An Enquiry Concerning Human Understanding
first published as —
revised, & included in —
Hume's standard, and his challenge
An epigram of Alfred North Whitehead says that The safest general characterization of the European philosophical tradition is that it consists of a series of footnotes to Plato. In a similar spirit, it could fairly be said that twentieth century philosophy of science is an elaboration on the views of David Hume (1711-1776). Logical positivists ranging from the working physicist Ernst Mach to the academic philosopher A. J. Ayer to the young Ludwig Wittgenstein, who was almost a saint of logic, adhered to the standard that Hume set forth in the final sentences of his An Enquiry Concerning Human Understanding:
If we take in our hand any volume of divinity or school metaphysics, for instance; let us ask, Does it contain any abstract reasoning concerning quantity or number? No. Does it contain any experimental reasoning concerning matter of fact and existence? No. Commit it then to the flames. For it can contain nothing but sophistry and illusion.
And if positivism's verification principle has been widely criticized in more recent philosophy of science, in favor of other ways of addressing Hume's challenge to inductive reasoning — for example, Popper's hypothetico-deductivism — both the validity of that challenge, and Hume's larger rejection of metaphysics, remain alive and well in the culture of the sciences and of secularist thought.
Hume's Enquiry is of interest, and worth reading, not only as the fountainhead of much of twentieth-century thought, but also as a literary work, which sets forth its theses and questions more elegantly than is common in philosophical writing. Hume's views gain persuasiveness from the impression made by Hume's style of writing, which has the elegance of the best prose of his century without the over-elaborateness of its worst. He presents views that, in his terms, naturally belong to accurate and abstruse philosophy, in a manner that makes them seem, again in his terms, easy and obvious.
Nonetheless, his real concern is with abstruse matters. A full review of his book can start out by noting its literary elegance, but needs to give most of its attention to the structure of his arguments, and to the points at which their soundness might be questioned.
In his first chapter, Hume frames his positive goal as "delineation of the distinct parts and powers of the mind", and in his second, he goes at once to a key distinction: that between impressions and ideas. An impression, he says, is such a thing as the sensation of excessive heat, or of moderate warmth; an idea, or thought, is such a thing as the recollection or imagination of either. The two are distinguished, in his view, by their different degrees of liveliness: impressions have an intensity that ideas lack. And ideas, he adds, are not merely weaker than impressions, but are characteristically copied from them: Someone who has not seen a color, or tasted wine, or felt anger or love, cannot form an idea of them.
Plausible though this is, it does not seem to exhaust the differences between the two categories. Impressions are marked not only by intensity, but by uncontrollability: We can't make ourselves feel warm, or hot, or make ourselves taste wine, by choosing to do so, and while we may be able to work ourselves up to intense emotion — as Hamlet says (at 2.2.528-534),
... this player here,
— this can't be done by simply choosing to feel passion; it takes a kind of manipulation of our own thoughts and our own physical behavior. Impressions are, as the word implies, impressed on us. And sensory impressions, such as the feeling of heat or the taste of wine, are impressed on us by the physical world we inhabit. Perception is not simply an event internal to consciousness; it's perception of something. In Aristotelian language, it has intentionality (see for example the discussion by Douglas J. Den Uyl and Douglas Rasmussen in "Ayn Rand's Realism", p. 11 of Douglas J. Den Uyl and Douglas Rasmussen, The Philosophic Thought of Ayn Rand, Urbana and Chicago: University of Illinois Press, 1986). In focusing purely on the element of intensity, Hume is assuming a "consciousness-first" approach as opposed to a "reality-first" approach, less overtly but no less surely than his predecessor Berkeley; and that assumption is, at the very least, not self-evidently correct. Hume makes a debatable position appear natural and even inevitable by pointing out a difference that is real but nonessential, and passing over a real and essential one. The same assumption can be found in the next (very brief) chapter, where Hume discusses connections among ideas, and groups them under three principles, resemblance, continguity, and cause and effect: The identification of, in particular, cause and effect as a relation among ideas within the mind, and not among objects, states, or events in the physical world, is another position that at best is hardly self-evident.
Hume goes on to recommend that whenever any idea is "faint and obscure", the proper way to deal with it is to trace it back to the impressions from which it is supposed to be derived. This advice seems akin to the Aristotelian motto that "there is nothing in the intellect that was not first in the senses." But implicit in that useful advice is that there is nothing in the senses that was not first in the physical world. Hume's unshakable rock seems to be not the physical world, but the sensations themselves, as events within consciousness.
