P. Jeffrey Brantingham has a new paper in the June 2006 issue of Current Anthropology in which he proposes a formal model to estimate forager mobility based on lithic raw material incidence in an assemblage. The paper is based largely on the development and testing of a model using a stochastic process called a Lévy random walk, followed by a short section which compares the expectations of the model to the Early Upper Paleolithic archaeological record from southwest France. Brantingham's theoretical perspective is grounded in human behavioral ecology and the approach he employs here derives from various the forms of optimality models that have become part and parcel of this field of study over the past quarter century (Winterhalder and Smith 2000). The math involved in building the model is rather involved but nonetheless imminently understandable, and Brantingham is explicit about the assumptions that underlie the model. In essence, he is trying to hold everything else constant in order to see how mobility alone influences raw material composition in an assemblages, so as to subsequently permit the utilization of that variable as a way to infer prehistoric mobility
I'm generally not too big a fan of the CA comment format because I think it lends itself too easily to veiled personal attacks and forces an unnecessarily polemical tone into scholarly discussion. However, here, it works quite well, with the commentators adopting a level-headed tone and providing constructive criticism on the paper. As a whole, they reasonably highlight the limitations that an archaeological record that is by nature incomplete and a time-averaged reflection of human behavior may have on the applicability of aspects of the model, problems which Brantingham acknowledges but leaves for future studies, a very reasonable position given the complexity of this first step in the process.
In my opinion, this is exactly the kind of work that needs to be done in hunter-gatherer archaeology, in order to move the filed forward, because it generates testable expectations that can be compared to pattern in the archaeological record. This does two things: First, it allows the identification of broad principles according to which specific facets of human behavior are (or can be) structured. Second, any discrepancies between expected and observed patterns (which are almost unavoidable) then lend themselves to further explorations and discussion, thus highlighting the virtually unlimited range of case-specific adaptive strategies humans and their ancestors devised to maximize their fitness and well-being in the past.
That said, I do have some minor reservations on some of the assumptions made by Brantingham's model. For one thing, while it may be necessary to assume this for conceptual simplicity, it is unreasonable to consider lithics simply as units of raw material. Of course, on one level, that is exactly what they are, so some might say this is warranted. However, humans have no inherent need for rock. What they have need for are the tools (broadly defined here) that are made from rock. As such, different types of lithics have different functions and different kinds of use-lives which may vary in length and nature. This argument complement those of commentator Michael Shott, although one can see how this facet of forager life could rather simply be incorporated in future refinements of Brantingham's model. My second reservation may betray my unfamiliarity with formal modeling or my misunderstanding of a specific aspect of Brantingham's model, but while I understand how he uses a Lévy random walk to generate distance-from-source distance rates, I'm not sure how these correspond to human-scale features. For instance, while I understand that lithic discard (and acquisition) may be constant in time and space, archaeologists and ethnographers mainly rely on 'sites' as the unit of analysis on which an assemblage is tied to a archaeological or ethnographic landscape. While the Lévy random walk generates patterns of distance-to-source discard rates, I'm unsure as to how these relate to sites writ large. In other words, does a Lévy random walk generate points on the landscape where more than single units of raw material are discarded or is each discard point completely independent? If the latter is true, then how do we reconcile Brantingham's model's observations about the expected patterns of raw material discard to the human reality where specific spots are the locus for multiple discard instances of raw material units? Again, this can probably be incorporated in the formal model rather easily, but it did make me ponder the one-to-one comparability of the model's patterns to what we document archaeologically.
Now, these interrogations aren't meant to detract from the value of Brantingham's paper. In fact, I think that it's exemplary in terms of how it links theory to methods, clearly explains the methods, and then compares expectations to archaeological data. Brantingham cites a number of other studies that are adopting formal modeling as a central methodological tool, so it looks as though the days of 'common-sense' approaches to defining expectations about the archaeological record may thankfully be drawing to a close, to be replaced by more rigorous research perspectives that lend themselves to direct testing.
Brantingham, P. J. 2003. A neutral model of stone raw material procurement. American Antiquity 68:487–509.
Brantingham, P. J. 2006. Measuring forager mobility. Current Anthropology 47:435-459.
Winterhalder, B., and E. A. Smith. 2000. Analyzing adaptive strategies: human behavioral ecology at twenty-five. Evolutionary Anthropology 9:51-72.