Origins of the Modern Mind

Three Stages in the Evolution of Culture and Cognition

Merlin Donald

• Publisher: Harvard University Press
• Year: 1993
• ISBN:
  9780674253704
• Topic: Biological Sciences
• Subtopic: Science General
• Index: Biological Sciences

This bold and brilliant book asks the ultimate question of the life sciences: How did the human mind acquire its incomparable power? In seeking the answer, Merlin Donald traces the evolution of human culture and cognition from primitive apes to artificial intelligence, presenting an enterprising and original theory of how the human mind evolved from its presymbolic form.

ONE

The Need for a Theory of Cognitive Evolution

Mental Architecture as an Emergent Phenomenon

Neuropsychology and cognitive science are concerned largely with the fundamental structure of the modern human mind. Although some attention has been paid to the phylogenetic succession of changes that must have led to the modern mind (see, for instance, Anderson, 1983), emphasis has been placed mostly upon the modern structure of human mental capacities, without taking their evolution into account. It is not an exaggeration to say that theories of cognitive structure are built mostly upon studies of the human mind as manifest in literate, postindustrial society and upon studies of the capabilities of computers. The extraordinary range of theory that has resulted was constructed for the most part without the constraints that must be applied to evolutionary hypotheses: continuity with previous forms, consistency with selection pressures, parsimony with regard to the number and complexity of successive adaptations, and so on. The result is that structural, that is, modular, models of mind proliferate without regard to their biological feasibility, even within neuropsychology.

Within the traditions of biopsychology and comparative biology, on the other hand, humans have been viewed “from below,” that is, in light of their mammalian predecessors. Phylogenesis has been a major concern to many thinkers in these fields, and efforts have been made to build the crucial conceptual bridge between the mental structures of apes and humans. A recent example is Lieberman’s (1984) attempt to specify the adaptations leading to human language. Since apes, and particularly chimpanzees, are genetically so close to humans, the key to the unique power of the human intellect is sought in its contrast with that of apes. However, theories emanating from these fields have not been rich in detail about mental structure, particularly about higher cognitive functions. This is partly because there are few equivalences in the terms used to describe the cognitive capacities of animals and humans; and those that are used often fail to do justice to the complexities of human cognition.

But whether a given theory of cognitive evolution starts “from below” or “from above,” it will wind up taking one of two broadly different approaches to higher function, which may be labeled “modular” and “unitary.” Modular theories, sometimes called faculty theories, propose a number of quasi-independent cognitive “modules” that are responsible for each dissociable or isolable aspect of higher function. The specific arrangement of modules in each theory usually varies, but generally each mental module encapsulates a definable higher mental function, or a stage of such a function. There may be separate structures (and, by implication, separate evolutionary adaptations) for spatial reasoning, mathematical ability, musical talent, phonological skill, the oral lexicon, the written lexicon, visual imagery, nonverbal thought, and verbal thought, to name a few.

Modular theories of language evolution find their biological roots in the writings of Darwin (1871) and Wernicke (1874), who both supported the idea of a series of special human adaptations leading to speech. These notions have found a modern expression in the biological theories of language proposed by Lenneberg (1967) and Lieberman (1975, 1984). The most recent influential example of a modular approach is found in the neurolinguistic theorizing of Shallice (1988).

Unitary theories, by contrast, generally hold that higher function is achieved by a single cognitive structure, that is, a single adaptation, with an exception sometimes being made for the peripheral, or sensorimotor, mechanisms of language. Anderson (1983) made a persuasive case for this approach in his book The Architecture of Cognition. His defense of a unitary approach rests on three arguments: (1) human higher function has a very short evolutionary history, and there would not have been time to evolve special faculties, for instance for mathematics; (2) human intelligence is highly plastic, or flexible, and displays a variety of special skills that could not have been anticipated in evolution; (3) the various higher cognitive functions of humans have many features in common.

