So my challenge as a reviewer has shifted. I still want to avoid "theoretical sectarianism" about cognitive development, but I especially want to avoid evangelizing too hard for the dominant constructivist "churches"-to avoid accepting SBTLs too easily, without critical appraisal. Accomplishing this new task can be challenging, given the support for SBTLs circulating in the current post-modernist Zeitgeist. Nonetheless, my hope in this chapter is to encourage an appraisal of SBTLs both directly and indirectly by describing aspects of SBTLs and alternatives theories, and the commenting on what each offers early childhood education and on how much each is partaking of the "new" constructivisms. The appraisals will not aim for full coverage, but emphasize the practical needs of early childhood educators, who I will define as any professionals who work with and support children day-to-day, regardless of institutional affiliation.

I will describe four major theories or models related to cognitive development in early childhood. The first will be learning theory, but learning theory primarily in its recent, relatively cognitive versions. The second will be information processing theory, with special emphasis on the newer ideas about pattern detection and parallel, rather than serial, processing. The third will be structural theory, with emphasis on what might be termed the "organizing inner mind." The fourth will be SBTLs proper, in both individualistic and social forms. This sequence will work, but not perfectly, because versions of SBTL have influenced all schools of developmental thought and early childhood education practice. The problem, it will turn out, is that distinctions among theories and concepts are not fully stable or clear. Blurring conceptual and pedagogical boundaries has mostly been healthy, but the reasons for blurrings need to be understood carefully, in order to make best use of the revisions in daily work with children.

To help fix the conceptual distinctions and shifts, I begin by discussing my understanding of what psychologists and educators tend to mean by the general term cognitive development, and its relevance (or lack thereof) for early childhood education. As we will see, the notion ofcognitive development is problematic for early childhood educators because it implies psychological distinctions not held by them in practice, or even in published reflections on their work. But the mismatch between "classic" cognitive development and the needs of professionals has apparently been noticed by developmental psychologists, some of whom have revised their research projects to bring them more in line with (and also help to create) newer meanings ofcognitive development. For better or for worse, the new meanings are all more socially based or "social constructivist" than before; hence my cautions in the opening paragraph of this chapter.

Three Problems with Cognitive Development

Structural Models of Cognitive Development

Structural models of cognitive development are intellectual descendants of rationalist philosophers such as Descartes and Kant, who argued that a person's mind (or "soul") imposes order on knowledge and experience, rather than simply absorbing or detecting it in the environment. When this view has been adopted by developmental psychologists, a top priority has been to discover exactly how a child imposes order on experience-the intellectual stages, mechanisms, or other order-making processes that are not a directly reflected in experience, but are somewhat independent of it and therefore somehow located "in" the child. Whatever the stages, mechanisms, or processes proposed, structural models assume they can be inferred from a child's observable behaviors, but are not identical with them. In recent times the most well-known example of this assumption is that of Jean Piaget and his associates at the "Genevan School" (Piaget & Inhelder, 2000). As undergraduates now learn in introductory psychology, Piaget postulated several broad cognitive stages that collectively formed a structural model of development, beginning in infancy with sensorimotor thinking and ending in adolescence with formal operational thinking. In the 1960s and 1970s, immense research energy in both Europe and North America went to working out features of these stages and to testing their breadth, universality, and logical inevitability (Gelman & Baillargeon, 1983).

Structural theory, and Piagetian theory in particular, enjoyed popularity among early childhood educators during this period of research (see Weikart, 1971; or Kamii, 1976), but critics identified both theoretical problems and practical limitations for education. For one thing, cognitive stages turned out to be much less general than Piaget had supposed. Slight changes in wording or procedures often affected success at experimental tasks and thereby altered a child's apparent cognitive maturity (Donaldson, 1979). Cross-cultural differences in cognitive performances were common, and related to differences in familiarity with tasks, task materials, and sociolinguistic expectations about conversation during interviews (e.g. teacher-like "test" questions were not construed uniformly). These problems did not invalidate structural theory entirely, but they did lead proponents to define stages more specifically than before (see also Seifert, 1993). Neo-structural studies of cognitive development still looked for broad endogenous ("inner") processes of developmental change, but hypothesized-and seemed to demonstrate-that they unfolded at different rates in different areas of cognition (Case, 1991, 1999). A child might develop numeracy skills and moral thinking through roughly analogous steps or stages, for example, but not necessarily at the same rates.

A more serious theoretical problem was that structural theories often sounded naive or simplistic about how cognitive change actually occurred. According to Piaget, cognitive structures (or "schemata") developed because of an interplay between assimilation (interpreting new experiences or concepts in terms of old ones) and accommodation (modifying existing experiences or concepts to fit with new ones). A child who sees a butterfly may at first name it a "bird" (assimilation to a prior concept), but after additional experience modify this representation to distinguish two kinds of flying things, "bird" and "butterfly" (accommodation). Although cognitive bootstrapping like this was all right as an explanation as far as it went, it was also paradoxical and philosophically unsatisfying. Some part of the child-an equilibrating "faculty"-had to make the necessary assimilations and accommodations, and that faculty already had to be cognitively developed enough at birth to do so. It was as if some small piece of the Piagetian child, a cognitively mature "homunculus" lurking inside, was cognitively equivalent to the whole child. Yet if this were true, then "whole children" did not develop-only their overt behaviors developed.

