Rationalism & Empiricism

What is distinctive of ‘rationalism’ as opposed to ‘empiricism’? I mentioned that Buckle and Millican approach matters by comparing Hume with Descartes, but both are fully aware that the philosophers falling under the traditional extension of ‘rationalist’ – Descartes, Spinoza, Malebranche and Leibniz – differ considerably among themselves on a whole panoply of issues. These include the nature and number of substance, the metaphysics of causation, the status of modal fact, and the nature of ideas and representation. Leibniz reintroduces elements of Aristotelianism and final causation, hardly things Descartes would embrace. Spinoza disagrees with Descartes and others on the relation of attribute to substance. Leibniz argues strenuously against Descartes' voluntarism about necessary truth, a doctrine also falling under Malebranche's sights. Furthermore, the traditional extension becomes somewhat fuzzy for Millican in that he holds that for some purposes aspects of Locke's philosophy, and other British philosophers, count as ‘rationalistic’, underscoring his own awareness of the complexities here.

These fairly obvious points made above throw into relief the limitations of an illustrative focus on Descartes. Without any independent grasp of what ‘rationalism’ means, what can we learn other than that Hume disagrees or agrees with Descartes? When is any such difference a difference between a rationalist element and an empiricist element? Consider, for example, Millican's use of Descartes. Millican identifies 23 theses held by Descartes but denied by Hume.Footnote ⁶ These include: a) we can have a priori knowledge of minds, b) the workings of the mind are transparent to introspection, c) the mind is better known than the body, d) the self is revealed through introspection as an indivisible unity whose essence is to think, e) the mind has innate ideas and f) we can prove with certainty – i.e. demonstrate – the existence of an external world. But it is not just Hume who denies these: so does Malebranche.Footnote ⁷ Is Malebranche here being an ‘empiricist’ in objecting to Descartes? Or do these differences between Descartes and Malebranche rest on something quite different? Is Malebranche really a ‘disciple’Footnote ⁸ of Descartes? Given that Malebranche and Descartes differ considerably, it is not immediately evident why any difference between Hume and Descartes marks a difference between an empiricist element in Hume and a rationalist element in Descartes. In the absence of any independent characterization of ‘rationalism’, we have only a list of differences between Hume and Descartes and nothing more.

Whatever Buckle or Millican might intend by ‘rationalism’, and its cognates, there are two clear and related features that serve to draw a principled distinction between the traditional ‘empiricists’ and the ‘rationalists’. First, rationalists share a commitment to a non-sensory form of representation, the intellect, which empiricists reject. Second, the intellect affords a grasp of the nature of substance, which empiricists also reject. Ironically, it is Descartes, in his famous discussion of a piece of wax in the Second Meditation, who offers the best illustration of these related theses. In considering a piece of wax, Descartes claims to ‘understand’ the wax distinctly, but that which is understood includes ‘none of the features arrived at by means of the senses’.Footnote ⁹ One reason offered for this claim is that we have a conception of something remaining unchanged throughout a change in its sensible qualities. But, more crucially, there is an element of modal knowledge that speaks in favour of a ‘purely mental scrutiny’ outstripping the perception of vision, touch or imagination.Footnote ¹⁰ Descartes says he can grasp that ‘the wax is capable of countless changes [of modes of extension], yet I am unable to run through this immeasurable number of changes in my imagination, from which it follows that it is not the faculty of imagination that gives me my grasp of the wax’.Footnote ¹¹ The piece of wax then ‘is not perceived by the senses or the faculty of the imagination but by the intellect alone’.Footnote ¹² The modal knowledge provided by the intellect provides a grasp of substance, the second mark of rationalism. For it furnishes a way of distinguishing that which is dependent – modes – from that upon which they depend, namely the essence. This knowledge involves non-sensory representations that are also inherently general, which allows for the modal knowledge required for knowledge of substance and mode. Malebranche (like Spinoza and Leibniz) is a rationalist inasmuch as he shares with Descartes a commitment to the intellect and its potential grasp of substance. Ideas of the intellect ‘represent things to the mind in a way so clear that we can discover by simple perception whether such and such modifications belong to them’.Footnote ¹³