Anderson has argued that a single general framework, or functional architecture, is sufficient to support all forms of higher function, even language. His approach echoes the faith of an earlier generation in a general-purpose “learning” capacity as the crucial adaptation underlying higher function. In fact, it is no coincidence that his ACT* framework is basically a diagram of memory structure. * What Anderson is saying is that the main “hardware” requirements of a computational model of human cognition (or of any complex mind) are general features of the system of storage and retrieval rather than separate specialized systems for various skills. It follows that the crucial human adaptation was some refinement of the memory system, especially of declarative and working memory, that enabled symbolic representation and language.

The unitary approach receives some support from primatology and comparative anatomy. A variety of studies over the past two decades have supported the encephalization hypothesis. Encephalization refers to the progressive increase in the relative size of the brain, particularly the cerebral cortex, during mammalian, and especially primate, evolution. Jerison (1973) and Passingham (1982), among others, have argued that the progressive growth in brain size characteristic of the great apes and hominids has followed a systematic trend, and that modern humans are simply an extension of that trend. Some particularly interesting data from Passingham show that the human brain has the proportions one would predict by extrapolating the trend of earlier primate brain expansions. Put simply, humans possess exactly the brains to be expected of large-brained primates. This proportionality extends even to the relative size of a variety of specific nuclei, ganglia, and substructures within the brain. In other words, there is no support in gross neuroanatomy for a completely novel language subsystem in the human brain, or for any other type of gross anatomical reorganization. This does not rule out some still-undetected molecular innovation related to higher function. But until such an innovation is uncovered, it appears that the most distinct property of the human brain is simply its extraordinary increase in relative size, which translates into increased numbers, and complexity, of memories, according to Jerison and Passingham.

The encephalization hypothesis in its strongest form holds that increased cognitive power is a direct result of increased encephalization. In principle, it argues for the kind of unitary structure of mind offered by Anderson. It implies that human cognitive power, including the higher aspects of language, is rooted in the superior computational capacity of the whole brain, and especially the cerebral cortex, rather than in any specific new structure. Lieberman (1984) walked the fine line between the unitary and modular approaches by assigning specifically linguistic adaptations, like control of the vocal tract and phonological processing, to specifically human neural speech processors, while leaving most higher functions to what he called a “central distributed neural computer.” In other words, the front-end aspects of speech may depend on special neural processors controlling phonation, but verbal thought and higher cognition in general are made possible by a single biological adaptation: the expansion of the cerebral cortex. This aspect of Lieberman’s theory is quite similar to an earlier proposal by Fodor (1983) (see chapter 3), which also placed the higher aspects of linguistic function in a general-purpose processor and the lower aspects in a self-contained language “module.”

The form of any evolutionary proposal regarding language and thought will be greatly affected by the number, and types, of cognitive modules associated with human higher function. The simplest evolutionary proposal would be a unitary system supported by the growth of the entire cortex, such as the one proposed by Jerison (1973). In his theory, the history of language and thought becomes the history of cortical expansion. As the encephalization quotient goes, so goes cognitive architecture and intelligence.

A more complex evolutionary scenario might involve many individual modules, each serving a different cognitive purpose. For instance, there might be many specialized language and thought modules that possess some degree of anatomical localization. The evolution of such a complex mosaic structure, unique to humans, would imply that a series of separate adaptations led us from the cognitive architecture of apes to our present structure. A selection advantage driving the acquisition of each aspect of the mosaic would have to be specified. Moreover, the number of possible sequences of change would be greatly increased, and the empirical basis for making a choice between them would be very limited. Such a scenario would be difficult to construct, particularly since, on present evidence, the evolution of language occurred so quickly. By way of comparison, the neural mechanisms of mammalian vision evolved over millions of years, through a large number of species. For a neural structure of at least equal complexity to evolve and eventually (surely not in its initial manifestation) support language, we should presumably expect an equally long and arduous process of evolution.

