The Evolution of Cognition

Abstract: With cultural evolution new processes of thought appear. Abstraction is universal, but rationalization first appeared in ancient Greece, theorization in Renaissance Italy, and model building in twentieth-century Europe. These processes employ the methods of metaphor, metalingual definition, algorithm, and control, respectively. The intellectual and practical achievements of populations guided by the several processes and exploiting the different mechanisms differ so greatly as to warrant separation into cultural ranks. The fourth rank is not completely formed, while regions of the world and parts of every population continue to operate by the processes of earlier ranks.

1 Evolution in Biology and Culture

Evolution is a comparative discipline. In biology one compares species, their anatomy, physiology, behavior, and ecology. In culture one compares social structure, economy, technology, and modes of thinking and feeling. The capacity for comparative analysis requires ways of describing the objects and phenomena being compared. The descriptive repertoire of biology is better suited to its subject matter than the descriptive repertoire of cultural studies. The purpose of this essay is to introduce a descriptive tool for cultural studies, the concept of cognitive rank.

The material and organizational achievements of cultures of different cognitive rank are apparent to any observer. The increase in knowledge can be measured in crude ways: more facts in books and other media. What is not obvious is that not only the contents but also the processes of thought changed repeatedly. We call the change in process cognitive evolution.

The concept of cognitive rank tells us nothing about why culture evolves, nor about why one society evolves and another doesn't, but it does allow us to conceptualize the difference between a more and a less evolved culture. It allows us to see how the difference is constructed, without telling us why that difference was constructed. That "why," important as it is, is beyond the scope of this essay--though we note our sympathy with Csikszentmihalyi's (1990) argument that much of the "why" is rooted in the mind's pleasure in and craving for complexity: culture evolves because people enjoy performing ever more complex tasks.

Cognitive rank is about the conceptual tools a culture has for the elaboration of abstract thought. People in all cultures reason abstractly, for that capacity inheres in language (see Benzon and Hays, 1988). But cultures differ in their capacity to order and generate abstract concepts. There are abstract ideas that cannot occur in the thinking of an Eskimo, or even a literate Florentine, or, for that matter, Darwin, Freud or Einstein. And, since there is no reason to believe that cognitive evolution has, or is about to, run its course, abstract ideas will be created which cannot occur in the most sophisticated systems of thought in current use.

The question of cognitive evolution is motivated by the observation that, of the many ways in which cultures differ from one another, complexity is one of the most obvious. An extensive body of research (Levinson & Malone, 1980) shows that when cultures are compared with respect to such things as number of levels of political organization, population of the largest community or of the whole polity, division of labor (number of crafts, technological diversity), external commerce, means of enforcing justice and equity, social stratification (caste and class), character of religion, legitimation of government, use of machinery, sources of power, kinds and amount of transportation, extent and quantity of communication, kind and amount of capital investment, some cultures have more than others. Further, the cultures which are more complex in one arena tend to be more complex in the others as well. The most obvious route to such differences is evolution: the more complex cultures emerged from the less complex by some evolutionary process.

Cognitive evolution, however, is not only about complexity; it is about sophistication as well. The arithmetic procedures we have been using since the Renaissance aren't more complex than those used by the Greeks and the Romans. They are, in fact, simpler. They are also more sophisticated and thereby much more effective. The cognitive processes of a culture of a high rank are more sophisticated than those of a culture of lower rank.

In dealing with this complexity and sophistication we need to avoid the racism which characterized nineteenth century thought on cultural evolution. Racism was intellectually respectable then and nineteenth-century thinkers really had no intellectual alternative. They thought that cultural evolution required better thinking, as indeed it does, and could ascribe improved thinking only to better brains. They had no theory of information, of learning, and were nowhere near the essential theory of learning to learn, which is the crucial link between cognition and cultural evolution.

The Nineteenth-century racists could not distinguish between culture and biology and assumed that cultural differences had to be explained by biological differences. But culture is a realm of being unto itself. Its phenomena cannot be reduced to biological phenomena. Hence a theory of cultural evolution need not imply anything about the relative merits of one gene pool over another. Cultural evolution isn't about biology; it is about culture. Cognitive evolution, as we see it, is a component of cultural evolution.

Yet there must be a biological substrate for culture. The possibility of culture, perhaps even its necessity (see Geertz, p. 1973), is inherent in human biology. The brain is the most obvious locus for our capacity for culture. The brain makes language possible and, by implication, culture. It is through language that humans elaborate their perceptual, motivational, navigational, and manipulative abilities into the complex mechanisms of culture. Language makes us capable of abstract thought, of conceiving, caring about, and inculcating such things as mana, gravity, justice, electromagnetism, and so forth. All cultures deal with abstractions but, as a culture evolves, the diversity and depth of its abstractions increases.

We are here concerned with characterizing this growth in abstractive power. This growth does not require any biological change. It requires only language. Any language which is a human language contains within itself the mechanisms necessary for evolutionary growth. In fact, it is probably more subtle than this. Whatever it is about the brain which makes language possible, that is the engine behind culture (see Benzon & Hays, 1988). That is a subtlety, however, which does not affect the main thrust of our argument.

2 Ontology, Abstraction, and Behavioral Mode

The first order of business is to set forth some of the conceptual machinery we need. We do that in this section, leaving a discussion of cognitive evolution to the next section. We begin with a discussion of ontology. Ontology is a branch of philosophy, but our interest is not philosophical. It is cognitive; we are interested in ontological thinking. For the basic nature of ontological categorization changes from cognitive rank to cognitive rank. From ontology we move to abstraction, thought about matters which cannot be seen, touched, smelled, etc., and the manipulation of abstraction by metaphor and rationalization. (Later we introduce some other methods of manipulation.) We conclude this section with a discussion of neural implementation.

2.1 Ontology

• (1) The table was made of wood.
  
• (2) The table was made of linguini.
  
• (3) The table was made of gossamer thought and airy whimsy.
  

The first two propositions are perfectly sensible, though the existence of a table of which (2) is true is very unlikely. But (3) doesn't make sense. "Thought" and "whimsy" aren't things from which one can build tables. A table is an inanimate concrete object. Thought and whimsy are mental events. Concrete objects cannot subsist in mental events.

