- Social Sciences
- African Studies
- American Studies
- Asian Studies
- Communication Sciences
- Ethnic Studies
- European Studies
- Gender Studies
- Physical Sciences
- Life Sciences
- Animal Communications
- Cell Biology
- Evolutionary Biology
- Food Science and Technology
- Human Anatomy
Where Do Research Problems Come From?
by Lewis R. Binford
Where Do Research Problems Come From?
Lewis R. Binford
Updated: November 5th, 2012
WHERE DO RESEARCH PROBLEMS COME FROM?
Lewis R. Binford
In science, understanding the relationship between data and problems is crucial to successful research. When problems are imposed on the data and its organization-instead of addressing the empirical subject matter ofrhe discipline-a major epis- remological dificulty arises. What domains of knowledge can reliably inform about the imposed problem? I suggest that the many claims that all data are "theory dependent" are primarily accurate in those instances when a problem is imposed on a body of data. This is the approach of humanities. When a problem is recognized in the context of pattern recognition studies that are focused on data generatedfrom the study of one's subject mattel; a different relationship exists between data and the- ory and fruitj5~1 theory building becomes possible. The latter is the approach of science. Central ro this contrast is the issue of where problems come from.
En las ciencias la comprensidn de la relacidn entre 10s datos y 10s problemas es la clave de la investigacidn exitosa. Cuando se imponen 10s problemas en 10s daros ysu organizacidn-en vez de dirigirse a la materia empirica de la disciplina-surge una difi- cultad epistemoldgica mayor: Cudles son 10s dominios del conocimiento quepueden informarnos en una forma confiable acerca delproblema impuesto? Yo sugiero que la ajrmacidn muy comin de que todos 10s datos "dependen de la teoria" es generalmente correctn en esos cnsos en 10s cuales se impone un problema en un grupo de daros. Esra es la manera de acercarse a1 asunro que caracteriza a las humanidades. Pero cuando se reconoce un problema denrro del contexro de 10s estudios de reconocimienro de patrones, que se enfocan en 10s daros producidos por el estudio de la materia de la disciplina, entonces existe una relacidn bas- rante distinta enrre 10s daros y la teoria y, al mismo tiempo, se hace posible la construccidn de la teoria. Esta liltima es la manera de acercarse a1 asunto que caracteriza a las ciencias. La clave para entender este contraste se ubica en la cuestidn de ddnde sur- gen 10s problemas.
The Rationality of Procedure accomplish." When this principle is applied to an
Relative to Goals examination of any scientific issue, the identity of
the empirical domains of phenomena that the science
growing body of archaeological literature has chosen as its subject matter is of paramount
addresses the questions of what ideas are importance. In the case of the science of archaeol-
valuable, what "heuristics" are unambigu- ogy (see Spaulding 1960), the target of study is the
ous, and what concepts are germane to the task of content and associational properties of the archaeo-
doing archaeology.' As a teacher, I have found some logical record.
of this literature not only confusing to students but Given this research target, we may also ask:
sometimes mystifying to me as well. "What guides or focuses our studies?'We might add
When I suggest to my students that they explore that since archaeology is the science of the archae-
the proliferating archaeological literature, I try to ological record, one would expect that the problems
prepare them for the experience by enunciating a we seek to solve are derived from a study of archae-
basic principle to guide their reading that relates ological remains. Many researchers, however, appear
directly to the goals of research: "I cannot make any to think that the problems to be investigated are
judgment about the rationality of the way you intend derived from contemporary social, cultural, or polit-
to proceed until you tell me what you are trying to ical issues (see Preucel and Hodder 1996:3-20). Such
Lewis R. Binford. Department of Anthropology, Southern Methodist University, P. 0.Box 750336, Dallas, TX 75275-0336
American Antiquity, 66(4), 2001, pp. 669-678
Copyright63 2001 by the Society for American Archaeology
approaches are common in the humanities. This, however, leads to a problematic linkage between empirical phenomena and the theories advanced from other sources about their causes (Merton 1968). Programs of research that are stimulated by questions arrived at independently of a study of the archaeo- logical record have as their imperative the discovery of something in the archeological record alleged to be informative about the "problem" of interest. It is suggested that this difficulty only arises in the social sciences and the humanities where problems do not generally derive from a study of the subject matter per se.
The recognition and discussion by archaeologists of "where do problems come from" is probably most characteristic of the contemporary era. Such a prob- lem does not appear to have been acknowledged dur- ing the era of traditional archaeology. That this is so is perhaps best illustrated by the fact that Walter Tay- lor (1948:7) had by 19432 sought to "focus the atten- tion of archeologists upon the nature of their objectives, their practices, and their conceptual tools." Taylor treated the issue of the archaeologists' dependence upon other disciplines to both define archaeologists' problems and to solve them as fol- lows:
Archeology per se is no more than a method and a set of specialized techniques for the gath- ering of cultural information. The archeologist, as archeologist, is really nothing but a techni- cian. When he uses his findings to study archi- tecture, he must employ the concepts developed in that field, and when he studies culture, he must use the theoretical structure erected by those who have made it their business to study culture, namely the anthropologists [1948:43].
