ARTIFICIAL CULTURE
Nicholas Gessler
In: Artificial Life IV: Proceedings of the Fourth International Workshop on the Synthesis and Simulation of Living Systems, eds. Rodney Brooks and Pattie Maes. MIT Press, Cambridge (1994), 430-435.
ABSTRACT
If we accept the propositions that archaeology is anthropology, and that anthropology is the study of culture process, then we now have at our disposal a new, untapped and rich paradigm for building theories of cultural evolution. Inspired by artificial life (AL), this is the paradigm of a computational anthropology assembled for the development of what I call artificial culture (AC). By studying the archaeology and ethnography of these virtual worlds, AC should help us develop and test theory for cultural evolution. It should neither confine us to those natural cultures accessible to ethnographers, comprised of modern participants and environments, nor to those accessible only to archaeologists. Rather, AC should enable us to explore alternative cultures and experiment with "what if..." scenarios. Perhaps we may in this way bring back to life the adaptations and environments of our early hominid ancestors. The foundations for such research are being built by AL in which we already see rudiments of AC. Beyond those cultures with which we’re familiar, a properly conceived AC should be capable of synthesizing and analyzing virtually any participants in virtually any world.
1. PROLOGUE
Professor Dobb's book is devoted to personetics, which the Finnish philosopher Eino Kaikki has called "the cruelest science man ever created..." The field is a recent offshoot of the cybernetics and psychonics of the eighties, crossbred with applied intellectronics... At present a "world" for personoid "inhabitants" can be prepared in a matter of a couple of hours. This is the time it takes to feed into the machine one of the full-fledged programs (such as baal 66, crean IV, or jahve 09)... A specific type of personoid activity serves as a triggering mechanism, setting in motion a production process that will gradually augment and define itself; in other words, the world surrounding these beings takes on an unequivocalness only in accordance with their own behavior... As hundreds of experiments have shown, groups numbering from four to seven personoids are optimal, at least for the development of speech and typical exploratory activity, and also for "culturization." On the other hand, phenomena corresponding to social processes on a larger scale require larger groups. At present it is possible to "accommodate" up to one thousand personoids, roughly speaking, in a computer universum of fair capacity; but studies of this type, belonging to a separate and independent discipline -- sociodynamics -- lie outside the area of Dobb's primary concerns, and for this reason his book makes only passing mention of them. (Lem 1978, 167ff.)
The foregoing fictional review of a non-existent book was published in 1971, and although personetics and sociodynamics have never emerged as disciplines, we now seem poised to create them.
2. EPISTEMOLOGY
For those anthropologists who subscribe to the new or processual archaeology, the metaphysics and epistemology of anthropology should become the metaphysics and epistemology of science. As its theorist proclaimed:
The external world exists in its own right, and that includes the properties of the archaeological record... It is the availability of the external world, regardless of the character of our cognitive devices, that makes it possible for science to work. We can learn the limitations of our own ideas, as science has demonstrated over and over again, through skillful interaction with the world of experience, the external world. (Binford 1986, 403.)
In a productive science the aim is to learn something of the limitations and inadequacies of one's received knowledge. (Binford 1989, 52.)
Reaction came from a self-named school of post-processual archaeologists who modeled their criticism after the literary post-modernists. To processualists these post-processualists were merely resurrecting the traditional objections to science. On the other hand, many of the post-modernists are studying science with a view towards a more cautious, yet still scientific epistemology. Post-modernism, then, is not necessarily opposed to the production of reliable knowledge about the external world.
Suppose we think about the reality "out there" as an unmediated flux... It interacts with and comes into consciousness through self-organizing, transformative processes that include sensory and cognitive components... On the other side are the constructed concepts that for us comprise the world... Within the representations we construct, some are ruled out by constraints, others are not... I am not saying constraints tell us what reality is. This they cannot do. But they can tell us which representations are consistent with reality, and which are not. (Hayles 1992, 78-80.)
We are increasingly stumbling across the limitations of our cognitive abilities and judging them unacceptable. Solutions include a philosophical restructuring of the mind as well as technological augmentations to it.
