1. Two Approaches


Recently there has been a pronounced shift in the interpretation of Hume on causation. The previous weight of opinion took him to be a Posit­ivist, but the new view is that he is a Sceptical Realist.[1] I hold no brief for the Positivist view. But I believe it needs replacing by something slightly dif­ferent, and that at best it shows an error of taste to make Sceptical Realism a fun­damental factor in the inter­preta­tion of Hume.

            Let us call any concept of one event producing another, or being neces­sarily a cause or consequence of another, and which involves something in the events beyond their merely being kinds of events that regularly occur together, a “thick” concept of the dependence of one event on another. Then on the Positivist account, Hume believes that no thick notion is intel­ligible. On the Positivist view there is very little that we can ever under­stand and mean by a causal connexion between events. All we can understand and properly mean by talk of causation is that events fall into certain regular patterns, and the Positivist inter­pretation is that Hume offered this as a reductive definition of causa­tion. This is the famous regularity theory, summed up in the _philosoph­ical' defini­tion: _an object, followed by another, and where all the objects similar to the first are followed by objects similar to the second'.[2] The Sceptical Realist view denies that Hume offered any such reduction or analysis of the notion of causation. It takes seriously the many passages in which Hume appears to allow that we are talking of some thick notion of dependence of one event on another, going beyond regular succession. It takes it that Hume acknowledges that there is some such thick relation, even if it will be one about whose nature and extent we are doomed to ignorance. Hence, in John Wright's phrase, Sceptical Realism. 

            At first sight the difference between Positivism and Sceptical Realism is reasonably clear, and it is plausible that if these are the two options then Hume is better seen as tending towards the second. But, as proponents of the Sceptical Realist interpretation realize, there is one big problem, arising from Hume's theory of meaning. Sceptical realism seems to demand that we understand what it would be for one event to depend thickly upon another, even if we are ignorant of the nature of this relation; Hume seems to insist that we have no impres­sion, and hence no idea of any such depen­dence.

            The problem here is a problem for any interpretation, and can be focussed on a con­tradiction, to which Hume seems to be committed:

(1) We have no ideas except those that are preceded by suitably related impressions.

(2) There are no impressions that are suitably related to the idea of a thick necessary connexion between distinct events.

(3) We have an idea of a thick necessary connexion between distinct events.

            The _suitable relation' spoken of includes direct copying, in the case of simple ideas, and whatever is covered by _compounding' in the case of complex ideas that are compounded out of simple ones.

            The Positivist interpretation takes Hume to be claiming that when we talk of causation we only mean something that strips out the thick element of neces­sity, and  substitutes regular contiguous succession. So (3) is false. The difficulty is that Hume apparently denies this:

            Shall we then rest contented with these two relations of con­tiguity and succession, as affording a compleat idea of causation? By no means. An object may be contiguous and prior to another, without being consider'd as its cause. There is a NECESSARY CONNEXION to be taken into consideration; and that relation is of much greater importance, than any of the other two above-men­tioned. [3]


The central problem in interpreting Hume is coping with the contradiction. The Sceptical Realist strategy is to downplay the importance of the theory of understanding, so that even if Hume officially said (2), it played a negligible part in his view of causation.


2. A Doubtful Distinction

            How then does the Sceptical Realist deal with the problem of meaning? Edward Craig and Galen Strawson draw attention to a distinction that occurs in Hume's writings.[4]  When the theory of ideas threatens our idea of external existence or _body', it is said that Hume invokes a distinc­tion between what we can _suppose' and what we can _conceive', the idea being that we can coherently suppose that there are things of some sort (external objects) even when strictly we have no idea of what it is that we are supposing. Another way of putting it is that we can have a _relative' idea of things whose _specific' dif­ference from other

things we cannot com­prehend. We could say that we have no representa­tive idea of what we talk about, but a relative or relational idea, locating it by its role. We would talk of a _something-we-know-not-what' that does something or bears some relation to an aspect of the world of which we do have an idea. This distinction solves the con­tradiction by distinguishing between the terminol­ogy of (2) and (3). Hume thinks we have no representative idea of causation: we have no impres­sion of it, and in some important sense it remains incom­prehensible, and we cannot represent to ourselves what it is. What we do have however is a relational idea of it—it is whatever it is that issues in regular succes­sions of events, or upon which such patterns depend, or whatever forces such regular­ities. The negative side is given in (2), but the positive side in (3).

