The role of parents in the socialization of children: an historical overview.

ELANOR E. MACCOBY

The approach of the next century is an appropriate time to take stock of psychology progress in the study of human development and to consider where is going. Attempting to understand the socialization process has been a long0standig enterprise in both social and developmental psychology. When broadly conceived, the outcomes of interest have not changed greatly over time. That is students of socialization continue to be concerned with cluster of processes that lead to adult being able to function adequately within the requirements of the social group or groups among which they live. Therefore, the target of outcome behaviors of interest have continue to be some aspect of adequate functioning.

What is meant by “adequate”? The meaning has varied, of course, but the is a common core of meaning stemming from the understanding that if children are to be adequate adults, they must acquire habits, skills, values, and motives that will enable them to (a) avoid deviant behavior, that is, avoid behavior that disrupt or place undue burdens on the functioning of other persons in the nested hierarchy of social groups within which individual live their lives (b) contribute, through work, to the economic support of self and family; (c) form and sustain close relationship with others. And (d) be able to rear children in their turn.

Although parents are not the only agents contributing to the socialization of children, the family has continued to be seen as a major – perhaps the major – arena for socialization. This reflects the pervasive assumption that ever though socialization and resocialization can occur at any point in the life cycle, childhood is a particularly malleable period, and it is the period of life when enduring social skills, personality attributes, and social orientations and values are laid down. The idea that “the child is father to the man” goes back to biblical times and probably before. So does the idea that an adult’s rectitude depends on having received proper training earlier in life from parents and other educators. Over many centuries, the writings of religious leaders and philosophers – as well as popular wisdom – have been repleate of theorys and speculations concerning what kinds of child training will produce well-socialized adults. It is only in this century, however, that childhood socialization processes have become the focus of scientific study.

Most students of socialization have understood that societies, particularly modern ones, cannot rely on the ubiquitous presence of policemen or monitors to keep individual members of society in line. (Among other problems, relying on monitors raises the recursive question of who would monitor the monitors.) Developmentalists have continued to recognize that socialization practices must be such as to bring children to some degree of self-regulation with respect to social norms. This aspect of socialization has been studied in various guises with various labels, among them conscience, resistance to temptation, internalization of values, postponement of gratification, moral development, and out-of-sight compliance to parental requirements. Changes in theoretical points of view have dictated changes in the way these outcomes have been defined, but some aspect of internalization has remained a common theme.

The affective aspects of relationships between parents and children—love, hate, fear, and empathy—have also continued to occupy a central place in most conceptions of the socialization process. As readers shall see, there have been changes in the role attributed to emotions, and some theories have given them minimal attention, but recognition of their importance has recurred repeatedly.

While these and other continuities can be seen in the field, of socialization research, there have also been sweeping changes. The first major change has to do with how inclusive the theories have been. There was an early period of grand, all-encompassing theories, which gave way to more modest theories that were more limited to specific behavioral domains or specific age periods. A second major change concerns the direction of effects. What began as top-down conceptions in which parents were seen primarily as trainers or transmitters of culture and children as empty vessels who were gradually filled up with the necessary social repertoires has shifted to a conception of socialization as involving mainly bidirectional and interactive processes. A third change has involved the development of more complex process models. Whereas early work consisted largely of a search for direct connections between given parental practices (or clusters of practices) and a given child outcome, current work adds a focus on processes that may mediate the way in which a parental practice affects a child.

Two Grand Theories

In the early part of this century, there was relatively little empirical research on family processes and their relation to children's development. Nevertheless this period did see the introduction of theoretical points of view that strongly influenced the empirical work that was to come. This period has been thoughtfully reviewed by Cairns (1983), and here I mention only some highlights

Behaviorism

The dawn of behaviorism ushered in a long period in which the socialization of children was seen by psychologists as analogous to the processes of learning being studied in the laboratory. Parents were teachers, and children were learners. The principles of classical and instrumental conditioning were seen as specifying the processes whereby children learned the required forms of behavior. Parents were the primary persons who set the agendas for what children were to learn and who administered the rewards and punishments that would strengthen desired behaviors and eliminate undesired ones from children's repertoires. Parents also set up the contingencies that enabled children to discriminate between situations in which a given behavior was permissible and situations in which it was not. (See the review by Gewirtz, 1969, for a detailed account of stimulus-response learning theories as applied to socialization.)

Habits once learned could be unlearned if they no longer received external reinforcement, and early-learned behaviors were not considered any more difficult to unlearn or replace than behaviors acquired later in life (Gewirtz, 1969, p. 61). In behaviorists' writings, the question of how to define behavioral units was seldom clearly posed (nor answered!), but it was recognized that emotions, as well as actions, could be conditioned, and both were subsumed under the more general concept of "responses." Responses could be quite small, isolated units (e.g., an eye blink or a smile), but smaller units of behavior could also be organized into smoothly articulated chains or clusters of acts that would make up a larger whole. Socialization, in summary was seen as a process of accretion of a repertoire of habitual social responses that had acquired a specifiable probability of occurring under specific conditions. The theory was not developmental except in its assumption that the younger the child, the more limited the repertoire and the more there was still to be learned. But new behaviors were thought to be acquired in the same way in childhood as at any other time of life

Psychoanalytic Theor

The second major theme, introduced early in the 1900s and strongly elaborated thereinafter, was psychoanalytic theory Many elements of this theory had to do with the socialization process and the role of parents therein. Some of the major propositions were as follows: 1. Early childhood is a time of high plasticity. Characteristics acquired at that time are nearly irreversible, although they may change the way in which they are manifested as children grow into adulthood.

2. There are two major intrapsychic forces—sexuality (libido) and aggression—and these progress through a series of predetermined psychosexual stages during the first 4 or 5 years of life. The theory was dynamic in that it was greatly concerned with children's emotional states (anger or love) rather than merely with the details of behavior.

