Abstract
This study examined whether, as mothers’ depressive symptoms increase, their expressions of negative emotion to children increasingly reflect aversion sensitivity and motivation to minimize ongoing stress or discomfort. In multiple interactions over 2 years, negative affect expressed by 319 mothers and their children was observed across variations in mothers’ depressive symptoms, the aversiveness of children’s immediate behavior, and observed differences in children’s general negative reactivity. As expected, depressive symptoms predicted reduced maternal negative reactivity when child behavior was low in aversiveness, particularly with children who were high in negative reactivity. Depressive symptoms predicted high negative reactivity and steep increases in negative reactivity as the aversiveness of child behavior increased, particularly when high and continued aversiveness from the child was expected (i.e., children were high in negative reactivity). The findings are consistent with the proposal that deficits in parenting competence as depressive symptoms increase reflect aversion sensitivity and motivation to avoid conflict and suppress children’s aversive behavior.
Depressive symptoms are common in mothers, which places children at risk for diverse developmental problems (Cummings & Davies, 1994; Downey & Coyne, 1990). These symptoms predict developmental risk, in part, because they promote negative, intrusive, and unresponsive parenting (Goodman et al., 2011; Lovejoy, Graczyk, O’Hare, & Neuman, 2000). An important proposal for why they do this is that depressive symptoms increase parents’ affective sensitivity to aversive inputs from children, which results in parenting often oriented not toward meeting children’s needs, but toward minimizing the depressed parents’ heightened negative affect (Dix & Yan, 2014; Lahey, Conger, Atkenson, & Treiber, 1984; Lorber, 2012; see Lewinsohn, Lobitz, & Wilson, 1973). The research reported here tested this proposal by examining whether expressions of negative emotion by mothers with depressive symptoms reflect attempts to minimize their ongoing distress and, furthermore, whether this explains why depressive symptoms at some moments predict highly negative reactions to children and, paradoxically, at other moments predict flat, unresponsive reactions.
It is commonplace to propose that depressive symptoms alter social behavior because they influence emotions and related motivational states. Depressive symptoms, even in nonclinical samples, are associated with impatience, irritability, stress, and altered hypothalamic-pituitary-adrenal axis functioning (Plotsky, Owens, & Nemeroff, 1998; Pruessner, Hellhammer, Pruessner, & Lupien, 2003). In the parenting literature, two contrasting proposals have been advanced for how affective sensitivity to aversive input might influence the behavior of depressed parents.
First, aversion sensitivity may promote negative actions and displays of negative emotion as a means of suppressing child behaviors that activate the depressed mother’s distress. It has been argued that parents have set points or upper limits of tolerance that regulate when they respond to reduce children’s aversive behaviors (Bell & Chapman, 1986; Lahey et al., 1984; see Rottenberg, Gross, & Gotlib, 2005). As depressive symptoms increase, these set points may decrease, which leads depressed mothers to become upset and to resist child behaviors that are less aversive than those that upset and elicit resistance from nondepressed mothers. The resulting negativity and frequent opposition to children, in turn, predict dysfunctional family interaction and diverse developmental problems in children (Dix, 1991; Lorber & O’Leary, 2005; Patterson, 2002).
Support for this proposal is found primarily in studies demonstrating simply that mothers with depressive symptoms are negatively reactive, that is, intrusive, controlling, and likely to express negative affect (Dix & Meunier, 2009; Field, Healy, Goldstein, & Guthertz, 1990; Lovejoy et al., 2000). Yet because researchers have not specified the child behaviors to which depressed mothers are reacting, such findings cannot verify that mothers’ negative, intrusive behaviors reflect low thresholds for activating negative emotion and motivation to minimize that emotion. These behaviors could reflect, instead, the difficult characteristics of children of depressed mothers (Richters, 1992), tendencies for strong negative emotion to get expressed directly in behavior regardless of its anticipated consequences (Baumeister, Vohs, DeWall, & Zhang, 2007), or a pervasive negativity that neither depends on, nor reflects attempts to control, aversive input from children.