Note, however, that what "makes sense" is relative to a culture. In one culture "The gods told me to act" is ordinary--everyone says it at one time or another; in another, it is a mark of sanctity; and in a third, a sign of insanity. The problem of the iron horse was that it was capable of autonomous motion, a predicate which did not "sensibly" apply to machines. Once that strange conjunction of machine and motion had become comfortable terms such as "locomotive" and "automobile" appeared to "naturalize" the new conception.

It is easy to think about these conceptual structures in terms of the roles objects are permitted to take in events and actions (see Benzon, 1986). "Robin," "nasturtium," and "sea cucumber" can all be the subject to verbs such as "grow," "breed," and "die." Things which can play this role with respect to these events are ontologically different from things which cannot, such as "rock," "cloud," and "creek." This pattern of participation distinguishes living things from inanimate things. Things capable of sensation (see, hear, touch, etc.) and autonomous locomotion (walk, swim, fly, etc.) are ontologically different froms things which cannot. By this criterion animals are differentiated from plants. That is to say, animals can play the agent or subject role with respect to verbs such as "hear," "run," "eat," and "look." Things which can speak and think are, in turn, different from those which cannot. By this criterion, human beings are differentiated from animals.

It is a mistake, however, to talk about the ontology of a particular system of thought. Beyond the simplest cultures, we have to talk about the culture's mechanisms for generating ontologies, which will generate an ontology for each specialized domain. All these ontologies will be made from the same model, but they will be different and only the specialist will grasp the ontology. The auto mechanic's sensorimotor experience of mechanical devices is richer than the gardener's, whose sensorimotor experience of plants is richer than the mechanic's. The ontological categories each constructs in his or her domain will reflect that sensorimotor richness. Each craft probably generates an ontology, which is part of growing up a craftsworker.

When we consider cognitive rank we will show that the mechanisms for generating new ontologies change from rank to rank, from implicit and uncontrollable to explicit and controlled.

2.2 Abstraction and Metaphor

Although the mechanisms are certainly not well understood, many things can be identified by sensorimotor characteristics. Things such as a red maple, a horse's gallop, rain, fish, a mountain, the technique for tieing a knot, and so forth, can be characterized by such things as color, form, texture, sound, smell, muscular efforts, and so forth. Much in human thought, however, is not readily characterized by the sensible and the manipulable. Honor, gravity, mana, esprit, these things are abstract. And they dominate our cultural life.

Metaphor is our most basic way of manipulating abstractions (Benzon & Hays, 1987). Consider, for example, the statement that "Achilles is a lion in battle." In a standard terminology (Richards, 1936), Achilles is the tenor , lion the vehicle , and the style they share is the ground , of the metaphor. In one view that metaphor is just a fancy way of saying: "Achilles is courageous." In this simple sentence the ground of the metaphor is directly conveyed rather than indirectly indicated. Consider, however, the situation of someone who doesn't know the word "courage"--which is, after all, an abstract concept--but who is familiar with the behavior of lions. Through the use of a metaphor that person can indicate the nature of Achille's character. Lacking both the word and the mechanism of metaphor, that person would be tongue-tied.

The metaphor mechanism, however, isn't restricted to cases like that of Achilles and the lion. Much of the world's wisdom is captured in proverbs--"The monkey doesn't see the hump on his own forehead" (Sesuto--see Cendrars, 1927), "A stitch in time saves nine," etc. In Kenneth Burke's formulation (1973, p. 296) proverbs "are strategies for dealing with situations ." The process by which the proverb is applied to the situation is a metaphorical one. The situation is the tenor, the proverb the vehicle, and the unstated ground is the abstraction, which is thus indicated in the process of applying it to the situation.

There are always going to be situations beyond the reach of a culture's lexicon. The use of metaphor is one way to advance into that territory. We are thus calling up the standard distinction between dead metaphor and robust metaphor. Dead metaphor is that which has been so used that its ground is now thoroughly familiar. The river's mouth, the road's shoulder, the foothills of a mountain range, these metaphors are so dead that we have to think twice simply to recognize the metaphor. Robust metaphor, however is the product of poets, of those writers who are called "real" writers by the critics, the ones who justify the special place in which poets and writers are put by cultures from ancient China to modern France. Robust metaphor is the only mechanism we have found for bringing new abstractions to consciousness.

2.3 Abstraction and Rationalization

Abstract concepts may also be exemplified by stories (Hays, 1973, 1976, 1981; Benzon, 1976, 1981; Phillips, 1978; White, 1975). These stories can be told but, of course, we have no direct access to the abstract concepts defined in these stories--no more so than we have direct access to the schemas which encode our sensorimotor apprehension of the world. Just as metaphor provides one way of manipulating abstractions, so rationalization provides another, more elaborate, way.

For example, consider "charity." One can't see, smell, taste, touch, or physically manipulate charity, thus charity must be abstract. We can, however, define charity as when "someone does something nice for someone else without thought of reward." This definition is the result of rationalizing the concept of charity. Metaphor can bring the abstraction "charity" to mind, but rationalization gives the mind something to work on. Any story which follows the general pattern of the definition is an instance of charity. In order to be charitable one need only identify a person in need, imagine a story in which one does something nice for that person, and then act to make the imagined story real.

Notice further that the definition of charity contains terms which themselves are abstract, such as "reward." There are many tangible things which may serve as rewards--an apple, a diamond pendant, a roller coaster ride--but there is no compact way to characterize this class of objects by its physical properties. Further, the class of things which serve as rewards will vary from individual to individual, and from time to time for a given individual. Some rewards, such as fame, are abstract. Similarly, "nice" and "thought" are abstract.

Finally, note the clause "without thought of reward." No exemplifying story is likely to contain anything directly corresponding to it in the way that every such story will contain a person who does something nice for someone else. These stories may well contain indications and statements of the thoughts of intentions of the charitable agent but "without thought of reward" is a statement of what those thoughts and intentions are not , not of what they are . This clause embodies a judgment passed on the whole class of exemplifying stories; it is a judgment about something which isn't in any of them, the lack of which suggests that the agent's motive must be charity. This clause thus embodies a sophisticated cognitive operation, an abstraction over the collection of exemplifying stories.