Acceptance of Taylor's pronouncement points to an additional step in the epistemological linkage between problem and data that is not characteristic of thought in the sciences. If the archaeologist's research questions come from the contemporary cul- ture of our field, as advocated by many contempo- rary archaeologists (see Gamble 1999; Gero and Conkey 1991; Hayden 1975: 56; O'Brien and Hol- land 1990; O'Brien and Lyman 2000; O'Connell
1995; Odell 1993: 109-1 lo), then the intellectual tools for uncovering answers to these problems must be taken from the contemporary knowledge base. We may reasonably ask-where does the past come in? The research strategy of Taylor's archaeological "technicians" must be to perpetually survey the broader, interdisciplinary landscape for concepts and interpretative principles that can be imposed on their data to render their work "meaningful" to discussions of the past. Taylor elaborated on the goals of the dis- cipline as follows:
archeology is neither history nor anthropology. As an autonomous discipline, it consists of a method and a set of specialized techniques for the gathering or "production" of cultural infor- mation [Taylor 1948:44].
Even though in today's lexicon Taylor's words might be translated into phrases such as "behavioral information" or "evolutionary information," I think most contemporary archaeologists would be fairly comfortable with Taylor's original choice of words. The question remains, nonetheless, how does one "gather or produce" cultural information by exam- ining the archaeological record? Where do archae- ologists obtain the interpretative knowledge and conceptual tools for translating archaeological data or observations on the form, arrangement, distribu- tion, frequency, and association among material items and their depositional contexts into "cultural information" about the past?
Procedural Means and Goals:
Picking on Lithic Studies
The problem outlined above is endemic in archae- ology and anthropology; I do not see the need to doc- ument it further for the thoughtful reader. However, in my exploration of the methodological conse- quences I will use a recent book on lithic analysis to illustrate some additional points when discussing the consequences of failing to address the general issue of where problems come from.
For instance, Odell comments as follows regard- ing recent introductions of "theory" to lithic studies:
major theoretical advances were being devel- oped in archaeology from bases in population ecology and ethnography. Optimal foraging theory (Winterhalder and Smith 1981) and the forager-collector model for the organization of hunter-gatherer mobility (Binford 1979, 1980; Kelly 1983) came to be perceived as appropri- ate frameworks for explaining archaeological site distributions, activities, etc., and it was not long before these models were applied to lithic data. . . [1996:3].
Odell's description of current practice is exactly how Taylor imagined theory to be obtained and used by archaeologists (Taylor 1948:29). The archaeolo-
gist is a "user" of theory to provide a Rosetta Stone for transforming the inert phenomena of the archae- ological record into "cultural information" or-in more contemporary terms-"behavioral information" about the past.3 In other words, archaeologists are exhorted to translate the observations made on the archaeological record into the conceptual terms of a theory, or more commonly a conceptual char- acterization of behavior or a particular observational situation. This process then invites researchers to produce an "interpretation" for their observations. Such derived intellectual products may also be referred to as reconstructions.
Problems with Interpretation as an Intellectual Goal
Many followers of the above interpretative strategy consider the uncertainty of the linkage between abstract theory and the attributes or characteristics observed archaeologically to be primarily responsi- ble for the many problems confronted by the archae- ological technician (sensu Taylor). Specifically, they want to know the "correct" meaning to assign to archaeological observations, which seems to be what Odell has in mind when he states that "stone tool curation is a concept employed to explain certain aspects of prehistoric hunter-gatherer behavior, and its effect on lithic assemblages . . . (19965 1, empha- sis added).4 It is, however, difficult to appreciate how
a "concept" can both explain unidentified aspects of "prehistoric hunter-gatherer behavior" and at the same time unambiguously identify the consequences of this explained behavior in the archaeological record.
Archaeological discussion and argument consis- tent with the Taylorian view usually appear as attempts to justify or warrant the act of assigning par- ticular meanings to particular phenomena. A nega- tive critique of interpretative conventions frequently consists of lengthy discussions about the ambiguity, or lack thereof, characterizing the link between one's archaeological observations and the meanings imparted by one's conceptual conventions. "New diagnostic" conventions may be labeled as unclear, ambiguous, or inappropriately applied to the empir- ical properties normally subjected to more conven- tional interpretations. Jelinek (1976:27), Nash (1996), Ode11 (1981), and Shott (1996) illustratepar- ticularly pertinent examples of the latter tactics.