If there is anything distinctive of empiricism, it is a rejection of the intellect. Locke famously holds that all representations emerge through outer and inner sensory experience, and Berkeley explicitly rejects the intellect. ‘Pure intellect’ he writes, ‘I understand not’.Footnote ¹⁴ And Hume is clearly aware that this is a fundamental point of difference between himself and Malebranche. Thus in the Abstract he writes:

Hume's challenge to Malebranche and the intellect rests on the ‘Copy Principle’, the thesis that ‘all our simple ideas, in their first appearance are deriv'd from simple impressions, which are correspondent to them, and which they exactly represent’.Footnote ¹⁶ All representation, for Hume, depends on sensory materials to furnish its contents. In connection with the intellect, he writes that some hold that there are ideas ‘of so refin'd and spiritual a nature, that they fall not under the conception of the fancy, but must be comprehended by a pure and intellectual view, of which the superior faculties of the soul are alone capable’.Footnote ¹⁷ Malebranche himself called intellectual representations ‘spiritual’.Footnote ¹⁸ Hume thinks that the Copy Principle ‘destroy[s] this artifice’.Footnote ¹⁹

I shall not comment on whether Hume's appeal to the Copy Principle really succeeds against the intellect. The point is simply that this is a key and explicit difference between the two thinkers. Before we continue, let me make two further comments on ‘rationalism’ and ‘empiricism’. First, the controversy over innate ideas tends to be associated with the rationalist/empiricist division, and it is quite true that Descartes and Leibniz explain the intellect by appeal to innate representations. However, the existence of the intellect is logically independent from the doctrine of innate ideas. Malebranche is a case in point, for he is a stern critic of innate ideas and yet subscribes to the intellect.Footnote ²⁰ For him, our representational capacities involve direct cognitive contact with quasi-platonic entities constituting the blueprints, as it were, for the created world.Footnote ²¹ So Buckle's claim that Hume's brevity over the issue of innate ideas is evidence that Hume does not really target rationalism does not touch the rationalist with whom Hume is most engaged.Footnote ²² Hume's strategy against the intellect need not be an attack on innate ideas but a matter of establishing empirically an alternative enshrined in the Copy Principle.

Second, a different feature is often said to mark an empiricist/rationalist distinction. Rationalists, supposedly unlike empiricists, subscribe to the thesis that the world is, in principle, ‘intelligible’. This is sometimes expressed as the idea that there is a reason for each and every state of affairs (the principle of sufficient reason). I have elected not to include this, and not merely because the meanings of ‘intelligible’ and ‘sufficient reason’ are difficult to determine. Rather, it is a matter of controversy as to whether Berkeley, Locke or even Hume rejected the ‘in principle intelligibility’ of the world. Berkeley's view of natural events as sign and signifier, constituting the language of God spoken in a providential world, seems to be a candidate for an intelligible world.Footnote ²³ Michael Ayers argues that Locke is an ‘ontological rationalist’ who holds that the world does have an intelligible structure, but not intelligible to us because we lack the intellectual faculty necessary to grasp it.Footnote ²⁴ Following John P. Wright,Footnote ²⁵ he also holds that the same is true for Hume, though this realist dimension to Hume is accompanied by a more thoroughgoing epistemological scepticism than Locke's. Rather than taking a stand on this issue, then, I shall bracket the issue of the ‘intelligibility’ of the world as a distinguishing mark of the rationalism/empiricism divide.

Setting the Scene: Malebranche's Influence on Hume

We begin with a summary of the extent and nature of Malebranche's influence on Hume. As this influence is documented elsewhere, I shall confine myself to a few brief remarks.Footnote ²⁶