Rationalization guides the culture in the use of language and in characterizing abstract concepts. We call the definitions that it makes and uses "metalingual" definitions; one chunk of language, by rationalization, is used to define another, the abstract term. Through the mechanism of metalingual definition language allows us to pile abstraction upon abstraction. That is one capacity we need to build a culture.

2.4 Neural Implementation

Our last piece of conceptual apparatus is peculiar in that we won't actually make use of it. All of the cognitive processes we discuss must be embodied in the human brain. However, that brain is essentially the same for all people in all cultures. We therefore want to make some general remarks on how significantly different cognitive processes can be embodied in one and the same neural structure.

The key concept is that of mode, as formulated by Warren McCulloch (Kilmer et al., 1969). The basic idea is that different tasks--exploration, feeding, courting, fighting, etc.--make different demands on the brain. Processing in a given brain region may be very important for one task, of moderate value for another, and of little use in a third task. Thus for a given task some brain areas will be very active, others less so in varying degrees, and still other regions will be quiescent. Such a global pattern of activity is a behavioral mode (cf. Thatcher and John 1977. pp. 220-221, 305-307).

Recent observations have brought this concept vividly to light. Studies of cerebral blood flow (Lassen et al., 1978) and cerebral metabolism (Phelps and Mazziotta, 1985; see also Posner at al., 1988) show dramatically different patterns of cerebral activity for different kinds of task. Visual perception tasks show one pattern of activation while voluntary movement has another; speaking, reading silently, and reading aloud, all exhibit different patterns.

In our view the neural correlate of cognitive rank is to be found in these patterns of brain activation. The activities of reading and writing require patterns of brain activity which don't exist in illiterate peoples. These new patterns of brain activity support modes of analysis and synthesis not possible in other modes; hence concepts of a new kind become possible. Other cultural inventions--we are particularly concerned about algorithmic calculation and computer programming--have similar effects. The creation of a new brain thus does not require genetically driven changes in brain structure; it requires only culturally driven changes in cognitive technology, what Marshall McLuhan (1962) called "media". McLuhan discussed the way different media change the "sense ratios" of a culture, making it more or less auditory or visual or olfactory, etc. To us a culture with a high auditory sense ratio (for example) is one in which many persons spend much time in an auditory mode.

3 Cognitive Rank

The basic idea of cognitive rank was suggested by Walter Wiora's work (1965) on the history of music. Wiora argued that music history be divided into four ages. The first age was that of music in preliterate societies and the second age was that of the ancient high civilizations. The third age is that which Western music entered during and after the Renaissance. The fourth age began with this century. (For a similar four stage theory based on estimates of informatic capacity, see Robertson, 1990.)

This scheme is simple enough. What was so striking to us was that so many facets of culture and society could be organized into these same historical strata. It is a commonplace that all aspects of Western culture and society underwent a profound change during the Renaissance. The modern nation state was born (Gellner 1983), the scientific revolution happened (Butterfield, 1957; Cohen, 1960), art adopted new forms of realistic depiction (Gombrich, 1960), attitudes toward children underwent a major change (Aries, 1962), as did the nature of marriage and family (Stone, 1977), new forms of commerce were adopted (Braudel, 1981-1984), and so forth. If we look at the early part of our own century we see major changes in all realms of symbolic activity--mathematics, the sciences, the arts--while many of our social and political forms remain structured on older models.

The transition between preliterate and literate societies cannot easily be examined because we know preliterate societies only by the bones and artifacts they've left behind, and the historical record of the ancient high civilisations is not a very good one. Instead we have to infer the nature of these ancient cultures by reference to modern anthropological investigations of preliterate cultures (just as biologists must often make inferences about the anatomy, physiology, and behavior of extinct species by calling on knowledge of related extant species). When we make the relevant comparisons we see extensive differences in all spheres.

Social order in preliterate societies may involve nothing more than family relationships, or at most the society extends kinship by positing ancient common ancestors. With little or no apparatus of government, civil order is maintained by feud or fear of feud. In literate societies, social order is kept by etiquette, contract, and courts of equity, and civil order is maintained by police and courts of justice. In preliterate societies each community, of 5 to 500 members (and generally less than 200) is autonomous until (about 6000 years ago) chiefdoms appear in a few places, and groups of villages forced into submission (Carneiro, 1981). In literate societies villages grow into towns and cities, which organize the villages of their hinterlands into kingdoms. Preliterate societies depend on the skills of hunting and gathering, of slash-and-burn farming, pottery, and a few more crafts, which are sound and effective where they exist. In literate societies certain persons trained to think choose to think about farming and write manuals for the agrarian proprietor--and eventually manuals of other crafts appear. Finally, Lawrence Kohlberg (1981, pp. 128 ff., 233 ff.) has found evidence that people in preliterate societies have less sophisticated moral concepts than people in literate societies.

The appearance of writing was followed by the Mosaic law and the prophets of Israel, and by the Periclean Age in Athens. The architecture, democratic political system, and above all the philosophy--both natural and moral--of the Hebrews and Greeks was so different from all predecessors that we tend to think of our civilization as beginning with them. In fact, a period of cultural regression followed the fall of Rome and before the Renaissance could begin a "little renaissance" beginning about A.D. 1000 and reaching its peak with Aquinas in the 13th Century was necessary to raise Europe once more to a literate level. Our civilization combines elements of Greek, Roman, and Hebrew antiquity with Moslem, Indian, Chinese, and Germanic elements.

We are suggesting that these four ages, the systematic differences between cultures at these four levels of cultural evolution, are based on differences in cognitive mechanism. As cultures evolve they differentiate and become more complex and sophisticated, the more sophisticated cultures having cognitive mechanisms unavailable to the less sophisticated. Over the long term this process is discontinuous. That is, at some point in the evolution of a culture a new kind of thinking becomes available which permits a dramatic reworking of culture. This new kind of thinking is engendered by a new capacity for manipulation of abstractions.

