Anyone exposed to the extended epistemologi-cal discussions that took place among archaeologists and philosophers of science in the 1970s, can appreciate the importance of understanding the underlying epistemological foundation of human behavioral ecology and foraging theory and how these approaches can and cannot be applied in archaeological contexts. The danger would lie in mistakenly assuming that foraging theory and behavioral ecology take any sort of strongly deductive or covering law explanatory form. The potential for mistaking human behavioral ecology as essentially positivist and deductive in logical structure is partially due to the occasional (mis)use by foraging theorists of terms such as "covering arguments," "hypothetico-deductive," deductive," and "deduce," which in formal logical contexts clearly indicate covering-law forms of inference; e.g. "HBE is an anomaly within sociocultural anthropology due to its hypothetico-deductive research strategy (Winterhalder and Smith 2000, 52; Optimal foraging theory can be classed with "analyses that begin with selection and that attempt to deduce its consequences for behavior in specified environmental circumstances" (Winterhalder and Goland 1997, 125).1
The possibility that foraging theory might be considered as qualifying for general covering law status, however, is not immediately evident, given that foraging models are "internalist" in that the causal focus is on humans making individual decisions: foraging theory is scaled at the level of individual or small-group intra-genera-tional decision making regarding changes in resource selection. In addition, the particular details of resource access, the situational specifics that provide the proximate context for such decisions, are recognized as central to understanding and explaining the choices that are made. Within this internalist, small-scale, intra-generational context of decision making, shaped and structured by the particular immediate composition of available resources, however, individual human choice mistakenly might be assumed to be dictated by overarching rules—as being derived from higher level biological principles of selection. In terms of the logical structure of foraging theory, then, particular individual events or decisions regarding resource selection might be "explained" in that they logically follow—can be deduced—from a general lawlike or universal principle that universally holds. The temptation would be to believe that predictive power and covering law explanatory strength could be derived from such overarching general laws of selection. It would then be these general or covering law principles which would determine the specific decisions that would be made, once the proximate variables regarding the availability and costs of available resources were factored in.
Something "loosely akin" (Salmon 1967, 18) to this general hypothetico-deductive method of inference provides the essential epistemological core for much of the research carried out in the sciences. There is no question that within the realms of physics, chemistry, biochemistry, etc., faith in, and employment of, overarching principles of varying reach are justified. But is such an epistemology appropriate for consideration of more complex biological and cultural behavior sets? Are there general laws or principles from which human decisions in any particular situation can be derived or predicted? By the mid-1970s archaeologists and philosophers of science had agreed that covering law models and hypothetico-deductive methods of confirmation were not appropriate for archaeological inference (Salmon 1975, 1976; W. Salmon 1967, 1973; Smith 1977, 1978; Wylie 1992).
Importantly, Winterhalder (2002a) clearly defines both the logical structure of foraging theory and the constraints inherent in archaeological inference in his discussion of a four-fold (Cells A-D) division of models according to their appropriateness of application in different scientific disciplines, based on the quality and quantity of data available (y-axis) and the degree of understanding in conceptual or theoretical terms (x-axis) that exists in different subject areas. Cells B and C, which contain many of the different kinds of models that can be employed in studies of human behavior or the historical and evolutionary sciences, differ from each other in significant respects. Cells B and C in Winterhalder's presentation are both constrained by limited data. Conceptual or theoretical understanding, however, is greater in Cell C than in Cell B, resulting in distinct differences between the two cells in terms of how models can be used, and how inference is constrained and structured. While certainly loosely drawn, this basic distinction between Cell B and Cell C—Cell C having a higher level of theoretical understanding—would seem to roughly correlate well with how confidently particular events can be predicted, derived, or deduced from overarching general principles; i.e. in Cell C they can, in Cell B they can't.
In concurrence with the consensus conclusion reached by interested philosophers of science, logisticians, and archaeologists a generation ago, Winterhalder places evolutionary archaeology squarely in Cell B, where it is appropriately paired with arguments by analogy ("analogic"—Winterhalder 2002a, 206) as the correct method of inference, and well outside the reach of Cell C deductive covering-law models. This does not mean that human behavioral ecology and foraging theory, even if they were proposed as qualifying for Cell C covering law hypo-thetico-deductive status, could not be employed in archaeological analysis and interpretation, only that they should be characterized and used under the same logical constraints as other Cell B models and explanatory frameworks.
