PS:DELETE NOW IF UNINTERESTED
FROM THE SPECIFIC TO THE GENERAL
The between-cultural universal, the within-cultural universal, and the context-specific
approaches compared and contrasted in the previous sections can be usefully summarized by
reference to Figures 2 and 3. Figure 2 represents what we refer to as a central-processor
approach found in the two univer
En
Figure 2.
CULTURE AND COGNITIVE DEVELOPMENT
EVENTS PROCESSING TASK/PERFORMANCE
"IQ"
"Conceptual level "
"Cogn itive style"
salistic theories. The central-processor approach assumes that experiences operate on the
current state of some central cognitive machinery, which in turn guides performance on the
range of tasks that individuals encounter. The domain of the processor and its hypothetical
structure are different for different theories. For Piaget, the processor corresponds to a
universal set of elementary facts about Homo sapiens and their shared world. It is endowed with
hierarchically organized structural units. For Witkin and Berry, the domains correspond to
ecocultural niches, and the processor is structured in terms of hypothetical amounts of
differentiation and integration.
Despite differences in terminology, data bases, and the internal structures they posit, both
approaches assume that each learning experience (E1, E2‹En) potentially contributes to an
increase in power (level, amount) to a central processor that is then deployed to deal with
individual performance tasks (Tl, T2‹Tn).
The achievements of these central-processor approaches have been considerable, and they may
represent a useful, even correct, approach to the issues. But we think that outstanding sources
of disagreement can best be minimized, if not eliminated, by taking a different tack, an extreme
version of which is charactered schematically in Figure 3.
Like the central-processor approach, the contrasting ''distributed-processor" approach
sketched in Figure 3 links experiences (E1) to task performance (T1) through discrete
schemata. However, a "distributed-processor" theory places little emphasis on processing that
is common to all tasks. Instead, this approach treats cognitive processing as distributed. It is
distributed in two senses; individual learning is assumed to be context-dependent in the first
instance (e.g., distributed by situation), and
processing is soc ially distributed among people within contexts, in the second. This context-
specific approach is strong where the central-processor theories are weak; it accounts for
variability by specifying the diverging lines between culturally organized practice and task-
specific performance. It also offers relatively specific models of culture-specific, taskspecific
cognitive processing. But the problem with this formulation, as we have noted, is its failure to
account for the generality in human behavior. Skills and knowledge acquired in one setting often
do appear in other settings under recognizably appropriate circumstances. In order for a
distributed processing approach to work, it must provide some way to represent the fact that
the individual events forming the base of the knowledge system are related to each other. The
content and distribution of those events
that is, their organization, will be an important source of generality that we can ascribe to
cognitive processes (e.g., that part of thinking controlled by the internal representations of
external events).
In order to explain how generality could be, and is, achieved if learning and development are, in
the first instance, context dependent, a theory must an-
E1 ----- Schema ----- T1
E2 ----- Schema ----- T2
E3 ----- Schema ----- T3
E4 ----- Schema ----- T4
. .
. .
. .
En ----- Schema ----- Tn
Figure 3.
swer several important questions. First, it must provide a way to describe the basic aspects of
the events that constitute the fundamental contexts for activity in many culturally central
domains of experience, and the basic unit of analysis in such an approach. Second, we need to
understand cultural theories (belief systems) concerning how events are connected by
members of any culture. At this point, we do not aspire to a general theory about the cultural
constructions of reality. But some systematic ideas concerning what Lave (n.d.) has called
"extensional domains descriptive of sign)ficant areas of experience" are crucial adjuncts to
identifying fundamental events, for the structure of these domains will provide important
hypotheses about event linkages and, thus, the generality to be expected from context-specific
learning. Third, we need descriptions of the interactions among people which assemble
behavior, indeed, the events that constitute their life experiences. Fourth, we need to consider
ways that cultures might control exposure to events so that adult behaviors emerge over time in
the behavior of children. The crucial question becomes: What conditions determine whether or
not a child will encounter events of the kind necessary to produce change from one stage of
generalizations to the next (understanding stage always to mean, "stage-withincontext")?
Context Selection
According to the approach we are advocating, the kinds of contexts that children spend their
time in are the fundamental units out of which cognitive development is constructed. Therefore,
a central role of culture in producing cognitive differences will be context-selection
mechanisms that operate on children as they grow up. A recent essay by B. B. Whiting (1980)
provides one illustration of how such mechanisms might operate in ways that clearly link to a
context-specific cognitive theory.
Previous work by Whiting had treated personality as a variable intervening between culture
and individual behavior reflected in expressive and projective behavior. But, in recent years,
she has begun to look at personality embodied in the everyday behavior of people and at the way
that adults regulate access to important cultural contexts.
We are interested in the contextual variables defined by culture that are associated with types
of social behavior.... Our model is designed with the aim of facilitating cross-cultural research
which purports to explore the regularities in the contextual components of social behavior. . .
Our present theory does not deny that there may be some lasting effects of early experiences but
dictates that we look as well to other experiences in the life course to explain social behavior.
We do not deny the importance of the mother and father in molding the child but our analysis of
samples of maternal behavior across cultures convinces us that the mother andfather's greatest
effect is in the assignment of the child to settings that have important socializing influences.
