal, matter to Hume? In the second and third paragraphs of I.iii.10, “Of the influence of belief,” Hume explains why steadiness is important:
Nature . . . seems to have carefully avoided the inconveniences of two extremes. Did impressions alone influence the will, we should every moment of our lives be subject to the greatest calamities; because, tho' we foresaw their approach, we should not be provided by nature with any principle of action, which might impel us to avoid them. On the other hand, did every idea influence our actions, our condition would not be much mended. For such is the unsteadiness and activity of thought, that the images of every thing, especially of goods and evils, are always wandering in the mind; and were it mov'd by every idle conception of this kind, it would never enjoy a moment's peace and tranquillity.
Were we moved to action only by sense impressions and memories, we could not make inferences to future events and thus could not make plans to avoid prospective pain or to enjoy prospective pleasure. Were we moved to action by every idea of pain or pleasure, we would not pursue a coherent plan of action over time, since our mental activity is unsteady, our ideas wander.33 Nature therefore provides a medium between these two extremes, so that some—but not all—ideas influence the will and action. These ideas are beliefs, nature's provision for a steady influence on the will and hence on action.
Nature has, therefore, chosen a medium, and has neither bestow'd on every idea of good and evil the power of actuating the will, nor yet has entirely excluded them from this influence. Tho' an idle fiction has no efficacy, yet we find by experience, that the ideas of those objects, which we believe either are or will be existent, produce in a lesser degree the same effect with those impressions, which are immediately present to the senses and perception. The effect, then, of belief is to raise up a simple idea to an equality with our impressions, and bestow on it a like influence on the passions. (T 118–19)
Hume's conception of the natural function of belief helps to explain the importance he attaches to the distinction between establishing that belief is (p.80) justified, other things being equal, and establishing that belief is justified, all things considered.34 As we shall see in § III.5, this distinction is critical to understanding the relationship between Hume's favorable attitude toward causal inference in Part iii and his negative or pessimistic assessment of causal inference in I.iv.7. Any belief is steady in its influence on thought, the passions, and action in the sense that it is infixed or steady in virtue of the mechanism that produces it. It does not follow that the belief is steady in its influence, all things considered. A belief might fail to be steady in its influence owing to the presence of beliefs with which it conflicts, beliefs that reduce the likelihood of the occurrence of its characteristic manifestations or its typical effects, which reduce its influence on the will and action. In Hume's view, the point of a distinction between establishing that belief is justified, other things being equal, and establishing that belief is justified, all things considered, is to call attention to the kinds of circumstances in which belief, a steady disposition, might nevertheless be unsteady in its effects. The “other things being equal” qualification is thus cashed out substantively, with reference to conditions in which states are infixed but nevertheless unsteady in their influence due to the operation of other mechanisms. Hume's concern is with circumstances that undermine the natural function of belief. Hereinafter, I use the term ‘stable’ as a shorthand for “steady in its influence on thought, passions, and action.”
There are a variety of circumstances in which beliefs produced by a mechanism, though steady in that they are infixed, could be unstable, unsteady in their influence. In one class of cases, a belief might be unstable owing to the presence of a contradictory belief. In I.iv.2, “Of scepticism with regard to the senses,” Hume discusses two related sets of contradictory beliefs. Hume characterizes “contradiction” as an “opposition” or “combat” (T 205–6); “contradiction” involves “struggle and opposition” between “two enemies” in an effort to “destroy” (T 215) one another (§§ I.2, V.3). The language of psychological conflict is not merely metaphorical or figurative; the conflicts have identifiable effects. For example, we find that we “successively assent” (T 266) to both sides of a contradiction (§ I.2). In holding contradictory beliefs, we are conflicted at the level of our underlying dispositions. These contradictory beliefs typically alternate in their influence, with one combatant, then the other, temporarily gaining the upper hand. These dispositions are infixed and steady, but—owing to the presence of the opposing belief—neither is stable.
(p.81) Hume's treatment of education provides an example of this sort of case. It is important that “education,” in Hume's sense, is inculcation, not inductively accredited testimony (§ III.3). In the final paragraph of I.iii.9, Hume explains why beliefs based on education are not justified. In the first paragraph of I.iii.10, he reiterates that “education infixes . . . opinion” (T 121). Hume's explanation of why beliefs based on education are not justified, though brief, contains a number of strands. We should expect Hume's conception of the natural function of belief, developed in the following two paragraphs, to be germane to his treatment of education. Hume observes that education's “maxims are frequently contrary . . . even to themselves” (T 117). Different unaccredited sources can inculcate “contrary” beliefs in a single person. Education tends to give rise to states that are beliefs and hence infixed, yet unsteady in their influence—owing to the presence of contradictory beliefs.35 In bringing forward these considerations, Hume establishes that education is not justified, all things considered. (I say more about Hume's treatment of education in § VII.6.)
The influence of a belief might also be undermined by the presence of second‐order beliefs that one takes to call into question whether a first‐order belief is true. An example would be a belief system that includes contradictory beliefs and also the second‐order belief that one believes this contradiction. To take another example, suppose one believes that p and also holds the second‐order belief that p results from a belief‐forming mechanism that is unreliable, leading to false beliefs more often than not. In these and related instances, one takes one's second‐order beliefs to indicate that some of one's beliefs are not true, or not likely to be true. Hume, I suggest, holds that reflection on such second‐order beliefs unsettles the first‐order beliefs that are called into question, so that one is less inclined to maintain them. The first‐order beliefs remain sufficiently infixed to constitute belief but are like a steady performer who becomes shaken or rattled or suffers a loss of confidence. To say that the second‐order belief “unsettles” the first‐order belief is to say that it unsteadies it—in the sense that its steadiness is somewhat reduced.36 The unsettled belief is less likely to produce its typical effects, less likely to influence the will and action. Here we have a second class of cases in which a belief might be unstable owing to the presence of beliefs with which it conflicts. Cases of this sort play a crucial role in I.iv.1, I.iv.4, and I.iv.7.
This second source of instability has an important feature, in that considerations with respect to truth undermine stability in belief. This might seem paradoxical. It is a familiar thought that “belief aims at truth.” In my interpretation, Hume maintains that stability is the natural function of belief. We might put this point by saying that nature endows us with belief in order to achieve stability, a steady influence on thought, the will, and action. But if stability is the (p.82) natural function of belief, how can there be room for belief to aim at something else, something other than stability, and at truth in particular? It thus might be thought that in my interpretation of Hume justification is a matter of stability rather than truth. The possibility of the second source of instability shows that this would be a mistake. It remains to understand why.
To make headway, it will be helpful to consider the sense in which belief aims at truth. Belief exhibits two kinds of characteristics. In the first place, belief is typically associated with a first‐order disposition (or set of dispositions) to display particular manifestations in relevant circumstances. These manifestations include other internal states, as well as verbal and nonverbal behavior. In the second place, belief is typically associated with a second‐order disposition to regulate one's belief that a proposition is true, and hence the first‐order dispositions involved in belief, by (what one takes to be) evidence or indicators of truth. The characterization of this disposition provides the sense in which belief “aims” at the truth; when we believe a proposition, we accept it as true with the aim that whether or not we believe it should be responsive to its truth, with the aim that belief should be sensitive to what is really true.37
There is room for controversy about the relationship between these features of belief. It might be held that possession of the first‐order disposition to display characteristic manifestations is sufficient for belief, so that one could believe a proposition even in the absence of the second‐order disposition to regulate one's first‐order dispositions by (what one takes to be) evidence of the truth. Alternatively, it might be held that the second‐order disposition is constitutive of belief, so that the presence of the first‐order disposition is not sufficient for belief unless the second‐order disposition is also present.38
An example will serve to make vivid the differences between these conceptions of belief. Suppose someone possesses the first‐order dispositions involved in belief that p. We present strong evidence that the belief has been acquired and sustained by a highly unreliable method. The person appreciates this evidence, but is not moved or bothered by it, and indeed continues to possess the first‐order dispositions. If a second‐order, regulative disposition is not required for belief, then the person possessing the first‐order dispositions maintains the belief that p. If a second‐order, regulative disposition is required for belief, the person's attitude toward p is not belief, but rather something else. Let us call it blind faith that p; the person regards p as true without possessing the second‐order disposition to regulate his first‐order dispositions by (what the person takes to be) evidence of their truth.
