Hume’s many problems with causation

Hume’s many problems with causation
David Hume denied the reality and knowability of causation due to numerous conceptual confusions and much fallacious thinking; and unfortunately, many philosophers have foolishly followed and continue to follow his lead to this day. There are two kinds of philosophers, the malevolent, destructive ones, who look for problems they hope unsolvable; and the benevolent, constructive ones, who look for solutions; Hume was definitely of the former variety. The following is an example[1] of Hume’s assertions concerning causation:
“We cannot discover any necessary connexion between the cause and the effect, even though we may observe their constant conjunction, in all past instances.”
Hume’s problems with causation are rooted in problems with modality. He did not sufficiently analyze the concepts of modality before making claims about causation. His main thesis that man cannot know necessity, and more deeply that there is no necessity, is self-contradictory. If something is impossible, then logically this implies that there is necessity for that thing not to exist. Moreover, it is logically impossible to claim the impossibility for man to know necessity, because that is itself a claim to the necessity of man’s ignorance about necessity. Impossibility and necessity are concepts that logically go hand in hand; impossibility is necessity of a negative and necessity is impossibility of a positive. One cannot accept one concept without the other: their forms do not differ, only their contents do.
Similarly, one cannot logically claim the negation of necessity without accepting the negation of impossibility, and vice versa. If everything is unnecessary, then nothing is necessary, in which case no necessity whatsoever can be claimed; likewise, if everything is possible, then nothing is impossible, in which case no impossibility whatsoever can be claimed. If nothing is necessary, then nothing is impossible; and if nothing is impossible, then nothing is necessary. These are not word games, but logical imperatives. Differences in polarity do not affect the concepts of modality. Hume ignored the square of opposition between the four concepts of necessity, impossibility (necessity-not), possibility (non-impossibility), and unnecessity (possibility-not). He thought he could pick and choose and accept one concept of modality and reject another. But no! – these four concepts are logically inextricably tied together; they cannot be taken up separately.
Furthermore, Hume confused different modes (or types) of modality, notably logical modality and natural modality. Logical necessity (the necessity of conceptual implications) is a broader concept than natural necessity (the necessity of natural law). Logical necessity logically implies natural necessity; but natural necessity does not imply logical necessity. Natural necessity is logically compatible with logical unnecessity. This means that, while we can indeed, using our imagination, conceive that natural laws (apparent natural necessities) could have been otherwise than they in fact are (i.e. that they are not logical necessities), that does not change the fact that they are natural necessities and we can know them by induction (instead of deduction), i.e. by generalization from empirical observations (until and unless further observations belie such generalization and impose particularization). Hume, through stupidity or dishonesty, conflated these two concepts of logical and natural necessity when he denied causation.
Hume was evidently also ignorant of the differences between the dynamic forms of causation based on ‘natural’ (and temporal) modality and the static forms of causation based on ‘extensional’ (and spatial) modality. These modes involve different forms of conditional propositions. In the natural mode (and likewise the lesser temporal mode), a dynamic event is followed or not-followed by another dynamic event, immediately or over time; here, causation refers to succession of events (changes) in time. The extensional mode (and likewise the lesser spatial mode), on the other hand, concerns the conjunction of static things (concrete entities or qualities or even actions, or abstract categories) irrespective of time. For examples: when a gas is heated, it expands, when cooled, it contracts (dynamic causation); whatever has legs, can walk, whatever hasn’t, can’t (static causation).[2]
Additionally, Hume vaguely claimed that causation refers to ‘constant conjunction’; but this is formally inaccurate, it is only half the story. Knowing that something A is always followed by or conjoined with another thing B (and therefore is never followed or conjoined with the negation of B) does not suffice to prove that B depends on A; for it is possible that the negation of A is also always followed by or conjoined with B, in which case B would surely be independent of A. Hume’s definition of causation as ‘constant conjunction’ refers to the positive relation of A and B, but fails to mention the negative relation of not-A to B. To claim causation between two items, one must always specify both the positive side and the negative side of their relation. Only when both these sides of the question are settled can we claim that there is, or may be, causation. The positive side is not enough by itself to do that.
