زمینه های روابط عاطفی منفی نوزاد به واکنش پذیری/حساسیت مراقبت

Abstract The aim of the study was to examine whether and how infant negative emotionality, depression in the caregiver, and marital emotional support are related to the caregiver’s reactivity/sensitivity. The central question concerned interaction effects between these risk factors. Furthermore, it was of special interest whether there were interaction effects of infant positive and negative emotionality on the reactivity/sensitivity of the caregiver. Participants were 101 mothers and their healthy 4-month-olds. The infant temperament characteristics were measured by parent report as well as in the laboratory. Caregiver behavior was observed during home visits. Hierarchical regression analyses were computed. Whereas there were no direct associations between either maternal depression or infant negative emotionality and maternal reactivity/sensitivity, the interaction of the two risk factors proved highly significant. Good marital support was directly linked to maternal reactivity/sensitivity. The relevance of positive infant emotionality could not be demonstrated.

. Introduction In the last two decades transactional models of development have replaced “nature-nurture” related perspectives, in which the development of personality was considered as either determined by environmental factors or genetic dispositions alone (Sameroff, 1975). There has been growing consensus that the course of development consists of a mutual process in which infant and caregiver constantly influence each other. The infant’s own contribution to his or her development, an aspect that has been neglected for a long time, has now been brought into focus: For example, Belskys process model of the determinants of parenting (Belsky, 1984) explicitly considers the impact of infant characteristics. In temperament research the notion that particular infant characteristics can have an effect upon the caregiver’s behavior has become a central issue Bates 1989, Rothbart 1989, Rothbart and Ahadi 1994, Rothbart and Posner 1985 and Thomas and Chess 1977. Furthermore, in the literature on child and adolescent psychiatry, difficult infant temperament is now regarded as a risk factor for the development of early behavioral problems as well as for the development of inadequate patterns in the initial infant-caregiver relationship Papousek and Papousek 1990, Rubin 1993, Rutter 1989, Warren et al 1997 and Zeanah et al 1997. In spite of these well established assumptions, there is still a lack of systematic knowledge regarding the postulated associations, especially for the time of early infancy (Seifer, Schiller, Sameroff, Resnick, & Riordan, 1996). More than 10 years ago, Crockenberg (1986) reviewed the existing research in an article that has since been frequently cited. Crockenberg was concerned with the question of whether and how infant temperamental characteristics (specifically negative emotionality or irritability) and patterns of parenting (specifically reactivity/sensitivity) work together and influence each other. Three possible models were presented: 1. Temperamental characteristics directly influence the caregiver’s responses. It is expected that different temperaments elicit distinct patterns of caregiving, e.g., infant irritability causes caregivers to respond in an insensitive manner to the infant’s needs. 2. Temperament influences caregiving only under specific conditions, i.e., it interacts with psychosocial attributes of the caregiver. This means that high irritability of an infant would not lead to decreased reactivity for all mothers, but only in the presence of particular attitudes, dispositions or current life circumstances. 3. Temperamental characteristics are not associated with caregiving at all. However, both classes of variables combine and interact in their effect on infant development. In the sense of “organismic specificity” this means that a difficult baby that happens to be paired with an unresponsive mother will have an unfavorable prognosis whereas the same lack of reactivity towards a child with an easy temperament would not have the same effect. Crockenberg (1986) reported 16 studies that dealt with associations between infant negative emotionality and maternal reactivity/sensitivity, nine of which yielded support for the notion that high irritability is linked to lower maternal reactivity. Seven, however, suggested the contrary, namely that mothers of highly irritable infants are more positive and show more commitment when interacting with their babies. The research following Crockenberg’s review continued to obtain inconsistent results. Two studies produced positive correlations between infant difficulty (assessed via parent questionnaire) and adequate maternal behavior. Washington, Minde, and Goldberg (1986) found that mothers who rated their six-month-old infants as temperamentally