Hume (and Popper) should take experience more seriously

It is truly remarkable that Hume (and before him al-Ghazali) was able to convince so many that there was something profoundly and systematically wrong (or, at least, seriously problematic) with our knowledge of causation, for there is surely nothing we have more intimate, varied and continuous experience of than cause-and-effect. It pervades our existence, from acts of breathing to the most elaborate projects.

How does Hume do so? By the same means al-Ghazali does. By putting the threshold of reason-to-believe very high and by putting the strength of the evidence of causation very low, both the experiential and what might be called the “metaphysical” evidence.

Regarding reason-to-believe, both Hume and al-Ghazali require proof in the sense of deductive validity. That is, they deny we can infer from cases to generalities, they deny the power of probabilistic reasoning. That all observed swans are white demonstrably does not prove that all swans are white: witness the black swans of Australia. All fine, but Hume and al-Ghazali go further, they claim that all observed swans have been white gives no reason to believe that the next observed swan will be white. For if one accepts that there are reasons to believe which are not proof, but nevertheless rational bases for belief, then Hume and al-Ghazali’s argument fails.

This is the key premise on which the entire argument sits. But, in order to make their case more plausible, both Hume and al-Ghazali recast our experience of causation into passive observance. That all we are doing is observing coincidental occurrence of cause and effect.

But that is not even remotely a full description of our experience of causation – or, indeed, our experience of experience. Experience is primarily not passive, but active. We actively – indeed interactively – experience pattern, structure and causation. Our prime experience of causation is not observing, but doing. We act, and by acting build up great experience of causation, of what does or does not happen when we do various acts. We are “inside” causation, and experience it thus. This is rather more than passive observance of coincidental cause and effect. It is only by making us think of causation as something “out there” – as something of removed, passive observation – that Hume and al-Ghazali make the characterisation of causation as mere coincidental occurrence remotely plausible and so seem as not having sufficient proof to constitute knowledge.

In his response to al-Ghazali’s argument, ibn Rushd (“Averroes”, known to medieval Scholastics as The Commentator, just as Aristotle was The Philosopher) appealed – alas, in a somewhat roundabout way – to the rationality of our belief in, and the depth of our experience of, causation when he stated that:

To deny the existence of efficient causes which are observed in sensible things is sophistry, and he who defends this doctrine either denies with his tongue what is present in his mind or is carried away by a sophistical doubt which occurs to him concerning this question. For he who denies this can no longer acknowledge that every act must have an agent. The question whether these causes by themselves are sufficient to perform the acts which proceed from them, or need an external cause for the perfection of their act, whether separate or not, is not self-evident and requires much investigation and research. And if the theologians had doubts about the efficient causes which are perceived to cause each other, because there are also effects whose cause is not perceived, this is illogical. Those things whose causes are not perceived are still unknown and must be investigated, precisely because their causes are not perceived; and since everything whose causes are not perceived is still unknown by nature and must be investigated, it follows necessarily that what is not unknown has causes which are perceived. ‘ The man who reasons like the theologians does not distinguish between what is self-evident and what is unknown, z and everything Ghazali says in this passage is sophistical.

Ibn Rushd then continues on to re-state the Aristotelian analysis of causes.

Hume and al-Ghazali are both correct to say we lack deductive proof of causation. As it is never certain that we have perceived and understood every aspect of some event, we never have more than probabilistic reasoning available to us about specific causal sequences. (So the notion that a “disproving” observation partakes of some deductive power that a “confirming” one does not is false: my old teacher David Stove used to make this point, but it took me years to fully understand what he was getting at.) Yet probabilistic reasoning can still be good reasoning, well up to the task of providing a rational basis for belief (and our systematic understanding of probabilistic reasoning has advanced quite a lot since Ibn Rushd’s day). Hume and al-Ghazali are wrong to insist that only deductive proof will do.
Read More...
While I do not find Ibn Rushd’s restatement of the Aristotelian analysis of causation in his response to al-Ghazali a satisfactory one, he is correct to take a metaphysical, which is to say a “structural” turn, for Hume and al-Ghazali’s “flattening” of experience is intimately connected to the way they discount considerations of pattern and structure. Our understanding of causation is very much built up on the basis of pattern and structure: that, for example, you can do things with a fork that you cannot do with a bat, and vice versa.