            The texts however give no direct support to this interpretation of Hume. While he does indeed use both a _relative' versus _specific' distinction and the pos­sibility of _suppos­ing' what we cannot _conceive', he uses them very sparingly indeed. In fact he never uses either, nor mentions either in connexion with causa­tion. He never uses or mentions either in the Enquiry or in the Dialogues in any context at all. This alone makes them an unlikely candidates for a central role in under­standing his mature philosophy [5]. But worse, there are warning signs to be noticed when they occur in the Treatise. There are four occurrences: on pp. 67 -68, p. 188 referring back to it, p. 218, and p. 241. In none of these cases is Hume actually con­trasting a specific versus a relative idea of any one property or relation, enjoining us that we can understand a property or object by its rela­tions even if we cannot understand it by some stricter standard derived from the theory of ideas. On the contrary, in each context it is the impossibility of conceiv­ing a _specific difference' between external objects and perceptions that is the focus of atten­tion. _Specif­ic' qualifies the proper­ties supposedly differen­tiating external objects from ideas, and of these specific qualities we know and understand nothing by any standard at all. Here are the two major passages with enough surrounding context to matter:

            'Now since nothing is ever present to the mind but perceptions, and since all ideas are deriv'd from something antecedently present to the mind; it follows, that _tis impossible for us so much as to conceive or form an idea of any thing specifically different from ideas and impressions. Let us fix our attention out of ourselves as much as possible: Let us chace our imagination to the heavens, or to the utmost limits of the universe; we never really advance a step beyond ourselves, nor can conceive any kind of existence, but those perceptions, which have appear'd in that narrow compass. This is the universe of the imagination, nor have we any idea but what is there produc'd.

            The farthest we can go towards a conception of external objects, when suppos'd specifically different from our perceptions, is to form a relative idea of them, without pretending to comprehend the related objects. Generally speaking we do not suppose them specifically different; but only attribute to them different relations, connexions and durations. But of this more fully hereafter. (p. 67-68)

            Philosophers deny our resembling perceptions to be identically the same, and uninterrupted; and yet have so great a propensity to believe them such, that they arbitrarily invent a new set of perceptions, to which they attribute these qualities. I say, a new set of perceptions: For we may well suppose in general, but _tis impossible for us distinctly to conceive, objects to be in their nature any thing but exactly the same with perceptions. What then can we look for from this confusion of groundless and extraordi­nary opinions but error and falsehood? And how can we justify to ourselves any belief we repose in them? (p. 218)

            It requires some daring to take these passages as a model for sceptical realism. Hume is far—about as far as can be—from saying that we actually possess a going idea of the external world, which allows us to understand, by some weak standard, what the externality is that we do not know about. Each of the two passages gives the strongest contrary impres­sion. The first affirms idealism ('... we never really advance a step beyond ourselves'). The second intro­duces the _supposes versus conceives' distinc­tion only while he simultaneously dismisses its effect out of hand. Its dismissal justifies Hume in describing his Philosophers (the culture whose spokesman is Locke) as actually inventing new perceptions, rather than inventing new things different from percep­tions. This is the very opposite of the view a Scepti­cal Realist Hume should take. He should admit that a Lockean succeeds in introducing a (relative) notion of an external object as something that has various relations to our perceptions, and then go on to worry how much we know about such objects. Hume does not do this: he simply dismisses the idea that we have a set of determinate, intelligible, proposi­tions about which, unfor­tunately, we shall never know the truth. It is not that we understand something, but cannot know whether it is true. It is that we give ourselves explanations which seemed to introduce an intelligible concept, but in fact fail (the demands put upon an external world independent of perception are simply inconsistent). We are in the domain of a _con­fusion of groundless and extraordinary opinion' where our only hope is to abandon reason altogether. So, even when it is used the _specific versus relative' distinction is not used as Craig and Strawson would have it used in the different area of causation, to which, as I have said, Hume never applies it.