The problem (is)... to compress a host of interlocking ideas, drawn from many sources... into a form coherently expressible in a linear script. (Rosen 1991, xiv.) The essence of it lies in what we call the modeling relation... the bringing of two systems of entailment into congruence. (Rosen 1991, xvi.)
The human mind is not adapted to interpreting how social systems behave... Evolutionary processes have not given us the mental skill needed to properly interpret the dynamic behavior of the systems of which we have now become a part. (Forrester 1971, 61.)
Physical models have a long standing as representations of both the external world and our ideas. Computational models join them mid-way between ourselves and reality. They exist outside of our perceptions and are indispensable to our quest for scientific knowledge and understanding.
3. THEORY
To many archaeologists, artifact characterizatrion and methodology was much more important than theory:
Students of contemporary archaeology must concern themselves with much more than seeking an understanding of the archaeological record. (Binford 1989, 62.)
To many anthropologists, theory was either unattainable or elusive. Building theory was the manifesto which the new archaeology proclaimed in 1962:
American archaeology is anthropology or it is nothing. (Binford 1962):
Archaeology must accept a greater responsibility in the furtherance of the aims of anthropology... Archaeologists should be among the best qualified to study and directly test hypotheses concerning the process of evolutionary change... As archaeologists, with the entire span of culture history as our "laboratory," we cannot afford to keep our theoretical heads buried in the sand. We must shoulder our full share of responsibility within anthropology. Such a change could go far in advancing the field of archaeology specifically, and would certainly advance the general field of anthropology. (Binford 1972, 32.)
Processual archaeology even set out to redefine the traditional hallmark of anthropology, the field practice of ethnography itself:
Explaining variability and change in human behavior is the primary goal of the social sciences, yet we expend a great deal of research effort that is of dubious value in this regard. (Hill 1977, 59.)
(The ethnographer) becomes increasingly dependent upon informants to provide him or her with information regarding their knowledge and beliefs in terms of which the local people operate... (Ethnographers) are still not operating in a scientific role. Instead, they have adopted the role of intercultural translators... (This) has compelled many social researchers to rely on their informants to create their data. In turn, these same informants guide the interpretation and ultimately mediate the understanding of the data. What ethnographers report is not data but information, the intellectualized expression of experience. (Binford 1986, 395-96.)
Systems theory held promise for many processualists, yet its limitations also brought disappointments:
Most of the systems literature does not deal with change in enough detail to be of much use. Systems research has been much more concerned with what systems are, how they can be described and classified, and how they are regulated. And further, since general systems theory is designed to be general, and to deal primarily with similarities among different kinds of systems, it provides very few useful ideas on how human social systems might differ from other kinds of living systems in terms of the processes of stability and change. (Hill 1977, 101.)
Only a handful of AL papers have been presented in archaeology and anthropology. So foreign do its practitioners appear, that ironically AL may more frequently be the object of ethnographic field studies than it is the paradigm for their understanding.
4. NATURAL CULTURE
Since the beginnings of anthropology, a definition of culture was as elusive as theory to attain. Repeated attempts have been made to survey the anthropological literature for some scientific foundations. A readable overview is THE RISE OF ANTHROPOLOGICAL THEORY (Harris 1968). With theory still beyond reach, Harris tried instead to formulate a cogent research strategy in CULTURAL MATERIALISM: THE STRUGGLE FOR A SCIENCE OF CULTURE (Harris 1979). Though not addressing computational anthropology, Harris did formulate what he called the universal pattern of culture (Harris 1991, 22). This useful schema may be expressed as a pyramid rising from the environment and ascending in three giant steps:
SUPERSTRUCTURE
(ideas -- cognition)
S T R U C T U R E
(social economy -- behaviors)
I N F R A S T R U C T U R E
(technology -- artifacts)
N A T U R A L - E N V I R O N M E N T
As a materialist, Harris sees causation as largely initiated from the bottom-up, constraining the variation available to higher levels in the pattern. He explicitly advocates a bottom-up research strategy. The scheme is computationally operational when equated with the objects shown in parentheses above. Failing this, the concept of culture has limited utility:
In fact, I believe we can profitably do without the concept "culture," since it appears to be unoperational in analysis. (Hill 1977, 103.)
