Department of Social Anthropology
University of Manchester
Manchester M13 9PL
Revised, October 1997
Second revision, February 1999
We human beings know a great deal. But we are able to be so knowledgeable only because we stand on the shoulders of our predecessors. As Durkheim noted long ago (1976 : 435), 'to that which we can learn by our own personal experience [is added] all that wisdom and science which the group has accumulated in the course of centuries'. The problem, which has remained at the heart of anthropological attempts to understand the dynamics of culture, is to fathom how this accumulation occurs. How is the experience that we gain during our lifetimes enriched by the wisdom of our ancestors? And how, in turn, does that experience make itself felt in the lives of descendants? More generally, in the creation and maintenance of human knowledge, what contribution does each generation make to the next?
One approach to answering this question, though it has venerable antecedents, has undergone something of a revival in recent decades thanks, in large measure, to parallel developments in cognitive science. It holds that knowledge exists in the form of 'mental content' which, with some loss and replenishment, along with diffusion at the margins, is passed on from generation to generation as the heritage of a culture-bearing population. Among the leading anthropological exponents of this approach is Dan Sperber. My aim here is to review some of the central strands of Sperber's arguments, and then to show why I think they are incoherent. I focus on Sperber's work not because it should be considered especially representative, but rather because it has the virtue of rendering unusually explicit the assumptions built into much contemporary theorising about culture and cognition, and of driving them through to their logical conclusions. If the conclusions are absurd, as I believe they are, then there must be something wrong with the founding assumptions.
These assumptions are, specifically, that knowledge is information, and that human beings are devices for processing it. I shall argue, to the contrary, that our knowledge consists, in the first place, of skill, and that every human being is a centre of awareness and agency in a field of practice. Building on these latter premises, I shall go on to suggest an alternative approach - one that owes more to phenomenological, ecological and 'practice-theoretical' perspectives on perception and cognition than to classical cognitive science - which, in my view, offers a more promising way forward. I realise there is currently a vigorous counter-movement within cognitive science itself, that is following a path very similar to the one proposed here. My critique, therefore, is directed against cognitivism in its 'classical' guise, rather than against its 'emergentist' alternative (for an excellent account of the latter, see A. Clark 1997). It is fair to say, however, that the classical perspective remains the dominant one in cognitive psychology; moreover its continued dominance is reinforced by a powerful alliance with evolutionary biology in its modern, neo-Darwinian formulation. Thus to take issue with classical cognitive science is inevitably to call into question some of the founding precepts of neo-Darwinism.
In both biology and psychology, as I shall show, the crux of the problem lies in understanding processes of ontogenetic development. Following an outline of Sperber's approach to the explanation of culture, I shall consider the problem of development as it figures, first, in neo-Darwinian biology, and secondly, in cognitive science. The solution, I argue, is to move beyond the dichotomy between innate capacities and acquired competencies, through a focus on the emergent properties of dynamic systems. Skills, I suggest, are best understood as properties of this kind. It is through a process of enskilment, not enculturation, that every generation grows into - and beyond - the wisdom of its predecessors. This leads me to conclude that in the growth of human knowledge, the contribution that each generation makes to the next is not an accumulated stock of representations but an education of attention.
More than fifty years ago, Alfred Kroeber was reflecting on the apparent analogies between cultural and biological phenomena. It would be wrong, he observed, to compare the individuals of a culture to members of a species. For the elements that combine to establish the particular pattern of thought and behaviour for a people cannot be traced to a common ancestral source but are of the most diverse origin. Following the convention of the time, Kroeber called these elements 'culture traits'. And if anything is comparable to a species, he thought, it is the trait (or trait cluster) itself. Where the species exists as a population of individual organisms of a certain kind, so likewise, the trait exists as a population of exemplars. Every act of making a stone axe of a particular form, or every utterance of a phrase with a particular grammatical construction, would be a member of such a population. And these individuals of diverse trait-species associate together to form the manifold patterns of human life, just as individuals of different organic species associate to establish the distinctive patterns of fauna and flora characteristic of particular locales. Thus, 'it is ecological aggregates to which cultures can be compared: local associations of species of diverse origin' (Kroeber 1952 : 93).
Kroeber's original idea has recently resurfaced, in a strikingly similar form, in Sperber's (1996) call for an epidemiology of representations. One has only to substitute for the currently unfashionable notion of 'trait', with its somewhat behavioural overtones, the rather more mentalistic notion of 'representation', and the two formulations become virtually identical. Just as the trait, according to Kroeber, is realised in its countless exemplars, so for Sperber, every representation exists on the level of concrete reality as the population of its 'tokens', whether these be found inside human brains or in the bodily behaviour which they set in train. There are, for example, millions of tokens of the story of Little Red Riding Hood, in the heads of everyone who can tell it, and in every event of telling. And where Kroeber compared traits to the plant and animal inhabitants of a locale, Sperber compares representations to the disease-causing micro-organisms that inhabit the body. Studying the proliferation and distribution of representations in brains is thus akin to the epidemiological study of the proliferation and distribution of micro-organisms in bodies: 'cultural phenomena are ecological patterns of psychological phenomena' (Sperber 1996: 60).
If knowledge consists in the representations that populate human brains, then the question which I posed at the outset concerning the contribution, in the accumulation of knowledge, that each generation makes to the next, may be rephrased as follows: How are representations transmitted? In other words, how does a representation in your brain find its way into mine, and from my brain into the brains of yet other people? To this question, Sperber offers a kind of answer quite different from that originally suggested by Kroeber. 'Man', Kroeber had pronounced (as long ago as 1917), 'is ... a tablet that is written upon' (1952: 32). The structure and nature and texture of humanity is such that it can be inscribed with any kind of cultural message you please; the critical distinction between the human and the animal is simply that the former is inscribable and the latter is not. This view of the human organism (or more specifically, of the brain) as a tabula rasa, however implausible, has served ever since to underwrite the more relativistic claims of cultural anthropology. It implies that students of culture need be no more concerned with the psychology of human nature than, say, journalists with the technology of paper-making.
Sperber's contention is that this traditional model of enculturation, as a simple process of inscription, rests upon an impossible psychology. Even if, in my behaviour, I 'write out' representations in my brain, this is not equivalent to 'writing them down' in yours. The chain of causation that leads from the concrete presence of a representation in one brain to its establishment in others is less direct. Sperber explains it by means of a distinction between 'mental representations' and 'public representations'. I have a tune in my head: that is a mental representation. I whistle the tune as I walk down the street: that is a public representation. It is public because it exists as a sound pattern that can be heard by other people in the vicinity. For some hearers it may form no lasting impression; for others, however, it may be not only heard but also remembered. For these latter, the tune is now established as a mental representation inside their heads. And they, too, may find themselves whistling it as they walk down the street, just as I did. A transmission of information has thus been effected, but only because every step of behavioural externalisation (which transforms the mental representation into a public one) is complemented by a further step of perceptual internalisation (which transforms the public representation 'back' into a mental one). And this latter step calls for the operation of a computational device capable of processing the input of sensory data, such as that generated by the impact of my whistling on the receptor organs of hearers, into an enduring representational form.
In short, some sort of cognitive processing device must already be installed, in human brains, before any transmission of representations can take place at all. A tabula rasa could not learn, since it would have no means to convert the sensory input into mental content. Once this point is recognised, however, we have also to admit that the processing device (or devices) may be better able to handle some kinds of input than others. We all know that some things, even when they are long and complicated like a story, are easy to remember, while other things - like lists of eleven-digit telephone numbers - tax our abilities to the limit. This, Sperber tells us (1996: 74-5), is simply because the brain's inbuilt cognitive mechanisms are specially equipped to handle objects with a narrative structure. If knowledge in non-literate cultures largely takes the form of myths and stories, this is because these forms are readily memorable. What cannot be easily remembered will naturally fall out of circulation and will not, therefore, be retained in the culture. In so far as the mechanisms of cognition determine what is memorable and what is not, they have a quite immediate impact on the organisation of cultural knowledge.
Given that these cognitive mechanisms, or processing devices, must be in place prior to any transmission of cultural information, where do they come from? Of course the devices could themselves have been learned. The child, in other words, may first acquire representations that specify the mechanisms for processing subsequent input. Yet how could those initial representations have been established in the child's mind unless some devices were already installed for selecting and processing the relevant, mechanism-specifying input? The problem is analogous to the one about how to send a message, in code, to a recipient who lacks the key. First you have to send another message, which specifies the key. But then the recipient must already be in possession of another key, in order to decode that message ... and so on and on, in an infinite regress. Unless, at some basic level, both sender and recipient possess a common set of interpretative devices, or 'framings', the communication of information could never get off the ground at all.