At its worst, argument is focused on the accept-
ability of specific words (Nash 1996:93), while at best, it focuses on the most likely suites of alterna- tive causal dynamics standing behind the observed properties. Almost all of the papers assembled in Odell's (1996) recent book are based on one or the other of these explanatory procedures. Within the Taylorian or humanistic milieu, a theory is evaluated in terms of how applicable it is to the archaeologi- cal observations it is claimed to explain. The best the- ories, it is argued by Odell and others, are reliable and unambiguous in referencing germane observa- tions. As Odell opines:
one of the challenees of the future will be to
assure that the elemental observations of an analysis provide accurate and believable sup- port for the theoretical structures that are cur- rently being created [1993:l 18].5
I could search for many years without finding another proposition with which I disagree more com- pletely. If accurate, the above opinion insures that all theory is "applied to data" as an accomodative argu- ment. Thus, researchers who believe that theory can be verified and act as Odell suggests insure that all verifications are tautologies. This is a danger that Popper long ago recognized, and as remedy argued for a falsificationist strategy in science (Miller
1985:162-170). In rebuttal to Odell, I offer the fol- lowing suggestions that, I hope, will demonstrate that elemental observations as such are not where one looks for support for theories and that one does not develop or adopt a theory in order to infuse frst-order observational units6 or basic data with meaning.
Curation (see Binford 1979, 1980) ought not to be thought of as a theory but as a descriptive term referring to a particular way of organizing tool use and production as observed in a living system. Archaeologists do not observe living systems, only their static byproducts. This assertion introduces the second principle to be discussed in this paper: what guides or focuses our studies?
Focusing first on "looking and ~eeing,"~
we may ask ourselves is curation something that can be seen directly, like an entity, within a living system? No, it has reference to a way of organizing raw materi- als, labor or workers, and tasks. It is only visible as a result of the study of the interrelations among things: actors, actions, and places. In this sense it is not a dimension as defined by Spaulding (1960:438). Curation has reference to a particular way of orga- nizing the relationships among a set of dimensional phenomena. It refers to a relational state and there- fore is not an "essential" property of any phenomena said to be so related. Looking for it as a directly visi- ble, discrete, and unambiguous diagnostic, as for instance aproperty of some stone tool, is unreasonable.
In science, diagnosis is usually a process of elim- inating a number of possible contributory factors rel- ative to some "primary observation." Qualitative and quantitative chemical analysis usually requires many different tests to identify a substance or to estimate the quantity that is present in a sample. Think about the complaint patients frequently make when they seek medical advice about a health problem, that they are subjected to too many tests or sent to too many specialists. It is important to remember that a diagnosis is rarely made as a result of a direct, uni- dimensional fit between symptom and "cause."
Successful diagnosis is the byproduct of a long his- tory of learning, tempered by the realization that the way we see the world and the reality of natural dynam- ics are two very different things. So, too, in archaeology, the perception of ambiguity and the inadequacy of our cognitive devices to comprehend directly the reality of natural dynamics-particularly from obser- vations on statics-are the byproducts of past suc- cesses or failures in our science. There is no justification, however, for throwing away what we have learned about dynamics just because our efforts have not produced unambiguous, diagnostic procedures for the identification of archaeological materials.
I have suggested elsewhere that the identification of ambiguity is one of the best ways to recognize an opportunity for learning (Binford 1987). Popper (Miller 1985:179-180) has long ago suggested the same general viewpoint:
we may say that the most lasting contribution to the growth of scientific knowledge that a theory can make are the new problems which it raises, so that we are led back to the view of science
and the growth Of as starting
from, and always ending with, problems-prob- lems of an ever increasing depth, and an ever increasing fertility in suggesting new problems.
On the other hand, in recent critiques of archae- ological progress, many writers have come to a dif- ferent conclusion. Ambiguity is cited as the justification for dismissing ethnographic observa- tions and descriptions-as well as observations about how living organizations work-because they fail to see dynamics directly in their static, end-state arti- facts (Ode11 1993:112, 113, 1996: 53)! The fact that the dynamics of most recent formation processes conditioning the character of the archaeological record are different from the large-scale dynamics affecting the organization of cultural systems seems to be lost on many researchers (for examples, see Bin- ford 198 lb). There are many different scales in terms of which research may be legitimately organized.
The above suggestions point directly to the impact on our work of our ways of "looking and seeing." In turn, recognition of different scales for observation demands a further consideration of how our tactics for thinking affect the factors that "guide or focus" our studies. Such contemplation awakens us to sev- eral additional germane issues. Our understanding of our focus and associated guidelines for research will be enhanced if we return to my initial principle, which states that "I cannot make any judgment about the rationality of the way you intend to proceed until you tell me what you are trying to accomplish." In keeping with this principle, we are forced to recog- nize that there are major differences between what scientific researchers and their counterparts in the humanities accept as desirable accomplishments. Such differences increasingly separate archaeologi- cal scientists from their humanistic companions.