3. Another Distinction


Before proceeding it is necessary to have in mind two things that might be asked of _thick' causation. When we think of a causally connected pair of events, such as the impact of the first billiard ball causing the motion of the second, we want there to be a further fact than (mere) succes­sion, or even mere regular succession of these kinds of event. We want there to be a dependency or connexion, a fact making it so that when the first happens the second must happen. Call this the desire for a causal nexus. But now suppose we shift our gaze to the whole ongoing course of nature. Again, we may want there to be a further fact than mere regular succession. We feel that the ongoing pattern would be too much of a coincidence unless there is something in virtue of which the world has had and is going to go on having the order that it does. We want there to be some secret spring or principle, some ultimate cause, “on which the regular course and succession of objects totally depends”.[6] This is whatever it is that ensures the continuation of the natural order, that dispels the inductive vertigo which arises when we think how natural it might be, how probable even, that the constrained and delicate pattern of events might fall apart. Call the desire for this further fact the desire for a straightjacket on the possible course of nature: something whose existence at one time guarantees constancies at any later time.[7]

            A fact alleviating this vertigo has to be a very peculiar fact, for the following reason. It has to be something whose own continued efficacy through time is subject to no possibility of change or chance of failure. For otherwise the fact that it keeps on as it does would itself be a case of coin­cidence or fluke, another contingency crying out for explanation and engendering inductive vertigo. Some think they can point us towards a fact with this potency. Some draw comfort from God's sustaining will (as if anything understood on the analogy of our own mental states could be timeproof!). David Armstrong believes that a kind of necessary, time­less, gridlock of universals will do.[8] Galen Strawson takes comfort in fun­damental forces constitu­tive of the nature of matter [9].

            It is easy to conflate the desire for a nexus, case by case, with the desire for a straightjacket. But Hume (sometimes—but see below) is clear that they are different. They are different because whatever the nexus between two events is at one time, it is the kind of thing that can in principle change, so that at a different time events of the same kind may bear a different connexion. Thus suppose we grant ourselves the right to think in terms of a thick connexion between one event and another: a power or force whereby an event of the first kind brings about an event of the second. Neverthe­less there is no contradic­tion in supposing that the powers and forces with which events are endowed at one time cease at another, nor in supposing that any secret nature of bodies upon which those powers and forces depend itself changes, bringing their change in its wake. Hume emphasizes this point in both the Enquiry and the Treatise.[10] It is his reason for denying that the problem of induction can be solved by appeal to the powers and forces of bodies. But it is equally a reason for separating the question of a nexus from that of a straightjacket. Nexuses by themsel­ves do not provide a straightjacket. The ongoing regularity and constancy even of a thick nexus between one kind of event and another is just as much a brute contin­gent regularity as the bare regular concatenation of events.[11] In each case we have something that can engender the inductive vertigo, or whose con­tinuation through time might be thought to demand some kind of “ground” or ultimate cause or straightjacket.

            The difference between a nexus, holding on some particular occasions, and a straightjacket guaranteeing the continuation of a pattern of con­nexions is easy to overlook. This is because of a lurking epistemological difficulty. Suppose one thinks that a particular nexus can be known for what it is, for instance by some observation whose content is more than the mere succession of events. One might report this by claiming to have seen that the one event had to happen, given the other. But if you see a _must' in one pair of events, would you not thereby see that it will hold for every pair of some kind that the original pair enables you to identify? How could you see it without seeing something with general implications, and ones that are immune to temporal change? In other words, you will take yourself to have seen a timeproof connexion: one that straightjackets how things could ever fall out. To put it the other way round, if things were not to fall out as expected, the original claim to have seen that the one event had to follow the other is refuted. This in turn makes it hard to see how a particular nexus could be an object of observation. Observation extends only to limited periods of space and time: how could we have within our view something that essentially casts its net over the whole of space and time?