The bottom line, then, is that every human being must come into the world pre-equipped with cognitive mechanisms that are specified quite independently, and in advance, of any process of learning or development. For John Tooby and Leda Cosmides, whose investigations into the psychological foundations of culture have followed a path very close to Sperber's, these mechanisms amount to what they call 'human metaculture', a bedrock of universals bequeathed to each and every one of us by virtue of our shared evolutionary ancestry. It is thanks to these metacultural framings that human beings are able to learn the variable features of their particular cultural traditions. Were it not for them, it would be impossible for the already enculturated adult to communicate with the new-born infant who 'arrives in the culture free of any knowledge about its particularities' (Tooby and Cosmides 1992: 92). Like the ethnographer in a world of strangers (Sperber 1985: 62-3), the infant can fall back on the ready-made information-processing devices which it shares with those around it, in order to gain admission to an otherwise closed world of cultural understanding.
I shall return below to the efforts of Tooby and Cosmides to establish an 'evolutionary psychology'; for the present it suffices to stress their conclusion, fully endorsed by Sperber, that the cross-generational transmission of variable cultural information depends upon the presence, in all human minds, of innate, species-typical mechanisms of cognition. These mechanisms, it is supposed, are the outcomes of a Darwinian process of variation under natural selection, and as such, they are built to specifications that are not cultural but genetic, included within the common biological endowment of humankind.
According to a now well-established scenario, it was during the Pleistocene era, while living as hunters and gatherers, that human beings evolved to be the kinds of creatures they are today. Since natural selection, as a rule, adapts organisms to their prevailing conditions of life, we can expect that the properties or 'design features' of the human mind, just as much as those of the body, would have evolved as solutions to the particular problems and challenges that would have been faced by ancestral populations of hunter-gatherers in Pleistocene environments. Moreover, there is good reason to believe that a cognitive architecture consisting of a collection of relatively discrete modules, each specialised in a particular domain of problem-solving, would have a selective advantage over a more general-purpose design. A specialised module that already knows, in a sense, what the problem is and how to deal with it, can deliver a more rapid and effective response, causing minimal interference to other cognitive tasks that may be going on at the same time. Thus there might be one module for navigation and orientation in the environment, another for handling social co-operation with conspecifics, another for the recognition and classification of animals and plants, another for language acquisition, another for tool-use, and so on (Hirschfeld and Gelman 1994).
The environments of ancestral hunter-gatherers, however, were very different from those encountered by the majority of the world's inhabitants today. Many of the challenges they faced have all but disappeared, while others have arisen that they could not have anticipated. Thus cognitive modules designed by natural selection for one purpose have, throughout history, been turned to account in other ways. Anyone can learn to drive a car, Sperber speculates (1996: 93), because the skills of driving call for computations of space and movement that the brain is innately pre-equipped to carry out. The requisite processing devices would have evolved in what Sperber calls their proper domain, that is, in the solution of cognitive tasks faced by hunter-gatherers in moving around in the terrain. But in driving a car they are mobilised in the actual domain of the motorist who has to make his way on the road. Though the circumstances could not be more different, the underlying cognitive operations are much the same. Indeed the implication of Sperber's argument is that any mode of locomotion that could not draw upon evolved cognitive capacities of one kind or another would probably be unlearnable, and could never become a part of culture.
More generally, while the natural environment of human beings has been largely replaced or overlain by a cultural one - that is, by an environment consisting of 'all the public productions ... that are causes and effects of mental representations' (Sperber 1996: 115) - the various domains of culture in which human cognition actually operates have been shaped by a selective bias in favour of representations that mimic the inputs of evolved cognitive modules in their original, proper domains. To rephrase the point in terms of another distinction suggested by Sperber, the human mind-brain is peculiarly susceptible to representations that are compatible with its innate dispositions. Such representations will proliferate and spread, and so establish themselves within the culture, whereas others - failing to satisfy the input conditions of the cognitive modules - will decline and disappear. Culture, in short, is parasitic upon the universal structures of human cognition.
Now behind the linked oppositions between dispositions and susceptibilities, and between proper and actual cognitive domains, there clearly lies a more fundamental distinction, namely between innate devices and acquired representations. A disposition is a function of a genetically specified cognitive device that has evolved within its proper domain as part of the human adaptation to the original environmental conditions of hunter-gatherer life in the Pleistocene. A susceptibility lies in the receptivity of a particular device to mental representational content of a certain kind, that happens to be current within the actual domain of the cultural life of a human population - a form of life which may not remotely resemble that of their earliest ancestors. It must follow, however, that those competencies that have an innate dispositional basis - that we are 'born with' - have to be clearly distinguished from those that are founded in acquired mental content. The first are products of an evolutionary process, the second belong to a process of history. Whereas evolutionary change is fundamentally genetic, the history of a population consists - according to Sperber (1996: 115) - in changes in its pool of cultural representations. And throughout the entire course of history, despite the turnover in the composition of this pool, the evolved architecture of the human mind has remained essentially constant.
I should like to begin my critique of Sperber's epidemiological theory of culture by focusing on its evolutionary implications. My purpose is to do away with the opposition between innate cognitive mechanisms and acquired cultural content by showing how the forms and capacities of human beings, like those of all other organisms, arise within processes of development. This leads me to a conception of evolution which, though radically at variance with the orthodox neo-Darwinian account, no longer forces us to reserve a separate ontological space for human history. And this, in turn, will eventually open the way to a quite different answer to our original question: how does each generation contribute to the knowledgeability of the next?
Sperber's obsession - widely shared in cognitive science - with the discovery of innate, genetically determined information-processing devices is at first glance rather puzzling, given his explicit appeal to neo-Darwinian evolutionary biology. For most biologists claim that they have long since discarded the distinction between innate and acquired structures. According to what is often called the 'first law of biology', the actual characteristics of organisms are neither innate nor acquired, but are products of the interaction, throughout the life-cycle, between endogenous, genetic causes and exogenous, environmental ones. Thus interactionism has long since replaced innatism as the dominant creed within biological science. In fact, however, a doctrine of genetic preformation still lurks beneath the surface of orthodox interactionism, since it is built into biology's own master theory - the theory of evolution under natural selection. The synergy between neo-Darwinian biology and classical cognitive science is thus, after all, closer than meets the eye, and both fail for the same reason: they are unable to offer an adequate account of ontogenetic development. Let me show why.
Interactionism describes development as an unfolding relation between genes and environment. In this relation, however, it is the genes that are supposed to hold the essence of form, whereas the environment is conceived merely to furnish the material conditions for its realisation. Each gene is taken to represent a unit of pure, digital information, encoded in the molecular structure of DNA. Put together, these units make up what is called the genotype, a formal specification of the organism-to-be which, by definition, is given independently and in advance of any real-world context of development. At the commencement of every new life-cycle, this genotypic specification is introduced, by way of the DNA of the germ cells, into a particular environmental context. In development, the information carried in the genes is then said to be outwardly 'expressed' in the phenotypic form of the resultant organism. Only the elements of the genotype, however, and not the characteristics of the phenotype, are transmitted across generations. Over many generations within a population, through accidents of mutation and recombination coupled with the effects of differential reproduction, the informational content of the genotype changes. These changes, it is claimed, add up to a process of evolution.
This is all very neat, save for one problem. To be sure, every organism starts life with its complement of DNA. But if genes are to be understood, as the theory requires, as the carriers of a formal design specification, shaped up through natural selection, from one locus of development to another, then there must be some systematic correspondence between the elements of this specification and the actual DNA of the genome that is independent of any developmental process. The existence of such a correspondence has been generally assumed, but it has never been demonstrated (see Cohen and Stewart 1994: 293-4). In practice, what happens is that biologists seek to redescribe the observed phenotypic characteristics of organisms as the outputs of a formal system of epigenetic rules (much as linguists seek to redescribe spoken utterances as outputs of a generative syntax). These rules are then read 'in' to the genome, so that development can be seen as the 'reading off' of a programme or specification that is already there, and that is imported with the genome into the site of inauguration of a new life cycle. In short, as an account of the evolution of form, neo-Darwinian theory rests on a simple circularity. That is one reason why it has proved so hard to refute.
At root, the issue comes down to one about copying. The orthodox account has it that the formal characters of the incipient organism are copied along with the DNA, in advance of its interaction with the environment, so that they can then 'interact' with the environment to produce the organism. I would argue, to the contrary (and as illustrated schematically in Figure 1), that copying is itself a process that goes on within the context of organism-environment interaction. In other words, the 'missing link' between the genome and the formal attributes of the organism is none other than the developmental process itself. There is, then, no design for the organism, no genotype - except, of course, as this might be constructed by the observing biologist. Organic form, in short, is generated, not expressed, in development, and arises as an emergent property of the total system of relations set up by virtue of the presence and activity of the organism in its environment. And if that is so - if form is a property not of genes but of developmental systems - then to account for the evolution of form we have to understand how these systems are constituted and reconstituted over time.