5. NORMATIVE ASSUMPTIONS
Despite frequent protestations to the contrary in the traditional and popular views of anthropology, culture whether it is defined as ideational or material, is often operationally defined as a corpus of shared traits held in common by a social group. While not denying that traits are sometimes shared, it seems to be the differences in these traits among participants that motivates societal evolution and change. Such normative assumptions of anthropology may be broken down into participant homogeneity, temporal consistency, and referential accuracy.
The assumption of participant homogeneity needs to be tested by incrementally varying the rules from individual to individual. Every division of labor or specialized activity along lines of age, sex, or class implies heterogeneity.
The assumption of temporal consistency needs to be tested by incrementally varying the rules within each individual as a function of temporal context. Daily life is full of inconsistent contextualized classifications.
And the assumption of referential accuracy needs to be tested by incrementally varying the correspondence between rule categories and their external referents. Misinformation is a fact of life.
An AC research strategy should test the extent to which those assumptions may be correct.
6. CULTURE AS EMERGENCE
The role of the individual in society is problematic. He is both the most obvious feature in society and the most neglected when it comes to describing group behaviors. Researchers either ignore the individual level and look for laws of culture at the population level of abstraction, or they ignore the population level and concentrate on the individual. No one has formulated a theory linking the two, although the need is recognized:
The dynamics between individual action -- which, ultimately, is the source for societal attributes measured at a more summary level -- and group properties, including societal organization and cultural systems, is a constant problem that archaeological theorizing has not adequately addressed. (Read 1990, 50.)
Behavior is the working out of deeper, structuring properties, hence the focus on the regularity of behavior and behavioral products characteristic of much of scientific archaeology incorrectly directs attention away from the structuring processes towards their consequences. (Read 1986, 16.)
Explanation... (may proceed) without the necessity for absolute, universal laws of behavior. (Read 1986, 11.)
As cellular automata illustrate, individual actors, operating under individual local rules, can automatically produce collective global patterns of behavior that emerge solely through their mutual interactions. Importantly, these global patterns of behavior are not programmed into the simulation. They have no existence within the individual actors themselves. Rather, they come into being only as the entire system operates. I suspect that many of the same processes are operating in culture.
It is not difficult to conceive of an emergent world, from the sub-atomic to the cosmological levels of abstraction, wherein quickly acting smaller elemental objects (or variables at lower levels), by carrying out only rules based upon local knowledge, give rise to more slowly emerging larger global patterns (or structures of behavior at higher levels). These higher level structures may then be taken as elemental variables for yet higher levels of interaction. Cause gives rise to consequence, and consequence becomes cause at an even higher level of abstraction. Structure in this view is a relative term: It is the abstraction of behavior as sensed from a higher level.
7. ARTIFICIAL CULTURE
By altering the interpretation of some models, it can be argued that research in AC is already underway by AL. There is no compelling reason why the differentiation and clustering behavior of a population of computational algorithms in a computer’s memory should resemble the emergence of co-adaptation and species-prey interaction in biology any more closely than it resembles the emergence of class structure or task specialization in social science.
Cellular Automata are perhaps the clearest examples of the emergence of global patterns from local rules (see Forrest 1991, Gutowitz 1991). They form a metaphor for human interaction on two-dimensional substrate. When cells are made mobile and their states include the ability to move, the social metaphor is even more complete.
Tierra, a highly formalized and abstract virtual world, is home for programs which evolve through natural selection. From an ancestral creature, speciation occurs with lineages specializing in different adaptations with minimal restraint. The species compete, cooperate, socialize, cheat, parasitize, and hyper-parasitize. (Ray 1991, 1993.)
Iterated Prisoners’ Dilemma with Choice and Refusal is populated by participants who are repeatedly matched as pairs, each having the choice of cooperating with its partner or defecting, the payoffs differing based on one partner's actions. With the faculty of memory, each can remember the past actions of the partner who chose it and can refuse to participate based upon that experience. As the population evolves, sub-populations emerge which can be characterized as cooperators, rip-off artists, and specialists who rip-off the rip-off artists. This is a distinctively human game with the formalism of co-evolution brought into the cultural arena (Stanley, Ashlock & Tesfatsion 1994).
