Computational Approaches to Archaeological Spaces

Andrew Bevan, Mark Lake, Andrew Bevan, Mark Lake

• Publisher: Routledge
• Year: 2016
• Print ISBN:
  9781611323467
• eBook ISBN:
  9781315431918
• Topic: Social Sciences
• Subtopic: Archaeology
• Index: Social Sciences

This volume of original chapters written by experts in the field offers a snapshot of how historical built spaces, past cultural landscapes, and archaeological distributions are currently being explored through computational social science. It focuses on the continuing importance of spatial and spatio-temporal pattern recognition in the archaeological record, considers more wholly model-based approaches that fix ideas and build theory, and addresses those applications where situated human experience and perception are a core interest. Reflecting the changes in computational technology over the past decade, the authors bring in examples from historic and prehistoric sites in Europe, Asia, and the Americas to demonstrate the variety of applications available to the contemporary researcher.

1 Introduction: Archaeological Inferences and Computational Spaces

Andrew Bevan and Mark Lake

Spatial analysis in archaeology today encompasses a wide range of experiential, fieldwork-based, deterministic, and/or stochastic approaches that vary considerably in their intended purpose and theoretical underpinnings. Although Hodder and Orton (1976) and Clarke (1977) both provided book-length treatment of the application of statistical and/or simulation approaches to spatial analysis in archaeology many years ago, the rapid uptake of computational methods such as Geographical Information Systems (GIS) and related methods in archaeology from the late 1980s and early 1990s marks a disciplinary step change, for enthusiasts and critics alike, increasing by an order of magnitude the quantity of spatial data that could be managed and analyzed, especially for those working at the scale of entire archaeological landscapes. The 1990s saw a string of methodologically focused edited volumes covering the development of archaeological GIS (Aldenderfer and Maschner, 1996; Allen, Green, and Zubrow, 1990; Lock and Stančič, 1995; Maschner, 1996). The most recent in this vein is now over 10 years old (Lock, 2000), perhaps partly reflecting the more mainstream adoption of standard GIS techniques, as also evidenced by the appearance of textbooks on archaeological GIS (Chapman, 2006; Conolly and Lake, 2006; Wheatley and Gillings, 2002) and recommendations for archaeological predictive modeling (Kamermans, van Leusen, and Verhagen, 2009; Mehrer and Wescott, 2005; Wescott and Brandon, 2000).

Since 2000 a major focus of methodological development in GIS has been integration with computer simulation, particularly agent-based modeling; the papers in Gimblett (2002) provide heritage management applications of this technology, while more research-oriented examples can be found in Kohler and Gumerman (2000). More generally, however, the last decade has seen something of a dichotomy in the presentation of new computational techniques as either forming part of something labeled Digital Archaeology (for example, Evans and Daly, 2006; more widely, the Digital Humanities, Bodenhamer, Corrigan, and Harris, 2010) or as the handmaiden of more explicitly model-based (Kohler and van der Leeuw, 2007) or spatial analytical (Robertson et al., 2006) approaches to archaeological inference (see also McCoy and Ladfoged, 2009). Gary Lock (2003) convincingly argued that the development of computer applications in archaeology has always reflected the entanglement of both methodological and wider theoretical advances, as documented in detail by others for the specific examples of GIS-based visibility analysis (Lake and Woodman, 2003) and computer simulation (several papers in Costopoulos and Lake, 2010). We are thus very conscious that any collection of computational approaches to archaeology will likewise be of the moment, so it is worth elaborating what makes this a particularly exciting time to be presenting such a collection.

Two broad sets of developments in the computational and social sciences over the last few years have conspired to create something of a research watershed and topical point of departure for a volume of this kind. The first set combines (1) some practical thresholds now reached in an otherwise continuous trajectory of growth in modern computing, (2) the sharply increased amounts of digital data now available, and (3) the wider climate of more open access to both data and software source code. A popular metric of the modern age has been the near exponential upward trend in computational power over the last six decades (so-called Moore's Law). However, there are good reasons to point to the last 10 years as unusually important, particularly with respect to the demands of research within the social sciences and humanities. On the one hand, laptop and desktop computers have reached a stage of development where they can be used for the vast bulk of programming, statistical analysis, modeling, and representational tasks in these disciplines. This is the hardware side of a wider democratization process in computational archaeology. On the other hand, high-performance computing, often leveraging clusters of otherwise idle processors rather than dedicated supercomputers, is enabling (1) far more intensive forms of Monte Carlo simulation with carefully constrained random samples and explicit hypothesis-testing, as well as (2) far higher-fidelity renderings of visually sophisticated spaces than has hitherto been possible (in this volume, see Bevan et al.. Chapter 2; Lake and Ortega, Chapter 9; Earl et al., Chapter 11).

