Emotion 2013, Vol. 13, No. 6, 1160 –1172

© 2013 American Psychological Association 1528-3542/13/$12.00 DOI: 10.1037/a0033923

Role of Adolescent and Maternal Depressive Symptoms on Transactional Emotion Recognition: Context and State Affect Matter Aaron M. Luebbe, Lauren M. Fussner, and Elizabeth J. Kiel

Martha C. Early and Debora J. Bell University of Missouri–Columbia

This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Miami University

Depressive symptomatology is associated with impaired recognition of emotion. Previous investigations have predominantly focused on emotion recognition of static facial expressions neglecting the influence of social interaction and critical contextual factors. In the current study, we investigated how youth and maternal symptoms of depression may be associated with emotion recognition biases during familial interactions across distinct contextual settings. Further, we explored if an individual’s current emotional state may account for youth and maternal emotion recognition biases. Mother–adolescent dyads (N ⫽ 128) completed measures of depressive symptomatology and participated in three family interactions, each designed to elicit distinct emotions. Mothers and youth completed state affect ratings pertaining to self and other at the conclusion of each interaction task. Using multiple regression, depressive symptoms in both mothers and adolescents were associated with biased recognition of both positive affect (i.e., happy, excited) and negative affect (i.e., sadness, anger, frustration); however, this bias emerged primarily in contexts with a less strong emotional signal. Using actor–partner interdependence models, results suggested that youth’s own state affect accounted for depression-related biases in their recognition of maternal affect. State affect did not function similarly in explaining depression-related biases for maternal recognition of adolescent emotion. Together these findings suggest a similar negative bias in emotion recognition associated with depressive symptoms in both adolescents and mothers in real-life situations, albeit potentially driven by different mechanisms. Keywords: depression, adolescence, emotion recognition, maternal depression Supplemental materials: http://dx.doi.org/10.1037/a0033923.supp

toms were associated with a negative bias in the perception of each others’ emotions across various interactional contexts that differed in the strength of the emotional signal present. Further, we tested whether one’s own state affect during the interaction accounted for recognition biases. This study advances the literature by (a) examining both maternal and youth symptomatology, (b) measuring emotion recognition in “real-time,” and (c) considering recognition across multiple contexts. Emotion recognition may be especially critical in mother– child relationships. Clearly, it is imperative for mothers to recognize their children’s emotions in order to provide adequate care and support for their child. Further, children who are more accurate in reading their mothers’ emotions may elicit care and be able to develop clear expectations regarding parental emotional responses to behavior. Adolescence, in particular, is a crucial period of development to explore emotion recognition. Family dynamics shift significantly during early adolescence, with an overall escalation in parent– child conflict (Steinberg & Morris, 2001). Also, although adolescents spend less time at home interacting with family members, parents continue to remain a salient influence throughout the teenage years (Steinberg & Morris, 2001). Accurately interpreting emotional cues is a critical element of successful communication and may potentially lead to reduced family conflict and lowered levels of psychological maladjustment.

Recognizing and interpreting emotional cues during interpersonal interactions are essential features of successful communication and the maintenance of social relationships. Through emotional expression and subsequent recognition, individuals convey social messages (e.g., sadness as a call for support) and reinforce one another’s behavior (Keltner & Kring, 1989). The majority of research investigating emotion recognition has focused on emotions conveyed through pictures of facial expressions and neglected to explore the influence of individual differences (e.g., depressive symptoms; one’s own emotion in a situation) and contextual factors (e.g., emotional signal of the context) during in-the-moment social interactions. The current study tested whether youths’ and mothers’ depressive symp-

Aaron M. Luebbe, Department of Psychology, Miami University; Lauren M. Fussner, Department of Psychology, Miami University; Elizabeth J. Kiel, Department of Psychology, Miami University; Martha C. Early, Department of Psychological Sciences, University of Missouri–Columbia; and Debora J. Bell, Department of Psychological Sciences, University of Missouri–Columbia. Funding for this study was provided by a University of Missouri Research Council (URC-07-081) grant awarded to Debora J. Bell. We thank the youth and parents who participated in this study. Correspondence concerning this article should be addressed to Aaron M. Luebbe, Miami University, Department of Psychology, 90 North Patterson Avenue, Oxford, OH 45056. E-mail: [email protected] 1160

