There is a difference between knowing that something happened, and knowing why something happened, but is there a difference between knowing why and knowing how?
What are we hoping to get in reply to the question 'Why is it raining?' - an explanation of meteorological systems, a discussion of what we have done to deserve a drenching, or maybe even an evolutionary explanation of how if we were never rained on we wouldn't be here to be debating it right now? The issue of explanation is of deep interest to philosophers of science, and we should hope to make our conception of it clear before attempting to go about explaining events and phenomena around us. I hope to show that explanation should be considered to be intimately related to causation, but that a more specific and focused account than that proposed by (amongst others) Lewis may be necessary if we are to truly reflect the ways in which people intuitively answer 'why?' questions, and how they attempt to explain events. I shall also address a number of problems that I feel might be posed as an attack on a 'simple' causal view such as Lewis', which holds that there is a direct relation between number of causal links exposed and the degree of explanation uncovered. I shall try to show that explanation cannot simply be considered a property that can be passed from cause to effect (which would benefit philosophers trying to establish a complete causal history) with the 'basketball' case; that outlining a complete causal history will not be simply difficult but impossible, with the 'dead parachutist' case; and that whilst most non-causal explanations will not satisfy our theory, they should not satisfy our intuitions about what makes a good explanation anyway, with the 'spinning coin' case.
In his essay 'Causal Explanation' David Lewis argues that the best formulation of what constitutes an explanation of a particular event is to outline the causes of the event, and that such a clarification is what is being sought when someone asks a 'why?' question. Such a conception fits well to our intuitions about what makes an explanation and to be able to explain the necessary conditions for an event to occur and perhaps certain rules governing the relationship between such conditions and the event in question would certainly seem a plausible formulation of what is needed to explain why the event in fact occurred. Furthermore, a causal explanation seems to be able to tackle (presuming we have the information involved) questions about why something didn't happen, as well as what it might take for something else to happen, so may well be even able to answer counterfactual 'why?' questions.
Do we need, however, as Lewis seems to suggest, to outline the entire causal history of an event to get a complete explanation of how the event came to happen? I would certainly argue that we do not, but further, in contrast to Lewis, would argue that for incomplete explanations it is not simply a matter of the larger the map of the causal history the fuller the explanation we have. Lewis does not seem to give adequate conditions as to what makes particular events in a causal history more useful for explanation than others, but rather seems to think it is just a matter of quantity. I would claim that not only is a large amount of information frequently less useful than a single specific piece of causal information, but that usually other information is entirely superfluous, and fails entirely to help explain.
Consider the example of a basketball game, that you have seen on television but that your friend has not. If he were to ask you later why a particular team won the game, your reply might be in the form of a brief commentary of how the game was played, or a weighing of the relative merits of the different teams, or a discussion of the injury which had prevented the star play from being able to participate in the game. Only your friend knows which particular response he wanted, and without a clearer question specifying which it is, any of these responses would be reasonable. It seems unlikely, however, that even if your friend is a scientist, you would try to give him a complete account of the causal history that led to a win for the particular team that was victorious, and do so by telling him the exact positions of every atom involved in the game (in the form of players, court and ball) and the velocities of each of these atoms. Given a supercomputer, it might well, with only the initial positions and velocities of all these things, be possible to predict who has won the game, since these are the only facts that causally determine a win (and also, of course, the amount of muscle each player has, and thus how fast they can run, how high they can jump, how far they can throw, etc.) yet, I would argue, even if this was a very full causal history for the game, it is no explanation at all, and your friend would most likely go and ask someone else. So an explanation cannot simply be any part of the causal history, and it is not a matter of quantity of causal relations, but quality of choice in which to tell.
I would argue, furthermore, that those philosophers who claim that a complete explanation can only be delivered by outlining a complete causal history are not just undertaking a very difficult task (whether or not it is unnecessary as I have argued), but an impossible one. If I was to ask someone why a parachutist died, I would not be happy with the 'explanation' 'Because he jumped out of a plane'. We do not normally expect that a parachutist jumping out of a plane will die (that's rather important for the sport, in fact!), and so to be told this piece of information, even if it is the most important event leading up to his demise will not be useful, and in fact would probably get sneers of 'pedant' in response. But a greater problem might be created by a different answer I might be given - namely 'Because his parachute did not open'.
