MOTHER-INFANT INTERACTION AND ITS STUDY’
Jucwh L . Gewirtz2 NATIONAL INSTITUTk ( I F M h N T A L HEALTH UIld
Elizuhetlz I;. Boyd2 UNIVERSITY O F M A R Y L A N D BALTIMORE COUNTY
I. INTRODUCTION
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142
11. DIRECTION O F INFLUENCE IN CHILD SOCIALIZATION . . . . A . EARLY ANALYSES . . , . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . B. RESEARCH ON CHILD INFLUENCE ON THE CAREGIVER . 111. MECHANISTIC AND ORGANISMIC MODELS O F DEVELOPMENT AND INTERACTION . . . . . . . . . . . . . . . . . . . . . A. THE MODELS CONTRASTED . . . . . . . . . . . . . . . . . . . . . . . . . B. IMPLICATIONS OF TkIE MODELS FOR DIRECTIONS OF INFLUENCE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
142 142 144
146 146 147
IV. STUDY OF INTERACTIONS BETWEEN MOTHER AND INFANT IN NATURAL SETTINGS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
149
..
152
V . LEARNING ANALYSIS 01;“SIMULTANEOUS” BEHAVIORS
VI. AN EXPERIMENTAL ANALYSIS OF MOTHER-INFANT INTERACTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153 A. DESIGN AND RESULTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153 B. DISCUSSION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 158
’
Several methodological issues conwlered in this paper are also discussed, but less technically, in a concurrent paper that deals with the flow of influence and mutual attachment acquisition in mother-infant interaction (Gewirtz & Boyd, in press-a). Authors’ address: NIMH Laboratory of Developmental Psychology, B2A-25 NIH Clinical Center, Bethesda, Maryland 20014. 141
142
Jacob L . Gewirtz arid Elizabeth t;: Bo)*d
V11. SUMMARY AND CONCLUSION . . . . . . . . . . . . . . . . . . . . . . . . . .
159
REFERENCES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
160
I. Introduction In recent decades, there has been an increasing realization that there is a mutual flow of influence in parent-child interaction. That is, it is now generally assumed that the behaviors of each interactor can influence the behaviors of the other, to greater or lesser degrees. Moreover, the complexities of the sequential and concurrent features of caregiver-infant interaction in life settings have come into focus. At the same time, at the empirical lcvel there has been an emphasis on (a) the difficulties of describing, or even organizing, these myriad interaction complexities, and (b) the study of the directions of influence in that interaction. In this frame, we will begin our analysis of the study of mother-infant interaction by tracing the modal approach to the direction of influence in environment-child interaction in child socialization from an emphasis on oneway caregiver influence on child t o a two-way caregiver influence on child and child influence on caregiver. Next, we will present two general perspectives on development, the mechanistic model and the organismic model, as convenient vehicles for considering some theoretical and methodological issues that are of current concern for the scientific study of child socialization in general and of caregiver-infant interaction in particular. We will attend particularly to the distinctive assumptions of the two models that are relevant to the study of environment/caregiver-infant interaction and the issue of direction of influence in that interaction. We will consider the implications of the models for the study of interaction and direction of influence in the context of several descriptive approaches to naturalistic caregiver-infant interaction, and will conclude that mechanistic and organismic perspectives have very similar implications for the empirical study of causal influences in that interaction. We will close with an example from our own work, from a mechanistic viewpoint, of an experimental approach to the mutual influence between caregiver and infant in nonsimultaneous interaction.
11. Direction of Influence in Child Socialization A.
EARLY ANALYSES
Traditional theoretical and empirical approaches t o human social and personality development (e.g., psychoanalysis, social learning) have been conceived in a
Mother-liifarit Iiiteraction and Its Studv
143
mechanistic frame and based, for the most part, on the one-sided assumption that parent (or other caregiver) behaviors are the primary environmental determinants of child behavior and development. If the child’s behavior could at all influence those of the parent, it was assumed that this influence could play but a minor role in the child’s sociali/.ation. Observed relationships between parent and child behaviors were routinely attributed t o the effect of the parent’s behaviors on the child’s behaviors. Within a social learning approach, this assumption generated numerous demonstrations that selected child responses could be conditioned by actual or simulated parent behaviors (e.g., Brackbill, 1958; Etzel & Gewirtz, 1967; Haugan & Mclntire, 1972; McKenzie & Day, 1971; Ramey & Ourth, 1971; Rheingold, Gewirtz, & Ross, 1959; Routh, 1969; Weisberg, 1963). The limitations of this model of child socialization have been increasingly recognized during the past quarter century, and the assumption of mutual influence between caregiver and child behaviors has come t o prevail in diverse theoretical approaches t o child development. In an important instrumental-learning analysis published in 1951, Sears discussed the dyad with its simultaneous focus on the interdependent actions and motives of two or more persons and pointed t o instances where the child could influence another’s behaviors (in particular, the parent’s). Concerned primarily with child development as influenced by the social environment, Sears did not elaborate the implications of the dyadic concept either for existing research data or for prospective research. Further, when Sears’ paper appeared there was an insufficient data base t o compel a reinterpretation within a mutual influence model of child socialization. It may be for such reasons as these that, at the time it was published, Sears’ analysis had little impact on research on the direction of influence in child social development. I n the same decade, Bowlby (1958), influenced by ethological theory rather than learning theory, proposed innate releaser mechanisms whereby specics-specific maternal behaviors may be released by such infant behaviors as crying and smiling. Although Bowlby assumed that the “released” behaviors of the mother became organized around her infant (through an unspecified form of learning) to bind her t o her infant, in that paper he did not actually advance a mutual influence model for mother-infant interaction. During the same period, GewirtL (196 I ) attempted t o fill this apparent gap in Bowlby’s and similar approaches. He detailed a conditioning model for the mutual acquisition of behavioral attachment of mother to infant and of infant to mother. In this model, it was assumed that both mother and infant respond differentially t o the other’s behaviors. With examples from diverse interaction situations, Gewirtz showed how the responses of each could influence (condition) the responses of the othcr. In this conditioning process, stimuli provided by the appearance and behavior of the mother might acquire discriminative and reinforcing control over the infant’s responses, while stimuli from the appearance and behavior of the infant might acquire discriminative and reinforcing