Scientific Versus Humanistic Goals and Approaches
Science seeks to build a cumulative learning trajec- tory through the study of its empirical subject mat- ter. Scientific study presupposes the goal of building an explanatory theory for the problems that are iso- lated during the course of study of a specific subject matter. On the other hand, archaeologists who con- sider archaeology a humanistic discipline attempt to write culture history, to reconstruct past life ways, or identify events such as the arrival of the fist Ameri- cans. Other archaeologists, who do not identify them- selves as humanists, share nonetheless a similar tactical approach in their research by attempting to argue that archaeological phenomena can be explained by selectionist, Marxist, structuralist, other theories, or even by accommodation to simple ethno- graphic observations. This refers to points made ear- lier about what archaeologists study and where they obtain their explanations. Do we, as Walter Taylor suggested, have to borrow our theory from elsewhere? Is archaeology a science in its own right, whose sub- ject matter-thearchaeologicalrecord-is the source
of the problems to be researched? Do archaeologists' arguments about their subject matter constitute the source in terms of which theories are developed about the archaeological record and the past?
Our answers to these questions can be guided by an exploration of the differences between interpretative arguments and referral arguments. Quite explicitly, an interpretation is an assertion that some characteristic or situation in the past is evidenced by some specific characteristic or property of the archae- ological record. In short, a knowledge claim is being made regarding the meaning of some archaeologi- cal observation. On the other hand, a referral argu- ment suggests the organizational context of past dynamics in which some properties observed in the archaeological record are thought to have arisen. No specific knowledge claim is being made regarding the details of causation nor the unambiguous mean- ing of any specific archaeological referent. One could say that an interpretation is an assertion or a knowl- edge claim about the past,whereas a referral argu- ment is an assertion or suggestion about the domain of dynamics that might be profitably investigated for clues to causal conditions operative in the past. As
such there are potential implications both to known archaeological phenomena as well as previously unobserved phenomena. Referral arguments are rich in empirical implications, whereas interpretative arguments are essentially sterile with respect to addi- tional empirical content. The goals of these two kinds of argument are different and, therefore, the logic and procedure for using them are different, as is the intel- lectual use and function of their product.
Referral arguments are always suggestions about the appropriate causal or organizational contexts that should be explored as part of a search for further understanding about the phenomena under investi- gation; as such, they further the learning goals of sci- ence. For instance, I stated aputative referral context for variability among Mousterian assemblages as follows:
If we assume that variation in the structure and content of an archaeological assemblage is directly related to the form, nature, and spatial arrangement of human activities, several steps follow logically. We are forced to seek explana- tions for the composition of assemblages in terms of variations in human activities. The fac- tors determining the range and form of human
activities conducted by any group at a single location (the site) may vary in terms of a large number of possible "causes" in various combi- nations. The broader among these may be sea- sonally regulated phenomena, environmental conditions, ethnic composition of the group, size and structure of the group, regardless of the ethnic identity of the group. Other determining variables might be the particular situations of the group with respect to food, shelter, supply of tools on hand, etc. In short, the units of "cau- sation" of assemblage variability are separate activities, each of which may be related to both the physical and social environment, as well as interrelated in different ways [Binford and Binford 1966:240-2411.
My arguments with Bordes in the 1960s and 1970s were not focused on how he classified stone tools. I questioned his assumption that variability in stone-tool frequencies was referable exclusively to the ethnic identity of the makers. Instead, I thought of assemblages as varying combinations of suites of items for which the determinant conditions of their regular association could be semi- or completely independent from one another. I thought that the first analytical step should be to find out how the "tool types" varied relative to one another. Then, we could proceed to study how subsets of dependent and inde- pendently associated types interacted with environ- mental and chrono-climatic variables, site-specific contexts, lithic raw materials, and associated fauna. Patterns identified at this scale of analysis would provide insights into how assemblage variability was conditioned, and we could then begin to build the- ory about its causes.
I also thought that such patterning might lead us to explore data based on ethnographic observations that would be germane to site-formation processes. My subsequent fieldwork among the Nunamiut rep- resented a strategy for understanding the differences in bone assemblages, described in terms of anatom- ical part frequencies among documented Eskimo sites (Binford 1978). Either investigative trajectory represents middle-range research that could be ini- tiated to expand our knowledge relative to that avail- able at the time a specific referral argument was suggested.
Interpretative Conventions and Their Warranting Arguments
Humanist interpretative goals, which include the reconstruction of past life ways in human terms or the documentation of specific events, differ radically from the knowledge goals of middle-range researchers. To humanists, referral arguments are seen as potential interpretative principles-derived from cognitive devices-that have the potential to provide pictures from a long-vanished past. If inter- pretation of the archaeological record is one's goal, what intellectual tools are required to render the past knowable?