(Whiting, 1980, pp. 96-97; emphasis added)
If one considers cognitive skills to be examples of social behavior (an easy allowance in that
Whiting and her students have included cognitive tests in their work), it is clear that Whiting
is suggesting precisely the kind of selection process that a contextspecific theory needs, that is,
a way to link contexts in terms of the cultural practices that sustain the group. Summarizing
the results of the massive Six Cultures studies and more recent work on sex differences,
Whiting characterizes the process linking contexts as follows:
This theory says that patterns of interpersonal behavior are developed in the settings that one
frequents and that the most important characteristics of the setting are the cast of characters
who occupy the setting.... The settings one frequents are in turn related to the activities that
occupy males and females of various ages in the normal course of living, activities that are
determined by the economic pursuits and social structure and organization variables. (Whiting,
1980, p. 103)
The parameters of human life are described by Whiting in a way that is reminiscent of the
ecocultural theory discussed above, and described at length elsewhere (Laboratory of
Comparative Human Cognition, in press). But, there is a significant difference between
Whiting's proposal and the standard psychological versions of the ecocultural approach. Instead
of the ever-present molding of behavior by the accumulated contingencies of history and
geography, we have a context-selection mechanism for developmental change.
In the normal course of living, in as much as the settings one frequents change as one grows
older and moves from childhood to adolescence to adulthood to old age, a person must be able to
learn new behaviors especially if the changes in his/her life style involve interaction with
differ
ent categories of individuals or are in settings focused around new activities, settings with
different standing rules of behavior.
Each setting is characterized by an activity in progress, a physically defined space, a
characteristic group of people and norms of behavior‹ the blue print for propriety in the
setting. Thus a child moving from the classroom to the playground interacts with adults and
peers in different manners. The standing rules for these settings do not prescribe the same type
of social interaction. (Whiting, 1980, pp. 103-104)
Many of the age changes that have been reported in the literature on child development may be
the result of frequenting new settings as well as gaining new physical and cognitive skills.
(Whiting, 1980,p. 111)
A context-selection approach of this sort is needed to begin to handle the problem of the
apparent generality of cognitive processes. Whiting's work offers a promising beginning, but a
great deal remains uncertain.
Missing almost entirely from Whiting's formulation of context selection and development is a
description of the "interpretive procedures" (Cicourel, 1973) that are necessary to account
for how people interpret rules in social situations, recognize the social circumstances they
confront, and otherwise answer the question: "When is a context?" (Erickson & Schultz,
1977). In order to handle the massive comparative enterprise represented by her theory,
Whiting and her colleagues made a series of strategic simplifications. Central was her decision
to represent each interaction involving a child by coding what instigated the child's action and
the child's response to this instigation. This coding required problematic judgments about
people's intentions. All the evidence we have about the assembly of behaviors within the crucial
contexts of enculturation indicates that they are complicated interactional events in which
stimulus (instigation) and response are very difficult to disentangle (e.g., see Cole, Hood, &
McDermott, 1978; Mehan & Griffin, 1980).
A second issue of doubtful status involves the mechanisms of transfer between one setting and
the next. Whiting is quite straightforward on this point:
Our theory also hypothesizes that the habits of interpersonal behavior that one learns and
practices in the most frequented settings may be overlearned and may generalize (transfer) to
other
settings and to other statuses of individuals. These transferred patterns may or may not be
appropriate to these new settings and can conceivably lead to maladaptive social behavior ....
(Whiting, 1980, p. 103)
The mature individual must learn setting-specific patterns of behavior but his dyadic patterns
are influenced by previous experience and habits. His/her perceptions of the responses of
people in the new setting may be blinded by expectations carried over from the old frequented
setting. (Whiting, 1980, p. 104)
What renders the status of these reasonable assertions doubtful is a great deal of evidence
indicating that transfer between settings as complex as those considered by Whiting may be
minimal or nonexistent (Shweder, 1979a,b, 1980). There is also a sizable literature in
psychology indicating that even transfer between problem isomorphs is extremely limited or
nonexistent when experimenters pose the problems, even when the problem solvers are college
students (Gick & Holyoak, 1980).
Contexts are not to be equated with the physical surroundings of settings‹classrooms,
churches, kitchens. They are constructed by the people present, in varying combinations of
participants and audience (Erickson & Schultz, 1977). As McDermott and Roth (1978) have
put it, contexts are constituted by what people are doing, as well as when and where they are
doing it. That is, people in interaction serve as environments for each other.
We characterize activities such as those from which Whiting obtains her data as cultural
practices, by which we mean activities for which the culture has normative expectations of the
form, manner, and order of conducting repeated or customary actions requiring specified skills
and knowledge (see Scribner & Cole, 1981). Cultural practices have to be learned as systems of
activity. These settings have "standing rules," what cognitive psychologists term "scripts"
(Schank & Ableson, 1977), anthropologists refer to as "contexts" (Frake, 1977), and
sociologists call "background expectancies" (Cicourel, 1973; Garfinkel, 1967) that orient
people to the behavior that is appropriate for a given situation.
A cultural-practice theory of culture and cognition resists the separation of individuals from
the environments in which they live their daily lives.
This means that the relation between culture and cognition represents neither a purely
subjective (in the head) nor purely objective (in the world) phenomenon; it is an
intersubjective phenomenon, to be found in the interaction between people. Goodenough's notion
that "culture consists of whatever one has to know or believe in order to operate in a manner
acceptable to its members" (Goodenough, 1964, p. 36) provides a good start toward an
interactional conception of culture and cognition. But this knowledge cannot be thought of in
static or purely internal terms. Rather, as Geertz (1973, p. 44) suggests, the knowledge is
akin to ''a set of control mechanisms‹plans, recipes, rules, instructions . . . for the governing
of behavior." These mechanisms, embodied in cultural practices, are largely accomplished
through the cooperations of individual members of the culture in contexts of practical activity
(Leont'ev, 1981).