144
Jacob I,. Gewirtz and Elizabeth F: Boyd
control over the mother’s responses and denote the acquisition by each of an “attachment” t o the other. Gewirtz’s model was open to the operation of unconditioned stimuli (releasers) for species-specific responses, as identified in animal behavior research and proposed by Bowlby ( 1 958). Paralleling Bowlby’s (1958) releaser-stimulus analysis, Gewirtz (1961) analyzed the learning processes that were potentially involved. He detailed how the infant’s smiles and cries, being reinforced by a mother’s systematic responding to them, in that very same context might be reinforcing other responses of the mother that the infant’s smiles and cries followed routinely. He also noted how often this mutual conditioning might occur without the mother’s awareness that her responses were changing systematically. More recent statements of this approach t o mutual conditioning and attachment can be found in a number of papers by Gewirtz (1969, 1972, in press-a, in press-b). The above-cited analyses had implications for interpreting directions-of-effects or influence in the literature of parent-child interaction. In that literature, cautions were occasionally raised that correlations did not necessarily imply a unidirectional flow of influence from parent t o child ( e g , Sears, Maccoby, & Levin, 1957). However, it was not until well into the 1960s that the issue of the direction-of-influence became central for numerous analyses of early child development and parent-infant interaction (e.g., Bell, 1968, 1971; Harper, 1971; Kessen, 1963; Korner, 1965; Wenar & Wenar, 1963). It was clear that diverse research results could not be interpreted exclusively as the effect of adult behaviors on child behaviors, and nunierous studies were subsequently directed toward assessing the effects of child features or behaviors on caregiver features or behaviors (e.g., Moss, 1967; Osofsky & O’Connell. 1972; Yarrow, Waxler, & Scott, 197 1). R . RESEARCH ON CHILD INFLUENCE ON THE CAREGIVER
We present here a sample of child features or behaviors that have been reported to influence adult features or behaviors. No attempt is made to evaluate the soundness of specific studies surveyed, t o present tlie specific results of those studies, or to evaluate specific authors’ inferences about the child’s behavioral influence where multiple interpretations of tlie results are possible. Further, this summary is limited to humans and is more representative of the mother-child interaction literature in earlier than in later life. Two designs have been used in the main t o examine the influence of child features or behaviors as independent variables on caregiver features or behaviors as dependent variables. In one design, demographic variables such as age, gender, race, socioeconomic status, developmental level, and clinical population, were used as gross indices of child behaviors and related to parent behaviors, whereas in the other design, individual differences in infant behaviors were related descriptively or experimentally t o variations in parent behaviors.
hi'otlier-lri/urrt bitcraction arid I t s Studv
145
Studies in which basically d ~ n i o g r a p h i cchild ~ variables were related t o observed parent variables include the following: infant gender was related to differential maternal smiling and talking (Thoman, Leiderman, & Olson, 1972), and infant vocalizations were related by gcridcr t o differential maternal responding (Lewis, 1972; Moss, 1967); infant fussing and crying were related by age to verbally expressed feelings of maternal attachment (Robson & Moss, 1970), and child age was related t o maternal speech (Ferguson, 1964; Snow, 1972); infant developmeiztal level was related to adoptive mother's speech (Beckwith, 197 1); and children's membership in clinical or normal population groups was related to personality characteristics of their parents (Bell, 1964; Connerly, 1968; Klebanoff, 1959; Prechtl, 1963). I n addition, mothers responded differentially to their own and other children in a stnicturetl tutorial situation (Halverson & Waldrop, 1970). Individual differences in child features or behaviors as independent variables affecting parent behaviors were studied descriptively with emphasis on naturally occurring child differences, aud cxperimentally with child diffcrcnces selected or manipulated. Individual difl'ereiiccs in in rant behavior were related to differential caretaking behaviors (Osofsky & DanLger, 1974; Yarrow. 1963); and child behaviors (e.g., dependency) were related to positive and negative teacher contact behaviors (Yarrow rt ol., 197 I ) . Some examples follow of studies that have used experimental or selective sarnpling techniques to assess parent or adult responses to child features 0 1 bchaviors as stimuli: Differential adult behaviors were related to IQ-score labels (liusenthal & Jacobsen, 1966) and to verbal-skill labels (Siegel, 1963) assigned to children with wliom the adults interacted. Parents responded differentially to those behaviors of their daughters characterized as either independent 01-dependent (Osofsky & O'Connell, 1972). The quality of infant cries was ielated to the temperature of the breasts of lactating mothers (Vuorenkoski, Was/-lliickcrt, Koivisto, & Lind, 1969) and to maternal responsivity to crying (Wolff, 1969). h i d the behaviors of adults instructed to train a nominal child to learn LI task t o a specified criterion were conditioned by simulated correct responses 01' that nominal child (Berberich, 1971 ). In the studies listed, only the adult behaviors were bases of the dependent variables;
'We note here inter a h that wlien gross demographic variables are used to index behaviors, the level o f analysis o f the independent, child variables is remote from the level of the dependent, parent variables. 'Illis use of independent and dependent variables at discrepant levels of analysis may be justified occasionally under a research tactic where the gross variable serves as a first appro\irnation to the independent variables at the appropriate analytic level. Nevertheless, a serious cost of this discrepancy between levels of analq-sis is that a number of interpretation< 01' thc f!inctional relationships reported in these studies become possible. Numerous plausihlc assumptions can be made to reduce the gross independent variables t o the child features o r behaviors that, under the theory used, are assumed to control the observed adult featurc? o r behaviors.
146
Jacob 1,. G ~ w i r t zand Elizabeth F. Boyd
however, in principle, both adult and child behaviors i n interaction could be studied as dependent variables. Several studies illustrate attempts t o preserve the sequential flow and integrity of dyadic behaviors in naturalistic observations of mother-infant interactions: infant fretting and crying were found likely t o initiate maternal behaviors of vocalization and rocking (Lewis & Lee-Painter, 1974); infant sucking patterns were found to be related to maternal movement (Richards, 1971); and conditional probabilities were reported for related shifts in mutual infant and maternal gaze states (Stern, 1974). We will elaborate further on such approaches to caregiver-infant interaction following the discussion of mechanistic and organismic models of development.