3. Parent practices determine the quality of a child's experience at each stage and are crucial in determining what the longrange consequences of these experiences will be. Parents must impose unwanted restrictions on the free expression of children's wishes and impulses. Children will become angry at parents when restrictions are imposed, and parents must then deal with this anger, suppressing it in some way. Parents are the ones who channel their children's aggression into acceptable channels during the early plastic years, and they must turn the child's sexual impulses away from the parents themselves.

4. Children experience intense conflict: They love their parents and need their devoted, nurturing presence; they intensely fear the loss of this nurturance and come to understand that their feelings of anger and sexuality directed toward parents entail the danger of parental rejection and the loss of parental nurturance.

 5. The conflict is resolved through identification. Children "internalize" their parents and "introject" their values, forming a superego or conscience that is an internal representation of the parents (primarily in their regulatory capacity). Because the child's incestuous wishes are directed primarily toward the opposite-sex parent, there is greater risk of retaliation or rejection by the same-sex parent, and conflict resolution therefore takes the form of identification primarily with the same-sex parent. This identification carries with it an adoption of appropriately sex-typed behavior and attitudes, along with an adoption of a more general set of prosocial values. A crucial outcome of the identification process is presumed to be the capacity to self-regulate primitive impulses.

Psychoanalytic theory underwent considerable criticism and modification during the years following the initial formulations by Sigmund Freud. In particular, many argued that the energy or drive manifested in young children's behavior was not specifically sexual. Still, the propositions listed here remained major ones that influenced the socialization research that followed

Each of the two theories just outlined was a grand, overarching theory that presumed to encompass most of what was significant about the socialization of children. The theories differed in that, for learning theorists, the child was close to being a tabula rasa (except for some inborn reflexes and need states such as hunger and thirst), whereas for psychoanalytic theory children entered the early childhood years equipped with a set of primitive impulses that needed to be brought under social control. Nevertheless, for both theories it was primarily through parental control and teaching that the adult culture was seen as being passed on to each new generation of children.

Early Efforts at Empirical Test

The late 1930s saw the initiation of an active period of research, which continued and expanded through the 1940s, 1950s, and into the 1960s. Some studies were straightforward applications of behavior theory and demonstrated that specific infant behaviors (such as smiles or vocalizations) could be instrumentally conditioned or extinguished. However, the most important development of this period was the effort to reconcile the two grand theories. Or more precisely, the effort was to derive hypotheses from psychoanalytic theory and to reformulate them into testable propositions stated in behavior-theory terms. This work had its origin in the Yale Institute of Human Development in the 1930s (see Dollard, Doob, Miller, Mowrer, & Sears, 1939—Frustration and Aggression; Miller & Dollard, 1941), and for convenience I refer to it as "the Yale school." A segment of this work that dealt with child rearing and its effects was undertaken in the 1940s and 1950s by psychologists Robert and Pauline Sears and anthropologists John and Beatrice Whiting (see Sears, Whiting, Nowlis, & Sears, 1953).

In this work, reinforcement and punishment, as well as their schedules and combinations, were the primary antecedent constructs. Secondary drives were seen as being created out of primary biological needs, by means of associationist principles. If the research had been based on behavior theory alone, it might have dealt with a range of socialization outcomes, such as children's learning to share toys, to comply with adult requests, to become polite and well mannered, to acquire language efficiently, and to learn to read. Instead, the choice of outcome variables was dictated primarily by psychoanalytic theory. Children's aggression and their seeking of parental nurturance (translated as dependency) were a central focus of study, as were sex typing and manifestations of identification with parents (e.g., conscience). Weaning and toilet training received research attention, on the grounds that manifestations of fixations at one of the early psychosexual stages might be detectable. Conditioned anxiety was emphasized as a mediating process underlying fixations and as having long-term inhibitory effects within each of the behavioral domains. The studies dealt almost exclusively with socialization events occurring in the first 5 years of life, the years that were crucial for personality formation according to psychoanalytic theory.

The results of this body of work were in many respects disappointing. In a study of nearly 400 families (Sears, Maccoby, & Levin, 1957), few connections were found between parental child-rearing practices (as reported by parents in detailed interviews) and independent assessments of children's personality characteristics—so few, indeed, that virtually nothing was published relating the two sets of data. The major yield of the study was a book on child-rearing practices as seen from the perspective of mothers (Sears et al, 1957). This book was mainly descriptive and included only very limited tests of the theories that had led to the study. Sears and colleagues later conducted a study with preschoolers focused specifically on the role of identification with the same-sex parent in producing progress toward social maturity. They used a much expanded range of assessment techniques, including observations of parent-child interaction. The hypothesis that identification with parents was a primary mechanism mediating children's acquisition of a cluster of well-socialized attributes was, once again, not supported (see especially Sears, Rau, & Alpert, 1965, Table 40, p. 246).

One can see then that these large-scale efforts to merge psychoanalytic and behavior theory, and then to predict children's personality attributes from parental socialization methods, were largely unsuccessful. For one thing, the two theories were probably more intrinsically incompatible than was acknowledged at the time. But it is important to note that the weakness of the research results did not disprove either of the grand theories. Rather, the theories began to fall of their own weight. New developments within the field of developmental psychology were making it more and more evident that neither theory, as originally formulated, could succeed. More limited theories began to appear, each growing up in a more limited domain and attempting to encompass a more limited body of data than the two theories of grand design. Before turning to these new developments, however, I need to mention a different line of thought and research that was occurring concurrently with the events just described.

