The Manifest Connection: Causation, Meaning, and David Hume

P Kyle Stanford. Journal of the History of Philosophy. Volume 40, Issue 3, July 2002.


Exciting recent Hume scholarship has challenged the traditional view that Hume’s theory of meaning leads him to deny the very intelligibility or coherence of supposing that there are objective causal powers or intrinsic necessary connections between causally related entities. Influential recent interpretations have variously held that Hume himself accepted the existence of such powers and connections, that he was genuinely agnostic about them, or that he denied their existence while nonetheless holding it to be a perfectly coherent possibility, indeed one that we routinely (albeit mistakenly) think actual. In this paper I will argue against all three of these lines of interpretation and in favor of what I consider a neglected alternative: that Hume rejects the existence of objective necessary connections or causal powers as literally incoherent or meaningless, but on subtle and sophisticated semantic grounds, rather than simplistic ones. I find support for this semantic reading and against the alternatives not only in passages whose significance to the debate is widely appreciated, but also in Hume’s discussions “Of Liberty and Necessity” and “Of the Immateriality of the Soul.”

The claim that Hume is a causal realist (i.e., that he takes genuine causation to involve the operation of causal powers in objects themselves) has recently been defended by, among others, John Wright, Donald Livingston, Edward Craig, and, in the most convincing detail, Galen Strawson. According to Strawson, Hume is concerned to deny only that we have any knowledge or comprehension of the causal powers in objects. He holds that Hume offers a regularity theory of our knowledge or experience of causation, and that only the legacy of positivism leads us to mistake this for a regularity theory of causation as it is in itself. Strawson claims that Hume’s unembarrassed references to “those powers and principles on which the influence of … objects entirely depends” (Enquiry 3) and “those powers and forces, on which [the] regular … succession of objects totally depends” (Enquiry 5) reveal that Hume himself believes that there are causal powers in objects, even though his skepticism prevents him from claiming to know that there are. Strawson goes on to argue that this very skepticism prevents Hume from making any knowledge claim about how causation is in itself, including the claim that it is definitely just regular succession or that there is definitely not any such thing as objective causal power.

This latter contention has achieved a wide currency even among scholars who seek to refute the claim that Hume himself believed in such powers, swelling the ranks of those who take Hume to be truly agnostic about the existence of objective causal powers or necessary connections. Kenneth Winkler, for example, mounts a thorough and convincing attack on the notion that Hume believed in objective causal powers, but he ultimately accepts Strawson’s claim that skeptical modesty prevents Hume from denying that there are causal powers in objects, concluding instead that Hume merely “refuses to affirm” that there are. Similarly, Terence Penelhum argues that Hume cannot claim to know that there is no necessity in objects (on pain of dogmatism), insisting that his point must be instead “that we can recognize that our ascription of necessity is mere projection, hallowed by habit and not by right.” Simon Blackburn also rejects the thesis that Hume believes in objective necessity, but he claims that its supporters merely “misplace the stress” of Hume’s argument and commit “an error of taste rather than an outright mistake,” because he takes Hume’s point to be that even if we can refer to regularity-transcending necessity or powers in objects, he is “utterly contemptuous of any kind of theorizing conducted in terms of such a thing.” The realist and agnostic camps, then, share an important interpretive thesis-that Hume’s skepticism prevents him from denying that there are causal powers or necessary connections in objects themselves-and when attacking this thesis I will group these two accounts together as the skeptical interpretations of Hume’s view.

This skeptical thesis is rejected by a third interpretation, which I shall call the false projectivist view, represented most famously, perhaps, by Barry Stroud (but see also David Pears). This interpretation takes Hume to deny the existence of any objective causal powers or necessary connections holding between objects themselves, but not because the existence of such powers or connections would be incoherent; instead, the false projectivist account insists that we do indeed successfully, although mistakenly, project necessity onto the objects themselves and that, in order to have this false belief in objective necessity, “we need at least an idea of necessity as something true of the connections between events.”

In opposition to all three of these positions, I will defend the explicitly semantic interpretation that Hume rejects the existence of causal powers or necessary connections in objects, and that he does so precisely because he finds the supposition that there are such objective causal powers or objective necessary connections to be strictly unintelligible and incoherent, rather than merely false or skeptically immodest. It is because the supposition that such powers and connections exist is not meaningful that Hume is not compelled by skeptical modesty to allow that it might be true. But this is not to revert to a naive or positivistic reading of Hume’s argument. As I argue in section 2, Hume’s semantic argument is considerably more sophisticated than its standard portrayal. Hume argues that it is only the felt determination of the mind that occasions our application of terms like ‘necessity’ and its synonyms, and that this felt determination must therefore be the referent of such terms if they are to refer to anything at all, even in the mouths of objectors who dispute this very claim about the meanings of their own terms and despite the possible existence of qualities in objects with which we are utterly unacquainted. Section 3 finds further support for this interpretation and against the alternatives in the section “Of Liberty and Necessity,” while sections 4 and 5 engage the strongest textual evidence for the projectivist and skeptical alternatives, respectively, arguing that this evidence actually constitutes compelling support for the semantic account.

Causation and Meaning in the Treatise

Skepticism indeed figures prominently in Hume’s attacks on his contemporaries’ proposals concerning the source of our idea or comprehension of necessary connection or causal power. Hume argues that this idea cannot be derived from any known or perceived quality of objects (or from reasoning about them), from the command of the mind or will over the body and over our thought, or from our understanding of God. But Hume is not content to conclude merely that his contemporaries do not know what they claim to know or think they know about causation. He goes on to insist that no attempt to ground our idea of necessary connection or power in objects can possibly succeed. He argues (Treatise161-2) that because “we can never distinctly conceive how any particular power can possibly reside in any particular object” (else we would be unable to conceive of that object existing without its effect following) and because general or abstract ideas are only “individual ones taken in a certain light” (see Treatise 17-25), it follows that “we deceive ourselves in imagining we can form any such general idea.”