These arguments are not new. One of their most influential proponents in recent years has been Susan Oyama (1985). The nature of an organism, as she points out, 'is not genotypic ... but phenotypic', and therefore 'depends on developmental context as profoundly and intimately as it does on the genome ... Evolution is thus the derivational history of developmental systems' (Oyama 1989: 5). Yet however much she has been at pains to distinguish her views from orthodox interactionism, her critics continue to confuse the two, claiming that there is nothing in a developmental systems approach that is not perfectly consistent with the premises of neo-Darwinian evolutionary biology (for an example, see Dunbar 1996). In this volume, Pascal Boyer reacts in precisely the same way to the arguments I adduce here. Indeed he portrays me as an advocate of the view that 'development consists of an "interaction" between prior information and external circumstances'. And he wonders why this should be incompatible with a neo-Darwinian view of the evolution of human capacities. Of course there is no incompatibility here! Some kind of mental block, however, appears to prevent Boyer, and many others who think like him, from realising that to regard form as emergent within the developmental process is anything other than a version of gene-environment interactionism.
The source of this block seems to lie in an assumption that organisms are effects of genetic and environmental causes. However the reality, as Daniel Lehrman warned many years ago, is far more complex. For the interactions from which the development of an organism proceeds are not between genes and environment, but between organism and environment. At every moment of the developmental process, formal structures or behavioural dispositions already established in the course of previous interaction are implicated, through further interaction, in the generation of new ones (Lehrman 1953: 345). And it is not in any of the components of the interacting systems, taken individually, that the constraints on the process are to be found, but rather in the relations among them (Oyama 1993: 8). Thus it is simply not possible to apportion causality between genetic and environmental factors. 'The web of causality', as Esther Thelen writes, 'is intricate and seamless from the moment of birth' (1995: 94). Or to put it another way, organisms are both cause and effect of themselves (Goodwin 1988: 108). In short, causation is not a relation between things - genes and environmental factors on the one hand, organisms on the other - that are external to one another, but is immanent in the developmental process itself.
Having said that, my claim that the genotype does not exist is likely to remain contentious, to say the least. Plotkin (this volume) goes so far as to call it 'an instance of biological illiteracy'! Let me, then, make my position absolutely clear. I do not doubt the existence of the genome, or that it sets in train processes that are crucial to the development of the organism at every stage of its life-cycle. Nor, moreover, do I deny that the composition of the genome changes across generations through a process of natural selection. What I do deny is that the DNA sequence in the genome encodes a context-independent design specification, and with it, the idea of natural selection as a design agent. Now Plotkin might possibly accept the first of these denials, since he dismisses as a 'red herring' the claim that biologists 'think the genome contains information'. It is difficult to know what to make of this dismissal, however, since it follows hard on the heels of an equally forthright insistence that the source of the order and information that cannot be found in environmental experience 'must lie in the genes'. I certainly have no problem with Plotkin's definition of evolution as 'the transformation of biological forms and systems in time', and he appears to agree with me (and Oyama) that development is the link between the genome and organic form. But what Plotkin fails to recognise is that if the idea of the genome as a carrier of information is indeed a red herring, then the theory of variation under natural selection, though it can account for changes in gene frequencies over successive generations within a population, is powerless to explain the evolution of biological form.
Genes, after all, are but segments of molecules that may or may not have unspecified consequences for the organisms they are in. It is perfectly possible, as Cohen and Stewart have demonstrated, for two quite different creatures to have precisely the same DNA in the genome. Not only can creatures evolve without any genetic change at all, they can also retain a more or less constant form despite considerable modification at the genetic level (Cohen and Stewart 1994: 309). Thus natural selection, leading to changes in the composition of the genome, occurs within evolution, but does not explain it. Nor does it even furnish a partial explanation, since to determine what part is due to natural selection, and what part is not, would require an apportionment of causal responsibility for the development of form between genes and environmental experience. And this, as I have shown, cannot sensibly be done. Only by going beyond the theory of evolution through variation under natural selection, and by considering the properties of dynamic self-organisation of developmental systems, can we hope to discover the possible consequences of those changes that can be explained by natural selection for the evolutionary process itself.
I should now like to return from biology to psychology, and to the problem that cognitive science has with its persistent appeal to innate structures. Just as neo-Darwinian biology has to posit a design for the organism, the genotype, that is specified in advance of its development, so cognitive science posits a design for the mind that pre-exists and underwrites all subsequent learning or knowledge acquisition. In so far as this design is assumed to have a genetic basis, and to have been fashioned by natural selection, it must form one component of the genotype. Yet here cognitive science runs into the very same dilemma that, as we have seen, derails neo-Darwinian theory - in an even more pronounced form. It is more pronounced since the genetically specified cognitive devices that are supposed to make possible the transmission of representations must already exist, not merely in the virtual guise of a design, but in the concrete hardwiring of human brains. In other words, it has to be assumed that DNA replication not only copies a design for the mind into the embryonic human being, but also copies out the mechanisms specified in that design into its head. Somehow, in order to kick-start the process of ontogenetic development, strands of DNA have miraculously to transform themselves into computational modules. This is rather like supposing that merely by replicating the design of an aircraft, whether on the drawing board or on the computer screen, one is all prepared for take-off.
By and large, in the literature of cognitive science, the postulation of innate structures is taken to require no more justification than vague references to genetics and natural selection (e.g., Johnson-Laird 1988: 35). Where the issue of development is addressed at all, the arguments are confused and contradictory. A case in point lies in the work of Tooby and Cosmides, to which I have already referred. Recalling the biological distinction between genotype and phenotype, Tooby and Cosmides argue that an equivalent distinction needs to be recognised in the study of mind - between evolved psychology and manifest psychology. Every situation, then, may be analysed into 'environmental conditions, evolved architecture, and how their interaction produces the manifest outcome' (1992: 45-6). This sounds like a restatement of orthodox interactionism. Yet if what they call the 'architecture' of the mind were truly analogous to the genotype, it would exist only in the programmatic form of a 'building design' regulating the construction, in the course of ontogenetic development, of a suite of cognitive mechanisms. Not all the features specified in this design, as Tooby and Cosmides go on to recognise, will be realised at once for any single individual. Thus different mechanisms will come on line at different moments in the life cycle; moreover they may be revealed in some individuals but not others, depending (among other things) on the environmental circumstances they have encountered.
From this it would appear that the mechanisms themselves should be understood as aspects of manifest rather than evolved psychology. Confusingly, however, Tooby and Cosmides more often use the phrase 'evolved architecture' to refer to these manifest structures than to the underlying design. Indeed they openly admit that their usage is ambiguous, that they do not bother to distinguish terminologically between 'expressed adaptive architecture' and the more fundamental programmes that underwrite its construction (1992: 82). Presented as a matter of expository convenience, this conflation of the manifest and the evolved is more in the nature of a cover-up that enables them to get away with talking about cognitive mechanisms as if they were already in place and fully operational, constructed by the great master-builder of evolution, natural selection, before any development can get under way at all. And it enables them to claim, almost in the same breath, that the 'evolved architecture' selects what is developmentally relevant in the environment, and that the environment is implicated in the very development of the architecture (1992: 84-7).
Sperber runs into precisely the same dilemma, though it is not elaborated to the same degree. In this case, the problem revolves around the status of the innate dispositions which, as we have seen, are defined as functions of evolved, genetically determined and domain-specific cognitive modules. Let us suppose, in accordance with orthodox evolutionary biology, that the genotype includes a set of instructions for constructing the modules, and the dispositions that are given in their operation. It would follow that the dispositions themselves emerge in the course of ontogenetic development. Sperber confirms this inference, noting that dispositions will only develop if 'appropriate environmental conditions' are fulfilled (1996: 67). Yet elsewhere, he assures us that the dispositions are there to begin with, and that they cause the developing child to attend selectively to information from the surrounding environment that is relevant to his or her becoming 'a competent speaker, a competent climber, thrower, catcher, eater, drinker, a competent manipulator of objects, a competent recogniser of animals, a competent predictor of other people's behaviour, and so on' (1996: 117). In this account of development, the process starts not with a plan for constructing cognitive modules that is as yet unrealised, but with preconstituted modules whose 'needs' for information are as yet unsatisfied. The structures are already in place, but they are initially empty of informational content. Development is about filling up the modules, not about their construction.