Thus Anderson’s principal objection to a modular theory stands: a complex modular solution to the problem of human higher function would not lend itself easily to an evolutionary model. Since we are not in a position to abandon evolution as an explanatory principle, we should be wary from the start of modular neuropsychological models which specify an architecture that depends on many built-in, specialized neural components. In any case, it is clear that any evolutionary hypothesis will be greatly affected by its creator’s initial stance on cognitive structure. The reverse is also true. The credibility of any structural hypothesis will be affected by the evolutionary scenario it implies. Phylogenesis and structure are two aspects of the same thing—the one a motion picture, the other a single frame, both trying to represent the same reality. There are times when the single frame is clearer, and there are other times when a blur in a single frame is resolved by motion. Since our view of higher function is not very good to start with, we cannot afford not to try to use both approaches.

Culture as Evidence for Cognitive Structure

One element frequently left out of cognitive modeling is the element of culture, that is, shared patterns of acquired behavior characteristic of a species. But the cognitive capacities of animals directly affect the kinds of culture they produce, and in the case of humans, the opposite is also true: specific types of human culture have direct effects upon individual cognition. In fact, the uniqueness of humanity could be said to rest not so much in language as in our capacity for rapid cultural change.

If we wished to put the proposition even more strongly, we might assert that what humans evolved was primarily a generalized capacity for cultural innovation. Part of that capacity was linguistic communication; part of it was the ability to think and represent the environment. This is a very easily defended proposition from the viewpoint of evolutionary theory; the selection advantages accruing to a species capable of cultural invention would be tremendous. Archaeological and anthropological evidence strongly supports the idea that the rate of cultural change increased during hominid evolution, at first slowly (Homo erectus changed only slightly over a period of a million years), then more rapidly (early Homo sapiens achieved several major innovations over a period of 200,000 years), and finally at the continuously accelerating pace of our own species, Homo sapiens sapiens.

Dunbar (1990) has recently proposed that encephalization was driven not by the cognitive demands of toolmaking or spatial mapping of the environment but by growth in the size of social groups. In other words, it was not instrumental intelligence that drove brain expansion but rather social intelligence. Complex societies make great demands on memory: large numbers of relationships have to be analyzed, understood, stored, and serviced regularly in order to sustain a large group organization. With certain exceptions, the more advanced primates cluster together into larger and larger social groups, culminating in the human capacity for organizing and sustaining very large groups. It could well be the case that the intellectual abilities needed to sustain large groups are identical to those that enable cultural invention. The first adaptations in the hominid line might have been driven by the demands of social grouping, and cultural invention might have been its by-product.

The main difficulty with such a proposal is its vagueness about mechanism. What biological mechanism would enable continuous cultural invention, or maintain larger social groupings? Inevitably such proposals lead one to speculate about cognitive mechanisms operating on the level of the individual brain. Thus, the bridge from cultural to biological realms is necessarily cognitive, and a complete evolutionary proposal should address the cognitive level, even if its ultimate objective is to explain our capacity for cultural change.

In any evolutionary theory, cultural evidence must play an important role, since it is not reasonable to expect that every “module” of higher function is functionally present in every neurologically normal human mind. To provide an example, if Morton’s (1980, 1981) well-known model of spelling* is taken at face value, the cognitive modules that constitute the heart of the visual path in the model—the visual input lexicon, and the graphemic output lexicon—are obviously absent in illiterate humans. Morton would not propose that illiterate humans are therefore missing certain essential biological modules. He has acknowledged that the modules do not in fact represent anatomical entities but only functional ones; and it is evident that the physical underpinnings of the visual path are normally present in the brain, since illiterate humans can usually be taught to read.

This, of course, would beg the question. The point is, some of his modules support species-universal traits that are not culturally bound (for instance, all the modules supporting speech); these probably represent distinctive biological adaptations. Other modules, like those supporting reading and writing, support behaviors that are obviously not species-universal and are largely or entirely bound by culture. The evolutionary basis for these, and thus the expectation of a distinct physical basis for them, may be very different. In other words, all modules, however closely reasoned on the cognitive or neuropsychological evidence, were not necessarily created equal, even though the neuropsychological criteria for their creation were similar.