The notions of assimilation and accommodation also did not account for the subtle, but common cases of cognitive change involving incommensurability: changes in which a new concept cannot, in principle, be understood either as part of an earlier concept or as a simple combination of earlier concepts. During the preschool years, for example, children differentiate the concepts of weight and density (Carey, 1991); yet neither concept, if understood accurately, may be either assimilated or accommodated from the other. Density is not simply a subcategory of weight, nor is it a superordinate category that includes weight plus other elements (e.g. volume); rather it is "weight per volume," usually represented mathematically as a ratio of the two. Developing the concept of density therefore requires accommodating (or modifying) twoinitial concepts, weight and volume, and combining aspects of their meanings without also losing track of their original senses. This is a relatively subtle and mysterious intellectual task, and more complex than simple assimilation or accommodation. It is also a task characterizing much of children's cognitive development, and increasingly so as they mature.

The emergence of incommensurable concepts can be explained a number of ways, but most explanations call upon social influences of some form, and therefore compromise the traditional assumptions of structural theory about cognitive development. One explanation, for example, is that new, incommensurable concepts are simply modeled and taught directly. A child witnesses others using the concept of density, for example, and (perhaps) is also taught definitions and examples of the concept explicitly. This explanation seems straightforward or even obvious, except that it is not clear whether explicit social transmission actually gives children an intuitive grasp of the new concept, rather than simply a new vocabulary or name for the concept. How is a young child supposed to sense the meaning of a previously incomprehensible idea (e.g. density) if he or she has no prior cognitive furniture with which to interpret it, and cannot be sure which prior experiences might help to build the furniture?

Another explanation for why incommensurable concepts develop is that a child experiences contradictions or (in Piaget's terms) disequilibrium which motivate learning the new concepts. At a water table in an early childhood classroom, for example, a child may note that floating and sinking do not depend on objects' weight; sometimes heavy things float and sometimes light things sink. Theoretically, this anomaly creates cognitive dissonance or instability and motivates development of a new concept, density, to account for the variations. This explanation also seems reasonable as far as it goes, except that it does not account for the uniformities often observed in new concepts that are incommensurable with former ones (Carey, 1999). After working at the water table, why do so many children eventually settle on the identical concept, density? Since density is initially so mysterious, why do not some children finally settle on other possibilities (e.g. wood always floats and rocks always sink, but some things cannot be predicted)?

A similar problem exists with a third explanation: that general reflective (or metacognitive) skills improve with age, and allow children to detect inconsistencies in their thinking and to revise concepts to reduce or eliminate them (D. Kuhn, et al., 1995). This explanation is also based the disequilibrium idea, that contradictions and disequilibrium stimulate cognitive change, but it also asserts that older, more developed children experience disequilibrium more fully and more effectively than younger, less developed children. Even if this idea were true, however, it is subject to the same criticism as its simpler version: it still does not explain why children tend to develop rather uniform conceptual changes, instead of idiosyncratic ones, even though their prior experiences are highly idiosyncratic and the concepts being learned cannot easily be mapped onto earlier concepts.

Is there a way to account for the uniform development of new concepts without reference either to simple (Piagetian) equilibration or to SBTL? The explanation based most fully on structuralist assumptions assumes that certain broad concepts and beliefs are for all practical purposes innate from birth, and that these "ancestor concepts" develop into a select number of core intuitive theories during infancy and early childhood (Carey, 1999). One ancestor concept, for example, is a belief that objects have underlying essences that are distinct from surface appearances and that determine their true identities. A related belief is that a human being keeps the same identity over time in spite of changes in appearance and setting-basically Piagetian object permanence applied to humans. A third is that living things, whether human or not, have intentions (Johnson, Slaughter, and Carey, 1998). These concepts, among others, allow a young child to develop an intuitive theory of human psychology, based on a central idea that "persons are a sort of living thing that has intentions." This core theory eventually accommodates (in the Piagetian sense) to experiences and to other initial core theories to beget intellectual descendants in the form of more sophisticated intuitive psychologies. The later psychologies include at least some concepts incommensurable with the earlier ones. Later core theories differentiate, for example, between accidental and intentional behavior, and still later ones differentiate unconscious behaviors from conscious ones-two distinctions not comprehensible earlier in development (Bartsch & Wellman, 1995)..

