The Archaeology of Hunter-Gatherers as Complex Adaptive Agents (1998) more

The Archaeology of Hunter-Gatherers as Complex Adaptive Agents (Jones’ Axed Dissertation Section, 1997) There have been important developments over the last decade in a new explanatory approach in the sciences which is commonly referred to as “Complexity Theory” (Waldrop 1992; Lewin 1992). Complexity theory focuses on the behavior of complex systems, that is, systems that contain myriad agents which interact with one another in a variety of ways. Complex systems which have received the most intense scrutiny to date include economies (at a variety of scales), the central nervous system, the immune system, ecologies, and the complex dynamics of ontogenetic development. Complexity theory holds that such systems undergo spontaneous self-organization. It is believed that self organization is an inevitable outcome of the interdependencies of complex systems. Such systems are attracted towards a limited number of more or less stable states. It is furthermore hypothesized that within complex adaptive systems, these self-organized states lie at the border between static and chaotic system behavior. It is believed that this position, coined “the edge of chaos” by physicist Norman Packard (Lewin 1992:53), is the most productive state in terms of the ability of the system to respond to and process information. The Santa Fe Institute in New Mexico is the Mecca of complexity theory research in North America (Waldrop 1992). Since 1985, scientists from a variety of fields have met at this think tank to share ideas. Economists, ecologists, physicists, biologists, and anthropologists have engaged in dialogue in the hopes of uncovering the principles that underlie all forms of complex adaptive systems. Their efforts have already resulted in scores of publications by the Institute. General principles applicable to many of the sciences are being investigated with the hope that new universal laws can be developed. At issue is the movement of the sciences away from a limited linear, reductionist way of thinking that has been unable to appropriately model the non-linear systems which dominate the real world. Humans are intimately enveloped by both ecological and economical systems, both of which are linked to one another at deep levels. Human social organization itself can be described as a complex, adaptive system in which numerous agents act for their personal gain, but where all such actions result in system permutations as a result of interdependencies and conflicting constraints (e.g. Kauffman 252-271). Complexity theory, and in particular the study of complex adaptive systems, is poised to shed light on patterns of human behavior, especially where the environments in which we live are undergoing rapid change. For these reasons I will examine how viewing prehistoric hunter-gatherers as adaptive agents within a multi-tiered complex system may advance our understanding of the prehistory of the Northeast at the close of the last Ice Age. The fundamentals of Complex Adaptive Systems Computer scientist John Holland has defined the essential elements which he believes are common to all complex adaptive systems (CAS) (Holland 1992; 1995a; 1995b). This particular synopsis is quite pithy, and I therefore include it here. It is an easy exercise to produce a list of significant characteristics common to all CAS: (1) All CAS consist of large numbers of components, agents, that incessantly interact with each other. (2) It is the concerted behavior of these agents, the aggregate behavior, that we must understand, be it an economy's aggregate productivity, or the immune system's aggregate ability to distinguish antigen from self. (3) The interactions that generate this aggregate behavior are nonlinear, so that the aggregate behavior cannot not [sic] be derived by simply summing up the behaviors of the isolated agents. (4) The agents in CAS are not only numerous, but also diverse. An ecosystem can contain millions of species melded into a complex web of interactions; the mammalian brain consists of a panoply of neuron morphologies organized into an elaborate hierarchy of modules and interconnections; and so on. (5) The diversity of CAS agents is not just a kaleidoscope of the accidental patterns; remove one of the agent types and the system reorganizes itself with a cascade of changes, usually "filling in the hole" in the process. (6) The diversity evolves, with new niches for interaction emerging, and new kinds of agents filling them. As a result, the aggregate behavior instead of settling down, exhibits a perpetual novelty, an aspect that bodes ill for standard mathematical approaches. (7) CAS agents employ internal models to direct their behavior, an almost diagnostic character. An internal model can be thought of , roughly, as a set of rules that enables an agent to anticipate the consequences of its actions. Even an agent as simple as a bacterium employs an "unconscious" internal model when it swims up a glucose gradient in the search for food, while humans make continual prosaic use of internal models, as in our unconscious expectation that room walls are unmoving structures (Holland 1995b: 45-46). Hunter-gatherers, and the societies in which they live can be described by all of the above