The most essential tools are cognitive conven- tions-true conventions-that guarantee when X is observed in the archaeological record, it means Y in terms of the past. Technically, in order to look at the static archaeological record and read a human nar- rative of the past, a huge array of laws-a dictionary of conventions-is required to translate static remains into human dynamics. In short, a full under- standing derived from basic scientific research is needed in order to have a reliable interpretative dis- cipline. If one is engaged in doing science, however, such a result is unacceptable, given the applied demands for producing a humanistic construction of the past. The daunting task faced by the humanist living in the shadow of a poorly developed basic sci- ence is to be able to find something among a set of archaeological remains that can tell the researcher what he or she wants to know. Normally this is accomplished by ~tipulation.~
Taylor maintained that the act of selection would be guided by "theory." Ironically, however, the implementation of Taylor's logic and its attendant procedures guarantees that the claim that all observations are theory-dependent would be ac~urate.~
This same intellectual trap stim- ulates much of the criticism and intellectual frustra- tion expressed in Odell's (1996) book.
Humanists regard referral arguments as potential conventions for use in interpretation. When poten- tial interpretative conventions are unjustified by sci- entific research, they represent unwarranted, conventional "truths" that are suspect and in need of investigation. Many years ago I argued that expla- nations rather than interpretations should be the goal of archaeological researchers (Binford 1973:253). This remains my goal today.
Much of my professional life has been devoted to learning about how the archaeological record was created and what agents produced its many disparate properties. When Ihave felt that I have made some progress toward this goal, I have presented my research, particularly in the realm of ethnographic and actualistic studies. I have frequently pointed out the inadequacy of many interpretative conventions used by archaeologists seeking to diagnose the char- acter of the past from their data, particularly those that were assumptively based. My criticism does not, however, imply that what I have experienced and observed represents all there is to know about the relationships between dynamics and statics. Why, then, do so many archaeological "consumers" of bor- rowed ideas that are used to "diagnose" the past demand unambiguous, single-contingency causes? Is this the way other sciences work relative to applied or interpretatively dependent associated disciplines? No.
Years ago I disagreed with Taylor's claim that in the 1940s he had initiated the changes in archaeo- logical thinking and practice that I and others were proposing in the 1960s.1° It was my contention that, in science, the arguments of the 1960s constituted a very different approach to both the use of words and the use of observations. I also suggested that archae- ologists could learn about many of the organizational and behavioral characteristics of living systems that some researchers had said could not be addressed through archaeological inference (Hawkes 1954). I stressed that
the practical limitations on our knowledge of the past are not inherent in the nature of the archaeological record; the limitations lie in our methodological naivete, in our lack of develop- ment of principles determining the relevance of archeological remains to propositions regarding processes and events of the past [Binford 1968:23].
I have since realized that some of the naivete I referred to resulted from my failure to see clearly that problems are best identified through the exploratory investigation of observations on whatever category of phenomena is chosen for research. Scientific prob- lems derive from the study of a subject matter, and research is therefore directed toward their solution. This procedure has a very important consequence: interpretative difficulties are avoided when problem recognition derives from pattern-recognition studies at the second- or third-order derivative level or scale of analysis. Primary or first-order observations refer to observations recorded during excavation or result- ing from descriptive synthesis from data generated for a particular bounded segment of the archaeolog- ical record. Second- or third-order derivative pat-
terning describes how the primary data or f~st-order observations interactively pattern with other infor- mation such as environmental, temporal, associa- tional, or comparative relative to other archaeological sites or scales of synthesis. Problems derive from the recognition of patterning that is unsuspected, new, and/or not understood. Problems are therefore at least partially recognizable in terms of synthesis from first-order observations. In other words, the conven- tions for looking are also the terms in which prob- lems may be recognized. The problem is at least partially identij7ed in terms of operationally defined observational units. One does not have to search for observations germane to the problem. The observa- tions that permitted the recognition of the problem will be at least partially germane to the arguments developed as one works toward a solution (see Bin- ford 1989:34-39, 231-234). Second, multiple researchers working on a given problem are likely to learn things that advance the knowledge base for
multiple investigators. Change can occur through knowledge growth rather than from cognitive style shifting as suggested by Kuhn (1964).
A Scientific-Humanistic Pastiche Leads to
Why do researchers appropriate the questions from one domain of experience and attempt to mine the archaeological record for answers when there is no compelling linkage between archaeological data and their questions? Why do archaeologists reject good descriptions of system dynamics because they claim that no unambiguous linkage exists between the described dynamics and the properties of the archae- ological record they are examining? It is those very dynamics that, at least partially, influence the struc- ture and content of some components of the archae- ological record! As I suggested elsewhere in this manuscript, the answers to the preceding questions are found in the different approaches to learning that characterize science and the humanities. I suggest that most of the interpretative conventions in archae- ology are a legacy of the humanistic approach that for many years has dominated anthropology and archaeology. Arguments about the utility of more recently posed conventions also are initiated by humanists who see "theories" as potential interpre- tative devices.
