If your theory explains the evidence does that mean it is probably correct? If “everything fits”, is your theory therefore surely right?
There’s a problem with that way of thinking and it is taken head-on by Paul Newall in a chapter titled “Logical Fallacies of Historians” in Tucker and Kane’s A Companion to the Philosophy of History and Historiography.
Affirming the consequent is a formal fallacy occurring when one or more potential premises are omitted from an argument. For example, “if my rain dance worked then it should be raining; it is raining; therefore, my rain dance worked.” Here other possible causes of the rainfall are left out and the argument fails. Affirming the consequent has the general form:
P1: If A then B;
C: Therefore, A.
Some of you may know of a 1970 “classic” by David Hackett Fischer, Historians’ Fallacies: Toward a Logic of Historical Thought. Newall acknowledges the strength of Fischer’s work but in a single chapter is addressing what he sees as the most common and unfortunately enduring fallacies among historians.
So how can one guard against being overcome by this fallacy? Newall continues:
However, we can add missing premises to show the error:
P1a: If C then B;
P1b: If D then B;
. . . and so on. For an argument of this form to not be fallacious, we would need an additional premise stating that A is the only possible cause of B. (Note that affirming the antecedent (or modus ponens) is not fallacious; that is, arguing “if A then B; A; therefore, B.”)
This fallacy is of considerable importance to historiography because it often forms the basis of models of confirmation and shows why a more sophisticated philosophical apparatus is required. A typical statement might be: “if our model/explanation of x [some event, say] is correct, we would expect to find y [some evidence, records or traces, for instance], as indeed we do.” Thus an account of some historical episode predicts that certain evidence of it will be found, and when this happens the account is confirmed or demonstrated to be correct. However, in this form the account is straightforwardly fallacious as an example of affirming the consequent.
For example, if Einstein were an extraterrestrial of superior intelligence, he should have made incredible intellectual achievements. He did, and therefore he was an extraterrestrial. We can attempt to undercut this objection by arguing that actually the evidence only makes the account more likely, but this requires further elaboration before it becomes philosophically tenable, for example by comparing different hypotheses with different prior probabilities (see Tucker 2004, for an example of a Bayesian approach that avoids this difficulty).
Bayes? Yes, historians use Bayes.
I’ve been trying to think through some other examples. Here are a few. Two political ones to start followed by some biblical ones, including some mythicist examples. Continue reading “Logical Fallacies of Historians: “If It Fits — Be Careful!””