Where does Hume stand with respect to these two conceptions? I have argued that Hume holds that beliefs are steady dispositions to particular manifestations (p.83) (§§ III.2–3). Hume thus takes the first‐order dispositions to be essential to belief. It does not follow that Hume holds that a (steady) disposition to display characteristic manifestations is all that is required for belief. Whether he does hold this depends on his view about the role of the second‐order, regulative dispositions.
Short of compelling evidence, we ought not suppose that a historical figure would deny that, in having belief, we seek to regulate what we regard as true by (what we take to be) evidence of truth. Even if such an aim is not constitutive, or of the essence, of belief, it seems closely associated with it. Consider the example of the person who has blind faith that p. At the least, we are inclined to look for some special psychological explanation of his indifferent response to evidence. Perhaps, for example, the person is so invested in regarding p as true as to fall into self‐deception about the evidential state of affairs. In that event, however, the person does not take the unreliability of the method used to acquire the belief to be evidence against the truth of p after all.39 There appears to be an important connection between belief and the second‐order, regulative disposition. We ought not suppose that Hume would deny this.
To the contrary, Hume is well aware of this connection. In I.iv.1, Hume observes that “our reason must be consider'd as a kind of cause, of which truth is the natural effect; but such‐a‐one as by the irruption of other causes, and by the inconstancy of our mental powers, may frequently be prevented” (T 180). Hume is explicit that reflection on the degree to which a faculty is fallible or unreliable can unsettle a belief and undermine its influence:
Similarly, Hume writes: “When I reflect on the natural fallibility of my judgment, I have less confidence in my opinions” (T 183). These claims presuppose that belief aims at truth. There is a loss of confidence in a belief precisely because one reflects on the degree to which a faculty (“my judgment”) is fallible or unreliable, unlikely to lead to truth.40 Reflection on second‐order beliefs that one takes to call into question the truth of first‐order beliefs would not unsettle the latter beliefs unless one sought to regulate belief by what one takes to be (p.84) evidence of truth. We might put this by saying that considerations of stability absorb considerations of truth; the regulative disposition operates through its impact on stability. Hume does not say whether the second‐order, regulative disposition of “aiming at truth” is—like the steady first‐order disposition to display particular manifestations—essential to belief; however, he regards the second‐order disposition to regulate one's beliefs by what one takes to be evidence of truth as generally characteristic of belief.
Having thus found in every probability, beside the original uncertainty inherent in the subject, a new uncertainty deriv'd from the weakness of that faculty, which judges, and having adjusted these two together, we are oblig'd by our reason to add a new doubt deriv'd from the possibility of error in the estimation we make of the truth and fidelity of our faculties. (T 182)
What happened to the claim that stability is the natural function of belief? Belief, for Hume, is a steady disposition to particular manifestations, typically associated with a second‐order disposition to regulate the first‐order disposition by what one takes to be evidence or indicators of truth. (In virtue of this regulative disposition, belief aims at truth.) Belief, so characterized, is nature's provision for a steady influence on thought, the will, and action. Stability is the natural function of belief; the first‐order dispositions to display characteristic manifestations and the second‐order, regulative disposition are the means for achieving stability, so that belief can serve its natural function or purpose. Within this framework, the aim of truth and the natural function of stability are wheels that engage one another; reflection on considerations with respect to truth can undermine stability. This is what happens in the second class of cases in which a belief is unstable owing to the presence of second‐order beliefs.41
Much as it seems plausible that contradictory beliefs will alternate in their influence, it seems plausible that first‐order beliefs will be unsettled by the presence of second‐order beliefs that question their truth. In §§ IV.3–5, I extend Hume's account of sources of instability to cases that are less straightforward. At this stage, I turn to some chief examples of instability owing to the presence of second‐order beliefs with which a belief conflicts.
Part iv of Book I contains a number of case studies of reflections that call into question the truth of one's beliefs. Here I have in view considerations Hume raises in I.iv.1 and I.iv.4 and elaborates in I.iv.7. The first of these, at Treatise 266, involves “a manifest contradiction” (§§ I.4–5). Hume writes in the final paragraph of I.iv.4, “Of the modern philosophy”: “There is a direct and total opposition betwixt our reason and our senses; or more properly speaking, betwixt those conclusions we form from cause and effect, and those that persuade us of the continu'd and independent existence of body” (T 231). (I examine Hume's grounds for this conclusion at § VII.2.) Hume claims that the senses lead us to believe that matter has a continued and independent existence but that causal reasoning shows that this belief is false. Hume appeals to this contradiction between the senses and causal reasoning in I.iv.7:
Hume's footnote is to I.iv.4. The passage continues: “How then shall we adjust those principles together? Which of them shall we prefer?” (T 266). Arising from “equally natural and necessary” principles, the contradiction is ineliminable (§ I.5). Hume writes that we “successively assent to both” (T 266), describing the cycle of alternating influence that can occur in the presence of opposing, equipollent beliefs (§ I.2). There is, moreover, an additional source of instability once a contradiction is recognized. Reflecting on the contradiction, we have a need to “adjust those principles together”; we are disposed to adjust our principles or beliefs. Though these beliefs are steady in that they are infixed, when one considers the contradiction one is less inclined to maintain the dispositions in question.
'Tis this principle [the imagination, or the vivacity of ideas], which makes us reason from causes and effects; and 'tis the same principle, which convinces us of the continu'd existence of external objects, when absent from the senses. But (p.85) tho' these two operations be equally natural and necessary in the human mind, yet in some circumstances they are *directly contrary. (T 266)
In I.iv.1, “Of scepticism with regard to reason,” Hume also introduces a consideration that calls into question the truth of one's beliefs. Indeed, he claims to identify a systemic source of instability. Hume argues that “all knowledge,” even demonstrative knowledge, “degenerates into probability” (T 180). “Probability” is “that evidence, which we employ in common life” (T 181), evidence based on causal inference. As we have seen earlier in this section, Hume argues that judgments of probability are subject to correction in light of the fallibility of judgment. Furthermore, this correction takes the form of a reduction in the estimate of probability, and the new judgment of probability is itself subject to correction and reduction, ad infinitum. The result of such a series of reductions, he claims, would be “a total extinction of belief and evidence” (T 183); it would “at last reduce [the original evidence] to nothing” (T 184).42 (I examine Hume's grounds for this conclusion at § VII.3.)
Hume takes up this material in I.iv.7. At page 267, he frames “a very dangerous dilemma” (§§ I.4, 5). We can rely on the understanding—demonstrative and causal inference—alone, subjecting probability judgments to repeated corrections. Or we can rely on the understanding together with “seemingly trivial” (T 268) properties. Hume writes of the first alternative: “I have already shewn,* that the understanding, when it acts alone, . . . entirely subverts itself, and leaves not the lowest degree of evidence in any proposition” (T 267). Hume's footnote is to I.iv.1.43 If, however, we admit the trivial propensities, we embrace “manifest (p.86) absurdities” (T 268). Recognizing this dilemma—much as in the case of noticing a contradiction—we have a need to adjust our faculties and beliefs: “What party, then, shall we choose among these difficulties?” (T 268). And much as Hume cannot find a way to revise his beliefs when he attends to the contradiction between the senses and causal inference, he writes of the present dilemma: “For my part, I know not what ought to be done in the present case” (T 268).
As it stands, the dilemma is artificial. It is not open to us to adhere to the understanding alone. The extinction of belief depends upon the assumption that the principles by which we correct probability judgments are “apply'd to every new reflex judgment” (T 184). It is only when adherence to the understanding is “steadily executed” (T 267) that belief is extinguished. Hume has argued that repeated correction of probability judgments “becomes forc'd and unnatural” (T 185), “the attention is on the stretch: the posture of the mind is uneasy” (T 185), and “we enter with difficulty into remote views of things” (T 268). Though we might “take a resolution” (T 267) to adhere to the understanding alone, this is not a resolution we could keep. The extinction of belief depends on a series of corrections and reductions that does not take place.