SimLife is a software package written for the computer game market, but which may be useful for prototyping scientific research. One defines a territory, sows it with vegetation, populates it with customized creatures, and then sits back to watch the ecosystem evolve (Karakotsios & Bremer 1993, Maxis 1993).
Strategic Theater of War (STOW) is a world-wide network of military vehicle simulators connected to semi-automated forces which are intelligent controllers for footsoldiers and single tanks requiring a minimum of human intervention. ARPA is presently working on fully automated representations for these players. (McDonough 1993).
Cooperative Robot Behavior is a goal of many robot builders, and its engineering is in some ways similar to constructing AC simulations, both in questions of how to achieve collective behavior and how to model the agents (Maes 1994, Mataric 1993, Steels 1994). In Brooks’s bottom-up approach robots have no locus of knowledge, reason, or representation, yet despite these limitations, the robot acts intelligently. What Brooks has observed about robotics may also be true for culture:
Just as there is no central representation there is not even a central system. Each activity producing layer connects perception to action directly. It is only the observer of the Creature who imputes a central representation or central control. The Creature itself has none; it is a collection of competing behaviors. Out of the local chaos of their interactions there emerges, in the eye of an observer, a coherent pattern of behavior. There is no central purposeful locus of control. Minsky gives a similar account of how human behavior is generated. (Brooks 1991a, 148-49.)
The titles of Brooks’s articles reinforce his bottom-up approach: "Elephants Don't Play Chess," "Intelligence Without Representation," and "Intelligence Without Reason" (Brooks 1989, 1991a, 1991b).
It is by no means clear that... a (perceptual world) is anything like what we actually use internally -- it could just as easily be an output coding for communications purposes... In fact, it may be the case that our introspective descriptions of our internal representations are completely misleading and quite different from what we actually use. (Brooks 1991a, 144.)
Simon noted that the complexity of behavior of a system was not necessarily inherent in the complexity of the creature, but perhaps in the complexity of the environment... We hypothesize... that much of even human level activity is similarly a reflection of the world through very simple mechanisms without detailed representations. (Brooks 1991a, 149.)
The goal of AC is to create a population of dynamically evolving terrain-based mobile autonomous agents serving as a complex of multiple interacting hypotheses for understanding human cultural behavior. Those agents can either be simulated in hardware as robots, or simulated in software as programs. The future of AC is optimistic:
Ideas, and other atomic particles of human culture, often seem to have a life of their own -- organization, mutation, reproduction, spreading, and dying. In spite of several bold attempts to construct theories of cultural evolution, an adequate theory remains elusive. The financial incentive to understand any patterns governing fads and fashion is enormous, and because cultural evolution has contributed so much to the uniqueness of human nature, the scientific motivation is equally great. (Taylor & Jefferson 1994, 8.)
8. COMPUTATIONAL OBJECTS.
Software for Artificial Culture is currently in the proof of concept stage. To facilitate communication among anthropologists, the code is written in C++ to run on a Comtrade EISA VLB 486-66 DX2 with 8MB ram and 256KB cache. A Digital Processing Systems, Personal Animation Recorder model DR-2100 may be added to collect frame-by-frame visualizations of data from long runs on a dedicated hard drive for later playback through a VCR at higher speeds. This configuration should force the search for simplicity in designing key modules and functions. As needed, software will be scaled up to run on a massively parallel machine. As it progresses, my implementation of AC should embrace the following objects:
Grid is the terrain of the artificial world on which virtual food, people, and artifacts will interact.
Virtual Objects are people, resources, artifacts, architecture, and geographic features. They may be further characterized by a food value, energy content, information load, storage properties, and persistence. They may evolve on their own.
Personoids are the agents having the properties of other virtual objects plus cognition and the ability to learn and remember. Among static characters will be a name, sex, birthdate, and pedigree. Among dynamic characters will be age, weight, fecundity, morbidity, caloric intake, and caloric surplus. Among cognitive characters will be "rules," whether responsive, significatory, or symbolic, which mediate between sensory and motor organs, or in robotic parlance between sensors and actuators. Learning will be established through techniques of evolutionary computation and cultural transmission. Basic emergent pattern detectors will be provided to sense abstract behaviors such as theft, trade, organizational groupings, seasonality. The code should allow new emergent pattern detectors to evolve.
