Go to Semiotic Review of Books Home Page
Go to SRB Highlights
Go to SRB Archives
SRB

SRB Highlights

This editorial appeared in Volume 9 (3) of The Semiotic Review of Books.

Human Cultural Adaptation and the Double Life of Models

by Bradd Shore

All scientists work through models in one way or another. Indeed, the basic goal of science may be thought of as modeling aspects of the world which are not otherwise comprehensible. Unless one takes the metaphysical view that the scientist's or mathematician's models are discoveries of the actual design-templates of a natural order or, religiously framed, the contents of God's mind, it is generally assumed that scientists' models are merely tools, heuristic constructs. But the human sciences are different, for in both psychology and anthropology, it is fair to say that models have a double status. They are, as in the other sciences, tools for studying the world. But in anthropology as well as in cognitive psychology, models of various sorts are also the object of study.

The idea that all humans understand the world through models or schemas has a distinguished intellectual pedigree in both cognitive psychology and in philosophy of mind. Kant proposed early on that human understanding involved a set of innate categories of the mind, innate mental schemas. In his 1932 classic study Remembering, Cambridge psychologist Frederick Bartlett invoked the notion of mental schemas as a way to understand how people remember by a kind of systematic distortion of input, bringing perceptions into line with already existing mental models. Meanwhile, across the Atlantic at Columbia University, at virtually the same moment, Ruth Benedict was using the parallel notion of "configurations" or "patterns of culture" to understand how traits of one society borrowed by another were systematically distorted or modified in the direction of a pre-existing model in the borrower culture. In her doctoral dissertation she demonstrated the power of such cultural templates internal to a group to transform what appeared at first to be a institution shared by many Native American groups -- the guardian spirit. In fact Bartlett's research was aimed at providing a psychological framework for understanding this very processes of cultural transmission and change that Benedict was trying to understand. And they both landed squarely on the notion of models to describe a general design feature of both individual cognition and culture.

Modeling is a good candidate for a basic activity of mind. The human brain is a virtuoso modeler, constructing models from external input, reading models already in the environment, storing models for future use and accessing pre-stored models as interpretive tools. Primate evolution left us with a dramatically expanded forebrain, language areas and association cortex. The old reptilian and mammalian brain cores didn't go away of course. But they were literally overrun by the brain's newly expanded cortex. This meant that most of our responses to the world would be inevitably filtered through layers of complex symbolic models. The brain is constantly generating models in the form of electrochemical patterns -- neural networks. The brain is also continually monitoring the external world through its sensory portals, seeking patterns in the world to model neurally. When our brains can match external patterns with those already stored in memory, we get "meaning."

All modeling simplifies input in the interest of intelligibility and the reduction of complexity. Complexity can be reduced in many ways. A change in scale (as in miniaturization or enlargement) can bring something into focus. Intelligibility can be enhanced by the reduction of detail, or the elimination of perceptible qualities likecolor, texture, movement or smell, or by a sharpening of features normally experienced as fuzzy. Or a complex whole can be modeled by one of its parts as a stand-in -- a salient exemplar, a best case or some other stand-in relationship, producing what psychologists call "prototype effects."

Evolution endowed our species with the ability to project our internal models into material forms. The sapient hominid is also Homo faber -- maker of artifacts. Our penchant for physical modeling is built in to the evolution of our hands, with their opposable thumbs and precision grip. With the help of a big forebrain, humans extended these manipulative capabilities through the invention of tools, both physical and symbolic. Many of these artifacts take the form of objects, both natural (such as the sun as a way of keeping time) and constructed (a clock). But the home-made models that people project from their minds into the world are not limited to things. Public models can also take less concrete forms like spoken words, or repeatable behavior patterns such as a handshake or a bow or a wink. All such models are projections of mind into the public world where they become objects for mutual orientation, the familiar symbolic furnishings of human environments.

It is not surprising that specialists in studying and understanding the world employ modeling as a basic tool. Scientists are in the business of constructing self-conscious models of reality. Scientific models tend to be called "theories." Sometimes, as in the case of molecular models, scientific models take a very concrete physical form. In other cases they take the shape of algorithms or mathematical equations. Statistical patterns are still another way to model the natural world. And finally, the typical case study is a particularly effective kind of model for revealing the characteristic qualitative features of something under study not brought out by statistical models. Case studies are especially useful in modeling certain aspects of human life.

Just like natural scientists, social scientists also study models and produce models. Rather than atoms, molecules, or cells, social scientists observe family patterns, personality structures, cycles of economic behavior, patterns of language use or cultural patterns that distinguish one community from another. In a sense social scientists create (academic) models or theories by observing behavior that is already modeled by those being observed a paradox to which I will return in a moment. If human behavior were not already patterned, social scientists would not have anything to discover. But human life appears to be so thoroughly modeled in so many ways, that social scientists are unlikely ever to run out of things to study.