The core of the problem, however, lies in the seemingly innocuous phrase, 'appropriate environmental conditions'. For the development of dispositions, according to Sperber, the appropriate conditions are those of the module's proper domain - the domain for which it originally evolved as an evolutionary adaptation. If the conditions are significantly altered, so that they are no longer appropriate in this sense, then we would expect that the dispositions would either not develop at all, or develop along different lines. Yet Sperber appeals to the 'innateness' of dispositions to assert the contrary: that the dispositions are there anyway, whatever environmental conditions happen to obtain, and that environmental differences are registered in development solely through the susceptibility of modules to diverse representational content. Tooby and Cosmides advance much the same claims, arguing on the one hand that the concrete mechanisms making up the evolved architecture are 'reliably constructed' under all normal environmental circumstances, but on the other that these universal mechanisms proceed to work on 'variable environmental inputs' to produce the diversity of manifest competencies and behaviours that we actually observe (1992: 45).
Let me unpack these claims (illustrated schematically in Figure 2) with reference to a specific and much-vaunted example, that of language acquisition. Here, the alleged universal mechanism is what has come to be called the 'language acquisition device'(LAD). During a well-defined stage of infancy, this device is said to be activated, operating upon the input of speech sounds from the environment so as to establish, in the infant's mind, the grammar and lexicon of the particular language (or languages) spoken in his or her community. An infant reared in social isolation, and thus deprived of the relevant environmental input, would not learn a language, but would still possess a fully formed LAD (Tooby and Cosmides 1992: 45). It would thus appear that language acquisition is a two-stage process: in the first, the LAD is constructed; in the second it is furnished with specific syntactic and semantic content. That, at least, is the theory, but is it borne out in practice? Is there any basis in reality for separating out the construction of 'innate' psychological mechanisms from the transmission of 'acquired' cultural representations, as shown in Figure 3, or is the division into these two stages no more than an artefact of our own analytic procedures? In what follows, I shall argue that the latter is the case.
The first point to note is that the mechanisms (if we can call them that) underwriting the child's ability to speak are not constructed in a vacuum, but rather emerge in the context of his or her sensory involvement in a richly structured environment. From birth, if not before, the infant is immersed in a world of sound in which the characteristic patterns of speech mingle with all the other noises of everyday life, and is surrounded by variously accomplished speakers who provide support in the form both of contextually grounded interpretations of the infant's vocalisations and of demonstrations, or 'attention-directing gestures' (Zukow-Goldring 1997: 221-3), to accompany their own. This environment, then, is not a source of variable input for pre-constructed mechanisms, but rather furnishes the variable conditions for the self-assembly, in the course of early development, of the mechanisms themselves. And as the conditions vary, so the resulting mechanisms will take manifold forms, each differentially 'tuned' both to specific sound-patterns and to other features of local contexts of utterance. These variably attuned mechanisms, and the competencies they establish, are of course the neurophysiological correlates of what appear to us as the diverse languages of the world. It is not, then, by way of the conveyance of specific syntactic and semantic content that the child develops the capacity to speak in the manner of his or her community. Language, in this sense, is not acquired. Rather, it is continually being generated and regenerated in the developmental contexts of children's involvement in worlds of speech (Lock 1980). And if language is not acquired, then there can be no such thing as an innate language acquisition device (Dent 1990).
What applies specifically in the case of language and speech also applies, more generally, with regard to other aspects of cultural competence. Learning to throw and catch, to climb, or to eat and drink, to cite a few of Sperber's examples (1996: 117), is a matter not of acquiring from an environment representations that satisfy the input conditions of preconstituted modules, but of the formation, within an environment, of the necessary neurological connections, along with attendant features of musculature and anatomy, that establish these various competencies. To underline the contrast between Sperber's position and mine, let me return to the question of copying. For Sperber, a design for the mind is copied, along with the DNA of the genome, at the inception of every new life-cycle. And this design, prior to being opened up to the differentiating influences of the environment, magically turns itself into concrete mechanisms in the brain, ready and prepared to process relevant environmental input. I have argued, to the contrary, that copying is itself a developmental process, that this process takes place in an environmental context, and that it alone provides a link between the genome and the formal properties of the organism - including those of its brain (see Figure 1).
In a sense, then, the architecture of the mind is a result of copying; that copying, however, is not an automatic transcription of cognitive devices (or instructions for building them) from one head to another, but a matter of following, in one's actions, what other people do. In this sense, of imitation rather than transcription, copying is an aspect of a person's life in the world, involving repeated tasks and exercises - or what Whitehouse (1996: 113) aptly calls 'the labours of maturation'. It is through the work of copying, then, that the neurological foundations of human competencies are established. This is not to deny that the resulting neural organisation may take a modular form; it is to insist, however, that modularity develops (Ingold 1994: 295), and that the precise way in which this packaging occurs will depend upon the particularities of environmental experience.
To dispel any possible misunderstanding, I should stress that my purpose is not to argue for the priority of nurture over nature, or to substitute for the innatist bias of Sperber's account a doctrine of the environmental determination of human capacities. My argument is not caught, as is Plotkin's (this volume), between the two poles of what he calls the 'great game', with one side positing the mind as a blank slate and the other insisting that it comes with a ready-made architecture. These are not the only theoretical alternatives, and indeed both are fallacious for the same reason, most succinctly expressed by Oyama, namely that the information specifying the capacities in question - whether its source be supposed to lie inside the organism or outside in the environment - must be presumed to 'pre-exist the processes that give rise to it' (Oyama 1985: 13). My point is that these capacities are neither internally prespecified nor externally imposed, but arise within processes of development, as properties of dynamic self-organisation of the total field of relationships in which a person's life unfolds.
An example is the ability to throw things with some precision, and to catch them by hand. Like walking on two feet, this appears to be one of the hallmarks of our species. Yet there are, in practice, any number of different modes of throwing, and of catching, appropriate to different activities and settings. Hurling a javelin, putting shot and bowling in cricket all call for different patterns and sequences of muscular tension, and for different judgements of pace, angle and spin. There is no 'essence' of throwing and catching, however, underwriting these variations in actual performance (Thelen 1995: 83). In every case, the particular capacities of perception and action constituting the motor skill are developmentally embodied into the modus operandi of the human organism through practice and training, under the guidance of already accomplished practitioners, in an environment characterised by its own textures and topography, and littered with the products of previous human activity. To adopt a felicitous expression from Kugler and Turvey (1987), the components that actually produce the limb trajectories involved in throwing and catching are not hard-wired but 'softly assembled'.
Though it is customary to speak of a capacity such as throwing/catching as a human universal, as distinct from its particular manifestations, we can do so only by artificially 'bracketing out' all variations of context, and by focusing exclusively on what every conceivable situation of development has in common. For comparative analytic purposes, such attempts to sift the general from the particular, or to establish a lowest common denominator of development, may have their uses. But if universal capacities are nothing more than abstract precipitates of this analytic procedure, to claim that they are concretely instantiated in the heads of individuals, in the form of evolved mental modules, is patently absurd (Shore 1996: 17). This is the crux of my disagreement with Boyer (this volume). Given the human genotype, he says, 'certain developmental consequences are practically inevitable in normal circumstances'. But how are we to decide what circumstances are normal? For Boyer, circumstances are exceptional rather than normal if 'they were not part of the conditions prevailing when the genes in question were selected'. In that sense, most children in the world today are growing up in rather exceptional circumstances. Yet barring accident or handicap, they are nevertheless supposed to be equipped with the full suite of evolved capacities specified by naturally selected genes.
It must, then, be possible to adduce a core of concrete circumstances - what Boyer calls a 'standard epigenetic environment' - that is common to all situations in which humans learn, for example, to throw things, from the hunting ground to the cricket pitch. I cannot fathom what these circumstances might be. Or to take another example, we might admit that infants, in learning to walk, normally find themselves in situations where there is ground to walk on. Yet how could the infant encounter 'ground', as a concrete condition of development, not only as distinct from, but also prior to, such diverse 'walk-on-able' surfaces as sand, asphalt, meadow and heath, all of which call for different modalities of gait, balance and footwork? And how, again, could such a ground be free from all contours? Bizarre as it may seem, this is precisely how the ground beneath our feet would have initially to be experienced, if we are to hold on to the notion that culturally and environmentally particular modalities of walking are added on to a universally innate capacity for bipedal locomotion. And just the same kind of partitioning of the child's experience of the environment is entailed by the notion that competence in his or her mother tongue is acquired on the basis of a preformed 'language instinct' (Pinker 1994).