People must display what they know to others and the meaningfulness of behavioral displays is
established by the interpretation of others. Production and interpretation are mutually
informing activities, conducted conjointly in interaction. Furthermore, the interpretation of a
behavioral display in the present informs the production of behavior in the future, just as the
production of present displays informs subsequent interpretations (G. H. Mead, 1934; Schutz,
1962: Voloshinov, 1973).
''Culture'' and "cognition,'' then, refer jointly to behavior assembled by people in concert with
each other. It is for this reason that a cultural practice theory takes cultural contexts, that is,
socially assembled situations, not individual persons or abstract cultural dimensions as the
unit of analysis for the study of culture/cognition.
Guided Change in Interaction
To our initial proposal that cognitive development is characterized by the mastery of context-
specific knowledge about the world, we have now added the ideas that (1) cultures arrange the
selection of contexts for children and that (2) one must study seriously the ways in which
interactions among participants construct and maintain behavior in those contexts according to
standing rules for the conduct of cultural practices. But we have not said much about how within-
context interactions result in within-context mastery of essential cultural knowledge. In order
to understand how the culture organizes for next steps of within-context development to occur,
we turn our attention to the sociohistorical school of
Soviet psychology, which explicitly connects ideas of interaction with the concept of
development.
The Relationship Between the Social and the Individual
The sociohistorical approach includes several proposals for how culturally organized social
interactional patterns can influence the psychological development of the child. These proposals
were made by Vygotsky and his followers in the process of developing a Marxist psychology
(El'konin, 1972; Leont'ev, 1978; Luria, 1976; Vygotsky, 1978; Zaporozhets, 1980). A
fundamental tenet of this approach is that human cognitive functioning emerges out of social
interaction. 6
The basic idea can be found in Vygotsky's ''general law of cultural development."
Any function in children's cultural development appears twice, or on two planes. First it
appears on the social plane and then on the psychological plane. First it appears between people
as an interpsychological category and then within the individual child as an intrapsychological
category. This is equally true with regard to voluntary attention, logical memory, the formation
of concepts and the development of volition. (Vygotsky, 1978, p. 57)
Vygotsky referred to the contexts organizing the social-to-psychological transformation of
thinking as " zones of proximal development. " Vygotsky defined this zone as the difference
between a child's "actual developmental level as determined by independent problem solving"
and the level of "potential development as determined through problem solving under adult
guidance or in collaboration with more capable peers" (Vygotsky, 1978, p. 86). He
demonstrated the usefulness of the notion of the zone of proximal development when dealing with
the issues involved in assessing mental ability.
For our present purposes, the most important application of the notion of the zone of proximal
development may be seen in Vygotsky's analysis of instruction. In this connection he argued that
children can benefit from interaction with more experienced members of their culture only if
the level of interaction falls within a certain range specified by the zone of proximal
development.
Instruction is good only when it proceeds ahead of development, when it awakens and rouses to
life those functions which are in the process of
maturing or in the zone of proximal development. It is in this way that instruction plays an
extremely important role in development. ( 1956,
p. 278)
>From the sociohistorical viewpoint, a culture maximizes its impact on a child's development by
providing regulative contexts that fall within the zone of proximal development. Of course,
there are many ways that more experienced and mature members of a culture can influence the
child's environment, but the following four seem particularly important.
First, culture arranges for the occurrence or nonoccurrence of specific basic problem-solving
environments embodied in cultural practices. Infants are taught to crawl or climb, sleep short
or sleep long. Preschoolers learn to model in wire or draw (i.e., model with pen and paper).
Students chant the Koran or read the Bible.
Second, the frequency of the basic practices is culturally organized. Does one read daily in class
or weekly in church? When is it necessary to sort grains? How many times a day does one
engage in pottery making and with how many products? Does one sell pottery as well as make
it? Culture exerts an overwhelming power in answering such questions.
Third, culture shapes the patterning of co-occurrence of events. One may use an interrelated
set of units to measure when selling and buying quantities of rice but measure cloth and tables
with unrelated quantities. One culture provides for recall of spatial arrays using verbal
rehearsal strategies and another without them. Written text is the vehicle for religious activity
in some cultures, but not in others.
Fourth, cultures regulate the level of difficulty of the task within contexts. This regulation both
in:reases the likelihood that potentially crucial learning events will occur and that costly
failure will be averted. Arranging for babies to learn to sit erect by propping them in a hole
with a blanket might be a tarter task in one culture. Sewing buttons on a shirt nay be the
starter task toward becoming a master :ailor in another. In each case, a series of difficulty
evels leading to mastery is elaborated.
In summary, the sociohistorical approach to uman cognition offers the following account of the
relationship between culture and cognition. First, it proposes that there is indeed a strong
connection between the social interactional processes that contitute activity in a culture and the
psychological processes of its members. This is so because an indiAdual's psychological
functioning is seen to emerge
through the process of internalizing various processes involved in social interaction which is
itself culturally organized. Second, the zone of proximal development provides the conceptual
Iynchpin in the process by which members of a culture, children and adults, produce the
relationship between social and individual functioning. It is here that the social becomes
individual and the individual becomes social.