In his 1916 study, C. W. Doxee claimed the influence to be singularly striking because ‘[r]arely do we discover an empiricist and a rationalist that teach identical doctrines, particularly when the doctrines in question are almost equally fundamental in the thinking of both’.Footnote ²⁷ That Hume and Malebranche teach ‘identical doctrines’ is rather an exaggeration, but Doxee identified extensive similarities between (i) their treatment of causal relations (ii) the unknowability of any substantial self and (iii) Malebranche's theory of natural judgment and what Kemp Smith called Hume's ‘natural beliefs’. Later work finessed and added to Doxee's study. In addition to those Doxee noted, the other discernible lines of influence include the following: Hume's view of truth as relations of ideas,Footnote ²⁸ the basic mechanics of sympathy,Footnote ²⁹ and Hume's treatment of the passions.Footnote ³⁰ Hume's view of secondary qualities and his projective metaphor of ‘spreading the mind’ come from Malebranche.Footnote ³¹ Hume's treatment of the associative relation of resemblance and its relation to error in the Treatise leans heavily on Malebranche's account of liaison naturelle (which Hume calls ‘natural relations’), as does his general account of association.Footnote ³²

This last example provides a good illustration of the kind of evidence we have for the influence. In accounting for association by resemblance, Malebranche appeals to ‘animal spirits’, writing that they

Now, compare the wording of this passage with one from the Treatise where Hume attempts to explain the source of the associative relation of resemblance:

The similarity of wording here is unmistakable,Footnote ³⁵ and indeed aside from the number of explicit references to Malebranche in Hume's works, the Treatise contains a significant number of passages that are near word-for-word translations of passages from the Search. Thus Malebranche:

And Hume?

One can find many such passagesFootnote ³⁸ (and nothing comparable from the work of Descartes as far as I can determine), showing that a direct line of influence of Malebranche on Hume is highly probable.Footnote ³⁹

It is not surprising, then, that the Search tops a list of works Hume advised his friend Michael Ramsey to read in preparation for the Treatise. Quite obviously, it would be absurd to say that Malebranche is Hume's sole or most significant influence. Nor should ‘influence’ be taken to mean mere slavish copying. As Charles Hendel wrote, Hume did not follow Malebranche ‘as a disciple’.Footnote ⁴⁰ Hume appropriates material drawn from Malebranche and uses it for his own ends. And clearly Hume differs from Malebranche in many fundamental ways. In what follows I show that these differences depend on a rationalist/empiricist distinction.

Causation

Malebranche's influence is at its most pronounced in Hume's treatment of causal power. At a superficial level, it is evidenced in T.1.3.14 ‘Of the idea of necessary connexion’, where there are near word-for-word translations of passages from the Search. Furthermore, many examples Hume employs, in both the Treatise and the first Enquiry, can be traced to Malebranche. The famous example of the billiard balls from §7 of the first Enquiry is in Elucidation 15 of the Search. Both thinkers, in rejecting the thesis that we can observe efficacy in the operations of the will, centre their discussions on the example of a paralyzed man attempting to move his arm (Search Elucidation 15; Enquiry §7). Both reject the claim that causal power is observable, but nevertheless offer an explanation of the mistaken belief that it is. Hume says that the vulgar ‘imagine that they perceive a connexion betwixt such objects as they have constantly found united together’,Footnote ⁴¹ and ‘suppose … they perceive the very force or energy of the causes’.Footnote ⁴² The source of this mistake for Hume is the mind's ‘great propensity to spread itself on external objects’,Footnote ⁴³ whereby we project an inner sentiment onto the world. This explanation again has its source in the Search. ‘When I see one ball strike another’, Malebranche writes, ‘my eyes tell me, or seem to tell me, that the one is truly the cause of the motion it impresses on the other’.Footnote ⁴⁴ But my ‘senses seduce me’:Footnote ⁴⁵ my experience of causal transactions is akin to misleading secondary property experience, owing itself to the fact that the soul ‘spreads itself (répandre)’Footnote ⁴⁶ onto external objects.Footnote ⁴⁷