Hume's point, I suggest, is that the considerations in I.iv.1 generate the second‐order belief that were we to subject judgments of probability to repeated corrections, the probability of the original judgment would reduce “to nothing.” This second‐order belief calls into question the truth of probability judgments and thereby unsettles them. When one considers that a series of corrections would reduce to zero the probability that first‐order beliefs are true, one is less inclined to maintain one's current probability judgments.44 As we have seen, the second‐order belief that we hold contradictory beliefs, based on the senses and causal inference, about the existence of matter also unsettles belief. In sum, I.iv.7 may be read as calling attention to the unsettling effects of some second‐order beliefs, which call into question the truth of other beliefs one holds.
Hume takes the instabilities that emerge in I.iv.7 to bear negatively on the justificatory status of the beliefs that are unsettled. He discusses the unsettling effects of considerations introduced earlier in Part iv at pages 265–68 of I.iv.7. At page 268, Hume writes: “I am ready to reject all belief and reasoning” (§§ I.4, II.2). Thus we find in Hume the judgment that no belief, not even belief based on causal inference, is justified. The considerations that lead Hume to this result are ones that call into question the truth of his beliefs, thereby unsettling them. This reduces the steadiness in influence these beliefs would otherwise have, in (p.87) virtue of being infixed, and thus undermines the natural function of belief. This is, at least, one reading of I.iv.7, Hume's difficult “Conclusion of this book.”45
A stability‐based interpretation provides a fruitful account of the relationship between Parts iii and iv of Book I. I have argued that the interpretation explains why the claims that causal inference results in belief and that causal inference is justified are entwined in I.iii.9 and neighboring sections (§§ III.1–2). Whereas Hume registers epistemic approval of causal inference throughout Part iii, in I.iv.7 he is “ready to reject all belief and reasoning.” Whereas Hume provides an extended account of “the several degrees of evidence” (T 124; cf. 153–54) in I.iii.11–13, in I.iv.7 he “can look upon no opinion even as more probable or likely than another” (T 268–69). Hume's destructive conclusions in I.iv.7 reverse his attitude toward causal inference in comparison to Part iii of Book I. To establish—as Hume does in Part iii—that the states produced by a psychological mechanism are beliefs, and hence steady in their influence in that they are infixed, is to establish that they are justified, other things being equal. To establish—as Hume does in Part iv—that the beliefs produced by a psychological mechanism are unstable is to establish they are not justified, all things considered. Hume's readiness to reject all belief as unjustified, at least for the reflective person, emerges in light of his claim that beliefs based on causal inference are unstable. My interpretation thus both accommodates the destructive conclusions in Part iv and explains the favorable results of Part iii.
III.5. Two Versions of the Stability‐Based TheoryIII. Integrating Belief and Justification III.5. Two Versions of the Theory
There is an ambiguity in my account of the relationship between Part iii and Part iv. I have said that the instabilities that emerge in I.iv.7 “bear negatively” on the justificatory status of beliefs based on causal inference. Would Hume want (p.88) to say that belief based on causal inference is never justified because it is susceptible to the instabilities he identifies in Part iv? Or might beliefs based on causal inference be justified, even though susceptible to these instabilities, provided they are not in fact infected by them?
These questions are closely related to an ambiguity in my distinction between establishing that a belief is justified, other things being equal, and establishing that a belief is justified, all things considered. Let us say that a person who searches for contradictions among beliefs, examines the reliability of belief‐forming mechanisms in producing true beliefs, considers the results of methodical application of the cognitive faculties, and so forth is (fully) reflective. There are degrees of reflectiveness. For the present purposes, it will suffice to group together persons who are not fully reflective as “unreflective.” With this classification in hand, we can explain the ambiguity I have in mind. Before doing so, it will be helpful to recall that Hume prefers to evaluate the justificatory status of a belief derivatively, with reference to the tendency of the belief‐forming mechanism that causes it to produce stability in belief (§ I.4). This can make for cumbersome formulations. Such shorthand locutions as “a belief is justified if it is stable” are to be construed to mean “a belief is justified if it results from a mechanism that tends to produce stable beliefs.”
In one interpretation, the beliefs produced by a psychological mechanism are justified, all things considered, if they tend to be steady in their influence for a reflective person, that is, for a person who fully examines beliefs in the ways I have characterized. To say that the beliefs are justified, other things being equal, is to say that for all that has been shown, they would tend to be steady in their influence for a reflective person. In this interpretation, to establish that the states that result from a psychological mechanism are beliefs is to establish other things being equal, to establish provisionally, that they are justified. Any gap between establishing justification other things being equal and establishing justification all things considered is a gap in our knowledge of what a reflective person's belief system would be like.
In the second interpretation, the beliefs resulting from a psychological mechanism are justified, all things considered, if they tend to be steady in their influence given the actual degree to which the person who holds the belief is reflective. To say that the beliefs are justified, other things being equal, is to say that they are steady, in that they are infixed by the senses, memory, or repetition. Such beliefs might nevertheless tend to be unsteady in their influence, all things considered, owing to the operation of mechanisms that tend to produce conflicts within the belief system of the person who holds the belief. In this interpretation, to establish that a psychological mechanism produces states that are beliefs is to establish that these beliefs are justified other things being equal, that they possess prima facie or defeasible justification. Any gap between establishing justification other things being equal and establishing justification all things considered is a gap between the steadiness that accrues to beliefs in virtue of being infixed and the steadiness in their influence of beliefs that do not conflict with other beliefs one holds. Though a mechanism might tend to produce beliefs, other mechanisms that operate (given the actual degree to which the person (p.89) is reflective) might tend to produce conflicts that render such beliefs unsteady in their influence. It is in this way that defeasible justification can be undermined.
In light of Hume's conclusions in I.iv.7 (§ III.4), the two versions of a stability‐based interpretation diverge in their epistemic assessments of the unreflective person. In the first version, the stable beliefs of an unreflective person are unjustified because they would be unsteady in their influence under reflection. In the second version, the stable beliefs of the unreflective person are justified because, given the degree to which the person is reflective, they are steady in their influence; it is the reflective person who is in the grip of instability.
Both versions assign a crucial role to considerations of stability; they differ in what that role is. In the second version, justification depends on whether a belief is stable given the actual degree to which the subject, the person who holds the belief, is reflective. This version imposes a requirement of stability under the subject's de facto degree of reflection, stability for the actual subject. In the first version, justification depends on whether a belief would be stable under (full) reflection, stable for a reflective subject. This requirement of stability under reflection is more demanding or stringent than that of stability under the subject's actual degree of reflection.46
It might be helpful to consider the character of the second version of Hume's theory, where the requirement for justification is stability for the actual subject. According to this version, the fact that a mechanism generates beliefs goes some way in conferring justification. A belief‐forming mechanism is a psychological mechanism that regularly produces, or tends to produce, beliefs. It follows from Hume's account of belief that the products of a belief‐forming mechanism tend to be steady in their influence in that they are infixed, and hence stable, other things being equal. In the second version of Hume's theory, to say that beliefs are justified, other things being equal, is to say that they are stable, steady in their influence, simply in virtue of being infixed by the mechanism that produces them. It follows that the products of a belief‐forming mechanism are automatically justified, other things being equal, simply in virtue of resulting from a belief‐forming mechanism; they automatically possess prima facie or defeasible justification.47
(p.90) The two versions of the stability‐based theory lead to different understandings of the relationship between Parts iii and iv of Book I of the Treatise. Let us begin with the first and more demanding version, which requires stability under reflection. In this interpretation, when Hume registers epistemic approval of beliefs based on causal inference in Part iii, he is claiming that, so far as he has examined causal inference, he has not uncovered any considerations to suggest that beliefs based on causal inference would be unstable if subjected to greater reflection. The only considerations in play in Hume's discussion of the nature and causes of belief are ones that relate to custom or habit, the mechanism that produces, and infixes, the dispositions that arise from causal inference. Indeed, as of I.iii.10, Hume has not fully scrutinized custom or repetition itself; in I.iii.12, he introduces a distinction between “imperfect” (T 131) and “perfect” (134, 135) habits (§ II.1). What is more, as of the close of Part iii, Hume has not yet considered the methodical application of causal inference, the belief in body, the relationship between causal inference and the senses, and so forth.