difficult showed more appropriate caregiving behaviors than mothers who did not perceive their child as difficult. Likewise, Zahr (1991) studied eight-month-old preterm infants and reported a significant association between mother-reported infant difficulty and more positive maternal behavior. In contrast to these findings, Spangler (1990), and Hagekull and Bohlin (1986), who studied a sample of healthy children in the first half of their second year, and Zahr (1991) with four-month-old infants in the above cited longitudinal study, as well as Clark, Hyde, Essex, and Klein (1997) in a large sample of four-month-old infants and their mothers found negative correlations. In these studies caregivers who perceived their infant as difficult exhibited a poorer quality in parenting behavior. Also, Feldman, Greenbaum, Mayes, and Erlich (1997) reported a negative relation between perceiving an infant as increasingly difficult in the course of three to nine months and the caregiver’s sensitivity when interacting with the baby. On the other hand, Hagekull, Bohlin, and Rydell (1997), who studied caregiver-infant dyads when the infants were 4, 10, and 15 months of age, did not find any associations between infant temperament and maternal sensitivity when dealing with the baby. All of the studies mentioned above used parent reports to assess infant temperament. Studies that assess temperament characteristics through behavior observation are less frequent and have revealed fewer apparent associations. In the work by Hagekull and Bohlin (1986) with 30 infants aged between 11.5 and 15 months, there was no connection between maternal and infant behavior. Butcher, Kalverboer, Minderaa, van Doormaal, and Ten Wolde (1993) studied preterms low on biological risks at three months corrected age. There was no significant relation between infant negativity as observed in interaction with the mother and the quality of maternal caregiving. Fish and Stifter (1993) did not obtain evidence for an association between infant fussing and crying in an ‘arm restraint procedure’ and maternal sensitivity during a free play session in the laboratory. Mangelsdorf, Gunnar, Kestenbaum, Lang, and Andreas (1990), and Seifer et al. (1996), however, were able to demonstrate moderate, but statistically significant positive relations between infant negative emotionality (at 6 and 9 months of age respectively) and the quality of maternal interaction behavior (especially reactivity/sensitivity). One reason for the inconsistency in the reported results probably lies in the fact that operational definitions of the temperament construct differ across studies. Defining infant temperament is one of the most essential problems in temperament research in early infancy (see Crockenberg 1986 and Seifer et al 1996). Parent report questionnaires are the most widely used instruments to assess infant temperament. Their validity, however, is limited to the extent that in parental judgments child characteristics are confounded with parental expectations and personality characteristics Affleck et al 1983, Bates and Bayles 1984, Crockenberg and Acredolo 1983, Mebert 1991, Mednick et al 1996, Pridham et al 1994 and Wolk et al 1992. Some authors maintain that parents’ perceptions of their children’s temperament are of particular predictive relevance just because of this subjective component Bates 1989, Mebert 1991 and Seifer et al 1996. However, if the parent questionnaire is the only instrument used, there is no way to differentiate between the “subjective component” and the “temperament component.” Regarding the more objective or unbiased methods of behavior observation, the assessment of infant temperament in the natural setting of caregiver-infant interaction is also problematic, because it is not independent of situational influences of the caregiver’s behavior (Crockenberg, 1986). The expression of positive or negative affect cannot exclusively be attributed to the infant’s temperamental characteristics, but is intermingled with influences from the caregiver’s behavior in that specific situation. Seifer et al. (1996) point to the related problem that many methods of rating caregiver’s reactivity/sensitivity also derive their measures from behavior the caregiver shows in reaction to the infant’s behavior, and as a consequence confound maternal reactivity with infant behavioral characteristics. These problems can be dealt with by using observations of infant behavior that occurs outside the situation of interacting with the caregiver, e.g., reactions to standard stimuli. On the other hand, the external validity of these measurements, usually obtained at only a single time point, is questionable. Goldsmith and Rothbart (1991) therefore propose using parent questionnaires concomitantly. This procedure provides a possible solution to the above described problems, because the necessary comparisons can be drawn between