In truth, to establish the strongest form of causation we must demonstrate both that A is constantly conjoined with B and that not-A is constantly conjoined with not-B; this is complete-necessary causation. If we find that A is constantly conjoined with B, while not-A is not constantly conjoined with not-B, that is still causation, though of a weaker sort, namely complete-contingent causation. Moreover, contrary to Hume’s naïve definition, there can also be causation without A being constantly conjoined with B. If A is not constantly conjoined with B, but not-A is constantly conjoined with not-B, that is partial-necessary causation. And if A is not constantly conjoined with B and not-A is not constantly conjoined with not-B, that is consistent with partial-contingent causation, the weakest sort of causation.
Thus, irregularity, whether on the positive or the negative side, is not per se proof of non-causation, because partial and contingent causation are logical possibilities. If a thing is a partial causative of another, they will not always be conjoined; and if a thing is a contingent causative of another, their negations will not always be conjoined. Complete (or sufficient) and necessary (or sine qua non) causation is only the strongest determination of causation (whence its paradigmatic role) – it is not the only possible form. In other words, causation may itself be conditional (whether on the positive or negative side); it does not only exist in unconditional form[3]. Hume did not take these crucial formal facts into consideration, but offered a very simplistic definition of causation, No wonder, then, that causation seemed to him something feeble.
Hume denied causation by referring to its lacking some vague, undefined, metaphysical “connexion” (suggesting a bond or cement or glue) between the putative cause and effect. His definition of causation as constant conjunction being inaccurate, he could well deny that this bond existed in the midst of what we call causation. But the truth is, when causation is properly defined with reference both to positive and negative regularities, as above pointed out, so that there is no way out of it with reference to the invisibility of a presumed natural glue between the terms – the problem that Hume imagined simply disappears. If both positive and negative sides are accounted for, no distinction can be drawn between regularity with glue and regularity without glue! Note this well. What is indeed regular (to any degree) is glued (to that degree); and only what is truly irregular truly lacks glue.
Causation is a complex abstract concept, not a simple concrete percept; so, to deny causation because one does not directly perceive it is as such absurd. The truth is that causation is so complex a relation, with four alternative full determinations possible, that it cannot just be denied point-blank, as Hume tried, without awareness of all its possibilities and thence exhaustive denial of them all, and that needs to be done for each and every pair of items in the whole universe. Before causation can be denied to exist, it must be shown empirically, for any pair of items, that none of the four determinations of causation are applicable to them. This is no easy task, because even if no regularity is immediately spotted in their relation, they may still be related by partial-contingent causation at a deeper level. Hume denied the reality and knowability of causation because he was unaware of the variety of its determinations and of the great difficulties in establishing non-causation. It is much easier to establish causation than non-causation.
Hume’s weak notion of causation may be construed post factum as being due to his mind’s tendency to resort to ‘material’ (Philonian) conditioning rather than ‘strict’ conditioning. This is evident in his speaking of causation in terms of ‘constant conjunction’ instead of ‘necessary conjunction’. In material implication, ‘A implies B’ means only that the conjunction of A and not-B is not actual; whereas in strict implication it means that such conjunction is not possible. In the former case, the relation is tenuous, mere happenstance at a given time; in the latter, the relation is utterly firm, one of actuality in all circumstances and throughout time and space, i.e. one of necessity. Behind this error of appreciation by Hume was his failure to understand inductive reasoning; that is, to grasp our ability to infer general actuality (necessity) from current actuality (happenstance), by means of generalization subject to corrective particularization in the event that evidence to the contrary is encountered at some later time.
Furthermore, Hume denied our knowledge of causation by claiming that it could only be strictly established through a putative ‘uniformity principle’ (used as a grand major premise); and then arguing that such principle cannot be relied on in practice since it is itself, at best, a mere product of generalization. But this argument, as will now easily be demonstrated, is absurd. The alleged uniformity principle is a red herring, in that there is no way to formulate such a principle so that it predicts uniformity of nature in specific cases. All we can say about the uniformity of nature is a vague statement that uniformity exists in nature, but we cannot in advance say where or when it does so. In other words, there is no principle of uniformity, only a fact of uniformity. And this fact is axiomatic and self-evident, because to claim otherwise is self-contradictory since any such explicit claim is based on concepts and all concepts presuppose similarity (uniformity of appearance) between two or more things (and incidentally, since we live in a diverse world, their dissimilarity from other things).