The empirical generalizations that I and others have presented often have been regarded as theories,
although, for example, there is nothing theoretical about the behaviors I observed among the Nunamiut. My descriptions of Nunamiut behavior and the terms that Iused to describe it are not theoretical. Nonethe- less, humanists (as outlined by Trigger 19 89:22)- in search of what Taylor would call "philosophy" as a source of interpretative conventions-have seized upon these observations and descriptions as new con- ventions. Not surprisingly, when used as interpreta- tive conventions, my empirical generalizations appeared to have no unambiguous referents in lithic assemblages from the midwest of North America! It was concluded by the humanistic critics, therefore, that "the relationship between curation, however defined, and scarcity-induced economizing behav- ior will never be totally clear, as tool conservation is a constituent of both" (Ode11 1996:76).
I reject the idea of equifinality (for an excellent discussion, see Rogers 2000). A belief in it is fos- tered by a false belief in single causes, or by seman- tically changing the scale of discourse regarding a particular variable. For example, "scarcity-induced economizing behavior" is abstract and must be made concrete by identifying the material elements or phe- nomena that are "scarce," and permitting the phrase "other things being equal" to have some empirical content that must be controlled in order to be able to see unambiguously another conditioning factor. Ambiguity of this kind is a general property of expe- rience and it is the fundamental reason why scien- tists do controlled research. Additionally, cognitive ambiguity has prompted scientists to become remarkably skillful in the design of experiments in which some variables can be held constant or con-
sidered irrelevant in order that the consequences of the interaction among a finite set of variables cho- sen for study can be observed unambiguously. This approach to learning demonstrates that scientific tac- tics are designed to insure learning, whereas the search for accurate and unambiguous interpretative conventions is a search for static and frequently vac- uous "truths" about reality.
I do not wish to imply that one cannot begin with a higher-level proposition and proceed to do pro- ductive science. Developed sciences do this all the time. They can reason from theories originally devel- oped to account for one set of observations, to addi- tional observations that-if the theory is correct-should be demonstrable. I have already dis- cussed the issue of middle-range research and the fea-
sibility of linlung "higher-level" propositions-or, more accurately, propositions for which there are no reliable links to observational data-to well-organized research in the present on living systems or sys- tems described from the perspective of either participants or observers of such systems (see Bin- ford 1981a:21-34). I am not suggesting that good work cannot come from exploring the experiential world with ideas not yet linked to that world. My point is that being a consumer of ideas rather than adopting an investigatory role insures that ideas will not act as portals to further knowledge but will instead be accepted or rejected for the wrong rea-
sons. Such a course will result in the imposition upon the archaeological record of an endless sequence of different interpretive conventions rather than to the growth of knowledge about the forces and factors that created that record over time.
Theory is not something one brings to data. The- ory is developed to explain relational patterns among data that are analytically generated among different observational domains or data sets. Arguments regarding the patterning will, therefore, be at least partially phrased in terms of the prior conventions for data production. The initial link to subsequent statements about the datais, of necessity, in terms of the operational definitions used in data production. This tautology is overcome when one takes the result- ing data, or parts thereof, to the pattern recognition stage of analysis (see Binford 1989:34-39,23 1-234, 394; 2001:54,123,182,243-44,316-317,374-375, 46 1-47 1). Pattern recognition studies expose
selected properties or characteristics of the archae- ological record (which at this point can correctly be referred to as archaeological data) to other indepen- dently reasoned and warranted bodies of data. Other things being equal, when there is demonstrable rela- tional patterning between the two or more data sets germane to the past, the relationships represent direct evidence of some systematic interaction between the two domains of observation in the organization of past dynamics. These are direct clues to cause and the dynamic operation of past processes.
At such a juncture, a scientist may engage in the- ory building designed to argue the way the causal or processual mechanisms worked to bring into being thedemonstrablepatterning. When these intellectual steps are followed, there is never a problem about what to look at when developing, elaborating, or exploring a theory. The theory is about the observa- tional units and their patterned mechanisms of inter- action with other observational units. An archaeol- ogist builds theory about second-order derivative patterning and not about the primary observations made at an archaeological site. This a dramatically different view of what we do than that cited earlier from Ode11 (1993:118).
I began this essay by stating a principle: "I cannot make ajudgment about the rationality of the way you intend to proceed until you tell me what you are try- ing to accomplish." Accepting this proposition, I then explored some fundamental properties of science, the most important of which was the recognition that the problems for which solutions are sought come from a detailed study of the subject matter of the science. In the case of archaeology, the subject matter is the archaeological record. I then asked if this viewpoint was widely held in contemporary archaeology. I have answered this question with a strong no, and I emphasized that the problem most researchers address is simply how to "interpret" their data. With that goal in mind, they survey other disciplines, review their own personal experiences and store of knowledge, or refer to writers of so-called archaeological theory for inspiration about the units of "meaning" they can impose on their data. This method does not advance
the study of the archaeological record nor does it translate the archaeological record into accurate statements about the past.