A More Modest Early Theory

In the 1930s and into the 1940s, a research group headed by Alfred Baldwin at the Fels Institute in Ohio undertook a longitudinal study of children and their families, making repeated home visits to observe parent-child interaction and assessing the development of the children at several successive ages. The group's thinking about what was important to observe and measure was influenced by the writings of Piaget and by the work of Kurt Lewin and his colleagues. Their theory, like the two theories of grand design, was a top-down theory, but it was also strongly developmental. They were the first to emphasize and to demonstrate that parenting must undergo systematic change with the increasing cognitive capacities of children. In conceptualizing what was important about parenting, they took their lead from the work on group atmospheres done by Lewin and colleagues in the late 1930s (Lewin, Lippitt, & White, 1939; see also Maccoby, 1992). In that work, it had been shown that under democratic leadership, groups of school-age children became more fully involved in group projects, displayed less hostility, and were able to work in the absence of supervision more effectively than children under autocratic leadership. Baldwin and colleagues contrasted democratic with autocratic home atmospheres and were able to identify meaningful connections between these atmospheres and the quality of children's functioning in out-of-home settings (Baldwin, 1949,1955).

Themes in the Work That Followed

Although elements of the two grand theories can be traced into the socialization research of the 1960s, 1970s, and 1980s, several profound changes occurred in theories about the nature of the socialization processes. These theoretical developments began to occur fairly independently of one another, being largely domain specific and not always compatible. I turn to them now.

The Decline of Simple Reinforcement Theory

The cognitive revolution that swept through much of psychology in the 1950s and 1960s did much to weaken the grip of simple S-R theorizing in general and reinforcement theory in particular. The impact on conceptions of socialization processes came first of all through the mushrooming work on developmental psycholinguistics.

Developmental psycholinguistic

It had long been understood that a child's early learning of the language of his or her culture was a necessary prerequisite for the smooth accomplishment of subsequent socialization steps. B. F. Skinner (1957) made a valiant attempt to bring language acquisition within the embrace of behavior theory, but the attempt was notably unsuccessful. The devastating review by Chomsky (1959), as well as reports by early students of children's language acquisition in the first few years of life, made it clear that the acquisition process had little to do with parental reinforcement. Parents were important in that they were the primary source of children's exposure to their culture's language. But children's early language was clearly not a simple imitation of phrases spoken by adults in their presence. Nor was it a process of selection, through parental reinforcement, from an initial large random repertoire of sounds, words, or phrases.

From the time of Chomsky's postulation of an innate language acquisition device to the current research on constraints in the acquisition of concepts (Markman, 1992), it has been clear that children do not come to language learning de novo. They are equipped with capacities or readinesses that are language specific, and they do much of the work of language acquisition themselves. While parents of course help to teach the meaning of words, children acquire the prosody and syntax of their language fairly independently of parental guidance, and their own inductive processes guide semantic development as well. These processes are clearly not top-down; the role of the learner is fully as important as that of the teacher. The revelations from developmental psycholinguistics served to demote considerably the power of parents as socialization agents. Parental inputs began to be seen as being used by children, rather than as determining what children do or learn. Readers will recognize that this was the same message that was conveyed by Piaget and his followers for a whole range of children's cognitive acquisitions.

Did researchers' increasing focus on cognition mean that the interest in socialization died out? By no means; only that the field of developmental psychology became bifurcated. Whereas earlier, when behavior theory had sought to encompass all aspects of development through its emphasis on learning as the central process, now one group studied language and thought, and another studied personality and socioemotional development. It was in the latter portion of the field that socialization processes remained important and where the language and concepts of behaviorism were retained for a longer period.

Attachment theory

orism were retained for a longer period. Attachment theory. A second body of work that served to weaken reinforcement theory was the research on attachment. As in the case of language acquisition, innate mechanisms were postulated. John Bowlby (1969) brought ethological theory squarely into the socialization arena. Drawing on work with nonhuman primates, he emphasized the evolutionary heritage that human beings brought to the early infant-mother relationship. Parent and infant were seen to be in a state of prepared readiness to develop reciprocal behaviors that would sustain the infant's development during the long period of dependency of human young. In this sense, parental behavior—in particular, maternal behavior—could be seen as instinctive, although it was by no means stereotyped or rigidly determined. There was a normal course of development of attachment behavior in human infants that depended on the responsiveness of the caregiver (Ainsworth & Bell, 1969). The process could be disrupted or deflected into maladaptive forms if the mother failed to perform her side of the interactive duet, but within a fairly wide range of maternal behaviors the adaptive function of the interaction would be achieved, making it possible for the infant to proceed smoothly to succeeding developmental steps.

Many of the behaviors identified by attachment theorists as part of the attachment syndrome were the same as what had previously been called dependency (e.g., clinging, staying close to the mother, and resisting separation), but their meaning and consequences were seen quite differently. In the "Vale school formulation, dependency was a kind of necessary side effect of the early nurturance of the infant—a set of habits that reflected the reinforcing properties of the mother but that needed to be dropped out if the child was to become able to function independently. The strongly contrasting view of the attachment theorists was that the attachment relationship was itself the necessary mediator of the child's ability to take on independent functioning. Parental responsiveness was not seen as reinforcing dependent behavior. In fact, parental responsiveness to children's dependency bids (crying or approaching) led children to behavior that was the opposite of what had been "reinforced": to a decrease in crying and to moving away from the parent to explore the environment (see Sroufe & Waters, 1977, for a detailed comparison of the two theories).

Nonreinforced learning through modeling

In the 1960s, Albert Bandura published some ground-breaking studies of imitative learning (e.g., Bandura, 1962,1965). Early work by the Yale school had included studies of imitation (Miller & Dollard, 1941), but they conceived of imitation in reinforcement terms. That is, children were so frequently reinforced for behaving like adults that adult behavior—or at least, the behavior of certain adults—acquired secondary reinforcer power. The Bandura formulation was quite different: performance was controlled by external contingencies, learning was not. He spoke of "no trial learning," and was able to demonstrate that children could acquire new behaviors without ever performing them overtly and without their ever being reinforced, merely by observing them being performed by others.