The next two chapters bring us to the heart of Hume's argument: His discussion of causality. At the outset, he makes a distinction between two subjects of human knowledge, the one that he refers to in his final sentences, quoted at the start of this review: relations of ideas and matters of fact. Relations of ideas, he says, are known to be true simply by the operation of thought, without any observation of things that actually exist; their truth is guaranteed by its denial being self-contradictory. In contrast, matters of fact are known by observing the world, and it is always possible to conceive, without self-contradiction, that the result of that observation might be otherwise. And reasoning about matters of fact always depends on the relation of cause and effect.
Now, in the first place, this is not easily reconciled with Hume's earlier statements about impressions and ideas. He asserts that abstract ideas "are naturally faint and obscure": but it is relations between such ideas that he holds up as being absolutely certain. And he recommends that if we want to make our ideas intelligible, we should trace them back to the impressions that gave rise to them: but those impressions inform us about matters of fact, of which, he claims, it is always possible for us to conceive the opposite. The more actual content we give to an idea, the less certainty we can claim for propositions involving it.
Hume puts forth, as an example of abstract ideas, the concepts and propositions of mathematics, such as the Pythagorean Theorem. These, he claims, can be known with absolute certainty; and they seem to have definite content. But, in the second place, in our actual experience of learning mathematics, we often encounter surprising conclusions — conclusions that we did not anticipate beforehand, and in some cases conclusions whose exact opposite seemed clearly right and perfectly intelligible. The mathematically naïve student, for example, might think that obviously there are fewer even numbers than there are whole numbers, and that the opposite is unintelligible; and might need hard work and careful instruction to convince them that there are exactly as many even numbers as there are numbers, or simply to make the claim understandable. It's also worth noting that though Hume thought that Euclid's conclusions could not be denied without self-contradiction, mathematicians such as Gauss and Lobachevski, not long after his time, showed that there were equally self-consistent geometries that reached different conclusions; and eventually Einstein made the case that which geometry applies to the real world is, in Hume's terms, a matter of fact.
Finally, when Hume says that it is not self-contradictory to suppose that a matter of fact might be otherwise, he is saying that when we put forth a proposition relating one idea (to use his terminology) with another, we can judge with full certainty that that proposition is true or false (a modern logician would use the technical term valid or invalid, but Hume does not); but when we relate an idea to an impression, or one impression to another, we cannot be so certain. But in his second section, he advised the reader that philosophical terms can easily be employed with no definite meaning, and the surest way to correct this is to seek the impressions from which the idea the term refers to are derived, in that it is not easy to fall into error or mistake with regard to impressions. It rather seems as if he's having things both ways, making ideas on one hand more certain, but on the other hand far less certain than sensory impressions or matters of fact!
Hume's central puzzle is the validity of reasoning on matters of fact. He traces such reason to cause and effect thinking; he argues that cause and effect thinking is founded on experience; but, he concludes, what is the rational basis for drawing conclusions from experience? Experience, perhaps, has shown that some object has always been accompanied by some effect; we infer from this that that object will always be accompanied by that effect. But this is to suppose that the future must resemble the past — and how is that supposition to be justified? All the futures we have seen so far may have resembled their pasts, but that does not show that the futures we have not yet seen will also resembled their pasts; to conclude otherwise is to assume that very resemblance, which is a circular argument.
Hume says that he would welcome a proof that this must be so by demonstrative argument; that is, a proof that it would be contradictory to suppose that the future will be different from the past. But, he says, no such proof has been presented. He adds that any such proof would have to be very simple, as ignorant people, and small children, and even beasts, are guided by experience, and thus by the assumption that the future will continue like the past; and any rational argument for this belief must be so simple that infants and beasts can grasp it. Therefore there is no such rational argument.
But what would such a rational argument accomplish? It would have to show that causality, or inductive reasoning, or the reliance on the past as a guide to what can be expected in the future (three principles that in Hume's account are equivalent) is demonstratively certain. But what Hume means by demonstrative certainty is that we cannot even think of the opposite claim: "whatever is intelligible, and can be distinctly conceived, implies no contradiction, and can never be proved false by any demonstrative argument or abstract reasoning a priori." That is, if causality were proven, it would mean that we could not form a consistent idea of its denial in our minds — though the inconsistency might be a subtle one, only identifiable by thinking through our ideas with great detail and specificity, as (Hume might have said) the impossibility of a triangle whose interior angles add up to more than 180° is not obvious to a child, but only to a geometer. But if we had such a proof, it would show that an acausal world was one we could not think of. Would it show that an acausal world was one that could not exist? Are the limits of what we can conceive the limits of what can exist? Given that Hume thinks a demonstrative argument is only about the relations of ideas within our own minds, a demonstrative argument for causality would seem still to be an argument about the idea of causality in our own minds; it still would not show that causality was a fact about the actual world. For that to be the case, we would have to suppose that logical contradiction was not only a limit on what we think, but a limit on what exists — that logic was not merely a rule of thought but a description of reality, and that in reasoning logically, we are conforming our minds to reality. This is something that would have made perfect sense to Aristotle or Aquinas, but as we have seen, the whole tendency of Hume's thought is against it. This is why he talks about "relations of ideas" in the first place.