Alongside this added capacity has also come a flood of digital information. Perhaps the most glamorous agents of this information revolution have been the automated or semiautomated collection routines enabled by Global Positioning Systems (GPS), mobile phones, social networking sites, and space-borne remote sensing instruments. Such datasets are often referred to as "big data" given that, at their most extreme, they can be large enough to demand innovative storage, retrieval, and analysis practices (though rarely so daunting in archaeology), but, in addition, there have also been major efforts at digitizing archival or legacy datasets, as well as a steady trickle of manually inputted information (for example, via citizen-led projects such as Wikipedia, Google Earth Community, and, in archaeology, the UK Portable Antiquities Scheme—such as Bevan, 2012a).

Within archaeology, some datasets offer richly structured formats and metadata in order to facilitate close semantic and technical interoperability with so-called Web 2.0 technologies, whereas others are archived in simple formats for direct download from institutional or national data repositories, and a third unknown, but most likely numerically dominant, group of datasets continues to float in a netherworld of eccentric colleague-to-colleague gifting. Even so, archaeology is increasingly in step with a wider social, political, and economic climate that emphasizes open approaches to both software development and data dissemination (Bevan, 2012b; Ducke, this volume; Lake, 2012) that stress transparency, universal access, and the promise of unanticipated onward use. Increasing instances of the tailoring of software applications to archaeological purposes is one outcome of this wider climate (for an early emphasis, see Lake, Woodman, and Mithen, 1998), and many of the papers in this volume reflect such efforts at customization.

The second set of developments that arguably make a volume of this kind a timely venture relate to the increasing maturity with which researchers have, over the last 10 years or so, been discussing some key trade-offs in analytical and interpretative approach: for example, with respect to the importance of realism versus abstraction, the primacy of bottom-up versus top-down processes, and the relative heuristic value of lumping or splitting human behavior. In this regard, and as Mark Lake and Damon Ortega argue in a later chapter, there has been a tendency for the subdiscipline of computational spatial approaches in archaeology to recapitulate broader disciplinary rifts. A good example is the dichotomy between space and place that pervades not only archaeology but also academic research more generally. Contrary to the traditional view of this as a confrontation between tribalized cultures and a few culture heroes (mirrored in the often stereotyped priorities of processual versus post-processual archaeology), we would argue that many researchers now consider more quantitative and more experiential approaches—for instance, as valid alternatives that together generate a useful dialectic. Indeed, the strong overlaps between, for example, complexity science and more avowedly humanistic approaches to space are visible in common interests with respect to historical contingency and path dependence or to behavioral emergence and structuration (see Portugali, 2006). Where there is diversity of approach and opinion moreover, we would suggest that this diversity should be thought of less as a set of generational pendulum swings or paradigm shifts and increasingly as desirable to-and-fro to encourage within a research institution, over the course of an individual research career or indeed within the same edited volume.

It should be no surprise therefore that contemporary practicalities and cross-cutting methods are salient features of the chapters that follow, but they manifest themselves slightly differently depending on different researchers' analytical and interpretative agendas. Certain differences of intention persist and suggest a structuring framework for this volume that (1) begins with a focus on the continuing importance of spatial and spatiotemporal pattern recognition in the archaeological record, (2) continues by considering more wholly model-based approaches that fix ideas and build theory, and then (3) goes on to address those applications where situated human experience and perception are a core interest.

1. Spatial Analysis

With these three emphases—loosely speaking, spatial analysis, spatial modeling, and spatial experience—in mind, the authors of the first three chapters in this volume all have a common interest in more inductive spatial approaches to archaeological inference. Together they take us across a sweep of different kinds of pattern recognition: moving from the spatial analysis of data recovered from primary fieldwork (for example, via case studies spanning a major research excavation, a field survey, and joint assessment of records from rescue archaeology; Bevan and associates) to extraction of archaeological features from ground-based or airborne remote-sensing imagery (Kvamme), to consideration of the collected records from museum archives and ethnographic observation (Clark).

Bevan, Crema, Li, and Palmisano (Chapter 2) return to a long-cherished tool of empirical archaeology, the distribution map, and argue that a range of new methods and theoretical perspectives should encourage renewed efforts to characterize such spatial patterns and the underlying generative processes behind them, even if we will often still be left with multiple competing explanations at the end of this effort. Their discussion ranges from patterns exhibited by typological subgroups of bronze weapons in the tomb complex of China's First Emperor, to early Iron Age settlements in the uplands of the West Bank, to the pithouses of Japan's early complex hunter-gatherers. Throughout, they stress the importance of grappling with an archaeological record that is chronologically fuzzy and spatially uneven.

