Nference to the best explanation: a common and effective form of archaeological reasoning

Citation: Fogelin, L. (2007) Nference to the best explanation: a common and effective form of archaeological reasoning. American antiquity, 603-625. (RSS)
Internet Archive Scholar (search for fulltext): Nference to the best explanation: a common and effective form of archaeological reasoning
Tagged:

Summary

Fogelin asks, how, considering the changes in theoretical approaches in archaeology over the decades, that archaeologists been able to still generate explanations about culture that seems right and have persisted through the paradigm shifts? The answer, Fogelin believes, is because the practice of post-processual and processual archaeology aren’t that different, that both utilize Inference to the Best Explanation (IBE) to generate their explanations. According to Fogelin. IBE has been standard practice, used implicitly, in archaeology for over a century. Even though new archaeology was based on deductive, empirical testing, deductive validity does not work well for archaeology. For example, the following is entirely logical in its deduction, however, it not a useful explanation:

All hominids have wings, H. ergaster is a hominid, therefore H. ergaster has wings.

Science then, is inductive as well, not in specific sense but in a wider sense, because deduction can’t determine if the premises of the hypothesis are good or true. Inductions, then, are always underdetermined by empirical evidence. As in the example above, they are empirical generalizations from a limited sample of past experiences. It should be noted that conclusions of an inductive argument are always under threat of the discovery of new evidence that could discredit them, and therefore can constantly be updated and are not edified.

Statistical induction is what most people consider when they think of inductive reasoning. Archaeology seldom has large enough sample sizes to do this, and additionally statistical induction has its own inherent problems - new evidence can disrepute previous calculations, and it requires a large set of past observations. IBE resolves some of these problems because it relies on multiple lines of evidence and can explain unique or infrequent archaeological phenomena, providing answers to both why and how questions. Kidder, Binford, Hodder, and Hegmon and Trevathan are all examples of IBE use in archaeological explanations.

IBE also resolves some of the problem inherent to the causal approach of explanation- at some point, the causal history will not be able to answer the “why” or will go so far back in the history of the "why" as to be irrelevant. Fogelin supports the use of constrastive explanations, which position two hypotheses against each other to explain why a phenomena occurred in one way, versus in another. For example, the question why a lithic debitage is concentrated in one area of a site rather thean across the entire site can result in the answer, that there was a lithic workshop, therefore no need to produce debitage across the entire site.

According to Fogelin, successful explanations have the following traits - empirical breadth (be able to explain multiple events), generality (applicable to wide variety of phenomena), modesty (generalizable, but not a law), refutability (testable, not edified), conservatism (shouldn’t throw out well-established explanations), simplicity (Occam's razor), multiplicity of foils (lenses through which we interpret/infer explanations).

Hanen and Kelly (1989) provide excellent case studies, mentioned in Fogelin's essay, that demonstrate the utility of the IBE method within archaeology. Haury's 1958 case study of Southern Arizona, for example, used migration hypothesis to explain anomalous feature shapes like L-shaped rooms and a D-shaped kiva, ceramics vessels painted in a northern tradition although made of local clays, archaeoobotanical materials believed to be restricted at the time to Northern Arizona, and additional clusters of evidence. Haury was able to infer from the multiple lines of evidence, that migration was the best explanation of that data. It is notable that he did not start the project assuming he would find evidence of migration, rather that hypothesis was induced after the fact to the explain the observed phenomena.

Another example is Kelley's 1979 work in El Salvador, in which a strange, partially buried feature was unearthed along the side of a temple. The existing evidence (depth of burial, content of feature, metal remains), was used to infer a hypothesis that the feature had a metallurgical function. Later radiocarbon dates ruled this out, determining the the feature dated to several centuries after the temple and was not contemporaneous. Nonetheless, multiple lines of evidence was used to induce that explanation, and the addition of new evidence like radiocarbon dates, demonstrates how new data can result in a reworking of inductively derived conclusion, as noted by Fogelin.

Theoretical and Practical Relevance

Reference

Hanen, M., & Kelley, J. (1989). Inference to the best explanation in archaeology. Critical Traditions in Contemporary Archaeology, Cambridge University Press, Cambridge