Second, aversion sensitivity may lead not to negative, intrusive, and excessive reactivity to children, but instead to low reactivity, unresponsiveness, and expressions of flat affect. This proposal stems from views of depressive symptomatology that emphasize its tendency to promote inhibition (Rottenberg et al., 2005) and conflict avoidance (Trew, 2011). Research demonstrates that mothers with depressive symptoms can be disengaged, unresponsive, and unemotional (Burke, 2003; Field, 1995; Puckering, 1989). Often they withdraw from difficult child behaviors, rather than act to suppress them (Goodman & Brumley, 1990; Kochanska, Kuczynski, Radke-Yarrow, & Welsh, 1987).
Although seemingly inconsistent with tendencies toward hyperreactivity, low reactivity may also reflect depressed mothers’ attempts to minimize their distress. Describing emotion-feedback theory, Baumeister et al. (2007) write, “when people anticipate negative emotional outcomes from taking action, they may choose not to act and hence to leave things as they are” (p. 193). Because children can be demanding and uncooperative, not engaging them is often less stressful than correcting or resisting their aversive behavior (Kochanska et al., 1987). In such instances, not expressing negative affect may reflect conflict avoidance. Ignoring children, in fact, predicts low reciprocal negative affect from them (Dowdney & Pickles, 1991). Consistent with this analysis, research demonstrates that parental distress predicts both lax and overreactive discipline among parents with depressive symptoms (Leung & Slep, 2006). Parents who display lax discipline also tend to display overreactive discipline, and both lax and overreactive discipline are related to negative emotion in parents (Leung & Slep, 2006; Lorber & O’Leary, 2005).
Yet research on depressive symptoms and low reactivity has not examined the child behaviors that depressed mothers choose not to engage. Thus, whether low negative reactivity reflects depressed mothers’ attempts to regulate their negative emotion is untested. Low negative reactivity might reflect, instead, the unresponsiveness that characterizes children of mothers with depressive symptoms (Dix, Meunier, Lusk, & Perfect, 2012; Dix, Stewart, Gershoff, & Day, 2007), blunting of the depressed mother’s experience of emotion (Rottenberg, Kasch, Gross, & Gotlib, 2002), or a pervasive inhibition or approach-motivation disturbance (Kring & Bachorowski, 1999) that neither depends on, nor reflects attempts to control, aversive input from children (Downey & Coyne, 1990).
If low parenting competence reflects depressed mothers’ attempts to regulate their negative emotions, their negativity should depend on their interactive context. Learning (Patterson, 1982) and emotion-feedback (Baumeister et al., 2007) theories predict that mothers express negative emotion not simply because they experience it, but because they expect that expressing it will reduce or exacerbate their distress. Research demonstrates that expectations about future aversiveness often regulate expressions of negative emotion more than does the intensity of the emotion itself (Baumeister et al., 2007; see Parkinson, 2005). When a depressed mother expects that expressing negative emotion will activate children’s resistance, she should suppress that expression; when she expects that expression to reduce the child behavior that is causing her distress, she should express that emotion.
These expectations should depend on at least two factors. The first is the aversiveness of the child behavior to which the mother is responding. When immediate child behavior is minimally aversive, the depressed mother may minimize her distress by suppressing her negative expression and impulse to resist the child. As long as the child’s behavior is below her set point, not reacting maintains tolerable levels of distress by minimizing children’s reciprocal negativity (Bell & Chapman, 1986; Dowdney & Pickles, 1991). In contrast, when children’s immediate behavior is above the mother’s set point and thus difficult to tolerate, her subsequent distress may be minimized if she acts to eliminate that behavior, that is, if she disapproves of or resists it.
Second, whether depressed mothers express their negative affect should depend on individual differences in children that determine whether mothers expect children to reciprocate negativity (Bell & Chapman, 1986). This expectation may reflect both the mother’s conception of her child’s general negativity (i.e., trait negativity) and her observation of the child’s negativity during the current interaction (i.e., mood negativity). Children often reciprocate mothers’ negative affect (Dowdney & Pickles, 1991), and some do so more than others (e.g., Patterson, 1982). A mother who expects her child to reciprocate should assume that expressing negative affect will increase, not reduce, her distress.