Even after learning theory was expanded to incorporate observational learning, the viewpoint concerning socialization remained a top-down one. By seeing siblings punished for certain behavior, children could learn that this behavior was not acceptable to their parents and avoid it without ever having been punished directly. But the contingencies applied by adults were still what children had to learn about and what mainly affected the probabilities that they would act in given ways. Social learning theory was revised and expanded over the subsequent years, becoming progressively more cognitive. (See Cairn's [1979] account of three generations of social learning theories.) In Bandura's writing, the label imitation was replaced by modeling, which in return was replaced by psychological matching, to indicate that the process was not one of simple mimicry of acts. Social learning theory remained essentially nondevelopmental, but developmentalists know that the body of knowledge concerning cognitive development is highly relevant to the processes of psychological matching. What information children will take in from the behavior of others, what kind of symbolic representation they will store, and how they will process the stored information in relation to specific situations they encounter—all these things must surely depend on their level of cognitive development. I return later to the implications of this fact for the processes known as identification.

Microanalytic analyses

In the 1970s, sophisticated computer technologies became available that made possible the analysis of moment-to-moment sequences of parent-child interaction. These sequences were first examined from the perspective of operant conditioning, in terms of the immediate contingencies members of a dyad were providing for each other's behavior. An important example of this approach is found in the work of G. R. Patterson and his colleagues (1980, 1982), who initiated a research program aimed at understanding the development of aggressive behavior in children (primarily in boys). They began with the top-down assumption that "most deviant behaviors, and particularly those relating to child aggression, are caused by inept performance of child management skills" on the part of the parents (Patterson, 1980, p. 1.) The group carried out detailed observations of the children's interactions with their parents, recorded in small units of real time. From the sequential data, they studied the moment-tomoment consequences that parents provided for aggressive children's behavior and compared them with the sequences found in the homes of nonaggressive (or at least normally aggressive) children.

The microanalyses had disappointing results from the standpoint of the initial theorizing. The studies did not show that parents of aggressive boys provided higher rates of either positive or negative reinforcement for aggressive behavior (see Maccoby & Martin, 1983, pp. 42-43, for detailed analysis of the findings). However, the yield of the studies was very rich in other respects. It was clearly shown that the interactions occurring between parents and children were indeed quite different in the households of aggressive children. There were long chains of mutually coercive behaviors; parents used somewhat more punishment with aggressive boys, but more important, punishment was less effective with these boys than with controls. Therapeutic interventions with the families demonstrated that it was possible in some families to improve the behavior of the children by establishing firm and rule-oriented (rather than capricious) parental control. By this means, the mutual avoidance among family members was lessened, so that joint problem solving became possible, and there was even a rebirth of mutual affection. The theorizing of the Patterson group evolved from a social learning approach to a social interactionist perspective (Patterson, Reid, & Dishion, 1992, pp. 2-4).

The moment-to-moment interactions of mothers with their infants and toddlers have also been observed and microanalyzed. With the availability of video-taping technology, it became possible to make detailed video records of interaction sessions, which could then be segmented into tiny time units for coding. It was quickly evident that mothers were getting into communication with their infants by, among other things, imitating them and coordinating their own activities to the infant's attentional states. In interaction with a skillful and responsive mother, the infant's social capabilities expanded, and the pair could build a more and more reciprocal system. While maternal responsivity could have been described as reinforcement, and there were attempts to do so, this conceptual framework turned out to be inappropriate on the whole. For example, infants did not simply increase the frequency of the responses for which they were being "reinforced" (e.g., smiling or crying). Instead, they learned a turn-taking schema and became able to wait for the mother to perform her response before initiating their own next action.

It is not possible in this space to do justice to the richness of the detailed work on mother-infant interaction, but a few contributions can be mentioned. The work has shifted attention from individuals to dyads, and this forms an important part of the growing "science of relationships" (Berscheid & Peplau, 1983; Hartup & Rubin, 1986; and see summary in Maccoby & Martin, 1983, pp. 26-36). The work has emphasized the importance of shared understandings, joint focus of attention, awareness by each of the other's intentions, and shared emotional states in the achievement of communication between parent and child (see Trevarthan & Hubley, 1978, on intersubjectivity). And the work has been highly influential in the redefinition of the socialization process, from one in which influence flows from adults to children to an interactive perspective.

Work on intrinsic motivation

In social psychology, there was a body of work on overjustification that dealt with whether individuals believed that their actions were self-motivated or were produced by external pressure. Studies with children were done in which their initial level of interest in a given activity was assessed, and then they were offered rewards for doing what they already wanted to do. It was shown that the children's interest in doing the activity declined (compared with pretreatment levels) when rewards were no longer offered. (Deci, 1975; Lepper & Greene, 1975) This work of course contributed to the decline of simple reinforcement theory and underlined the importance of children's interpretations of their own behavior and its causes.

Changing Conceptions of Children's Identification With Their Parents

As mentioned earlier, Freud's theory conceived of identification as a unitary process that brought about advances in several aspects of personality, including the adoption of sex-typed behaviors, taking adult roles in enforcing rules on self and others, resisting temptation, and feeling guilt over transgressions. As was noted, this theory was not sustained by research evidence. A major problem was that the different presumed outcomes of identification did not cluster together, nor did they have similiar socialization antecedents. Subsequent developments involved reformulations of the concept of internalization, and studies looked for relevant parental antecedents for different personality domains separately.

The work on modeling

Bandura explicitly discarded the terms identification and internalization on the grounds that they had too many surplus meanings. In early experiments, he did draw on psychoanalytic conceptions of identification to examine whether preestablished relationships between model and child would affect the likelihood that the child would imitate the model. The experiments showed that children were more likely to imitate a model who had previously been nurturant toward them or who had power over resources that children wanted (Bandura, 1965). The implications for the role of parents were clear: Because parents are both nurturant and powerful, children should be more likely to learn by observing them than by observing strangers.