If, however, there are no innate psychological structures - no built-in architecture nor even any context-independent design specifications - then what, if anything, evolves? Earlier, with regard to the general problem of form in biology, I argued that since form arises within developmental systems, to account for its evolution we need to focus on the temporal unfolding of these systems and on their properties of dynamic self-organisation. Reconceptualised in this way, the evolutionary process becomes one in which organisms, through their presence and their activities, set up the developmental conditions under which their successors are destined to live out their lives. Likewise the manifold capacities of human beings, from throwing stones to bowling cricket-balls, from climbing trees to scaling ladders, from whistling to playing the piano, emerge through the labours of maturation within fields of practice constituted by the activities of predecessors. It makes no sense to ask whether the capacity to climb lies with the climber or the ladder, or whether the capacity to play the piano resides in the pianist or the instrument. These capacities exist neither 'inside' the body and brain of the practitioner nor 'outside' in the environment. They are rather properties of environmentally extended systems that crosscut the boundaries of body and brain (A. Clark 1997: 214). It follows that the work people do, in establishing environments for their own and future generations, contributes quite directly to the evolution of human capacities.
In his study of the computational tasks involved in maritime navigation, Edwin Hutchins observes that 'humans create their cognitive powers by creating the environments in which they exercise these powers' (1996: 169). This, for him, is the process of culture, though one could just as well call it history. But is there anything specifically human about this process? Hutchins compares the human navigator to the ant, which owes its apparently innate ability to locate food sources with uncanny accuracy to the trails left in the environment by countless predecessors. Wipe out the trails, and the ant is lost. So, indeed, would humans be, without culture or history. Hutchins's conclusion is that the capacities of the ant, too, are constituted within a historical process of culture. Alternatively (and amounting to much the same thing) we could conclude that the supposedly 'cultural' capacities of human beings are constituted within a process of evolution. My point is that history, understood as the movement wherein people create their environments, and hence themselves, is no more than a continuation of the evolutionary process, as defined above, into the domain of human relations (Ingold 1995a: 207-12). Having dissolved the distinction between the innate and the acquired, we find that the parallel distinction between evolution and history falls with it.
Up to now, I have used the words 'capacity' and 'competence', loosely and interchangeably, to describe aspects of human knowledgeability. Neither word, however, seems entirely appropriate for this purpose. The trouble with the concept of capacity is that it is rooted in the metaphors of container and content, of human psychology as a set of preconstituted, modular compartments or 'acquisition devices', waiting to be filled up with cultural information in the form of mental representations. The image of mind as container is shared both by Sperber and by many of those with whom he claims to take issue: orthodox cultural relativists who, as he puts it, are naive enough to believe that 'human mental abilities make culture possible and yet do not in any way determine its content and organisation' (1996: 57). What I find naive, to the contrary, is Sperber's belief that culture stands as 'content' to human psychology (Lave 1988: 85). Without doubt, people raised in different environments learn to perceive their surroundings, and to act within them, in different ways. We may even agree to call these differences cultural. As such, however, they are not so much received into the 'capacities' of a universal psychology as immanent in that field of relations wherein human beings undergo the organic processes of growth and maturation, and in which their powers of action and perception are developed and sustained.
But the notion of competence is equally problematic, largely because of the way its meaning has come to be constituted, particularly in the literatures of psychology and linguistics, through an opposition to performance. The notion suggests a knowledgeability that is detached from action and from the contexts of actors' bodily engagement with the world, and that takes the form of interior rules or programmes capable of specifying, in advance, the appropriate behavioural response to any given situation. Competence, as Dreyfus and Dreyfus have pointed out (1986: 26-7), underwrites the kind of process that, according to cognitive science, lies at the heart of all intelligent action, namely 'problem-solving'. The approach of the intelligent problem solver, in this view, is always to act on the basis of a plan, which is formulated by bringing a given set of decision rules to bear upon a representation of the existing situation. Thus the notions of capacity and competence are closely bound up with one another: where the former implies a built-in readiness to accept certain types of rules and representations, the latter inheres in this received mental content. The individual equipped with a capacity for language can acquire competence in English; the individual equipped with a capacity for throwing/catching can become a competent cricketer, and so on.
To think in these terms, however, is to treat performance, such as that of the English speaker or the bowler in cricket, as nothing more than the mechanical execution, by the body, of a set of commands generated and placed 'on line' by the intellect. It is to suppose that the performance begins with a plan which, since it contains a precise and complete specification of the behaviour to follow, is necessarily a structure of a very complex kind. The process of implementation, on the other hand, is assumed to be of clockwork simplicity. This is the approach taken by Sperber, and indeed generally in the classic tradition of cognitive science. But as David Rubin has pointed out (1988: 375), we could equally well take the opposite tack: that is to presuppose a simple structure, or even no structure at all, and to account for the performance as the unfolding of a complex process. Consider, for example, the movements of the woodsman, in felling a tree with his axe. A complex structure, simple process model would regard every swing of the axe as the mechanical output of a mental computational device installed inside the woodsman's head, designed to calculate the optimal angle of the swing and the precise force of the blow. A complex process model, by contrast, would regard the movement of the axe as part of the dynamic functioning of the entire system of relations constituted by the presence of the man, with his axe, in an environment that includes the tree as the current focus of his attention (Bateson 1973: 433). More generally, a model of the latter kind would treat performance not as the discharge of representations in the mind but as an achievement of the whole organism-person in an environment (Thelen 1995).
This calls for a fundamentally ecological approach, and that is the approach I adopt here. Its basic premise is that human knowledgeability is founded not in some combination of innate capacities and acquired competence, but in skill (Rubin 1988, J. E. Clark 1997). How, then, might an ecological account of skilled practice take us beyond what classical cognitive science describes as competent performance? We have seen that the competent performer of cognitive science is bound to the mechanical execution of a predetermined plan. Once set upon a course of action, he cannot therefore alter it without interrupting the execution in order to reconfigure the plan in the light of new data. The movement of the skilled practitioner, by contrast, is continually and fluently responsive to perturbations of the perceived environment (Ingold 1993a: 462). This is possible because the practitioner's bodily movement is, at one and the same time, a movement of attention; because he watches, listens and feels even as he works. It is this responsiveness that underpins the qualities of care, judgement and dexterity that are the hallmarks of skilled workmanship (Pye 1968: 22). As Nicholai Bernstein wrote some fifty years ago, the essence of dexterity lies not in bodily movements themselves but in the 'tuning of the movements to an emergent task', whose surrounding conditions are never precisely the same from one moment to the next (Bernstein 1996: 23, original emphases).
Moreover the more skilled the practitioner, the less 'working out' is needed: thus what distinguishes the expert from the relative novice is not the complexity or elaborateness of his plans or representations but the extent to which he can dispense with them. 'When things are proceeding normally', as Dreyfus and Dreyfus put it, 'experts don't solve problems and don't make decisions; they do what normally works' (1986: 30-1). This is not to deny that experts lay plans or formulate objectives. The woodsman has to select which tree to fell, and to decide upon the orientation of the notch so that it will ultimately fall in a direction that will not risk damage to surrounding vegetation. To observe him doing this, however, is to watch as he paces the woods, casting his eyes over different trees, sizing them up. In other words, it is to observe him feeling his way, in an environment, towards a goal that is conceived in anticipation of a future project. This kind of preparatory work, as Leudar and Costall have pointed out, 'is itself a mundane social activity, not a purely "intellectual" enterprise' (1996: 164). As such it calls, just as any other skilled practice, for powers of perceptual discrimination finely tuned through previous experience. Furthermore the 'plans' that the woodsman arrives at through this activity in no sense specify or determine the movements that follow, or the circumstances attending them, in all their concrete detail. What they do, rather, is to place him in a position of readiness, from which to launch into the subsequent project with a reasonable chance of success. Once the project is underway, he must fall back on bodily skills that he has already perfected (Suchman 1987: 52).
Now it is one thing to characterise the knowledge of the expert in terms of skill, quite another to claim, as I have done, that skill is the foundation of all knowledge. It would of course be foolish to imagine that we are experts in everything we do. In the course of anthropological fieldwork among reindeer herdsmen in Finnish Lapland, I had to learn to cast a lasso. Despite my best efforts, I remain incapable of capturing a moving animal of my choice amidst the swirling mass of deer in the roundup enclosure. The trouble is that lacking the intimate coordination of perception and action of the expert herdsman, I have to stop to think before casting the rope, in order to figure out how to proceed - by which time the reindeer, which is a good deal more skilled at evading the lasso than I am in casting it, has already moved out of reach. Indeed an account of how I set about casting the lasso, according to a series of precalculated moves, would at first glance appear to conform rather closely to the picture that classical cognitive science presents of the competent practitioner at work. For me it is indeed a puzzle, a problem to be solved. In what sense, then, can it be claimed that my rudimentary competence is nevertheless founded in skill? To answer this question I need to introduce a distinction, which is critical to my argument, between knowledge and information.