If we accept the validity of the sorts of modules Morton and others routinely construct from neuropsychological evidence, it follows that the actual cognitive structure of an individual mind is heavily influenced by culture. Styles of reading and writing are culture-bound. Thus, depending upon the specific skills demanded by literate culture, prospective model-builders have to introduce modules for different functions. For example, in the brain of a reader of modern English, as Morton and many others would agree, there must be a module that performs direct grapheme-to-phoneme translation. But in a largely ideographic writing system, such as ancient Chinese, there is no need for such a device; instead, there must be a device that performs very complex image-to-meaning-to-syllable translation, as well as direct image-to-syllable matching. This follows from the way the symbols are configured. There are other solutions to reading: hieroglyphs employ a rebus principle, rather than an alphabet. They may be read in various ways, depending upon the context, and on certain marker symbols that can be embedded in the text. Reading them involves a very different structural arrangement of cognitive modules than reading alphabetic material.

Neuropsychological dissociation of modules is taken as good evidence that the modules exist as functional units in the brain (Shallice, 1988). Presumably Shallice would scour the hospitals for speakers of ancient Egyptian who happened to read hieroglyphs, hoping to encounter various types of dyslexics; the academic purpose of the exercise would be to establish that ancient Egyptian reading did in fact break down in ways consistent with its putative internal modular structure. Thus, one might find dyslexics who have specifically lost the linkage between image, meaning, and sound at various points or in various combinations. The point of studying these pathologies would be to construct a functional model of the ancient reader’s brain.

But why stop with reading? Every culturally bound skill deserves a similar neuropsychological dissection, and any dissociable, informationally encapsulated module that emerged would similarly receive the distinction of its own modular structure within the larger cognitive architecture. Reading attracts more attention because of its importance, but virtually any highly overpracticed skill is bound to have a distinctive modular structure that can break down in various predictable ways. This is not a frivolous notion: the brain of a professional tennis player is undoubtedly employing its resources in a very different way than it would have if, for cultural reasons, the same individual had grown up instead to become a very unathletic biblical scholar.

This idea receives some support from recent neurophysiology, especially in the study of cortical plasticity, or the malleability of the central nervous system. It has been known for some time that the immature brain is highly plastic; that is, it can grow connections, and lose connections, in many different ways, depending on early experience. Changeux (1985) has proposed an epigenetic theory of brain development, in which the young brain proliferates new connections fairly indiscriminately, that is, invents many possible routes of development, of which only a few will survive, due to the selective effects of experience. His ideas are a mirror-image of the neurological solution offered earlier by Hebb (1949) of selective synaptic growth.

Changeux was aware of the cultural corollary of this notion: namely, that culturally specific, highly redundant patterns of brain use would imply the existence of cultural traditions that have an indirect neurological instantiation—that is, the brains of many individuals in a particular culture are broadly programmed in a specific way, while in another culture they may develop differently, because use patterns are fundamentally different. This idea has gained credibility from recent studies of monkeys, using single-cell mapping of cortical regions. Merzenich (1987) and his colleagues have demonstrated that the cortical sensory region for the hand expands and contracts with demand, even in the adult. That is, the detailed topographic maps, or sensory fields, that project various skin sensations onto the somatosensory cortex are not static, fixed maps but dynamic computational resources. They recruit adjacent cortical columns when demand is heavy, that is, when the animal is required to make fine discriminations, and shrink back when de...

Table of contents

• Cover
• Title
• Copyright
• Contents
• Prologue
• 1 The Need for a Theory of Cognitive Evolution
• 2 Darwin’s Thesis
• 3 Wernicke’s Machine
• 4 The Chronology of Anatomical and Cultural Change
• 5 Primate Cognition: Episodic Culture
• 6 First Transition: From Episodic to Mimetic Culture
• 7 Second Transition: From Mimetic to Mythic Culture
• 8 Third Transition: External Symbolic Storage and Theoretic Culture
• 9 Consciousness and Indeterminacy
• References
• Acknowledgments
• Index