More importantly, Hume immediately and explicitly points out that this inability to form any general idea of necessary connection or power in objects carries significant consequences for what we can possibly mean when we use the terms ‘power’ and `necessary connection.’ He argues that when we talk of beings “as endow’d with a power or force” and when we speak of a necessary connection between objects while supposing this connection to depend on something in the objects themselves, “in all these expressions, so apply’d, we have really no distinct meaning, and make use only of common words, without any clear and determinate ideas” (Treatise 162, original emphasis; see also Enquiry 74). Hume thus argues that applying terms like ‘power,’ ‘force,’ and `necessary connection’ under the supposition that the power, force, or connection is something that objectively inheres in beings or objects is illegitimate. So applied, these terms have no “clear and determinate ideas” annexed to them and thus no “distinct meaning.” But Hume wants, reasonably enough, to avoid the simplistic conclusion that no meanings could even possibly be applied to expressions attributing powers, forces, and necessary connections to being and objects; thus, he continues, as ’tis more probable, that these expressions do here lose their true meaning by being wrong apply d, than that they never have any meaning; `twill be proper to bestow another consideration on this subject, to see if possibly we can discover the nature and origin of those ideas, we annex to them. (Treatise i6z, original emphasis; see also Enquiry 74)

Hume does not think, then, that expressions attributing powers, forces, and necessary connections to beings and objects are mere nonsense, but neither can their meaning be construed literally, that is, under the supposition that such powers, forces, and necessary connections are an objective feature of beings and objects. Thus, he suggests that for these expressions to be applied under this supposition is for them to be “wrong apply’d” and that to avoid the conclusion that such expressions “never have any meaning” at all we must consider an alternative way of applying or construing them. Let us call this “The Choice” that Hume offers: he insists that we must choose between attempting to construe expressions that attribute powers and necessary connections to beings and objects literally and objectively (in which case we must find them to be altogether meaningless) and adopting an alternative understanding of the content of such attributions on which a defensible meaning can be attributed to them after all. In other words, Hume is preparing to tell us what we could possibly mean by attributing powers and necessary connections to beings and objects, and offering us the option of taking expressions that do so to have that meaning, rather than none at all.

Hume proceeds to argue that necessity is a purely subjective phenomenon. He points out that a single conjunction of objects does not lead us to conceive of a necessary connection between them: it is only when the conjunction is repeated in resembling cases that we are led to have this idea. From this Hume concludes that this repetition must itself “either discover or produce something new, which is the source of that idea [of connection]” (Treatise 163). But, he continues, merely repeating the same conjunction between pairs of objects certainly does not discover anything new in them, nor does repeating a conjunction produce anything new “either in these objects, or in any external body” (Treatise 164). Thus, the ideas of necessity, power, and efficacy “represent not any thing that does or can belong to the objects, which are constantly conjoined” (Treatise 164).

Hume’s famous conclusion, of course, is that repeating a conjunction produces something new only in the mind: an “impression of reflexion” or the “felt determination of the mind” to move from the idea of the cause to that of the effect. He insists that “This determination is the only effect of the resemblance [of the similar conjunctions] and therefore must be the same with power or efficacy, whose idea is deriv’d from the resemblance” (Treatise 165; see also Enquiry 75-6). Moreover, he explicitly points out the implications of this conclusion for the meanings of our causal expressions:

Without considering [necessity] in this view, we can never arrive at the most distant notion of it, or be able to attribute it either to external or internal objects, to spirit or body, to causes or effects. (Treatise 165)

Hume thus rescues himself from the threatened absurdity of finding expressions that attribute necessity and power to beings and objects to be absolutely nonsensical: such attributions are coherent and meaningful, provided we take their content to be a claim about the effects of the repeated conjunctions of the objects upon our minds. The false projectivist and skeptical interpretations notwithstanding, Hume here claims that unless we “consider [necessity] in this view” we will be utterly incapable of making such attributions altogether, and it is this reasoning that supports his conclusion that “necessity is something that exists in the mind, not in objects” (Treatise 165).

But Hume’s ultimate conclusion makes sense only in light of The Choice he has offered us throughout. He claims,

Either we have no idea of necessity, or necessity is nothing but that determination of the thought to pass fray causes to effects and from effects to causes. ac to their experienc’d union. (Treatise 166)

We are free to decide whether we have an idea of necessity or not only in the derivative, linguistic sense that we must choose whether to let the subjective idea Hume has identified (the only real candidate) count as what we mean by necessity. If we insist on construing necessity as an intrinsic property of external objects then we can have no idea of it at all, while our discourse about necessity can be meaningful only on the subjective construal he offers. Thus, Hume glosses his position in the Conclusion of Book I (Teatise 267 my emphasis) by claiming that “when we say we desire to know the ultimate and operating principle, as something, which resides in the external object, we either contradict ourselves or talk without a meaning”: we contradict ourselves if we grant “the ultimate and operating principle” its genuine subjective referent but say we desire to know what it is in the objects, while we talk without a meaning at all if we insist that its referent must be something objective. The Choice reappears yet again in the “Abstract,” where Hume claims “either we have no idea of force and energy, and these words are altogether insignificant, or they can mean nothing but that determination of the thought, acquir’d by habit, to pass from the cause to its usual effect” (Treatise 656-7).