111. Mechanistic and Organismic Models of Development and Interaction In this section we outline the mechanistic and the organismic views of development, as discussed by Overton and Reese (1973; Reese & Overton, 1970; Overton, 1973, in press), to provide a context for considering environment/caregiver-infant interaction and the flow of influence within that interaction. These two general models provide a useful vehicle for examining some theoretical and methodological approaches to interaction and to the direction of influence within it. A. THE MODELS CONTRASTED
Mechanistic models of development assume that the organism’s behavior may be completely predicted from a knowledge of the antecedent material and efficient causes of that behavior. As a corollary, the organism has been termed “reactive” t o external events. However, this “reactivity” is a purely formal metaphor, entirely a function of the assumptions such models make about causes and effects. The behavior of the organism may act, or operate, on the environment as “cause” at some points in time. However, as cause, this behavior is conceived to be entirely a function of the antecedent material and efficient causes of it. It is assumed also that the interactions of the behaviors of the organism and of the environment can be meaningfully analyzed into sequential components t o demonstrate influences between the two behavior sets. Such interactions are defined as mutual actions in that they imply sequential causality and a unidirectional influence from antecedent cause to effect (Baltes & Reese, in press; see, also, Baer. as quoted in that same paper). In this frame, development is viewed as quantitative change in behavior that, ultimately, can be reduced to or predicted from the history of the material and efficient causes of
the organism’s behavior. Further, within mechanistic models this prediction and control constitute an explanatiou of behavioral development. Experimental manipulation of the behavior o f the organism and the environment to identify the material and efficient causes o f behavior-the controlling variables-is often the preferred method of investigating research questions derived from niechanistic models. Organismic models of developiiient view the organism (and the environment) as inherently active and constantly changing. It follows from the assumption or metaphor of an “active” organism that the organism’s behavior cannot be predicted completely under those niodels. O n a priori grounds, organismic models assume that the behavior of the organism and the behavior of the environment are interdependeiit and, Iherefore, that their interaction cannot be meaningfully analyzed into sequences of cause and effect (Overton, 1973). Such interactions are defined as strong rcciprocal interactions and arc assumed to involve simultaneous, rather than sequeiitial, operations of cause and effect, i t . , reciprocal causality. That is, the heliavior of the organism may be “effect” and the behavior of the environment “cause” and, simultaneously, the behavior of the environment may be “et’fect” and the behavior of the organism “cause” (Overton & Reese, 1973). I t is ill the sense that each variable may be both the cause and the effect of the other, siniultaneously, that the influence of the organism and the environment is termed “bidirectional” by Reese and Overton. Natural developmental changes i n behavior are viewed as qualitative, emergent changes, the products of strong I-eciprocal interactions, that may not be predicted from an analysis of component parts or antecedents of the behavior. Under the organismic model, t h e basic causes of behavior are final and formal, i.e., teleological, in that be1iavior:il change is viewed as directional and due to the essential nature of the form o r structure of behavior (Overton & Reese. 1973; Reese & Overton, 1970). Howevci, inaterial and efficient causes are assumed t o function either to inhibit or t o facilitate these natural structural changes, which are not themselves open to spccific empirical determination. Hence, empirical research on such qualifiers of natural development cannot, in itself, lead to the complete explanation of this developnient. Under the organismic approach, descriptive rather than experimental study of the behaviors of the organism and the environment as a unit is typically preferred for discovery of the teleological principles of organization and structure that explain behavioral development. B
IMPLICATIONS or THI MOD€ LS I OR DIRECTIONS OF INFLUENCI
Overton (1973, p. 86) has noted that, in principle, organismic assumptions of an organism and environment that are interdependent and spontaneously and constantly changing actually preclude the experimental analysis of directions of influence in strong reciprocal interactions. However, it appears that direction of
148
Jacob I.. Gewirtz arid Elizabeth E Boyd
influence in interaction may be conceptualized in the frame of an organismic approach in at least three contexts, when: (1) strong reciprocal interactions are assumed and the model incorporates assumptions about the relative influences of causes such that some influences comprising these interactions are more effective than others; (2) strong reciprocal interactions are at issue and some of the behaviors of individual members of the dyad are efficient or material causes that inhibit or facilitate changes within these interactions; or (3) the “convenient fictions” of independence and sequential causality arc adopted, and some interactive behaviors of the organism and of the environment (or of a second organism) may be meaningfully analyzed into independent, static components such that routine descriptive, experimental, and statistical methods apply t o their study. Given these working assumptions, such interactions are defined as weak reciprocal interactions (Overton, 1973, pp. 86-87; Overton & Reese, 1973, p. 79). Under an organismic model, it is apparently not possible t o distinguish strong from weak reciprocal interactions on an a priori basis. When routine empirical analyses of interactions yield results consistent with an organismic theory, it can be concluded that they were meaningfully analyzed and are, therefore, weak reciprocal interactions. If the empirical analysis of components of the interaction yields results that are anomalous or inconsistent with a theory, it can be concluded that these interactions were not meaningfully analyzed and are, b y definition, strong reciprocal interactions (Reese, personal communication). On a priori grounds, it would not be possible t o study experimentally the assumptions of the relative degree of influence within strong reciprocal interactions. However, the study of facilitative or inhibitory effects on strong reciprocal interactions by material and efficient causes, and of weak reciprocal interactions, would not differ operationally from that of mutual actions described for mechanistic models. That is, in operational terms, the empirical study of direction of influence in both weak reciprocal interactions and mutual actions would be the study of unidirectional influences or one-way causality. On this basis, therefore, the two disparate perspectives, mechanistic and organismic, appear to have similar implications for studying the direction of effects in interaction. At the same time, mechanistic models with their emphasis on sequential causality routinely would attempt to order, or relate to antecedent or concurrent variables, phenomena that, t o organismic theorists, might connote reciprocal causality and/or facilitative or inhibitory effects on the formal course of development. In principle, mechanistic models assume that the complete empirical determinants of development can be specified, whereas, in principle, organismic models assume that development is a function of strong reciprocal interactions that have no specific empirical determinants. In the context of our earlier discussion of influence in caregiver-child interaction during child socialization, the study of the direction of influence neces-
Mother-hiiafalit Iriteractiori and Its Study
149
sarily involves unidirectional or sequential causality: the behavior of one member of the dyad is considered to be constant or is held constant-the independent variable-and its effect on the behavior of the other member of the dyad that is free to vary-the dependent variable-is assessed. Thus, in an empirical, though not necessarily in a theoretical, context we have raised the issue of direction of influence to correct traditional approaches that focused almost exclusively on the study of unidirectional influence from parent (or environment) t o child and hence t o provide a focus for the unidirectional influences of both caregiver and chdd in the interchange. The theoretical assumptions about the basic chai-acteristics of the organism and behavior and the interpretations of the observations depend on the theoretical approach employed. Under an organismic model, the observations could connote simultaneous bidirectional influences. Under a mechanistic model, these same observations could connote simultaneous unidirectional influences. In Section V we will describe approaches t o the study of such simultaneous unidirectional influences within a mechanistic approach.
