In this essay I hope to vindicate Hume’s claim from the Enquiry Concerning Human Understanding that neither experience nor reason can support the principle of induction. In so doing I suggest we take a close look at how experience is constituted, following Karl Popper in arguing for a reversal in the relations Hume supposes between observation and hypotheses. The result of this will point the way to criteria which removes the requirement of rational or empirical justification for first principles on condition that they are empirically falsifiable.
First Principles: Induction
Hume sees our human understanding as wrought in two: (1) relations of ideas and (2) matters of fact. Relations of ideas are a priori truths, e.g. that three plus five makes eight, which we can know without having to look at how things are in the world. Matters of fact, contrarily, are known a posteriori which means that we have to look at how the world is in order to come to have such knowledge, e.g. that motion from one billiard ball communicates motion to another. In section IV of the Enquiry, Hume is keen to investigate the nature of our evidence of matters of fact “beyond the present testimony of our senses, or the records of our memory”.
To begin he supposes that our knowledge of matters of fact is founded on knowledge of cause and effect, citing the example of finding a watch on a barren desert island. Finding this watch would lead one to conclude that there had once been people there through simple inference from the effect (watch) to the cause (people). As we know watches exclusively as an effect of human activity we come to associate the phenomena and so are able to infer, upon finding the one, the former presence of the other. So now the problem becomes one of discovering how we arrive at knowledge of cause and effect.
Hume argues that we can’t manage this a priori because we cannot infer the causal powers of anything from sensible qualities alone and so we must become empirically acquainted with causes and effects in order to make such an inference. One could not, for instance, deduce upon first seeing two pieces of smooth marble that they would become extremely difficult to separate except by lateral pressure - only by having first experienced such a phenomenon could we come to expect it when next presented with two pieces of marble. If a priori reasoning cannot ground our knowledge of cause and effect that leaves us with one remaining candidate: experience. Following on from this in Part II of section IV Hume endeavours to find out what the foundation of our conclusions from experience is.
Remarkably, Hume claims that such conclusions are “not founded on reasoning, or any process of the understanding”. He establishes this by considering two propositions of the sort we find in all reasoning concerning matters of fact:
This second statement is known as the principle of uniformity. In Humean terms, it is the notion that the hidden causal powers governing the behaviour of the natural world are uniform across time. But the question is: how do we infer (2) from (1)? We begin with such experience as we have in witnessing the constant conjunction of two events, one billiard ball communicating motion to another. We might witness this simple phenomenon hundreds, maybe thousands of times – but how could we possibly derive (2) from such temporally limited cases?
(1) "I have found that such an object is always attended by such an effect".(2) "I foresee, that other objects, which are, in appearance, similar, will be attended with similar effects".
Hume admits that he cannot discover the device which leads us from the one to the other, but is not willing to say that as he cannot presently think of the solution to our conundrum that no such solution is therefore possible. However, in order to bolster his negative conclusion he will take apart the two forms of human understanding and try to show why we can’t in principle discover the reason for our inference from (1) to (2) in either of them. We’ve seen why we cannot make this inference a priori given that we can certainly conceive of the causal laws governing the universe changing in the future. We must therefore focus instead on probabilistic reasoning regarding matters of fact. Recall Hume’s argument up to now:
i. Our reasoning concerning matters of fact is based on cause and effect (p. 26)
ii. All reasonings and conclusions based on cause and effect are derived from experience (p. 29)
iii. All our experiential conclusions derive from the supposition that the past will resemble the future (principle of uniformity).
We see that the foundation of our conclusions based on experience is the principle of uniformity and so we cannot then secure this principle on experience without falling into circularity! We find ourselves with a problem: all of our empirical knowledge happens to be based on causal inference and yet when we look for the ground upon which this knowledge rests we cannot find it in the understanding at all. No a priori reasoning and certainly no amount of experience can justify our supposition of uniformity, for Hume.
How then are we led to anticipate future events based on our finite past experience? Hume proposes that it is habit or custom which leads us to this expectation. He defines habit/custom as the propensity of a repeated operation or act to cause the reproduction of the very same operation or act, reiterating that no process of the understanding is factored in. But, significantly, as this kind of faculty is not cognitive/propositional in nature (owing to its not being an operation of what Hume calls the understanding) it means that the principle of induction is not a principle at all but rather a mode of behaviour. What we must now consider is the possibility that behaviour can function as a mode of the understanding.