These several kinds of thinking are cumulative; a simpler kind of thinking does not disappear from a culture upon the introduction of a more complex kind. A culture is assigned a rank according to the highest kind of thinking available to a substantial fraction of its population (cf. Kohlberg, 1981, p. 129). That a culture is said to be of Rank 3 thus doesn't imply that all adult members have a Rank 3 system of thought. It means only that an influential group, a managing elite if you will, operates with a Rank 3 cognitive system. The rest of the population will have Rank 1 and Rank 2 conceptual systems.

Each cognitive process is associated with a new conceptual mechanism, which makes the process possible, and a new conceptual medium which allows the mechanism and process to become routine in the culture. This is an important point. The general effectiveness of a culture is not determined by the achievements of a few of its most gifted members. What matters is what a significant, though perhaps small, portion of the population can achieve on a routine basis. The conceptual medium allows for the creation of an educational regime through which a significant portion of the population can learn effectively to exploit the cognitive process, can learn to learn in a new way. Our basic scheme is as follows:

In an earlier paper (Benzon & Hays, 1988) we argued that the human brain is organized into five layers of perceptual and cognitive processors. We called the top layer the gnomonic system and thought of it as organizing the interaction between the lower four layers (see also Hays, 1981). All abstractions form in the gnomonic system and the cognitive processes we are concerned about here are all regulated by this gnomonic system. Hence for the purposes of this paper it is convenient to collapse this system into a two-level structure, with the gnomonic layer on top in an abstraction system and the other four layers on the bottom, collectively, the concrete system.

With a Rank 2 structure, Aristotle was able to write his philosophy. He presented it as an analysis of nature but we take it to be a reconstruction of the prior cognitive structure. In the Renaissance, some thinkers developed cognitive structures of Rank 3. Exploitation of such structures produced all of science up through the late nineteenth century. Beginning, perhaps, with Darwin and going on to Freud, Einstein, and many others, a new kind of cognitive process appears. To account for it, we call on Rank 4 process. We understand earlier science to be a search for the one true theory of nature, whereas we understand the advanced part of contemporary science to be capable of considering a dozen alternative theories of nature before breakfast (with apologies to Lewis Carroll). The new thinker can think about what the old thinker thought with. And indeed we use that sentence to summarize the advance of each cognitive rank over its predecessor.

3.1 Rank 1: What's In a Name?

Rank 1 cultures run the gamut from simple roving bands of hunter-gatherers to quasi-states living in large permanent settlements. The peoples living in the Americas when the Europeans first arrived are all Rank 1 cultures, from roving bands which followed the buffalo in the central plains of North America to the Aztec, Maya, and Inca empires. The range of complexity across these societies is great, but all are within the compass of Rank 1 cognitive mechanism.

The emergence of language catalyzed the emergence of Rank 1 culture. What is critical about language is that it enabled people willfully to manipulate their mental processes, to gain control of consciousness (see Jerison, 1976, Vygotsky, 1962; Csikszentmihalyi, 1990, discusses the manipulation of consciousness in a way which is relevant here). With language people can call up thoughts of events long past and far away, thoughts of events which haven't happened yet, even stories about purely imaginary beings and deeds. Through language it is possible for an individual to summon a multiplicity of concepts and images to mind in a relatively short time and to bring the mind's synthesizing and analytic capacities to bear on this multiplicity. In ten minutes, a half-hour, an hour, more or less, a story can cover a much greater range of experience than would be accessible in the same period of time if one had to go there to see, hear, taste, and do it, whatever the "it" might be. In a short time one can tell a story which ranges across the entire geography inhabited by one's culture. But to travel that territory might take days, or weeks, or more.

Much of the abstract knowledge of Rank 1 cultures is carried in myth. Levi-Strauss (1969) has shown that, however bizarre the events of a myth may seem to us, myth is governed by a rigorous and relentless logic in which schemes of kinship, geography, the satisfaction of biological needs, and cosmography are subject to the same ordering principles. In his analysis of a Tsimshian myth about the culture hero Asdiwal, Levi-Strauss (1976) shows that the story is based on the need to resolve these underlying oppositions:

These four sets of concepts are in metaphorical correspondence. Low, earth, man, and endogamy are metaphorically equivalent through their opposition to high, heaven, woman, exogamy. Myth serves to bring these concepts into abstract relationship with one another thereby bringing order to the conceptual world. Such symbolic constructions constitute the norms by which individuals in Rank 1 cultures live.

A Rank 1 ontology consists simply of a listing of the types recognized by the culture, with some subcategorization. General categories such as "plant" and "animal" are very rare at this level and even categories such as "bird," "beast," and "fish," are not routinely used (Berlin et al., 1973). The commonest categories are at the level of "oak," "eagle," and "trout," with some subcategories, "white oak," "bald eagle," and "rainbow trout." Rank 1 peoples certainly have a practical knowledge of differences between plants and animals--they don't set snares for plants or expect animals to stay in the same place, but the conceptual basis of that practical knowledge is not made explicit in their systems of categories.

To illustrate this, let's consider an example recorded by the Russian psychologist A. R. Luria. In 1931-32 Luria made observations on the effect which literacy training had on the thought processes of Uzbekistani peasants. The following exchange took place with an illiterate thirty-eight year old adult (Luria, 1976, pp. 81-82).

What do a chicken and a dog have in common?

"They're not alike. A chicken has two legs, a dog has four. A chicken has wings but a dog doesn't. A dog has ears and a chicken's are small.

You've told me what is different about them. How are they alike?

"They're not alike at all."

Is there one word you could use for them both?

"No, of course not."

What word fits both a chicken and a dog?

"I don't know."

Would the word "animal" fit?

"Yes."

Immediately after this exchange the subject was asked about fish and crow. When the subject denied that they had anything in common he was asked whether one word could be used for both. He replied, "If you call them animals, that wouldn't be right. A fish isn't an animal and a crow isn't either. A crow can eat a fish but a fish can't eat a bird. A person can eat a fish but not a crow."

While the subject is acquainted with the word "animal," he doesn't thoroughly knows its meaning. It took a great deal of prompting for him to agree that chicken and dog were both animals and, having so agreed, he was unable to apply the term to fish and crow. The concept clearly was not one he routinely used. The subject's comments about the difference between chicken and dog suggests that he cannot form a generalization which covers both. That is, there is no easy way to eliminate extraneous detail from his concepts of dog and chicken so that the same conceptual core remains in each case. The similarity between wings and forelimbs is not at all compelling to this peasant, nor would it be to any but a biologist or those whose view of the world has been informed by the biologist's thought.