Environment is a relative term comprising everything external to the object under consideration, which also influences it.
Ethnographer is the trajectory of selected program parameters and states, restricted in ways which one might expect actual ethnographic data collection may be biased. The ethnographer may be provided with higher level emergent pattern recognizers to detect aggregation, dispersion, residence, and marriage. The ethnographer must also be embodied, operating within the physics of the virtual world and perceivable by its objects. Its cognitive skills may also be allowed to evolve. This reflexivity should enable us to examine the biases of various natural-world ethnographic practices.
God is omniscient. S/He is the ideally unbiased record of the trajectory of all program parameters and states. S/He exists for us only in a less-than-ideal state. Unlike the ethnographer S/He is disembodied and objective. Higher level emergent pattern detectors may be employed.
As these objects are implemented, the AC will recursively be scaled up and pared down in order to enhance its performance and its congruence with natural cultures, beginning at the level of foraging hunter-gatherers.
9. PRELIMINARY RESULTS
Since this project is in its formative stage, preliminary results are only indicative of future complexities. Already it is evident that foraging efficiency (measured as the agents’s ability to extract all the food resources before death) varies widely and often counterintuitively with differing search strategies, distributions of resources, and rations eaten. This is partially the result of agents destroying the gradient which would lead them to higher food concentrations. Agent and contour-following, hill-climbing, and random path search behaviors are observed. Resource distributions are transformed: random distributions are made uniform, resource desert patches are created, and secondary resource patterns are generated some quite unlike the original distributions. From these preliminary results I would expect that a computationally based AC would make an excellent test platform for equationally based foraging theory (Stephens & Krebs 1986), behavioural ecology (Krebs & Davies 1991, 1994), and human evolutionary ecology (Winterhalder & Smith 1988, Smith & Winterhalder 1992). I would expect that AC’s terrain embodied interactivity would eventually provide us with new insights into the complex adaptive functionalities of human cultural and bio-cultural evolution (Lovejoy 1992).
10. ETHICS
The promulgation of AC will be accompanied by ethical debate on its potential uses for prediction, control, and social change. The results of AC modeling used to direct social policy may be somewhat tempered by its use by an increasingly computer literate public. Nevertheless, as with any new technologically driven enterprise, the mastery of such evolutionary computational techniques will probably fall to a technical elite. Intriguingly, at least two new concerns have been raised from science fiction for serious discussion.
Given the strong claim that AL is in fact real life, we must anticipate the same assertion for a developed AC. If artificial entities become as real as natural ones, should the rights of natural entities be extended to the artificial? The door swings both ways. If some artificial entities are judged to be in ways more acceptable than their natural counterparts, should the rights of certain deviant natural entities be abolished?
A related concern is the effect that socially unacceptable behaviors directed towards artificial worlds may have on the individual’s interaction with the natural world. An immersive virtual reality coupled with a developed AC may make the natural and artificial indistinguishable (as happens in contemporary war games). When that distinction is suspended, then must human behavior in the two realms be judged by the same conventions?
REFERENCES
Binford, Lewis R. 1962. "Archaeology as Anthropology." American Antiquity, 28, 2, 217-225. cited
Binford, Lewis R. 1972. "Archaeology as Anthropology." An Archaeological Perspective. New York: Seminar Press. cited
Binford, Lewis R. 1986. "Data, Relativism and Archaeological Science." Man (N.S.) 22, 391-404.
Binford, Lewis R. 1989. "The 'New Archaeology,' Then and Now." Archaeological Thought In America, by C.C. Lamberg-Karlovsky, 50-62. Cambridge: Cambridge University Press. gessler cited
Brooks, Rodney A. 1989. "Elephants Don't Play Chess." Cambridge: MIT Artificial Intelligence Laboratory. gessler
Brooks, Rodney A. 1991a. "Intelligence Without Representation." Artificial Intelligence 47 (1991) 139-159. Elsevier. gessler
Brooks, Rodney A. 1991b. "Intelligence Without Reason." International Journal Of The Conference On Artificial Intelligence (1991). gessler
Forrest, Stephanie, editor 1991. Emergent Computation -- Self-Organizing, Collective, And Cooperative Phenomena In Natural And Artificial Computing Networks. Special Issues of Physica D. Cambridge: MIT Press, A Bradford Book (Elsevier Science).