Knowing-How & Rhetoric
Italian rhetorician Ernesto Grassi sees the problem of first principles as arising from a failure to appreciate the metaphysical relation between two types of discourse: the rhetorical and the rational. Rational discourse consists in proving things, and to be proven means to be demonstrated on the basis of something. In order to prove the notion that a moving billiard ball will communicate motion to a stationary one, for instance, we point to our past experiences of such phenomena on the basis of the principle of induction. Raw experience alone only gets us as far as constantly conjoined perceptions - we must make an inference based on the principle of induction in order to arrive at the general claim adverted to.
But rational principles cannot themselves be demonstrated by rational means, as to be proven demands the supposition of just such first principles. This is another way of expressing Hume’s problem above, that induction cannot ground induction without presupposing itself and thereby running into circularity. So when we posit these first principles, Grassi asks, what is the form of discourse through which we do it, if not rationality? As Hume has eliminated the possibility of experience filling this role, perhaps we ought to consider Grassi’s claim that it is from out of rhetorical, creative discourse that these suppositions come into being. Grassi understands rhetoric metaphysically as the origin of rational discourse and not, as is typically understood, a set of clever argumentative tricks. Rhetoric is understood as emerging from praxis, the day-to-day concerns of human existence, which are historically contingent.
As we’ve seen, Hume concludes that neither demonstrative nor probabilistic reasoning can ground the principle upon which induction rests – it is only habit which leads us to suppose this. But what if the phenomenon proposed here as “habit” was in fact a form of the understanding? Certain philosophers in fact do see this non-propositional habitual comportment as a form of the understanding, one upon which propositional knowledge rests. Gilbert Ryle was one such philosopher:
“…both philosophers and laymen tend to treat intellectual operations as the core of mental conduct […] they tend to define all mental-conduct concepts in terms of cognition. They suppose that the primary exercise of minds consists in finding the answers to questions and that their other occupations are merely applications of considered truths…”
While this might not be totally true of Hume (see footnote 12), nevertheless he is partly implicated for imagining that the understanding performs only cognitive operations, either through a priori deduction in formal systems or a posteriori inference based on past experience. But our readiness to deal with situations in our daily lives exhibits understanding too, for instance we know-how to use all sorts of things without needing to make explicit inductive or deductive inferences based on considered truths. Ryle saw the application and consideration of discovered truths as an operation of intelligence, and not vice-versa. Praxis, then, is perhaps a form of the understanding too.
With respect to this the first thing which strikes us about Hume’s account is his restriction of the understanding to propositional forms, both demonstrative and probabilistic. While his insistence that sceptical abstract reasoning will never affect the day to day comportment of human life is very astute, nevertheless I feel his cognitive-propositional picture of the understanding prevents him from fully grasping any possible non-propositional origins of logical principles - leaving them shrouded in mystery. In the next section we will consider more evidence regarding the notion of practical concern and the role in plays in human understanding.
The Necessity of Perspective
What Hume says about the specific nature of experience aside from its relation to ideas is scant, suggesting he thinks the concept is unproblematic and in need of no elucidation. However, if we take a closer look at how experience is resolved to us we might gain significant insight into Hume’s problem. Karl Popper helpfully deflates the myth of directly intelligible experience, insisting first that experience is experience-for something and that as such any perceived repetitions are repetitions-for-something. The perspective of the subject is paramount as any perceived repetitions have to be interpreted as repetitions from within a perspective. This is as against the “naïve view” that events are similar and their being recognised as such is down to our unmediated perception of that fact.
In order to make this clear consider two separate cases involving one person passing an item to another person. Both instances are formally identical with respect to experienced sense-data (which is to say in both instances we witness one person passing an item to another). Now suppose that these instances have different meanings - in one a gift is being given and in another a debt is being repaid. Sense-data alone cannot convey these different meanings as the sense impressions themselves are identical and therefore contain no indication of how to discriminate the different meanings. The meanings of the two instances are found in the purposive activities involved. In the gift-giving example, for instance, let’s say it’s your birthday and so when the gift is presented to you, you can interpret the item as a gift according to your expectations of the situation. Contrast this with the debt repayment instance – one is not expecting a gift, but they are expecting repayment from their friend and so they interpret the item as repayment, not as a gift.
Popper furthers this challenge, inviting us to appreciate the absurdity of the instruction: “take a pencil and paper; carefully observe, and write down what you have observed!” This instruction immediately solicits the question “observe what?” – alluding to the requirement of a purposeful direction to guide any observation:
“Observation is always selective. It needs a chosen object, a definite task, an interest, a point of view, a problem. And its description presupposes a descriptive language, with property words; it presupposes similarity and classification, which in its turn presupposes interests, points of view, and problems.”