Still, subsequent studies on the acquisition of aggressive behavior from filmed models showed that children would copy the behavior of unfamiliar models. It was clear, then, that the process of observational learning did not depend on a child's preestablished relation of dependency or fear in relation to a given model. Observational learning, then, was a much broader process than what had been previously called identification, and the interpersonal dynamics that had been presumed to underlie children's spontaneous taking on of parental characteristics were deemphasized.

Work on the acquisition of sex-typed behavior raised further questions about the role of identification. The psychoanalytic viewpoint was that boys became masculine and girls feminine by identifying with the same-sex parent. However, studies on parent-child similarities did not find that a child's sex-typed characteristics were related to those of the same-sex parent. And work by Perry and Bussey (1979) indicated that, with respect to the behavior displayed, children would preferentially imitate same-sex models only when there was consensus within a same-sex group of models and clear differentiation from a group of the other sex. This work helped to explain why children's sex-typing was so poorly related to that of the same-sex parent. It indicated that the child's acquisition of sex-typed behavior could not be seen as a process of incorporation of the characteristics of a single model. It reflected the more general point that children were choosing whom and what to imitate on the basis of their growing conceptions of what was relevant to their own self-definitions and aspirations. Identification then became more a consequence than a cause of children's sex typing.

The view from attachment theor

Following the period of frequent, overt manifestation of attachment behaviors in the first 2 years of life, attachment theory postulates the formation of an internal representation of the attachment relationship, an internal schema that then affects the nature of new relationships formed later in life (see Main, Kaplan, & Cassidy, 1985). How similar is this schema to the internalized parent that Freudians postulated? Not very similar. It is quite unlike the conscience or superego presumably derived from identification with a powerful, punitive, often critical parent. Nor is it equivalent to anaclitic identification, which is a child's internalization of a parent's nurturant qualities. Rather, the child's internal representation of the early attachment relationship reflects how secure and trusting that relationship was (and probably continues to be) and determines whether the child will be open and trusting, or apprehensive and wary, in approaching a new person with whom an intimate relationship might be formed. What is being "internalized" from a child's attachment experience is the quality of a relationship with a parent—importantly, the roles of both partners—not the personality characteristics of a parent (Sroufe & Fleener, 1986).

Work on morality and altruism

Hoffman (1975) assessed several dimensions of morality in school-age children, including whether their moral reasoning was based on fear of punishment or hope of reward, as distinct from principled reasoning or concern for the well-being of others. The child's level of social responsibility was assessed from the reports of teachers and classmates. The researchers looked for relationships between children's moral status and their parents' child-rearing methods. Psychoanalytic theory would suggest that parental use of withdrawal of love would be a predictor of these various aspects of internalization, but it did not prove to be. Rather, parents who used other-oriented induction—that is, those who frequently reminded their children of the effect of their actions on others—were the ones whose children were most likely to manifest internalized morality.

The importance of other-oriented induction emerged again, with much younger children, in the work of Radke-Yarrow and colleagues (Zahn-Waxler, Radke-Yarrow, & King, 1979). They found that children as young as 2 years old would respond to the distress of others by efforts to help or comfort. The children most likely to do so were those whose mothers, when the children were guilty of wrong doing, most often stressed the effects of their actions on others, rather than simply threatening or punishing.

One theme that emerged clearly from the large body of work on conscience was that parental use of power-assertive socialization techniques was counterproductive for children's development of internalized standards and controls.

Out-of-sight complianc

e. In several studies, it has been found that the parental behaviors that are associated with children complying to adult demands in the parent's presence are different from those associated with compliance when no parent is present (see review by Maccoby & Martin, 1983). If out-of-sight compliance can be encompassed within the range of meanings of internalization this research bears on my topic and shows some consistency with other findings. The findings are that even though power-assertive methods are often effective in obtaining immediate compliance, delayed, out-of-sight compliance is more likely if parents have used other-oriented indue tion and attribution of prosocial motives to their children. Parents behave as though they were aware of these connections. As Grusec and Kuczynski (1980) have shown, most parents use a variety of socialization techniques, including both power assertion and induction or reasoning. Their choice depends largely on the nature of the child's infraction; power assertion is more often used for immediate control, induction and reasoning for moral training.

Scaffolding.

There has been a strong recurrence of Vygotskian thinking concerning the role of parents in children's cognitive development. According to this point of view, cognitive as well as social development occurs mainly in a social context, through interaction with trusted, more competent partners (see Rogoff, 1990, for a review). Of course, it is stretching the concept of internalization to include the scaffolding work here. It is similar to internalization work in that it attempts to understand the processes whereby children come to be self-regulating. However, students of scaffolding describe a different route to this outcome, one that does not involve the dynamic aspects of identification. And although the parent is seen as teacher in one sense, the scaffolding work does not see the parental teaching role as one of applying "contingencies"—rewards, punishments, or corrections—following children's correct or incorrect responses. Rather their role is to provide a structure for learning that will increase the likelihood of children's succeeding in their attempts to learn. The kind of scaffolding that parents need to provide changes greatly with the child's age. At any age, however, the parent can function to simplify problems by breaking them up into component parts so that the child only has to do one simplified part at a time. The parent also functions to focus the child's attention on each successive element of a task as it becomes most relevant. Rogoff says,

In addition to the executive role of adult-child interaction, struc- turing the goals and subgoals of an activity, adult-child interaction may provide children with routines that they can use as their contribution to more complex activities. That is, routines of adult- child interaction may provide ready-made pieces of meaningful actions on which children can build their further efforts. (Rogoff, 1990, p. 95)