Worry about the negativity of highly reactive children should vary as a function of the aversiveness of their immediate behavior. At a given moment, if a highly reactive child is only mildly aversive—below the mother’s tolerance set point—she would minimize her distress by suppressing expression of her negative affect. This minimizes reciprocal negativity anticipated from children who are high in negative reactivity, but not from children who are low in negative reactivity. When the aversiveness of children’s immediate behavior is above the depressed mother’s tolerance threshold, she should be motivated to express negative affect to reduce that behavior, and this motivation should be stronger with children who are high, rather than low, in negative reactivity. When displayed by highly reactive children, difficult-to-tolerate behavior signals that an extended aversive exchange may occur and that a particularly negative reaction may be required to overcome the child’s resistance. In contrast to when children’s immediate behavior is tolerable (low aversive), when it is significantly aversive, mothers with high levels of depressive symptoms may minimize their distress by reacting more, not less, negatively to children who are high, rather than low, in reactivity.
To evaluate these proposals, we tested three predictions. Hypothesis 1 was that when children’s immediate behavior is low aversive, depressive symptoms will reduce mothers’ expressions of negative affect. Hypothesis 2 was that as children’s immediate behavior becomes increasingly aversive, depressive symptoms will heighten mothers’ expressions of negative affect. Hypothesis 3 was that children’s general tendency to be negatively reactive will moderate relations between mothers’ depressive symptoms and their expressions of negative affect. When children’s immediate behavior is low aversive, depressive symptoms will predict low expression of negative affect with children high in negative reactivity more than with children low in negative reactivity. As children’s immediate behaviors increase in aversiveness, mothers’ depressive symptoms will predict increases in expressions of negative affect with children high in negative reactivity more than with children low in negative reactivity.
Method
Participants
To ensure a sample with high levels of depressive symptoms, we studied 319 recently divorced or divorcing mothers and their 5- to 11-year-old children (mean age = 7.8 years, SD = 2.0). Mothers were identified from court records and assessed initially within 120 days of filing for divorce; 25% were divorced by the baseline assessment. Boys and girls were about equally represented (52% female, 48% male). Mothers’ ages ranged from 21 to 53 years (median = 36.8). On average, they attended but did not finish college (38% of the sample; 9.4% had less than a high school diploma, 1.3% held doctoral degrees).
Procedure
Data were collected during home visits over a 2-year period. Participants were evaluated at baseline, with subsequent assessments occurring when changes in repartnering occurred (i.e., a new relationship, cohabitation, engagement, remarriage). If no such transition occurred, assessments were done at 12 and 24 months. Nineteen percent of participants completed four to six assessments, 58% completed three assessments, 12% completed two assessments, and 11% completed only the baseline assessment. As Singer and Willett (2003) note, multilevel models such as those examined here can accommodate variations in the number and spacing of observations across participants (see the Results section for details of the three models).
Measures
Mothers’ depressive symptoms
At all assessments, mothers completed the Center for Epidemiologic Studies Depression Scale (CES-D; Radloff, 1977). This scale consists of 20 items that assess symptoms over the past week (e.g., “I had crying spells,” “I felt lonely,” “I felt sad”). The scale has good internal consistency, test-retest reliability, and construct validity. Because they were adapting to divorce, mothers in this sample had elevated symptomatology. Thirty-four percent scored above the clinical cutoff (≥ 16), higher than the 20% typically observed. Sixty percent were above the clinical cutoff at some point in the study.
Mother-child interaction
Each mother identified an area of disagreement with her child and a divorce-related concern her child had expressed previously. During a 12-min interaction, they discussed these topics. Interactions were coded using the Family and Peer Process Code (Stubbs, Crosby, Forgatch, & Capaldi, 1998). This consists of 24 content codes (e.g., tease, attack, advise, command, coerce, agree, comply, refuse, self-disclose). For each content code, one of six emotions is also coded: happy, caring, neutral, sad, distressed, and aversive. Assessed on 15% of interactions, reliability was good. Kappas averaged .80 (with 92% agreement); 85% were above .70, 91% above .60, and 96% above .50. A new code was recorded whenever content or affect changed; thus, a speaker could have multiple codes in sequence. Consequently, talk turns were created by selecting the most negative content and affect that occurred during a sequence.
Aversiveness of children’s behaviors
An exhaustive set of 144 child behaviors was generated by combining each of the six affect codes with each of the 24 content codes. To ensure adequate data for each behavior, we eliminated behaviors with base rates below the median of 28 (< 1.6% of turns). This yielded a final set of 47 child behaviors that included 170,357 talk turns, on average 187 per dyad per interaction. Child behaviors were ranked from least to most aversive based on the probability that across the entire sample, mothers’ expressed negative affect in the next turn.