            This problem probably explains one puzzling feature of Hume's proced­ure. He repeatedly affirms that someone who has a full apprehension of a thick causal connexion would be in a position to make an a priori claim about the way events will fall out and what kind of event will be caused by another. He argues that because we cannot have this time-proof knowledge we do not ap­prehend the causal connexion, for instance in the exercise of our own will.[12] The argument seems initially to be, as Craig describes it, a muddle, since there is no evident reason why someone apprehending a nexus on one occasion should thereby know that the same nexus will obtain on another—the very point Hume himself emphasizes when arguing that powers and forces will not solve the problem of induction.[13] I suspect that Hume sees that nothing would really count as apprehension of a particular _must' unless it carried with it implications of uniformity for the general case. It is to be(per impossibile) a particular apprehension, but one with the consequences of ap­prehending a straightjacket. Someone ap­prehending a straightjacket for what it is will as a conse­quence know its immunity to time and chance: he will know the timeless must that it guaran­tees. He will be apprehending the impos­sibility that events should ever transpire other­wise. He has therefore a piece of knowledge that, although it took an empirical starting point in the apprehension of an individual thick necessary con­nexion, can be seen a priori to have implications for all other places and times. And it is this that Hume treats as his target, even when the issue ought to be the ap­parently lesser one of the particular nexus.[14]

            There may be some room for man­oeuvre over the lesser claim to have apprehended a particular, but not necessarily timeproof, thick connexion. One might try allowing the par­ticular apprehension not to carry any implica­tions for what might be present on other occasions.[15] The difficulty will be that an apprehension of a mutable thick connexion does not give us quite what we want from knowledge of causation. That knowledge has to have a conse­quence: the subject possessing it must be prepared to foretell the one kind of event on the appearance of the other. It is not at all clear how apprehen­sion of a particular relation obtaining at a particular place and time could automatically carry any such consequence: one might, as it were, say that this is how events are con­nected today, and form no expectation, and not know what to expect to happen tomorrow.

            Sceptical realism might characterize Hume's position on either the nexus or the straightjacket. But unless we understand the extraordinary demands on a straightjacket we shall fail to see that realism concerning it is hardly important compared to his scepticism. Thus when Strawson opposes the Regularity Theory, with its ongoing flukes, by citing _fundamental forces' essentially constitutive of _the nature of matter', and invokes these to soothe away inductive vertigo, he is surely forgetting Hume's point.[16] Even if forces are taken _to latch on to real, mind independent, observable-regularity-transcendent facts about reality'[17] they need someth­ing further in order to serve as a straightjacket. They need neces­sary immunity to change; they need to be things for which the inductive vertigo does not arise. Equally if the _nature of matter' is to help, then the continuation of matter must not be just one more contingency, whose falling out the same way instant after instant, time after time, is a cosmic fluke. The force that through the green fuse drives the flower might falter, and so might the fuse and the flower, but a straightjacket must not. Its immunity to change must be necessary, for if it is contingent then either it is a fluke that of any changes that might occur, none ever does, or else this regularity is itself not brute but demands some further straightjacket in the background, of which we have even less inkling. The point is that we will not locate it by ordinary talk of _force' and its cognates. For even if Hume can countenance understanding of a thick nexus the theoretical demands on a straightjacket are a great deal more demanding. [18]

            Hume's main interest in causation is to destroy the idea that we could ever apprehend a straightjacketing fact: we have no conception of it, nor any conception of what it would be to have such a conception nor any conception of how we might approach such a conception. In par­ticular we must not think of the advance of science as targeted on finding such a thing. The lesson drawn from Newton is that just as Principia gives us the operation of gravitation­al force, but does not “tell us what it is”, so any conceivable advance in science can only do more of the same. It can put events into wider and more interesting and exception-free patterns, and that is all. _The most perfect philosophy of the natural kind only staves off our ignorance a little longer.' [19]

             Would it be easy for Hume to allow us a _relative' idea of a straightj­acketing fact—a _something we know not what' that governs/brings about­/ex­plains the continu­ing order of nature'? We understand this only insofar as we under­stand the relation involved of governing or bringing about. But can we understand the relation? Can Hume say the relation part of the relational idea is intel­ligible? The question is whether we know what governing or bringing about would be when we have no example, and indeed no conception of the kind of fact alleged to be doing it. Hume, given his endorse­ment of Berkeley's theory of ideas, must say that we cannot take rela­tional ideas (governing, forcing, grounding, issuing in, bringing about) out of the context within which they have intel­ligible applica­tion, and apply them without blush in contexts in which they do not [20]. We can only generate the general idea if we have particular ex­amples. Otherwise com­prehen­sion fails.