Clearly, this kind of analysis could be applied much more broadly, in particular to children's understanding of social routines and scripts. The work of Trabasso, Stein, Rodkin, Munger, and Raughn (1992) on children's progress between the ages of 3 and 5 in understanding the intentions, goals, and plans of storied characters indicates that when mothers tell pictured stories to their children, they incorporate information about characters' goals and plans that is slightly in advance of their child's current level of understanding, thus providing a structure for the child's next steps in achieving social meanings. At a more molar level, the role of parents both in structuring the household environment so that children will be able to explore without getting into trouble and in managing daily routines so as to be predictable and satisfying to children has surely not been given the attention in traditional socialization research that it deserves. However, instances of attention to these processes are beginning to surface in the socialization literature. For example, in a study of the children of divorced families, the structure provided by predictable household routines has proved to be a strong predictor of adolescents' adjustment (Buchanan, Maccoby, & Dornbusch, in press). I should note here that students of interaction between parents and adolescents stress that this interaction strongly affects adolescent ego development, but not through a process of identification or internalization (Powers, Hauser, Schwartz, Noam, & Jacobson, 1983). They point out that parental influence need not take the form of making the child similar to the parent; indeed, the ego development of the two generations is not directly related (see Powers et al., 1983, p. 20). Here one can see that Vygotskian approaches to self-regulation have moved very far from the earlier conceptions of the origins of self-regulation in internalization.

Changing Definitions of Optimal Parentin

Popular interpretations of psychoanalytic theory and some early research findings on the undesirable effects of punishment or rigid restriction of children led to the view that the ideal parent was the permissive parent. As noted earlier, Baldwin and colleagues advocated democratic parenting, by which was meant minimal restrictions and the involvement of children in family decision making to the maximum possible extent. Diana Baumrind began her career studying adult leadership styles (see Maccoby, 1992, for more detail on this history), and she became convinced that an optimal leadership style was not best described as a collegial arrangement in which leaders essentially became resource persons and co-workers, but rather that elements of democracy needed to be combined with elements of authority. She applied this viewpoint to studies of the socialization of children, being motivated in part by a conviction that a simple permissive philosophy of child rearing did not lead to optimal outcomes for children's adjustment.

Baumrind's initial typology of parenting styles is well known (Baumrind, 1973). Compared with either an authoritarian or permissive style, she regarded the authoritative style as optimal, and this style involved a combination of affection and attentive responsiveness to children's needs, along with parental imposition of clear requirements for prosocial, responsible behavior (to the degree consistent with the child's developmental level). Achievement of this pattern was seen to require considerable negotiation—even confrontation—with children, and parents needed to be firm, as well as kind and understanding.

Baumrind's concept of authoritative parenting has been widely adopted by other students of socialization and has been notably successful in distinguishing effective from ineffective parenting. Current examples may be found in the work of Dornbusch, Ritter, Leiderman, Roberts, and Fraleigh (1987) and in the large-scale study of family structures by Hetherington and colleagues (Hetherington & Clingempeel, 1992).

A number of definitions and redefinitions of parenting types have emerged as new groups of families have been studied and the age range of the children has been expanded. It is difficult indeed to identify the "same" parental attributes across time, because a parent deals with a child who is first an infant or toddler, then a preschooler, then in middle childhood, and then an adolescent. Especially for families in which the children are preadolescent or adolescent, a disengaged parenting pattern can now be seen to be frequent, and in its outcomes contrast with those of the authoritative or authoritarian patterns (Baumrind, 1991; Hetherington & Clingempeel, 1992; Lamborn, Mounts, Steinberg, & Dornbusch, 1991; Maccoby & Martin, 1983). Increasingly, researchers are finding that some parents are more fully committed to their parenting role than others (see Greenberger & Goldberg, 1989; Pulkkinen, 1982)— or at least maintain a fully engaged commitment over a longer period of the child's development—and that the degree of commitment may be even more important than the style with which that commitment is expressed.

For Steinberg, Elmer, and Mounts (1989), the optimal parenting cluster includes not only high acceptance-warmth and firm control (the two major Baumrind elements), but also a quality which Steinberg et al. call psychological autonomy or democracy. Democracy may be particularly important as children grow older and progressively more skillful in negotiating with their parents. Steinberg and Dornbusch (in press) have shown that when parents and adolescents are jointly involved in making decisions that affect the children's lives, the children have better self-regulation and impulse control than when parents either impose decisions unilaterally or leave the decisions to their children.

However authoritative parenting is defined and whatever the age of the child, there appears to be a common core of meaning that defines the optimal cluster, and it has to do with inducting the child into a system of reciprocity. An authoritative parent assumes a deep and lasting obligation to behave so as to promote the best interests of the child, even when this means setting aside certain self-interests. At the same time, the parent insists that the child shall progressively assume more responsibility for responding to the needs of other family members and promoting their interests as well as his or her own within the limits of a child's capabilities.

The Role of Affect

Freud and subsequent psychoanalytic theorists have stressed the power of children's emotions—love, anger, and fear—in the formation of their internal personality structures. In a different way, so have the attachment theorists, who have stressed the function of the child's attachment to the parent in lessening children's anxieties. Recent work has greatly enriched the picture of the way in which affect is involved in parent-child interaction.

Emotions constitute the first language whereby parents and children communicate with one another before the child acquires speech. Infants respond to their parents' facial expressions and tones of voice. Parents in their turn "read" the affective quality of their infants' arousal states, responding appropriately when their infants are either distressed or in a happy, playful mood. It is important to note that in responding appropriately to a child's mood state, parents are not simply matching it empathically. They respond to a child's distress with soothing, rather than by manifesting distress themselves, a reaction that sometimes calls for considerable emotional control on the parent's part.

Fernald (1992) has identified some cross-cultural similarities in the affective meaning of the tones of voice mothers use to infants, and facial expressions too have been found to have similar meanings for people from different cultures (Ekman, 1972). The initial phases of parent-child bonding, then, are based on affectively charged parent-child exchanges. Furthermore, the work on social referencing suggests that by the end of the first year, a mother's facial expression—either a smile or a fear face—influences whether an infant will explore an unfamiliar environment. And Cummings's (1987) work on overheard quarrels between adults shows that young children react to adult anger with distressed facial expressions and inhibited play.