Mothers’ and children’s negative reactivity
To measure tendencies to respond with negative affect, we combined aversive and distress codes to form a negative-affect code. Sadness was excluded because it is often unresponsive to immediate inputs and linked less to signals about the impact of partners’ immediate actions (Frijda, 1986; Horstmann, 2003). Mothers’ negative-reactivity scores were computed for each child behavior; these scores were the proportion of times the mother responded to a particular behavior with negative affect during the next turn. Children’s general negative reactivity, an individual-difference measure, was the proportion of maternal expressions of negative affect to which children responded with negative affect during the next turn.
Results
We examined mothers’ rates of negative affect in response to each of the 47 child behaviors. To determine the extent to which mothers’ negative reactivity changed as the aversiveness of children’s immediate behaviors went from low to high, we generated individual intercepts and slopes for each mother at each assessment. These were calculated by regressing the probability of each mother reacting with negative affect to each child behavior onto the expected probability of reacting with negative affect to each behavior (i.e., the probability for the overall sample). This yielded two dependent variables, a reactivity intercept (the conditional probability of maternal negative affect when children’s immediate behavior was low aversive) and a reactivity slope (the rate at which that conditional probability increased as children’s behavior varied from low to high aversive). Thus, tests of intercepts examined predictors of mothers’ negative reactivity when children’s immediate behavior was low aversive; tests of slopes examined predictors of the rate at which negative reactivity increased as children’s immediate behavior varied from low to high aversive (Fig. 1 displays an example).
Relation between the observed probability of mother’s negative reactivity and the expected probability, as indexed by the aversiveness of children’s immediate behavior. The graph shows the average across the entire sample, as well as data points and a best-fitting regression line for an individual mother as an illustrative example.
Three multilevel models were used to examine these repeated measures data. Each assessed predictors of the intercepts and of the linear trajectories of mothers’ negative reactivity across variations in the aversiveness of children’s immediate behavior. The first two models assessed, in separate analyses, mothers’ depressive symptoms (Model 1) and children’s general negative reactivity (Model 2) as predictors. The full model, Model 3, assessed depressive symptoms, children’s general negative reactivity, and their interaction. Full-information maximum-likelihood estimation was used to handle missing data. Because control variables (child sex, child age, mother’s age, length of separation) did not predict mothers’ reactivity or measures predicting it, these variables were excluded from final models. Analyses of child age and sex as moderators revealed no clear moderating effects. Findings are displayed in Table 1 and Figures 2 and 3.
Results of Models Predicting Mothers’ Negative Reactivity Intercepts and Slopes Across Variations in the Aversiveness of Children’s Immediate Behavior
For fixed effects, unstandardized coefficients are reported.
Results from Model 3: interaction of child general negative reactivity and mother’s depressive symptoms in predicting mothers’ negative reactivity intercept. Low = 1 SE below the average, high = 1 SE above the average.
Results from Model 3: interaction of child general negative reactivity and mother’s depressive symptoms in predicting mothers’ negative reactivity slope as children’s immediate aversiveness increased. Low = 1 SE below the average, high = 1 SE above the average.
Analyses of mothers’ reactivity intercepts supported Hypothesis 1: Whether evaluated alone or in the full model, mothers’ depressive symptoms predicted less expression of negative affect when children’s immediate behavior was low in aversiveness. In contrast, whether evaluated alone or in the full model, children’s general negative reactivity did not predict mothers’ reactivity when children’s behavior was low in aversiveness. Consistent with Hypothesis 3, the full model revealed that children’s general negative reactivity moderated relations between mothers’ depressive symptoms and mothers’ reactivity intercepts. As predicted, simple slopes from this interaction (see Fig. 2) demonstrated that, when the immediate behavior of highly or moderately reactive children was low in aversiveness, mothers’ negative reactivity declined as their depressive symptoms increased (β = −0.0003, p < .001; β = −0.0001, p < .001, respectively). In contrast, when the immediate behavior of children who were low in reactivity was low in aversiveness, mothers’ depressive symptoms were unrelated to mothers’ negative reactivity (β = 0, p > .50). Note that when children’s immediate behavior was relatively nonaversive, mothers with high levels of depressive symptoms were less negatively reactive with children who were high rather than low in negative reactivity, whereas mothers with low levels of depressive symptoms were more negatively reactive with children who were high rather than low in negative reactivity.