Whatever the merits of this structural explanation for conceptual change, it offers only limited help to early childhood teachers because it is still based on traditional assumptions about the nature of cognitive development-that cognitive change is long-term, that it happens "inside" the child even though influenced by "outside" forces, and that cognitive processes can be distinguished from non-cognitive ones. Taken together, these assumptions limit relevance to early childhood education, because they render young children relatively inaccessible to educational influence, and therefore make the early childhood teacher less responsible for the future betterment of her children. As Robbie Case and others have pointed out, the dominant educational strategy in programs guided by structural theories of cognitive development has been guided discovery: providing children with a rich, supportive environment so that development can proceed unimpeded (Case, 1998; Baker, 1999). The chief job of the teacher, according to structural theory, is to understand what children are capable of and where they are headed developmentally; intervention is only a secondary purpose. Emphasizing indirect guidance is certainly consistent with developmentally appropriate practice (Bredekamp & Cottle, 1997), but it is not sufficient advice for most programs in early childhood education. In general early childhood educators seek not just understanding, but appropriate intervention strategies as well. So it is necessary to look for additional ideas or theories in order to supplement structural theories of development. This task in turn requires modifying the traditional working assumptions of cognitive development embedded in structural theory. One such alternative source of ideas that meets these requirements is learning theory.

Learning Theory as Cognitive Development

Traditionally, learning theory has focused on forms of human change that were by definition somewhat peripheral to human development. Learning theory was about relatively rapid associations between stimuli or between responses and reinforcements. When stimuli, responses and consequences were all simple enough and when conditions were properly controlled, the fast-paced (by developmental standards) stimuli and responses displayed lawful properties and allowed for testable predictions about (short-term) behavior change. The resulting changes were called "learning." Connections between stimuli were named associative learning or "classical" conditioning, and those between a response and its consequence were named operant conditioning. Those between a response and consequence that were merely observed by an individual were called modeling, observational learning or "vicarious conditioning."

Assessed in terms of developmental theory and of SBTL, the problem with behaviorally oriented learning theory had three blind spots: it did not say enough about long-term change, about the experience of learning, or about how learning and behavior became organized in the minds and lives of learners. Observational learning theory was an early effort to remedy the last two of the three problems (Bandura, 1977), and represented an initial turn toward what SBTL theorists eventually called meaning-making. In observational learning, a person somehow "sees" relationships between a relatively complex behavior and its consequences. A student does not merely see a classmate elevate an arm and vocalize, for example, but sees the classmate "answer a teacher's question." If the teacher then tightens facial muscles, the student sees not merely muscular action, but sees the teacher "smile in response to the student." In observational learning theory, observed concrete behaviors become expressions of meanings perceived as happening in others (Bandura, 1995). This approach is still far from the thorny issues of interpretation discussed by social constructivists, but it is nonetheless a nod toward students' lived experience and personally constructed meanings.

Recent versions of learning theory have moved even more toward meaning-making than did Bandura and his colleagues. In one relatively radical formulation of learning theory, for example, conditioning has been interpreted as not about behavior change per se, but about making meaning (DeGrandpre, 2000). In this view, linkages among overt stimulus cues, responses and consequences are merely cases of this broader process. Operant conditioning is not merely about reinforcing or strengthening specific responses as a result of specific consequences, but about creating perceived relationships among responses, the stimuli prior to responses, and the consequences that follow. "Reinforcement" of a behavior is therefore not simply the consequences of a behavior, as in earlier versions of learning theory, but the relationship among all elements of a learning trial-stimuli, responses, and consequences. If a teacher smiles pleasantly at a young child's interesting comment during a circle-time discussion, for example, what is reinforced is not just the child's tendency to make verbal comments in the future. What is reinforced is actually the child's tendency to associate or connect the teacher's smile, the child's own verbal contributions, and the setting or conditions when doing so is appropriate. To say that "the teacher reinforces the child with her smile," in fact is misleading because it implies more control on the part of the teacher than the teacher really has. She does contribute to the reinforcement-relationship by smiling, but so do the classroom circumstances and the child's own behavior. All must happen together, and together they constitute reinforcement-and the creation of meaning.

Learning theory has also become more constructivist by developing more complex models of how children learn, and therefore more able to explain concepts and skills that truly matter in the lives of children and caregivers. An example is the "overlapping waves theory" proposed recently by Robert Siegler (1996). This theory makes three assumptions: 1) that children normally use a variety of strategies for solving problems, 2) that strategies can co-exist for long periods even when some are suboptimal or incompatible, and 3) that experience changes the relative salience of particular strategies, rather than eliminating any outright. Changes in salience alter the likelihood of particular strategies actually appearing as behavior. When graphed across time, relative salience rises and falls like a series of overlapping waves; hence the name of the model. Learning theory-or at least Siegler's overlapping waves version of learning theory-is about studying these changes in salience and their eventual expression as altered behavior.

The result, compared to older models of learning, are predictions about learning that are more realistic. In actual classroom practice, a child is indeed likely to try a variety of strategies in ways that are often somewhat trial-and-error. A child may work two-digit subtraction problems, for example, by using "borrowing" algorithms officially recommended by the teacher, but also persist in counting on fingers some of the time, and at other times also try personalized (often bug-ridden) one-digit procedures that work only with particular subsets of problems. Even when a child gains skill with a new strategy-in this case the "borrowing" algorithm-he or she still often regresses to earlier methods now and then, just as adults do. The child is sometimes also likely to spend time perfecting existing, suboptimal approaches rather than trying new ones-learning to count on fingers faster or more surreptitiously, for example-since doing so usually has short-term payoff. New, more optimal strategies appear both suddenly and intermittently when and if they finally do. The relative importance of a child's alternative subtraction strategies will gradually change with practice, but suboptimal ones will not totally disappear for quite a while, if ever. The results are patterns of behavior that resemble Siegler's "overlapping waves" model of learning (Siegler, 2000).