If we try to adopt a Heraclitean premise that no two things are the same, anywhere, at any time, we are using concepts – namely, ‘no’, ‘two’, ‘things’, and so on – concepts which all depend for their comprehensibility and realism on there being things indeed similar enough to be included in the same classes and denoted by the same terms. A world of utter chaos, for which the dictum that ‘you cannot step into the same river twice’ is literally and universally true, is simply unthinkable. Any world containing two or more things is logically bound to involve some uniformity, if only the shared fact of existing in the same world. Heraclitus’ philosophy was sophistry, in that he did not apply his insight of multiplicity to his own discourse[4]. All we need to acknowledge, in order to form concepts, is that there are apparent similarities in characteristics between diverse things; we do not need to enumerate or prove these similarities a priori. Likewise, all we need to acknowledge to form causative propositions is that there are apparent similarities in behavior between diverse things; we do not need to enumerate or prove these similarities a priori. To repeat, this is not an appeal to a principle of uniformity, but only to a fact of uniformity.
After introducing the uniformity principle, Hume rightly rejected it as involving a circular argument (the principle being itself a product of generalization could not be used to logically justify generalization). But he did not realize that such a general principle was logically impossible to even formulate; i.e. that there is no way to state a priori how far uniformities can and must go for each class of things – we can only discover this a posteriori, by means of pedestrian ad hoc empirical and theoretical considerations. Also, for any class of things, we know for sure that there will at some point be an end to uniformity – but we cannot foretell where that point lies. Nevertheless, having rejected what he wrongly thought to be the theoretical basis of our knowledge of causation (and other concepts), namely generalizations from observations, Hume concluded that causation could not be proven and therefore was in reality non-existent. But since his major premise was wrong, namely that knowledge of causation depends on an alleged uniformity principle, it is not surprising that his conclusion was wrong.
The true basis of generalization is simply that when we observe (or less directly induce from observations) a set of things to have some (positive or negative) property or behavior, we are logically bound to assume that things of the same kind which we have not yet encountered will indeed be found to have the same (positive or negative) property or behavior, until and unless we happen to come across one or more things of that same kind which do not in fact have that property or behavior (in which event we would adjust our judgment by means of particularization). This is not arbitrary inference, but is firmly based on the fact that the property or behavior of the known cases is empirically based, whereas (for the time being, at least) any expectation of opposite property or behavior for the yet unknown cases is based on nothing but imagination[5]. We rightly prefer to base our knowledge on experience rather than on imagination. This is the true validation of generalization – it is the genuine verdict of intelligent reason and the very basis of induction.
Sure, we too can well, as Hume did, imagine, in a general way, that cases we have not observed (or induced) might turn out to be different from cases we have already observed, since such changes of polarity have indeed in the past been observed ex post facto in other specific contexts. But notice the absurdity of this alleged critique: it is itself an attempt at generalization, moving from past observations of changes in polarity (in other contexts) to prediction of future changes in polarity (in the present context)! Note this well: Hume denied generalization from same polarity to same polarity by implicitly arbitrarily admitting generalization from polarity change to polarity change. This was logical inconsistency on his part; and the fact that his attempted refutation of causation is self-contradictory proves irrefutably that causation cannot logically be denied to exist. The truth is that until and unless we actually do observe (or induce) a change of polarity in the present specific context, we cannot logically assume or claim that it will ever happen. Because observation is epistemologically more reliable than imagination, we are logically bound to prefer the former to the latter, and generalize accordingly, until if ever some new observation shifts the dividing line.[6]
Hume, having rejected the validity of generalization (due to his unawareness of the processes of rectification of overgeneralization by particularization, and due to his resort to sophistical rationalizations, among other things), sought to explain our belief in causation with reference to mere habit or custom. He sought to deny causation by claiming that what we perceive as ‘regularity’ is really mere association of ideas rather than constant conjunction of facts. He claimed our belief in causation is not based on reason but on imagination; so that there is no objective connection between cause and effect, but merely a subjective feeling of connection (that elusive ‘glue’). This pseudo-explanation of our belief in causation was also fallacious. We do not assume causation by introspective observation of our mental processes, but from awareness of the behavior of the external (or, where applicable, internal) phenomena that these mental processes are about. Looking inward is not our usual way; it involves a different effort from our more common outward orientation of awareness.