Analyses of mothers’ reactivity slopes supported Hypothesis 2. Whether evaluated alone or in the full model, mothers’ depressive symptoms predicted a faster rate of increase in maternal negative reactivity as children’s behavior varied from low to high aversive. Children’s general negative reactivity predicted faster increases in mothers’ negative reactivity in Model 1, but not in the full model. Consistent with Hypothesis 3, the full model revealed that children’s general negative reactivity moderated relations between mothers’ depressive symptoms and mothers’ reactivity slopes. Simple slopes from this interaction (see Fig. 3) demonstrated that, as the behavior of highly or moderately reactive children increased in immediate aversiveness, mothers’ negative reactivity increased more quickly as their depressive symptoms increased (β = 0.054, p < .001; β = 0.028, p < .001, respectively). In contrast, as the behavior of children low in negative reactivity increased in immediate aversiveness, mothers’ negative reactivity did not increase more quickly as their depressive symptoms increased (β = 0.009, p > .15). Note that mothers with high levels of depressive symptoms increased their negative reactivity more quickly with children who were high rather than low in negative reactivity, whereas mothers with low levels of depressive symptoms increased their negative reactivity more quickly with children low rather than high in negative reactivity.
Discussion
These results, based on within-mother changes in depressive symptoms over 2 years, are consistent with an emotion-regulation perspective on the negative reactivity of mothers with depressive symptoms. When mothers’ depressive symptoms were high, they expressed (a) particularly low negative affect when the aversiveness of children’s immediate behavior was low, a tendency that increased as children’s general negative reactivity increased, and (b) particularly high and accelerating levels of negative affect when the aversiveness of children’s immediate behavior varied from low to high, a tendency that also increased as children’s general negative reactivity increased. This pattern supports the proposal that aversion sensitivity leads mothers with depressive symptoms to experience increased motivation to suppress their expressions of negative affect when children’s behavior can be tolerated—thus avoiding children’s reciprocal negativity—and increased motivation to express negative affect when children’s immediate behavior is significantly aversive—thus reducing immediate aversive stimulation. The data clarify why depressive symptoms predict both high and low negative reactivity and specify affective processes that may contribute to the problematic parenting and child behavior problems linked to mothers’ depressive symptoms.
Depressive symptoms and low negative reactivity
When responding to low-aversive child behavior, mothers with depressive symptoms were low in negative reactivity. This tendency was pronounced with highly reactive children. At high levels, depressive symptoms appeared to suppress mothers’ expressions of negative emotion in response to the low-aversive behavior of those children most likely to react negatively to that expression (i.e., children who are high in negative reactivity), but not with children least likely to react negatively (i.e., children who are low in negative reactivity). In contrast, when their depressive symptoms were low, mothers expressed high, not low, negative emotion in response to the low-aversive behavior of highly reactive children, potentially to socialize, or communicate disapproval of, mildly problematic behaviors. These findings are consistent with learning (Patterson, 1982) and emotion-feedback (Baumeister et al., 2007) theories; they stress that expression of emotion is regulated by its anticipated consequences. They suggest that, when children’s behavior is low aversive or tolerable, depressive symptoms increase mothers’ conflict avoidance. Researchers stress that such avoidance contributes to lax discipline and reduced parental teaching and socialization (Dix, 1991; Leung & Slep, 2006; Patterson, 1982).
Researchers offer other explanations for why mothers with depressive symptoms are unresponsive or low in negative reactivity. One explanation is that depressive symptoms lead mothers to prefer low-effort responses (Downey & Coyne, 1990; Kochanska et al., 1987). Yet depressive symptoms were linked here to reduced expression of emotion, which appears to require little physical or mental effort. More important, no additional effort is required to express negative affect to children who are high in negative reactivity than to children who are low in negative reactivity, yet depressive symptoms predicted low maternal reactivity principally with children who were high in negative reactivity. Thus, effort-based proposals seem unable to account for the low negative reactivity in this community sample. A second explanation for low affective responsiveness is that depressive symptoms blunt the experience of emotion (Rottenberg et al., 2002). Yet, in the present study, when depressive symptoms were high, mothers displayed low negative reactivity for only a modest portion of child behaviors. As children’s immediate behavior became aversive, depressive symptoms quickly predicted high, not low, negative reactivity. Thus, blunting of negative affect would have to have occurred only for that range of behaviors least likely to have aroused it (i.e., low-aversive behaviors). Blunting of emotional experience therefore seems a poor explanation for the low reactivity observed in this community sample.
