Internet Archaeol 15. Clive Orton. Models of what?

3. Models of what?

Since modelling may involve a considerable investment of effort, it is worth examining which aspects of archaeology are likely to benefit from the simplification that a model is intended to bring. A broad division can be made into models of the archaeological record and models of archaeologists' thought processes. The former covers a wide range of needs. At one end of the spectrum, we have simplifications (or emphases on some particular aspects) of a physical entity, such as the Harris matrix representation of archaeological stratigraphy or the Thiessen polygon structure imposed on a settlement pattern (see above). At the other end, we have models that are claimed to represent a whole system or sub-system of human activity, such as Thomas' model of Shoshonean subsistence and settlement patterns (Thomas 1972), or Mithen's models of hunter-gatherer foraging (Mithen 1990). Generally speaking, the tightly-focussed and more restricted models seem to have been the more useful (Lake 2001, 727). Between them come a wide range of specific events and processes that might usefully be modelled, for example, the inputs and outputs of a pottery production process in order to examine its working practices (see below).

A particularly useful example of the former category is the modelling of site formation and taphonomic processes in the attempt to be able to draw archaeologically reliable conclusions from partial and biased data (Shott 1998; Orton 2000, 40-66).

The latter (models of archaeological thought processes) often arise from the question 'what are we actually doing when we...?' A good example is the process of classification of artefacts — the division of a coherent body of material into sub-groups or 'types', possibly with some ideas about the relationships between the types so defined. This might be seen as just 'something that archaeologists do', an intuitive exercise born out of training and experience, tempered by the recognition that there may be different 'types of types': chronological, functional, stylistic, for example. However, considerable insight may be gained by modelling the process as an exercise in multi-dimensional geometry, as the partitioning of an n-dimensional space in which each dimension represents a chosen attribute or characteristic. The intuitive archaeological processes can be modelled by a series of formal decisions — choice of variables, definition of a metric (the 'distance' between points), choice of partitioning strategy — which may help to explain the differences between different approaches, and to guide us towards the assistance of an appropriate computer technique (Doran and Hodson 1975, 159-86). Questions such as 'do I need to transform my data?' and 'which version of cluster analysis should I use?' become easier to answer within this broad framework. Kendall's analysis of Petrie's discovery of the technique of seriation (Kendall 1971) can also be put into this category — the archaeological aim of putting grave assemblages into chronological order led to an implicit mathematical model of the processes involved, which in turn led to the development of a practical technique.

The most inclusive model at this end of the spectrum is probably the 'Research Cycle' (Orton 2000, 10), a model of a way in which quantitative research is undertaken, linking theory, models and data in a cycle of model-building, data collection, statistical analysis and archaeological interpretation.