Notice further that in justifying his account of fish and crow the subject talked about the roles which "fish," "crow," and "person" can take with the verb "eat." This is the sort of consideration which generates ontological categories, but this subject clearly couldn't get to a meta level from which he could explicitly grasp this categorization.

The Rank 1 thinker can form metaphors and proverbs, and thus get abstract ideas into his or her higher level of cognition. These abstractions influence perception and behavior, giving an organization to sense-data that no animal can have. But the Rank 1 thinker lacks a structure that permits comparative judgment between alternative abstractions, or any other intellective assessment. (See Le Pan's, 1989, critique of African religion for illustrations.)

3.2 Rank 2: The Letter of the Law

Writing is critical not only because it allows the stable representation of thoughts but also because it forces thinking about thought. In contemplating the written word, scribes and sages begin systematically to think about words, the sequencing of words, and, inevitably, about the mental processes behind words, about thought. They were able to do this because they could see a text as a whole, they could get outside the flow of discourse as no illiterate speaker or hearer could do. They had to do this in order to create and refine the conventions of written discourse. For writing imposes much stricter requirements of completeness and grammaticality to make up for the lack of paralinguistic and contextual clues available in face-to-face conversation, not to mention the ability to question the speaker about anything one doesn't understand. The writer is inherently more self-conscious than the speaker and such self-consciousness is likely to engender thought about writing, language, and ultimately, about thought itself.

While the capacity to formulate discourse about language is intrinsic to language--Roman Jakobson (1961) called this the metalingual function--the use of writing facilitates such metalinguality. With writing the text becomes a sense-data entity of a much more palpable sort than the airy nothings which carry speech from tongue to ear. Metalinguality thus can first be about the text, and then by abstraction about the language and the thought that the text represents. The problem of creating and using a writing system thus moved metalinguality to the point of engendering a new process of thought, rationalization, through the form of metalingual definition.

Recall the analysis of charity. "Charity" could be given meaning by recounting various instances of charity, just as "apple" can be given meaning by pointing to various instances. The rationalization, "when someone does something nice for someone else without thought of a reward," expresses an abstraction over the collection of exemplifying instances. The use of such rationalizations is what we are asserting is new to Rank 2 culture. People in Rank 1 cultures have abstractions, but have only proverbs and myths for expressing them. The mechanism of metalingual definition allows a Rank 2 culture explicitly to convey its abstract knowlege through rationalizations. With conveyance comes the possiblity of intellectual analysis, of philosophy, and of mathematics as well.

With the emergence of philosophy comes the explicit construction of ontology. In his Categories Aristotle states that all assertions are about substance, quantity, quality, relation, time, position, action, or passivity. These categories constitute an ontology, an assertion of the main aspects of reality. In his treatise On the Soul Aristotle analyzes the soul as a substance and argues that it has a hierarchy of functions: the nutritive, the perceptive, the locomotive, and intellective. Living things can be arranged in a hierarchy according to how many of these faculties each possesses, thus giving us an account of the varying capacities of plants, animals, and humans.

The Rank 2 abstractive system thus has two mechanisms available to it: metaphor and metalingual definition. Metaphor, the basic mechanism for managing abstraction, rides at the "top" of the system, as it did at Rank 1, and brings new abstractions into awareness. The metalingual mechanism rationalizes the ontology implicit in the relationships between objects and events in the concrete domain. This yields the categorical thinking of Rank 2 philosophical thought. The world is distributed into a set of categories--which, in the West, became the Great Chain of Being--and explanations are formulated in terms of these categories. Each thing acts in accordance with the capabilities of its category. The problem is to determine its category and assimilate its actions to the capacities inherent in its category.

Consider how much of philosophy is devoted to the definition of terms, to rationalizing their meaning. To be sure, the discussion quickly moves beyond the scope that dictionaries allot to definitions, so far beyond that scope that we generally do not think of philosophical thought as elaborating definitions. But that is what is happening. Plato writes his Republic to define "justice" and, in the process, gives us a doctrine of the state, the just state. The rationalizations of abstract terms are not generally the simple sorts of statements we used in discussing "charity." These rationalizations are pages and pages and books and libraries of thought. Rank 2 thought consists in the explicit and extensive elaboration of metalingual definitions. In the process of defining one term many other terms must be used, and the meanings of those terms must be elaborated as well, and so forth. This whole process is guided, intuitively, by metaphor (cf. Pepper 1942, White 1973, Lakoff and Johnson, 1980). The Great Chain of Being is, itself, one such master metaphor.

The development of writing took several thousand years, from the first mnemonic marks to the fixing of an alphabet with distinct marks for all the consonants and vowels of Greek. The philosophies of the Hebrews (with a consonantal alphabet) and the Greeks (with a full alphabet) followed in less than a thousand years after their alphabets.

Alphabetic writing is critical, in part, because it is the only form of writing which can be readily routinized. As Erik Havelock has argued (1982) a logographic or ideographic system such as the Chinese used requires the memorization of thousands of visual symbols. Syllabaries, while having a much more limited number of symbols, often produced ambiguous texts requiring skilled experts to determine just what words were intended.

3.3 Rank 3: Subject and Object

The role which speech plays in Rank 1 thought, and writing plays in Rank 2 thought, is taken by calculation in Rank 3 thought (see Havelock, 1982, pp. 341 ff.). Writing appears in Rank 1 cultures and proves to be a medium for Rank 2 thinking. Calculation in a strict sense appears in Rank 2 and proves to be a medium for Rank 3 thinking. Rank 2 thinkers developed a perspicuous notation and algorithms. It remained for Rank 3 thinkers to exploit calculational algorithms effectively. An algorithm is a procedure for computation which is explicit in the sense that all of its steps are specified and effective in the sense that the procedure will produce the correct answer. The procedures of arithmetic calculation which we teach in elementary school are algorithms.