The educational value of a more meaning-oriented version of learning theory can be seen in the recent professional writing of early educators. Discussions of behavior problems in young children, for example, still draw heavily on concepts of applied behavior analysis (Kaiser and Rasminsky, 1999; Scarlett, 1998). Teachers are often still urged to frame behavior problems in terms of operant conditioning (or as antecedents, behavior, and consequences, which amount to the same idea), in order to influence a child's unwanted behavior. But though this literature is quite behaviorist, it often is also concerned with the meaning of antecedents, behavior, and consequences for the child: with the relationships she perceives among these elements, and with how a teacher might influence the perceptions in desirable directions. One result is that teachers are urged to alter not just consequences, but also the connections (or lack thereof) among antecedents, behavior, and consequences. A temper tantrum may be highly disruptive, for example, but the solution may not be simply to avoid paying attention to it (withholding reinforcement in the narrow sense). It may also be necessary to avoid connecting antecedents to the tantrum behaviors-avoid the triggers for tantrums in the first place. Not only is the broader strategy more likely to alter behavior, it is also more likely to help a child "see" his tantrum in new, more acceptable ways-to alter, that is, its personal meaning.

Much of the literature on behavior management also implies a learning theory of "overlapping waves." In discussing how to deal with a tantrum, for example, teachers are reminded that relapses are likely to occur, and that teachers should persevere with altering relationships among antecedents, responses, and consequences in spite of them. Expecting relapses implies that a child is guided by competing responses that are gradually shifting in relative importance or salience. Helping a child respond to frustration in consistently desirable ways is really a matter of developing a high ratio of desirable responses to undesirable ones, and thereby also making the desirable responses highly "meaningful" to the child in the sense of strengthening their connections to antecedents and consequences.

In addition to becoming more compatible with early childhood practice, such revisions to learning theory have had significant influences both on learning theory itself and on traditional meanings of human development, including cognitive development. The ideas about meaningful reinforcement described earlier bring learning theory closer to SBTLs in that it becomes more concerned than before about how learning is experienced by the child and how the child can take responsibility for "constructing" new, learned behaviors. The shifting, overlapping strategies described by Siegler, on the other hand, blur the conventional distinction between short- and long-term change which has been central to the idea of cognitive development. At the heart of the model is a process that is continuous, yet never complete, and in this sense it is neither short- nor long-term, but always both and always lifelong.

Mingling the short- and long-term is consistent with the professional needs of teachers and others with direct responsibilities for children. As professionals, we are fundamentally concerned with connecting the past and the future to a child's present: with helping children to relate previous experience and future intentions to current skills and plans. Classroom practice can be interpreted, in fact, as an active effort to combine the short-term (the current activity) with the long-term (the child's prior skills and knowledge with the child's goals), even though doing so confuses the conventional distinction between immediate learning and long-term growth implied by the term cognitive development.

In addition, of course, good teaching of the young integrates children's actions in other ways. Teachers often attend to a child's unstated intuitions and fully conscious reasoning simultaneously; or they deal with a child's individual academic achievements and social relationships at the same time. In order to understand these concerns in parallel, integrated ways, even the revised versions of learning theory are not fully helpful, and early childhood educators must again turn elsewhere, such as to information processing theory or to SBTL itself. The next section considers the first of these alternatives, information processing theory, and its relation to cognitive development and to early childhood education. Following this discussion, we can look directly at what SBTLs have to offer.

Information Processing Theory as Cognitive Development

Information processing theory has intellectual roots in the everyday metaphors of the mind as a container, filing cabinet, bank, and the like. Depending on the metaphor, knowledge (or "information") is an object or substance that either fills the container, gets filed, or is deposited for later use (Olson and Bruner, 1998). Learning consists of sorting and organizing knowledge or information, as well as coordinating already-organized information to determine how and when a person should respond. Knowledge itself is inert or passive, and the mind is neutral with respect to what it stores or organizes. The "energy" to motivate learning is distinct from both knowledge and its storage place, the mind.

The computer and its operations are the most recent version of the container metaphor, and lie at the heart of information processing accounts of learning and cognitive development. Information processing models often use computer-like language to explain thinking (Bereiter & Scarmadalia, 1998). They speak of working memory (like a computer's buffer) for short-term manipulation of information, and they speak of long-term memory (like RAM and/or a hard drive) for permanent organized storage. There are executive processes analogous to computer software, and these direct how information is transformed, categorized, and saved. As with real computers, there are size and speed constraints on each aspect of information processing. The computer-driven language implies that learning and cognitive change are essentially a matter of following already-known rules or instructions, much as a well-programmed computer does. Information processing therefore has been especially successful at explaining learning and thinking with a conscious, linear character, such as solving multi-step problems in arithmetic. To solve a word problem about long division, for example, a child (theoretically) applies a known algorithm, and success depends on correctly mapping the conceptual relations in the word problem to steps of the algorithm. As a child gains experience with word problems involving long division, he or she becomes more skillful at this sort of task.