Hume insists that even an objector who finds it a “gross absurdity” to claim that power must be removed from all causes and bestowed upon the mind that perceives them has no more idea of necessity or power in objects than a blind man has of the color of scarlet, and he argues that “We do not understand our own meaning” in claiming that “an efficacy is necessary in all operations” (Treatise 168, my emphasis). Hume thus raises the surprisingly contemporary suggestion that we can be mistaken or misled about the meanings of our own expressions: while we imagine that we are attributing something to the objects in claiming that an efficacy is necessary in all operations, the associated idea that (by Hume’s lights) gives this expression its actual semantic content is, unbeknownst to us, that of a subjective “felt determination of the mind” to move from the idea of a cause to that of its effect; and it is this felt determination itself, the effect in us of an experience of repeated conjunction, to which the expression therefore refers. Similarly, terms like `power,’ ‘efficacy,’ and ‘necessity’ have their semantic content fixed by the idea of a subjective felt determination of the mind and refer to this felt determination itself whether a speaker who uses them realizes this or not. It is thus “of little purpose” (Treatise 168) for Hume’s objector to prove that “an efficacy is necessary in all operations,” because the objector turns out to be saying something only about the subjective determination of our thought, if she is really saying anything at all.

This aspect of Hume’s argument is neglected by Blackburn, whose agnostic brand of projectivist interpretation is otherwise sensitive to a number of the semantic sophistications raised above. Blackburn rightly argues, for instance, that Hume does not hold our first-order use of causal language to be an error or mistake, because he accepts that use of causal language as fundamental and proceeds to try to show us what we mean by it and why we are right to use it when we do. But Hume’s point, according to Blackburn, is not to provide a “lexicographer’s analysis” of the meaning of our causal language; instead it is simply that our use of causal language must be explained by our exposure to regular successions and the consequent functional, nonrepresentative change in our minds and mental dispositions, rather than by appeal to some alleged apprehension of objective necessities. But Blackburn insists that, for Hume, this foundation for explaining our use of causal language does not exhaust the semantic content of that language, for the surface content of our causal locutions (and our use of them) demonstrates that “cause becomes objectified so as to be spoken of as a feature of the real [i.e., objective] world.” And indeed, Blackburn complains” that Hume gives us insufficient guidance concerning how this semantic projection of causal necessity to the objects can possibly work.

There is a textual foundation for the interpretation, of course, in Hume’s famous remark that the mind “spreads itself on the world.” “What Blackburn misses, however, is that Hume is there describing the source of our mistaken belief in the objectivity of causal necessity and not a semantic function or operation that gives additional content to our causal language at all. Contra Blackburn, Hume is not guilty of providing insufficient guidance concerning how we project necessity onto the world because he does not think that we actually manage to accomplish any such projection! Instead, Hume holds that our causal language does not actually have the content that it appears to have; that is why he insists (above) that “we do not understand our own meaning” in proving that an efficacy is necessary in all operations, and, even more explicitly, in the Enquiry, that:

When we say, therefore, that one object is connected with another, we mean only that they have acquired a connexion in our thought, and give rise to this inference, by which they become proofs of each other’s existence: A conclusion which is somewhat extraordinary, but which seems founded on sufficient evidence. (Enquiry 76, my emphasis)

Notice that Hume expects us to find this conclusion “extraordinary” because it is a thesis about the actual content of our causal talk that violates the content that such talk appears or seems to us to have: I can be surprised to learn what I mean by ‘necessary connection,’ because the term has its true (subjectivist) meaning despite appearances or beliefs to the contrary, my ignorance of this content, or even my unwillingness to agree to it.

Blackburn’s projectivist interpretation is also inconsistent with the way Hume proceeds to reconcile his semantic analysis of our causal language with his skepticism. Blackburn’s Hume is forced to allow that our talk of necessary connections, powers, and forces might, after all, “latch on to real, mind-independent, observable-regularity-transcendent facts about reality.” He must also insist that Hume’s point is simply that any relative idea of such a connection “plays no role in our real understanding” and that “the realism concerning it is hardly important compared to his scepticism.” But this allowance is inconsistent with the way that Hume actually proceeds to reconcile his skepticism with his claim to know (on semantic grounds) the subjective nature of necessity and causal power themselves. This reconciliation occurs in a passage that has been widely misconstrued as a concession that there may be objective causal powers or objective necessary connections after all:

I am, indeed, ready to allow, that there may be several qualities both in material and immaterial objects with which we are utterly unacquainted; and if we please to call these power or efficacy, `twill be of little consequence to the world. But when, instead of meaning these unknown qualities, we make the terms of power and efficacy signify something, of which we have a clear idea, and which is incompatible with those objects, to which we apply it, obscurity and error begin to take place, and we are led astray by a false philosophy. This is the case, when we transfer the determination of the thought to external objects, and suppose any real intelligible connexion betwixt them; that being a quality, which can only belong to the mind that considers them. (Treatise 168)

Having just argued that terms like ‘power’ and ‘efficacy’ have an inherently subjective referent in the mouths of all speakers, whether they realize it or not, Hume is not now saying that these terms could refer after all to unknown objective qualities that are somehow responsible for the constant conjunctions we observe. He is, instead, being careful not to draw any dogmatic or anti-skeptical conclusion from his semantic argument. He allows that there might be “several qualities both in material and immaterial objects with which we are utterly unacquainted,” and he recognizes that an objector could insist, say by fiat, that when she says ‘necessity’ or `power,’ she does not mean the “felt determination” Hume has identified but instead some utterly unknown objective quality of external bodies (since none of the known qualities will do). What Hume points out is that it will “be of little consequence to the world” for the objector to salvage objective referents for the mere terms ‘necessity’ and ‘power’ in this way, for she will simply be forced to give up any claim to mean what we do by them. Hume has argued that we employ expressions like ‘necessity’ and ‘power’ when a mental association has been established between the ideas of two constantly conjoined entities, but nothing turns on the name we choose for this phenomenon. If the objector insists that the terms ‘necessity’ and ‘power’ will refer to some intrinsic feature of the conjoined objects themselves, she simply divorces these terms (in her perverse senses) from the phenomena that we use them to characterize. She cannot by linguistic fiat guarantee both that the terms ‘necessity’ and ‘power’ refer to something objective and that they refer to those phenomena that occasion their application for us and in which we take an interest. If we want to talk about what we ordinarily mean by the terms `efficacy,’ `power,’ and `necessary connection,’ we cannot simply agree to mean something objective, any more that we can establish by agreement that the term `apple,’ as we ordinarily use it, refers to pears. We could, of course, agree that the term ‘apple’ will refer to pears in the future, but we would only bear out this agreement if we changed the way we used the term ‘apple’ as well. And if the objector were to bear out her convictions regarding the terms ‘necessity’ and `power,’ she would have to use them as synonyms for “utterly unknown qualities of material and immaterial bodies” and not for our ordinary terms ‘necessity’ and `power.’ Hume insists that the objector must either use causal terms to “signify something of which we have a clear idea” (“a quality, which can only belong to the mind” and that “is incompatible with those objects, to which we apply it”). Or, if she insists on construing causal expressions objectively, she cannot coherently “transfer [or project] the determination of the thought to external objects,” but must instead give up any claim to be talking about what we mean by ‘necessity’ or `power.’