These procedures are so familiar to us, and so obviously elementary, that we forget that their creation was a major cultural achievement--attempting long division in Roman numerals, however, should remind us of just how very difficult computation can be without a good system of notation. Nor did the ancients have explicit rules of procedure. Marrou, in describing education in the Hellenistic period, writes (1956, p. 158):

Strange though it may seem at first, it is nevertheless quite clear that addition, subtraction, multiplication and division ... were, in antiquity, far beyond the horizon of any primary school. The widespread use of calculating-tables and counting-machines shows that not many people could add up--and this goes on being true to a much later date, even in educated circles.

The mathematical texts are simply concrete examples of different problems worked out in full. They illustrate to the reader how to do sums of various kinds. But by themselves such series of examples could hardly suffice to enlighten a novice as to new methods nor impart to him fresh knowledge. They must have been intended as supplements to oral instruction.

3.4 Rank 4: Modern, Post-Modern, and All that Jazz

Two decades ago counter-cultural savants in the West proclaimed the coming of an Aquarian Age. A decade ago different savants eagerly anticipated the emergence of an Information Age (Toffler, 1973). Is this just millenial fever triggered by anticipation of the turn of the next millenium--as years ago our European ancestors anticipated the second coming of Christ with approach of the year 1000--or is something really happening?

There is no definitive way of refuting the view that nothing is happening. But, those who maintain it must at least admit that, in the past, there have been major changes in culture and so such changes are at least possible. At the very least, those who argue that we are undergoing some form of radical change are not arguing for something we know has never happened.

We are of the view that, once again, culture is evolving toward a new rank and that most of the intellectual and artistic displacement we are seeing reflects these growing pains. This new cognitive rank, Rank 4, has its origins in two developments in late nineteenth century thought: the creation of formal systems of logic and metamathematics and the emergence of non-mechanistic science.

The metamathematical work of the late nineteenth and early twentieth centuries was intended to provide rigorous deductive foundations for all branches of mathematics and to unify these branches into one coherent system. The work in logic was concerned, on the one hand, with developments in mathematics, but was also concerned about formulating general mechanisms for the formulation of any type of proposition so that the propositions of science could be formulated in a giant deductive system from which truths could be cranked out on a routine basis. Thus the development of the propositional calculus and set theory, whatever their importance for metamathematics, also looked toward the world and got some thinkers into the habit of translating propositions from ordinary language into logical forms. Thus logic became a set of formal tools in which other conceptual systems--mathematical and scientific--could be represented. As the physical sciences moved more abstractly into the physical world, logic moved more abstractly into the informatic world.

The new scientific style was forced further and further from a mechanistic universe by hard facts of the most intransigent and nonmechanistic sort. Thermodynamics provides the prototype, with biology (evolution) right behind (Prigogine & Stengers, 1984). Perhaps the most deeply unnerving case, however, is that of quantum mechanics. That light behaved in some experiments like waves and in other experiments like particles was uncomfortable. To explain what they could see, physicists had to imagine a quantum world that they could not see: In principle and not in mere practice, the quantum world is not observable (Penrose, 1989). Yet it provides a framework for mathematical derivations that explain, if that is the right word, the observations that can be made in our world. The fact of the matter is, Rank 4 science is as different from Rank 3 science as Rank 3 science is from Rank 2 natural philosophy. Sophisticated logic and mathematics become ever more necessary to thought. To admit the forces and the intangible particles without logic and mathematics to regulate explanation would be to readmit magic and superstition.

Time itself comes under new scrutiny. Not until the turn of the 20th century was motion was studied in enough detail so as to provide descriptions of complex irregular movement. Motion pictures, photographs showing the paths taken by hands performing a task, time-motion studies in factories, and paintings of a single subject at several stages of an action all turned up more or less together. This conceptual foregrounding of temporality, when combined with metamathematical and logical reasoning, led to the development of the abstract theory of computing between the world wars.

The central work is Turing's explication of the algorithm. Rank 3 had concocted and used algorithms, but Turing explained what an algorithm was. In order to formulate the algorithm Turing had to think explicitly about the control of events in time. He described his machines as performing an action at a certain time; then another action at the next moment; and so on for as many consecutive moments and actions as necessary. His universal machine, a purely abstract construction, was an algorithm for the execution of any algorithm whatsoever. With it, he showed that no interesting formal system can be complete in the sense of furnishing a proof for every true statement. (Others reached the same conclusion by other methods, and he missed being first by a brief interval.)

It took the work of von Neumann, and others, however, to embody Turing's account of the algorithm in a physical device: the computer (Bernstein, 1964, pp. 60 ff.). Of particular importance is von Neumann's use of the conditional jump as a control structure. In a computer all of the data and instructions are kept in a long list. The computer operates on this list by taking an item, operating on it, and then moving to the next item, operating on it, and so forth, item after item. For purely mathematical purposes such an arrangement is fine--it is how Turing thought of his abstract machines. For practical computing, however, this is not a good arrangement. For practical purposes you want to be able to access any piece of data, and any instruction, whenever you need it, regardless of whether it is next in line. There is no practical way of arranging things so that the next item needed is always the next one in line; you need to jump around. Von Neumann's conditional jump allows this. The basic idea is that one applies a test to the current state of the computation. Where the computation moves next depends on the outcome of the test. In principle the computation could move to any instruction and any piece of data regardless of where that item is; it need not be the next one in the list. Much of the art of programming consists in the ordering and manipulation of such control structures. From a purely mathematical point of view they are unnecessary. But without them there would be no practical programs, no art of programming. For this reason we think of control structure, starting with von Neumann's conditional jump, as the Rank 4 conceptual mechanism, corresponding to the Rank 3 algorithm and the Rank 2 metalingual definition.

Yet the thinker of Rank 4 is not a Turing machine but a user of both objective models and algorithmic theories of them. The way out of the impasse found by Goedel, Turing, and others is to remember always that it is possible to construct a new Turing machine to prove what the former could not prove. The loss of sense data as a faithful representation of reality on acceptance of quantum mechanics and the loss of total deducibility in metamathematics led to a crisis in Rank 3 thought. But Rank 4 can accept the situation and exploit it to obtain both the transistors and the programs of digital computers--pragmatic justification for Rank 4's extraordinarily sophisticated conceptual processes. Its models of nonobservable phenomena are useful.