DEPRESSION AND EMOTION RECOGNITION IN CONTEXT

This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

Depressive Symptomatology and Emotion Recognition Broadly, depression symptoms are associated with biased interpretation of social information (Gotlib & Hammen, 1992) including the emotional signals displayed by others (Feinberg, Rifkin, Schaffer, & Walker, 1986). In a recent review, Bistricky, Ingram, and Atchley (2011) concluded that there was general support that depressed adults and those at-risk for depression typically demonstrate biased information processing of facial affect. Recognition biases seem to emerge most often when emotions are ambiguous or judgments about intensity are needed. Studies typically fail to find associations between adult depressive symptomatology and errors in simple identification of unambiguous emotional displays, especially in community or less distressed clinical samples (e.g., Beevers, Wells, Ellis, & Fischer, 2009; Frewen & Dozois, 2005). Yet, several studies show that adults with depression attribute sadness to neutral faces and neutral emotional states to happy faces (Gur et al., 1991; Leppänen, Milders, Bell, Terriere, & Hietanen, 2004) as well as require greater intensity of facial expressions than control participants to correctly identify faces displaying positive emotions (Joormann & Gotlib, 2006). Other aspects of emotion processing also seem to be impaired for depressed adults including selective attention toward negative faces and away from positive faces and biased memory for faces displaying negative emotions (and sadness in particular; see Bistricky et al., 2011, for a review). Although considerably outnumbered by studies of adults, studies with child or adolescent samples suggest similar biases in emotion recognition tied to depression. For instance, adolescent daughters of depressed mothers were more likely than controls to demonstrate a bias for identification of pictures of sad, but not angry, faces (Joormann, Gilbert, & Gotlib, 2010). Similarly, in middle childhood, depressive symptoms have been associated with perceptions of greater anger, but not sadness, in pictures of lowintensity faces (van Beek & Dubas, 2008). Overall, existing evidence suggests that individuals with elevated levels of depression, regardless of whether an adult or child, are likely to exhibit a negative emotion recognition bias such that they “underperceive” positive emotional displays and “overperceive” negative displays. One limitation of extant research, however, is that it has typically investigated emotion recognition using pictures of facial expressions depicting a single emotion (e.g., Ekman and Friesen’s Pictures of Facial Affect; Eckman, Matsumoto, & Friesen, 1997) or varying intensities of a given emotion. Significantly less research has explored emotion recognition during in-the-moment social interactions, despite the fact that most emotion recognition in daily life happens during in-person interactions in which multiple factors (e.g., posture, tone of voice, context) influence judgments of emotion (Barrett, Mesquita, & Gendron, 2011). Emotional expression during social encounters is neither static nor unambiguous like pictorial stimuli may be. Also, an important limitation of many studies of emotion recognition is that the pictures of faces used as stimuli typically have no personal relevance. Recognizing the emotions of one’s parent or child may be influenced by past learning or various expectancies for behavior developed over the course of repeated interactions. As such, what past studies may have gained using rigorous experimental control may have come at the expense of some generalizability. To our knowledge, only two studies have examined the influence of depression on emotion recognition in actual parent– child

1161

interactions. Sanders and colleagues (1992) found that although a correlation existed between children without depression’s rating of maternal sadness and anger in a problem-solving interaction and independent observers of the same interaction, there was no correlation between the ratings of children with depression and the independent observers’ rating. Although this investigation suggests that youth with depressive symptomatology are biased in recognizing emotional signals, whether they perceived more or less anger and sadness than independent observers is unknown. In an innovative study, Ehrmantrout, Allen, Leve, Davis, and Sheeber (2011) likewise investigated emotion recognition during a family problem-solving discussion using self-report and a semistructured video-mediated recall procedure. During video-mediated recall, youth watched the problem-solving interaction they had just participated in with their parent and were asked how their parent was feeling every 20 s of the videotaped interaction. Elevated levels of youth depressive symptomatology were associated with fewer reports of parental happy affect and more frequent reports of parental aggressive affect, controlling for observed and selfreported parental affect (Ehrmantrout et al., 2011). Although suggesting that youth depression is associated with emotion recognition biases, additional questions remain. For instance, although we expect similar relations between maternal depression and recognition biases as has been found for adolescents, neither of these studies actually investigated maternal depressive symptomatology. The current study addresses this gap by testing relations of maternal depressive symptoms to emotion recognition. Additionally, as described below in greater detail, it is possible that biases in emotion recognition associated with depression may arise only in certain contexts or be influenced by other depression-related factors (e.g., one’s own affective state, speaker’s expressions) that occur during in-person mother–adolescent interactions.

Emotion Recognition Across Context Multiple factors shape how emotion is perceived during inperson encounters. Aside from encoding variations in facial musculature, perceivers often integrate aspects of the situational context when making judgments about others’ emotions (Aviezer et al., 2008; Barrett et al., 2011; Carroll & Russell, 1996). Indeed, recent work has emphasized the pivotal role of contextual factors on emotion perception and the balance between facial expression and situational information. When a basic emotion is displayed, facial expression appears to override any additional information that is acquired regarding the situation, a concept referred to as facial dominance (Carroll & Russell, 1996; de Gelder & Hadjikhani, 2006). Yet, facial dominance may still be influenced by context. For instance, when shown pictures of a face displaying a certain emotion (e.g., disgust) that contrasted with an emotional context suggested by the background (e.g., an angry scene), eyetracking data revealed that perceivers looked more to facial regions known to be associated with the emotion suggested by context rather than to regions associated with the emotion clearly shown by the pictured face (Aviezer et al., 2008). Thus, contexts with a stronger emotional signal may guide recognition more than those with a weaker signal. Further, context appears to be particularly salient when facial expressions are more ambiguous (Carroll & Russell, 1996). In interpersonal interactions, emotions are displayed continuously, and individuals must utilize all resources

LUEBBE, FUSSNER, KIEL, EARLY, AND BELL

This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

1162

(e.g., contextual cues, verbal comments, body language) when perceiving emotion in others. Although these studies emphasize the malleability of emotion perception and the importance of exploring contextual factors in emotion recognition, very little (if any) work has explored the influence of contextual factors in individuals with elevated symptoms of depression. Evidence does indicate that adults with depression display larger deficits in emotion recognition when emotional signals are less intense (Joormann & Gotlib, 2006). Thus, we expected that in contexts eliciting clear emotional signals (e.g., the excitement of a winning a game) more depressed individuals would not exhibit recognition deficits; yet, in ambiguous contexts, when others’ emotional displays may also be more uncertain, individuals higher in depressive symptoms would likely experience even greater difficulty perceiving emotion.