Hume concludes his discussion with his famous dual definition of ’cause’: one in terms of the contiguity, succession, and constant conjunction of objects, and one in terms of the felt determination in our ideas of those objects. His account thus includes the necessity that he says “makes an essential part of causation” (Treatise 407), but only as an inherently subjective mental phenomenon, as he has found any other construal of it to be either devoid of meaning altogether or utterly disconnected from the meanings of our causal expressions.

“Of Liberty and Necessity”

A compelling further source of textual support for my semantic interpretation is found in Hume’s argument in “Of Liberty and Necessity,” which Hume begins by stating his goal of showing “that all men have ever agreed in the doctrine of necessity and of liberty, according to any reasonable sense, which can be put on these terms” (Enquiry 81). He goes on to point out that the account of necessity he has already given, in terms of “the constant conjunction of similar objects, and the consequent inference from one to the other” (Enquiry 82), applies as easily to conjunctions of human motivations with actions as it does to conjunctions of objects in the physical world, and that “the conjunction between motives and voluntary actions is as regular and uniform, as that between the cause and effect in any part of nature” (Enquiry 88). This establishes, at most, that our experience or knowledge of necessity is the same in the case of moral and physical causation (i.e., constant conjunction and consequent inference), but from this Hume moves to the conclusion that the nature of necessity is itself identical in these two domains:

when we consider how aptly natural and moral evidence link together, and form only one chain of argument, we shall make no scruple to allow, that they are of the same nature, and derived from the same principles … Here is a connected chain of natural causes and voluntary actions; but the mind feels no difference between them, in passing from one link to another: Nor is less certain of the future event than if it were connected with the objects present to the memory or senses, by a train of causes, cemented together by what we are pleased to call a physical necessity. The same experienced union has the same effect on the mind, whether the united objects be motives, volitions, and actions; or figure and motion. We may change the names of things; but their nature and their operation on the understanding will never change. (Enquiry go-i, last emphasis mine; see also Treatise 406-7)

Here Hume begins by claiming simply that natural and moral evidence are of the same nature, but ends by claiming that the nature of necessity itself is identical in the two cases: that is, Hume concludes this passage by insisting that the “effect on the mind” of an “experienced union” has precisely the same nature and the same operation on the understanding in moral causation whether we are pleased to call this “necessity” (as we do in the case of physical causation) or not. But if Hume allowed that necessity or power, in itself or in the objects, might be more than, or different from, what our experience captures, he could not argue from the identity of our experience or evidence of necessity in physical and moral causation to the identity of necessity itself in the two cases: the argument is not merely invalid, but saliently so, given Hume’s recognition (see below) of a strong intuition on our part that necessity and causation are not the same in these two spheres. To draw the stronger conclusion, Hume must take himself to have established not merely that our experience of necessity is limited to the felt determination of the mind in both cases, but rather that this subjective determination of thought is what necessity itself is (of course, he has repeatedly said just this). This is rendered yet clearer in the Treatise, where the passage above follows Hume’s insistence that “the necessary connexion is not discover’d by a conclusion of the understanding, but is merely a perception of the mind” (Treatise 405-6).

Hume goes on to suggest that we imagine a difference between moral and physical causes because we mistakenly take ourselves to perceive necessary connections between external objects, but not between our own motives and actions. But finding that our experience of causation is the same in these two cases, he suggests, should lead us to recognize that the necessity of the first is equally present in the second:

being once convinced, that we know nothing farther of causation of any kind, than merely the constant conjunction of objects, and the consequent inference of the mind from one to another, and finding, that these two circumstances are universally allowed to have place in voluntary actions; we may be more easily led to own the same necessity common to all causes. (Enquiry 92)

Hume’s willingness to recommend the conclusion that necessity is the same in the case of physical and moral causation is straightforwardly inconsistent with the skeptical interpretations: if his only point were that our experience of necessity is identical in the two cases, the most Hume could possibly conclude is that we have no reason to think that there is a difference in the nature of necessity in the two cases and that our belief in such a difference is unjustified. But he explicitly encourages us to go beyond this merely skeptical conclusion and to accept the positive claim that necessity itself really is the same in both cases. This constitutes some evidence against Stroud’s false projectivism as well, for it would seem that Hume’s willingness to deny the existence of objective necessity or causal power really cannot be reconciled with his skepticism unless he has eliminated even the meaningful possibility of their existence.

Hume rejects the skeptical interpretations even more clearly in a footnote in which he offers yet another diagnosis of our mistaken belief in a difference between the necessity of physical and moral causes:

viz. a false sensation or seeming experience which we have, or may have, of liberty or indifference, in many of our actions. The necessity of any action, whether of matter or of mind, is not, property speaking, a quality in the agent, but in any thinking or intelligent being, who may consider the action; and it consists chiefly in the determination of his thoughts to infer the existence of that action from some preceding objects…. (Enquiry 94n., emphasis mine; see also Treatise 408)

Hume proceeds to argue that the sense of looseness or indifference we feel when performing actions is simply the lack of any determination in our own thoughts, and furthermore, that a spectator could usually infer our actions from our motives and character, which “is the very essence of necessity, according to the foregoing doctrine” (Enquiry 94n., my emphasis; see also Treatise 409). In one breath, then, Hume offers not only his usual claim that necessity itself is the felt determination in the thoughts of an observer, but also the explicit denial that necessity is a “quality in the [moral or physical] agent, “just the claim that the skeptical interpretations deny he can make.