An Example of Zone-of-Proximal
Development Analysis
An example from research on American children illustrates the way in which adults organize
the learning environments of children, thereby creating effective zones of proximal
development. Wertsch and his associates (Wertsch, 1978, 1979, in press-a, in press-lo;
Wertsch, McNamee, Budwig, & McLane, in press) have conducted a series of studies on how
mothers help their young children carry out tasks such as assembling a simple puzzle in
accordance with a model. The puzzle pieces were cut-out parts of a truck which 5-year-old
children were supposed to insert into a frame so that the end product would be identical with the
model. Each child was helped by his or her mother through two assemblies of the same puzzle.
The course of the interaction as "the child assembles the puzzle while the mother assists "
usually went something like this. During the first try at making the puzzle, the child might
insert the pieces in the puzzle, but there is no attempt to make the puzzle in accordance with
the model. If the model is used at all, it is because the mother negotiates the dyed's activity so
she can include the model in the overall strategy. For example, she may make decisions about
which pieces are to be used by looking at the model herself, or she may instruct the child to look
at the model even though he or she does not understand what role it plays in the task.
During this early stage the child understands only very simple directives by the mother. These
directives (which may involve both verbal and nonverbal communicative behaviors) are
simple in the sense that they involve minimal understanding of what the mother and
experimenter see as the overall task. For example, a regulative utterance such as, "Put the red
one here" (with the mother pointing to the correct location) can lead to the appropriate task
behavior on the part of the child without his or her understanding the relationship between "the
red one," the model, and the puzzle. As the task session progresses, the mother is often able to
utilize more complex speech to elicit appropriate task behaviors from
the child. For example, she may use a regulative utterance such as, "Where does the red one
go?" This utterance is more complex in the sense that, unlike the earlier one, it requires the
child to identify and execute several substeps of the overall task in order to respond
appropriately. Thus in order to identify where the red one goes the child must be capable of
regulating her or his own activity to check the model.
Of course, this does not mean that whenever a mother uses a complex regulative utterance the
child will respond appropriately. In those cases where an inappropriate response occurs,
however, the mother is likely to follow up the child's response by switching to a simple
directive that makes explicit the substeps implicit in the complex directive. For example, in
the case of the complex directive we mentioned above, the mother might say, "Look over here
[pointing to the model puzzle] and you can see where the red one goes." In this way, the child is
given a demonstration of how to " unpack" complex regulative utterances in the task situation
and of how to expand the context of the task and thus to move closer to the overall task definition
shared by the mother and experimenter.
The range of semiotic options available to the mother makes it possible for her to provide
assistance to the child at various levels. This is a crucial aspect of development since the zone of
proximal development can be expected to change as the child's experience with the problem
increases. Initially, orientation to the task, selection, comparison, and even motor components
of the problem may in large measure be carried out by the mother, who elicits required motor
and verbal compliance from the child. The mother is doing more than the child can do, but not so
much more that the child cannot participate. As the child comes to take over more of the task,
the mother shifts the nature of the work she does (e.g., offering praise or pointing to troubles a
few steps ahead).
In a manner reciprocal to adult options, the child can participate in the task at several different
levels. These include bare participation where an adult guides the child through the appropriate
steps in the task when the child understands little more than that there is a task, or maybe even
only that there are a series of small tasks that the child does not yet see as components of an
overall task. By participating in interaction understood by the adult (the interpsychological
functioning characteristic of the early stages of learning a task), a child can "accomplish" the
task before the child understands what he or she is doing. Rather than understanding the task
first and then carrying it out, in this sequence of events the child carries out the task (on the
interpsychological "plane'') and then understands it. The child's understanding of the task and of
the associated complex regulative speech of the adult is a consequence, rather than a
prerequisite, of going through the task. (It is what Cazden, 1981, has aptly called
"performance before competence.")
The interactive nature of the learning process is highlighted by two facts, so mundane that they
invite inattention. First, one never sees the mother sitting next to the child, blithely putting
the puzzle together. The child is always a participant, and that participation is made possible by
the adult. The nature of the participation is interactively negotiated by child and adult. Second,
the puzzle always gets put together. This puzzle problem is well within the independent
problem-solving capacity of one of the participants, so, of course, it gets done. Putting these
two facts together, we can see the basis for a claim that development always occurs in a zone of
proximal development. Additionally, that zone is dynamically achieved by the child and others in
a social environment. Initially, it may seem that for any problem the social environment may
be doing more than its share of the work, but the achievements of the child/mother in
interaction with each other (with the mother carrying a heavy cognitive load) are
progressively transformed into achievements of the child, with the mother serving as a distant
prop.
The central insight embodied in these ideas about the immediate contexts where development
occurs is that crucial events causing change from one level to another heavily involve other
people; and in the case of young children, it is generally older people who provide an
environment that makes likely the necessary learning. In this sense, important aspects of
cognitive development "come from the outside" in the form of socially organized information
about the goals and constraints regulating behavior. In some cases, the child will quite literally
be told the necessary information ("You better check the model before you choose a puzzle
piece, silly!"); in others, adults only make the important factors salient ( "Which piece will fit
into the corner?");-and in still others, they may do no more than make it possible for the child
to be present while potential learning events are in progress ("Come sit on my lap while I help
your sister figure out this puzzle. " ). Directly or indirectly, the social environment is likely
to be providing important information to sustain and increase the efficiency of thinking. This is
not to imply that children have no role in structuring the environ
meets that in turn structure their behavior. As our example from Wertsch's work indicates,
the child's behavior provides the adult with crucial information as well as the other way
around.