For both thinkers, this vulgar error in the epistemology of causation contributes to a dialectic between a ‘true’ and a ‘false’ philosophy of causal power. The misleading impression of power becomes yoked to a philosophy of substantial form. Malebranche holds that ‘pagan philosophy and … the impressions of the senses’ lead to a false and ‘detestable philosophy’,Footnote ⁴⁸ whereby objects have genuine causal powers qua substantial forms. Hume thinks that ancient philosophy is a ‘false philosophy’,Footnote ⁴⁹ an ‘invention of the words, faculty and occult quality’Footnote ⁵⁰ in the face of causal transactions that we cannot understand. For despite vulgar appearances, we have ‘no idea of power or agency, separate from the mind, and belonging to causes’.Footnote ⁵¹ Malebranche agrees: ‘[w]hatever effort I make to understand it, I cannot find in me any idea representing to me what might be the force or power they attribute to created objects’.Footnote ⁵² Hence our attribution of power is ‘a fiction of the mind of which we have naturally no idea, [and] it is the fancy that leads everyone to imagine it’.Footnote ⁵³ It is clear then that Hume was deeply engaged with Malebranche. It is equally clear, however, that they arrive at radically different conclusions regarding causal power. We shall see that their respective argumentative strategies are again strikingly similar, but the differences fall out of the empiricist/rationalist divide on the intellect.

What though are their respective conclusions? Malebranche argues for occasionalism, the thesis that all power resides in God's volitional activity. Occasionalism entails the rejection of ‘second causes’. To believe in second causes is to believe that causal transactions involve the exercise of powers possessed by the particular objects involved. Occasionalism, by contrast, holds that causal relations among objects consist in those objects standing in regularities, underwritten by God's will, that are identical to laws of nature. The truth of ‘A causes B’ depends not on A's power to bring about B, but in there being a God-sustained regularity whereby B follows A. Hume, on the other hand, is famously sceptical about power, holding that all that we can understand of the causal relation is captured in his ‘two definitions’. No objective power is manifest in any instance of a relation between objects that we deem causal. Our idea of necessary connection is not derived from any observation or impression of genuine power but instead from a feeling of compulsion, acquired through repeated observation of effect following cause. Beyond all this we cannot conceive or understand any more to the causal relation.

These then are very different conclusions. But these differences in conclusion do not by themselves mark any rationalist/empiricist contrast. It is true that Hume explicitly criticizes Malebranche's theory of causation, but then so does Leibniz. Berkeley, on the other hand, accepted a similar, though by no means identical, view of the causal relations in the natural world to that of Malebranche. It is not so much the differences in conclusions that are important for our concerns but how they are reached. We can approach this matter from what is the central point of contact between Hume and Malebranche. Both conceive of powers as necessary connections.Footnote ⁵⁴ Both hold that any instance of a power must relate A and B with absolute necessity, such that in any world where A exists, B exists. This metaphysical relation is understood to imply a certain modal epistemology. Any such power, when grasped, is supposed to render it impossible for the thinker to conceive cause A without effect B. It is sometimes said that because Malebranche and Hume view causal powers in this way, they are guilty of conflating ‘natural’ with ‘logical’ necessity,Footnote ⁵⁵ or of setting too high a standard for the causal relation. But, somewhat briefly and dogmatically, this is not so. First, the relevant modal relations hold between objects or events, whereas logical necessity (whatever else it might be) relates conceptual or semantic items. So although this metaphysical relation is reflected in a conceptual relation grounding the impossibility of conceiving A independently of B, the relevant modality is a metaphysical one. It is not constituted by relations of necessitation between the concepts that pick out the relevant relata. The impossibility of finding anything else conceivable expresses a grasp of the metaphysical impossibility of A without B: it does not constitute it.

Second, the identification of causal necessitation with metaphysical necessitation is not an arbitrarily high standard for power. To see why we need to distinguish sharply what our concept of A causing B might be from what might be the metaphysical grounds for any power, anything more than regular succession. Our concept of ‘cause’ commits us to more than a simple relation of regular succession holding between types A and B. We might, for example, believe A and B are causally related just in case we believe (implicitly) that B is counterfactually dependent on A. So, pace Hume and Malebranche, we have no need of metaphysical necessity to distinguish genuine causal relations from mere regularities, for the conceptual difference can be marked by mere counterfactual dependence. But even if we grant that our ordinary concept of causal necessity implies only counterfactual dependence and not absolute necessity, this still leaves open a metaphysical question. Is there anything more metaphysically speaking to such relations other than regular succession? At the metaphysical level we are offered a stark choice: either such relations hold contingently, which means that ultimately speaking causal relations are a matter of brute regularities – contrary to our supposition that there is more to causation than regularities - or they must be grounded in absolute necessity. For at the metaphysical level there are two and only two modalities: contingency and necessity. Thus David Lewis's counterfactual analysis of the concept of causation is made against a metaphysic of contingent regularities, some of which are dignified with the term ‘law of nature’.Footnote ⁵⁶ If you maintain that there is metaphysically speaking more to the world than mere regularities, then there is no other modality other than absolute necessity. Hume and Malebranche do not therefore set some arbitrarily high standard for the existence of causal powers but know quite rightly that if there is anything more to the relation than regularity it has to be absolute necessity.