            Nevertheless, it will be said, even if this shows that we have no idea at all of what would count either as a straightjacket, or as knowledge that some kind of fact provides one, it seems plain that Hume allows that there is one, even while insisting on scepticism about its nature. Sceptical Realists might be right that he allows us a _relative' idea of such a fact, silently betraying the Berkeleyan background. Even if this were technically correct—and we have seen how far it stretches the texts—it would still misplace the stress; this is why I originally described it as an error of taste rather than an outright mistake. The point is that Hume is utterly con­temptuous of any kind of theorizing conducted in terms of such a thing. We are at the point where anything we say _will be of little conse­quence to the world', or in the world of _notion(s) so imperfect that no sceptic will think it worth­while to contend against (them)' [21], [22]. His attitude must be the same as that he holds to an equally noumenal substratum, supporting the qualities of matter:

            But these philosophers carry their fictions still farther in their senti­ments concerning occult qualities, and both suppose a substance supporting, which they do not understand, and an accident supported, of which they have as imperfect an idea. The whole system, therefore, is entirely incomprehensible...(Treatise. p. 222)

He is here directly echoing Berkeley:

            ...Lastly, where there is not so much as the most inadequate or faint idea pretended to: I will not indeed thence conclude against the reality of any notion or existence of any thing: but my in­ference shall be, that you mean nothing at all: that you imply words to no manner of purpose, without any design or signification whatsoever. And I leave it to you to consider how mere jargon should be treated. [23]

Craig, especially makes the case that there is importance in the positive claim that something-we-know-not-what exists, and the importance is scepti­cal: it enables Hume to destroy any pretension to finding what we might antecedently have hoped to understand about nature. I agree entirely that this critical aim is essential to Hume, and at least as important as the theory of understanding itself. But Hume enjoys this realignment without himself making any positive claim about the existence of any mysterious, straighj­acke­ting fact or facts. The realignment of our self-image, our philosophy of what real discovery and understanding might be, is independent of any such assertion. We do not ourselves have to think the other side of the line to learn how tightly the line defining the limit of all possible empirical enquiry is drawn. The point is that our real engagement with the world, in our under­stand­ing and our science, and our self-image or philosop­hical under­standing of the notions we actually use must sail on in complete indif­ference to any facts transcending our ideas. _Relative' ideas of such facts play no role any more than relative ideas of many things: Car­tesian Egos (simple, in­divisible entities whose per­manence ensures the identity of the self); the substratum in which proper­ties inhere; objective goods commanding the will of all those who apprehend them and so on. Since the actual business of making judgements about the identity of the self, or the possession of proper­ties by things, or what is good or bad, goes on in complete indifference to these things, they play no role in our real understand­ing.[24] They have no use at all: nothing will do just as well as something about which nothing can be said.

4. The Nexus

Perhaps the same is not true of individual thick connexions, that is, the par­ticular causal nexus obtaining between specific events at a time. Don't we give every employ­ment to such a notion? And if Sceptical Realists are right that Hume is not giving us a positivist reduction, do they remain in possession of the field here at least? I do not think so, for there is a third option: a truer description of Hume on ordinary empirical causation would be that he is neither a Positivist nor a Sceptical Realist, but rather a not-so-sceptical Anti-Realist.[25] That is, he gives us a story explaining and even justifying our use of the vocabulary of causation, while denying that we represent a real aspect of the world to ourselves as we use it.

            The outline of Hume's positive theory of causation is well known. The mind's perceptions, which form the material with which it must work, reveal only a regular succession of events. However, upon experience of such a regular succession the mind changes. It does not change by forming an impression or idea of any external property invisible in one instance alone. It changes functional­ly: it becomes organized so that the impression of the antecedent event gives rise to the idea of the subsequent event. No new aspect of the world is revealed by this change: it is strictly non-representative, just like the onset of a passion, with which Hume frequently compares it.[26] But once it takes place we think of the events as thickly connected; we become confident of the association, we talk of causation, and of course we act and plan in the light of that confidence.