IV. Study of Interactions between Mother and Infant in Natural Settings Mother-infant interaction in natural settings often involves numerous actor turns and behaviors that can occur concurrently and/or in rapid succession, with overlapping durations. Even with devised behavior categories available. the interaction flow between a mother and her infant could be so extensive as to make difficult even the recording of the details of the flow of interaction in such unrestricted life settings. Moreover, a few minutes of observation of the behavior flow could generate very considerable amounts of data, including myriad unique sequences. Some might conceive of such phenomena as reflections of strong reciprocal interactions, as units that cannot be readily analyzed into components and that involve simultaneous rather than sequential operations of cause and effect. We concluded in the preceding section that, on a priori grounds, under an organismic approach the experimental study of direction of influence within strong reciprocal interactions is precluded. It may be argued that the organismic assumptions of interdependence and spontaneous change, when taken to their logical extremes, preclude even the descriptive study of interaction phenomena that are assumed to be a function of strong reciprocal interactions of the behaviors of the organism and environment, without the “convenient fictions” of independence and behavioral stability, as the categorization of these behaviors implies independent and static characteristics. Yet assuming that such phenomena may be studied empirically, we will here sample alternative descriptive methods that have been used t o study motherinfant interaction in natural
150
Jacob I,. Gewirtz and Elizabeth F. B o ~ d
settings, to compare a descriptive approach to mutual actions w i t l n a mechanistic frame by J . L. Gewirtz and Gewirtz ( I 965) to three descriptive approaches to the complex details of the interchange flow that may reflect the organismic concepts of strong reciprocal interactions. In this comparison we will attempt to determine whether the organismic approach, which emphasizes reciprocal causality with interdependent and emergent phenomena, implies descriptive methods that are different from those implied by the mechanistic approach, which emphasizes sequential causality in terms of independent and stable categories for environment and behavior. It should be noted that the authors of the latter three approaches are not necessarily committed to the study of development as a function of strong reciprocal interactions. A procedure was specifically devised by Gewirtz and Gewirtz to record mutual actions in interaction occurring in natural settings(J. L. Gewirtz & Gewirtz, 1965; H. B. Gewirtz & Gewirtz, 1969). Retaining actor reference, the method emphasized the sequential order of behavior and stratified behavior elements according t o (a) whether or not the behaviors occurred in interaction, (b) background factors and persons, and (c) thematic settings (e.g., feeding). The determination, in interaction, of the conditional probabilities of each behavior of an actor being followed by each behavior of the other actor could be made. Those authors also employed a behavior-free pattern categorization of interaction in terms of actor turns and identities. Thus, the symbol “CMCM” represented a four-turn interchange sequence initiated by a behavior of the infant and terminated by a behavior of the mother. Gewirtz and Gewirtz related distributions of various of these interaction indices to such variables as identity of the interactor, thematic setting, place of interaction, and time of day, as well as t o such demographic independent variables as sibling order, group child-rearing conditions, and age. Lewis and Lee-Painter (1974) presented four methods of analysis of naturalistic observations of mother-infant interactions: (1) overall frequency of types of behaviors of mother and infant; (2) frequency of types of behaviors occurring within a 10-second period, without actor reference; (3) two-actor sequences in a 10-second period with each actor’s behavior referenced as initiator or as response; and (4) conditional probabilities of transitions from the following states on single 10-second trials: neither mother nor infant behaves; infant behaves; mother behaves; both behave; or mother behaves t o another and infant behaves. Lewis and Lee-Painter (1974) have pointed out that these methods of analysis are of single actor or static elements of interaction and are not of interaction as independent of these elements. Also, they have suggested that alternative methodologies may need to be developed to study the “flow” of interaction, independent of its elements. In the context of the formation of attachment, Hartup and Lempers (1973) proposed that a description of the structural properties of interaction is provided by group profiles of the overall or niean frequency of the behaviors of both
Motlzer-Iiifuiit Iritcrurriori a i d Its Stud)>
151
infant and mother, without actor reference. Further, these authors considered that the second method of Lewis and Lee-Painter (1974), that of giving behavior frequencies within a 10-second period without actor identification, describes interactions as units rather than in terms of their individual components. Hartup and Lempers presented three response indices of attachment that were described as a composite of the behaviors of both interactors without referencing either actor in descriptions of the interaction unit: (1) distance between interactors, (2) dyadic gazing, and (3) dyadic mimicry. Brazelton, Koslowski, and Main (1974) developed a method of describing one-minute observations of mother-infant interactions, which they sampled longitudinally. Their intention was to maintain the integrity of the interaction chain by delineating cyclical patterns of attention and nonattention on the basis of the type and number of concurrent mother and infant behaviors. They described the infant’s attending behaviors as building up gradually to a peak intensity and frequency and then declining gradually. They described three general patterns of maternal response to the infant’s patterns of attention and nonattention: (1) increasing behavior during attention and withdrawing during nonattention; (2) increasing behaviors during nonattention; and (3) maternal attention and nonattention that is out of phase with the infant’s cycle. The four descriptive approaches surveyed are similar in that, at the observational or data collection level, behavior elements of the interaction are recorded. The data collection methods differ on the durations of the interval in which behavior elements are sampled and in the categories of behavior elements observed. At the data reduction level, the approaches are similar in their attempts to extract regularities or patterns of either concurrent or nonconcurrent behaviors of the two actors. Those who extract regularities or patterns of change without actor reference (e.g., Hartup & Lempers, 1973; Lewis & Lee-Painter, 1974, method 2 ) represent novel approaches. As the actors are not referenced, the resulting profile of behaviors docs index the interaction as a unit. However, without actor reference, there is a loss uf interesting regularities (such as behavior duration and identities o f initiators) and of sequences of actor behaviors. Further study can indicate whether the use of this technique will provide useful additional leverage on the process of interaction. As expected, inferences about directions of influence made from the observations are similar also in that sequential causality and independence are assumed. For example, Hartup and Lempers (1973) report that the frequency of dyadic gazing concurrent with other behaviors decreases over time. From this observation they conclude: “As the cognitive and motor apparatus of the child changes, the proportion of gazes in conjunction with other modes of contact decreases” (Hartup & Lempers, 1973, p. 247). I n this instance, the inferred developmental changes in the child’s capacities influence the frequency of beliaviors in the
152
Jacob 1,. Gewirtz arid Elizabeth F. Boyd
interaction. In a similar fashion, Brazelton ef al. (1974) propose that in motherinfant interactions the mother who reduces the intensity and frequency of her responses t o decreases in infant attentive behavior maintains longer periods of interaction. Therefore, data collection methods and inferences about directions of effects of the disparate mechanistic and organismic models are similar. Data observation depends on the analysis of the stable and independent components of interaction. Inferences about, or definitive empirical demonstrations of, directions of influence in interaction depend on assumptions that allow for the study of weak reciprocal interactions (under organismic models) or efficient and material causes with routine experimental and statistical methods. The heuristic functions of the two theoretical perspectives, for guiding and interpreting research, would be dissimilar. Within the organismic approach, research would focus on the (1) descriptive study of the course of development to discover the formal causes of development, which are not open t o empirical determination, and (2) facilitative and inhibitory effects on the course of development. Mechanistic models have been developed t o order sequential relations between environmental and behavioral events. Within such an approach, research would focus on the study of the functional relations among sequences of variables, t o provide the description and determinants of development.