The most obvious B Cell constraint or limitation to be imposed on foraging theory would of course be to sever any perceived explanatory or deductive, predictive link between an overarching principle of selection and the specific subsistence choice faced within a particular context of resource availability. Foraging theory, like other theoretical approaches and perspectives, are a rich source of situation-specific hypotheses regarding resource selection. Once formulated, however, such individual, situated hypotheses would derive no strength, no power, no advantage over other alternative hypotheses, from the foraging theory from which they were generated. The only appropriate measure of the relative worth of alternative competing hypotheses, including those based on foraging theory, is how well they account for and are supported by currently available archaeological information, and the value and depth of plausibility considerations. Their status is achieved, not ascribed (Wylie 1992).
Winterhalder (2002a) also provides a description of the general "exploratory" and "evaluative" role of models in Cell B subject matter areas, and in so doing provides clear guidelines and reasonable expectations regarding the appropriate application of foraging theory to transformational questions documented in the archaeological record, including the process of domestication and agricultural emergence:
With limited data and understanding (Cell B), models necessarily are speculative and provisional. The main activity is their evaluation. Here models help to define and isolate problems. They facilitate preliminary analyses. Evolutionary Archaeology and some life history and behavioral ecology models sit here. As theoretical understanding increases Cell C models can be used in an interpretive or inferential fashion. (Winterhalder 2002a, 208).
The evaluative [Cell B] mode is focused on testing the model against observations. Evaluative statements commonly take the form: If the assumptions of the model accurately reflect the referent situation, then we expect the following observations. Here [Cell B] models provide provisional hypotheses and the analytical work lies in testing those hypotheses . . . The model is made subordinate to the empirical evidence (Winterhalder 2002a, 209). [bracket text added]
The phrase—"the model is made subordinate to the empirical evidence"—reiterates the key point that in Cell B subject matter situations (which encompasses evolutionary archaeology and domestication), any hypotheses regarding resource choice which are generated from foraging theory expectations and principles carry no predictive or explanatory weight with them. Whatever value they may have relative to other alternative hypotheses can only be measured in how well they account for the available relevant data. As Winterhalder points out (2002a), the seductive danger for anyone working in Evaluative Cell B situations with models characterized as also having Interpretive C Cell capacities is deluding oneself into the false belief that in fact one is working in Cell C Interpretive mode, where covering laws hold and specific hypotheses derive predictive weight and explanatory power from their mother ship covering laws:
An analyst willing to pursue evaluation is seldom so disinterested as to eschew completely extrapolating to interpretive uses . . . as a consequence it can be difficult to discern the relative balance of evaluation and interpretation in a particular application. Judgements are further complicated if authors are not explicit. It is tempting for the advocates of a model to slip unintentionally from evaluative to interpretive modes with few discernable signals that they have done so. (Winterhalder 2002a, 2°9)-
Winterhalder goes on to describe the evaluative mode as grading into the Cell C interpretive mode, and indicates that it is possible to offer "a plausible covering argument that the model applies to a situation" (2002a, 209), and this argument of plausibility "rather than by direct test" enables one to explain and interpret observations, and to further extend or generalize from them (Winterhalder 2002a, 209):
[Cell C] statements take the form: because of its theoretical generality for social foragers, the producer-scrounger game likely characterized hominid hunter-gatherers, with the consequences x, y, and z for subsistence and social behavior. Interpretive use implies a model that is only weakly validated in either the theoretical or the empirical sense in that situation [Bracket text added.]
How does one decide if a model has a high enough plausibility rating to qualify for Cell C interpretive mode applications? Winterhalder (2002a, 209) suggests that "willingness to engage in the interpretive use of a model depends on" three criteria:
a. general confidence that the model captures the relationship it claims to represent; that is on prior testing in related contexts;
b. on the covering argument for suitability in this context;
c. on inability to make a more direct evaluative use of the model in this context [e.g. in the historical sciences such as archaeology] (Winterhalder 2002a, 209).
Given the use of modifiers such as "plausible covering arguments" and " its theoretical generality . . . likely characterized" (italics added), Winterhalder clearly is not suggesting these three criteria provide a justification on logical grounds for the elevation of any theory or model to "covering law" or "hypothetico-deductive" status, where a previously demonstrated universal application of the overarching principle substitutes for, and reduces the value of, any actual empirical evidence or testing.