This at first looks highly satisfactory, and, whilst if I were a grieving family member I might seek a more specific reason why the parachute did not open, it does give more information than I had before (after all, he could have got entangled in the cords, or been unable to breathe, or landed incorrectly), yet on closer inspection it seems problematic. The use of negative explanations begs the question of why we have mentioned the particular things we did. Even more than with positive explanations where one might be tempted to argue that we're leaving out certain ceteris paribus considerations, with negative explanations we have a problem as we seem to be invoking one of these things as our explanation. Why do we need to say that his parachute did not open, but not that there wasn't a haystack to break his fall, or an ocean, or a temporary reversal of gravity? A reply to this might be that the only reason that we mention the parachute in particular is that that is out of the ordinary, and all the other factors are implied but need not be stated explicitly. I do not think that this is a satisfactory response, because this seems to ignore what an explanation is. We can't say that an explanation always contains many extra implicit facts, because clearly if we never state these facts then we never use them to explain. An explanation cannot be said to include an enormous number of other factors if we take a response without any of these factors involved to be a perfectly adequate explanation. Someone who is unaware that it was only a bizarre fluke that a farmer removed a haystack from where the parachutist landed the day before is no less able to give an adequate explanation because of this ignorance. Whilst a complete survey of the causal history would require the inclusion of an infinite number of non-events, an explanation does not.
A distinction should definitely be made between cases of negative explanations (where something is explained by an event not occurring) and cases where there simply is no explanation. If someone were accused of stalking another person, their response might be that there was another explanation for their always being where the 'victim' was, such as that they always share the same lectures, but there is another possibility - that there simply is no cause at all, that it was merely coincidence. These sort of failed conspiracy theories rest on the absence of a cause, yet still make a satisfactory explanation, so what is to stop us arguing that just about any fact can be explained in non-causal terms, i.e. by saying they just happen to coincide. The first response might be that this could well be the case, but happens not to be, since many things do have added causal explanations, but this is problematic since then the burden seems to be just that we will agree that things that look like they have an explanation can be explained causally, and that with other things we should be satisfied with the argument that there is no cause at all.
A favourite example of mine comes in the form of an article I read about a year ago in a newspaper. It was about some research into, of all things, coins spun of a flat table. The basic premise was that when a coin that has been spun falls over it spins faster and faster as the vertical movement is converted into horizontal movement, or something along those lines - the specifics are not important - what was important was that it was suggested that this acceleration should continued right up to the moment when the coin was flat, in inverse proportion to the angle between the coin and the horizontal. Basically, it was argued, the coin should spin faster and faster until, the moment before it landed, it was moving infinitely fast. The research was investigating why this did not happen, and the conclusion settled upon was that if it did spin infinitely fast it would create a singularity, but nature doesn't like singularities, so it doesn't create one, so it doesn't spin infinitely fast. This struck me as a ridiculous reductio, and yet was seriously being suggested as an explanation.
Lewis seems to come up with an analogous argument, yet seems entirely satisfied with it. I would argue that he was mistaken to. His argument is about a collapsing star. He explains that stars that collapse shrink to a certain size and then cease their collapse. Why does this happen? Because, he argues, the Pauli Exclusion Principle, which outlines how dense a star can get, says that it would be impossible for it to collapse any further. He is very explicit that there is no underlying reason, that there are no opposing forces stopping the collapse, and suggests that this is an example of a non-causal explanation. My response would be that this is not only an unsatisfactory causal survey, but luckily no explanation at all either. We should not be satisfied with the quoting of a law to explain something, but hope that underlying it is some further reason, in terms of properties of the parts involved. This might, at the smallest level, cause some problems - how far should we go? We may eventually have to concede that we have gone as far as we can (though I strongly suspect human knowledge will run out before we can get beyond a level where we suspect that there is an explanation but are unable to find it - though quantum theory would suggest that at the lowest level things cannot be explained, and are simply random) but certainly at this level we can fairly demand a further explanation of the law in terms of atoms and forces. Further, if we believe the (somewhat contentious) claim that where we know all the initial conditions and causal connections we can know what must happen, then if we argue that it is impossible for the star to collapse further, we are simple restating the fact that the causes will lead to it not doing so.
Overall, therefore, I cannot see how there is anything more to explaining why something happened, than explaining how it came about to - yet this need not, and cannot, involve a complete causal history, but instead should involve the pertinent, immediate causes. We can always, of course, hope that the person asking the 'why?' question is a small child, since then we will get to outline the entire causal history, as they ask 'why?' in response to our explanation, requiring one step further back in the causal history, ad infinitum!