Hume and Malebranche agree on this point and the associated epistemology of conceivability, but differ over whether one can determine any relation of necessary connection. For Malebranche there is one, and only one, relation of necessary connection, and hence only one genuine power, namely that between God's will and its objects:

Clearly here Malebranche is arriving at his occasionalist conclusion by appeal to the notion of power as absolute necessity, and the related epistemology of conceivability. Hume, on the other hand, uses conceivability to reject the claim that any necessary connexion can be found among the supposed relata of causation. He writes

The fundamental difference between the two then is that Malebranche finds God's acts of will and their effects inconceivable from one another, whereas Hume simply denies any such relation. But there has to be more to matters than this, otherwise we have not much more than a relatively uninteresting standoff between the two, a matter of Hume simply denying what Malebranche claims. The difference lies in the account of conceivability. Now, as we shall explore in connection with the self, both thinkers connect conceivability with ‘separability’. To conceive A apart from B is to ‘separate’ A from B. So what further difference is there between them? Notice Hume holds that we have no idea of necessary connection because a grasp of such a connection would ‘imply the absolute impossibility for the object not to follow, or to be conceiv'd not to follow upon the other’. To support this claim he states ‘as all distinct ideas are separable, ’tis evident there can be no impossibility of that kind’.Footnote ⁵⁹ So because ‘distinct’ ideas are ‘separable’ we can never therefore grasp necessity: for to find B inconceivable apart from A is to find them inseparable, but given that the ideas are ‘distinct’, we can always separate A from B, and so find them independently conceivable. So the following seems to be on the cards. Hume holds that God's volition and its effect are distinct, and so, contra Malebranche, they are independently conceivable, and hence not necessarily connected.

By itself, however, this gets us little further than a simple standoff. For unless we are given some independent conception of what ‘distinct’ means here, Malebranche is not forced to accept this conclusion – he could simply deny that God's will and its objects are ‘distinct’ in the relevant sense. The key, and principled, difference rests in the theory of conceptual representation with which each operates. Recall that Hume's Copy Principle holds that ideas – concepts - are exact copies of sensory impressions, such that the only difference between sensory experiences, on the one hand, and ideas or concepts one the other, lies in their relative degree of vivacity. So any ‘distinctness’ in ideas must be inherited from the character of the sensory impressions from which such ideas are derived. Now, the only kind of ‘distinctness’ that sensory experience can yield is phenomenal distinctness: it cannot itself be conceptual (on pain of turning Hume into Kant). Hence given that ideas are exact copies of sensory impressions, and sensory impressions are essentially phenomenal items, the distinctness of ideas must be phenomenal.

This is a direct result of the thesis that all ideas are derived from, and exactly resemble, sensory impressions. Any conceptual difference is grounded in the phenomenal distinctness of sensory experience. Any cause and effect are separately conceivable because they are phenomenally distinct. But Malebranche holds that our concepts are of a different nature, for he holds that ideas are concepts that can represent independently of sensory experience, and so conceptual distinctness is not constrained by phenomenal distinctness. So his conception of conceptual distinctness need not imply that the idea of God's will and its effect are separable, and independently conceivable. For him, we cannot conceive God's volition failing to have its effect, presumably on the grounds that omnipotence conceptually implies necessity, and conceptual implication is not determined by phenomenal character. Hence the difference between the two thinkers has to be down to the rationalist/empiricist divide. Hume ultimately disagrees with Malebranche because his account of conceptual distinctness is governed by the thesis that all concepts are determined by their sensory origins, and so distinctness must have a phenomenal character. Hume's theory of mind is a matter of separating images, and so that shows that there is no necessary connection knowable among distinct ideas. It is because Malebranche does not subscribe to this thesis about concepts that Hume and Malebranche part company. This is why Hume thinks we have no idea of necessary connection.