The Treatise version of this discussion repeats this rejection of the skeptical thesis several times, arguing that “the actions of matter have no necessity, but what is derived from [constant union and the inference of the mind] ” (Treatise 400) and that uniformity in characters and actions “forms the very essence of necessity” (Treatise 403). Hume then proceeds to explicitly connect these claims with his earlier semantic line of argument: he argues, first, that if all known characteristics of natural causation are present in moral causation, it is a “manifest absurdity” to “attribute necessity to the one, and refuse it to the other” (Treatise 404), and, second, that whoever draws “a conclusion concerning the actions of men, deriv’d from a consideration of their motives, temper and situation…. does ipso facto believe the actions of the will to arise from necessity, and that he knows not what he means when he denies it” (Treatise 404-5, last emphasis mine).

Hume’s concluding discussion of liberty and its contrast with necessity undermines both the skeptical and false projectivist interpretations. Because necessity is present whenever actions can be inferred from motives and character, Hume famously argues, liberty is not the absence of necessity, but is instead the absence of constraint. That is, one’s actions exhibit liberty when they are caused by one’s own motives and character:

By liberty, then we can only mean a power of acting or not acting, according to the determination of the will; that is, if we choose to remain at rest, we may; if we choose to move, we also may. Now this hypothetical liberty is universally allowed to belong to every one, who is not a prisoner and in chains. (Enquiry 95)

Hume claims here that liberty can mean only the absence of constraint, and later, that “liberty, when opposed to necessity, not to constraint, is the same thing with chance; which is universally allowed to have no existence” (Enquiry 96). But if a realist, agnostic, or even a false projectivist account were correct, Hume could not possibly offer the absence of constraint as the only defensible definition of `liberty.’ If the existence of objective causal powers and necessary connections were even a coherent possibility, there would be a perfectly legitimate definition of ‘liberty’ as opposed to necessity: liberty in this sense would be simply the absence of such objective power or necessity, and it would be perfectly reasonable to ask whether physical or moral causation exhibits liberty of this opposed-to-necessity kind (even if the answer were unknowable or invariably negative). But Hume claims that liberty can only be opposed to constraint and not to necessity, because liberty opposed to necessity would simply amount to the absurd denial of uniformities and consequent inferences in the realm of human motives, inclinations and conduct. Nevertheless, the realist, agnostic, and projectivist accounts all require the conception of liberty that Hume here rejects as illegitimate, for each seeks to defend a substantive claim that involves it: either that physical and/or moral causation does (false projectivist) or does not (realist) exhibit liberty of this opposed-to-necessity sort, or that it is an open question whether or not it does so (agnostic). But the absence of constraint is the only definition of ‘liberty’ Hume is prepared to countenance as legitimate. This implies that he does not recognize the existence of objective causal powers or necessary connections as even a meaningful possibility. Similarly, if the existence of liberty in the sense opposed to necessity were either an open question (agnosticism) or a controversial matter of truth or falsehood (false projectivism and realism, respectively), Hume could not open “Of Liberty and Necessity” by announcing that “all men have ever agreed in the doctrine both of necessity and of liberty, according to any reasonable sense, which can be put on these terms” (Enquiry 81).

It is unsurprising that the skeptical interpretations run afoul of passages in “Of Liberty and Necessity,” for they are inconsistent with the whole argumentative point of this section as well. In “Of Liberty and Necessity” Hume seeks to show that our long-standing fear of determinism, and our disputes about it and free will, are rooted in a misunderstanding of what determinism involves. The thrust of his argument is that our actions are surely constantly conjoined with particular motives and character traits (and consequently inferable from them) and that, as he has shown, this is all that something’s being “caused” or “necessitated” does or can amount to. Thus, we have all forever agreed that our actions are determined (a much less frightening prospect than it seemed) even when they are the result of our free choices, and the correct contrast to liberty is therefore not the necessity which attends every constant and predictable sequence of events, but the constraint present when our actions are not in fact determined by our own motives and characters. But the desired conclusion patently fails to follow from the weaker premise that all we know or experience of necessitation is constant conjunction and consequent inference, for this leaves open the possibility that our actions are indeed ultimately compelled by an unperceived objective necessity over which we have no control-just the prospect we found so frightening. Hume cannot be offering us (with such fanfare) the cold comfort that, even if we are the puppets of objective necessity, we are too blind to see the strings by which it moves us. His argument makes sense only if we take it to assert that necessity is nothing more than constant conjunction and consequent inference, and thus to reject the skeptical interpretations outright.

Let us move on, then, to consider the strongest textual evidence that has been offered in support of the projectivist and skeptical interpretations. The final two sections will argue that this evidence need not be explained away, since it turns out to offer crucial support for the semantic reading.

Rejecting Projecting

Stroud’s classic argument for the false projectivist position considers and rejects the semantic interpretation’s contention that Hume holds our talk of necessity to assert something about the states of our own minds, rather than to ascribe an intrinsic property to external objects themselves. Stroud points out that this would commit Hume to the “subjectivistic” view that our causal claims are, at least in part, claims about ourselves, and he suggests that this is an implausible account of the content of our causal beliefs that Hume should want to avoid:

Even if there is nothing in reality which our belief adequately represents, still we do seem to have the belief that the connections between things are necessary in themselves, and would remain so whatever happened to be true about US.

