Archaeological Inference and Inductive Confirmation

From AcaWiki
Jump to: navigation, search

Citation: Smith, B. D. (1977) Archaeological Inference and Inductive Confirmation. American Anthropologist 79(1977) (RSS)
Internet Archive Scholar (search for fulltext): Archaeological Inference and Inductive Confirmation
Tagged:

Summary

In this article, Smith discusses how the paradigm shift at the time of the New Archaeology motivated archaeologists towards philosophy of science, in attempts to resolve theoretical problems related to creating a scientific archaeology. Binford was one of the early proponents of this, arguing for the use of a deductive model of explanation that could be utilized by processual archaeologists to conduct empirical and logical hypothesis testing. It should be noted that Binford was not particularly in support of an hypothetico-deductive model in particular, rather was in support of the use of a model in general.

Because of the nature of archaeological research, it is impossible to use entirely deductive methods. Poor preservation, small sample sizes of past observations, and poor recovery rates make it difficult to explicitly uncover the logical, deductive steps between our hypotheses and the observed evidence. Smith believes that archaeologists have actually been using a more inductive method, largely because of the nature of our data and also because it is a key component of everyday reasoning.

Smith goes on to describe the hypothetico-analog model (HA), essentially a precursor of the Inference to Best Explanation model. The key feature of this model is that multiple hypotheses are tested against to determine which best explains the multiple lines of evidence. Hypotheses are determined to be sound by plausibility considerations, which state that a hypothesis is plausible if there is significant prior probability of its occurrence. Hypothesis are then tested against reference classes (essentially ethnographic analogies to determine the prior probability), which is determined by using observational prediction to choose the reference class.