In my last post, I reflected on Elizabeth Anscombe's 1979 essay "What Is It to Believe Someone?"; that essay is, more or less, a continuation of a line of thought she began exploring in her 1973 article "Hume and Julius Caesar," which is concerned with knowledge of history. In this post, I want to sketch Anscombe's argument in this earlier work, and raise a question or two about it.
Anscombe argues that Hume was sadly mistaken in Section IV of Part III of his Treatise Concerning Human Understanding. There Hume claims that when we reason about history, we infer effects from causes, and that whenever we infer effects from causes, we must “establish” the existence of the causes either by a sense impression, or by an inference from other causes (i.e., of which it is an effect). Hume illustrates this by describing how it is that we believe that Julius Caesar was killed in the Roman Senate on the Ides of March: we infer, from some present testimony (perhaps a passage in a history book) that there was a chain of testimony stretching back to an eye-witness account of the event, and so conclude that the event must have occurred.
Anscombe first observes that Hume’s example is surprising, since the reasoning there is not inferring effects from causes, but causes from effects. However, changing Hume’s thesis to being about either inferring causes from effects, or effects from causes, doesn’t save his model. Hume has said that the end of the chain must be in our perception, but now the beginning of the chain is our perception, and the end is something in the remote past. She analyzes Hume’s argument as entailing four theses:
First, a chain of reasons for a belief must terminate in something that is believed without being founded on anything else. Second, the ultimate belief must be of quite a different character from the derived beliefs: it must be perceptual belief, belief in something perceived, or presently remembered. Third, the immediate justification for a belief p, if the belief is not a perception, will be another belief q, which follows from, just as much as it implies, p.
She goes on to note that there is a corollary that must also be believed: “When we believe in historical information belonging to the remote past, we believe that there has been a chain of record."
She then argues that this (evidentialist and foundationalist) model of historical knowledge must be mistaken. First, the written records, if they are grounds of belief, are the grounds of our belief in the killing of Caesar, “that the assassination is a solid bit of history." It is the belief in the event that is the ground for believing “in much of the intermediate transmission." She says that we believe that there were eyewitnesses to the event because we believe the event happened, and not vice-versa.
Then she says that she’s not just catching Hume out in a mistake, but is wanting to correct the mistake, which is the more difficult and significant task. Regarding how we know that there ever was a Julius Caesar, she says: we were told. Furthermore, we can’t check whether there was such a person other than by “finding out the status of the information about him."
She argues that, at least for some historical figures, the hypothesis that the figure “never existed” cannot be taken seriously. But this is not because it is a contradictory concept, but because seriously entertaining the hypothesis would undermine anything that could count as evidence of its truth. As Anscombe puts it, “that that man, Caesar, existed and that his life terminated in assassination: this he could call in question only by indulging in Cartesian doubt." Someone who first heard of Caesar in Shakespeare's plays and wondered whether he was just a fictional invention of Shakespeare’s could check up on whether Caeser ever actually existed by checking history books. But if she doubts it in spite of what is in all the historical records, then she is stuck--there's no other way to answer the questions. So Anscombe says that the hypothesis that Caesar never existed can’t even have the status of a “wild hypothesis” that is “vastly improbable.” To engage in such a doubt would place her in a “vacuum” in which she could not even produce reasons to regard the hypothesis as more or less doubtful (i.e., of having greater or lesser a degree of probability).
She ends by saying that not everything can be “put up for checking.” She rejects the analogy of our knowledge (at least of the past) as a boat that we can repair while staying afloat. Some parts of our knowledge of the past cannot be removed, and our historical knowledge as a whole “stay afloat”, as it were. (I take this last point to be a critique of at least some forms of epistemological coherentism--the "raft" that Sosa speaks of in his "The Raft and the Pyramid.")
Anscombe also makes an interesting observation about how we generally don’t postulate historical figures (such as her great-great- grandmother) in order to explain current phenomena (such as her own existence).
And that brings us to the end of the essay. As is typical of Anscombe, she brings to bear an insightful analysis, makes some great points, and some suggestive comments, but leaves a lot of questions still unanswered, and a lot of work still to be done. And at the end, I find the constructive thesis, the story of what is “really happening” in the case of historical knowledge, not entirely clear. I take it that the point is this: we believe in historical events/persons because of what we are told, whether by contemporaries of ours, or through written/recorded records from our predecessors, and these testimonies are the direct reasons why we believe; they are not the beginnings of some long chain of inferences that we must trace in order to be justified in believing. Rather, for at least those particularly well-supported (in a coherence sort of way), we cannot even reasonably doubt them without falling into a sort of radical skepticism. So we are epistemically dependent on these testimonies in such a way that we cannot verify them by other means. Something like that seems to be the take-away.
She will go on, in future essays, such as "What Is It to Believe Someone?", to offer more critiques of Hume's view of testimonial knowledge, and to say more suggestive things about how it is that we really come to our historical and testimonial beliefs (if not knowledge). But here it seems that she is arguing that the standard Humean foundationalist-evidentialist epistemology won't work, and also that coherentism won't work (at least if coherentism is supposed to imply that absolutely any belief we have can, in principle, be doubted and examined without threatening the whole structure of our web of beliefs). This is, at least for its time, a rather radical claim, since for a while it seemed that one must either be a foundationalist or a coherentist about epistemic justification. What is this third way that is needed to account for our knowledge of history (or justification of our beliefs regarding historical matters of fact)?
Ernie Sosa's "The Raft and the Pyramid" is credited with initiating virtue epistemology, and in that essay he proposes it as a third alternative to the "raft" of coherentism and the "pyramid" of foundationalism. I think of virtue epistemology as a form of foundationalism, though one rather unlike the evidentialist foundationalism of David Hume. My own inclination (and research program) is that a virtue epistemology, like that of Sosa's (or John Greco's), can be developed to answer these puzzles presented by Anscombe regarding testimonial knowledge, and, by extension, historical knowledge. There are some propositions that we believe on trust, and must believe on trust if we are to believe them at all. But trust can manifest intellectual virtue.