The Influence of One’s Own Emotions on Recognition Aside from depressive symptomatology, other characteristics of the perceiver can contribute to biases in recognition (Niedenthal, Halberstadt, Margolin, & Innes-Ker, 2000). In theory, a perceiver’s own emotional state may impact the ability to recognize emotion (Niedenthal et al., 2000), yet relatively little empirical work has explored this effect. In an early study, Schiffenbauer (1974) investigated how induced emotional states (disgust or amusement) in perceivers influenced their judgment of pictures displaying various facial expressions. Compared with controls, individuals in the disgust condition labeled faces as relatively more negative, whereas individuals in the amusement condition labeled the same faces as relatively more positive. These results are consistent with an emotion congruent effect—that individuals are more prone to recognize emotions congruent with their internal feeling state. Likewise, Niedenthal and colleagues (2000) found that participants’ induced emotional state influenced their ability to detect a noncongruent emotional state. Compared with control participants, individuals who were induced to feel happy labeled more neutral and sad faces as happy, and individuals who were induced to feel sad labeled more neutral and happy faces as sad. Collectively, findings suggest that a perceiver’s internal feeling state strongly impacts the emotional expression they recognize in another. To the extent that individuals with elevated symptoms of depression experience more intense negative emotions or less intense positive emotions than do others in a given situation, recognition biases may be more likely to occur. For example, an adolescent with elevated depressive symptomatology may feel angrier during a mother– child discussion than a youth low in symptomatology. That anger, in turn, may prime the adolescent to perceive more anger in their mother even if she was expressing non-negative affect.

Current Study The current study examined the role of depressive symptomatology on emotion recognition during parent–adolescent interactions across multiple emotion-eliciting contexts. Three research aims guided our investigation. First, we examined whether adolescent depressive symptomatology was associated with biased recognition of maternal emotions during dyadic interactions across various contexts that differed in the valence and intensity of

emotional cues. We hypothesized that elevated symptoms of youth depression would be associated with higher reporting of negative affective states and lower reporting of positive affective states compared with parent’s own reports. We predicted that this bias in emotion recognition would be most salient in contexts with a weaker emotional signal. Second, we tested whether maternal symptoms of depression were associated with biased recognition of youth emotions across situational contexts. Similar to hypotheses for adolescents, we expected mothers higher in depression to perceive less positive affect and more negative affect compared with adolescents’ self-report of emotions, especially in contexts with a weaker emotional signal. Although few studies find that gender moderates relations of depression to emotion recognition, because there is a small but significant gender difference in emotion expression (Chaplin & Aldao, 2013) and there appears to be a general female sex advantage in facial emotion processing (McClure, 2000), it may be reasonable that depression is associated with recognition biases differently for girls and boys. As such, we tested whether adolescent gender was an additional moderator of emotion recognition biases for both mothers and adolescents. Finally, using actor–partner interdependent models (APIMs), we investigated whether one’s current affective state in a given context may account for recognition biases. We hypothesized significant indirect effects from symptomatology to biased recognition via changes in one’s own affective state. Given that emotion recognition occurs in a dyadic interaction and that evidence suggests that characteristics of the speaker can influence emotion expression (e.g., depression is related to more frequent and extended displays of negative affect; Eckman et al., 1997), the use of APIMs for Aim 3 also allowed us to test if recognition biases may be an effect not of the perceiver’s depression or own affect, but rather of the expresser’s depression.

Methods Participants Participants came from a larger project (N ⫽ 150 adolescents) examining family emotional processes and internalizing symptoms. Original study participants were recruited from public school district enrollment rosters for students in seventh to ninth grades in central Missouri via letters sent home and follow-up phone calls to legal guardians of potential participants. Of the larger data set, 13 youth participated with only a male caregiver. Because fathers respond to emotions differently than mothers (Klimes-Dougan et al., 2007) and because father data were underpowered for analyses in this data set, the current study only included those participants who were accompanied by a female caregiver (either alone, N ⫽ 113, or with a male caregiver, N ⫽ 24). Although data from 24 male caregivers are available for youth included in analyses, the small number of fathers precludes conducting inferential statistics and these data are not considered further. Adolescents in families with fathers present were more likely to be male, ␹2(1) ⫽ 5.68, p ⫽ .02, and older, M ⫽ 14.00 vs. M ⫽ 13.51, t(126) ⫽ ⫺2.41, p ⫽ .02, for present vs. absent, respectively. Importantly, no other differences in demographics or major constructs of interest (i.e., depression, ratings of affect) emerged between families in which the father was absent or present for the lab visit. Nine additional youth were missing critical data for the current study (i.e., they

This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

DEPRESSION AND EMOTION RECOGNITION IN CONTEXT

only rated one emotion for self or mother rather than all five emotions following experimental tasks), and they were also excluded from analyses. In total, participants included 128 adolescents (60 girls; Mage ⫽ 13.60, SD ⫽ .92) and their primary female caregivers. Adolescents were approximately evenly divided across seventh (32%), eighth (31%), and ninth grades (37%). Parent-identified race/ethnicity of the youth was mostly European American (86%; one participant was further identified Hispanic/Latino), with 6% African American, 3% Asian/Asian American, and 4% Biracial/Mixed Race. Female caregivers (referred to hereafter as “mothers”) were primarily biological mothers (82%). Two were step-parents, two were adoptive parents, and one was a grandmother. A fraction of respondents (14%) did not indicate their relationship, but living in the home with the adolescent was a primary inclusion criterion for participation. Mothers ranged from 29 to 60 years in age (M ⫽ 43.88; SD ⫽ 6.02), and most (75%) were married or remarried (i.e., had a spouse in the home). Families were primarily middleto upper-class: 22% of families’ yearly incomes fell below $40,000, 52% between $40,000 and $100,000, and 26% over $100,000.