Hume defers such questions to Part iv. In I.iv.7, he appeals to the instabilities located in I.iv.1 and I.iv.4 in route to his destructive conclusions about justified belief. Thus, an interpretation in which Hume holds the more demanding theory provides a natural reading of the recurrent claim in Part iii that causal inference is justified. This is a provisional judgment; for all Hume has shown at that stage, beliefs based on causal inference are stable. So far, so good. Once the considerations advanced in I.iv.7 are in view, this provisional judgment is withdrawn in favor of the “all things considered” judgment that belief based on causal inference is unjustified. Such belief is unjustified because it would be unstable for a reflective person.48 This is a reading of the relationship between Parts iii and iv from the perspective of an interpretation that ascribes to Hume the requirement of stability under reflection.
The interpretation that attributes to Hume the less demanding theory, requiring stability under the subject's actual degree of reflection, can also make sense of the developments in Part iv that I have reviewed. If Hume's requirement is stability for the actual subject, we can think of his discussion of the two systems of realities—based on the senses, memory, and custom—as exhibiting (p.91) a specimen belief system that someone might hold. The beliefs it contains are automatically justified, other things being equal, simply in virtue of resulting from belief‐forming mechanisms. Given that the subject does not engage in the sorts of reflections characteristic of Part iv, there is no tendency to instability in belief, which would undermine the prima facie or defeasible justification within this belief system, so that the beliefs it contains are also justified, all things considered.
This is different from the upshot of attributing to Hume the more demanding theory. If justification requires stability under reflection, as of Part iii, beliefs based on the senses, memory, and causal inference are merely justified for all that has been shown. Since instabilities would emerge upon reflection, the beliefs are not justified, all things considered. In the less demanding reading, if there is no tendency to instability in the specimen belief system, all the considerations relevant to that person's belief system have been taken into account; the other things being equal condition is satisfied, and the person has justified belief, all things considered.
In sum, Hume's claim that to establish that states produced by a mechanism are beliefs, and hence steady, is sufficient to establish that they are justified, other things being equal, admits of two interpretations. Both afford a natural resolution of the original puzzle: to explain why the claim that causal inference is justified arises in tandem with the claim that causal inference results in belief. In the more demanding reading, to establish that the products of causal inference are beliefs, and hence steady in that they are infixed, is to establish that they would be steady in a reflective person's belief system, for all that has been shown; the claim in Part iii that causal inference is justified is provisional. In the less demanding reading, to establish that the products of causal inference are beliefs, and hence steady in that they are infixed, is to establish that these beliefs are justified, other things being equal. I believe one must adopt one or the other of these interpretations—interpretations that link justification to stability—in order to explain the textual phenomena in Part iii.
III.6. A Defense of Attributing the Less Demanding Version to HumeIII. Integrating Belief and Justification III.6. Attributing the Less Demanding Version to Hume
I have argued that the fact that instabilities that infect causal inference emerge upon reflection in I.iv.7 does not in itself adjudicate between the two readings of Hume's theory of justification. Is there reason to believe that Hume subscribes to one version of the theory rather than the other? My claim that Hume develops his theory of justification in two stages, constructive and destructive (§ I.4), is a useful starting point for considering this question.
If justification requires stability under reflection, no belief is justified. The more demanding version of the stability‐based theory thus readily explains Hume's reversal in his expressed attitude toward the epistemological distinctions he has drawn. But we are left to wonder why Hume abandons his pretheoretical epistemic commitments rather than the theory that fails to sustain them. (p.92) I think it fair to respond that Hume has evidence for taking a stability‐based theory as giving the most promise of sustaining his pretheoretical commitments and that he has no more promising theory in view. After all, the more demanding theory has a great deal of provisional success; it explains Hume's pretheoretical intuitions about causal inference (§§ III.2, 5) and also, as we shall see, explains his epistemic judgments relating to the probability of causes (§ IV.1), “unphilosophical” probability (§§ IV.2–6), and beliefs resulting from associative propensities other than causal inference (§§ V.1–4).
At the same time, attributing to Hume the less demanding version of the theory does a better job of explaining Hume's reversal in his epistemic judgments, without thereby seeming to mandate abandonment of the stability‐based theory. If justification requires stability given the actual degree to which the subject is reflective, it is open to Hume to refine his pretheoretical commitments in light of the developments in I.iv.7. The less demanding version overturns those commitments only insofar as they are construed as epistemic assessments of the beliefs of the reflective person; it sustains the commitments construed as assessments of the stable beliefs of unreflective persons. This asymmetry leads to a qualified reversal, with Hume retaining the less demanding version of the theory in light of its success in systematizing his pretheoretical distinctions in their application to unreflective persons. The less demanding theory offers a more nuanced account of Hume's reversal.
Hume's remarks in I.iv.7 subsequent to pages 268–69 confirm this interpretation. If Hume inclines to the more demanding theory, his position is that no belief is justified, for either the reflective or the unreflective person. It is unclear, however, that Hume has any epistemic objection to the stable beliefs of the unreflective person. At page 272, Hume writes of landholders: “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” (T 272). We can think of these unreflective persons as confining their beliefs to the senses, memory, and everyday causal inference. Hume's discussion continues: “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” (T 272). Far from criticizing the beliefs of the landholders, Hume writes that “they do well to keep themselves in their present situation” (T 272). This has an approving ring.49 The less demanding version of Hume's theory—in which beliefs are justified provided they are steady in their influence given the degree to which the subject is reflective—can account for this. The belief system of the unreflective person is not infected by the instabilities endemic to the beliefs of someone who (p.93) is more reflective. Treatise 272 thus supports attributing the less demanding theory to Hume.50
There is, however, an important line of objection to interpreting Hume in accordance with the less demanding version of the stability‐based theory. The worry is that if justification depends upon the actual degree to which a person is reflective, justification comes too easy. It would seem that beliefs arising from dogmatism, deference to authority, and ignoring evidence—to take some leading examples—could be stable. If justification simply requires stability under the subject's actual degree of reflection, such beliefs could be justified. Furthermore, the objection proceeds, Hume surely wants to condemn the beliefs in question.
I believe this worry to be more limited in scope than its general formulation suggests. In considering this point, it is instructive to observe that the objection can also be pressed against C.S. Peirce. Hume's emphasis on stability and infixing belief has affinities with Peirce's contention that the goal of inquiry is the settlement of opinion or fixation of belief (CP 5.375, 377). For Peirce, as well as for Hume, an unreflective person can achieve settled belief. Tenacity, authority, and the a priori method—three of the methods for fixing belief Peirce discusses—are unsettled by reflection. In particular, one must reflect on the consideration that it is a “mere accident” (5.380) or “accidental” (5.383) that they have led us to form particular beliefs and not others—that these beliefs have been “determined by [a] circumstance extraneous to the facts” (5.383). As Peirce writes: “A man . . . wishes his opinions to coincide with the fact, and . . . there is no reason why the results of those three first methods would do so” (5.387). Beliefs resulting from methods other than the method of science are unsettled by second‐order reflection on truth (§ III.4). But not everyone engages in the reflection necessary to unsettle such beliefs:
Much as Hume has no epistemic objection to landholders' stable beliefs, Peirce has no criticism of an unreflective person who succeeds in using the method of tenacity or authority to maintain settled belief.51
A man may go through life, systematically keeping out of view all that might cause a change in his opinions, and if he only succeeds . . . I do not see what can be said against his doing so. It would be an egotistical impertinence to object that his procedure is irrational, for that only amounts to saying that his method of settling beliefs is not ours. (CP 5.377)
(p.94) This is not to say that Peirce has no response to the objection that justification comes too easy. He accepts the consequence that the stable beliefs of unreflective persons are justified. His strategy is to seek to mitigate the difficulty by contending that there are few actual cases of this sort. The reflection required to unsettle the methods of tenacity and authority is likely to occur. Peirce writes of the method of tenacity, for example:
Owing to the social impulse, one is likely to find that others—in one's own community or “in other countries and in other ages” (CP 3.381)—hold different opinions. This discovery tends to stimulate the unsettling reflection that one's method of attaining belief is sensitive to accidental circumstances extraneous to the facts. Peirce has no epistemic objection to settled belief in cases where we successfully “make ourselves hermits” (5.378), but he contends that such cases are infrequent.