Consider a cookbook. The book is packed with information about how to prepare a selection of mouth-watering dishes. But is it in this information that the knowledge of the cook consists? Sperber would have it so. Thus the recipe for Mornay sauce, to cite one of his favourite examples, includes everything you need to know to prepare the sauce in your own kitchen. Nothing more is presupposed than the ability to read. Once the instructions have been transcribed into your head, all you have to do is to 'convert [them] into bodily behaviour' (Sperber 1996: 61). The conversion, however, is easier said than done. No known cookbook comes with such precise instructions that its recipes could be converted into behaviour just like that. When the recipe instructs me to 'melt the butter in a small pan and stir in the flour', I am able to follow it only because it speaks to my own prior experience of melting and stirring, of handling such substances as flour and butter, and of finding the relevant ingredients and utensils from the various corners of my kitchen (Leudar and Costall 1996: 163). The verbal commands of the recipe, in other words, draw their meaning not from their attachment to mental representations in my head, but from their positioning within the familiar context of my activity in the home. Like signposts in a landscape, they provide specific directions to practitioners as they make their way through a field of related practices - or what I have elsewhere called a 'taskscape' (Ingold 1993b: 158). Each command is strategically located at a point which the original author of the recipe, looking back on previous experience of preparing the dish in question, considered to be a critical juncture in the total process. Between these points, however, the cook is expected to be able to find her way around, attentively and responsively, but without further recourse to explicit rules of procedure - or in a word, skilfully.
Thus the information in the recipe book is not, in itself, knowledge. Rather, it opens up a path to knowledge, thanks to its location within a taskscape that is already partially familiar by virtue of previous experience. Only when placed in the context of skills gained through this prior experience does information specify a route that is comprehensible and that can practicably be followed, and only a route so specified can lead to knowledge. It is in this sense that all knowledge is founded in skill. Just as my knowledge of the landscape is gained by walking through it, following various signposted routes, so my knowledge of the taskscape comes from following the several recipes in the book. This is not knowledge that has been communicated to me; it is knowledge I have built up for myself by following the same paths as my predecessors and under their direction. In short, the growth of knowledge in the life history of a person is a result not of information transmission but of guided rediscovery. I shall return to this distinction below, since it has a critical bearing on how we understand the process of copying.
It is here, finally, that we can identify what distinguishes the real-life novice from the competent practitioner as classically depicted by cognitive science. Both may proceed in the same 'stop-go' fashion, periodically interrupting the flow of action in order to take stock of the situation and to plot a new course. But the deliberations of the novice are carried on not in an inner mental sanctum, closeted from the multiple domains of practical life, but in a real world of people, objects and relationships. The environment, then, is not merely a source of problems, of adaptive challenges to be resolved; it becomes part of the means for dealing with them. As Andy Clark nicely puts it, the mind is a 'leaky organ' that refuses to be confined within the skull but shamelessly mingles with the body and the world in the conduct of its operations (A. Clark 1997: 53). Every step of problem-solving is itself an exploratory movement within that world.
When, for example, lasso in hand, I adjust my body posture so as to be suitably poised for the next cast, I do not place my limbs into precalculated positions; rather, the preparatory adjustments of my posture are integral to the process of calculation itself. The 'calculator', in other words, is not a device inside the head but the whole person in the world (Lave 1988: 154). Along with parts of the body, aspects of the environmental setting are incorporated as integral parts of the 'device'. But this means, too, that calculation and implementation, far from being separate, successive stages of any operation, are merged as one and the same. As with a journey through the landscape, to have found your way to a place is to have actually arrived there, and not merely to have plotted a course for the trip. In the final analysis, then, there can be no distinction between solving a problem and carrying out the solution in practice, since every step of problem solving is itself a step along the way to implementation. And each step follows the next as successive movements of the perceptually aware agent - the 'person-acting' (Lave 1988: 180-1) - in the settings of practice.
How, then, to return to my original question, does each generation contribute to the knowledgeability of the next? The answer cannot lie in the transmission of representations - or of what D'Andrade (1981: 179) calls '"pass it along" type information' - since, as I have shown, this rests upon the impossible precondition of a ready-made cognitive architecture. In this penultimate section I want to offer an alternative suggestion: namely, that the contribution each generation makes to its successors amounts to an education of attention.
Earlier, I argued that there is no 'reading' of the molecules of DNA making up the genome of an organism apart from the process of development itself. We have now arrived at an almost identical conclusion with regard to human learning: there is no reading of a verbal script - such as that contained in the cookbook - that is not part of the novice's practical engagement with his or her environment. Once again, the issue comes down to one about copying. Recall that in Sperber's epidemiological model of cultural transmission, knowledge comes in the form of representations which pass from head to head by way of successive steps of behavioural externalisation and perceptual internalisation. But the copying of a public representation, perceived in the world, into a corresponding representation in the mind forms no part of the process by which the representation, once copied, is put into practice.
In its public form, for example, the recipe for Mornay sauce exists in the cookbook as 'an ink pattern on a piece of paper which can be read' (Sperber 1996: 61). To turn the public representation into a mental one, the aspiring cook need only be equipped with cognitive mechanisms for processing the input from this pattern into a corresponding set of images in the mind. In principle, the recipe could be reread, by generation after generation, without any cookery having to take place at all. Cooking, in Sperber's account (and as shown schematically in Figure 4), is not copying, it is the expression of copies that are already established in the mind of the cook. This view has its precise analogue in orthodox evolutionary biology which, as we have seen, supposes that development is the expression of a formal specification that has already been copied into the organism, through genetic replication, at the point of conception. Just as evolutionary theory imagines that the specifications of organic form, coded in genes, can be passed down the generations independently of the processes of development, so cognitive science imagines that cultural knowledge, coded in words or other symbolic media, can be passed on independently of its practical application in particular tasks and contexts.
But Sperber plays a trick with his Mornay sauce example. For if reading the recipe were really a matter of processing the visual input from an ink-pattern, then the resulting mental representation would consist of nothing more than an image of the written script. And the reader would implement the representation not by cooking at all, but by writing out the words of the recipe onto another piece of paper. The trick is to suppose that by forming an image of the ink-patterns, the aspiring cook has also formed in his mind an image of what those patterns are supposed to represent, namely a programme for activity in the kitchen. However Sperber's other example, the story of Little Red Riding Hood, gives the game away. He explains that the child whose mind contains a representation of the story may, if she so wishes, turn it 'into bodily, and more specifically vocal, behaviour' (1996: 62), simply by telling the tale. By the same token, the cook would convert the remembered recipe into behaviour by writing it down. To claim, as Sperber does, that the cook turns the mental representation of the written recipe into bodily behaviour by preparing the sauce makes about as much sense as supposing that the child converts the story of Little Red Riding Hood into behaviour by setting off to her grandmother's with a pot of butter and a cake!
A device capable of scanning the pages of a cookbook, and of processing the input data into stored images, could also be equipped, at least in principle, with a facility for printing out the received information. Yet such a device - which is what Sperber imagines the human being to be - would be quite unable to cook. How, then, do real-life human beings, as opposed to the animated processing devices of Sperber's imagination, learn the arts of cookery? They do so, of course, by copying the activities of already skilled cooks. To reiterate my earlier point, copying is not the automatic transcription of mental content from one head to another, but is rather a matter of following what other people do. The novice watches, feels or listens to the movements of the expert, and seeks - through repeated trials - to bring his own bodily movements into line with those of his attention so as to achieve the kind of rhythmic adjustment of perception and action that lies at the heart of fluent performance (Gatewood 1985). As Merleau-Ponty has observed, we do not so much copy other persons as copy their actions, and 'find others at the point of origin of these actions' (1964: 117). This process of copying, as I have already shown, is one not of information transmission but of guided rediscovery. As such, it involves a mixture of imitation and improvisation: indeed these might better be understood as two sides of the same coin. Copying is imitative, insofar as it takes place under guidance; it is improvisatory, insofar as the knowledge it generates is knowledge that novices discover for themselves. Thus conceived, improvisation - in Bourdieu's terms - is 'as remote from a creation of unpredictable novelty' as is imitation 'a simple mechanical reproduction of the initial conditionings' (Bourdieu 1977: 95). Both are aspects of the situated and attentive engagement that is fundamental to becoming a skilled practitioner (Ingold 1996b: 179).