Hume goes on, in the next chapter, to offer what he calls a solution to doubts about causality. That solution lies, in the first place, in saying that what causal reasoning is founded on is custom or habit; that is, when we have repeatedly seen fire heat things, or felt water quench our thirst, we come to rely on their doing so, without being able to give a reasoned argument as to why they should. In the second place, he says, what custom or habit gives rise to is a peculiar mental state that he calls belief. He defines this as "nothing but a more vivid, lively, forcible, firm, steady conception of an object, than what the imagination alone is ever able to attain." Again, Hume's focus is entirely on the internal mental state and not at all on the physical world! He takes as possible any combination of ideas that the mind can come up with, however contrary to experience — the head of a man on the body of a horse, or a billiard ball striking another and both coming to a stop — but distinguishes the possibilities in which we believe, because we have seen them repeatedly, from the ones we only imagine, because we have never seen them.
After a brief discussion of probability, Hume takes up the idea of "necessary connexion". This seems to be closer to what is usually thought of as causality, as can be seen from Hume's taking up after a few paragraphs the concept of power, or the ability to make things happen. This is normally taken to be an attribute of an object that enables it to act on another object in some way. Hume gives this idea two characteristic turns:
First, he takes it that if such a power existed, its outward existence should give rise to some inward impression of its presence. This is very much in accord with the conceptual realism of Plato and the rationalists' emphasis on intuitively known self-evident truths. He seems to be accepting the idea that truth must make itself manifest, and arguing that per contra, since the existence of power (over and above habitual conjunctions of events) does not make itself manifest, we have no basis for regarding power as true.
Second, he turns to the question of whether we may know of power by experiencing it in ourselves, through our actions. But his example of "power" is the human will, or volition; that is, he thinks we experience power by willing our hands to gesture, or our legs to walk, and seeing that our bodies do as we will. This seems needlessly abstract and needlessly mentalistic. In the ordinary course of events, we don't first will our legs to walk, and then walk; the two happen at once and inseparably. If the mental act is separated from the physical, it's precisely in those cases where the physical does not happen, or happens slowly and with difficulty — and those are experiences not of power but of impotence. The experience of power seems to occur more in the physical domain — for example, in lifting a heavy object, or pushing a stuck one, and feeling our muscles overcome a resisting force — and this is much more clearly an experience of the power of a physical object, the human body. Hume's exclusive focus on impressions, ideas, sentiments, and other entities within the human mind leads him not to look at the human body as an organ of knowledge.
Hume ends up by defining causation as either the regular succession of one state of affairs by another, or the expectation of the second in our minds when we perceive the first. He seems to grant that this expectation can be treated as a certainty, and relied on. But he is not prepared, it seems, to grant that either state of affairs is produced by an object whose properties give it the power to bring that state about — for example, that a mass has the power to attract another mass, as in Newtonian mechanics — because we do not immediately perceive such a power when we first see a particular succession of events, but only come to believe in it after many experiences of that succession. That is, it seems that he wants power to be known by perceiving it. His whole mode of thought takes as real only what is perceived, and treats abstract concepts as faint and faded versions of perception. But this seems to be another version of Berkeley's skepticism about matter! Hume once again is treating sensory impressions, or states of his own consciousness, as real and certain, but physical objects, or constituents of the physical universe, as doubtful speculations.