The Zone of Proximal Development in
Soviet Cross-cultural Studies
Sociohistorical psychologists never exploited the potential of the zone of proximal development
as the locus of development in cross-cultural work. Only one cross-cultural expedition was
undertaken at the time these ideas were being developed (described by Luria, 1976), and few
have been taken since that time (Tulviste, 1979). Almost all of the scanty Soviet work sought
to demonstrate qualitative shifts in the basic activities that underpin cognitive systems for
preliterate and industrialized people. No research went mto an exploration of how these
different kinds of activity systems came into being and how they operate to reproduce
themselves across generations.
In view of the fact that the sociocultural theory posits the zone of proximal development as the
focus of learning and use of higher psychological functions, it may appear surprising that no
work was put mto comparative studies of concept acquisition. One can point out the limited
attention given to crosscultural work as an explanation for this neglect. But the failure runs
deeper. A commitment to the sociohistorical approach applied cross-culturally is a commitment
to looking at how cultures organize learning environments for their members, especially their
young. Following his theory, Luria went to Uzbekistan to discover the cognitive consequences of
the dramatic shift from traditional pastoralism to literate, technological activities. He
conducted interviews and experiments to tap these consequences. But it is in interaction
between Uzbeks not in interactions between Uzbeks and Russians that the theory predicted the
operation and acquisition of Uzbek concepts and problem-solving modes. Nowhere in the world
has the sociohistorical research program been carried out. However, a great deal of work has
been done that illustrates hypothetical pieces of the overall process.
The Zone of Proximal Development in
Anthropological Research
In a wide variety of studies, cultural anthropologists have described the patterns of family
interactions called socialization or education "in the broad sense" (M. Mead, 1958; Raum,
1940; for a review, see Mead, 1958). Overwhelmingly, in pretechnological societies, whether
of hunter-gatherers (Lee & DeVore, 1976) or agriculturalists or pas
toralists (Whiting & Whiting, 1975), children are described as participants in a wide variety
of social activities that we consider adult. Their role as participants varies as they grow older,
but not the fact of their participation. The more detailed ethnographies of the socialization
process show that children are routinely assigned tasks commensurate with their current
abilities as elements in a larger task guided by their older siblings or adults. Just as
developmental psychologists can point to stages of understandmg corresponding to logically
connected aspects of the environment, anthropologists have pointed out that the sequences of
child acquisitions in naturally organized learning environments have a strong element of
necessity imposed by environmental constraints. The idea that one must be able to walk before
it is possible to run exemplifies this central fact about psychological development's dependence
upon the constraints imposed by biological structure and environmental contingencies.
An example of a specialized skill engaged in by adults that is learned in specific contexts
regulated by older children and adults is given in the work of Kulah (1973). Kulah studied the
use of proverbs in the formal and informal rhetorical discussions of Kpelle (Liberian) elders.
He was interested in the way that young Kpelle children come to learn the meaning of the
proverbs. His investigation showed that in a very important sense, proverb content and
interpretation are not taught; they are "arranged for" through the organization of linked
activities. The arranging starts long before any child is expected to know or use proverbs.
All Kpelle children engage in a variety of verbal games including riddling and storytelling. One
genre of this game requires teams of children to pose riddles to each other. The riddles consist
of two parts roughly akin to a ''question" and an "answer." Both questions and answers are part
of the traditional lore of the group. They must be learned as pairs. The children line up in two
rows and sequentially challenge each other with riddles. The team that answers the most riddles
correctly is the winner.
The teams of children are age graded. Children of a wide span of ages (say, from 5 to 15 years)
may play; the oldest on each team takes the first turn, then the next oldest, down to the
youngest. In this way, even the youngest member of a team is important, and even the youngest
is around to learn many new riddles.
This activity is related to adult proverb use in the following way. The question or answer half of
the riddles that the children learn are key phrases that will appear in adult proverbs. It is as if
the riddle
learning serves to teach children the ''alphabet" along the way to learning to "read words." For
example, a "question" might be something like "rolling stone" and the answer, "no moss."
Kulah's research shows that the potential meaning in combining ''rolling stone" and ''no moss''
is not well understood by young children, even if they know a lot of riddle question answer
pairs. In a task designed to see if the children would group different riddles by the common
meaning that the adult interpretation specifies, young children did not respond as if one riddle
was related in any way to the other. But as the children grew older, they came more and more to
approximate adult groupings of riddles according to their "message." By the time they are old
enough to participate in the adult discussions where these proverbs are a rhetorical resource,
they show the adult pattern of proverb interpretation. They are ready to learn how to use their
now-organized alphabet in a new context, as a component in new, adult, tasks. The adult
contexts, in turn, reorganize the "old skill" into a new activity.
While traditional societies such as that described by Kulah provide examples of age-graded
activities such as the riddle game, the major educational contexts are unlikely to separate
children and adults.
Fortes emphasizes the unity of the social sphere of adults and children among many traditional
African peoples.
As between adults and children . . . the social sphere is differentiated only in terms of relative
capacity. All participate in the same culture, the same round of life, but in varying degrees,
corresponding to the stage of physical and mental development. (Fortes, 1970, p. 18)
This observation is especially well borne out in recent studies of adults and children engaged in
common activities that serve both as contexts for important economic pursuits and for
socialization of the young into necessary adult cultural practices.
Lave (n.d.) has recently completed a detailed study of the process of becoming a tailor among
Liberian tribal people living in the capital city of Monrovia. A typical tailor shop is peopled by
men and boys ranging in age from 6 to 60 years. The range in tailoring expertise is as great as
the range in age. Rank beginners and masters work side by side in generally crowded quarters.
