Imagine this: you go and bash the bell of a doorbell with your hand hard enough for it to ring forth. But then some philosopher comes up to you and says 'Ah, but do you really know it was your bashing that caused the ringing? Might this not just be one thing being correlated with another but not being caused by it?' My question today is: How does one get to be such a philosopher? What must have gone wrong with the bent of her thinking such that questions like these start to seem sensible to her?
Well: now imagine that you only thought of cases like the above after first having considered wireless doorbells. The postman comes with a parcel. He presses the button. The bell rings inside. But later your neighbour presses the button and the bell doesn't ring. What's going on? You realise that the receiver was tuned to the wrong frequency and in fact, when the postman came, it was the coincidental pressing of another neighbour's button, to which your receiver was accidentally tuned, that caused your bell to ring when it did.
Consider, then, that your philosophical thinking about causality was itself fundamentally conditioned by such cases in which we have to try to distinguish causes and coincidences. To get this thinking off the ground you'll need to make sure you have a fairly alienated set of cases to work with. You'll want your causes and your effects to very much be 'distinct existences'. We won't want to be thinking of cases like squishing an egg in your hand or crushing a nut with your teeth. We'll want to be thinking of cases like bells and buttons. We will want cases in which there is necessarily a mechanism involved. Not cases in which we have to do with the workings of proximal components of the mechanism - which presumably, on pain of infinite regress, do not always have their own linking mechanisms.
When you think about it, the same kind of set of examples are required here as in any other situation when we want to get a philosophical problematic off the ground so we can set to important 'work on' it. First ignore the engaged human being living effectively and spontaneously as part of her proximal environment itself containing objects in proximal relations to one another. Instead sit back and go for the disengaged perspective, contemplating objects which themselves may or may not really be engaged with one another. Hive off, at least for a while, your understanding of the being of causality from your living encounter with your environment, imagining that it is some kind of an open question whether what you take yourself to encounter really does instantiate causality. (If you want you can console yourself with the notion that you can always plop causality back in later with a transcendental argument or some other train of thought borrowed from transcendental idealism. Causal animation might not be encounterable in nature: never mind - just animate nature with the form of your own thought!)
In this way you can happily start to make engagement, relatedness - be it causal, intentional, or rational - seem somewhat unsafe, happenstance. Start with the uncertain cases: start with mere correlations, belief, arm risings, appearances - and then ask yourself what your warrant is - since if you start from there you can be sure that warrant will be required - to ascend to causes, knowledge, arm raisings, realities. Now you can start too to think that causality must always be realised in a mechanism - since mechanisms are required when we have two distinct things that require linking up. Your alienated perspective on the natural world is now pretty much complete. And now you can start to feel confident that you've got enough of a rationale to start developing 'theories' to help us all distinguish between causes and mere correlations. That distinguishing, you now urge, not being something we ought to have allowed ourselves to do with such blithe confidence beforehand. Now it's obvious that 'how do you know that x caused y?' is not just sometimes, but always, a good question to ask. Estrangement now being secured, the grant application can go in in good faith and one can once again sit comfortably in all those oddly decontextualised locutions like 'the theory of x' and 'the problem of y'.