V. Learning Analysis of “Simultaneous” Behaviors In some analyses of mother-infant interaction, there is a premium placed on assessing simultaneous or concurrent behaviors of both members of the dyad. We explore here briefly the feasibility of applying an instrumental-conditioning paradigm to the study of direction of influence in the interaction of concurrent behaviors. It might be concluded that the assumption in a conditioning approach of sequential causality between stimulus and response units as cause and effect implies that the independent and dependent variables can only be defined discretely and sequentially in time. However, this is not the case. Consider the example of interaction patterns observed in the naturalistic study of caregiver-infant interactions cited earlier (Brazelton er al., 1974). Suppose it is hypothesized that the caregiver monitored the infant’s patterns of nonattention and attention and maintained optirnal durations of attentive interaction by decreasing the level, in number and intensity, of behaviors concurrent with decreases in attentive infant behaviors. To test this hypothesis of the direction of influence between concurrent interactive behaviors, the caregiver’s behaviors could be defined as being at one of two levels of an independent variable: (1) providing a constant level of stimulation and (2) decreasing the level of stimulation. The caregiver could then be instructed to behave, alternatively, in these two manners toward the infant, concurrent with the infant’s decrease in atten-
Mother-Injaiit I:iteractioii arid Its S t u d y
153
tive behavior. In this way the effect of caregiver behavior on the concurrent duration of the infant’s attentivc behavior could be assessed. Similarly, the stability of the patterns of the infant’s attentive behavior, either on cue or simulated, could be manipulated systematically to assess its effect on the caregiver’s maintenance of level of stimulation t o the infant. If then it were found that each interactor’s behaviors affected the other’s, when one interactor’s behaviors were held constant and the other’s behaviors were free t o vary, there would be some basis for the inference of mutual influence at a theoretical level. In order to analyze the effects of a caregiver’s behaviors on the infant’s behaviors, or vice versa, it has bceri thought necessary t o manipulate or t o hold constant either the caretaker’s or the infant’s behaviors. However, the working assumption has been that the observed functional relations would be valid when these behaviors are free t o vary i n natural interaction. Further, for cases of multiple determinants operating in an experimental context, where the effects of some factors are controlled, it has been routinely assumed that the furictional relations resulting from these procedures are operative in a context in which these multiple factors are free to vary in natural interaction. Further, the study of the effects of each interactor’s behaviors on the other’s might reveal ways in which different features of the behaviors of each are affected, that would better explain the pattern of their interaction when the behaviors of both are free to vary. There is another way t o look at the study of concurrent influences, in the frame of a mechanistic model like the instrumental conditioning paradigm. It may be assumed that a response of the caregiver (C, ) may function t o condition a response of the infant ( I , ) and that, simultaneously, another response of the caregiver (C,) may be conditioned by another response of the infant (I2). Wirhin response systems, the influence of C , on I , and the influence of Iz on C, would be conceived to be unidirectional. However, between response systems the influence is concurrent: at the same time that caregivcr behaviors can be influencing infant behaviors (C, on I , ) , infant behaviors can be influencing caretaker behaviors (I2 on C2). We now turn t o a first approximation of the experimental study of motticrinfant interaction, with emphasis on the direction of influence in that interaction. Examples will be presented from sonie contrived studies of the conditioning of maternal responses by infant behaviors.
VI. An Experimental Analysis of Motherlnfant Interaction A. DESIGN AND RESULTS
The design and results of two studies are described here. These studies begin an experimental program on the direction of influence in mother-infant inter-
154
Jacob L . Gewirtz and Elizabeth F. Boyd
action. They illustrate research derived from a mechanistic model that emphasizes sequential causality via an instrumental conditioning paradigm (Gewirtz & Boyd, in press-a). Under a mechanistic model, this interaction is defined at both theoretical and empirical levels as mutual action (Baltes & Reese, in press). The influence that caregiver behaviors can have on infant responses in interaction has abundant experimental documentation (e.g., Brackbill, 1958; Routh, 1969). Hence, in these two completed experiments, emphasis was placed on studying the influence that infant behaviors can have on a mother’s response in interaction. The mother-infant interaction was contrived experimentally and the infant’s behaviors in the interaction were simulated. The mother and (nominal) infant were on opposite sides of a one-way mirror, with the mother facing the mirror side (Fig. 1). In these studies, each of eight middle-class mothers was asked to serve as “experirnenter” in an investigation of whether infant “subjects” could learn that their behaviors can have social consequences. Each mothcr was asked to provide these consequences by saying a short phrase (M) immediately following (i.e., when cued by) representations of one of her infant’s behaviors (I,,, ). A seemingly responsive but incidental behavior ( I R + ) , attributed t o the mother’s own infant, was identified for her as merely occurring in
Fig. 1. View from the infant’s darkened room, showing positions of infant (with a doll in infant’s seat) and o f mother (demonstrated by a modell in lighted room on the other side of one-way mirror.
Mother-Iiijar~tltireractiori and Its Studv
155
the situation. An explicit attempt was made t o give the mother no indication to attend t o that incidental infant behavior (IK+), much less to foster it. That infant behavior (IR+) was then presented contingent on one of the mother’s The actual purpose of the learning verbal and/or expressive responses (MI<). experiments was to determine the reinforcing efficacy of those seemingly responsive, but incidental, infant behaviors--vocalizations or headturns (IR+ )-for ). everyday maternal responses-verbali/,ations and/or facial expressions (ME< Specifically, in Experiment 1 , each mother was asked t o say a phrase (M) immediately following her infant’s headturns to her (I,,, ). Incidental vocalizations ( I R + ) attributed t o her infant were then presented contingent on the mother’s verbal phrases accompanied by her smiles @IR), to determine the reinforcing effectiveness of the infant vocalizations (IR+ ) for those maternal social responses (MR). In Experiment 2 , the elements of this paradigm of instrumental conditioning were reversed and extended to a study of the influence of a different infant behavior, the headturn to the mother (IR+), as a potential reinforcer for different maternal expressive response ( M R ) . Each mother was asked t o say a short phrase (M) immediately following (i.e., when cued by) her infant’s vocalizations (I,,, ). To detemiine the reinforcing effectiveness of those infant headturns to the mother ( I R + ) for her expressive responses (MR), the infant headturns were then presented contingent on her facial expressions (in one case crossed with her verbal responses). The maternal expressive responses (MK)that were reinforced were full smiles, partial smiles, and nonsmiling, clipped and sustained oral expressions, and smiles and nonsniiling accompanying particular verbal classes. The results of the two experinients are summarized in Figs. 2, 3, and 4. Contingencies were in effect in the two studies until the responses of each mother reached a criterion increase; therefore, the group curves charted in Figs. 2 and 3 are representative of the individual curves. For Experiment 1, Fig. 2 shows the mean response frequency curves for the reinforced verbal category accompanied by smiles on the first two arid the final three 45-trial blocks of the three conditions the four mothers had in common. Under Conditioning 1 and Conditioning 2, a maternal verbal category accompanied by smiles was followed by a vocalization attributed to the mothers’ infants. Under the Reversal 1 condition, this contingency was removed from that maternal response class and was, instead, provided following the alternative verbal response accompanied by smiles. Inferential statistical tests applied to the pattern of slopes of the conditioning and reversal curves of Fig. 2 showed that the incidental infant vocal behavior, when applied contingently on the maternal response charted, functioned as a reinforcer t o condition that response. Consequently, in this instance the incidental vocalizations attributed t o the offspring of the mothers were shown to influence maternal behavior, as indexed by the learning effected in the mothers’ responses (Cewirtz & Boyd, in press-b).