This method of fixing belief . . . will be unable to hold its ground in practice. The social impulse is against it. The man who adopts it will find that other men think differently from him and it will be apt to occur to him, in some saner moment, that their opinions are quite as good as his own, and this will shake his confidence in his belief. (CP 5.378)
Hume is in a better position than Peirce to maintain that such belief‐forming mechanisms as tenacity, dogmatism, and deference to authority are unlikely to sustain settled belief. In III.iii.2, “Of greatness of mind,” Hume appeals to the operation of sympathy (as introduced in II.i.1) to ground a social source of instability:
The disturbance to thought results in an internal “commotion” (T 593), an unstable condition that would not arise if other persons' beliefs remained “conceal'd” (T 593) from us. When we acquire beliefs about the opinions of others, sympathy “draws along . . . judgment” (T 592; cf. 316, 319, 365), “in some measure” making those beliefs “our own” (T 593). In cases where we also hold contrary beliefs, “the evident conflict” (T 593) takes the form of an internal contradiction, which unsettles the beliefs with which we began (§ III.4).52
So close and intimate is the correspondence of human souls, that no sooner any person approaches me, than he diffuses on me all his opinions, and draws along my judgment in a greater or lesser degree. And tho', on many occasions, my sympathy with him goes not so far as entirely to change my sentiments, and way of thinking; yet it seldom is so weak as not to disturb the easy course of my thought. (T 592)
The mere fact that one becomes familiar with “opposite” (T 593) opinions unsettles beliefs that arise from dogmatism and deference. For Hume, unlike Peirce, the unsettling effects of awareness of opposing opinions do not require reflection. (I elaborate in § IV.5 on nonreflective mechanisms that unsettle belief.) Hume advances psychological theses that can be exploited to support the claim that the range of actual cases in which unreflective persons achieve stability is narrow.
Two men traveling on the highway, the one east, the other west, can easily pass each other, if the way be broad enough: But two men, reasoning upon opposite principles of religion, cannot so easily pass, without shocking. . . . Such is the nature of the human mind, that it always lays hold on every mind that approaches it; and as it is wonderfully fortified by an unanimity of sentiments, so is it shocked and disturbed by any contrariety. (Essays 60–61)
Hume has an additional account of why dogmatic adherence to belief and deference to authority typically become unsettled, even absent second‐order reflection on their reliability. These mechanisms lead to beliefs that conflict with causal inference. It is difficult to imagine actual cases where persons confine themselves to dogmatism or deference. That many beliefs arise from dogmatism or deference does not preempt or prevent the operation of custom. Habitual causal inference will tend to produce beliefs that are in opposition to those based on dogmatism and deference. As a result, contradiction and instability (§ III.4) will infect the belief systems of unreflective persons who are dogmatic or deferential, quite apart from the social source of instability. Hume's criticism in I.iv.4 of beliefs based on superstition proceeds along these lines (§ V.4). These considerations again point in the direction of concluding that cases in which dogmatism or deference lead to settled belief are rare.
In thinking about the unsettling effects of conflicts with causal inference and of encountering persons with different opinions, it is important to bear in mind that justification depends upon the tendency of a belief‐forming mechanism to produce stable beliefs. In assessing justification, we identify the degree to which the subject is reflective. We then consider whether the belief‐forming mechanism tends to produce stable beliefs in persons who are reflective to a similar degree. The fact that an individual's beliefs are stable does not in itself imply that they are justified, for they might result from mechanisms that do not tend to produce stable beliefs (§§ I.4, III.5). There is, in this way, an imperfect match or looseness of fit between stability for a person and justification. This leverages the import of Hume's arguments to show that stable beliefs resulting from dogmatism and deference are infrequent.
(p.96) The case of ignoring evidence is somewhat different. To address it, we could attribute to Hume an account of justification with reference to comparative stability: a person's belief is justified if it results from a belief‐forming mechanism that tends to produce stable belief, provided there is no alternative belief‐forming mechanism that is available to the subject that has at least as great a tendency to produce stable belief and that would lead the subject not to hold the belief.53 (“Availability” would have to be characterized narrowly, in keeping with the spirit of the requirement of stability under the de facto degree of reflection.) Consider, for example, Part V of the Dialogues concerning Natural Religion. The proponent of the argument from design argues from the similarity of the universe to machines and from the observation that machines are the product of intelligent design to the conclusion that there is a single God who designed the universe. In observing that “a great number of men join in building a house or ship” (DNR 167), Hume is making the point that the proponent of the argument is overlooking available evidence—that the universe resembles complex machines designed by teams of intelligent beings. This evidence is far from recherché; it is readily available to anyone who has access to the premises for the argument from design. Recognizing such evidence would lead one to withhold belief in a single designer. Furthermore, a belief‐forming mechanism that takes such evidence into account presumably tends to produce beliefs that are at least as stable as those produced by ignoring such evidence. A comparative stability account therefore implies that persons who ignore obvious facts about the design of complex machines are not justified in believing in a single designer on the evidence of the design argument. As we shall see, there are contexts in the Treatise where it is fruitful to suppose that a comparative stability account comes into play (§ IV.2).
I have sketched a number of resources at Hume's disposal for blunting the claim that justification comes too easy on the less demanding version of the stability‐based theory. We still need ask whether Hume does wish to condemn the beliefs of the unreflective person insofar as they are stable. The textual evidence that he does wish to condemn them as unjustified is inconclusive. We have seen that Hume is unwilling to criticize “honest gentlemen.” The less demanding theory can account for this. Hume takes the position that the stable beliefs of the unreflective person are justified while observing that these beliefs would not be justified were the person more reflective. What about Hume's treatments of religion in “Of Miracles,” “Of a particular Providence and of a future State,” and the Dialogues? We need not read these discussions as impugning beliefs based on dogmatism and deference in an unreflective person. By and large, these works can be interpreted as calling attention to what a person would believe on these topics upon reflection. In other words, we can take Hume to combine a favorable epistemic assessment of some stable beliefs of an unreflective person with the observation that the beliefs would not be stable, and hence (p.97) would not be justified, were the person more reflective. We need not take Hume to find the permissive epistemic evaluations of the less demanding theory to be unacceptable.
Whatever the case in Hume's later publications, we are in a position to explain why he adopts the less demanding theory in the Treatise. The observation that the stable beliefs of the unreflective person would not be fully justified, were the person sufficiently reflective, raises the question of how reflective a person ought to be.54 As we have seen, Peirce is a notable proponent of a stability‐based theory who is not prepared to claim that the person who clings tenaciously to belief ought to be more reflective or ought to adopt a different method. Peirce seems unable to assign pride of place to higher levels of reflection, even though he thinks that reflection leaves one method of fixing belief—the method of science—standing. Similarly, in discussing the landholders, Hume does not suggest that unreflective persons ought to be more reflective. Within a framework that prizes belief‐forming mechanisms insofar as they contribute to stability or settled belief, it is difficult to locate a rationale for saying that justification requires stability under reflection.
Hume's position in the Treatise compounds the difficulty of setting stability under reflection as a requirement for justification. This more demanding requirement assigns reflection a privileged epistemic status. Hume cannot derive this status for reflection from its contribution to stability, since he concludes in Part iv that reflection is deeply destabilizing. In this respect, Hume parts company with Peirce. For both Hume and Peirce, an unreflective person can achieve settled belief, albeit belief that would be unsettled by greater reflection; Hume, however, does not share Peirce's optimism that heightened reflection, should one engage in it, leads to a better method of settling belief. To the contrary, he claims that reflective persons cannot achieve stability in belief. This is a philosophical obstacle to Hume adopting the more demanding theory in the Treatise.55 (The systemic source of instability in I.iv.7—the dangerous dilemma deriving from I.iv.1—vanishes from the first Enquiry.) The less demanding version, by contrast, does not require any argument that reflection is epistemically privileged. In the Treatise, Hume can more readily motivate the requirement of stability under the de facto degree of reflection. This explains why Hume declines the opportunity to criticize the beliefs of “honest gentlemen” and instead writes of these landholders approvingly.