In this setting, immediate economic necessity and longer term economic security combine to
organize the learning activity of apprentices. From their first day in the shop, apprentices have
to make themselves useful and masters
must begin the task of making them independently productive.
Because the masters cannot sacrifice productivity, a well-worked-out series of steps in the
mature practice of tailoring (in this case, making trousers and suits) has evolved, beginning
with elementary tasks such as sewing button holes and progressing through a series of tasks
(cutting, measuring, sewing zippers) of increasing difficulty. From time to time, Lave
observed direct instruction (e.g., when a master ripped out a badly sewn seam to show how it
should be done). At other times, masters would arrange special lessons using scraps of cloth or
paper to permit practice in cutting or sewing on the machine. However, for the most part
instruction was arranged for by including the apprentice in elementary steps as a part of the
adult activity. In these circumstances, apprentices participate in production of the adult
product at the highest level possible because this level is maximally profitable for all
concerned.
A crucial feature of such arrangement of this kind of learning environment that fits precisely
the idea of a zone of proximal development is that all stages of the mature practice are a salient
part of the learner's environment, regardless of what "level'' in the process he is working at.
Thus, the learner is gaining direct practice at one level while observing the skills necessary
for later levels. As Lave and many others note, this kind of arrangement ensures progress with
relatively few errors while providing constant motivation toward mastery.
Another description of a culture practice organized as a zone of proximal development for
novices is provided by the Childs and Greenfield (1982) study of Zinacantieco weaving,
referred to earlier. As a part of their research, Childs and Greenfield conducted a careful
analysis of the role of social guidance in the mastery of weaving. The guidance was of two kinds.
First, there was direct intervention by an adult at points where the learner was making, or was
about to make, a mistake. At the beginning of any step in the weaving process (which Childs and
Greenfield separate into six major steps), the adult was found to intervene heavily; toward the
end of each step, there was little or no guidance. On the child's first garment, the adult spent
93% of the time weaving with the child. If a girl had completed one garment, adult participation
was reduced to about 50% of the time. After as many as four successful garments, the adult was
still involved directly in weaving about 40% of the time.
This guided instruction is by no means a silent process. Childs and Greenfield show that adult
talk
is closely tied to the level of skill manifested by the learner and the specific circumstances that
the child is facing. Commands dominate early in learning, and these are overwhelmingly of the
sort " Do x. " In later stages of learning, when the novice weaver's actions are less problematic,
the adult talk shifts to statements that point out salient features of the present stage of the work
or links between stages.
A second way in which adult Zinacanteco weavers guide practice is to provide children direct
exposure to all of the steps of weaving and associated activities as a part of the process of
learning to carry out eac h of the steps. We mentioned above that Childs and Greenfield assign
six steps to the weaving process. From an early age, long before we might notice that they are
learning to weave, girls witness the whole process with all of its six parts. To borrow another
phrase from Fortes, ''the child is from the beginning oriented towards the same reality as its
parents and has the same physical and social material upon which to direct its cognitive and
instructional endowment" ( 1970, p. 19). When it comes time to learn to weave the first
garment, the process of applying what one has learned in the past to the circumstances at hand
hardly arises.
A number of researchers in anthropology and psychology have focused on the mother/child
interaction as the focus of developmentally sign)ficant learning experiences. In all cultures,
the early life of the child presents child and family alike with a set of problems that are
common to all of our species. But depending upon the complexity of environmental factors, the
systems of socialization activities that deal with those common problems will be different. They
must differ because passing on the culture to children is only a part of what must be done to
maintain the species on a day-to-day and moment-to-moment basis. The organization of these
other activities must modulate the organization of the child's immediate environment.
Kirk's work relating mother/child interaction to cognitive performance among Ga children
illustrates cultural differences in the ways that mothers guide children's problem-solving
efforts as a key mediator of improved performance. Kirk (1977) found that certain maternal
behaviors correlated with greater skill on conservation tasks, while differences in subcultural
groupings (rural, urban, suburban) did not so correlate. Her research shows that mothers who
most frequently used specific referents to indicate relationships, and who just)fied and
explained events, had children who performed on higher levels on conservation tasks. This
parallels an earlier finding (Kirk & Burton, 1977) from Kenya, where the
nonverbal communicative specificity of mothers in teaching interactions was closely associated
with the cognitive performance of children. In these studies, as well as those by Rogoff (
1978), the terms used to describe the more effective maternal activities are the same terms
we would use to describe environments that constitute effective zones of proximal development.
Connections Between Contexts
Thus far we have considered several important issues that have to be resolved in order to build
a context-specific approach to culture and cognition into a theory that can encompass the major
known phenomena: the ideas that ecocultural constraints operate through context selection, that
contexts represent systems of activity, and that cognitive change is often interactionally
managed within the significant contexts of socialization.
Still missing is an explicit statement of how a context-specific theory accounts for the ways in
which past experience carries over from one context to the next. In both the Piagetian and
psychological differentiation approaches, the mechanism invoked to provide for intersituational
consistency is called transfer. But, as we will see, traditional invocations of this term will not
solve the problem.
Current Psychological Evidence
A standard procedure for assessing transfer between different contexts in which the same
behavioral principles are believed relevant is to train subjects to solve problems in one form
and then to test the influence of this learning in the new, isomorphic problem context. For
example, Reed, Ernst, and Banerji (1974) gave college students two isomorphic problems to
solve. The first was the classical missionaries and cannibals problem, the second, a logically
identical problem involving husbands and wives. The similarity of the task structures was not
sufficient to induce transfer between the two problems. Tran.sfer occurred only when the
subjects were explicitly told about the relationship between the two problems. A similar
finding is reported by Wason and Johnson-Laird ( 1972) concerning a logical problem
presented in the form of abstract symbols (vowels and numbers) or in everyday language
concerning postmasters, mail, and stamps.