It is becoming evident that ambient mood states have a good deal to do with the quality of parent-child interaction. Lay, Waters, and Park (1989) have shown that when a positive mood has been induced in a child, the child is more likely to comply with a mother's directions. Dix (1991) has shown that when a mother is in an angry mood—one that has not arisen as a result of anything the child has done—she is more likely to believe that subsequent interaction with her child will be unpleasant and that more sternness will be required. Patterson, in studying a group of exceptionally well-functioning families, observed that an important ingredient of their interactions was humor: They made one another laugh, and light, pleasant mood states served to defuse conflicts. Several theorists have pointed to parent-child interaction as a context in which strong emotions, both positive and negative, are especially likely to be aroused (Berscheid, 1986; Dix, 1991), because the achievements of each individual's goals depend on the coordinated actions of the partner. Dix (1991) has assembled evidence to show that strong emotions, once aroused, serve to organize, motivate, and direct parental behavior. He has also noted that the frequency of episodes of mutual anger is greatly reduced if parents are able to adopt children's goals as their own.

Affective exchanges between parents and very young children appear to play an important role in whether toddlers will react empathically to others' distress (Zahn-Waxler & RadkeYarrow, 1990; Zahn-Waxler et al., 1979). Follow-up work at ages 8-10 years with children who had been studied as toddlers (Kochanska, 1991) points to a connection between the parentchild affective exchanges in early childhood and the children's subsequent empathy with victims of others' wrongdoing. Kochanska, building on Hoffman's earlier work, argues that children's ability to experience discomfort, guilt, and anxiety associated with actual or anticipated wrongdoing is a necessary but insufficient condition for the emergence of conscience (a selfregulatory component is needed too), and she sees parental affective responses as central for the development of the relevant emotions.

From Individuals to Interactions to Relationships

As was mentioned earlier, in early socialization research, parent behaviors were called antecedents, and child behaviors were called outcomes, and when a correlation between the two was found it was usually interpreted as an effect of one person's behavior (the parent) on another individual (the child). In the 1970s, insistent voices began to be raised pointing out that the causal arrow might point the other way (Bell & Harper, 1977; Parke, 1977). Clearly, a simple concurrent correlation between attributes of two interacting persons tells one nothing about the direction of effects. Researchers began to use time sequences and change scores to try to identify the direction of influence within a dyad. Several microanalyses of parent-infant interaction sequences provided strong evidence for the view that the infant's behavior, more than the parent's, was driving momentto-moment sequence as they unfolded (see summary, Maccoby & Martin, 1983, p. 30). Evidence for the power of children to affect the course of bouts of parent-child interaction has continued to appear up to the present time (e.g., Kuczynski & Kochanska, 1990; Lytton, 1990; Patterson, 1986; Patterson, Bank, & Stoolmiller, 1990). Even more commonly, it has been evident that cycles of successive and mutual influence prevailed (Patterson, 1982).

The idea of bidirectional influence was originally one in which each participant in the interaction of the parent-child dyad was seen as shaping the other, by providing reinforcements or aversive consequences for one another's behavior (e.g., Sears, 1951). In recent years, however, this conception has given way to one that stresses the development of reciprocity and linked streams of behavior between the members of a familiar pair. In short, the interest has shifted to relationships (Hartup & Rubin, 1986; Hinde, 1987; Youniss, 1983). From this point of view, children are socialized mainly through participating in the interaction within close relationships. Interactions between intimate pairs (friends or family members) are quite different from those between strangers. Relationships are constructed over time. Patterson and colleagues have shown how the frequent occurrence of coercive cycles between parent and child can undermine the child's acquisition of prosocial behavior and positive social interactional skills. In part, this failure of socialization occurs because mutually coercive cycles do not allow children to gain experience in sustained joint activity with others. When a parent-child pair are able to engage in noncoercive joint activity, their streams of behavior become interwoven, so that the smooth continuation of one person's behavior depends on the partner's performing the reciprocal portion of the action. Partners develop coherent expectations concerning each other's behavior, joint goals, shared scripts from which each acts, and shared meanings that make fuller coordination of their activities possible.

"Vbuniss (1983) argues that socialization should not be described as a process whereby control of children is shifted from adults to the children themselves, who become progressively more autonomous and self-regulating. Rather, he says, at every stage of life, relationships involve coregulation, and individuals never graduate to being free of the regulatory requirements of intimate others unless they become social isolates. (See also Maccoby, 1984, on coregulation between parents and children in middle childhood.) This view implies that any enduring parental influence stems mainly from the nature of the relationships parents have coconstructed and continually reconstructed with their children. These relationships can vary in many ways from one parent-child pair to another. Some are such as to foster children's development, others inhibit it (see Hauser et al, 1987, on parents' enabling and restrictive interactive styles).

Present and Future

This article has shown that the study of socialization has been a highly active research field and has been undergoing major change. What questions remain open, and what directions are promising? One broad question concerns the sequencing of steps or phrases in socialization. To what extent do early socialization events constrain what the parent-child relationship can become at later times? That is, to what extent does the ability of a parent to cope effectively with socialization issues at one period of a child's life depend on the socialization that occurred previously? Although there are several domain-specific accounts concerning how early parent-child relationships feed into the child's subsequent social behavior (e.g., early secure attachment is associated with later positive reactions to the social initiations of new partners), we do not yet have a coherent theory of the ways in which parent-child relationships themselves evolve. What does a parent's ability to use an authoritative parenting style with a child aged 6-8 years old depend on? A secure attachment in infancy? Sufficiently skillful earlier scaffolding so that the child acquired the competencies needed for him or her to be treated democratically? Can one trace trajectories in which early parenting styles set a given parent-child relationship in a given direction so that its subsequent characteristics are predictable? Can such trajectories be redirected?