Forrester, Jay W. 1971. "Counterintuitive Behavior of Social Systems." Simulation, February, 61ff.
Gutowitz, Howard, editor 1991. Cellular Automata -- Theory And Experiment. Special Issues of Physica D. Cambridge: MIT Press, A Bradford Book (Elsevier Science).
Harris, Marvin 1968. The Rise Of Anthropological Theory: A History Of Theories Of Culture. New York: Thomas Y. Crowell. gessler cited
Harris, Marvin 1979. Cultural Materialism: The Struggle For A Science Of Culture. New York: Random House. gessler cited
Harris, Marvin 1991. Cultural Anthropology. Third edition. New York: Harper-Collins. gessler cited
Hayles, N. Katherine 1992. "Constrained Constructivism: Locating Scientific Inquiry in the Theater of Representation." New Orleans Review, Winter, 76-85. cited
Hill, James N. 1977. "Systems Theory and the Explanation of Change." Explanation Of Prehistoric Change. A School of American Research Book, 59-104. Albuquerque: University of New Mexico. gessler
Karakotsios, Ken and Michael Bremer 1993. SimLife -- The Official Strategy Guide. Rocklin: Prima.
Krebs, J.R. & N.B. Davies 1991. Behavioural Ecology: an Evolutionary Approach. Oxford: Blackwell.
Krebs, J.R. & N.B. Davies 1994. An Introduction to Behavioural Ecology. Oxford: Blackwell Scientific.
Lem, Stanislaw 1978. "Non Serviam." In Stanislaw Lem's, A Perfect Vacuum, 167-196. San Diego: Harcourt Brace Jovanovich.
Lovejoy, C.Owen. 1992. "Modeling Human Origins." Handout for the Center for the Study of the Evolution & Origin of Life Symposium, UCLA, Los Angeles.
MacDonough, James 1993. "Doorways to the Virtual Battlefield." Paper presented to Virtual Reality -- The Commitment to Develop VR. Los Angeles.
Maxis 1993. Sim Life -- The Genetic Playground. Software by Maxis, Orinda, California.
Ray, Thomas S. 1991. "An Approach to the Synthesis of Life." Artificial Life II, SFI Studies in the Sciences of Complexity, vol. X, edited by Langton et al. Reading: Addison-Wesley.
Ray, Thomas S. 1993. "An Evolutionary Approach to Synthetic Biology." Manuscript. (October 21.)
Read, Dwight W. 1986. "Mathematical Schemata and Archaeological Phenomena -- Substantive Representations or Trivial Formalism?" Science And Archaeology, 28, 16-23.
Read, Dwight W. 1990. "The Utility of Mathematical Constructs in Building Archaeological Theory." Albertus Voorrips, editor, Mathematics And Information Science In Archaeology: A Flexible Framework. Studies in Modern Archaeology, Volume 3. Bonn: Holos. read
Rosen, Robert 1991. Life Itself: a Comprehensive Inquiry... New York: Columbia University Press.
Smith, Eric Alden & Bruce Winterhalder 1992. Evolutionary Ecology & Human Behavior. New York: Aldine de Gruyter.
Stanley, E. Ann, Dan Ashlock and Leigh Tesfatsion 1994. "Iterated Prisoner's Dilemma with Choice and Refusal of Partners." Artificial Life III, SFI Studies in the Sciences of Complexity, vol. XVII, 131-175, edited by Langton. Reading: Addison-Wesley.
Stephens, David W. & John R. Krebs 1986. Foraging Theory. Princeton: Princeton University Press.
Taylor, Charles and David Jefferson 1994. "Artificial Life as a Tool for Biological Inquiry." In Artificial Life. Volume 1, Number 1/2, 1-13. MIT Press.
Winterhalder, Bruce & Eric Alden Smith 1981. Hunter-Gatherer Foraging Strategies. Chicago: University of Chicago Press.