Stroud first argues that the source of our idea of necessity must be a feeling or impression of the determination of the mind to pass from one idea to another, rather than this determination itself (as Hume sometimes seems to suggest), for ideas arise from impressions, and a determination of the mind, one idea’s causing another, “could scarcely be that impression, or any impression.” Of course, this feeling cannot be a direct perception of one mental event causing another, or of something connecting the two events, else we would have just what Hume denies: an impression of necessity from observing a particular instance. The only remaining possibility, Stroud suggests, is that the impression of necessity “is just a peculiar feeling that accompanies, or is simultaneous with, the occurrence of that second event in the mind.”

Although we can offer no further characterization of the content of the impression or idea of necessity, Stroud points out that we are in the same boat with respect to every simple impression or idea: asked to characterize the content of our idea of red, we can reply only that it is an idea of red. Most importantly, if the impression of necessity is simply a specific, peculiar, feeling that accompanies some mental transitions, Stroud insists, then there is no reason that its content cannot, like that of other simple impressions, be successfully attributed to objects (albeit falsely). Furthermore, Stroud argues, Hume himself claims, in the following famous passage, that we do indeed successfully accomplish such projection to the objects themselves:

`Tis a common observation, that the mind has a great propensity to spread itself on external objects, and to conjoin with them any internal impressions, which they occasion, and which always make their appearance at the same time that these objects discover themselves to the senses. Thus as certain sounds and smells are always found to attend certain visible objects, we naturally imagine a conjunction, even in place, betwixt the objects and qualities, tho’ the qualities be of such a nature as to admit of no such conjunction, and really exist no where. But of this more fully [here a footnote refers us to Part IV, section 5] hereafter. Mean while `tis sufficient to observe, that the same propensity is the reason, why we suppose necessity and power to lie in the objects we consider, not in our mind, that considers them; notwithstanding it is not possible for us to form the most distant idea of that quality, when it is not taken for the determination of the mind, to pass from the idea of an object to that of its usual attendant. (Treatise 167)

Here Stroud finds Hume to be saying that we do indeed attribute necessity to the objects themselves, just as we attribute colors and other secondary qualities to them, although this is a mistake on our part:

Just as we ascribe redness to certain things in the world only because something happens in our minds when we observe things that … actually possess no redness, so we ascribe necessity to certain things in the world only because something happens in our minds when we observe things (viz. correlations or conjunctions) that possess no necessity. (Stroud, 82)

The “projecting” passage’s analogy to secondary qualities seems to put Stroud’s interpretation on firm textual footing: after all, Hume finds nothing incoherent in our projection of colors to the objects, and he does say that we “suppose necessity and power to lie in the objects we consider.” Nonetheless, a careful reading of this very passage reveals that Hume is indeed denying the coherence of projecting necessity to the objects themselves. Notice first that Hume’s analogy to the projection of sensory qualities does not appeal to colors, perhaps the most natural example of such projection, but to sounds and smells instead. Furthermore, Hume says that sounds and smells are “of such a nature as to admit of no such conjunction” (my emphasis), suggesting that it does not simply happen to be the case that there are no such qualities in objects, but that the nature of these qualities themselves somehow precludes them from being in things.

The full significance of these points becomes clear when we pursue Hume’s explicit invitation to hear “of this more fully hereafter” in “Of the Immateriality of the Soul” (Part IV, section 5), where Hume proceeds to argue in considerable detail that we are not capable of ascribing non-visual, non-tactile sensible qualities to objects after all. He acknowledges that just as in the case of necessity) we commonly suppose that we project such qualities (e.g., sounds and smells) to the objects themselves, but he offers several arguments for the conclusion that such qualities cannot be coherently understood to occupy spatial positions, and that our attempted projection of them to objects is therefore ultimately unsuccessful. He argues, for example, that whatever occupies a spatial location must do so either by being extended or in the manner of a mathematical point, conditions that could only “agree to” impressions of sight and touch: desires and tastes are not circular or triangular, and if they existed as mathematical points we would be able to arrange, for example, four of them into a square pattern (at least in thought)—a suggestion that is “evidently absurd” (Treatise 235, 239). He concludes that the “greatest part of beings” (in fact, all but their visual and tactile qualities) not only in fact exist “nowhere” (i.e., in no physical location), but must do so: “so far from requiring any particular place, [they] are absolutely incompatible with it, and even the imagination cannot attribute it to them” (Treatise 235-6).

Hume offers a characteristically Humean account of why we try to conjoin these perceptual qualities with those that are extended and divisible, blaming the mind for trying to ease our mental transitions by adding a further relation (conjunction in place) to two relations we find to hold between perceptions (causation and conjunction in time). But the attempted attribution must fail, as Hume insists in his clearest argument against it, expressed in the form of a dilemma:

But whatever confus’d notions we may form of an union in place betwixt an extended body, as a fig, and its particular taste, ’tis certain that upon reflection we must observe in this union something altogether unintelligible and contradictory. For shou’d we ask ourselves one obvious question, viz. if the taste, which we conceive to be contain’d in the circumference of the body, is in every part of it or in one only, we must quickly find ourselves at a loss, and perceive the impossibility of ever giving a satisfactory answer. We cannot reply, that ’tis only in one part: For experience convinces us, that even, part has the same relish. We can as little reply, that it exists in every part: For then we must suppose it figur’d and extended; which is absurd and incomprehensible … We suppose, that the taste exists within the circumference of the body, but in such a manner, that it fills the whole without extension, and exists entire in every part without separation … Which is much the same, as if we shou’d say, that a thing is in a certain place, and yet is not there. All this absurdity proceeds from our endeavouring to bestow a place on what is utterly incapable of it…. (Treatise 238)

Thus Hume holds that we try but fail to project non-visual, non-tactile secondary qualities onto objects: such qualities are intrinsically incapable of being understood as occupying spatial locations, and not even the imagination can accomplish this incomprehensible projection. Of course, this evidence weighs just as heavily against agnostic projectivists like Blackburn and Penelhum as it does against false projectivists like Stroud (and Pears), for Hume here argues that the very attempt to project necessity onto the world in the first place is incoherent and therefore doomed to failure.