Modeled in this rather straightforward way, the explanation seems clear enough, although there is little mention of children's common tendency to vacillate among solution strategies, or for them to regress following newly learning strategies, as discussed earlier in the context of behavioral learning theory. Instead, the model portrays children not as becoming more predictable, but as becoming more flexible or strategic in switching from one solution method to another when a first method does not work. Research on mathematical problem solving has found, for example, that five- and six-year-olds often fail to solve arithmetic word problems because they have only a limited repertoire of algorithms with which to interpret the problems (Riley, M., Greeno, J., & Heller, J., 1983). Two or three years later, their repertoire has expanded, and so has their success rate at solving word problems. In this and other ways, information processing is framed as "cognitive development" in the classic sense: long-term gains are distinct from short-term gains, and cognitive behaviors can be described and predicted separately from motivations, social relationships, and emotions. Note, though, that this picture of learning and development most effectively explains problems and cognitions describable as a sequence of steps. When a problem requires several subtasks completed in parallel, classic information processing fares less well, a limitation that is addressed in newer versions of information processing theory, discussed later in this section.

How useful is a linear view of information processing for early childhood education? Its chief educational implication is to highlight the value of direct instruction-activities organized and/or directed by the teacher having clear, focused goals. Such teaching does have a place in early childhood education, but dominant philosophies in the field tend instead to emphasize open-ended, indirect methods such as guided discovery or authentic, apprentice-like practice (Bredekamp and Cottle, 1997). Because of this mismatch of philosophical starting points, classic information processing theories, based on linear models of thinking, have had only limited application in early childhood education, for much the same reason that "classic" models of behavioral learning have had only limited impact. They are simply too focused and sequential compared to the classroom realities which they are modeling.

To some extent newer versions of information processing theory have remedied this problem by formulating models that account for non-linear learning and thinking. They have, for example, addressed classroom behaviors like these: 1) carrying a cup of juice without spilling, or 2) formulating an intuitive hunch during a class discussion, and 3) sensing when a classmate's behavior is covering up a feeling rather than expressing one (e.g. a smile is hiding anxiety). These are important cognitive skills for children, but they do not have a linear structure and therefore are difficult to explain as a series of linear steps with logical solutions akin to math problems.

The need to describe and predict non-linear thinking led in the 1980's and 1990's to a form of information processing theory called connectionism or parallel distributed processing(McClelland, Rumelhart, and Hinton, 1987; Sigel, 1999). These accounts portray the mind as large, multiple networks of associations, each responding to new experience by becoming either stronger or weaker, and each operating independently of and in parallel with the others. After any new experience, the relevant associations in the network individually change so as to optimize a child's performance on future occasions-although perfectly optimal behavior may never be achieved. Balancing a cup of juice, for example, stimulates various sensory associations within a child's hand, arm, and body. Some of these become stronger and others weaker as experience with cups of juice accumulate, but the eventual result is an optimal (if not perfect) ability to balance a cup. Likewise, immersion in a second language activates numerous associations among phonemes, words, and grammatical forms. As the associations shift in relative strength as a result of language immersion experiences, skillful (if not perfect) ability with the language emerges.

The assumptions supporting connectionism are more compatible with the realities of classroom life than are linear versions of information processing. As early childhood teachers themselves frequently note, learning by children often is less a matter of focused, sequenced steps, than of the broad, simultaneous learning of many associations combined with the optimal integration of separately learned connections. The early childhood teacher's role often consists of assisting this broadly focused process. Developmentally appropriate practice also gives a major role to incidental learning, to serendipitous learning, and to active roles for the child in shaping his or her own learning. All of these cognitive processes-breadth of focus, integration of associations, and child initiation-are consistent with imagining learning as a vast network or pattern of associations operating in parallel, rather in linear sequences.

Connectionism also sits more easily than linear processing with notions of individuality and diversity in learning. Instead of focusing on the learning of relatively pre-set, step-by-step content like algorithms, logical thinking, and decision rules, connectionism highlights the more open-ended challenges of recognizing patterns, integrating or balancing competing associations, and optimizing responses to diverse stimuli. Since these processes inevitably unfold differently for each child, the model also implies tolerance for the individuality that inevitably accompanies learning. Teachers should expect, for example, that no two children will learn to balance a cup of milk in exactly the same way, nor acquire exactly the same version of a second language. Expressed in terms that are less connectionist but more post-modernist, connectionism predicts that each child will be "positioned" by his or her experiences as they unfold; diversity will be the rule, not the exception.