Jacob I,. Gewirtz and Elizabeth F. Boyd
156
REVERSAL 1
CONDITIONING 1
L-4-
1
2
3
4
5
CONDITIONING 2
1 y - 5
45TRIAL BLOCKS
Fig 2. Mean frequencies o f the verbal-with-smile response first reinforced in the Conditioniizx 1 treatment, under three treatments with infant-vocalization reinforcement (Mothers 1, 2, 3, and 4 ) .
For Experiment 2, Fig. 3 shows the mean response frequencies, for the reinforced maternal facial expressions, of the terminal 45-trial block of Conditioning 1, Reversal 1, and Conditioning 2, of the three mothers who participated in all three treatment conditions. Under Conditioning 1 and Conditioning 2, the representation of a headturn t o the mothers, attributed t o their infants, followed a maternal facial expression. Under the Reversal 1 condition, this contingency was removed from that maternal facial expression and was, instead, presented after an alternative facial expression of the mother. Inferential statistical tests applied to the pattern of means charted in Fig. 3 showed that the incidental infant headturn to the mother functioned as a reinforcer t o condition the facial expression of the mother. Hence, as in Experiment 1, the incidental infant response functioned as a reinforcer t o influence the maternal responses (Boyd & Gewirtz, 1976). This learning can be illustrated for an individual mother. Figure 4 presents response data for Mother 5 , whose full smiles, and then whose partial smiles, were conditioned by contingent infant headturns t o her. As it was necessary for that mother to withdraw from the study, she was the only one of the four mothers in Experiment 2 who was not subjected t o a Conditioning 2 procedure. Hence, her data did not contribute t o the means that are charted in Fig. 3. Figures 2, 3, and 4 demonstrate that stimulated infant vocalizations (in Experiment 1 ) and headturns to mother (in Experiment 2) functioned as reinforcers t o
/:;I 1 Mother-Infant Interaction and Its Study
CONDITIONING 1
KEVEKSAL
157
CONDITIONING 2
2
t a w: 15
10
4 FINAL 45-'TRIAL BLOCK
Fig. 3. Mean frequencies (Mothers 6, 7, and 8) of the expressive response first reinforced under Conditioning I , for the terminal 45-trial block in each of three treatments.
condition the mothers' verbal phrases and/or expressive responses. Hence these maternal social responses canie readily under the control of what, in the interactive context, appeared to be the responsive vocal or headturn behaviors of the mothers' own infants. Systematic postexperimental interviews and unsolicited comments during the course of the studies suggested that learning
full smiles
"
Partial smiles
It I
Fig. 4. Frequencies of full smiles, partial smiles, and nonsmiles for Mother 5 under three treatments with infant-headturn reinforcement.
158
Jacob L. Gewirtz and Elizabeth F. Boyd
occurred without the mothers’ being aware either that (a) there were infant vocal or headturn behavior contingencies in effect for their verbal and/or expressive responses, (b) their responses had been conditioned, or (c) the behaviors of their infants had been simulated. These conclusions received support also from the fact that, in Experiment 2 , expressive responses were conditioned definitively even though mothers were not asked to make them. B. DISCUSSION
In the studies reported, mother-infant interactions were experimentally contrived to facilitate investigation of the interaction process by demonstrating selective functional relations among the behaviors of mother and infant. This procedure markedly reduces the rich and complex detail of natural mother-infant interaction, on the assumption that it is not necessary to capture all or even many possible features of life interactions to demonstrate the plausibility and utility of a theoretical conception. Therefore, our initial approach has been to delineate a context for interaction that involves very few actor turns and a single imposed sequence. With this experimental tactic, functional relationships or direction of influence among the sequenced behaviors of mother and infant can be determined definitively. This tactic also facilitates a demonstration of the relevance and utility of an instrumental-conditioning model to account for the influence of the behavior of one actor, the infant, upon the behavior of the other actor, the mother. Because interaction phenomena in natural settings may involve behavior elements that occur concurrently, in rapid sequence, and overlap differentially in duration, it is difficult to analyze them in the detail required to demonstrate their relevance or irrelevance to a particular model such as operant conditioning. This is not to conclude that such conditioning principles operate only under highly restrictive conditions. Rather, they can be shown to operate definitively only under highly specified conditions. At the same time, the simplified interaction contrived in the present studies shares salient features with natural mother-infant interaction. Although simplified and in part simulated, the sequence of maternal verbal and expressive behaviors followed by infant vocalizations, or of maternal facial expressions followed by infant turning toward the mother, are like behavior sequences that occur routinely in mother-infant interaction in life settings. Therefore, it is proposed that the reported results have validity for a portion of mother-infant interaction in natural settings and that mutual-conditioning processes may underlie many of these interactions. At present, the usual descriptive approaches to mother-infant interaction in life settings sample and summarize interaction instances that are relatively simple, involving few response elements in discrete or in convenient overlapping time periods. Indeed, the examination of the four descriptive approaches indi-
Mother-Infant Interaction and Its Study
159
cates, on an empirical level, that the interactions they described are not unlike those studied in the contrived situation we described here. By restricting focus to a small sample of interaction, the experimental tactic employed in the studies we reported is not unlike those descriptive tactics used to sample interactions in order to extract regularities within and between actor turns in the natural setting. Indeed, we know of no observation tactic that is capable of efficiently organizing the complexities of mother-infant interaction in life settings, except when (a) only a limited segment of the interaction instances is sampled, (b) those instances are relatively simple, involving few response elements, which occur in discrete or otherwise convenient periods of time, or (c) features of interaction sequences are abstracted. Admittedly, it is important to map mutual behavioral regularities as they occur in the natural environment. However, the functional implications of these regularities must often be presumptive in the absence of experimental or statistical manipulations. Because it facilitates the determination of functional relations between interactor behaviors, the experimental tactic employed in the studies reported offers a promising alternative to routine descriptive approaches of observing natural interaction on its own terms. It is planned to extend the results of these studies to a demonstration of the mutual conditioning of the actual behaviors of both the infant and the mother. Initially, it is planned to demonstrate a discriminative-cue function of an actual infant’s behavior for a mother’s behavior and the reinforcement control of this actual infant behavior by another of the mother’s behaviors. In these contemplated studies, the simulated mother-infant interaction will more closely approximate natural-setting interactions than was the case in the studies reported. As our experimental simulations of mutual conditioning in mother-infant interactions come to approximate more closely naturally occurring interactions, a conditioning account of the development of a substantial portion of these interactions will become increasingly plausible.