Depressive symptoms and high negative reactivity
When mothers were responding to moderately or highly aversive behavior, their depressive symptoms predicted greater negative reactivity and steeper increases in reactivity as children’s immediate aversiveness increased. For these aversive behaviors, mothers with high and low levels of depressive symptoms differed little when they interacted with children who were generally low in negative reactivity, but differed significantly when they interacted with children who were generally high in negative reactivity.
This pattern is consistent with theories that emphasize that expression of emotion is regulated by its anticipated consequences (Baumeister et al., 2007; Patterson, 1982). The pattern is consistent as well with the proposal that, as children’s behavior comes to exceed the depressed mother’s low threshold for tolerating aversive input—and primarily when high and continued aversiveness from children is expected—higher levels of depressive symptoms predict greater expression of negative affect as a means of discouraging children’s immediate aversive behavior. Strong motivation to reduce children’s aversive behavior may be one mechanism underlying the negative and intrusive parenting associated with depressive symptoms (Lovejoy et al., 2000). Strong negative emotion in mothers is associated with dysfunctional family processes and problematic child development (Dix, 1991; Leung & Slep, 2006). It predicts harsh, overreactive discipline (Leung & Slep, 2006; Lorber & O’Leary, 2005), poor child adjustment (Dix, 1991; Maccoby & Martin, 1983), and coercive parent-child exchanges associated with child abuse and child-conduct problems (Patterson, 1982, 2002). High sensitivity to the aversive inputs from children may underlie interactional and developmental problems associated with depressive symptoms in mothers.
Other perspectives imply that heightened negative expression among mothers with depressive symptoms might reflect not their attempts to regulate their emotional states, but their inability to inhibit expression of intensely felt emotions. Yet direct links between depressed mothers’ felt and expressed emotion cannot explain why at statistically comparable levels of aversiveness, children low in negative reactivity elicit greater expression of negative affect from depressed mothers than do children high in negative reactivity. Furthermore, the idea that felt emotion is expressed more as depressive symptoms increase cannot explain why, when children’s immediate aversiveness is low, depressive symptoms predict less, not more, negative affect with highly reactive children. Relative to children who are low in negative reactivity, children who are high in negative reactivity should arouse more, not less, anxiety or irritation in mothers, which if directly expressed would result in greater, not less, negative reactivity. Thus, a motive to minimize subsequent distress appears to be necessary to explain the contextual variations in affective displays observed in this community sample.
Summary and Conclusion
This study suggests that during periods of high family change and conflict, depressive reactions in mothers set in motion patterns of emotional expression that contribute to poorly regulated mother-child interactions. Consistent with feedback theories of emotion (Baumeister et al., 2007) and behavioral approaches to coercive family process (Patterson, 1982), these patterns appear to result from the tendency of depressive symptoms to increase aversion sensitivity and motivation to minimize ongoing distress. Research is needed that addresses the mechanisms underlying these processes. In particular, the complex roles played by depression-related changes in internal cognitive and neuroendocrine factors, on the one hand, and in external levels of family conflict and negative emotion, on the other, need to be addressed. Furthermore, although maternal negativity and low responsiveness have been linked to mothers’ depressive symptoms in diverse samples, aversion sensitivity with children may be particularly high during divorce; research with other samples is needed to determine how general the processes observed here are. If similar results occur in other samples, depression-related aversion sensitivity and motivation to reduce discomfort may prove significant for understanding the strained social relationships, not only of mothers and children, but of individuals with depressive symptoms generally.
Footnotes
Declaration of Conflicting Interests
The authors declared that they had no conflicts of interest with respect to their authorship or the publication of this article.
Funding
This research was supported by Grant No. 1-5 R01 HD41463-01A1 from the National Institute of Child Health and Human Development to Edward R. Anderson and Shannon M. Greene.