One thing study of Hume, his influences, and responses to him has taught me is that the single most important area of philosophy, in terms of its ramifications, is philosophy of causation. Everything leads back there; it pulls everything together; and a very tiny change there can lead to massive changes everywhere else. While not all reasoning is causal, most reasoning involves causal reasoning of some sort. It is crucial in the strictest sense; it is the major crossroads of philosophy. One thing that I think needs to be worked out more clearly is the way in which historical reasoning is causal reasoning, and these are just some notes in that direction.
(1) To say that a given description of the past is reliable is to say that this description is an effect of such sort that it can only be traced back to actual historical actions that are such as the description describes, allowing for approximation (or, what is the same thing, in such a way as to be an appropriate means for the ends of the kind of inquiry one is doing). By saying 'it can be traced back' to these causes, I mean that it is caused by them, through intermediating causes.
(2) All historical accounts are effects; all effects have causes. The big problem for historical inquiry is determining how the features in the effect can be accounted for by causes, and what kinds of causes those must be. This genus of causal reasoning has not gone unnoticed in the history of philosophy; in fact, most undergraduates in philosophy have read one of the most influential early modern texts on the subject, although the discussion there was for different ends than clarifying historical inquiry. I speak, of course, of Descartes's Meditations III. There the question is of how to determine the features of the cause of an idea
from the features of the idea
, considered as an effect. But it is structurally the same problem, and what follows are some rewritings of parts of the argument of Meditations III in terms of historical accounts rather than ideas in the mind.
If we only take historical accounts to the extent that they are accounts, they seem on par. But if we take historical accounts in terms of what they represent to us as objects of thought, they end up being rather different. And this is the key point. Everything in the effect must be traced back to the causes of the effect. Thus, taking historical accounts as effects, as they obviously are, we must be able to identify causes not merely of the formal reality of these accounts, that is, for the fact that they exist and are accounts, but also for the objective reality of them, that is, for their being accounts saying this rather than that. What is more, we must take the causal chain resulting in the historical account to explain everything in the historical account without remainder, although obviously for practical purposes there will often be gaps. But it must be so: it is an a priori
precondition for any inquiry that effects can be fully explained by their full set of causes. Given a principle of this sort, rational inquiry is possible; without it, it will never get off the ground.
Now accounts may, with the help of additional intermediate causes, give rise to other accounts, serving as the partial cause of later accounts (partial because accounts tend not to be self-propagating). This cannot regress infinitely, though; we must in the end reach a first account, the cause of which is the archetype from which the objective reality or content of the historical account comes. (There need not, of course, be one single cause for the whole content of the historical account, but every bit of content in the historical account must have some originating cause.)
Historical inquiry, insofar as it deals with historical accounts, is the attempt to identify, from the features of the effect and what is known about the intermediate causes, what the features of this archetype must be, may or may not be, or must not be.
(3) It is clear from thinking about these chains of intermediate causes that the most important thing to determine about them is whether any of the parts of the total cause of our historical account are defective or deficient causes. This genus of causation, too, has been given some thought, although again not with the specific purpose of elucidating historical inquiry. It has arisen in the context of privation theories of evil. As Augustine noted, if evil as such is a privation of good, then, while there is an oblique sense in which evil has an efficient cause, it is more accurate to say it has a deficient or defective cause. That is, evil as privation traces back not to an efficient cause insofar as it is effective, but an efficient cause insofar as it fails
to be effective.
This is actually going to be true of privations generally, including failures to be accurate in recording something. When we ask whether an account is reliable, we are asking whether it lacks what is required to be a true account; or, in other words, we are asking whether any of the partial causes that constitute its total cause are defective causes precisely to the extent that they are causes of the account at all; and, if so, whether these deficiencies are significant for our purposes and whether any other causes compensate for them. All errors must have some defective cause; and in assessing the reliability of an account historians spend their time ruling out the possibility of various deficiencies in the causes, such as malice, deception, confusion, gullibility, or mistake.
(4) We could also say that claiming an account to be reliable involves claiming that it can be trusted. It's important to understand that this is not a different account of reliability from that found in (1). When we trust an account to tell us what happened we are assuming something about how it relates to its causes. Either we have reason to think it can only be traced back to actual historical actions that are such as it describes, or we have no reason to think it doesn't and are willing for our purposes to work on the assumption that it does. The primary value of putting it in terms of trust is, besides the fact that this is often the way people put it anyway, that talk of trust brings out the fact that every evaluation of the reliability of an account is in terms of how reliable it is as a means for whatever precise ends we have in view. It is this that raises questions about whether the account is reliable to a sufficient degree of approximation. Sufficient for what? Sufficient for whatever we are doing with it. Putting it in terms of trust brings this out nicely.
(5) All of the above talks about what historians do with accounts -- that is, testimony. Obviously historians do not only work with testimonies about what happened. They work with artifacts (papyrus, monuments, coins, etc.) as well, and even with natural objects. The reasoning in these cases is causal as well -- artifacts and natural objects are effects, with causes of their own, and we can ask broadly similar questions. The main reason for focusing on testimonies or historical accounts first, though, besides the fact that historical accounts are very, very important for historians, is that causal inferences to positive conclusions about the content of a historical account are generally quite direct and straightforward. The full-scale causal inquiry to establish these inferences can be quite complicated, and a single account can require many, many such inferences. But the inferences themselves are fairly straightforward, and tend all to be of the same kind. This is not necessarily so with other things historians look at. In talking about historical accounts
, we have already narrowed down the type of cause we will be considering: accounts are by their nature direct or indirect effects of cognitive agents. Other causes only have to be considered very indirectly, insofar as you might want to know why a cognitive agent might make this mistake, or that modification, or anything else. Any involvement of a noncognitive cause that is direct (i.e., not mediated by a cognitive cause) would be rather unusual. Almost the only time it ever comes up at all is when the medium in which the account itself is preserved can be a defective cause for the account; and the way in which the medium can be a defective cause of the account is simply by loss of information (holes develop, edges shred, magnetic structure is disrupted). When we deal with artifacts as such, however, there are lots of other causes besides cognitive agents that consistently are important, despite cognitive agents still being fairly important; and when we talk about natural objects, cognitive agents are often very minor parts of the full explanation, if at all, and there are jillions of other factors to consider. Thus historical analysis of testimony, while not simple in itself, tends to be simpler in overall structure than other kinds of historical reasoning.