Artists and designers create new models out of old ones. Artists model the world of sight and sound, word and movement. Good art produces pleasure and insight by crystallizing significant forms in every sensory mode. Aesthetic pleasure is probably conditioned by the human penchant for finding emotional and intellectual meaning in models. "Pure art" reflects the joy of modeling for its own sake.

Models are also a useful way to approach the concept of culture. Culture may be thought of as a vast stock of models used by communities to understand and manipulate the world in a more-or-less coordinated way. This produces the paradox described at the beginning of this essay that cognitive psychologists and cognitive anthropologists both study models and study with models. We make models about people's models, and so we tend to be at pains to distinguish scientific theories or models (our tools) from what we call folk-theories or models (their tools and ourobjects of study). It can get pretty tricky, but that is in the nature of the enterprise.

Adaptive intelligence employs many kinds of models, only some of which are cultural. Some models, like the perception of the basic facial gestalt, or certain basic emotion expressions are shared by most or all humans, and are good candidates for hard-wired mental models. At the other end of the spectrum are personal models, the idiosyncratic dimensions of experience sometimes encountered in dreams, and sometimes projected into public forms as highly personal expressions. Cultural models are subject to social coordination and inter-subjective sharing. We eat the same food, greet the same kind of people in the same kind of way, fall in love and express that love in roughly predictable and recognizable ways. To the extent that our external experiences are similarly constrained in this way, our mental representations of life experiences, and memories and our predispositions and expectations of life will tend to overlap. Such shared-in models are conventional instituted models.

Cultural models serve a wide variety of cognitive and social functions which may be summed up as orientation, conceptualization, communication, and control. Models make possible our orientation to the world and the world of time and space and social relations. Models allow conceptualization, making it possible for us to remember, to think and even to feel. Models enable communication of these thoughts, memories and feelings to others. And models of all sorts are used to manipulate and control our social, cognitive and physical environments.

With these general features of cultural models in mind, we can now turn to some basic distinctions between different types of models. In terms of their location, models come in two basic types: models-in-the-world and models-in-the-mind. Although this distinction is often not clear in discussions of models, it is very important. Models-in-the world are "instituted models," realized as public artifacts. Such artifacts include material objects such as paintings or model-cars. They also include less concrete institutions like gestures, words, rituals and everyday conventional routines like "sales meetings" or "breakfast." Mental models have a way of finding their way into concrete forms. Explicit representations like words, or paintings, dramas or gestures are literally re-presentations of mental images, the mind made matter. And conversely, many of our most intimate thoughts and feelings are modeled on something or someone in the world.

A second important distinction is between users' models and generalized models. This is an issue of whose perspective is modeled. A generalized model attempts to represent something simultaneously from many points of view, or from no particular point of view. A good example is a city map. The actual point of view seem to be that of a bird equipped with special vision capable of seeing street outlines and general outline contours but no buildings or other objects. This kind of map rises above any normal person's point of view in order to provide a generalized perspective usable by anyone anywhere in the range of the map. It tries to coordinate lots of potential perspectives.

By contrast a users' model takes the point of view of a particular individual situated somewhere on that map. When someone asks for directions, you can respond in terms of either kind of model. If you use perspective-independent coordinates like "east" and "west" in your directions, you are implicitly using a general model. But such models are hard for most people to use. More common in giving directions are user's models inwhich the direction-giver places herself mentally in the shoes of the direction-asker and walks or rides him from landmark to landmark, using markers like traffic lights, street names, and user-specific coordinates like left and right.

The distinction between users and generalized models has great significance for both commercial design and the problems of how our scientific models represent others' reality. While user models are much more intuitive for individuals, many products and environments appear to be designed from generalized models, i.e., an overhead point of view (the way conventional architect's renderings or mechanical diagrams conceptualize an environment or design object). The same goes for anthropological accounts of culture, which often use generalized models like general kin terminologies or overhead site maps to represent "culture" from a birds-eye-view perspective alien to the individuals whose knowledge we are trying to represent.

A third distinction is between descriptive models and prescriptive models.

A detailed description of a landscape, a sketch of the landscape and a photograph of that landscape are all models of a pre-existing reality. They are descriptive models. By contrast, a landscape architect's plans for a new park and his mental images that he used to imagine the park are both models for an intended reality. They are prescriptive models, or templates. The fact that we call both of these things "models" suggests their intimate links, the transformations models undergo as we take in what is already there, transform it mentally and then project a new vision back onto the world. Models of reality are often in transit, on the way to being converted into new models for reality. And vice versa.

This give-and-take between prescriptive and descriptive models has something to do with the key role of analogy in creative intelligence. Cognitive psychologists, poets and linguists have long suspected that our ability to make analogies is a basic form of human intelligence. Analogies use what we already know as a bridge to something new. This is why we find ourselves grasping for similes and metaphors when we struggle to comprehend something unfamiliar or hard to understand. An analogy takes a model of something more familiar and turns it into a model for something less familiar. In this way what is becomes what might be. And under the right conditions what might be can eventually become what is. And so on. We know this process as creativity.