The process of learning by guided rediscovery is most aptly conveyed by the notion of showing. To show something to someone is to cause it to be made present for that person, so that he or she can apprehend it directly, whether by looking, listening or feeling. Here, the role of the tutor is to set up situations in which the novice is afforded the possibility of such unmediated experience. Placed within a situation of this kind, the novice is instructed to attend particularly to this or that aspect of what can be seen, touched or heard, so as to get the 'feel' of it for him- or herself. Learning, in this sense, is tantamount to an 'education of attention'. I take this phrase from James Gibson (1979: 254), whose attempt to develop an ecological psychology - that treats perception as the activity of the whole organism in an environment rather than that of a mind inside a body - has been a major source of inspiration for the approach I have adopted here. Gibson's point was that we learn to perceive not by taking on board mental representations or schemata for organising the raw data of bodily sensation, but by a fine-tuning or sensitisation of the entire perceptual system, comprising the brain and peripheral receptor organs along with their neural and muscular linkages, to particular features of the environment (Gibson 1979: 246-8).
This education of attention is quite different from what Sperber has in mind, when he suggests that the child is innately disposed to attend to information that specifies particular competencies such as speaking, climbing, throwing and so on (1996: 117). The difference is twofold. First, the structures of attention to which Sperber refers, namely 'dispositions', are given from the start, and do not themselves undergo development within an environmental context. Secondly, their 'education', if we can call it that, lies not in their sensitisation to certain features of the world, but in their receipt of specific informational content. To put it rather crudely, Gibson's novices are 'tuned up', Sperber's are 'filled up'. Thus from a Gibsonian perspective, if the knowledge of the expert is superior to that of the novice, it is not because he has acquired mental representations that enable him to construct a more elaborate picture of the world from the same corpus of data, but because his perceptual system is attuned to 'picking up' critical features of the environment that the novice simply fails to notice. The accomplished woodsman - to revert to our earlier example - looks around him for guidance on where and how to cut: he consults the world, not a picture in his head. The world, after all, is its own best model (A. Clark 1997: 29). Adopting one of Gibson's key metaphors, we could say that the perceptual system of the skilled practitioner resonates with the properties of the environment. Learning, the education of attention, is thus equivalent to this process of attunement of the perceptual system (Zukow-Goldring 1997).
We are now, at last, able to provide an answer to the question with which I began. In the passage of human generations, each one contributes to the knowledgeability of the next not by handing down a corpus of disembodied, context-free information, but by setting up, through their activities, the environmental contexts within which successors develop their own embodied skills of perception and action. Rather than having its evolved capacities filled up with structures that represent aspects of the world, the human being emerges as a centre of awareness and agency whose processes resonate with those of its environment. Knowing, then, does not lie in the relations between structures in the world and structures in the mind, mediated by the person of the knower, but is immanent in the life and consciousness of the knower as it unfolds within the field of practice - the taskscape - set up through his or her presence as a being-in-the-world. Cognition, in this sense, is a process in real time. 'Rather than speaking of ideas, concepts, categories and links', Gatewood suggests, 'we should think of flows, contours, intensities and resonances' (1985: 216). These are terms that describe, on the one hand, the features of the taskscape in which practitioners are situated and, on the other hand, the course of their own attention as they make their way through it. But since the taskscape through which any person moves is constituted by the practices of all the others, each plays a part in establishing the conditions of development for everyone else. Thus the process of cognition is tantamount to the historical process of social life itself. And the latter, as I have already shown, is but a continuation, into the human domain, of a more encompassing process of evolution.
By way of an epilogue, I should like finally to dispose of three fallacies that lie at the heart of Sperber's approach to the explanation of culture. The first is that cultural knowledge takes the form of representations; the second is that these representations, in their mental modality, are stored within the containers of a universal psychology, whence they have to be retrieved prior to their practical enactment; the third is that in this enactment, a boundary is crossed between mental and public domains.
Sperber's 'representations' are peculiar things. It is not at all clear what they represent, if not themselves. True, the recipe for Mornay sauce is, in some sense, 'about' the activity that goes on in the kitchen. When it comes to the story of Little Red Riding Hood the situation is not so clear, and Sperber's claim that a myth such as this is 'an orally transmitted story which is taken to represent actual events' (1996: 95) is decidedly unconvincing. As for the tune that I whistle as I walk down the street, the claim that it represents something other than itself (ibid.: 32) is quite absurd. Indeed, if we accept that the representation is a 'concrete physical object' (ibid.: 61), which can turn up just as well inside human brains as outside them, in the public form of artefacts, then nothing seems to have been gained by the substitution of 'representation' for the old-fashioned concept of 'trait'. At least the latter does not send us off in a fruitless search for what the object in question is supposed to stand for. Might we not do better, as a last resort, to return to the language of Kroeber, and to regard every cooking of a sauce, every telling of a story and every whistling of a tune as a particular exemplar of a trait, rather than a particular token of a representation?
The answer is 'no'. Cookings, story-tellings and whistlings are not representations, they are not traits, indeed they are not objects of any kind; they are rather enactions in the world. When you whistle an original tune, or tell a story for the first time, you do not merely convert, into manifest behaviour, a structure that exists already full-formed in your mind. Rather, the form of the melody or the story arises and is suspended within the current of the activity itself, situated as it is within an environment that includes me, the listener. And when I listen, I do not convert the pattern of acoustic stimulation back into a mental structure, but align the movement of my attention so that it resonates with that of your action. To put this another way, I am not - as Sperber would have it (1996: 32) - the 'user' of a form that you have yourself 'produced', but have rather joined with you, albeit silently, in the process of its production. The same goes for watching the activity of the cook in the kitchen: again, the movement of my attention 'follows' that of the cook as he goes about his task. To listen or to watch, in that sense, is to accompany another being, to follow - even if only for a short while - the same path as that being through the lifeworld, and to share in the experience the journey affords. Critically, in this journey, both observer and observed travel in the same direction. Listening, in short, is not the reverse of whistling or speaking, nor is watching the reverse of doing, for 'both are oriented in the same direction by the movement of consciousness' (Ingold 1986: 273).
How, then, is it possible that having heard you whistle a tune, which I might never have heard before, I am later discovered whistling the very same tune myself? Surely, it will be argued, there is no way to explain this remarkable ability of recall save by supposing that the tune exists, in some form, 'inside my head' - that is, as a mental representation. This, of course, is Sperber's view. It implies that the mind itself, with its several domain-specific compartments, is a repository for an immense accumulation of such representations, and that the act of recall is a matter of accessing or 'calling up' a particular representation from its place of storage, so that it can then be turned into overt behaviour. To adopt this view, however, is to force a distinction between recall and performance. The former is conceived as a purely intellectual operation, carried on exclusively within the space of the mind, whereas the latter is conceived as a purely physical or behavioural operation, subsequently undertaken by the body. This essentially Cartesian distinction is redolent of one to which I have already referred, and which is central to classical cognitive science's account of intelligent action, between problem-solving and the bodily implementation of the solutions reached.
Criticising this account, I argued that problem-solving is inseparable from the actual movements of the person-acting in the settings of practice, and therefore that to have solved a problem is ipso facto to have implemented the solution. On the same grounds, I would now argue that recall and performance are one and the same: that whistling a tune is remembering it. This - harking back to Rubin's (1988) point - is to understand remembering not as the accessing of a complex structure, but as the unfolding of a complex process. To whistle a tune or tell a story that you have heard before is like taking a walk in the country along a route that you previously travelled in the company of someone else. You remember as you go along, where to 'go along' means to find your own way through the terrain of your experience. Thus the tune or story is a journey made rather than an object found, and to have remembered the way is already to have reached one's destination. Yet if each whistling, or telling, is a separate journey, how can we say of any two journeys that they are actually the same? For Sperber, every performance is the token of a representation, so that to whistle the same tune on different occasions is simply to run off two replicas from the same template. The similarity is given from the start. Journeys, however, can only be compared retrospectively in terms of the ground covered. Just as to follow someone is to cover the same ground through the world of lived experience, so to remember is to retrace one's steps. But each retracing is an original movement, not a replica.
Finally, if the forms things take, whether in the imagination or on the ground, arise within the current of involved activity in a field of practice (Ingold 1995b: 76), then there is no longer any need to suppose that to be perceived by others, a form has to have crossed a threshold from the interiority of my brain to the world outside or, conversely, that to be known by me, it has to have crossed the threshold in the reverse direction, from the surrounding environmnent to my brain. It would be wrong, as I have argued elsewhere (Ingold 1992: 51), to think of the interface between brain and environment as one of contact between two mutually exclusive domains, respectively mental and public; rather each - to use Bohm's (1980: 177) term - is 'implicate' in the other. Thus in the course of development, the history of a person's relations with his or her environment is enfolded in particular, neurologically grounded structures of awareness and response. Likewise, enfolded within the manifold forms and structures of the environment are histories of the activities of persons. In sum, the neurological structures and artefactual forms that Sperber calls representations are not causes and effects of one another, but emerge together as complementary moments of a single process - that is, the process of people's life in the world. It is within this process that all knowledge is constituted.