Procedure All procedures were approved by the university Institutional Review Board (IRB). Data were collected in a university laboratory setting after parents provided written consent (for self and minor), and adolescents provided written assent. Self-report measures of depression were included in a larger packet of survey measures completed individually by mothers and youth in separate rooms during the first portion of the experiment. Following the surveys, mother–adolescent dyads (and when present, fathers) participated together in three 7-min experimental tasks designed to elicit various emotions. Before the first task and in between each of the subsequent tasks, adolescents and parents reported on their own and each other’s state affect for five discrete emotions (see Measures below). The first task (labeled success) was a computer-based trivia game. Participants were instructed to work together to correctly answer questions as quickly as possible and were told that correct answers would earn additional entries into a raffle for a $100 mall gift card. Questions were adapted from the board game “The Worst Case Scenario Survival Game” (®University Games) and asked about obscure survival techniques (e.g., the proper technique for surviving an alligator attack). In reality, regardless of the actual answers chosen, all dyads were provided computerized feedback at a predetermined success rate (93%) indicating that they had correctly answered items. Additionally, following the task, the experimenter provided each family with standardized verbal feedback that they had scored a high number of points, that they were one of the highest scoring families in the study to date, and that they had earned 11 extra entries in the raffle. For the second task (labeled discussion), dyads engaged in a problem-solving discussion regarding two sources of conflict (e.g., curfew) for parents and adolescents. Topics for discussion were based on items previously endorsed by the mother–adolescent dyad on the Issues Checklist (Prinz, Foster, Kent, & O’Leary, 1979). The two topics chosen by the researcher were items that participants indicated were recently discussed at home and that

1163

caused some anger or frustration. Although research assistants were trained to intervene should the conflict discussion escalate to the point where at least one member of the dyad was visibly distressed, no such instances occurred. The third task (labeled failure) was a puzzle game in which dyads were instructed to construct designs that appeared on a screen using a set of various-colored blocks. Participants were also told they would earn an additional entry into the raffle for every correctly constructed design, but lose an entry for each incorrectly constructed design. Of the five designs presented to participants, four were impossible to construct with the given blocks. The fifth design (which appeared second in the progression and was included to increase realism for the task) was correctly constructed by all families. Following this task, all dyads were provided with standardized verbal feedback indicating that they had correctly copied one design but incorrectly copied four designs and therefore lost a total of three entries in the raffle. Finally, to alleviate any lingering negative emotions, dyads engaged in a positive mood induction consisting of watching and discussing funny video clips. Following the positive mood induction, all dyads were debriefed about the study including the rigged nature of two of the tasks completed. Additionally, family members were thanked, compensated $10 apiece for their time, and entered into the drawing for a $100 mall gift card (given to one family chosen at random at the project’s completion).

Measures Adolescent depression. Adolescents reported on their own depressive symptoms using the Children’s Depression Inventory (CDI; Kovacs, 1992). For each of 27 items, participants endorse one of three statements about their feelings in the past 2 weeks (e.g., “I like myself,” “I do not like myself,” “I hate myself”). Items are scored from 0 to 2 from least to most depressive and summed to provide a total score. Because of IRB concern, the suicidal ideation item (Item 9) was excluded. The CDI has demonstrated acceptable internal consistency across varying adolescent samples (Kovacs, 1992; Smucker, Craighead, Craighead, & Green, 1986) and test–retest reliability over varying lengths of time (Smucker et al., 1986) and has demonstrated good convergent validity with other measures of depression (Hodges, 1990). For the current study, coefficient alpha was acceptable (␣ ⫽ .89). Maternal depression. Mothers reported on their own depressive symptoms using the depression subscale of the Brief Symptom Inventory (BSI; Derogatis, 1993). The BSI assesses general psychopathology symptoms. On a scale from 0 (not at all) to 4 (extremely), participants indicate how often in the past week they experienced distress or were bothered by each of 53 symptoms, and mean items scores are used for total scores. For the current study, only the six-item depression subscale (e.g., “Feeling blue”) was used (␣ ⫽ .88). The BSI is a reliable (original psychometric study ␣ ⫽ .85 for the depression subscale) and valid screener of psychological distress (Derogatis, 1993). State affect. As described above, state affect was measured prior to the interactions tasks (baseline) and again following each of the tasks (success, discussion, failure). Participants indicated on a five-point scale from 1 (Not at all) to 5 (Extremely) how they “feel right now” on five discrete emotions (i.e., happy, excited, sad, angry, and frustrated). “Happy,” “sad,” and “angry” were

This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

1164

LUEBBE, FUSSNER, KIEL, EARLY, AND BELL

chosen given that they are basic emotions likely to be elicited in our observational tasks and are conceptually consistent with past work (e.g., Ehrmantrout et al.’s [2011] use of happy, dysphoric, aggressive). “Excited” and “frustrated” were chosen to tap a positively and negatively valenced emotion that had face validity for the observational tasks that pulled for success and failure, respectively. Adolescents and mothers provided self-report and also reported on each other’s same five emotions. Importantly, the baseline measure was completed in separate rooms, and as such, each member of the dyad had no visible information on which to base their rating of their partner’s affect. In essence, this rating was a “best guess” at how the other felt. To reduce data, composite scores for positive affect (PA; mean of “happy” and “excited”) and negative affect (NA; mean of “sad,” “angry,” and “frustrated”) were calculated. Across contexts, mean alphas were acceptable for adolescents’ reports of their own affect (PA mean ␣ ⫽ .82; NA mean ␣ ⫽ .87) and for their perception of their mothers’ affect (PA mean ␣ ⫽ .80; NA mean ␣ ⫽ .83). Somewhat lower, but acceptable, alphas were found for mothers’ reports of their own affect (PA mean ␣ ⫽ .73; NA mean ␣ ⫽ .60) and for perceptions of their adolescents’ affect across contexts (PA mean ␣ ⫽ .76; NA mean ␣ ⫽ .67). Because of a study error, the term “happy” was not included on the state affect forms for the first 16 participating dyads (12.5% of the ratings for “happy” in total). Analyses indicated no differences on any study variable between these 16 dyads and the remaining 112 participants (all ps ⬎ .05). Additionally, sporadic missing data on state affect was present in an additional 22 of 8,192 total ratings of state affect (.003%), and these data were also missing at random. As such, values for PA and NA for individuals with missing data on a given discrete emotion constitute the mean of remaining emotions that did have data.