I said Hume explicitly criticizes Malebranche, and indeed he does.Footnote ⁶⁰ But his key criticism presupposes the more fundamental difference regarding the intellect we have just discussed. Against Malebranche, Hume argues that since the Copy Principle is true, and the doctrine of innate ideas is false, any idea we have, including that of power or necessary connection, as well as the idea of God, must be constructed from sensory materials. But we have no sensory-derived idea of power, and without an idea of power we cannot ascribe power to the deity.Footnote ⁶¹ The premises of this argument though presuppose the more fundamental rejection of the intellect we have just been discussing. The sensory nature of Humean ideas makes all ideas ‘loose and separable’, allowing us to conceive the independent existence of all constituent ideas. Grasping a necessary connection, on the other hand, would supposedly render it impossible to conceive A apart from B. Such a grasp is thereby impossible because of the loose and separable nature of sensory impressions.Footnote ⁶²

Self and Substance

We now turn to Hume and Malebranche on self-knowledge. Both deny that we have an idea of a substantial self, maintaining that introspection reveals only particular perceptions rather than any substance upon which those perceptions depend. Although Hume rejects the existence of a substantial self, and Malebranche endorses it, commentators have been very struck by the similarity of the two thinkers on the self.Footnote ⁶³ Thus Andrew Pyle remarks that Malebranche ‘arrives … at a position, as regards our knowledge of our own souls, not far removed from the scepticism of Hume’.Footnote ⁶⁴ I shall show that the kind of reasoning Hume exploits in rejecting the existence of a substantial self is, like the case of necessary connection, very similar to that employed by Malebranche to establish a different thesis, namely that there is an unknowable self. Again the difference in conclusions here, though fostered by similar kinds of considerations, depends crucially on the rejection of the intellect. Briefly, Malebranche's argument that the soul is an unknown substance depends on (i) an intellectual grasp of the nature of material substance as extension and (ii) an argument about conceivability and independent existence. Hume rejects (i) and exploits (ii) to show that the self is not a substance.

We begin by examining Malebranche's reasons for thinking that there is a substantial self, of which we have no idea, and why it is that we have no such idea. The claim that we have no idea of a substantial self relates to an asymmetry between our epistemic position with respect to the self and with respect to the body. Recall that intellectual representations yield a certain kind of a priori modal knowledge. They ‘represent things to the mind in a way so clear that we can discover by simple perception whether such and such modifications belong to them’.Footnote ⁶⁵ We possess a grasp of body as pure extension provided by the intellect. This consists in a grasp of extension as essence, allowing us to determine a priori all the modes of which matter is capable. This knowledge is expressed in the science of geometry, which offers an exhaustive account of the properties of matter. Now, this kind of understanding is simply not available with respect to the mind, and it is in this sense that we have no idea of it. We have ‘no idea of our mind which is such that, by consulting it, we can discover the modifications of which the mind is capable’.Footnote ⁶⁶ Any knowledge we have is of particular modes or sensations and is a posteriori: we only come to know those modes by experience. The soul ‘knows itself capable of a given sensation not through the perception it has of itself in consulting its ideas but only through experience, whereas it knows that extension is capable of an infinite number of figures through the idea it has of extension’.Footnote ⁶⁷ Any knowledge of self is limited to an ‘inner sensation’ or sentiment intérieur which (somehow) intimates that there is a self, but not what that self is.Footnote ⁶⁸