Interestingly, the three criteria listed above closely conform to what takes place during plausibility consideration of hypotheses—an initial step in various supplemented hypothetico-deductive methods of inductive confirmation which are generally considered appropriate for archaeological inference (Salmon 1976; W. Salmon 1967, 1973; Smith 1977, 1978). When considered within the context of such methods of inductive confirmation, the logical outlines and limitations for useful application of foraging theory in archaeology can be outlined:
1. Covering Law and Hypothetico-deductive frameworks of inference are not appropriate for foraging theory, which can only reasonably be employed within an inductive format of confirmation;
2. Foraging theory potentially can be a rich and valuable source of situated small-scale hypotheses regarding domestication and other resource-selection decisions along the food production developmental trajectory;
3. Specific hypotheses generated on the basis of foraging theory principles or expectations derive no predictive power or explanatory status from their source;
4. Once formulated, hypotheses are subjected to plausibility considerations—their prior probability is taken into account as an initial screening of their plausibility: "such plausibility considerations are logically separate from and prior to the actual testing of a hypothesis" (Smith 1977, 605). A number of authors have addressed the need for explicit and careful attribute class and reference class selection during plausibility consideration (Ascher 1961; Salmon 1967, 90-91; Salmon 1975, 461; Smith 1977);
5. Plausibility considerations in archaeological reasoning invariably take the form of argument by analogy, which calls for an explicit and detailed determination of the prior probability of all alternative proposed hypotheses on the basis of seven nonquantitative criteria (see Copi 1972, 358-362; Salmon 1975; Smith 1977):
1. The number of situations shown to share the attributes in question.
2. The dissimilarity of the situations shown to share the attributes. This is what M. Salmon is referring to when she discusses employing ethnographic cases from dissimilar subgroups of the reference class
3. The number of shared attributes.
4. The number of inferred attributes.
5. The significance of the shared attributes.
6. The specificity of inferred attributes
7. The number of points of difference between situations.
(Smith 1977, 608)
The prior probability rating of a foraging theory generated hypothesis would thus be considered within the framework of argument by analogy and would involve explicit presentation and assessment of well documented particular past situations of a similar kind (see Winterhalder's criteria 1 and 2 above) where empirical evidence supported the hypothesis;
6. Plausibility consideration in archaeological reasoning is followed by the formulation of a set of alternative hypotheses of non-negligible prior probability (multiple working hypotheses; Chamberlain 1965), each having a set of observational predictions. Hypotheses generated from foraging theory should not, then, be viewed in isolation, as the only potential solutions under consideration.
7. The relative strength of alternative competing hypotheses would then be established through empirical testing. This would involve developing sets of test implications or observational predictions for each, along with bridging arguments and, when needed, auxiliary hypotheses. Determining which hypothesis is best supported by the available archaeological evidence would then be based on the significance, number, and variety of observational predictions shown to be empirically true and false.
When considered within this larger context of the general consensus logical structure of archaeological inference, the central epistemologi-cal challenge facing foraging theory comes into clear focus. Given the inherent complexity and limited strength of archaeological arguments, researchers interested in employing foraging theory, as Winterhalder (2002a) points out, will constantly be confronted by the seductive lure of simply accepting, a priori, the universal applicability of foraging theory principles, rules, theorems, etc., and applying them in an interpretive, explanatory mode in specific transition to food production case study situations in lieu of any requirement for empirical data-based testing and confirmation. This dilemma facing foraging theory applications in terms of universal applicability versus a reliance on empirical testing and confirmation is well illustrated in the ongoing discussions regarding the economic role of acorns during the Natufian lead-up to initial plant domestication in the Near East (Barlow and Heck 2002). Based on the general expectations of foraging theory, acorns should have been a significant component in Natufian diets, given their energetic benefits and costs. But representation of acorns in Natufian archaeobotan-ical assemblages falls far short of foraging theory predictions. On one side, scholars adhering to a perspective of strict empirical observation might argue that given their relative absence in the archaeological record, acorns played at most a minor role in Natufian economies. Scholars strongly convinced of the predictive power of foraging theory could argue that the general model is more likely to reflect past reality than what is present in the archaeological record, and that acorns must therefore have been an important food source, irrespective of archaeological evidence to the contrary. The important contribution that foraging theory can make in such situations, and its appropriate epistemological application, is to show that theory and available data are at odds, while opening potentially rewarding new avenues of inquiry leading to a better understanding of the procurement, processing, and dietary significance of acorns in Natufian economies and possible reasons for their absence either from Natufian settlements or from Natufian diets.
Was this article helpful?