Freud, also, was driven to imagine a world that cannot be observed. From the unconscious come dreams, tics, and all the psychopathology of everyday life; but the unconscious mind cannot actually be made conscious by any method. The mathematical derivations of the physicists make possible precise tests, and the results are too good to admit of argument. The hermeneutic derivations of the psychoanalyst are not algorithmic, and make possible only rather vague tests. Freudian theory remains a topic of much contention.

And now we join the long parade, with a component of cognitive structure that, we say, lies in principle beyond direct observation, the gnomonic component that forms abstractions. It may have correlates that can be observed with electronic instruments, but they are not the thing itself. For the present our theory cannot escape contention, but we are convinced, first, that our apparatus is as simple as any that can account for thought, and second, that the possibility of close experimental verification waits only on development of the theory and of psychological methodology.

3.5 Cognitive Rank and Cultural Diversity

The vast majority of the world's cultures have Rank 1 cognitive systems. All of the cultures indigenous to North and South America, sub-Saharan Africa, Australia, Polynesia and Micronesia, Siberia, and the hill country of Southeast Asia are Rank 1 cultures. Rank 1 cultures are very diverse because the evolutionary criterion--effectiveness--is relatively lax in the demands it makes on culture patterns. The range of things a people must do to ensure their physical survival is relatively narrow in comparison to the mind's need and capacity to bring conceptual order to the world. The collective cultural inventiveness of Rank 1 cultures far exceeds the requirements of physical survival. The number of distinct and independent Rank 1 cultures thus numbers in the thousands.

There have been only 3 or 4 independently arising Rank 2 cultures, the rest all arising through diffusion or conquest. Around the eastern Mediterranean, from Egypt and Mesopotamia up through the Hebrews and Greeks to Rome: that is one. Harappan civilization in the Indus basin: that is two, but with possible linkage to the eastern Mediterranean. China: that is three, and the possibility of remote linkage exists there as well. Islam is perhaps regressive, and altogether derivative. Muscovite and Byzantine are derivative as well. By our standards the American cultures (Aztec, Mayan, Inca) did not quite reach rank 2.

The range of diversity among Rank 2 cultures is high, but perhaps not so high as that among Rank 1 cultures. Rank 2 polities, however, are more internally diverse than Rank 1 polities, in part because many Rank 2 polities have incorporated a variety of Rank 1 polities through conquest.

Rank 3 cognition happened, independently, only once, in Western Europe during the period called the Renaissance. It remains to be seen whether or not, or in what sense, Rank 3 culture admits of variety. Certainly in one respect, science and technology, there is only one Rank 3 culture. But, is democracy of one sort or another the only viable rank 3 political system? Possibly the Rank 3 contents of culture are so constrained under the force of three mechanisms of abstraction that there is no variety possible--or perhaps possible, but not obvious.

Edwin O. Reischauer's (1977, p. 288) remark that we should think of Japan as becoming modernized, as opposed to becoming Westernized, suggests that Japan is evolving a Rank 3 culture which is, in many respects, different from the Rank 3 culture of the West. Perhaps, for example, we should think of the much-studied Japanese corporation (Ouchi, 1981; or Halberstam, 1986) as a non-Western way of organizing Rank 3 business and manufacturing tasks. Within the United States, the emergence of jazz provides a different example. Jazz is quite different from classical, or European (and European-derived) music, and some authorities maintain it has a sophistication and expressive power which is equivalent to classical music (Collier, 1978; Williams, 1983). If so, then jazz is a Rank 3 music which is quite different from the Rank 3 music of Europe, suggesting that Rank 3 culture isn't restricted to the lines of development implicit in Western Europe of the Renaissance.

As far as we can tell Rank 4 culture exists only in various intellectual and artistic realms. The earliest teaching of rank 4 was presumably in a few graduate departments of physics and mathematics. After World War II, a few other disciplines reached that level, but only in certain universities. In the proliferation of new "interdisciplinary" ventures one can readily distinguish between those that combine elements of two recognized disciplines without generating much that is new, and those that lay out two disciplines side by side only to transcend them; the latter are likely to be teaching Rank 4 ideas. In the 1980s, several graduate schools of management adopted curricula incorporating such things as operations research, decision theory, and game theory in a way that rises above Rank 3. The remarkable suggestion, made by the government of the USSR in 1989, that a world-wide referendum determine whether East and West Germany could reunite, gives some hint of what might be expected in a Rank 4 political system. Twenty years ago, most countries claimed that human rights were a matter of national sovereignty; more recently, many countries have agreed that suppression of rights anywhere is a matter of universal concern.

4 In Medias Res: The Current Flux

As we have noted, it is our view that Rank 4 cognition exists in only a few arenas. Beyond that, we think that computing has not advanced to the point where it can serve as a means of routinizing Rank 4 thought in the way that algorithmic calculation served Rank 3, writing served Rank 2, and speech, Rank 1. The evolution of Rank 4 cognition is thus very much in progress. The purpose of this section is to examine this situation and outline the cultural process which will resolve the situation.

4.1 The Thinking Machine or Electronic Brain

One of the problems we have with the computer is deciding what kind of thing it is, and therefore what sorts of tasks are suitable to it. The computer is ontologically ambiguous. Can it think, or only calculate? Is it a brain or only a machine?

The steam locomotive, the so-called iron horse, posed a similar problem for people at Rank 3. It is obviously a mechanism and it is inherently inanimate. Yet it is capable of autonomous motion, something heretofore only within the capacity of animals and humans. So, is it animate or not? Perhaps the key to acceptance of the iron horse was the adoption of a system of thought that permits separation of autonomous motion from autonomous decision. The iron horse is fearsome only if it may, at any time, choose to leave the tracks and come after you like a charging rhinoceros. Once the system of thought had shaken down in such a way that autonomous motion did not imply the capacity for decision, people made peace with the locomotive.

The computer is similarly ambiguous. It is clearly an inanimate machine. Yet we interact with it through language; a medium heretofore restricted to communication with other people. To be sure, computer languages are very restricted, but they are languages. They have words, punctuation marks, and syntactic rules. To learn to program computers we must extend our mechanisms for natural language.