The less demanding theory has a noteworthy implication. The stable beliefs of the unreflective person are justified, whereas the beliefs of the reflective person are unstable and hence unjustified. The beliefs of the unreflective person occupy a preferred epistemic status. I believe that securing this paradoxical result was among Hume's intentions in the Treatise. Hume seeks to show that an epistemic preference for reflection is a prejudice. As a matter of temperament, I (p.98) suspect, Hume took delight in disparaging intellectual reflection (§§ I.4, VII.1). Part of the evidence for this is the dramatic fashion in which Hume introduces reflection's unsettling effects. There is but the barest hint of the destabilizing character of reflection in Part iii (§ IV.2). And though I.iv.1 and I.iv.4 develop the materials for this result, Hume withholds his destructive conclusion about reflection until I.iv.7 (§§ I.4, VII.1–3). I contend, in addition, that Hume is overzealous in signing off, in I.iv.7, on arguments for the claim that reflection is destabilizing (§§ VII.2–3). It is because he contends that reflection is destabilizing that Hume has no grounds for criticizing the stable beliefs of unreflective persons. This is a box of Hume's own making, one that puts the more demanding version of a stability‐based theory out of reach.56 My procedure, in the remainder of this study, is to adopt the less demanding of the two stability‐based interpretations, while remaining alert for new evidence in this regard (§§ IV.5, V.4).
III.7. Further Remarks on I.iv.7III. Integrating Belief and Justification III.7. Further Remarks on I.IV.7
According to Hume, even the reflective person can escape reflection's unsettling effects for periods of time. The instabilities dissipate when one dines, plays a game of backgammon, makes merry with friends, or enters into some other amusement (T 269). Would Hume wish to say that the reflective person has justified belief at least during such relaxed periods?
Michael Williams finds an affirmative answer tempting. He is inclined to attribute to Hume a contextualist approach to justification, where “context includes features of the doxastic as well as the external environment.” Let us focus on the doxastic environment, one's other beliefs. Williams writes that “knowledge is subject to quite different substantive constraints” in different contexts. In Williams's interpretation, a person can have justified belief in an everyday context and lose that justification in a reflective context. There can be a “transient” or “temporary” loss of knowledge, but only that, within the context of reflective periods.57 The results of reflection do not undermine everyday justification and knowledge.
The more demanding version of a stability‐based interpretation is incompatible with Williams's position. The requirement of stability under reflection has the consequence that our beliefs during relaxed periods are not justified because they would not be steady in their influence during reflective periods. But I have rejected this interpretation, so I should evaluate Williams's position in light of the requirement of stability under the de facto degree of reflection.
Williams takes knowledge to require, at least, justified, true belief. Williams wants to attribute to Hume the view that knowledge is unstable because justification is unstable; standards for justification depend upon context, and skeptical (p.99) contexts raise the standard. Williams's paradigm for skepticism is “modern or ‘Cartesian’” skepticism, that is, skepticism with respect to our knowledge of the external world.58 Such global or radical skepticism relies on the availability of a hypothesis systematically incompatible with the truth of our beliefs in the relevant domain. Perhaps there is no material world and all of our sensory experiences are the product of a dream or deceiver. The skeptic's point is that there is no non–question‐begging argument to show that belief in the external world is more probable than the skeptical hypothesis. The skeptic does not claim that there is any positive reason to believe his hypothesis. Cartesian hypotheses give us no reason to think that our belief in the external world is false or even probably false.
Hume's arguments in I.iv.7 are not Cartesian in Williams's sense. At Treatise 265–66, Hume claims to identify a manifest contradiction between reason and the senses, such that one belief we hold (upon reflection) must be false. This is a kind of antinomy; Hume's strategy is to show that the belief in matter is inconsistent with another belief (§§ III.4, VII.2), not that a skeptical hypothesis undermines its justification. The discussion of the dangerous dilemma at pages 267–68 depends upon the prior result that, if subjected to repeated corrections, the probability that any first‐order judgment is true reduces to zero. This is again different from showing that there is no reason to prefer an everyday hypothesis to a merely possible skeptical alternative; the argument for rejecting all belief proceeds from the premise that the understanding is in fact fallible. Both sets of considerations, unlike Cartesian hypotheses, purport to show that beliefs we hold are false or probably false. It is therefore misleading to assimilate the developments in I.iv.7 to Cartesian skepticism. Any interest that Hume might have in Cartesian skepticism about the external world is not in play; no global skeptical hypothesis to raise the standards for justification is operative in I.iv.7. In Hume's view, his reflections in the “Conclusion of this book” give us positive reasons to reject beliefs we hold; this seems far removed from having justification cut out from under us.59 For this reason, I am not sympathetic with attributing to Hume a “contextualized” variant of the requirement of de facto stability. The more reflective moods of interest to Hume do not so much raise the standards for justification as introduce new substantive considerations, which Hume takes to tell against the truth of other beliefs one holds.
There is a point of a different sort. Consider the reflective person during relaxed periods in comparison to the landholder at any time. From the perspective of Williams's interpretation, both would seem to have justified beliefs, since (p.100) both are in an unreflective context. There is, however, a difference between the landholder and the reflective person during relaxed periods. Hume insists that these periods will give way to renewed reflection; after engaging in everyday amusements, “I am uneasy to think I . . . decide concerning truth and falshood, reason and folly, without knowing upon what principles I proceed” (T 271). We should expect this to lead to a new round of reflection and attendant instability.60 A relaxed posture cannot itself be sustained once the reflective posture has been achieved. A person who has attained a reflective posture can relax, but the reflective dispositions will reassert themselves in periods of renewed reflection. Hume could allow that reflective persons are justified during unreflective periods by tinkering with the time frame for stability. It remains the case that someone who has entered into the reflections of I.iv.7 will oscillate between periods in which beliefs are unsettled by reflection and those in which they are not. Though reflective persons can achieve periods in which their beliefs are not unsettled, these periods contain the seeds of their own destruction. This is the point Hume wishes to emphasize, whatever we say about the status of the beliefs of the reflective person during relaxed periods.61
(1.) What “we”—in more recent epistemology—call “knowledge.” Hume's “proofs” (§ II.1) fall within the scope of the term as I am using it. Though Hume officially reserves ‘knowledge’ for belief arising from the comparison of ideas (T 69–70, 124), he allows in the Abstract: “No matter of fact can be proved but from its cause or its effect. Nothing can be known to be the cause of another but by experience” (A 654). See also the uses of cognates of ‘know’ at T 103, 104, 148.
(4.) Passmore, 1952/1968, pp.62–63. Price is perhaps fastening on to this puzzle when he writes that Hume “want[s] to stick to his original definition of belief as a lively . . . idea associated with a present impression by a relation of experienced constant conjunction,” but that Hume “now wants to say that it is a definition of reasonable or sensible or sane or intelligent belief” (1969, p.174). See also Walsh, 1972, pp.104–6.
(7.) Noting that “[Loeb] thinks that, in explaining how beliefs are produced, Hume produces an account of their justification,” Owen objects: “When [Hume] has explained the production of belief, he has explained the production of a state we take to be true. . . . But this is a much weaker claim than modern claims about justification” (1999, p.140 n. 38). (See § II.1, n. 13.) But in my interpretation, Hume's theory of justification is linked to beliefs qua stable or steady.
(8.) In the Oxford Philosophical Texts edition of the Treatise (Norton and Norton, 2000b), which follows the Clarendon critical edition (Norton and Norton, forthcoming) in incorporating Appendix passages into the text of Book I, references to pages 628–29 and 630–32 may be found at the penultimate paragraph of I.iii.7 and the final three paragraphs of I.iii.10, respectively. In the latter case, the first twelve words at the first of the three Appendix paragraphs are identical to those at the final paragraph of I.iii.10 (at T 123) as originally published. These two paragraphs, though similar thematically, diverge thereafter. In 2000a, Norton and Norton write that the Appendix “addition makes the [original, final] paragraph . . . redundant and thus it has been omitted [in 2000b]” (p.252), though it will be included in the critical apparatus of the Clarendon edition. (There is no notice of the omission in 2000b.) As a consequence, the passages I quote at T 123 in my discussion of poetical enthusiasm below in this section are not included in Norton and Norton, 2000b.