Data Analysis To assess if our manipulation of emotional context (i.e., the various tasks) elicited different mean levels of each discrete emotion for both adolescents and their mothers, repeated-measure ANOVAs with subsequent Tukey post hoc tests were calculated for self- and other-reported state affect. To test Aim 1 (i.e., replication of findings in the literature), multiple linear regressions were used to assess if adolescent self-reported depression was associated with adolescents’ reports of their mothers’ affect, controlling for mothers’ own self-reported affect. The interaction of adolescent gender and depression was also included to test for moderation. Extending the literature, effects were investigated separately at baseline and for each separate experimental context. Also extending the literature, a similar set of regression analyses were conducted to test whether maternal depression was associated with mothers’ perceptions of their children’s affect over and above the youths’ self-reported affect (i.e., Aim 2). Although the analytic approach in Aims 1 and 2 allow for comparison to past work (e.g., Ehrmantrout et al., 2011), it is limited in that it does not account for interdependence of dyadic data. To account for interdependence, for Aim 3, APIMs with distinguishable dyads (Kenny, Kashy, & Cook, 2006) were used to test whether adolescent depression was related to biased emotion recognition over and above the effects of maternal depression (and vice versa; see Figure 1, top panel, for a schematic of these

models). Additionally, these models were used to test if an individual’s own reported affect in a given situation accounted for any relations between depression and ratings of their dyad partner’s affect (i.e., indirect effects; Aim 3). All APIM models were conducted within a path-analysis framework using full-information maximum likelihood estimation in Mplus 5.1 (Muthén & Muthén, 1998 –2007). Paths from an individual’s own depression to the rating of their partner’s affect can be considered the influence of depression on emotion recognition (i.e., actor effects in the APIM model), whereas the influence of one’s own depressive symptoms on their partner’s accuracy in rating affect can be considered the path that captures the emotional expression process (i.e., partner effects in the APIM model). Because indirect effects tend to be non-normally distributed, bias-corrected bootstrapping procedures were used to create 95% confidence intervals (CIs; N ⫽ 5,000 bootstrap samples; MacKinnon, Lockwood, & Williams, 2004).

Results Preliminary Analyses For descriptive purposes regarding sample characteristics, we examined means, standard deviations, and ranges for measures of adolescent and maternal depression. Regarding total scores, most adolescents (M ⫽ 7.86, SD ⫽ 6.89, Range ⫽ 0 –38) and mothers (M ⫽ 0.37, SD ⫽ 0.63, Range ⫽ 0 –3.83; recall that mean item scores are used) were in the nonclinical range on measures of depression. Although the exclusion of Item 9 prevents calculation of total scores, projected scores based on the remaining 26 items indicated that 5% of youth scored in a borderline clinical range or higher (i.e., gender and age-adjusted T score ⬎65), approximately comparable to prevalence estimates in the population for youth in this age range (Costello, Erkanli, & Angold, 2006). For mothers, roughly 4% scored in the borderline clinical range or higher, which would be lower than expected in the population (Derogatis, 1993). In order to ensure that our various experimental contexts were, in fact, resulting in changes in participants’ experienced affect, we examined differences in mean levels of self-reported affect across contexts. Separate analyses were conducted for mothers’ and adolescents’ reports of both PA and NA. As demonstrated in Table 1, mean levels of affect changed in an expected pattern given the context of the interaction. For both mothers and adolescents, PA initially increased from baseline following the success task and then decreased following the discussion task and decreased further subsequent to the failure task. Reflecting a similar contour of change, again for both mothers and adolescents, NA decreased from baseline to the completion of the success task and then increased after the discussion and again after the failure tasks. For both PA and NA, and for each reporter, there was a significant (p ⬍ .05) change from the preceding time point of measurement. For both mothers and adolescents, ratings of their dyad partner’s affect showed identical patterns to ratings of one’s own affect across contexts (see Table 1). In general, these results provide statistical evidence that the manipulation resulted in the desired change in reported felt affect, on average, for participants.

This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

DEPRESSION AND EMOTION RECOGNITION IN CONTEXT

1165

Figure 1. Actor–partner interdependence models (N ⫽ 128 dyads) for influence of adolescent and maternal depressive symptoms predicting accuracy in rating each other’s affect as mediated by self-reported emotional experience. For ease of presentation, residual terms for “adolescents’ own affect” and “maternal own affect” are not shown. In all models, these were estimated and the correlation between the two was also estimated. (Top) A schematic of the eight models run, with bolded paths corresponding to the estimates of interest presented in Table 3. (Bottom) An example for baseline positive affect to assist the reader with Table 3 (i.e., the estimates in the first row). Adol. ⫽ adolescent; PA ⫽ positive affect. ⴱ p ⬍ .05. ⴱⴱ p ⬍ .01.

Aim 1: Is Adolescent Depressive Symptomatology Associated With Biased Recognition of Maternal Emotions Across Situational Contexts? Regression analyses indicated that, over and above mothers’ ratings of their own state affect, adolescent depressive symptoms were associated with a negative bias in recognition of their mothers’ emotions, but only for certain affective states and only in specific contexts (see Table 2, top, for estimates and t tests). Although adolescent gender

was tested as a moderator, no significant interactions emerged, and gender was removed from further analyses. For PA, adolescents who were higher in depressive symptomatology, compared with those lower in symptoms, rated their mothers as lower in PA at baseline. Recall that this context is devoid of emotional cues given that adolescents and mothers were in separate rooms. In the three joint experimental contexts, adolescents’ depressive symptomatology was unrelated to their rating