Hume distinguished between ‘impressions’ and ‘ideas’, the former term apparently including concrete mental phenomena and the latter more abstract ones. But the expression ‘association of ideas’ may reasonably be taken as intended to cover all mental phenomena, i.e. both the primary concrete data and the derived abstract data. That is, here the term ‘ideas’ has a broad connotation, including sensations, intuitions, percepts, and imaginations, and also concepts of these, propositions with the latter, and complex thoughts built up from those. Additionally, feelings, emotions, moods, attitudes, and even habits and behavior patterns, and more complex psychological elements, could here be considered as ‘ideas’. However, Hume, when speaking of association of ideas in the context of his theory of causation probably had in mind mainly the cognitive aspect of ideas, since in his view they replaced objective ‘facts’. Thus, for him, here, the ideas associated were principally memories of sensations, intuitions, perceptions, imaginations, concepts, and discourses. He seems to have imagined these mental entities (as he evidently saw them in his mind’s eye) as succeeding each other like billiard balls in motion.
In any case, it is naïve to view ‘ideas’ (here including ‘impressions’) as literally mental entities, they are rather the ideating subject’s more or less conscious intentions of various sorts – they have ‘intentionality’, they point to external and/or internal objects in various ways (with or without word labels). They do not succeed each other completely independently of the subject’s consciousness and volition, like mere billiard balls, but do so through much more complex and varied processes. The way Hume effectively pictured ideas was ultimately very simplistic and visual. He lacked the perspicacity necessary for accurate introspection. His observations of his own inner life were very superficial and approximate. He was satisfied with very brief incursions into his own mind, not investing enough effort in his meditations, not looking inward very carefully and deeply, and not persevering. And he evidently lacked the intelligence needed to properly evaluate what he did observe. He also lacked logical skill, and the self-criticism required for that, i.e. the ability to direct his critical eye on his own discourse.
In truth, observed facts can only incidentally be called ideas (mental objects), insofar as once observed they usually or often remain in the mind as memories (which are then used in further imagination and thought). Observed facts, be they external or internal, are phenomenologically given – in that sense, they are all necessarily objective and in no way subjective. Although objects of consciousness (and thereby new contents of the mind), they are facts, not fictions. One cannot claim that consciousness is necessarily distortive without putting one’s own claim in doubt and disqualifying it; that is, skepticism is logically self-contradictory and therefore self-defeating. Moreover, to refer to the constant conjunction (and thence causation) of ideas instead of facts is to willfully ignore that ideas are also, in themselves, facts. From which it logically follows that if constant conjunction of ideas is admitted as possible, then there is no reason to deny that constant conjunction (and thence causation, if the negative side is taken into consideration) of other facts is equally possible, whether these are facts to which ideas refer or facts that have not yet produced ideas in the mind.
To admit the one and deny the other would be baseless and arbitrary, because there is no formal difference between these two sorts of constant conjunction, only a difference of content. The propositions ‘X causes Y’ and ‘The thought of X causes the thought of Y’ are formally the same because they concern the same relation (‘causes’), even though they differ in content (the terms ‘X’ and ‘Y’, and the terms ‘the thought of X’ and ‘the thought of Y’, respectively); a partly symbolic term like ‘the thought of X’ is just an instance of the generic, fully symbolic term ‘X’. If generalizations of observations regarding subjective ideas are acceptable (as inductive truths), then the same form of reasoning can inevitably and undoubtedly be applied to objective observed facts.
Moreover, note well, the concept of ‘association of ideas’ is wider than that of causation; Hume stole this wider concept when he limited it to apparent causation[7]. For example, I may remember that time I sat having coffee with my friend, and then go off mentally thinking about someone we talked about among other things, and then from there go off thinking about that other person’s deeds or character or whatever. The ideas associated in this example would not be considered as causatively related; they would be viewed as accidentally succeeding each other, because they do not invariably follow each other[8]. We do induce by generalization from repetitive successions of observed facts (or, more broadly, using adduction, theory formulation and confirmation) that these facts are causatively related (until and unless exceptions are found). But this does not mean that all ideas which happen to succeed each other occasionally are causatively related; in truth, very few ideas, if any, are constantly conjoined[9]. Yet Hume relied on the absence of generality and fortuitousness in the latter case to throw doubt on the firmness of the causative link.