In the following chapter, Hume turns to "liberty and necessity", or the question of free will, one of the great classical puzzles of metaphysics — and seeks to show that there is no actual puzzle in it. His argument, in essence, is that there is not one opposition, but two: between necessity and chance, and between liberty and constraint. Necessity, he says, lies in the regular and predictable succession of events, which applies as much to human conduct as to inanimate nature; in a striking example, he points out that the prisoner tries to tunnel out through the walls or floor, rather than hoping to persuade his jailers to open the doors for him, regarding inanimate stone as less intractable than human beings. Not being subject to necessity would mean being governed by chance, or as we would now say, by randomness. But there is no such force as chance, in his view. (It would be interesting to see what Hume would have made of quantum mechanics. Would he join Einstein in saying that God does not play dice with the universe?) If we oppose liberty to necessity, we are compelled to treat liberty as nonexistent, or else abandon the very idea of an intelligible order of nature. Liberty, he says, is rightly opposed to constraint, as by the bonds of a prisoner; liberty is the power of acting or not acting according to one's own will, and constraint is anything that prevents such action. That is, he accepts the practical or political sense of liberty, but rejects the metaphysical as meaningless.
It's somewhat surprising that he takes this route in his argument, actually. He starts out by denying that there is any actual power that necessitates anything. Necessity in human events, in his view, must be either the regular succession of one human action by another, or the expectation of that succession when one observes the former action, and the belief that the latter action must follow it. But when we think about our own actions, we do not and indeed cannot believe that we must perform the later action after the earlier; the sense that there are real alternatives before us, and that we may choose one or the other, is inescapable. So Hume's own criterion of necessity fails when we try to apply it to ourselves. And there is, in Hume's system, no external reality or necessity as a reference point; in the last analysis, necessity exists because we believe in it. The lack of belief in necessity in our own case seems to stand as a conclusive argument against Hume's assertion that it applies to ourselves.
Hume might like to argue that the assumption of necessity, in dealing with inanimate nature, or with other human beings, is logically incompatible with the assumption of metaphysical liberty, in relation to ourselves; and that we can reject the simultaneous adherence to both beliefs as logically contradictory. But logical contradiction, as Hume has it, applies only to relations of ideas, and not to matters of fact; and in dealing with ourselves and our own actions we must be dealing with matters of fact. Hume's own view ought then to be that there can be no more contradiction in supposing that we are free to break the regular succession of events in our own actions, than in supposing that we are bound to maintain it. Indeed, we might perfectly well take it as a brute fact that we see necessity in the world all around us, but cannot apply it to ourselves. Hume's distinction between political and metaphysical liberty certainly is worth making, but his argument against metaphysical liberty seems to confuse his two categories of objects of human reason.
At the end of this chapter, Hume turns to the question of divine omnipotence, and of whether God must be counted as the cause of human sins; and this is the first of several theological discussions that occupy the next chapters, followed by his famous discussion of the grounds for belief in miracles, and for belief in divine providence and a better future state of existence. These discussions are made more complex and indirect by his need to avoid contradicting the theological beliefs that he is giving grounds to doubt, for fear of public outcry or perhaps even legal consequences. Hume claims not to be denying the Christian faith, but only to be showing that it is purely a matter of faith and that claims to ground it in reason are failures. But his rejection of a reasoned defense of Christianity is the substance of these chapters. The treatment is often ingenious — it's here, for example, that his earlier short discussion of probability comes to the foreground in his examination of miracles. However, it's something of a digression from his main argument.
He returns to his main argument in his final chapter, which explores various versions of the skeptical philosophy, which he recommends, not in its pure form, but in moderation, as a remedy for the excessive dogmatism of most human reasoning. He calls for us to draw only modest conclusions, staying close to matters of fact, and rejecting most theorizing. And this leads to his great peroration, with which I began this review.
But the heart of that peroration, the distinction between abstract logical systems and factual assertions, is more questionable than Hume supposes. Hume himself undermines it, with his discussion of ideas as fainter, less specific residues of impressions. But even that doesn't go far enough. An abstract idea, considered on its own, without the means of relating it to observations through the senses, really has no content at all, not even fantasy content, like the idea of a centaur. To use a fiscal metaphor (appropriately for Hume, one of the pioneers of economic analysis), relating an abstraction to concretes is like exchanging a bank note for goods or services; but relating an abstraction to another abstraction — which is all that Hume's "relations of ideas" can amount to — is like trading paper for paper, or accounting entries for accounting entries, in one vast cycle, with no way of knowing how much the value of the paper has been inflated, or whether any of it is worth anything. If logic and demonstration were only ways of keeping this exchange of irredeemable counters straight, the game wouldn't be worth playing. It certainly doesn't hold up as a standard of cognitive value that could be used to discount our actual knowledge of the physical world.
© 2013 William H. Stoddard
An Enquiry Concerning Human Understanding
W. E. Stoddard's
Philosophy at Troynovant
Mentality at Troynovant
Coining at Troynovant