We can also now see why the famous projection passage says that we suppose necessity and power to lie in the objects, a claim that falls into a consistent pattern (noted by Strawson) of distinguishing what we can conceive from what we can merely suppose. As Winkler documents, a distinctive late eighteenth-century usage draws this distinction by permitting us to “suppose” absolutely anything, even contradictions, as mathematicians do in reductio proofs. Thus, we may take Hume at his word in the famous projection passage, as saying that we do indeed suppose necessity to be in the objects and holding, nonetheless, that this supposition is itself incoherent, just as he says above that we “suppose” the taste of a fig to exist within the circumference of its body, despite the absurdity of its so doing.

Hume’s discussion in “Of the Immateriality of the Soul” makes clear that the famous projection passage is not claiming that we attribute necessity to objects just as we attribute secondary qualities in general to them. Instead, Hume is analogizing the projection of necessity to the objects to the “absurd,” “unintelligible,” and “contradictory” projection of specifically non-visual and non-tactile secondary qualities to them. Furthermore, Hume’s conclusion is explicitly semantic: no coherent meaning can be assigned to these “suppositions”; thus the attempted projections fail and the central textual evidence for projectivism actually supports the semantic reading after all.

Strawson’s “Double Meaning” Theory and Our Relative Idea of Objects

Although he discusses neither “Of Liberty and Necessity” nor “Of the Immateriality of the Soul,” Strawson is certainly aware that Hume sometimes seems to reject outright the intelligibility or coherence of attributing necessity or power to objects, and his reply to this line of thought threatens to undermine any direct textual evidence for the semantic interpretation. Strawson claims that Hume’s discussions of meaning reveal a “Meaning Tension” (in fact no more than an ordinary ambiguity in the term ‘mean’) and that Hume wants to allow that we can refer to objective causal powers (and therefore to causation in the objects as more than mere regularity) even if our idea of it cannot be positively contentful or approved by the Theory of Ideas. Thus, when Strawson’s Hume claims that we have no “idea” of objective causal powers, that our terms cannot “mean” them, and that it is “unintelligible” to suppose that they exist, he means only that we can form no Theory-of-Ideas-approved idea of objective causal powers, that our terms do not stand for a contentful conception of them, and that such powers are “not positively conceivable by us.” And Strawson’s Hume recognizes looser senses of semantic terms like `idea,’ `meaning,’ and ‘intelligibility’ on which we can have ideas of, refer to, and understand objective causal powers after all.

The most important line of evidence Strawson offers in support of this “Double Meaning” theory, is Hume’s discussion of our ideas of external objects. As Strawson points out, Hume’s discussion of causation is analogous in some ways to his discussion of external objects, of which Hume thinks “it is impossible for us so much as to conceive or form an idea of any thing specifically different from ideas and impressions” (Treatise 67). But instead of concluding that we have no idea whatsoever of objects distinct from our impressions, or that the term ‘object’ can only “manage to mean” a collection of ideas and impressions, Hume allows that we can have what he calls a “relative idea” of external objects (Treatise 68): an idea of them as whatever it is that stands in a certain relation to our impressions themselves. This idea does not have a descriptive, Theory-of-Ideas content, Strawson says, but it is “not no idea at all.” He goes on to suggest that Hume must allow us an analogous relative idea of objective causal powers, that is, of “that about reality which is in fact the reason why reality is regular in the way that it is,” and that this idea enables us to refer to causal powers even if we can form no positive or contentful idea of what they are like. Nonetheless, a careful reading of Hume’s discussion of our relative ideas of the external world shows why this parallel cannot ultimately be sustained, and why Hume could not have embraced the relative idea Strawson proposes.

The apparent problem for my semantic reading arises most perspicuously in Hume’s consideration of the “philosophical” (Lockean or representative realist) view that our perceptions are merely sensory representations of continuously and independently subsisting material objects. While Hume certainly argues that this Lockean view of the external world is rationally unjustified, and at times even seems to hold that it is outright false, he does not seem to hold that it is meaningless, incoherent, or unintelligible, claiming that “The farthest we can go towards a conception of external objects, when supposed specifically different from our perceptions, is to form a relative idea of them, without pretending to comprehend the related objects” (Treatise 68). Here Hume seems to allow that we can refer to objects considered “specifically different” from our perceptions by means of a relative idea of them, not based directly upon any impression.

We must ask, then, why Hume would deny that the term ‘necessity’ could refer to anything beyond features of our experience, but not consign the term ‘object’ to the same semantic fate (i.e., allow that the Lockean conception of objects is meaningful), for an examination of his text makes clear that Hume does not accept an analogous relative idea of causal power or necessary connection. In the first place, Hume nowhere asserts or even suggests that we have the kind of relative idea of causal power or necessary connection that Strawson claims he allows. Furthermore, he explicitly discusses our relative idea of external objects only in the Treatise, where he emphatically rejects the notion that we have any such idea of necessity or power. He claims instead that if we do not consider necessity simply “an internal impression of the mind … we can never arrive at the most distant notion of it,” that it is not “possible for us ever to form the most distant idea of [necessity], consider’d as a quality in bodies” (Treatise 165, my emphasis), and that “it is not possible for us to form the most distant idea of [necessity and power], when it is not taken for the determination of the mind” (Treatise 167, my emphasis).