            There are two separate components in this story: the contribution of the world to our apprehension, and the functional change in the mind itself.[27] These are the two aspects separated in the famous two _definitions' of cause: 

            An object, followed by another, and where all the objects similar to the first are followed by objects similar to the second.

            An object followed by another, and whose appearance always conveys the thought to the other.[28]

The first _philosophical definition' describes the contribution of the world, insofar as we can apprehend it, and the second _natural' definition describes the non-representative, functional dif­ference in the mind that apprehends the regularity. The parallel with Hume's philosophy of ethics is so far com­plete: again, there is a neutral starting point in the mind's apprehension of some non-ethical facts, and then the onset of non-representative passions ready to be voiced in our moralizing. 


            It is only after this point that complexities start, but unfortunately Hume gives less help with them than one would wish. The theory so far tells us of a non-representative change, a change in the structure of our expecta­tions, that gets expression when we deem two events to be causally con­nected. But it has not yet conjured up a full theory of the content of propositions about cause. It does not tell us, for example, what we are bothered about when we wonder if A caused B, what we are saying when we say that every event has a cause, or whether we can sensibly talk of unknown causes. We need more detail about the way in which cause becomes objectified so as to be spoken of as a feature of the real world, if its origin is in a feature of our own minds. Hume shows little interest in such questions, and indeed against the background of the theory of ideas, he can only point in misleading directions. He says, for example, that by a necessary connexion we _mean' a con­nexion in the mind, leaving himself open to interpretation as a kind of Berkeleyan, taking the idea of necessity to be a representation of some thick connexion we are aware of in our own minds. He then has to spend Part 1 of Section VII of the Enquiry averting this misunderstanding.[29] In his theory of morality he similarly seems unclear whether he is saying that virtue and vice are _nothing in the objects', but only sentiments in us, or that they are the qualities of objects that tend to arouse those sentiments.[30] What he lacks is a link between the real functional dif­ference, and the thick content we contrive to give causal judgements: the way we talk and think in terms of a projected property of things.[31] A telling point here is that in both the Treatise and the Enquiry he produces the _two definitions' only at the end of the discussion, and in each place he does so apologetically, in effect telling us that they are not to be regarded as strict definitions. On the view I am recommending this is right: they separate the two different aspects of the matter -- the contribution of the world, and the change in us. But they do not give us a lexicographer's analysis, and we should not expect one. There is no way of moralizing without using a moral vocabulary, and no way of causalizing without using the vocabulary of cause, efficacy, or power.

            Notice, however, how many cards Hume holds in his hands. The basic theory is flexible enough to accommodate many points that are usually raised against him. Our reactions to nature are subtle: not all regularities betoken cause, and sometimes we attribute cause after miniscule experience of regularity.[32] Well and good: the basic theory need put no limits on the input to our causalizing, any more than his theory that in moralizing we voice a passion puts a limit on the input to our moralizing. On the output side, the change in the structure of our thought after we have deemed a sequence to be causal may also be complex. Its heart is that we _make no longer any scruple of foretelling' one event upon the appearance of another. But there may be other changes. We may become willing, for example, to hold the sequence constant as we think about what would have happened if something else had happened, or what would happen if something else were to happen. Once we view a sequence as causal, it is held fixed as we conduct counterfactual and conditional deliberations. Well and good: the basic theory puts no limits on such consequences either. The theory also happily predicts the _intui­tions' that lead people to detest the Positivistic _regularity' theory of the content of our causal sayings. Someone talking of cause is voicing a distinct mental set: he is by no means in the same state as someone merely describing regular sequences, any more than someone who appreciates some natural feature as good is in the same state of mind as someone who merely appreciates the feature. The difference in this case is in the sentiment or passion that the feature arouses, and in the causal case in the fixity that the sequence of events takes in our thinking. Finally, the contradiction I identified at the beginning of this paper is sides­tepped by distin­guish­ing a represe­ntative idea of a con­nexion, which we do not have, from a capacity to make legitimate use of term whose function is given non-representatively, which we can have.