Yet in spite of nods toward diversity, connectionism does not specify how diversity among children might develop or about how early childhood educators might deal with it under classroom conditions. The connectionist child remains somewhat "dis-heartened" in the sense of "lacking heart," motivation, or a context for learning. In connectionist theory, the optimizing of associations unfolds automatically, almost machinelike, even though in professional practice, early childhood teachers obviously do not think of their children as machines, no matter how complex their mental mechanisms may be. Normally, teachers concern themselves not only with optimizing children's skills and cognitive associations, but also with the social relationships, settings, and scripts that guide children's cognitive activity and allow it to develop. For insight about these social dimensions, teachers must therefore look elsewhere, and most obviously to SBTLs-the socially based theories of learning, described next.

Socially Based Theories of Learning

Socially based theories of learning (SBTLs) call attention how learning and development are interactive or socially mediated. Internal cognition is not denied, but treated as derivative to social interactions and cultural processes. In some versions of SBTL, even learning that appears solitary, such as reading a book or writing an essay, is interpreted as interactive: the reader has a (mental) dialogue with the author, for example, and the writer communicates with an (imaginary) audience.

As pointed out earlier, SBTLs have gained prominence-even dominance-in educational circles recently. The philosophical roots of the framework, however, are far from recent. Two hundred years ago, a form of socioculturalism was put forth by George Hegel when he argued that knowledge is synthesized from an intellectual dialogue or "dialectic" among ideas. One hundred years ago, a version of SBTL was also used by Karl Marx in arguing that knowledge and beliefs result from individuals' material positions in society: owners of property, in particular, conceive of human nature and human motivations quite differently than do workers, who are positioned only to "sell" their labor. Some form of SBTL is in fact implied by any deliberate teaching, simply because deliberate teaching necessarily entails human interaction, no matter what the local arrangements or practices for doing so (O'Connor, 1998).

Unfortunately the popularity of SBTL has led to a proliferation of models and theories based on broadly social premises, and therefore fostered ambiguity and misunderstanding about what a "social" perspective on learning means for education (Salomon and Perkins, 1998). For the purposes of understanding cognitive development in early childhood education, it is helpful to distinguish among three main varieties: 1) social mediation of individual cognitive change, 2) the use of cultural tools to learn or develop new concepts or ideas, and 3) participation in a group's efforts to create knowledge jointly. The first variety is the most familiar and compatible with everyday and "traditional" views of education and cognitive change, and the third is the least familiar and most radically challenging. The second falls between these extremes, with implications that are at once both more and less familiar.

Social Mediation of Individual Cognitive Change

The most familiar way of framing a SBTL is simply to think of an individual or group helping another individual to learn. A teacher helps a student to write an essay, for example, by offering suggestions or encouraging self-reflection about initial drafts. Three or four children work together to master a common set of mathematics problems, each offering help-but also receiving it-from the others individually. Parents rephrase an incoherent, but well-intentioned comment from their preschool child so that the comment makes more sense. And so on. In all cases help is offered and received, individual learning happens as a result, and over time the learning leads to broader "development" in the sense described at the beginning of this chapter. Note that this sort of social mediation is not confined to classroom instruction. It can happen anywhere, regardless of the setting-at home or school, during class or afterwards, at work or play. In fact social mediation is virtually universal in human societies (Kruger and Tomasello, 1998). When it does happen in a classroom, it takes familiar forms, such as a teacher assisting a child with reading a story, or assisting many children either in groups or one at a time. The constraints of schooling, however, make classroom interaction simply a special case of social mediation, one among many ways to assist individual cognitive change.

Social mediation by and for individuals feels familiar because it relies on the container metaphor for learning, discussed earlier as part of information processing theory. As pointed out before, the metaphor assumes that learning and development happen "inside" individuals and that resulting knowledge is in some sense portable from one situation to another. A solution strategy for math may have to be learned first with the help of others. Once learned, though, it belongs to the individual and can be used by him or her to solve other similar math problems; assistance will not be needed later. Likewise, new vocabulary or grammar may first have to be acquired by modeling language from others and being reinforced by them; but once learned, the new vocabulary and grammar belongs to the individual and can be used without further social assistance. Learning may be social, but the resulting knowledge is individual.

Of course, some social experiences assist learning more than others, and cognitive changes does not happen automatically just because more than one person is involved. In general mediation works best when a learner responds to the mediating person(s) actively, because he or she must take the information or strategies offered by others and transform them into forms that can be truly the learner's own (Rogoff, 1991). Suppose, for example, a preschooler who is outdoors gathering autumn leaves. If the teacher wishes to stimulate skills in classification effectively, the child cannot simply be told to group them in particular pre-set ways. Better (according to SBTLs) is to invite the child to devise categories personally, and to explore the potential of the personally constructed taxonomy that results.

Active construction of knowledge is crucial to the success of social mediation; hence the oft-used name for this perspective, constructivism. Active construction is well-supported by research on individual tutoring-and, of course, by many early childhood teachers' intuitions as well. Adults who tutor children in remedial mathematics, for example, are less successful if they provide only information and corrections, and more successful if they engage in intense interaction, give highly individualized or personal guidance, and ask for frequent active responses from their tutees (Rogoff, Matusov, and White, 1998; Lepper, et al., 1997). Interestingly, expert children who tutor less knowledgeable peers can achieve similar success, but their successes are less frequent so than adults'. Why? Presumably because child tutors, even when comparative experts, are more prone than adults to ignoring a learner's need for active engagement with new knowledge, and more focused on simply completing problems or tasks at hand (Briggs, 1998).