Furthermore, if we look more closely at the said example, we see that it consists of a random leap of attention from some detail of one topic to a related detail of another topic which is not substantially or causally related to the former. The two (or more) topics are very incidentally related, by means of a volitional shift of attention. It cannot be said that there is causation between the ‘ideas’ involved – rather, they involve some common detail which influences the subject to will a change in direction of his attention. The common detail may be viewed as a thread traversing both topics; but the association between them is an isolated event, not a case of constant conjunction. There are perhaps, said offhand, ‘ideas’ that are constantly conjoined (though much depends on what we admit as constituting an ‘idea’). But even if so, such causative links in the mental sphere are not to be confused with the notion of association of ideas. Perhaps the constant conjunction between the emotion of fear and that of hate, or the constant conjunction between the attitude and practice of careful scrutiny of information and the personality trait of perspicacity, would be fitting examples.
Finally, I would like to draw attention to a further self-contradiction in Hume’s skeptical thesis. By categorically denying the existence of volition, he was effectively claiming, a priori, the logical necessity of determinism and/or natural spontaneity, as there are no other conceptual possibilities besides these three. But what is logically necessary is necessarily naturally necessary (though not vice versa). Logical necessity logically implies (though is not implied by) natural necessity (i.e. physical necessity in the material realm and mental necessity in the realm of mind)! Yet Hume denies the knowability, and indeed the very existence, of natural necessity (i.e. natural law); and as for natural spontaneity (i.e. natural happenstance not subject to law), if this was his preference, this cannot logically be upheld as universally applicable, because that would make it a natural law (a generality which Hume denies). So, he was tacitly contradicting himself; and his skeptical thesis was (and is) untenable.
From all the above arguments, we see that Hume’s denial and critique of causation was utter nonsense, the product of a confused or malignant mind. The same, of course, can be said of his denial and critique of soul and volition (free will), and thence of ethics; I have done this elsewhere in considerable detail[10]. His ideas were not the product of a superior intellect, but the result of a poisonous mélange of lack of reflection, ignorance, stupidity, and bad will. It is amazing that there were, and still are, people so naïve and unintelligent as to be convinced by his absurd arguments.
 
 

[1]              An Enquiry Concerning Human Understanding (1748), Section IV, Part I.
[2]              See my work Future Logic (1990) for a full analysis of the different modes of conditioning.
[3]              For a more detailed analysis, see my earlier listing of the four determinations of causation, drawn from my work The Logic of Causation.
[4]              For the same reason, of course, Parmenides was a sophist, since he formulated a philosophy of absolute, universal, and exclusive monism by means of discourse involving multiple concepts. The concepts of similarity and dissimilarity are inextricably tied together.
[5]              That is, if the initial experience (or more broadly, induction) is positive, any expectation of a negative case is merely imaginary until and unless it is actually experienced (or induced). Inversely, if the initial experience (or induction) is negative, any expectation of a positive case is merely imaginary until and unless it is actually experienced (or induced).
[6]              See my work Future Logic for more detailed treatment of inductive reasoning, and further reflections in subsequent works of mine. If Hume had reflected more carefully on Francis Bacon’s “negative instance” he might have realized its significance and saved the world centuries of intellectual imbecility and perversion.
[7]              He also referred to association of ideas to explain resemblance and contiguity in time and place. But the refutation of the three claims (which are intertwined, anyway) is the same.
[8]              Note in passing that two or more people having a conversation may jointly engage in association of ideas. One says something, this brings to mind something else to another, which in turn reminds the first of something else entirely, and so forth. The same people having a conversation the next day, from the same starting point, would likely proceed quite differently (although, to be sure, some people tend to repeat themselves, as if they have one track minds).
[9]              Even if, for me, these days, a certain tune awakens in me fond memories (e.g. of a woman I once loved), it is very improbable that this tune will do that again at some future time, say at a time I happen to have other concerns or when I am no longer interested in the object of those memories. Furthermore, even if this sequence of events remains with me all my life, it does not follow that other people experience the exact same sequence (with the same tune and fond memories). Constant conjunction of ideas is a very rare phenomenon, assuming it ever at all occurs. Hume’s claim that such mental ‘habits’ are the source of our belief in causation is not based on any common experience, but is rather a figment of his undisciplined imagination.
[10]            In my book Hume’s Problems With Induction (2008). Regarding the no-soul claim, see my essay “Critique of the Buddhist Five Skandhas Doctrine.”