data from different sources. Apart from the problem of operational definition, a further potential explanation for the inconsistent results concerning the associations between infant temperament and maternal reactivity/sensitivity is the possibility of interactions between temperament and other variables (Crockenberg, 1986). Given the assumption—as Crockenberg proposes in her second model—that certain instances of infant temperament (e.g. high negative emotionality) correlate with a lack of reactivity in the caregiver only under conditions where further risk factors take effect. In this case—if we test only the linear links—we are likely to find either high, low, or zero correlations, depending on characteristics of the specific sample we are working with. Crockenberg (1986) therefore strongly recommended searching for interaction effects. Besides poor socioeconomic conditions, a depressive tendency in the caregiver and a lack of social support, especially in the marital relationship, are considered important risks for the development of inadequate parenting Belsky 1984, Crockenberg 1986, Field 1994, Papousek and Papousek 1990 and Zeanah et al 1997. A positive impact of social support by the spouse on the interactive behavior of mothers with their 4-month-olds was demonstrated by Crnic, Greenberg, Ragozin, Robinson, and Basham (1983), and Crnic, Greenberg, Robinson, Ragozin, and Basham (1984), as well as by Teti and Gelfand (1991). However, contrary results also have been reported. Clark et al. (1997) found social support by the partner and maternal reactivity/sensitivity to be inversely related. Fish and Stifter (1993) did not obtain any association between marital satisfaction and sensitivity in the caregiver-infant interaction. Recent findings from studies on the impact of depression on the early infant-caregiver relationship correspond to the described presumption of interaction effects: Several studies did not find any differences between mothers diagnosed as depressive and healthy mothers in their caregiving behavior Moore et al 1997, Goodman and Brumley 1990, Campbell et al 1995 and Seifer et al 1992. On the other hand, studies which have used high scores on symptom checklists for the definition of depression have consistently revealed a reduced reactivity and disturbed affective expressiveness in the ‘depressive mother’. These differing results led Campbell et al. (1995) to suggest that not maternal depression per se but only the cumulation of risks which were captured by the symptom lists is responsible for the reduction of the adequacy of the caregiving behavior. Accordingly, in a rather large and representative sample of primiparous women, Moore et al. (1997) were able to show that the diagnosis of depression does not necessarily go along with low sensitivity, high negative, or lack of positive affect in interaction contexts with the infant. These limitations were observed only if the depression that was measured postpartum later turned out to take a chronic course. The authors suppose that further circumstances appending to the depression might be responsible for the later course of the depression as well as for the lowered maternal sensitivity. Similarly, Teti and Gelfand (1991) conclude from their results that only if additional strains or demands accompany the depression, will inappropriate caregiving occur. In the following study we examined the question of whether and how infant negative emotionality, proneness to depression in the caregiver, and emotional support in the partnership are associated with the caregiver’s reactivity/sensitivity in interaction with the baby. In particular, we examined whether the temperament characteristic of negative emotionality in its relation to maternal reactivity interacts with the other two risk factors mentioned. In the work of many theorists negative emotionality/irritability is considered an important aspect of infant temperament, and compared to other components of infant temperament, its relation to parenting behavior has been studied quite often. Frequent infant crying and lack of soothability are supposed to be important determinants of perceived infant difficulty Bates 1980, Bates 1989 and Hubert and Wachs 1985. Belsky, Hsieh, and Crnic (1996) pointed out that emotionality should be treated as a two-dimensional construct, i.e., with a separate positive and negative dimension. Furthermore, they emphasized that positive emotionality plays an important role in parental interactive behavior in the context of negative child behavior (see Belsky, Fish, & Isabella, 1991). With reference to these considerations, we included the behavioral aspect of positive emotionality in our research questions. Specifically, we examined whether the association between infant negative emotionality and the reactivity/sensitivity of the caregiver was moderated by the degree of infant positive emotionality.