One sense in which knowledge of the soul is not a priori is that we cannot understand its modes unless that mode is actualized: we cannot, that is, read off the taste of a pineapple from a prior grasp of the essence of the soul. A different sense in which our knowledge is not a priori is that one cannot tell by introspection whether any particular mode is a mode of the soul or a mode of body. One cannot tell whether pain is a mode of body or soul, or whether perceived colour is a mode of external bodies or merely an aspect of subjective experience. This feature of our knowledge of the mind figures in Malebranche's attempt to establish that there is some (a priori unknowable) substantial self. The argument goes like this. Anything is either a mode or a substance. We can determine what is a mode of extension by considering ‘what is inseparable [inséparable] from it’, and what we ‘we conceive of as inseparable’Footnote ⁶⁹ is extension or ‘relations of distance’. Since colours, pains etc., are evidently not substances, and not modes of extension, they are thereby modes of a non-extended, immaterial substance. Thus Malebranche's spokesman in the first of the Dialogues on Metaphysics and Religion says ‘my own substance, is not [therefore] a body, since my perceptions, which surely belong to me, are something entirely different from relations of distance’.Footnote ⁷⁰

Turning now to Hume's account, the most evident difference between the two thinkers consists in the fact that Hume rejects any substantial self, famously holding that the self is nothing but a bundle of perceptions. According to Ralph Church,Footnote ⁷¹ Hume reaches this conclusion by pushing Malebranche's epistemological claim – that we have no introspective knowledge of a substantial self- to the ontological conclusion that there is no substantial self. This leads Church to say that Malebranche a ‘better empiricist than Hume’Footnote ⁷² just because Malebranche is agnostic where Hume calls for rejection. But this misconstrues both Hume's argument and its relation to Malebranche's views. We should note first that Malebranche holds that we grasp one substance – namely extension – whereas Hume does not. To have an idea of substance requires an impression, but we have no such impression. Thus in T 1.4.5 ‘Of the immateriality of the soul’,Footnote ⁷³ Hume writes that as every idea ‘is deriv'd from a precedent impression, had we any idea of … substance … , we must have an impression of it’.Footnote ⁷⁴ We have no idea of substance, material or immaterial, because we have no impression of it. This is clearly an application of the distinctly empiricist thesis that all ideas must relate suitably to sensory experience. Perhaps this is what motivates Church's claim that Hume pushes Malebranche's ignorance claim to Hume's rejection. But this misses Hume's more fundamental argument, which exploits conceivability and separability considerations similar to those Malebranche uses.

Consider the definition of substance as ‘that which may exist by itself’. To determine what satisfies that definition we should, with Malebranche, identify that which can be conceived as having independent existence through the separability of the relevant item from others. Hume applies this principle to perceptions. All perceptions are distinct, and can be separately conceived: they ‘may be consider'd as separately existent, and may exist separately, and have no need of any thing else to support their existence. They are, therefore, substances, as far as this definition explains a substance’.Footnote ⁷⁵ Since perceptions are supposed not themselves to be substances, but modes of the soul, the traditional route to determining substancehood delivers quite the wrong result. Thus either we have no idea of substance at all or, per impossible, perceptions are substances:

These considerations do the real work in Hume's rejection of a substantial self in T.1.4.6 ‘Of personal identity’. Hume does not reject a substantial self simply on the grounds that such a self is not the object of introspection. It is true that he says that introspection reveals only a ‘bundle or collection of different perceptions, which succeed each other with an inconceivable rapidity, and are in a perpetual flux and movement’,Footnote ⁷⁷ and I ‘never can catch myself’.Footnote ⁷⁸ But the real argument rests on conceivability and separability. Thus he writes:

Hume therefore uses modal considerations to reject a substantial self, an appeal to separability that Malebranche used to establish that the self is an immaterial substance. It is quite wrong then to think with McCracken that Hume's use of separability and distinction is among the ‘fundamental empiricist principles’ operative in the rejection of the self.Footnote ⁸⁰ Instead rationalists like Malebranche and Descartes embrace the connection between distinctness, separability and conceivability in the epistemology of substance. The difference between them and Hume rest on the rejection of the intellect, and hence a grasp of substancehood. For Hume, either we have no idea of substance (since we have no impression of it) or the use of separability to determine substancehood delivers the result that perceptions are substances, and hence cannot be related to some substance that is identical to the self.