This paper originates from discussions in the series of seminars on 'Memory and Social Transmission', held at Queen's University, Belfast, through 1994 and 1995, and funded by the Economic and Social Research Council. I would like to thank the organisers of the series, Elizabeth Tonkin and Harvey Whitehouse, for making these discussions possible, and all the participants for many hours of fruitful exchange, from which I learned a great deal. In revising the paper, I have benefited greatly form the comments of two very helpful, but anonymous readers.
Bateson, G. 1973. Steps to an ecology of mind. London: Granada.
Bernstein, N. A. 1996. On dexterity and its development. In Dexterity and its development, eds. M. L. Latash and M. T. Turvey. Mahwah, NJ: Lawrence Erlbaum Associates.
Bohm, D. 1980. Wholeness and the implicate order. London: Routledge and Kegan Paul.
Bourdieu, P. 1977. Outline of a theory of practice. Cambridge: Cambridge University Press.
Clark, A. 1997. Being there: putting brain, body and the world together again. Cambridge, Mass.: MIT Press.
Clark, J. E. 1997. A dynamical systems perspective on the development of complex adaptive skill. In Evolving explanations of development: ecological approaches to organism-environment systems, eds. C. Dent-Read and P. Zukow-Goldring. Washington, DC: American Psychological Association.
Cohen, J. and I. Stewart 1994. The collapse of chaos: discovering simplicity in a complex world. Harmondsworth: Viking.
D'Andrade, R. G. 1981. The cultural part of cognition. Cognitive Science 5: 179-95.
DeCasper, A. and M. Spence 1986. Prenatal maternal speech influences newborns' perception of speech sounds. Infant Behavior and Development 9: 133-50.
Dent, C. H. 1990. An ecological approach to language development: an alternative functionalism. Developmental Psychobiology 23: 679-703.
Dunbar, R. I. M. 1996. How not to do biology. Cultural Dynamics 8: 363-368.
Dreyfus, H. L. and S. E. Dreyfus 1986. Mind over machine: the power of human intuition and expertise in the era of the computer. New York: Free Press.
Durkheim, E. 1976 . The elementary forms of the religious life, trans. J. W Swain (2nd ed.). London: Allen & Unwin.
Gatewood, J. B. 1985. Actions speak louder than words. In Directions in cognitive anthropology, ed. J. W.Dougherty. Urbana, IL: University of Illinois Press.
Gibson, J. J. The ecological approach to visual perception. Boston: Houghton Mifflin.
Goodwin, B. C. 1988. Organisms and minds: the dialectics of the animal-human interface in biology. In What is an animal?, ed. T. Ingold. London: Unwin Hyman.
Greenfield, P. M. 1991. Language, tools and the brain: the ontogeny and phylogeny of hierarchically organized sequential behaviour. Behavioral and Brain Sciences 14: 531-95.
Guillaume, P. 1971. Imitation in children, trans. E. P Halperin. Chicago: University of Chicago Press.
Hirschfeld, L. A. and S. A. Gelman 1994. Toward a topography of mind: an introduction to domain specificity. In Mapping the mind: domain specificity in cognition and culture, eds. L. A. Hirschfeld and S. A. Gelman. Cambridge: Cambridge University Press.
Hutchins, E. 1995. Cognition in the wild. Cambridge, Mass.: MIT Press.
Ingold, T. 1986. Evolution and social life. Cambridge: Cambridge University Press.
Ingold, T. 1992. Culture and the perception of the environment. In Bush, base: forest farm. Culture, environment and development, eds E. Croll and D. Parkin. London: Routledge.
Ingold, T. 1993a. Technology, language, intelligence: a reconsideration of basic concepts. In Tools, language and cognition in human evolution, eds K. R. Gibson and T. Ingold. Cambridge: Cambridge University Press.
Ingold, T. 1993b. The temporality of the landscape. World Archaeology 25: 152-174.
Ingold, T. 1994. Tool-using, toolmaking and the evolution of language. In Hominid culture in primate perspective, eds D. Quiatt and J. Itani. Niwot, CO: University Press of Colorado.
Ingold, T. 1995a. 'People like us': the concept of the anatomically modern human. Cultural Dynamics 7: 187-214.
Ingold, T. 1995b. Building, dwelling, living: how animals and people make themselves at home in the world. In Shifting contexts: transformations in anthropological knowledge, ed. M. Strathern. London: Routledge.
Ingold, T. 1996a. Why four why's? A response to my critics. Cultural Dynamics 8: 377-386.
Ingold, T. 1996b. Situating action V: the history and evolution of bodily skills. Ecological Psychology 8: 171-182.
Johnson-Laird, P. N. 1988. The computer and the mind. London: Fontana.
Kroeber, A. L. 1952. The nature of culture. Chicago: University of Chicago Press.
Kugler, P. N. and M. T. Turvey 1987. Information, natural law, and the self-assembly of rhythmic movement. Hillsdale, NJ: Erlbaum.
Laughlin, C. D. Jr., J. McManus and E. d'Aquili 1992. Brain, symbol and experience: toward a neurophenomenology of human consciousness. New York: Columbia University Press.
Lave, J. 1988. Cognition in practice: mind, mathematics and culture in everyday life. Cambridge: Cambridge University Press.
Lehrman, D. S. 1953. A critique of Konrad Lorenz's theory of instinctive behavior. Quarterly Review of Biology 28: 337-363.
Leudar, I. and A. Costall 1996. Situating action IV: planning as situated action. Ecological Psychology 8: 153-170.
Lock, A. 1980. The guided reinvention of language. London: Academic Press.
Maynard Smith, J. 1969. The status of neo-Darwinism. In Towards a theoretical biology, 2: Sketches, ed. C. H. Waddington. Edinburgh: Edinburgh University Press.
Meltzoff, A. N. 1993. The centrality of motor coordination and proprioception in social and cognitive development: from shared actions to shared minds. In The development of coordination in infancy, ed. G. J. P. Savelsbergh. North Holland: Elsevier Press.
Merleau-Ponty, M. 1964. The primacy of perception and other essays on phenomenological psychology, the philosophy of art, history and politics, ed. J. M. Edie. Evanston, IL: Northwestern University Press.
Oyama, S. 1985. The ontogeny of information: developmental systems and evolution. Cambridge: Cambridge University Press.
Oyama, S. 1989. Ontogeny and the central dogma: do we need the concept of genetic programming in order to have an evolutionary perspective? In Systems in development: the Minnesota Symposia on child psychology, Vol. 22, eds. M. Gunnar and E. Thelen. Hillsdale, NJ: Erlbaum.
Oyama, S. 1993. Constraints and development. Netherlands Journal of Zoology 43: 6-16.
Pinker, S. 1994. The language instinct. New York: Morrow.
Pye, D. 1968. The nature and art of workmanship. Cambridge: Cambridge University Press.
Rubin, D. 1988. Go for the skill. In Remembering reconsidered: ecological and traditional approaches to the study of memory, eds U. Neisser and E. Winograd. Cambridge: Cambridge University Press.
Shore, B. 1996. Culture in mind: cognition, culture, and the problem of meaning. New York: Oxford University Press.
Sperber, D. 1985. On anthropological knowledge. Cambridge: Cambridge University Press.
Sperber, D. 1996. Explaining culture: a naturalistic approach. Oxford: Blackwell.
Suchman, L. A. 1987. Plans and situated actions: the problem of human-machine communication. Cambridge: Cambridge University Press.
Thelen, E. 1995. Motor development: a new synthesis. American Psychologist 50: 79-95.
Tooby, J. and L. Cosmides 1992. The psychological foundations of culture. In The adapted mind: evolutionary psychology and the generation of culture, eds J. H. Barkow, L. Cosmides and J. Tooby. New York: Oxford University Press.
Whitehouse, H. 1996. Jungles and computers: neuronal group selection and the epidemiology of representations. The Journal of the Royal Anthropological Institute (N.S.) 2: 99-116.
Varela, F. J., E. Thompson and E. Rosch 1991. The embodied mind: cognitive science and human experience. Cambridge, Mass.: MIT Press.
Zukow-Goldring, P. 1997. A social ecological realist approach to the emergence of the lexicon: educating attention to amodal invariants in gesture and speech. In Evolving explanations of development: ecological approaches to organism-environment systems, eds. C. Dent-Read and P. Zukow-Goldring. Washington, DC: American Psychological Association.