Results 3.1. Controlling for influences of infant sex and maternal educational level To control for the possible effects of infant sex and maternal educational level on the reactivity/sensitivity measure, a series of 3-way ANOVAs were conducted with infant sex, mother’s educational level, and each of the following variables (dichotomized by median split) as factors: negative and positive emotionality, hopelessness, ‘mutuality in partnership’. Main effects of infant sex (F(1,98) = 1.19, p = 0.28) and maternal educational level (F(1,98) = 1.47, p = 0.23) were not statistically significant. None of the 2-way and 3-way interaction effects reached statistical significance. 3.2. Intercorrelations of negative and positive emotionality scores To determine associations between the different positive and negative emotionality scores (correlations with the scores in the infant-caregiver interaction are given, too), we used Spearman rank correlations (to account for skewed distributions). Results are given in Table 1a and Table 1b. 1 Table 1a. Correlations witnin infant negative emotionality measures and Laboratory Assessment Caregiver-Infant Interaction Caregiver Report .38∗∗∗ .24∗ Laboratory Assessment .08 legend -Spearman rank correlation coefficients are listed. legend -Significance: ∗∗∗ : p < .001, ∗∗: p < .01; ∗ : p < .05. Table options Table 1b. Correlations within infant positive emotionality measures and Laboratory Assessment Caregiver-Infant Interaction Caregiver Report .32∗∗ .25∗ Laboratory Assessment .28∗∗ legend -Spearman rank correlation coefficients are listed. legend -Significance: ∗∗∗: p < .001, ∗∗ : p < .01; ∗ : p < .05. Table options Interrelations between caregiver reports and laboratory assessments were highly significant. Furthermore, caregiver report ratings were correlated with infant affect scores derived from the infant-caregiver interaction. With regard to associations between affect scores in the interaction with the caregiver and laboratory scores, only the intercorrelations of the positive emotionality scores proved significant. 3.3. Relations with maternal reactivity/sensitivity Mother’ perceptions of low infant negative emotionality (rSpearman = −0.22; p < .05) and a high degree of mutuality in the partnership (r = 0.21; p < .05) were associated with high maternal reactivity/sensitivity. No bivariate significant correlations were found between maternal reactivity/sensitivity and hopelessness, perceived positive infant emotionality and lab assessment scores. As proposed by Cohen and Cohen (1983), a series of hierarchical multiple regression analyses were performed to test the hypothesized interaction effects on the reactivity/sensitivity score. Negative and positive emotionality measures were split at the median because they were not normally distributed. To determine whether there were specific results for the different temperament assessment methods, regression analyses for lab assessment and caregiver report of infant negative emotionality were conducted separately (see Table 2 and Table 3). Table 2. Hierarchical regression analysis involving infant negative emotionality as reported by the mother with maternal reactivity/sensitivity as criterium variablelegend Variables R F Sign. p < FChange Sign. p < 1.: sex of infant, maternal educational level mutuality in partnership maternal depression negative emotionality (caregiver report) positive emotionality (caregiver report) .31 1.54 ns 2.: maternal depression × negative emot. (caregiver report) .45 3.22 .005 12.10 .001 2.: mutuality in partnership × neg. emot. (caregiver report) .34 1.62 ns 2.00 ns 2.: pos. emot. (careg. report) × neg. emot. (caregiver report) .32 1.42 ns 0.69 ns legend -ns: not statistically significant. Table options In each analysis, the two covariates (infant sex and maternal educational level) were entered first into the equation, followed by entry of the hopelessness score, the ‘mutuality in partnership’ score, and negative and positive emotionality as a block. Finally, to examine the interactions, the last entry in each of these analyses was a 2-way interactive term involving negative emotionality and one of the other factors (i.e., negative emotionality × maternal hopelessness, negative emotionality × mutuality in partnership, and negative emotionality × positive emotionality). Table 2 and Table 3document the results. After entry of the covariates and the factors, the equation did not reach statistical significance. Table 3. Hierarchical regression analysis involving negative emotionality as assessed in the laboratory with maternal reactivity/sensitivity as criterium variable legend Variables R F Sign. p < FChange Sign. p < 1.: sex of