As a consequence it is easy for many people to think of computers as people. Thus Joseph Weizenbaum (1976), with considerable dis-ease and guilt, tells of discovering that his secretary "consults" Eliza--a simple program which mimics the responses of a psychotherapist--as though she were interacting with a real person. Beyond this, there are researchers who think it inevitable that computers will surpass human intelligence and some who think that, at some time, it will be possible for people to achieve a peculiar kind of immortality by "downloading" their minds to a computer. As far as we can tell such speculation has no ground in either current practice or theory. It is projective fantasy, projection made easy, perhaps inevitable, by the ontological ambiguity of the computer. We still do, and forever will, put souls into things we cannot understand, and project onto them our own hostility and sexuality, and so forth.

A game of chess between a computer program and a human master is just as profoundly silly as a race between a horse-drawn stagecoach and a train. But the silliness is hard to see at the time. At the time it seems necessary to establish a purpose for humankind by asserting that we have capacities that it does not. To give up the notion that one has to add "because . . . " to the assertion "I'm important" is truly difficult. But the evolution of technology will eventually invalidate any claim that follows "because." Sooner or later we will create a technology capable of doing what, heretofore, only we could.

Perhaps adults who, as children, grow up with computers might not find these issues so troublesome. Sherry Turkle (1984) reports conversations of young children who routinely play with toys which "speak" to them--toys which teach spelling, dolls with a repertoire of phrases. The speaking is implemented by relatively simple computers. For these children the question about the difference between living things and inanimate things--the first ontological distinction which children learn (Keil 1979)--includes whether or not they can "talk," or "think," which these computer toys can do.

These criteria do not show up in earlier studies of children's thought (cf. Piaget, 1929). For children who have not had exposure to such toys it is perfectly sensible to make the capacity for autonomous motion the crucial criterion. To be sure it will lead to a misjudgment about cars and trains, but the point is that there is no reason for the child to make thinking and talking a criterion. The only creatures who think and talk are people, and people also move. Thus thinking and talking will not add anything to the basic distinction between living and inanimate things which isn't covered by movement.

The child who must deal with computer toys has a different environment to deal with. However, if the child gets used to machines which "think" and "talk," that child might develop an ontology in which the "thinking machine" had a natural and comfortable place--as we have developed an ontology which is comfortable with locomotives, cars, and airplanes. By what criterion will that child differentiate computers and people? We can't answer this question yet, but we trust that the child will find a way. When that child has become an adult, that adult may not be so puzzled about the essential nature and potential of computers.

4.2 The Software Problem

Ontological ambiguity aside, computing technology clearly is still in its early stages. Most of the advance in computing has been in hardware, not software. We know how to build the machines; but we do not know how to program them. For the last four decades hardware has gotten faster, more compact, more reliable, and cheaper. That is what has allowed computers to spread thoughout the industrialized world.

The development of software has not fared so well (Bacon, 1982; Branscomb, 1982; Hays, 1982). The very best programmers are very good indeed. But their expertise does not travel well to the legions of programmers who write most of the world's software. The concepts involved in routinizing a high standard of software seem much more arcane than those involved in hardware. Part of the problem is certainly that software structures are much more complex than hardware. Computers contain millions and millions of circuits, but they are all of a relatively few types, assembled in a relatively few ways. The control structure of an intricate program is an intricate structure, and each of its elements is inherently sophisticated. We have no deep and powerful insights into how to manage control structures.

We note, however, that the vast majority of those with a deep intellectual involvement in software, whether as programmers or as theoreticians, arrive at this interest from a childhood and early adolescence which has been free of any significant contact with computing. We suggest that, when computing moves so deep into the culture that significant numbers of children are programming, then the culture will be able to produce a cohort of adults capable of solving the software problem, and many others as well. Then Rank 4 thought will have arrived.

4.3 The child is father to the man

In general we assume that the growth of thinking and knowledge in individuals is epigenetic, that later mechanisms of thought are constructed from the materials made available by earlier mechanisms (Benzon & Hays 1988, pp. 314-319). In particular we remain partial to the concept of stages pioneered by Jean Piaget (Piaget & Inhelder, p. 1969). The abstractions one begins learning in adolescence are based on the more concrete structures of thought acquired earlier. The preadolescent matrix will vary from rank to rank.

Let us begin by considering Rank 2. The child of Rank 2 parents will be exposed to the Rank 2 cultural environment of those parents. Much of that world will be as mysterious to the Rank 2 child as the world of Rank 1 adults is mysterious to the Rank 1 child. But, for example, the Rank 2 child can see his parents read and write, while the Rank 1 child cannot, for that is a mystery which doesn't exist in the Rank 1 world.

Consider, specifically, the language to which a Rank 2 child is exposed. It will have a more deeply developed system of superordinate and subordinate categories, including categories such as plant and animal. The child may not have an immediate grasp of the conceptual structure underlying these categories--such ontological knowledge develops gradually (Keil, 1979)--but he or she can learn to use those words correctly in many contexts and that knowledge will be a good foundation on which to construct the appropriate abstract justification for the categories. The conceptual environment of the Rank 2 child is thus significantly different from that of the Rank 1 child, and the difference is of the sort which will make it easier for the Rank 2 child to acquire the abstractions necessary for a Rank 2 adult.

We expect childhood exposure to computing to have a similar effect. But we do not as yet see anything significant happening on a large scale. Computers may be in every primary school in the nation, and in a small percentage of homes, but children do not spend much time on these computers. And most, if not all, of the time they do spend is devoted to using the computer in the most superficial way, not in learning to program it. And that, programming, is where the major benefit lies. It is in programming that the child has to deal with control structure, the element which is new to Rank 4 thought.

We know that children can learn to program, that they enjoy doing so, and that a suitable programming environment helps them to learn (Kay, 1977; Pappert, 1980). Seymour Pappert argues that programming allows children to master abstract concepts at an earlier age. In general it seems obvious to us that a generation of 20-year-olds who have been programming computers since they were 4 or 5 years old are going to think differently than we do. Most of what they have learned they will have learned from us. But they will have learned it in a different way. Their ontology will be different from ours. Concepts which tax our abilities may be routine for them, just as the calculus, which taxed the abilities of Leibniz and Newton, is routine for us. These children will have learned to learn Rank 4 concepts.

References