A more recent study by Gick and Holyoak ( 1980) demonstrates just how difficult it may be to
obtain transfer among problems that are extremely similar from the experimenters' point of
view, and ex
tremely similar for the subjects too, once they are told about the similarity. What makes the
failure of transfer seem odd is that Gick and Holyoak went to a lot of trouble, short of giving
verbal hints, to make certain that the relevant analogous information was known to the
subjects. But availability of information was not aufficient to induce transfer, because subjects
failed "spontaneously" to apply the known, relevant solution. Gick and Holyoak's summary of the
obstacles to transfer in their study pinpoints precisely limitations on the amount of
"spontaneous" transfer to be expected among the sign)ficant contexts of children's lives:
A potential analogy may often be encoded in a very different context from that in which the
[current] problem appears. Indeed, the basic problem in using an analogy between remote
domains is to connect two bodies of information from disparate semantic contexts. More
generally, successful transfer of learning generally involves overcoming contextual barriers.
This may not be easy .... (Gick & Holyoak, 1980, p. 349.)
Thus, it appears from the recent psychological literature that transfer (as spontaneous
application of analogies among remote contexts) is a weak theoretical reed to use as a central
mechanism in any theory of culture and cognitive development.
The perplexity of this work on transfer, when combined with evidence that learning is based on
within-context skill mastery, is that all aspects of the emerging theory point to the isolation of
cognitive achievements. But our experience of the world does not appear as a mosaic of
unconnected fragments. One possible way out of this conundrum is to conclude that the
appearance of order is itself an illusion. This position, an extreme version of the "constitutive"
perspective described above, is suggested by Shweder's wry comment that:
The everyday mind accomplishes a very difficult task. It looks out at a behavioral world of
complex, context-dependent interaction effects and unsubstantial intercorrelations among
events, yet it perceives continuities, neat clusters, and simple regularities. (Shweder, 1980,
p. 77)
Although it is certainly true that Homo sapiens is engaging in creating order out of disorder, and
while it may be true that Homo sapiens is more a rationalizing than a rational species, no one,
including Shweder, is denying that past experiences operate in the present to influence
behavior. The problem is
that detecting an analogy between disparate contexts seems to be a relatively rare individual
achievement. Valuable as they are, current psychological approaches to the problem of transfer
are not likely to provide us the evidence we need to build a theory of culture and cognitive
development.
Yet, it seems clear that people do use past experience to conduct present behavior; and in this
sense, all behavior reflects the transfer of past learning. This is confirmed in the psychological
literature on reasoning, particularly that which deals with the reasoning processes that occur
in mundane settings. Several contemporary theories of thinking are compatible with the notion
that a great deal of knowledge is context specific. According to such views, thinking consists
largely of the retrieval of context-specific information that is appropriate for the task at hand.
An early theory of this type was proposed by Bartlett ( 1958) who claimed that in everyday
thinking (of the sort that one can sample on afternoon talk shows, dinnertime conversations,
and faculty meetings) conclusions are reached (problems are solved) with little consideration
of logical alternatives because ''in popular thinking the end of the preferred argument sequence
itself takes charge of the selection of particular items of evidence" (p. 175).
Furthermore, Bartlett tells us that in everyday life the generalities and conclusions that are
put forth and the evidence that is selected are strongly socially determined. The generalities and
conclusions are usually a part of common wisdom and the selected evidence is more than
personal recall; it is "social knowledge, socially distributed'' (Schulz, 1962), evidence known
as well to others in the group who would be likely to use the same evidence in the same
circumstances.
Our problem of relating past context-specific [earnings to new or future contexts can be
formulated, in the light of Bartlett's work, as a problem of socially determined retrieval in the
new or future contexts of socially determinded structures and processes. Clearly, there is a
possible next step here. One particular process that may be socially determined is the process
by which analogies are retrieved. In fact, we suspect that analogy-retrieving processes may
well be a form of a culturally elaborated tool for generalizing or transfer (Scribner & Cole,
1973). But, such analogy-retrieving processes arise in domains in cultures, and an
understanding of their use would still require a framework like the culture-practice approach
described below.
A Culture-Practice Approach to Transfer
Instead of searching for a central, general mechanism that exists in the head of individual
persons as a
way to account for transfer, a cultural-practice orientation urges us to look to the organization
of the environments in which interactions occur. Emphasizing the importance of social
organizations does not dismiss the research on transfer-as-individualactivity. Rather, it
emphasizes that in several important ways, transfer is arranged by the social and cultural
environment. This shift of focus does not so much solve the transfer problem as it dissolves it.
As the work of Lave, of Childs and Greenfield, and of other anthropologists (e.g., Lee & Devore,
1976) strongly suggests, contexts are the "threads" from which are woven the fabric of a
society's total adaptation to its circumstances. LeVine (1970) and the Whitings' work (Whiting
& Whiting, 1975) contain the same idea cast in a different mold, as does the entire ecocultural
movement. If the implications that we are drawing from the research reviewed here are
correct, these approaches do not have to depend upon the notion of generalized transfer to
accomplish their goals. Overlap in environments and the societal resources for pointing out
areas of overlap are major ways in which past experience carries over from one context to
another.