Nonetheless, almost any social mediation of learning may be better than none at all. Whether child or adult, and whether highly skilled or not, live interaction requires a learner to treat emerging ideas with reflective distance, simply to participate in the interaction. The reflective stance thus created teaches an implied lesson, that ideas are capable of being manipulated, refined, and elaborated as if they were objects (O'Donnell and King, 1999). The same metacognitive "lesson" may sometimes occur to adults or older children without mediation, particularly if they have had appropriate prior experience with self-reflection. But among the very young, it is hard to imagine metacognition and self-reflection developing far without the expectation and incentive of dialogue, with its attendant requirement of cognitive self-awareness. The importance of mediation for initiating cognitive reflection may account for early childhood teachers' widespread reliance on discussion and on loosely structured group activities as participation structures in the classroom. SBTLs promise to support and enhance these teaching practices by identifying specific conditions where they may be especially successful.

Cultural Tools To Aid Cognitive Development

An extension of the idea of social mediation is to conceive of socially made "tools" which extend individual learning and development. Here the term tools refers not just to physical implements (e.g. a desk calculator or a pencil and paper), but also to technical procedures (methods for solving arithmetic problems) or symbol systems (natural languages, musical notations). A cultural tool qualifies as a social mediator in three ways. First, it has conventional structure and use which are agreed on by a social community, though often unconsciously: a pencil is conventionally designed with a certain size and shape and conventionally used for writing, not for stirring soup. Second, a cultural tool extends thinking in particular ways intended and valued by society: a young child can perform more complex multiplication with a desk calculator, for example, than without one. The enhancements are much like what happens during live, in-person mediation, as described previously, though perhaps with effects that are less individualized. Third, a cultural tool stimulates particular new concepts, metaphors or assumptions about thinking and learning that go beyond immediate, practical uses of the tool. A computer not only facilitates children's efforts to solve logic problems, for example, but encourages children (and others) to regard the human mind as a sort of computer. Printed books, for another example, teach content, but they also encourage a modern belief that memory consists not of literal recall of words, but of the essential ideas underlying printed words (Egan, 1997; Bloch, 1998).

By pointing out these effects, SBTLs imply that early childhood teachers have an important responsibility to introduce young children to cultural tools. Computers, for example, should not be limited to later grades in elementary schools, any more than pencils and pens are. Nor should the acquisition of new vocabulary or new uses for language, whether in a child's first language or in alternate "foreign" languages. Cultural tools can, and often should, be introduced immediately-provided that early childhood teachers understand that the tools are simply being introduced, and that perfection must wait for additional exposure, practice, and maturity (Healy, 1998).

Participation in Joint Creation of Knowledge

The least familiar and most radical version of social mediation sees learning as participation in joint or group creation of knowledge. Sometimes called social constructivism,this framework deliberately blurs distinctions between individuals and their social contexts, distinctions that are crucial in other psychological theories. Knowledge, concepts, and meaning emerge through extended interaction and become distributed in complex ways throughout the interacting group. No one person ever "knows it all," but different individuals do learn different parts of "it all" as a result of differences in their roles and responsibilities within the group.

The fact that general knowledge is distributed implies that individuals' knowledge is situated or grounded in particular activities and social contexts, and that it is in some sense non-transferable. An early childhood teacher learns much about teaching, for example, through interacting with her actual current students. But her knowledge is grounded in the unique social chemistry of a particular classroom group, a combination that can never be fully duplicated in other classes or in subsequent years. The knowledge of teaching that she acquires from any one class therefore transfers to the next year's class only imperfectly. No matter how experienced a teacher is, therefore, each new school year quite literally poses new challenges, and inevitably feels both exciting and anxious.

Such a restricted view of transfer is a serious limitation from the point of view of more individualistic models of learning and development. But it is a less central issue within the framework of social constructivism than another problem, that of insuring effective participationin joint activities. Portability of knowledge is less important than whether everyone knows at least something relevant to the mission of the group. How can marginal individuals (e.g. children) find a legitimate place in the group's activities, and gradually become able to participate more centrally (Lave & Wenger, 1991; Sfard, 1998)? And if knowledge is indeed situated in particular activities or roles, then the very term transfer is a misnomer, since it implies movement of an intact object from one container of knowledge (e.g. my brain) to another (e.g. your brain). When social constructivists engage the problems of participation and transfer, they frame them not in terms of transfer, but of recontextualizing experience: what was learned previously is not transported to new situations, like a suitcase full of tools, but rather combined with new experiences into broader, more enriched knowledge and skills. When classroom learning fails to "transfer," it is because teachers have not helped children to reframe or recontextualize the new with the old: numeracy activities, for example, are not extended clearly enough from numeracy activities engaged in at home or on the playground (Dyson, 1999). The unfortunate result is children who not only fail to develop numeracy skills, but who also risk becoming marginal to the learning community of the class. In this version of SBTL, the latter is the more serious of the two problems, since it interferes with further learning.