Kvamme (Chapter 3) offers a focus on two aspects of spatial and computational research that are common in archaeology but otherwise rare in this volume: the demands of cultural resource management (CRM) and the digital riches offered by remote-sensing imagery. The challenge he articulates for the archaeological community is to come up with automated or semiautomated pattern recognition techniques that can leverage the vast amounts of remote-sensing data that are now available to provide cost-effective CRM solutions across the large swaths of the planet currently under threat from development and destruction. The challenge is therefore one of seeking to upscale the techniques that would otherwise provide nice vignettes at the level of individual sites to ones that can be used routinely or can have a more global reach (see also Ducke, Score, and Reeves, 2011; Menze and Ur, 2012).

Clark (Chapter 4) argues for a more joined-up approach to multivariate analysis of large cross-cultural datasets. Drawing in particular on analytical advances in ecology, he suggests that we can squeeze far more out of complex archaeological datasets (1) by collapsing such data into distance matrices expressing dissimilarities among different sample sites (for instance, taken from different settlements or different perceived culture groups) via one or more ordination techniques, (2) comparing them with patterns of geographic or social proximity, and (3) conducting careful outlier analysis. His case study using an ethnographic example from the Pacific coast of North America highlights how evidence as seemingly distinct as social structure, material technology, subsistence practices, and ordinary distance can be brought together to offer useful analytical traction on one another.

The level of simplification and abstraction advocated in these three chapters is often considerable. Bevan and colleagues choose to ignore the complex 2D or 3D morphologies and alignments of their artifacts, houses, and settlements in favor of pure point locations. Clark chooses to collapse the complex relationships that exist among a host of different cultural variables into a series of distance matrices. Both chapters adopt relatively simple, Euclidean representations of geographic distances and social proximities that are in reality far more complex. Kvamme argues for the use of similarly idealized geometric types, such as the circle, as benchmarks for comparison with the features that appear in the raster grids produced by remote sensing. In all three cases, these trade-offs are deliberate and ones that are meant to be effective only at certain spatial scales and not at others (see also Lock and Molyneaux, 2006).

2. Spatial Modeling

In partial contrast to these first three chapters, the subsequent four by Rivers and colleagues, Premo, Barton, and Herzog can all be characterized as more deductive and model-based in that they start with explicit theories or more informal understandings of human behavior and then seek to establish how these might play out under specific circumstances that did or could have obtained in the past. It is a measure of the diversity within this broad approach, however, that the conceptualization of space in these four chapters ranges from the more abstract topological model favored by Rivers, Knappett, and Evans to the geographically referenced, continuous field, elevation models used by Barton and Herzog. Another axis on which these chapters vary is the extent to which they focus on individuals or groups as the unit of analysis.

Rivers, Knappett, and Evans (Chapter 5) offer a graph theoretic model of the evolution of trade and exchange links in the Bronze Age Aegean. They formalize the spatial relationships assumed to be in operation between Bronze Age sites as an evolving set of weighted networks and in so doing provide important discussion of the sense in which network models can incorporate properties of the actual geographical space in which the network is ultimately embedded. Their exploration of the different insights provided by looking at the most likely configuration of links and nodes, on the one hand, versus the most efficient one, on the other, brings ...

Table of contents

• Cover
• Half Title
• Title
• Copyright
• Contents
• List of Illustrations
• Acknowledgments
• 1. Introduction
• 2. Intensities, Interactions, and Uncertainties: Some New Approaches to Archaeological Distributions
• 3. An Examination of Automated Archaeological Feature Recognition in Remotely Sensed Imagery
• 4. An Introduction to Integrative Distance Analysis (IDA)
• 5. Network Models and Archaeological Spaces
• 6. Multilevel Selection and the Evolution of Food Sharing in Fragmented Environments: A Spatially Explicit Model and Its Implications for Early Stone Age Behavior
• 7. Stories of the Past or Science of the Future? Archaeology and Computational Social Science
• 8. The Potential and Limits of Optimal Path Analysis
• 9. Compute-Intensive GIS Visibility Analysis of the Settings of Prehistoric Stone Circles
• 10. Reconsidering the Concept of Visualscape: Recent Advances in Three-Dimensional Visibility Analysis
• 11. Formal and Informal Analysis of Rendered Space: The Basilica Portuense
• 12. Reproducible Data Analysis and the Open Source Paradigm in Archaeology
• Index
• About the Editors/Contributors