VII. Summary and Conclusion We have discussed the direction of influence in child socialization in general and in caregiverinfant interaction in particular. Theoretical and methodological approaches to caregiverinfant interaction and the directions of influence within that interaction were discussed in the context of two general theoretical perspectives for human development, the organismic and the mechanistic. These disparate models imply very different conceptualizations of behavioral development. It was concluded that, as they are presently conceived, the implications of the two disparate models are very similar for the empirical study of the directions of effects within caregiverinfant interaction. In the context of an organismic
160
Jacob L. Gewirtz and Elizabeth F. Boyd
model, it appears that the empirical study of interaction and of the direction of influence in interaction is possible either under simplifying “fictions” of independence and sequential causality or when the facilitative or inhibitory effects on interaction of material or efficient causes are at issue. The assumptions of independence and sequential causality, of course, underlie a mechanistic model. The heuristic functions of the two theoretical perspectives, for guiding and interpreting research, would be very dissimilar. Within the organismic approach, research would focus on the (1) descriptive study of the course of development to discover the formal causes of development, that are not open to empirical determination, and (2) facilitative and inhibitory effects on the course of development. Mechanistic models have been developed to order sequential relations between environmental and behavioral events. Within such an approach, research would focus on the study of the functional relations among sequences of variables, to provide the description and determinants of development. We noted that researchers operating under the aegis of very different theories (e.g., mechanistic or organismic, cognitive or learning) may employ descriptive research methods that are very similar, and that no existing observation tactic appears capable of providing differential leverage on the complexities of what some conceive to be simultaneous effects comprising mother-infant interaction in life settings. In this context, we concluded that the experimental tactic involving the contrived procedure we reported offers a promising method for the study of interaction. REFERENCES Baltes, M. M., & Reese, €1. W. Operant research and operant paradigm: Resolution and apparent contradiction. In B. C. Etzel, J. M. LeBlanc, & D. M. Baer (Eds.), New developnzents in behavioral research: Theory, method, and application. In honor of Sidney W.Sijou. Hillsdale, N.J.: Erlbaum Associates, in press. Beckwith, L. Relationships between attributes of mothers and their infants’ IQ scores. Child Development, 1971,42, 1083-1097. Bell, R. Q. The effect on the family of a limitation in coping ability in the child: A research approach and a finding. Merrill-Palmer Quarterly, 1964, 10, 129-142. Bell, R. Q. A reinterpretation of the direction of effects in studies of socialization. Psychological Review, 1968, 75, 81-95. Bell, R. Q. Stimulus control of parent or caretaker behavior by offspring. Developmerital Psychology, 1971,4,63-72. Berberich, L. P. Do the child’s responses shape the teaching behavior of adults? Journal of Experimental Research in Personality, 1971, 5, 92-91. Bowlby, J. The nature of the child’s tie to his mother. hzternatiorzal Jounial of Psychoanalysis, 1958, 39, 350-373. Boyd, E. F., & Gewirtz, J. L. An infant’s headturns to its mother condition her facial expressions. (Submitted for publication, 1976.) Brackbill, Y . Extinction of the smiling response in infants as a function of reinforcement schedule. Child Development, 1958, 29, 115-124.
Motlzcr-hifairr Iritcraction and Its Stud>,
161
Brazelton, T. B., Koslowski, B., & Main, M. The origins of reciprocity: The early motherinfant interaction. In M. Lewis & L. A. Kosenblum (Eds.), The effects o f t h e infant on its caregiver. New York: Wiley, 1974. Pp. 49-76. Conncrly, R. J. A comparison of personality characteristics of parents of brain-injured and normal children. Dissertation Ahstracts, 1968, 28, 1291-1292. Etzel, B. C., & Gewirtz, J. L. Experimental modification of caretaker-maintained high rate operant crying in a 6- and a 20-week old infant (Infans tyrannotearus): Extinction of crying with reinforcement o f eye contact and smiling. Journal ofExperimenta1 Child PS.Vchology, 1967, 5, 303-317. Ferguson, C. A. Baby talk in six languages. In J. Gumperz & D. Hymes (Eds.), The ethnography of communication. Amvican Anthropologist, 1964, 66, Part 2, 103-114. Gewirtz, H. B., & Gewirtz, J. 1.. caretaker settings, background events, and behavior differences in four Israeli child-rearing environments: Some preliminary trends. In B. M. Foss (Ed.), Determinants of inJant behaiiour I V . London: Methuen, 1969. Pp. 229-252. Cswirtz, J. L. A learning analysis o f the effects of normal stimulation, privation, and deprivation o n the acquisition of social motivation and attachment. In B. M. Foss (Ed.), Determinants of infant behaviorrr. London: Methucn (New York: Wiley), 1961. Pp. 21 3-299. Gewirtz, J. L. Mechanisms of social Icnrning: Some roles of stimulation and behavior in early human development. I n D. A . Goslin (F,d.), Handbook of socialization theory and research. Chicago: Rand McNally, 1969. Pp. 57-212. Gewirtz, J. L. Attachment, dependency, and a distinction in terms of stimulus control. In J. L. Gewirtz (Ed.), Attachment and tlependencj’. Washington, D.C.: Winston, 1972. Pp. 139-177. Gewirtz, J. L. Maternal respondinp and the conditioning of infant crying: Directions of influence within the attachment-acquisition process. In B. C. Etzel, J. M. LeBlanc, & D. M. Baer (Eds.), New developments in behavioral research: Theories, methods, and applications. In honor of Sidnry W .Bijoii. Hillsdalc, N.J.: Erlbaum Associates, in press. (a) Gewirtz, J. L. The attachment acquisition process as evidenced in the maternal conditioning of cued infant responding (particularly crying). Human Development, in press. (b) Gewirtz, J. L., & Boyd, E. 1:. Experimcnts o n mother-infant