(9.) For commentators who place some emphasis on steadiness, see Laird, 1932, p.88; Wolff, 1960, pp.112–14, 128; Pears, 1990, pp.12, 36, 44; Baier, 1991, pp.72–74, 80; and Johnson, 1995, p.171. See also the references to MacNabb at n. 20.
(10.) The quotations may be found at Stroud, 1977, pp.9, 17 (cf. 10 and 27), and 23 (cf. 26). (Cf. Johnson, 1995, pp.37–38.) For the pressure on the theory of belief, see pp.72–73, 224–26. Note that Stroud tends to treat Hume's associationism as a component of the overall theory of ideas (1977, pp.8–9, 35–37).
(11.) Here I diverge from Stroud, who sees Hume as focusing on what must be “added to” the theory (1977, p.9; cf. 11). Stroud thinks Hume overlooked the possibility of a dispositional account of belief. See n. 22.
(12.) Hume thus often seems oblivious to the disparity between his treatments of belief both as a lively idea and as a steady disposition. For example, in the final paragraph of I.iii.7, he claims that someone who reads a book “as a true history” has “a more lively conception” than someone who reads it “as a romance” (T 97–98). At the same time, he directs that the material at pages 628–29 of the Appendix—a passage with clear dispositional elements—be inserted just before this paragraph.
(13.) I cannot accept the emendation to “external establish'd” (Norton and Norton, 2000a, p.253) persuasions to be adopted in the Clarendon critical edition of the Treatise (Norton and Norton, forthcoming). Though not citing any textual basis for this change, Norton and Norton maintain that it is “even less likely that [Hume] meant to speak of ‘eternal’ [rather than “external”] persuasions” (p.253). They suggest that in context an “internal/external” contrast makes “good Humean sense” in that the external beliefs “arise from experience” (p.253); however, one can remember one's own internal states. They also note that Hume distinguishes internal and external perceptions at T 190, but this is in I.iv.2, a specialized context dealing with the belief in body. Nothing there suggests that memories might be external perceptions. In any event, there is no problem to solve; “eternal” is rhetorical flourish. In the Appendix, Hume is contrasting genuine belief with the “mere phantom of belief” (T 630) that arises from poetical enthusiasm. In this context, he exaggerates the steadiness of memory and custom.
(15.) My interpretation of Hume on belief has been influenced by MacNabb, 1951/1966, esp. pp.71–81. Everson, 1988, adopts a related “causal” and “functional” interpretation. Price credits Hume with “a hint or suggestion” of a dispositional account (1969, p.187; cf. 165, 188). Armstrong maintains that Hume “wavers” between dispositional and occurrent analyses of belief (1973, p.71). For others who attend to the dispositional passages, see Basson, 1958, ch. 3; Bricke, 1980, pp.121–22 (cf. 30–31, 46–58); and Pears, 1990, pp.50–51.
(16.) There is an intermediate hypothesis, that beliefs are lively ideas that are steady in their influence on the thought, will, and action—that lively ideas, when they exist, have uniform effects. Given that ideas come and go, this hypothesis does not explain the steadiness in those effects.
(18.) Cf. Price, 1969, pp.172–73. Price appeals to examples other than poetical enthusiasm. My response derives from MacNabb, who also uses the terminology of “pseudo‐belief” or “quasi‐belief” (1951/1966, pp.76, 79). I extend his treatment to poetical enthusiasm, a topic he does not discuss. I reserve the terminology of ‘quasi belief’ for a different use, in § V.5.
(19.) I do not think Hume can be taken to identify belief with a persistent occurrent idea token that has a stable vivacity. A poetical enthusiasm can be sustained and thus satisfy this condition.
(22.) Stroud cites passages where Hume discusses belief's “effects on the mind” but comments: “Hume seems never to have entertained the idea that this connection between belief and the passions and the will might constitute the very difference he seeks between belief and mere conception” (1977, p.74). Similarly, Pears takes note of Hume's “remarks about the effect of belief on action” in the course of criticizing Hume for “treating [belief] . . . as a mental event or occurrence” (1990, p.50). Overlooking the insights—see n. 15—of Armstrong, MacNabb, and Price during the period 1951–71, many recent commentaries do not even acknowledge dispositional elements in Hume's account: Bennett, 1971, pp.294–95; Fogelin, 1985, pp.53–58; Waxman, 1994, pp.33–42; Garrett, 1997, pp.27–28, 143, 208–15; and Mounce, 1999, pp.35–37.
(23.) Rather than allowing that resemblance can intensify a belief, Hume should say that resemblance can result, strictly speaking, only in a pseudointensification of belief.
(26.) MacNabb, 1951/1966, p.79; Passmore, 1952/1968, p.62; and Baier, 1991, p.75. At Treatise 117, Hume devotes a paragraph to a number of “instances” that he takes to be “parallel” to education. See § IV.2, n. 7.
(27.) Laird reads Hume as placing education in the same category as poetical enthusiasm and resemblance, as mechanisms that do “not usually produce full conviction” (1932, p.114). He writes that for Hume education “might acquire a force even stronger than actual experience” (p.113). This does not do justice to Hume's position at Treatise 116 and 121.
(28.) Price observes that the requirement of association to a present impression would have the effect of restricting Hume's theory to beliefs about particular matters of fact, excluding general empirical beliefs (1969, pp.179–81). As the cases of liars and inculcation show, even particular beliefs need not be related to a present impression. An advantage of taking belief to arise from repetition is that it allows a unified treatment of particular and general beliefs; both are dispositions, though the general beliefs are dispositions to form particular beliefs under appropriate conditions. See MacNabb, 1951/1966, p.77; and Price, 1969, pp.181–83.
(29.) Cf. Baier, 1991, p.76; and Owen, 1999, pp.206–7. I do not agree with Passmore that Hume “now substantially abandons the pretence of giving a general theory of belief, admitting, for example, that ‘education’ produces in us ‘beliefs’ which are not associated with impressions” (1952/1968, p.100).
(31.) Of course, argument is required to establish the operative theory of justification. I am attempting, however, to locate the theory of justification implicit in Hume.
(33.) Winters also draws on Treatise 118–19 to make this point (1981, pp.640–41). Though Passmore dismisses Hume's preference for coherent and settled belief as a prejudice (§ I.5), he does not cite Treatise 118–19, much as he overlooks the potential role for uneasiness in this context.
(34.) One could press Hume in the direction of bringing evolutionary or biological considerations to bear on justification. Cf. Schmitt, 1992, esp. pp.67–72. (See also § IV.5, n. 36.) For discussions of protoevolutionary views in Hume, see Pike, 1970, pp.176–82; and Monteiro, 1976. The latter includes helpful references to a wide body of literature. I think Pears overstates the distance between us and Hume owing to the development of Darwinian theory. See his 1990, p.68, though cf. 98. In claiming that, according to Hume, operations of the imagination “are found to be only trivially or accidentally connected with the truth of the beliefs which are their effects” (1991, p.274), Stroud seems to overlook the possibility that Hume takes many beliefs to result from mechanisms that are adaptive. Also see § I.5, n. 29.
(36.) Here I prefer ‘unsettle’ to ‘unsteady’ because the latter admits an alternative reading, such that to unsteady a belief is to render it not steady, sufficiently unsteady that it no longer constitutes belief.
(37.) I owe the central distinction in this paragraph to David Velleman, in conversation. The formulations are based on Velleman, 1992, pp.13–15 (including n. 24), and 1996, p.709; and Leon, 1992, pp.299–302. Note that the claim that belief aims at truth is not the simple tautology that to believe that p is to believe that p is true.
(39.) I owe this point to David Velleman.
(40.) As evidence that Hume subscribes to a reliabilist theory of justification, Schmitt writes: “A systematic review of Book I shows that Hume is persistently concerned to judge the reliability of our operations”; the passage I quote from Treatise 180 is one of Schmitt's examples (1992, p.56). Schmitt draws on a range of evidence, but it is important that the sorts of passages he cites do not immediately explain the nature of Hume's concern for reliability. A coherentist, and hence an internalist, could be concerned to assess reliability, in the service of incorporating into a belief system second‐order beliefs about the success of belief‐forming mechanisms, beliefs that increase overall coherence.