Both Hume and al-Ghazali are reacting against Aristotelian Forms, deemed to be direct apprehensible by the human mind from which causation can be deduced from our knowledge of said defining forms. But, in the way of human thought, their reaction turned into over-reaction. The belittling of the role of pattern and structure went rather too far. Hume was, for example, quite wrong to claim all knowledge of the world comes from mere experience, treated as a mass of sense-impressions, because without pattern and structure there is no basis for connecting one experience with another. We come into the world primed to identify pattern and structure: we have to be, to be able to learn, to be able to connect one experience to another. In the words of a early childhood educator, babies:

… apply limited Action Schemes to everything they come across, which is part of learning about the world. The earliest action schemes are things like grab and suck. Babies are born knowing how to do this—or at least they can do it within hours of being born. Later there is grasp & shake, which is why we give babies rattles. There's also banging, dumping (around 9 months) and turning things over to see the other side. As babies develop they get to combine these basic action schemes into more complicated patterns. So younger babies are kind of looking at the world of objects as: is this something I bang, something I shake or something I ignore? Babies & toddlers put everything in their mouths too, just like animals do. Touch it, to feel its texture, grasp and shake and turn it over, taste it. Watch a gorilla do the same thing to an unknown item the next time you're at the zoo.

Anyone who has to deal with a complex and crowded intersection for the first time will testify how different that confusing experience is to later experiences of the same intersection when one becomes familiar with its patterns and structures. Pattern and structure is basic to, and pervades, both our experience and our learning from experience, for without pattern and structure there could be no learning to speak of at all. (Nor, for that matter, any existing things at all.)

Which is not to say this priming to perceive pattern and structure is itself definitive. As a brain scientist Gary Marcus says:

The initial organization of the brain does not rely that much on experience … Nature provides a first draft, which experience then revises … ‘Built-in” does not mean unmalleable; it means organised in advance of experience” (Gary Marcus, 2004).

The priming is malleable and interactive with experience. We are not compelled to wander off into Kant’s world of phenomena and unknowable things-in-themselves noumena.

What is even more surprising about the intellectual success of scepticism about our knowledge of causation is that no one actually believes we have no understanding of causation: indeed, all of us continue to act on the basis of our built-up understanding causation based on our experience of (and inferences from) pattern and structure. (Which is then expanded by information transferred to us by language from others.) Hume’s monetary economics, for example, is all about generalised causation.

Al-Ghazali’s success is marginally less surprising than Hume’s, however, since al-Ghazali achieved extra plausibility by appealing to God: his take on causation becomes another manifestation of the submission to God that is the essence of Islam.

Aristotle was not wrong to be so concerned with pattern and structure: it is necessary for things to exist, including for thought to happen, and for any learning to be possible at all, for one experience to have any implication or connection to another. Since pattern and structure are necessary for things to exist, the universe can therefore be described using mathematics, the science of pattern and structure. The trouble was being too Platonic about pattern and structure: to put too much confidence in ideas, in the human capacity to directly categorise accurately and definitively.

But we need to be wary of the opposite problem, of being too dismissive to human perception and reason. To be sure, that mathematics has the “gold standard” of deductive proof available to it is a seductive example. But our logical and cognitive apparatus is not so impoverished that only deductive proof is good enough. This cannot be so, since so little is available to us to connect on the basis of deductive reason from what is known with certainty. We live in a world of uncertainty and so of probabilities.

And of reasoning perception based on a capacity to categorise which is far from perfect and definitive but demonstrably good enough to get by. And good enough due to reasons beyond mere happenstance. That something is not certain is not a reason not to believe it. Mere possibility is certainly not a superior reason to believe than experience. Indeed, it is not even as good a reason to believe.

Which is why the cool rationality of Hume’s scepticism about our knowledge of causation is, in fact, deeply irrational since it elevates mere possibility (which is what uncertainty is) as more defining of experience than experience itself, in the sense that mere possibility trumps inferring from experience.

Hence Hume, and those whose philosophy descends from him, such as that of Karl Popper, should take experience, including of pattern and structure, as a basis for inference more seriously in their philosophy. Indeed, as seriously as they did every waking moment of their lives.

So, Ibn Rushd has a point, the reasoning is sophistical, as long as we grant that the first deceived were the reasoners themselves.