characteristics typical of complex adaptive systems. Further definitions of CAS are spelled out in more detail in other works, and all appear appropriate to the description of hunter-gatherer societies (e.g. Holland 1995a; Gell-Mann 1995). By viewing huntergatherers as agents within a complex, adaptive social system, a tremendous body of literature concerning CAS becomes pertinent to anthropologists. Applications of CAS Theory for Anthropologists Social Organization CAS theory may shed light on the principles underlying human social organization at a number of levels. Here I am most interested in the organization of hunter-gatherer societies. The observation of the crystallization of hierarchical order from the structure of many complex adaptive systems is reminiscent of observations made around the world concerning the recurrent modes of hunter-gatherer social organization. Hierarchically, hunter-gatherer social organization can be said to increase in degrees of complexity (as measured by the degree of interconnectedness between elements) as one moves from the individual, to pair-bond, to family group, to extended family group, to local band, to regional macro-band or deme, and possibly beyond. CAS theory helps us to visualize the social organization of hunter-gatherers as multi-tiered, highly connected systems. Similar “crystallized” self-generating structure may also be at the root of political and economic organizations such as those defined originally by Service as the band-tribechiefdom-state hierarchy (Service 1962). It seems possible that CAS theory may provide a new way of understanding how these systems evolve and what spurs them to change. CAS theory could provide an explanation for the rapid shifts in prehistoric economies which have sometimes been observed. If social organizations are in fact balanced “at the edge of chaos,” also referred to as the state of “self-organized criticality” (Bak and Chen 1991), where the system is at its greatest potential for information processing, we may be able to understand, and perhaps in part explain, the phase-shift-like transformations which societies sometimes undergo. Human Economy I am of the strong opinion that the principles underlying CAS theory will help anthropologists to resolve problems with current economic models. The optimization theories which dominated the literature of the 1980’s proved helpful in the modeling of hunter-gatherer ecologies and economies. Unfortunately, there were obvious shortcomings, and fieldwork often failed to demonstrate the theoretical underpinnings of most optimal-foraging models. I believe that the major problem with such models was the underlying assumption of the omniscience and unbounded rationality of the agents modeled. Omniscience of environmental conditions and the decision-making process of all other agents permitted such models to view optimal goals as attainable – as have all classical economic models. Agents also tended to act as individuals, rather than elements of dynamic social systems where decisions are limited by historical contingency, and none come without conflicting restraints. Unfortunately, omniscience and unbounded rationality can never be met in real world situations, and agents never act in a vacuum. Complex adaptive systems, on the other hand, do not require all-knowing states, but are able to achieve “excellent” if not optimal solutions to very complex problems such as foraging efficiency (Kauffman 1995: 245271). CAS are able to achieve this exactly because their elements are so deeply integrated. Non-optimal but excellent decision-making behaviors lend themselves to modeling as CAS. Human Ecology In the realm of human ecology, CAS studies may greatly deepen our understanding and appreciation of the union of humans with their environment (e.g. Waldrop 1992: 333). In hunter-gatherer studies this could shed light on predator-prey relationships as well as population dynamics. Answers raised to problems relating to the Lotka-Volterra and Verhulst-Perl population equations may be forthcoming. Both of these have fallen out of favor as ecologists come to embrace much more dynamic models of species interaction, at the expense of the classical successionist and climax-state models (Botkin 1991). When an ecology is seen as a complex adaptive system, it becomes apparent that self-regulating mechanisms develop as a result of the interactions within the system itself, and from its ability to respond effectively to change. As anthropologists, we need to keep pace with such theoretical shifts in other fields. They may aid our pursuit to better understand the role of humans in the “natural” world, and the ways in which we both alter amd are altered by the environment which surrounds us. Human Technology CAS models may also shed light on the evolution of human technologies. When technologies are viewed as deeply interconnected webs of individual parts and tools, it is clear how these, too, can behave as complex, adaptive systems. Fitness landscape algorithms, designed originally for the study of the behavior of the gene variants of an organism, have been used to study a variety of CAS (Kauffman 1995, Lewin 1992: 57). Such models indicate a number of characteristics of CAS, and in this case, of the evolution of technologies. The