The idea of joint construction of knowledge seriously challenges traditional assumptions about cognitive development, the nature of individual differences and the distinctiveness of cognition itself. With knowledge distributed in a social context and not portable in the usual sense, and with it not primarily located in individuals, the outcomes of learning and development become hard to identify in particular children. "What develops" is not the individual so much as the group or activity in which individuals participate. As odd as it may sound, a social constructivist might argue that a group or activity can "learn" in some sense, and that such learning does not necessarily imply that the participating individuals also learn.

The joint construction of knowledge, furthermore, is never a strictly "cognitive" activity, since by definition it involves social activity as well as individual thinking. The interactions among participants in a successful activity setting are not only social, but even significantly moral and emotional, and take on the qualities of caring relationships as described by Noddings (1992; Goldstein, 1999). In a group constructing knowledge jointly, the central participants focus attention not just on their own contributions, but also on those of marginal participants, and concern for the marginal individuals allows them to frame or scaffold the tasks of the activity with more sensitivity and effectiveness. More effective distributions of tasks, skills, and problems in turn also helps group performance. Under these conditions, some learning usually does accrue to individuals, but only part of it: much learning is better described as happening to the group or community as a whole. When a kindergarten class rehearses a Christmas pageant, for example, individual children do (hopefully) learn their parts, and many also learn bits and pieces of others' parts. But it is the class, not individuals, that "learns" to perform the play.

Because this way of thinking about cognitive development discards the container metaphor of learning, it has proved challenging to use in classroom activities and communities organized around individual learners as the fundamental educational clients. Yet a situated, social perspective on learning has much to recommend it for early childhood education teachers, or indeed for any teacher seeking to offer culturally and developmentally appropriate practice. Why? Because in line with recommendations for developmentally appropriate practice, the perspective highlights participation as a key problem for teaching and emphasizes its importance in early childhood classrooms (Bredekamp & Cottle, 1997). The situated, social perspective also steers teachers away from measuring content knowledge in conventional test-like ways, and toward more authentic forms of assessment. Socially situated learning suggests that instead of expecting children's to accumulate knowledge or skill along a few key dimensions, teachers should expect numerous qualitative differences-sometimes even radical ones-in what children learn from the "same" project or activity setting. Given a radical expectation of diversity, the planning and testing of pre-set, specified goals becomes less important than in other views of learning or development. Instead, curriculum goals and their assessment grow out of individual children's numerous differences and out of children's dialogues and activities together. These teaching implications are not new to early childhood education (Bredekamp, 1987), but their intellectual alliance with SBTLs, and particularly with social constructivist forms of SBTL, has not been fully recognized.

Conclusions

In spite of philosophical compatability between SBTL and developmentally appropriate practice in early childhood teaching-or perhaps because of it-there are dangers if early childhood educators adopt "social revisionism" too easily. The dangers are both theoretical and pedagogical. Theoretically, SBTLs cannot stand alone: to be truly meaningful, they must say something precise about how individual children learn: what they "get" from mentors, groups, classrooms, and communities. SBTLs must also say something precise about how activity settings are shaped or influenced by the individuals in them, and not just the other way around (Anderson, Greeno, Rider, & Simon, 2000). These conceptual bridges are what early childhood teachers need and seek, given their responsibilities. They can be realized only by taking seriously the more individually oriented theories of cognitive development and learning described earlier in this chapter. As the recent revisions to these theories suggest, rapprochements with SBTLs may indeed be possible, although more philosophical and empirical work still needs to be done.

Pedagogically, the dangers of SBTLs lie with their very popularity among early childhood teachers. As happened with Piagetian constructivism a few decades ago, SBTL may become identified with only parts of itself, and the parts will sanction distorted ideas of what constitutes good teaching practice. Ideas currently vulnerable to becoming stand-ins for entire SBTLs might be Vygotsky's notion of the zone of proximal development, for example, or activity theorists' notion of distributed cognition. In and of themselves, these two ideas suggest the value of certain socially oriented teaching practices, such as peer tutoring (for the zone of proximal development) or cooperative group work (for distributed cognition). Yet unless concepts like these become fully situated alongside ideas about individual cognitive development, they are likely to be overused, or else treated as mere philosophical window dressing divorced from actual practices, developmentally appropriate or otherwise (Slavin, 1995).

At bottom maybe Freud was right a century ago, when he argued that an individual and the social community do not usually have congruent qualities, needs, or interests, and that the two must therefore constantly negotiate their differences, explicitly or implicitly, in order to realize their respective goals (Freud, 1989/1930). Early childhood teachers may succeed best if they can realize that "civilization and its discontents" can happen in early childhood programs just as much as elsewhere. Translated into the work of teaching, this means that a child's (cognitive) development is one thing, the classroom community is another, and at times their purposes and needs cannot be reduced to each other. Understanding this feature of the human condition, and of fostering cognitive development in particular, may paradoxically help the child and (classroom) community become more congruent: help them both see what they truly hold in common, and where they can mutually thrive.