It’s helpful to think of this in terms of a library. If it was disorganised it would be very difficult to find any book we were after, so we organise it as best fits its particular purpose. In a university library, for instance, the books are typically divided along disciplinary lines; Philosophy, German, Psychology, etc. Popper’s point is that we perform a similar feat in our experience, interpreting/organising it according to how what we’re perceiving matters to us, relative to our interests, points of view, and problems.
The import of this result is two-fold. The first outcome is that Hume is right that experience alone cannot support the principle of induction (as we have discovered the necessity of a perspective from which to interpret experience). The second outcome is that Hume’s psychological solution runs into the exact same difficulties as we found when trying to explain the supposition of the principle of induction from experience. This is owing to the fact that repetitions need to be interpreted as repetitions from within a perspective so that even habit presupposes experience within an organising perspective. Two avenues within the problem now point us to the perspectival character of experience.
If the first outcome stands we see that the proper starting points for securing knowledge are problems emerging from our particular interests/concerns – not raw perception itself. We have seen how we cannot subtract our active sensitivity to the meaning of a situation and try to rebuild it from the sensory remainder as one needs their concernful perspective to interpret sensory experience. An experiment involving young dogs which Popper mentions offers a great example of this. After being exposed to a lit cigarette for the first time the dogs turn and move away from it, unable to be coaxed into returning. Days later they were shown the cigarette again and responded in like manner, even turning away from a rolled up piece of white paper, interpreting it as similar in kind to the cigarette which had previously disturbed them.
The dogs’ past experience informs their current perspective and enables them to anticipate discomfort when presented with another object which looks like the offending cigarette. But if (1) a perspective allows our experience to be intelligible while, (2) experience simultaneously informs and shapes that very perspective - are we not again stuck at circular reasoning? Not this time, as our expanded notion of experience shows us that the dog’s initial apprehension of the cigarette as unpleasant was not a passive/direct experience but a concernful experience mediated by the dogs’ simple aversion to unpleasant smells. The experiment shows that no foregoing experience is required for an object to be initially interpreted as something.
Now, as we’ve seen that knowledge does not proceed from unmediated experience but from experience within an organising perspective, we may suppose that the origin of the principle of uniformity is found in those very activities in which the anticipation of regularity is discovered. What Hume called habit is not any passive reception of repetitive impressions, in fact we’ve seen that it need not even be repetitive but can be formed after one instance. Habit, rather, is the active appropriation of things relevant to our interests. This leaves us with the possibility that no justification is required for our expectation of regularity save its genesis in dealing with things in a world organised by our interests.
But Popper’s solution does not cast our principle of uniformity to the hounds of irrationality. While it may originate in the behavioural anticipation of uniformity this does not mean that it isn’t subject to another empirical process: falsification. If no amount of verifying experience can justify the leap from constantly conjoined perceptions of a certain kind to a universal rule governing all possible perceptions of that kind, induction is troubled from the ground up. However, a falsified theory is falsified deductively – meaning that you can logically demonstrate that something isn’t the case rather than inferring that it probably is. Finding a black swan, for example, renders it deductively certain that the hypothesis “all swans are white” is false. The aim therefore is not to verify the principle of uniformity by experience but to ensure that it is able to be falsified by experience. The condition is that falsification takes place within a perspective wherein a falsifying observation is interpreted as a falsifying observation.
Hume, D. (1909). Enquiries Concerning the Human Understanding and Concerning the Principles of Morals. Clarendon Press.
Popper, K. (1972). Conjectures and Refutations. Routledge and Kegan Paul.
 (Hume, 1909), p. 26
 Ibid. p. 27
 Ibid. p. 28
 Ibid. p. 32
 Ibid. p. 34
 Ibid. pp. 34-35
 Ibid. p. 43
 Ibid. p. 43
 (Grassi, 2001), pp. 18-24 (note:)
 Ibid. p. 19
 Ibid. p. 19
 Ibid. pp. 18-19
 Ibid. pp. 6-7 (if we were not concernful beings for whom the natural world has meaning with respect to those concerns, nothing like induction would be necessary as we would have no need to make any inductions).
 (Ryle, 2000), p. 27 (my italics)
 (Hume, 1909), p. 41
 (Popper, 1972), p. 44
 Popper doesn’t explicitly use the word ‘perspective’ however I feel it is a convenient catch-all expression for the various phenomena he lists as helping organise and make sense of experience (see footnote 15).
 Ibid. p. 45
 Ibid. p. 46
 Ibid. pp. 46-47
 Ibid. p. 45
 Ibid. p. 155
 Ibid. p. 44
 Providing all background assumptions stand, etc.