LUEBBE, FUSSNER, KIEL, EARLY, AND BELL

1166

Table 1 Results of Repeated-Measures ANOVAs for Adolescent and Maternal Report of Self and Other’s State Affect Across Situational Context Context Baseline

Success

Discussiona

Failurea

F

␩2

3.06 (1.09) 1.37 (0.76)

3.84 (0.98) 1.09 (0.45)

3.05 (1.09) 1.34 (0.73)

2.80 (1.10) 1.78 (0.96)

22.77ⴱⴱ 34.57ⴱⴱ

.15 .21

3.04 (0.95) 1.38 (0.64)

3.88 (0.96) 1.09 (0.44)

3.16 (1.04) 1.27 (0.69)

2.95 (1.16) 1.61 (0.75)

46.84ⴱⴱ 22.53ⴱⴱ

.27 .15

2.61 (0.79) 1.22 (0.44)

3.70 (0.90) 1.02 (0.10)

2.68 (0.97) 1.26 (0.53)

2.53 (0.99) 1.60 (0.57)

23.56ⴱⴱ 44.93ⴱⴱ

.16 .26

2.50 (0.90) 1.32 (0.58)

3.66 (0.99) 1.04 (0.15)

2.61 (1.06) 1.31 (0.55)

2.42 (1.02) 1.73 (0.68)

103.51ⴱⴱ 48.72ⴱⴱ

.45 .28

This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

a

Adolescent self-report PA NA Adolescent report on mother PA NA Mother self-report PA NA Mother report on adolescent PA NA

a

Note. PA ⫽ positive affect; NA ⫽ negative affect. Post hoc contrasts using least-squared difference tests compared means in each context with the preceding context (e.g., success vs. baseline; discussion vs. success). Means across emotion were significantly different from their respective preceding context mean ps ⬍ .05. a Values are means with standard deviation in parentheses. ⴱ p ⬍ .05. ⴱⴱ p ⬍ .01.

of their mother’s PA. For NA, effects emerged in contexts in which the emotional signal was more ambiguous (i.e., discussion) or absent (baseline), compared with tasks with more obvious emotional signals. Even controlling for mother’s self-reported affect, adolescents who reported greater levels of depression rated their mothers as higher in NA in these two instances. In contexts designed to elicit a stronger emotional signal (i.e., success and failure), adolescent depressive symptomatology was unrelated to ratings of maternal affect.

Aim 2: Is Maternal Depressive Symptomatology Associated With Biased Recognition of Adolescent Emotions Across Situational Contexts? A similar pattern of findings emerged, with two exceptions, for the relations of maternal depressive symptomatology to perceptions of adolescent affect, over and above adolescents’ own ratings (see Table 2, bottom). For PA, and in contrast to adolescents, maternal depression was unrelated to perceptions of lower PA at

Table 2 Relations of Perceiver’s Depression and Expresser’s Self-Reported Affect to Perceptions of Expresser’s PA and NA Across Context Context Baseline b

SE

Success t

b

SE

Discussion t

b

Failure

SE

t

b

SE

t

2.01ⴱ ⫺0.85

.23 ⫺.39

.03 .31

2.57ⴱ ⫺1.26

DV ⫽ Adolescent Report of Maternal Affect (Aim 1) PA Mother’s PA Adol. depression NA Mother’s NA Adol. depression

.14 ⴚ1.24

.09 .28

1.63 ⴚ4.46ⴱⴱ

.24 ⫺.46

.09 .30

2.76ⴱⴱ ⫺1.53

.19 ⫺.27

.09 .31

.35 1.02

.09 .25

4.00ⴱⴱ 4.04ⴱⴱ

.59 .33

.11 .30

5.42ⴱⴱ 1.09

.34 .59

.10 .30

3.53ⴱⴱ 1.97ⴱ

.44 .36

.09 .25

4.89ⴱⴱ 1.45

.33 ⫺.15

.08 .12

4.08ⴱⴱ ⴚ1.23

.25 .13

.08 .13

2.97ⴱⴱ 1.00

.53

.06

8.59ⴱⴱ

.21

.08

2.76ⴱⴱ

.29

.09

3.21ⴱⴱ

.08

.11

0.80

DV ⫽ Maternal Report of Adolescent Affect (Aim 2) PA Adolescent’s PA Mom depression NA Adolescent NA Mom depression

.20 .02

.08 .12

2.33ⴱ 0.20

.22

.08

2.74ⴱⴱ

.40

.12

3.51ⴱⴱ

.38 .15

.08 .12

.27 .27 .59 ⫺.03

.06 .06 .20 .11

4.67ⴱⴱ 1.23 4.26ⴱⴱboys 4.26ⴱⴱgirls 2.99ⴱⴱboys ⫺0.27girls

Note. N ⫽ 128. Boldfaced values represent a depression-related emotion recognition bias. PA ⫽ positive affect; NA ⫽ negative affect; Adol. ⫽ Adolescent; DV ⫽ dependent variable. ⴱ p ⬍ .05. ⴱⴱ p ⬍ .01.

This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

DEPRESSION AND EMOTION RECOGNITION IN CONTEXT

baseline. Like adolescents, maternal depression was also unrelated to PA in the other three contexts as well. Consistent with Aim 1, maternal depression was associated with perceiving their adolescent as experiencing more NA at baseline and in the discussion task. As with adolescents, no significant associations emerged in the two contexts with a stronger emotional signal. Unlike adolescents, there was also one significant adolescent gender ⫻ maternal depression interaction that occurred regarding recognition of NA in the success task (b ⫽ .61, SE ⫽ .23, t ⫽ 2.72, p ⬍ .01). Specifically, higher levels of maternal depression were associated with perceiving greater NA in the success condition for their sons (b ⫽ .59, SE ⫽ .20, t ⫽ 2.99, p ⬍ .01), but not their daughters (b ⫽ ⫺.03, SE ⫽ .11, t ⫽ ⫺0.27, p ⫽ .79). No other significant gender moderations emerged.