In addition, in both works Hume explicitly rejects a Lockean proposal very close to the sort of relative idea of causal power that Strawson proposes:

Mr. Locke, in his chapter of power, says that, finding from experience, that there are several new productions in nature, and concluding that there must somewhere be a power capable of producing them, we arrive at last by this reasoning at the idea of power. But no reasoning can ever give us a new, original, simple idea; as this philosopher himself confesses. This, therefore, can never be the origin of that idea. (Enquiry 64n; see also Treatise 157)

This passage makes Strawson’s “Double Meaning” theory look quite implausible, for Hume here explicitly applies the standards of the Theory of Ideas to reject what is essentially a relative idea of causal power. Of course, Strawson will surely insist that Hume means here to argue only that Locke’s proposal fails to offer any positive conception of causal power. But even if we accept this supposition, Hume would have to be deliberately trying to mislead us about his views if he were to reject Locke’s suggestion without explicitly acknowledging that we indeed have a relative idea of causal power and insisting that it is simply not an idea of the right (Theory-of-Ideas-approved) kind. The more plausible conclusion is that Hume rejects Locke’s suggestion because he does not think we have any relative idea of causal power or necessary connection at all.

Finally, Hume points out that our idea of power is a relative one on his own account, but one that picks out the unknown cause constantly conjoined to a particular effect, and not any connection between the two:

According to these explications and definitions, the idea of power is relative as much as that of cause, and both have a reference to an effect, or some other event constantly conjoined with the former. When we consider the unknown circumstance of an object, by which the degree or quantity of its effect is fixed and determined, we call that its power… (Enquiry 77n)

This suggests, of course, that Strawson has misread his favorite passages from Hume on powers: when Hume claims that “nature … conceals from us those powers and principles on which the influence … of objects entirely depends” (Enquiry 32-3) and that we are “ignorant of those powers and forces, on which [the] regular succession of objects totally depends” (Enquiry 55), he is bemoaning only the fact that we are frequently ignorant of which particular characteristics of objects are constantly conjoined with particular effects, and not our ignorance of some supposed objective tie or connection between them. More importantly, however, this passage suggests that Hume carefully considered the question of a relative idea of causal power and concluded that a relative idea of the unknown cause of some particular effect (a Humean “power”) was legitimate, while a relative idea of the objective or intrinsic connection between a cause and its effect (a Strawsonian “power”) was not. We may therefore conclude that when Hume talks of hidden or secret powers it is in the sense of the former, legitimate, relative idea and not the latter, illegitimate one, at least when he intends these expressions to refer at all.

That Hume drew this sophisticated distinction suggests that he had some reason for refusing to accept a relative idea of Strawsonian power or necessary connection, and there is indeed a compelling, albeit simple, ground for such a refusal: while the apparatus of Hume’s Theory of Ideas can generate relative ideas of both external objects and Humean causal powers, the Theory has no resources for generating a legitimate relative idea of objective necessity or Strawsonian power. Let us see why.

On Hume’s account (Treatise 13f), our ideas of relations are of seven general (and jointly exhaustive) kinds: resemblance, identity, relations of space and time, relations of quantity, relations of quality (and degree), contrariety, and cause and effect. These complex ideas are obtained from considering some particular respect in which two particular entities are related, following which we may let such an idea be either specific or general in its signification (e.g., represent either “being separated by ten feet” or “being spatially separated”). We use such ideas of relations to form relative ideas by uniting the idea of a single entity (relatum) with that of some relation it can bear to other entities, generating relative ideas like “whatever is ten feet north of me” or “whatever caused your hiccups.” It is by this means that we may form the relative idea of Lockean external objects, or of Humean causal powers. The Lockean conception of an external object is simply “whatever it is that causes these sense-impressions,” while the relative idea of a Humean power is “whatever causes this particular effect,” generated from the idea of some particular phenomenon and (again) the relation of causation. But of the seven kinds of Humean relations, only causation itself is even a plausible candidate for grounding a relative idea of objective necessary connection or Strawsonian power: that is, the relative idea of objective necessary connection or Strawsonian power must be “whatever causes observable regularities of succession (or causes some particular regularity) to occur.”

The problem with this relative idea of Strawsonian causal powers is that articulating the conditions something would have to meet in order to satisfy it lands us in an infinite regress. On this account of what causal powers are, a particular regularity of succession is not genuinely causal unless its instances are not merely attended by or occur in conjunction with a causal power, but instead occur in virtue of or because of the presence of that power; this, after all, is what is required for the causal power to be the thing that causes the observable regularity of succession to obtain. But this has the unfortunate implication that the relationship between the power and the instance of the regularity cannot itself be a causal one, unless there is a further power in virtue of which the relation between the first one and the instance of the regularity obtains; for, as we have already noted, on this account of the matter, truly causal successions must occur in virtue of, or because of, causal powers. Of course, the conjunction between the second power, on the one hand, and the conjunction between the first one and the instance of the regularity, on the other, will require yet a further power if it is in turn to be truly causal, and so on (see figure below).

In other words, the only thing that could satisfy the proposed relative idea of causal power, as “whatever it is that causes regularities of succession to occur,” is a metaphysically preposterous infinite stack of such powers. A single objective power cannot satisfy the relative idea, because a further power will be required to render it the genuine cause of (rather than being merely conjoined with) the instance of the regularity, which will, in turn, require a further power to render the relation between the second and the conjunction between the first and the instance of the regularity causal, and so on, ad infinitum. There is, therefore, a very simple reason why Hume must reject the relative idea of Strawsonian causal power but not that of Lockean external objects or Humean causal powers: his Theory of Ideas provides the resources for generating the latter relative ideas, but not the former.

As in the case of the projectivist account, then, a careful reading of the evidence offered against the semantic interpretation turns out to support it after all: Hume’s doctrine of relative ideas supports a Humean, but not a Strawsonian, relative idea of causal power. Without Strawson’s “Double Meaning” theory and its perverse reconstruction of Hume’s attacks on objective causal power and necessary connection, the textual considerations offered in favor of the semantic interpretation and against the realist, agnostic, and false projectivist alternatives reassert their full force, driving us to conclude that Hume can, and does, on fairly sophisticated semantic grounds and in a manner fully consistent with his skepticism, deny the existence of objective causal powers or objective necessary connections as even a coherent or meaningful metaphysical possibility.