interaction underlying mutual attachment acquisition: The inl‘ant conditions the mothcr. In T. Alloway, L. Krames, & P. Pliner (Eds.), Attachment bchai,ior. Advances in the stud]>of communication and a j j k t . Vol. 3. New York and London: Plcnum, in press. (a) Gewirtz, J. L., & Boyd, E. 1‘. Infant vocalizations condition maternal verbalizations. Child Development, in press. (b) Gewirtz, J. L., & Gewirtz, H. B. Stimulus conditions, infant behaviors, and social learning in four Israeli child-rearing environments: A preliminary report illustrating differences in environment and behavior betwccn the ‘only’ and the ‘youngest’ child. In B. M. Foss (Ed.), Determinants o.f infant hehaviour III. London: Methuen (New York: Wiley), 1965. Pp. 161-184. Halverson, C. F., & Waldrop, M. I:. M;iternal behavior toward own and other preschool children: The problem of “ownness”. Child Deidopment, 1 9 7 0 , 4 1 , 839-845. Harper, L. V. The young as a source of stimuli controlling caretaker behavior. DevelopmentalPsychology, 1971,4, 73-88. Hartup, W. W., & Lempers, J. A problem in life-span development: The interactional analysis of family attachments. I n P. B . Baltes & K. W. Schaie (Eds.), Life-span
162
Jacob L. Gewirtz and Elizabeth F. Boyd
developmental psychology: Personality and socialization. New York: Academic Press, 1973. Pp. 235-252. Haugan, G. M., & McIntire, R. W. Comparisons of vocal imitation, tactile stimulation, and food as reinforcers for infant vocalizations. Developmental Psychology, 1972, 6, 201-209. Kessen, W. Research in psychological development of infants: An overview. Merrill-Palmer Quarterly, 1963,9, 83-94. Klebanoff, L. B. Parents of schizophrenic children: Parental attitudes of mothers of schizophrenic, brain-injured and retarded, and normal children. American Journal of Orthopsychiatry, 1959, 29,445-454. Korner, A. F. Mother-child interaction: One- or two-way street? Social Work, 1965, 10, 47-5 1 . Lewis, M. State as an infant-environment interaction: An analysis of mother-infant interaction as a function of sex. Merrill-Palmer Quarterly, 1972, 18,95-121. Lewis, M., & Lee-Painter, S . An interactional approach to the mother-infant dyad. In M. Lewis & L. A. Rosenblum (Eds.), The effect of the infant on its caregiver. New York: Wiley, 1974. Pp. 21-48. McKenzie, B., & Day, R. H. Operant learning of visual pattern discrimination in young infants. Journal of Experimental Child Psychology, 1971, 1 1 , 4 5 4 3 . Moss, H. A. Sex, age, and state as determinants of mother-infant interaction. Merrill-Palmer Quarterly, 1967, 13, 19-36. Osofsky, J. D., & Danzger, B. Relationships between neonatal characteristics and motherinfant interaction. Developmental Psychology, 1974, 10, 124-1 30. Osofsky, J. D., & O’Connell, E. J. Parent-child interaction: Daughters’ effects upon mothers’ and fathers’ behaviors. Developmental Psychology, 1972, 7, 157-168. Overton, W. F. The active organism in structuralism. Human Development, in press. Overton, W. F. On the assumptive basis of the nature-nurture controversy: Additive versus interactive components. Human Development, 1973, 16, 74-89. Overton, W. F., & Reese, H. W. Models of development: Methodological implications. In J. R. Nesselroade & H. W. Reese (Eds.), Life-span developmental psychology: Merhodological issues. New York: Academic Press, 1973. Pp. 65-86. Prechtl, H. F. R. The mother-child interaction in babies with minimal brain damage. In B. M. Foss (Ed.), Determinants of infant behaviour 11. London: Methuen (New York: Wiley), 1963. Pp. 53-59. Ramey, C. T., & Ourth, L. L. Delayed reinforcement and vocalization rates of infants. Child Development, 1971,42,291-297. Reese, H. W., & Overton, W. F. Models of development and theories of development. In L. R. Goulet & P. B. Baltes (Eds.), Life-span developmental psychology: Research and theory. New York: Academic Press, 1970. Pp. 115-145. Rheingold, H. L., Gewirtz, J. L., & Ross, H. W. Social conditioning of vocalizations in the infant. Journal of Comparative and Physiological Psychology, 1959,52,68-73. Richards, M. P. M. Social interaction in the first weeks of human life. Psychiatria, Neurologia, Neurochirurgia, 1971, 1 4 , 3 5 4 2 . Robson, K. S., & Moss, H. A. Patterns and determinants of maternal attachment. Journal of Pediatrics, 1970, 77,976-985. Rosenthal, R., & Jacobsen, L. Teachers’ expectancies: Determiners of pupils’ I.Q. gains. Psychological Reports, 1966, 19, 115-1 18. Routh, D. K. Conditioning of vocal response differentiation in infants. Developmental PsychologjJ, 1969, 1,219-226.
Mother-Iirfant Interaction and Its Study
163
Sears, R. R. A theoretical framework for personality and social behavior. American Psychologist, 1951,6,476-483. Sears, R. R., Maccoby, E. E., & Levin, H. Patterns of child rearing. Evanston, Ill.: Row, Peterson, 1957. Siegel, G. M. Adult verbal behavior with retarded children labeled as “high” or “low” in verbal ability. American Journal of Mental Deficiency, 1963,68,417424. Snow, C. Mothers’ speech to children learning language. Child Developmenf, 1972, 43, 549565. Stern, D. N. Mother and infant at play : The dyadic interaction involving facial, vocal, and gaze behaviors. In M. Lewis & L. A. Rosenblum (Eds.), The effect of the infant on its caregiver. New York: Wiley, 1974.Pp. 187-213. Thoman, E. B., Leiderman, P. H., & Olson, J. P. Neonate-mother interaction during breast-feeding. Developmental Psychology, 1972,6,1 l(tl18. Vuorenkoski, V., Wasz-Hockert, O., Koivisto, E., & Lind, J. The effect of cry stimulus on temperature of the lactating breast of primipara: A thermographic study. Experientia,
1969,25,1286-1287.
Weisberg, P. Social and nonsocial conditioning of infant vocalizations. Child Development,
1963,34,377-388.
Wenar, C., & Wenar, S . C. The short-term prospective model, the illusion of time, and the tabula rasa child. Child Development, 1963,34,697-708. Wolff, P. The natural history of crying and other vocalizations in early infancy. In B. M. Foss (Ed.), Determinants of infant behaviour IV. London: Methuen, 1969.Pp. 81-109. Yarrow, L. J . Dimensions of maternal care. Merrill-Palmer Quarterly, 1963,9,101-1 14. Yarrow, M. R., Waxler, C. Z., & Scott, P. M. Child effects on adult behavior. Developmental Psychology, 1971,5,300-311.