Aim 3: Does an Individual’s Own Affective Response Account for Emotion Recognition Biases? Of course, factors other than one’s depression may influence perceptions of another’s affect. One’s own emotional responding in a given situation may color perceptions of others. Indeed, in our sample, bivariate correlations were high for both adolescents’ rating of their own and their perceptions of their mothers’ affect (range r ⫽ .63 to .87; rmean ⫽ .74) and for mothers’ rating of their own and their perception of their children’s affect across valence and context (range r ⫽ .46 to .88; rmean ⫽ .72). Similarly, in a dyadic interaction, the other’s depressive symptomatology may influence perceptions (e.g., via changes in emotional expression). In our sample, bivariate relations between adolescent depression and mothers’ ratings of adolescent’s PA ranged from ⫺.27 to .04 (rmean ⫽ ⫺.12) and NA from .01 to .47 (rmean ⫽ .26) across contexts. Relations of maternal depression to adolescents’ ratings of their mothers’ PA ranged from ⫺.24 to .10 (rmean ⫽ .01) and NA from ⫺.01 to .34 (rmean ⫽ .08). Using APIMs, we tested if mothers’ or adolescents’ selfreported depressive symptoms continue to predict perceptions of the other’s affect over and above the effects of one’s own affective experience (which we consider as a mediator of depression to perceptions) and controlling for a partner’s symptoms. Such models provide the ability to test for unique effects of symptomatology and own affect, as well as account for interdependence in symp-

1167

toms between adolescents and mothers and for interdependence in that portion of emotion recognition not accounted for by symptomatology (i.e., correlated residuals). Because indirect effects may exist even where there are no direct effects between the independent and dependent variables, we conducted separate analyses for both PA and NA in each condition (i.e., not just models for which significant relations with depression and ratings of affect were found in Aims 1 and 2; eight models in total). Recall, that the top panel of Figure 1 is a schematic of the APIMs conducted. Of interest to Aim 3, the indirect effects, labeled “Adol. AB” and “Mom AB,” test whether relations between depressive symptoms and perceptions of the other’s affect is mediated by one’s own affective responding in the situation. This path can be decomposed into the A paths that represent the effect of depressive symptoms on one’s own affective response and the B paths that represent the relation of one’s own affective response on perceptions of a partner’s affective response. The paths labeled “Adol. D” and “Mom D” test whether a partner’s depressive symptomatology influences emotion perception (over and above other effects in the model). The paths labeled “Adol. C” and “Mom C” represent the effects of adolescent and maternal depressive symptoms on perceptions of their dyad partner’s affect, respectively. In essence, these paths are the same as those in Aims 1 and 2, but now control for the other dyad member’s symptoms and the individual’s own state affect. Although not immediately germane to Aim 3, all other paths were estimated in the models, including the correlation between the residual terms for adolescents’ and mothers’ own affect (full models can be accessed in Online Supplemental Materials). Standardized path coefficients and indirect effects for APIMs are shown in Table 3. To assist in understanding Table 3, the bottom panel of Figure 1 displays coefficients for PA at baseline (i.e., the first row of Table 3). With one exception for adolescents and one for mothers, depressive symptoms were not directly related to lower or higher ratings of either PA or NA across contexts (i.e., C paths). Adolescent depressive symptoms were related to perceptions of lower maternal NA in the success task. Maternal depressive symptoms were related to perceptions of greater adolescent PA in the failure task. Both effects reflect suppression effects such that no recognition biases in these contexts were found in Aims 1 or 2.

Table 3 Standardized Path Coefficients and Indirect Effects of Interest for Actor–Partner Interdependence Models Across Affect and Context

Positive affect Baselinea Success Discussion Failure Negative affect Baselinea,b Successb Discussiona,b Failure

Adol. A

Adol. B

Adol. C

Adol. AB

Adol. D

Mom A

Mom B

Mom C

Mom AB

Mom D

⫺.31ⴱ ⫺.23ⴱ ⫺.37ⴱⴱ ⫺.28ⴱⴱ

.62ⴱⴱ .74ⴱⴱ .77ⴱⴱ .88ⴱⴱ

⫺.14 .01 .14 .08

⫺.19ⴱⴱ ⫺.17ⴱ ⫺.28ⴱⴱ ⫺.25ⴱⴱ

⫺.03 .03 ⫺.02 ⫺.15ⴱⴱ

⫺.11 .17ⴱ .07 .03

.60ⴱⴱ .80ⴱⴱ .83ⴱⴱ .86ⴱⴱ

.09 ⫺.05 ⫺.07 .15ⴱ

⫺.07 .14ⴱ .06 .03

⫺.12 .09 ⫺.03 ⫺.03

.51ⴱⴱ .48ⴱ .50ⴱⴱ .30ⴱⴱ

.58ⴱⴱ .73ⴱⴱ .83ⴱⴱ .77ⴱⴱ

.00 ⫺.16ⴱ ⫺.10 ⫺.09

.37ⴱ .63ⴱ .51ⴱⴱ .76ⴱⴱ

.05 .07 .04 .00

.23 .00 .11 .09

⫺.03 ⫺.03 ⫺.03 ⫺.01

.30ⴱⴱ .35ⴱ .41ⴱⴱ .23ⴱⴱ

.27ⴱ ⫺.03 .14 .01

.62ⴱⴱ .00 .21ⴱ .11

Note. All models run using Full-Information Maximum Likelihood. Adol. ⫽ Adolescent. a Adolescent depression was associated with biased recognition in Aim 1 for this affective state and context. biased recognition in Aim 2 for this affective state and context. ⴱ p ⬍ .05. ⴱⴱ p ⬍ .01.