infant, maternal educational level mutuality in partnership maternal depression Negative emotionality (laboratory assessment) Positive emotionality (laboratory assessment) .30 1.33 ns 2.: maternal depression × neg. emot. (lab assessment) .43 2.51 .03 8.80 .005 2.: mutuality in partnership × neg. emot. (lab assessment) .36 1.69 ns 3.64 .10 2.: pos. emot. (lab assessm.) × neg. emot (lab assessm.) .31 1.15 ns 0.14 ns legend -ns: not statistically significant. Table options The interaction between negative emotionality in the lab assessment and maternal hopelessness accounted for a highly significant portion of the variance in maternal reactivity/sensitivity. Furthermore the interaction between the parent report measure and the mothers’ depression scores proved highly significant. The effects are illustrated in Fig. 1 and Fig. 2. Under the condition of high maternal depression and high negative emotionality of the infant, mothers’ reactivity/sensitivity was very low. This result was obtained for both caregiver report and negative emotionality as assessed in the laboratory, respectively. Furthermore, with regard to negative emotionality in the laboratory, in all other instances, i.e., if neither or only one of the two risk factors (high depression and high infant negative emotionality) was present, the caregiver exhibited comparatively higher reactivity/sensitivity than in the presence of both risk factors. The interaction effect of maternal depression and perceived high infant negativity revealed a different result: mothers who perceived their infants as low in negative emotionality but who describe themselves as depressed were observed to be relatively high in reactivity/sensitivity. Mothers who were depressed and saw their infants as high in negative emotionality tended to display relatively low reactivity/sensitivity. Interaction of negative emotionality (caregiver report) and maternal depression ... Fig. 1. Interaction of negative emotionality (caregiver report) and maternal depression on maternal reactivity/sensitivity. Figure options Interaction of negative emotionality (laboratory assessment) and maternal ... Fig. 2. Interaction of negative emotionality (laboratory assessment) and maternal depression on maternal reactivity/sensitivity. Figure options The interaction between laboratory negative emotionality and mutuality in partnership was marginally related to reactivity/sensitivity, but the whole equation was not significant. The corresponding effect involving caregiver-perceived negative emotionality as well as the interactions between positive and negative emotionality scores did not reach statistical significance. To answer the question of whether the different methods of assessing temperament account for common or specific variance in maternal reactivity/sensitivity, and to control for possible influences of the infant’s affective performance on the ratings of caregiver reactivity/sensitivity, a final set of hierarchical regression analyses was performed (see Table 4). Together with the other factors and covariates, infant positive and negative affect scores—as observed in coincidence with maternal reactivity/sensitivity during the home observations—were entered in a first step, followed by the entry of the interaction-term involving negative emotionality (as measured by one method) and depression. In a last step, the interaction between negative emotionality (as measured by the other method) and depression was entered. Table 4. Hierarchical regression analysis with maternal reactivity/sensitivity as criterium variable: control for assessment methodslegend Variables R F Sign. p < FChange Sign. p < 1a, b: sex of infant, maternal educational level maternal depression negative emotionality (mother-infant interaction) positive emotionality (mother-infant interaction) negative emotionality (caregiver report) positive emotionality (caregiver report) negative emotionality (laboratory assessment) positive emotionality (laboratory assessment) .36 1.31 ns 2a: maternal depression × neg. emot. (lab assessment) .45 1.99 .05 7.20 .01 3a: maternal depression × negative emot. (caregiver report) .51 2.41 .02 5.46 .03 2b: maternal depression × negative emot. (caregiver report) .48 2.30 .02 9.91 .003 3b: maternal depression × negative emot. (lab assessment) .51 2.41 .02 2.92 .09 legend -ns: not statistically significant. Table options After controlling for the other methods, the amount of variance in reactivity/sensitivity that was explained by the interaction effects decreased. However, the interaction between negative emotionality as reported by the mother and depressive tendency was still significantly related to maternal reactivity/sensitivity.