There is also an important distinction between mechanical models and statistical models. In a well-known essay "Social Structure," Claude Lévi-Strauss distinguishes mechanical from statistical models in terms of the relation between the scale of the model and that of the phenomena:

According to the nature of these phenomena, it becomes possible or impossible to build a model, the elements of which are on the same scale as the phenomena themselves. A model the elements of which are on the same scale as the phenomena will be called a "mechanical model"; when the elements of the model are on a different scale, we shall be dealing with a statistical model. (Lévi-Strauss 1967: 275-76)
Marriage norms which specified an individual's preferred mate in terms of actual group or category membership (e.g., clan endogamy or cross-cousin marriage) regulated marriage through a mechanical model. On the other hand where marriage is determined by such probabilistic factors as the size of the primary or secondary groups of likely mates, or the degree of social fluidity, or the chance that two availablepotential mates would be in a position to meet, we have a statistical or probabilistic model regulating marriage. This distinction also accounts for the nature of how social phenomena are modeled and studied by social scientists. If we want to study the French family, we could employ a mechanical model by presenting a vivid and detailed case study, a description of several representative French families. Alternately we could construct a statistical model of the family by collecting a sample of characteristics of a cross section from a large number of French families and model French family life in the form of a statistical profile of behavior distributions.

Both kinds of model simplify the French family in the interest of intelligibility, but in different ways. The mechanical model is qualitatively rich, descriptively subtle but uses a very restricted sample. The statistical model is qualitatively reduced (particular people and events are reduced to quantitative instances of general types), but broadly representative of an entire population. Mechanical and statistical models are both powerful ways to model reality, and have complementary advantages and disadvantages. In terms of understanding complex behavior and events, there is an interesting disparity between the kinds of models scientists like to think with and those which ordinary people prefer to use in understanding and predicting.

Whereas empirical science constructs mechanical models of behavior as hypotheses or laws, the data which is generally preferred to test such hypotheses constitutes statistical models based upon sophisticated sampling techniques to guarantee representativeness over a large data sample, so that the data models reality, in Lévi-Strauss's terms, at a different scale from the initial hypothesis or the final law. On the other hand ordinary people often base their inferences and hypotheses on particular vivid cases -- mechanical models -- whose credibility is derived less from their statistical typicality than from one or another prototype effect (e.g., a good exemplar, a memorable case, the first instance, the most recent instance, a personally salient example, etc.).

A final distinction, and one closely related to that between mechanical and statistical models, brings us to our concluding point. When culture is conceptualized as models, "culture" no longer suggests the discrete and bounded entity which our folk models suggest. Instead of a categorical model we have an epidemiological model of culture, a distributed and unevenly shared system of instituted models and their mental analogs. Epidemiological models suggest gradients, clines and other features of statistical distributions. They lack the precise boundaries of categorical models. So cultural communities are always internally differentiated and cultural models are characterized by different degrees of sharing.

The trouble is that epidemiological models are not especially good to think with. For many people they are unimaginable. But this does not mean they are not true. The result is an ironic modeling problem with cultural models: a tension between the stochastic character of culture as a distributed system and the rather neat categorical folk models that most of us (including anthropologists) seem to be comfortable with. The human mind appears to have evolved with a strong bias towards understanding the world through readily communicable categorical models, while many of the phenomena it is trying to understand are better suited to being represented as less clear-cut epidemiological models. Which is why human understanding is inherently a constructive activity, and why some complex concepts (like "culture" or "race") aresubject to habitual misrepresentation.

References

Bartlett, F.C. 1932 Remembering. Cambridge: Cambridge University Press.

D'Andrade, Roy 1995 The Development of Cognitive Anthropology.Cambridge and New York: Cambridge University Press.

Holland, Dorothy , and Naomi Quinn (eds.) 1987 Cultural Models in Language and Thought. Cambridge: Cambridge University Press.

Hutchins, Edwin 1995 Cognition in the Wild. Cambridge, MA.: MIT Press.

Johnson-Laird, P.N. 1983 Mental Models: Towards a Cognitive Science of Language, Inference, and Consciousness. Cambridge: Harvard University Press.

Lévi-Strauss, Claude 1967 Social structure. In Structural Anthropology. New York: Anchor Books.

Shore, Bradd 1996 Culture in Mind: Cognition, Culture and the Problem of Meaning. Oxford and New York: Oxford University Press.

Spradley, James (ed.) 1972 Culture and Cognition: Rules, Maps and Plans. New York: Harper and Row.

Strauss, Claudia and Naomi Quinn 1997 A Cognitive Theory of Culture. Cambridge and New York: Cambridge University Press.

Bradd Shore is Professor of Anthropology at Emory University (GA)


Go to Semiotic Review of Books Home Page
Go to SRB Highlights
Go to SRB Archives