Mutual cognitions no doubt play a role in the carryover of one phase of socialization to the next. As parents and children accumulate a long history of interacting with one another, each acquires a set of expectations concerning the other's behavior and stereotyped ways of interpreting the other's reactions. Probably, each progressively reacts to the other more in terms of these stereotypes than in terms of the other's actual momentto-moment behavior. (Note that Patterson, 1980, has found that aggressive children often react coercively to other family members' approaches even when those approaches are benign, as though they interpreted them as having hostile intent.) If this is so, the rate of new learning that is derived from the interaction of a given parent-child pair should decline as time goes on, and relationships could become more rigid, because each person's stereotypes function to keep the partner from changing. However, stereotypes can and do change, and we do not know the extent to which early-formed stereotypes can set the trajectories that relationships will take over extended periods of time.

On a related issue, is each period of childhood a kind of window of developmental time during which certain socialization lessons are best learned? Perhaps it is time to reconsider the old question of critical periods. Are children more open in early childhood to the emotional conditioning processes that underlie empathy and interpersonal trust than they will be later? Learning to self-regulate affect is one of the major achievements of childhood, but it seems likely that as children achieve it, they are rendered less susceptible to any aspect of socialization that involves the arousal of intense affect. Or to put the matter differently, it is possible that the effects of socialization experiences involving strong affect are more enduring and more resistant to change than less intense encounters and that the encounters of early childhood are more affectively intense. There are echoes of earlier psychodynamic thinking here. While I do not believe developmental psychology will go back to psychoanalytic socialization theory in any serious way, the question of whether and in what ways early childhood is a period of special importance still exists and needs to be dealt with.

The most central assumption of all the socialization viewpoints I have examined is that events that occur in the context of parent-child interaction affect children's social behavior in other settings and at later times. Nowadays, considering the intellectual history I have reviewed, one can hardly see this as a matter of simple generalization of habits learned in one setting (with parents) to interaction with new partners. Every partner is different; each new relationship is coconstructed with a partner who brings something different to the relationship than any previous partner has done. How then do children build on their previous interactive experience with parents in the context of new relationships? Attachment theorists argue that even though specific behaviors are not carried over, the quality of relationships is likely to be: Children seek to reconstruct, with peers and later with intimate romantic partners, the relationship they had with a parent (Sroufe & Fleener, 1986). Others appear to believe that new developmental periods, new settings, and new partners open up possibilities for qualitatively quite new relationships. I suspect that the experiences children have in their same-sex peer groups in middle childhood (Maccoby, 1990) must be integrated in some way with in-family experiences to influence the quality of the relationships young adults form with romantic partners, but these questions remain to be explored.

As mentioned, research on adult relationships has some connections with the recent advances in the understanding of parent-child relationships. Nevertheless the fact remains that the parent-child relationship is unique in a number of respects, most especially in its asymmetry. Although this article has shown that influence is a bidirectional matter from infancy onward, there can be no doubt that the differential between parents and young children in power and competency is enormous. Parents select and design the settings in which children will spend their time, and to some extent, the identity of the cast of characters with whom the child will have an opportunity to interact. Parents control access to things children want. Parents are larger and stronger and can control children's movements physically (witness Japanese parents applying gentle pressure to the top of a small child's head to cause the child to bow—an early bit of training in deference). Parents have vastly greater knowledge, and children need to rely on this knowledge, especially in unfamiliar situations.

What are the implications of this asymmetry? A number of themes have emerged. One is that parents derive authority from their greater power and competence, and they cannot abdicate this authority without endangering the children. The work on authoritative parenting points to the importance of parents' carrying out the managerial and control functions in family life, and the writings of family systems theorists stress that the boundaries between the parent and child generations should be kept clear: Families become dysfunctional if roles are reversed so that children become the ones who nurture or control par ents; furthermore, there is evidence that children are less competent when their parents "disengage."

The research reviewed earlier indicates that it matters how parents exercise their authority. Simple unqualified power assertion seems effective for immediate behavioral control but appears to undermine children's progress toward becoming independently prosocial and self-regulating. In other words, although parents have great power (especially when children are quite young), they had best use it sparingly or selectively in disciplinary encounters. But the implications of the parentchild asymmetries in power and competence go much beyond the question of whether and how parents should punish or issue orders. Parents must use their greater interactive skills to adapt themselves to the child's capacities and current states. Thus in the first year, they xneed to speak the emotional language that infants understand, even though this language is quite different from speech to adults. The work on scaffolding points to the importance of the way parents arrange situations and event sequences so that the demands of a situation will be within the child's "zone of proximal development"; the time is ripe for extending the caffolding work beyond cognitive development and exploring its relevance to the growth of social competence.

The parent role calls for a very demanding admixture of childlike and adult perceptions. Clearly, parent-child interaction goes more smoothly when parents adopt the child's momentary goals as their own. To do so calls for considerable empathy with children's emotional states and ways of thinking. Yet a parent cannot fully adopt a child's point of view. There are longer term goals, in the child's best interests, that the child cannot appreciate and may indeed resist. And some of the parent's own goals may be independent of what is needed to serve the child's momentary goals or long-term interests. There is plentiful evidence that skill in role taking is important for smooth and mutually beneficial interaction among status equals, but socialization researchers have devoted little attention to what is required for successful role taking with much less mature partners. More important, little is known about the process whereby parents maintain multiple perspectives at the same time: taking the child's perspective affectively and cognitively while at the same time maintaining their own adult orientations. One thing seems obvious: To maintain these multiple perspectives calls for considerable effort and skill, and these in their turn must rest on parents having accepted almost unlimited, long-term commitment to promoting the child's welfare. It is in this respect that the parent-child relationship continues to be unique, and it may be expected to remain a distinct branch of the growing science of relationships.