            There are, I believe, only two ways in which this kind of theory could be opposed. One is to deny that a Humean could forge the missing links, between the functional difference we are expressing, and the surface content of our causal judgements. The other is to deny that we have here a distinctive position, by assailing the limits on _representation' under which Hume operates. The first attack presses the point that Hume needs to tell us what happens not just when we think that A causes B, but also when we think that there exist unknown causal connexions, that regardless of whether we had ever existed there would still have been causal connexions and so on. We think in terms of causation as an element of the external world, and there remains a real question of how much of this thought Hume can explain, and how much he has to regret. However, his prospects for deflecting this first criticism must be quite bright. For as we have seen Hume is working with exactly the same ingredients in the case of ethics. Here too there is the task of explaining the apparently objective content of moral judgements given their source in the passions, but here it is much harder to believe that the problem is insoluble, and Hume certainly did not believe it to be so.

            The second attack need not deny Hume his ingredients. It simply claims that we can cook with them in a different way, awarding ourselves the right to a genuinely representative concept of causation. For when should we say that we have a representa­tive idea of a property or relation? One answer would be: when we can picture it holding, or exhibit to ourselves in imagination a scene in which the property or relation is visibly instanced. This is a natural empiricist answer, and the one that leaves Hume poised to argue that we have no representative idea of thick causal con­nexions. For a view (or succes­sion of other experiences: sounds, felt pressures, and so on) in which there is given a certain succession of events, and in which one event causes another, need be no different from a view in which it does not, but in which the same succession happens anyhow. This is why we have to interpret sequences as causal, and however automatic this act is, it is still one that needs to be performed. But empiricism nowadays sounds like prejudice: why should we not have a theoretical concept of a thick causal connexion, allowing both that there is a step from the raw appearance of a scene to the belief that it instances such and such con­nexions, but also insisting that we know what it is for such connexions to exist? We have a theoretical idea of them, and the idea represents the way the world is when they are present.

             The real problem with this is that it only works if we also understand the relation between the thick connexions and the ongoing pattern of events. Thick connexions make events happen; they guarantee outcomes, they issue in patterns of events, and so on. But these are terms of dependency or causation, so we understand the theory only if we understand them. And this understanding in turn is queried by the problems described in above: any Realist theory needs to tell us how the _musts' present on one occasion throw their writ over others. Otherwise it fails to give us what we want from a causal understanding of the world. For all the story goes, someone might be a virtuoso at detecting particular thick connexions, yet have no idea what to expect or how to conduct counterfactual and conditional reasoning.

            The net result is that any such realist theory looks extravagant. It asks from us more than we need. To see this, imagine a character we might call the Bare Humean. The Bare Humean misses out this capacity for apprehen­sion or theory, so does indeed lack the representative idea of thick connexions that these are supposed to give us. But she goes through the functional change which Hume describes, and conducts her expectations and actions accordingly. She can be an enthusias­tic natural scientist, finding concealed features and concealed patterns in nature to aid prediction. She can understand that finding ever more simplic­ity and ever more general patterns may be _set us as a task', so that there will always be more to know about nature. She will need a vocabulary to express her confidences and her doubts, and to communi­cate them to others; she will be a virtuoso at the salient features that are usable day by day to control her world. What else does she need? Are we sure she is missing anything at all—isn't she a bit like you and me?

Notes to Hume and Thick Connexions (5)


[1]. I have in mind Edward Craig, The Mind of God and the Works of Man (Oxford, Clarendon Press, 1987), Galen Strawson The Secret Connexion (Oxford, Clarendon Press, 1989) and John Wright, The Sceptical Realism of David Hume (Manchester, Manchester University Press, 1983).

[2]. Enquiry Concerning Human Understanding, ed. Selby Bigge, Section VII pt II, p. 76

[3]. Treatise of Human Nature, ed. Selby Bigge (Oxford, Oxford University Press, 1888) p. 77.

[4]. Edward Craig, ibid. p. 124. Galen Strawson, ibid. Ch. 12.

[5]. It is particularly odd that Strawson relies upon them, since he conceives of the Enquiry as embodying Hume's official theory of causation.

[6]. Enquiry, p. 55.