Science advances through the study of its subject matter and the invention of theories that guide fur- ther exploration of the subject matter in a search for additional clues to causal processes. This means that archaeologists should develop their research prob- lems through analytical study of the archaeological record. Problem solving is directed by theory that is developed to explain how the archaeological record got to be the way it appears to be. This means that archaeologists must work at a more comprehensive analytically designed scale of synthesis than the "ele- mental" or directly observedproperties of the archae- ological record. They must analyze data for patterning and must display such observations against different frames of reference (see Binford
1999 and 2001) in search of patterning. By frames of reference I mean systematically organized bodies of data consisting of phenomena that are not, strictly speaking, archaeological (for example, environ- mental, ethnographic, temporal, geographic, or some other domain of synthesized data).
It is the interaction between facts of the archeo- logical record and other systematically organized bodies of knowledge-synthesized at comparable scales-that yields patterning that provides and defines the nature of appropriate problems for research. At this point, the linkage between the prob- lem and at least some of the archaeological charac- teristics germane to the problem are already known, so explanatory investigation can begin. Theory build- ing is focused on the synergistic relationships iden- tifiable through pattern-recognition studies using selected properties of the archaeological record. This inductive inquiry is very different from the imposi- tion of borrowed "theory" on the archaeological record, as advocated by Taylor in the 1940s and, regrettably, still routinely practiced today.
Acknowledgments. Four people have provided extraordinary intellectual help to me during the preparation of this manu- script: Tera Bond-Freeman, Russell Gould, Amber Johnson, and Tim Kohler have all made substantive suggestions as to how to best present the ideas developed in this manuscript. Nancy M. Stone edited an early version of the manuscript for style and the use of the English language. I am most grateful to all five for their interest and assistance.
Binford, L. R.
1968 Archaeological Perspectives. In New Perspectives in Archeology, edited by S. R. Binford and L. R. Binford, pp. 5-32. Aldine Publishing Company, Chicago.
1973 Interassemblage Variability: The Mousterian and the 'Functional Argument.' In The Explanation of Culture Change: Models in Prehistory, edited by C. Renfrew, pp. 227-254. Duckworth, London.
1978 Nunamiut Ethnoarchaeology. Academic Press, New York.
1979 Organization and Formation Processes: Looking at Curated Technologies. Jounzal ofAnthropological Research 35:172-197.
1980 Willow Smoke and Dogs' Tails: Hunter-Gatherer Set- tlement Systems and Archaeological Site Formation. American Antiquity 45: 1-17.
1981a Bones: Ancient Men and Modern Myths. Academic Press, New York. 1981b Behavioral Archaeology and the "Pompeii Premise."
Journal ofAnthropologica1 Research 37: 195-208. 1983 Working at Archaeology. Academic Press, New York. 1987 Researching Ambiguity: Frames of Reference and Site
Structure. In Method and Theor)> forActivity Area Research: An Ethnoarchaeological Approach, edited by S. Kent, pp. 449-512. Columbia University Press, New York.
1989 Debating Archaeology. Academic Press, New York. 1999 Time as a Clue to Cause. Proceedings of the British Academy, 101: 1-35. The British Academy, London. 2001 Constructing Frames of Reference: An Analytical Method for Archaeological Theory Building Using Ethno-
graphic and Environmental Data Sets. University of Cali- fornia Press, Berkeley.
Binford, L. R., and S. R. Binford 1966 A Preliminary Analysis of Functional Variability in the Mousterian of Levallois Facies. American Anthropologist 68:238-295.
Clark, G. A. (editor) 1991 Perspectives on the Past: Theoretical Biases in Mediter- ranean Hunter-Gatherer Research. University of Pennsyl- vania Press, Philadelphia. Gamble, C. 1999 The Palaeolithic societies of Europe. Cambridge Uni- versity Press, Cambridge. Gero, J. M., and M. W. Conkey (editors) 1991 Engendering Archaeology: Women and Prehistory. Blackwell, Oxford. Hawkes, C. 1954 ArcheologicalTheory andMethods: Some Suggestions from the Old World. American Anthropologist 56: 155-168. Hayden, B. 1975 Curation: Old and New. In Primitive Art and Technol- ogy, edited by J. S. Raymond, B. Loveseth, C. Arnold, and
G. Reardon, pp. 47-59. University of Calgary, Alberta. Jelinek, A. J.
1976 Form, Function, and Style in Lithic Analysis. In Cu1tural Change and Continuity: Essays in Honor of James Bennett Grifin, edited by E. Cleland, pp. 19-33. Academic Press, New York.
Kuhn, T. 1964 The Structure of Scient$ic Revolutions. University of Chicago Press, Chicago. Merton, R. K. 1968 Social Theory and Social Structure. Free Press, New York. Miller, D. W. 1985 Popper Selections. Princeton University Press, Prince- ton. Nash, S. E.
1996 Is Curation a Useful Heuristic? In Stone Tools: Theo- retical Insights into Human Prehistory, edited by G. H. Odell, pp. 81-89. Plenum Press, New York.
O'Brien, M. J., and T. D. Holland
1990 Variation, Selection, and theArchaeologica.1 Record. In Archaeological Method and Theory, Vol. 2, edited by M. B. Schiffer, pp. 31-79. University of Arizona Press, Tucson.