Lave emphasizes another point concerning sources of intercontext transfer which applies well
to the Childs and Greenfield weaving case and, we believe, to a great deal of our everyday,
culturally organized experience. In speaking about the arithmetic problems that tailors
encounter, she says:
Most of the arithmetic problems encountered in everyday life have been seen many times
before. They are routine occurrences. This follows from the general routineness of our
everyday lives. The tailors come to work six days a week, make trousers, shirts and hats,
alongside the same people they have been sewing next to for months or years, for customers
many of whom they have known for years. (Lave, 1979, p. 4)
Repetition and redundancy minimize the problem of transfer posed by new and unusual
problems such as those that constitute the backbone of psychological research on problem
solving.
The single most pervasive resource, and the one that is easily overlooked in a discussion
concentrated on learning and problem solving, is language itself. We have already had occasion
to note the routine and repeated nature of a great deal of our experience. This routineness and
repeatedness is coded in the exicon, reinforcing whatever analogy-supporting data there may be
in the physical characteristics of outine events. Thus, two instances of ''weaving a hammock,"
described with those words, may be
responded to as the same in part because of the eonventional and known meanings of the words
"weaving" and "hamniock," whereas alternative descriptions might not evoke transfer of
knowledge. Understanding of this point is a major motivating force behind Whorf's (1956)
insistence on the importance of language as a molder of thought; language represents a distilled
cultural theory of what goes with what in the world. Children master their culture's theory of
the connections between contexts as they master their language. This fact is the basis for
Stefflre's (1965) assertion that "an individual will behave toward an object or event in a
manner that is similar to the way he behaves toward objects and events that he encodes in the
same way" (p. 12).
The importance of language as a code for the sediment of past wisdom concerning the relatedness
of elements of experience can be illustrated by reference to the Gick and Holyoak (1980) work
on transfer between logically identical problems. Their college students could not transfer
problem-solving solutions from one example of a problem to another because they did not apply
the appropriate analogy. One problem was embedded in a story of a brain operation; the other
involved a military dictator. As separate as these problems sound, audiences hearing a
description of this work at a professional conference have no difficulty understanding and
instantaneously applying the appropriate analogy. The difference between the lecture and
experiment arises because, in describing the two problems, speakers or authors refer to the
two problems as "radiation problems" as a part of their description. When language encodes the
relevant relation between distinct contexts, the contexts are no longer distinct; no transfer as
an individual invention is required. We suggest that this phenomenon is extremely widespread
and accounts for a great deal of the way that cultures render past problem-solving solutions
available for analysis of present problem contexts.
Language, which codes the culture's theory of what goes with what, is a universal resource
organizing transfer. In addition, there are culturally elaborated tools for organizing and
manipulating information that accomplish generality of cognitive skills. Perhaps the single
most important cultural tool for associating contexts among which transfer might occur is
literacy. Its history can be seen as a case of the movement from the context specific to the
general. If Schmandt-Besserat's (1978) account of the earliest precursors of writing is
correct, the earliest writing forms represented no more than tallies for a very circumscribed
set of objects in the earliest agricultural settlements. They were devices for recording the
number of animals or amount of grain in tiny,
protoagricultural settlements. While their contexts of use must be considered quite
circumscribed, these tallying devices became standardized for linking information in one
immediate context (the amount of grain in a wagon) and the information in another context (the
total amount of grain in a shed, or the amount loaned to a neighbor). For something on the order
of 6, 000 years, these devices remained context specific; only increasing in kind very slightly.
But with the advent of bronzeworking technology, improved agicultural techniques, and the rise
of trade, a more powerful system of recordkeeping was required. There were many more
contexts in which such tallies were needed. As the connections between people became looser
because of increased community size, intercommunity trade, and division of labor, people came
to inhabit very different contexts from each other, even within the same culture. The tokens
proliferated, modes of representing them changed, and eventually an alphabetic system was
formed, the system on which contemporary literate practices in American schools are based.
Even from this brief account, it should be clear that writing is not an all-or-none invention.
We cannot simply say, "The Greeks invented the alphabet." As the record clearly shows (cf.
Gelb, 1963), the alphabet was the end product of many centuries of context-specific
adaptations to changing circumstances. Both the way in which literacy is a tool designed and
implemented in specific contexts and the linkages it provides between situation have been
illustrated in the research by Scribner, Cole, and their colleagues described earlier (Scribner
& Cole, 1978, 1981).
Alphabetic literacy is a powerful tool for storing and transferring information across time and
distance. It is a transfer-producing tool. But it is not a context-independent tool. Rather, it too
is tied very closely to the contexts of activity that constitute adult practice (Cole & Griffin,
1980).
Summary
Throughout this section, we have taken pains to point out that the way in which past experiences
carry over from one situation to another is conceived of differently in cultural-universalistic
and in context-specific theories. The cultural-practice theory deemphasizes transfer as a
central process occurring within the minds of individuals and emphasizes movement of
information across contexts as a soc ial accomplishment. The tuition of young children by
adults, their direct intervention, especially when a mistake is about to be committed, and
adults' practice of embedding learning in everyday experiences,
are some of the ways in which environments are arranged for events to reoccur. In fact, the
massive redundancy and repetitiveness of learning situations minimizes the occurrence of new
imizes the occurrence of new situations. In those unusual circumstances when people are confronting new situations, the
physical features of those environments, the social distribution of social knowledge, and the
presence of a number of cultural resources, notably language and literacy, assist in providing
bridges between contexts.