(Originally posted: 4/8/2011)
Recently, someone asked me what me dissertation research was about. That isn’t a particularly unusual question for any scholar, but the truth is my research was complex enough that I suspect many archaeologists wouldn’t understand it without actually looking at large chunks of the 400+ pages. Nevertheless, in trying to explain the questions I’d asked and the approach I’d taken, I also began thinking about some more intricate issues about how I and other archaeologists construct arguments. Follow through to the whole story for my thoughts.
What really caught my attention in the discussion was when the person asked if I’d used deductive or inductive reasoning. In fact, my dissertation was abductive, a reasoning process that moves back and forth between deductive and inductive logic. My argument was a particular kind of abductive argument that I’ve never found a formal description of, but which I’ve always called a “coherence argument.”
The major problem that confronts most archaeological arguments is one of epistemological warrant. To interpret and understand the archaeological record, we must begin with certain assumptions. Good assumptions coupled with good logic will teach us about the past. An argument that begins with bad assumptions must necessarily lead to misconceptions about the past. Therefore, in order to have any confidence in our conclusions, we must have reason to believe that our assumptions are current–that is, an epistemological warrant. Many archaeological arguments thus begin with long, torturous discussions of why particular assumptions are warranted, and even then the resulting argument is subject to a critic’s simply disagreeing. This is what one might call the “I just don’t buy it” criticism, and it’s one reason American archaeology is so theoretically fragmented today. If two people disagree on basic assumptions, there’s little chance of them agreeing on anything else.
On the other hand, good theories should give good assumptions. What, then, makes for a good theory? I would offer that a good theory is one whose assumptions are plausible (but not necessarily logically certain, a priori), which accurately predicts empirical observations, and which has explanatory power. Importantly, the converse is also true: a bad theory has implausible assumptions, does not predict observations, and cannot explain.
A coherence argument, then, is one that is built from a theoretical foundation with plausible asusmptions that are assumed (more or less) uncritically–that is, without a strong epistemological warrant. Those assumptions are then used to interpret empirical data. If the assumptions are bad, the interpretation of the data that they provide will not be coherent; it will make no sense. Conversely, if they are good assumptions, the interpretation will not only make sense, but will have real explanatory power. A coherent explanation of data does not prove assumptions are valid, but it does greatly increase the likelihood that they are.
Thus, by simply using a theory without any good reason to believe in it, one can test the theory. Also, the application of an otherwise unwarranted theory to data can explain the empirical data. A coherence argument therefore works from both sides of the equaton simultaneously: testing the theory at the same time it is using it to explain data.
Examples of this kind of argumentation are actually quite common in archaeology, but I will offer only two. First, consider Michael Ventris’ deciphering of Linear B in the 1950’s. He began with two largely unfounded assumptions: that the name of Knossos had not changed since the Bronze Age, and that Linear B was a method of writing Greek. Neither asumption could be justified in any rigorous sense, though both were plausible. Using them as a basis, he was then able to decipher Linear B, translating it into readable Greek that carried real meaning. Had the assumptions been incorrect, the likelihood of his finding meaningful content in the Linear B tablets was vanishingly small. Thus, he proved his assumptions were correct at the same time as revealing a wealth of data about Bronze Age Mycenean civilization.
A second example of this sort of reasoning also comes from Classical archaeology, but is much more recent. A few years ago, several astronomers published an article demonstrating that astronomical clues in Homer’s Odyssey pinpointed the day of Odysseus’ Slaughter of the Suitors to April 16, 1178 B.C. They began with a plausible, but by no means universally accepted, assumption that a passage in Book 22 refers to a solar eclipse visible from Ithaca on that date. Then, using that assumption, they calculated other astronomical phenomena that would have been visible in the skies over the Mediterranean on various dates just prior to April 16. These astronomical phenomena neatly explained a variety passages elsewhere in Homer’s text. Their conclusion was twofold: first, that the Book 22 passage did indeed refer to the April 16, 1178 B.C. eclipse, and second that the other passages referred to similar astronomical phenomena. They confirmed the assumption while also illuminating the text.
This kind of argument is very useful in archaeology. It is a way to circumvent the “I just don’t buy it” criticism. Had either group of researchers had to justify their assumptions with strong epistemological warrants, neither case could every have been made. Instead, by simply pushing ahead and not worrying about proving their case beforehand, they helped advance knowledge. As I said, this kind of argument is actually rather common in archaeology, though seldom acknowledged as such. Scholars are wary of leaving themselves open to criticism, or possibly wasting valuable research time only to discover that their assumptions were in fact wrong. On the other hand, for many kinds of archaeological investigation, time is the only thing that might be lost, and withstanding criticism is part of the job. I think coherence arguments, consciously formulated and carefully applied, are a strong way to build archaeological knowledge.