first is that when technological systems undergo a transitional shift (which may be the result of a seemingly minor perturbance – such as a slight technological innovation), rapid changes throughout the system might occur in chain reactions. During this period of “technological upheaval,” tool forms may go through rapid changes (of morphology, material, manufacturing technique, etc.). The examples most often cited include bicycles and cars (Lewin 1992: 70). After such explosive experimentation, innovation and design improvements slow exponentially (Kauffman 1995: 205). It is reasonable to extend these observation to situations in which technology is closely tied to the environment (or other variables) as well. In this case, where environmental and technological systems are intricately tied, changes in one can affect the other. Thus novel, or rapidly changing, environments may well affect the rate of technological innovation. Similarly, technological changes may substantially affect the environment (think of plows, axes, and nuclear power plants, among others). Clearly feedback scenarios can be expected. Society and the Individual Perhaps one of the most important consequences of viewing human social systems as CAS is the opportunity this provides to empower the individual. The individual seems to have lost stature in recent decades within the field of anthropology, and perhaps particularly within the sub-discipline of archaeology. In the 1950’s, Leslie White promoted a sense that culture change occurs as an inevitable wave within which the individuals of a particular society seemed to bob about, as if adrift at sea (e.g. White 1949, 1959; 1963). In archaeology, the processual movement helped the individual gain a foothold on more solid ground, as he or she was seen as the agent of numerous site formation processes (e.g. Schiffer 1987; Binford 1979). Unfortunately, the human element was all too often lost again in a snowstorm of statistics, flowcharts, and graphs. With the post-processualist era, the archaeologist him- or herself shared the stage with the individuals of the past as skepticism concerning our efforts towards an objective scientific approach spread. Research energy seemed to flow more into the philosophical assessment of our own ability to know the past, than it did into analyzing the archaeological data itself. The individual is king in CAS. While CAS may crystallize into relatively robust states able to reconfigure in the face of network change, inevitably it is the actions of the individual agent that cause cascade reactions to avalanche throughout the system. Adaptive systems perched at the edge of chaos are sometimes referred to as balanced in a state of self-organized criticality, where small changes may have profound consequences (Bak and Chen 1991). Humans can again be seen as the agents who create history. The actions of an individual may repercuss throughout a society for generations, and may bring about phase-shifts to a new level of social organization. Final Thoughts on the Modeling of Human Societies as CAS The beauty of the use of CAS models by anthropologists is that it does not require a new way of thinking. There is no need to debate the effects of paradigm shifts or epistemological revolutions. Anthropologists for over a century have studied humans as elements of a larger, complex social order. It has been evident from the start that this social order is hierarchical and adaptive to both internal and external change. CAS models promise to help us to better understand how that social order functions, on what it is dependent, and how it comes to be. One might imagine a multi-tiered system (a system of correlated fitness landscapes) in which all of the above CAS systems, and others, are closely tied to one another. In this case, even slight changes in one tier might bring about profound changes in another, depending on the degree of system interdependency. By way of example, picture a seven-tiered correlated system composed of the following seven levels: climate, environment, subsistence organization, settlement organization, social organization, technology, and belief systems. Depending on the degree of internal (intra-tier) and external (inter-tier) system connectedness (epistatic coupling), different tiers of the system will be more or less prone to change. CAS models suggest that such systems, when adaptive, “settle into” locations on the edge of chaos (that is, into states of selforganized criticality) where the entire system is most able to respond to and process dynamic information. As CAS dynamics become better understood, it is possible that scenarios such as this one can be profitably modeled through computer simulations. In fact, such models are currently being formulated for the prehistoric Southwest. CAS studies are still in their infancy. While no universal laws of complex adaptive system behavior are currently known, there is hope that they may exist. Complexity theory is a new form of inquiry into the sciences. It is believed by many that it will one day revolutionize the way we understand the world around us. While it was too early to apply the developing models of CAS in this dissertation, I highly recommend that this approach be used by students of human social systems in the coming years.