[7]. At least. It may be that its existence at one time should entail its existence at any previous time as well. But one way of gesturing at what is wanted is to imagine God creating it by some kind of fiat or act of law-giving, whose writ would run only into the future.

[8]. What is a Law of Nature, (Cambridge, Cambridge University Press, 1983) p. 88 ff.

[9]. E.g., p. 91,  pp. 254—255.

[10]. Treatise, pp. 90—91; Enquiry p. 37

[11]. One might seek to avoid this by the verbal manoeuvre of identifying kinds of events by their causal powers, in which case it will follow that events of the same kind will bear the same causal connexions. But as Hume in effect points out, inductive vertigo then transfers itself to the contingent question  of whether future events with the same sensory ap­pearance will turn out to be of the same kind.

[12]. Enquiry, Section 7, passim.

[13]. Treatise, p. 91, Craig, p. 97

[14]. On these issues, see also Peter Milligan, 'Natural Necessity and Induction', Philosophy 61, 1986.

[15]. Anscombe, G.E.M. 'Causality and Determination', in Metaphysics and the Philosophy of Mind (Oxford, Blackwell, 1981).

[16]. p. 91.

[17]. Ibid

[18]. Strawson is probably betrayed into this conflation by using the one term 'Causation' (with a capital 'C') equally for a thick nexus and a thick straightjacket.

[19]. Enquiry, p.31. This is famously the point where Newton said 'hypotheses non fingo', and the point that left contemporary scientists such as Huygens and Leibniz, who had wanted to know what gravity was, and not merely how bodies moved under its influence, feeling badly let down. Newton was quite within his rights to want more scientific understand­ing of gravitational attraction, and Hume does not oppose the goal. But if Newton and his contemporaries wanted a different thing -- an understanding of the impos­sibility that events should ever fall out otherwise -- then Hume stands in his way. Hume does not magnify the difference between himself and Newton, but if Newton was aiming at this superlative piece of understanding, and thought that the methods of natural science might give it, then Hume is clearly opposed. He was the first to see that what Newton did was the only kind of thing that could ever be done.

[20]. Berkeley's rigour on this is apparent in his constant polemic against 'abstraction', and in such matters as his embargo on taking causal relations away from the domain of the will, given that it is this which is the basis of our understanding of them. More directly relevant is his insistence that if you try to introduce a "relative notion" of matter as whatever-it-is-which supports various properties, you mean nothing. As well as the passage quoted in the text, see Principles of Human Knowledge, Pt. 1, Section 80.

[21]. Enquiry, p. 155.

[22]. Treatise, p. 168. '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'.

[23]. Berkeley, Three Dialogues between Hylas and Philonous, Dialogue 2, Para 121.

[24]. Strawson is at pains to show that not all Hume's reference to straightjacketing facts are ironic, but I do not think he shows that they are not contemptuous.

[25]. Or, quasi-realist, in the sense of these essays.

[26]. Norman Kemp Smith, The Philosophy of David Hume, (London, MacMillan, 19242), Chapters I and II. These present convincing evidence that this comparison was the prime mover of Hume's theory of causation. It opened up the 'New Scene of Thought' of which he speaks in the 1734 letter to (probably) George Cheyne.

[27]. As clear a statement as any is Hume's recapitulation, Enquiry p.78 - 79.

[28]. Enquiry pp. 76 - 7.

[29]. An interesting scholarly question, to which I do not know the answer, is why he took such elaborate care in the Enquiry, Section VII, to distinguish his theory from Berkeley's, when the Treatise contains no corresponding passages. It is one of the very few cases where the Enquiry is fuller than the Treatise. Did some review or correspondence make the need evident to him? 

[30]. His lists of virtues (e.g. Enquiry Concerning the Principles of Morals, Appendix I, p. 258) specify the properties of people such as benevolence, serenity, and so on that make us love them; his official position (e.g. Treatise, p. 471, p. 614) identifies the virtue with the sentiment itself.

[31]. Hume is quite prepared to allow that our common notion of cause contains defective elements - see the footnote on p. 77 of the Enquiry. But overall he is perfectly friendly to the way we think.

[32]. Hume discusses these complexities in Treatise I, iii, xv: 'Rules by Which to Judge of Causes and Effects'.