O'Brien, M. J., and R.L. Lyman 2000 Applying Evolutionary Archaeology: A Systematic Approach. Plenum Press, New York. O'Connell, J. F. 1995 Ethnoarchaeology Needs a General Theory of Behav- ior. Journal ofArchaeological Research 3: 205-255. Odell, G. H.
1981 The Morphological Express at Function Junction: Searching for Meaning in Lithic Tool Types. Journal of Anthropological Research 37:3 19-342.
1993 A North American Perspective on Recent Archaeolog- ical Stone Tool Research. Palimpsesto: Revista de Arque- ologia 3:109-122.
1996 Economizing Behavior and the Concept of "Curation." In Stone Tools: Theoretical Insights into Human Prehistory, edited by G. H. Odell, pp. 51-80. Plenum Press, New York.
Odell, G. H. (editor) 1996 Stone Tools: Theoretical Insights into Human Prehis- tory. Plenum Press, New York. Preucel, R. W., and I. Hodder 1996 Communicating Present Pasts. In Contemporary Archaeology in Theory: A Reader, edited by R.W. Preucel
and I. Hodder, pp. 3-20. Blackwell Publishers, Oxford. Renfrew, C., and P. Bahn 1991 Archaeology: Theories, Methods and Practice. Thames and Hudson, Inc. New York. Rogers, A. R. 2000 OnEquifinality in FaunalAnalysis.AmericanAntiqz~ity 65:709-723. Shott, M. J.
1996 Innovation and Selection in Prehistory: A Case Study in the AmericanBottom. In Stone Tools: Theoretical Insights into Human Prehistory, edited by G. H. Odell, pp. 279-315. Plenum Press, New York.
Spaulding, A. C. 1960 The Dimensions of Archaeology. In Essays in the Sci- ence of Culture: In Honor of Leslie A. White, edited by G.
E. Dole and R. L Carneiro, pp. 437456. T. Y. Crowell Co., New York. Taylor, W. W. 1948 A Study of Archaeology. Memoirs of the American Anthropological Association No. 69, Menasha.
1972 Old Wine and New Skins: A Contemporary Parable. In ConternporaryArchaeology,edited by M. Leone,pp. 28-33. University of Southern Illinois Press, Carbondale.
Trigger, B. G. 1989 A History ofArchaeologica1 Thought. Cambridge Uni- versity Press, Cambridge. Wylie, A.
1989 The Interpretive Dilemma. In Critical Traditions in Con- temporary Archaeology, edited by V.Pinsky and A. Wylie, pp. 18-27. Cambridge University Press, Cambridge.
1. See also many of the papers in Clark (1991).
- Taylor's 1943 doctoral dissertation at Harvard was later published as Memoir number 69 of the American Anthropological Association (1948).
- To most humanists and many social scientists, this lack of fit between what is observed and what is conceptually organized in a theory is similar to what Merton (1968) meant in his call for middle-range theory. My view was that the abstract character of theory in sociology arose because research problems were not derived from the subject matter of sociology but, instead, from an applied interest in chang- ing soclety for the better. My use of the terms "middle-range research" and "theory building" referred to the problems
arising from the study of the properties of the archaeological record itself and the development of arguments regarding the referents in dynamic terms for static observations (see Binford 1983:3-10, 411-433).
It is such stipulated conventions that I suggested should be "tested" through the use of deductive reasoning in the early 1960s. Unfortunately, this suggestion was raised by some authors to the position of a general principle for doing "science." That of course was nonsense.
This claim is a mantra of the critics of science and defenders of humanistic endeavors (see Renfrew and Bahn 1991:432; Wylie 1989:19). In one very real sense the claim is accurate up to a point, but the conclusion drawn is false. Observations may be theoretically grounded, but the pattern- ing manifested by those observations relative to observations of other phenomena is not necessarily dependent upon the same theories, or upon any formal theory at all. It is only when observations are interpreted as I have outlined here-if the linkage is accepted-that they become theory-dependent.
See Taylor's (1972) article entitled "Old Wine and New Skins: A Contemporary Parable" for his view of intel- lectual developments in archaeology in the 1960s.
- I addressed this same problem some time ago in my discussion of "behavioral archaeology" (Binford 1981b). The issue of behavioral reconstruction, motive, and ambiguity is strongly restated in much of the recent lithics literature.
- While ambiguous, I read this statement as implying that one expects the "theoretical structures" to have been "created" independently of the elemental observations. Only then would it be reasonable to expect any "support" to be meaningful!
- See Binford (1989:34-36, 267-277) for earlier discus- sions of issues bearing on this discussion.
- I have treated the issue of the differences between what we see and how such perceptions are conceptualized, relative to what interpretations they may seem to justify, numerous times (Binford 1983:48, 57-58, 240, 378, 403, 406, 420-421; 1989:3440).
Received May 25, 2000; Revised January 15, 2001;
Accepted February 26, 2001.