(42.) Owen contends that Hume's “concern is not about justification, but about truth” (1999, p.189). In most any account of justification, there is some connection between justification and truth. This observation also bears on Garrett's position that the line of development at I.iv.1 and pages 267–68 of I.iv.7 unfolds entirely within cognitive psychology, rather than normative epistemology (1997, pp.222–27). The corrections are for the likelihood of “error” (T 180, 182) in light of “our fallible . . . faculties” (T 180). For related criticism of Garrett, see Fogelin, 1998, p.168.
(43.) There is the suggestion that the argument that probability reduces to zero is directed at non‐Humean conceptions of belief: “If belief, therefore, were a simple act of the thought, without any peculiar manner of conception, or the addition of a force and vivacity, it must infallibly destroy itself” (T 184). This suggestion is canceled, twice over. First, Hume writes in the next paragraph: “But here, perhaps, it may be demanded, how it happens, even upon my hypothesis [that belief is more properly an act of the sensitive, than of the cogitative part of our natures], that these arguments above‐explain'd produce not a total suspense of judgment” (T 184). Second, at Treatise 267 Hume appeals to I.iv.1 to support a general result about the subversion of belief. Cf. Lynch, 1996, esp. pp.100–101; and Owen, 1999, p.198.
(44.) I provide a somewhat different treatment of this material in my 1991, pp.261–68; 1995a, pp.103–12; and 1995b, pp.323–27. The interpretation in the text is closer to that of my 1998, p.222.
(45.) Garrett maintains that Hume endorses a normative epistemic principle in I.iv.7: “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” (T 270). Garrett thinks this “Title Principle” (1997, p.234) avoids the first horn of the dilemma (§ III.4) because the repeated reflection required to subvert belief is strained rather than lively and mixed with some propensity (p.235). That is fair enough, but we need not strain to reach the higher‐order belief that were we to subject judgments of probability to repeated corrections, the probability of the original judgment would reduce to “nothing.” Also, the title principle cannot show us a way around the manifest contradiction. Garrett thinks that at Treatise 265–66 Hume is merely addressing difficulties that arise specifically for modern philosophers (p.220). Garrett does not mention the first sentence of the following paragraph: “This contradiction wou'd be more excusable, were it compensated by any degree of solidity and satisfaction in the other parts of our reasoning” (T 266). Hume introduces the dangerous dilemma in the next paragraph, which suggests a continuous normative argument; he finds nothing to mitigate the epistemic problems the manifest contradiction raises for parts of “our reasoning.” In addition, the title principle, which appears but once in the Treatise, is wedged between his discussions of periods of amusements at Treatise 269 and the description of the return of a reflective posture at 271 (§ III.7). The title principle thus itself seems to reflect a momentary mood, rather than Hume's considered epistemic view. Cf. Winkler, 1999, pp.198–200. See also Singer, 1995, pp.609–12.
(46.) The first version of a stability‐based interpretation has affinities with reflexive approval interpretations due to Baier and Korsgaard (§ I.5). The second version, however, allows that an unreflective person holds beliefs that are justified though they would not survive self‐scrutiny or reflection, so that this less demanding theory is not an instance of a reflexive approval view.
(47.) The second version of the stability‐based theory does not imply that beliefs are justified, other things being equal, simply in virtue of being beliefs. A belief might not result from a belief‐forming mechanism because it does not result from any mechanism or because it results from a mechanism that only infrequently produces beliefs. (Perhaps association by the relation of resemblance occasionally produces belief.) Rather, the requirement of stability for the actual subject implies that those beliefs that result from belief‐forming mechanisms are automatically justified. In my 2001b, esp. sec. 4, I develop an interpretation where the justification of a belief depends upon its own stability, rather than the tendency of the mechanism that causes it to produce stable belief. Elaborated in this way, Hume's position does imply that justification, other things being equal, is an intrinsic property of belief. This thesis is characteristic of a “negative coherence theory” of justification, in the sense of Pollock, 1979, esp. pp.105–11, and 1986, esp. pp.72–73, 83–87. I am indebted to Frederick Schmitt for helping me distinguish the implications of the two developments of my position.
(48.) Proponents of reflexive approval interpretations want to contain this result. Baier, I believe, would look to reflection that is suitably socialized (1991, pp.284–85, 119–20), but I do not see how the element of cooperation can avoid the subversive results of the reflections in I.iv.1 and I.iv.4. See my 1994, pp.473–74. Korsgaard takes note of the problem (1996, p.63 n. 31), referring to a Hume Society Conference paper. As I read this unpublished work, Korsgaard takes Hume to allow that under full reflection we approve beliefs held unreflectively, e.g., beliefs that we will hold unreflectively in the future. This approach enables Korsgaard to interpret Hume as retaining the full reflection test, by applying such reflection to unreflective beliefs. But if under reflection we disapprove of our beliefs held reflectively, would we under reflection approve of the same beliefs held unreflectively?
(49.) Flew describes Hume as “commending” the landholders (1986, p.115). As Hookway notes, “The condition of a country gentleman . . . may be enviable but is not available to everyone” (1990, p.101). The point that the condition is not available to those who are more reflective is as close as Hume comes to objecting to it.
(50.) Fogelin attributes to Hume “perspectivism”: “What we believe and what we think it appropriate to believe is a function of the level of investigation we are indulging in” (1998, p.164, emphasis added). Consider the stronger position: that what we believe and what is appropriate to believe is a function of the level of investigation. Fogelin seems to reject this interpretation: “Hume's writings exhibit a radical form of epistemological, or better, doxastic perspectivism” (p.164). On the other hand, Fogelin suggests the stronger position in his remark that in Hume's writings there is “normativity reflecting the scene that progressively unfolds as the investigation develops” (p.168). The stronger form of perspectivism falls out of the less demanding version of a stability‐based theory.
(52.) Brand calls attention to the III.iii.2 passages for a different purpose (1992, pp.88–89). Though she does not cite Treatise 592–93, Korsgaard relies on this doctrine in her interpretation of Hume's moral theory (1999, pp.24–26). See notes 32 and 40, in §§ IV.4–5. We can wonder whether Hume should allow that sympathy strictly results in belief (Árdal, 1966/1989, pp.46–48; and Jenkins, 1984, pp.93–97). Hume might try to treat the operation of sympathy as akin to perception or memory in that it infixes belief without a past repetition. Or he might treat it as akin to poetical enthusiasm and resemblance in producing pseudobeliefs that do not manifest a steady disposition (§ III.2). Hume's main point is to emphasize that sympathy creates a disturbance, commotion, or conflict; pseudobeliefs could fulfill this role.
(56.) I am indebted to Frederick Schmitt for suggesting some of the arguments in this section.
(59.) Williams does not discuss the manifest contradiction. His brief treatment of the dilemma (1991, pp.5–6) focuses on the difficulty in sustaining the series of corrections, whereas I focus (in § III.4) on the higher‐order belief that performing the series of corrections would reduce probability to zero. Broughton is clear about the difference in character between Cartesian arguments and those of I.iv.7, which do “not turn simply on the possibility that my belief may be false, consistent with what I observe” (1983, p.16).
(60.) Williams would agree on this (1991, pp.8–9). See also Laird, 1932, p.179; Hookway, 1990, p.103; Johnson, 1995, pp.323, 325; and Singer, 2000, pp.237–38. For a different reading, see Baier, 1991, esp. p.22.
(61.) In Fogelin's interpretation (n. 50), the results of every level of investigation are on a par (1998, pp.164–69). This parity would be epistemic, on the stronger version of perspectivism. I am inclined to think that, in Hume's view, what it is appropriate for us to believe depends upon the most reflective level of investigation the actual subject has achieved. Fogelin cites Treatise 272 (p.164): “We shou'd yield to that propensity, which inclines us to be positive and certain in particular points, according to the light, in which we survey them in any particular instant.” I am reluctant to put too much weight on this passage, where Hume is facing the immediate problem of how he can proceed to Books II and III, in the aftermath of the destructive results of I.iv.7. But I think that the force of his answer at 273 is that we can proceed insofar as we manage to focus on “particular points” rather than the sorts of general reflections that undermine them.