b

Maternal depression was associated with

LUEBBE, FUSSNER, KIEL, EARLY, AND BELL

This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

1168

In all models run, there was a significant indirect effect (i.e., AB path) from adolescent depression to ratings of their mothers’ affect via their own affective response in the moment. That is, at baseline and across all contexts, an adolescent’s depressive symptoms were associated with decreased PA, as well as increased NA. In turn, these affective responses were strongly associated with their ratings of their mothers’ affect in that situation. Thus, biases found in Aim1 appear to be accounted for by adolescents’ own emotional responses in a given situation.1 In contrast to adolescents, there was only one significant indirect effect from maternal depressive symptoms to their perception of their child’s affect via their own affective response. This effect occurred for PA in the success task and was smaller in size compared with all indirect effects found for adolescents. Note, too, that this model was not an instance in Aim 2 in which maternal depression was related to decreased perceptions of PA. When examining the A and B paths separately, in contrast to adolescents, maternal depression was infrequently related to self-reported emotional responding in the situation (i.e., in only three of eight models, with one of those being the counterintuitive finding of depression being related to greater PA in the success task). Similar to adolescents, however, maternal self-reported PA and NA were strongly related to their perceptions of adolescents’ PA and NA, respectively. Like adolescents, mothers’ own affect does influence their ratings of their children’s affect, but unlike adolescents, mother’s depression does not appear to influence their own state affect as strongly. There was limited support for the idea that a partner’s depressive symptoms influence the other’s perceptions of the partner’s emotions (i.e., D paths). Adolescents who were higher in depression had mothers who rated them as being lower in PA in the failure task and higher in NA at baseline. Mothers’ depression did not influence adolescents’ ratings of maternal affect.

Discussion The current study tested whether adolescents’ and mothers’ depressive symptoms were associated with a negative bias in the perception of each others’ affect across various contexts that differed in the strength of the emotional signal present. Depression was associated with perceiving more NA in dyad partners in the baseline and discussion conditions for both mothers and adolescents. Mothers’ depressive symptoms were also associated with perceptions of greater NA in their sons in the success task. For PA, only one relation with depression emerged: adolescent depression was associated with perception of lower maternal PA in the baseline condition. Additionally, we tested whether the perceiver’s own affective response in a given context (as well as the expresser’s depression) accounted for emotion recognition biases. Findings differed for adolescents and mothers. Adolescents’ own affective state mediated the relation between depression and perceptions of their mothers’ affect in all models tested. Although mothers’ own affective state did strongly influence their ratings of their adolescents’, their depressive symptoms did not generally influence their own affective state. Finally, neither adolescents’ nor mothers’ depressive symptoms appeared to influence the others’ perceptions of affect.

Aims 1 and 2: Adolescent and Maternal Depression’s Influence on Biased Recognition of Emotion Our findings replicate findings in the literature suggesting a negative bias in emotion recognition for adults and adolescents who are higher in depressive symptomatology (e.g., Bistricky et al., 2011; Ehrmantrout et al., 2011; Joormann & Gotlib, 2006; van Beek & Dubas, 2008). Adding to the existing literature, however, it appears that the negative bias may arise only for some affective states in specific contexts. As Barrett and colleagues (2011) argue, extant research on emotion perception primarily occurs using methodology devoid of context. Our findings lend additional support to the hypothesis that the emotional signal present in a given context influences one’s perceptions of others’ emotions. Consistent with situational dominance findings (Carroll & Russell, 1996), recognition biases were associated with both adolescent and maternal depression primarily in contexts in which emotional signals were either more theoretically ambiguous (i.e., the discussion task) or nonexistent (i.e., the baseline condition). Our baseline condition findings suggest that “recognition” biases may exist even apart from what is being observed in another. Thus, for individuals with greater depressive symptoms, difficulties in emotion accuracy may not only stem from information-processing biases in identification, selective attention, or memory processes associated with reading others’ cues accurately (see Bistricky et al., 2011), but also overcoming general negative cognitive biases about interpersonal relationships (e.g., Beck, 1987) that may lead the individual to believe another is experiencing more negative and less positive emotion than he or she truly is. In fact, in our data, as emotional cues in the environment became stronger, depressive biases were generally reduced or not evident. Although considerably more studies have examined adult rather than adolescent depression in relation to emotion processing biases, a major contribution of the current study was investigation of the influence of mothers’ depressive symptoms on recognizing emotions in their own children. In our study, mothers higher in depressive symptoms exhibited an almost identical pattern of bias regarding NA as did adolescents higher in depressive symptoms. Specifically, both at baseline and in the discussion task, mothers higher in depression overattributed NA to their adolescents (and vice versa). This consistent finding suggests that emotion processing biases may exist across development and do not appear to be influenced by potential role-differences (i.e., being the parent vs. being the child) in intimate family relationships. Developmental similarity is not completely surprising because emotion identification reaches adult-like levels by approximately age 10, an age younger than the current sample (van Beek & Dubas, 2008). Coupling findings for both adolescents and mothers, depressiverelated negative biases in ambiguous situations may have significant consequences for parent– child relationships, especially dur1 A second set of four APIMs (one for each context) were conducted that included both PA and NA simultaneously in models. That is, models included both adolescent and maternal depression as independent variables, self-reported PA and NA for each partner as mediators, and ratings of others’ PA and NA for each partner as dependent variable. Results were completely consistent with the separate PA and NA models described in Aim 3. Results are presented individually for PA and NA models for clarity of presentation.

This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

DEPRESSION AND EMOTION RECOGNITION IN CONTEXT

ing attempts to resolve conflict that, unfortunately, peaks in early adolescence (Laursen, Coy, & Collins, 1998). Unsatisfying social interactions fraught with misreading of emotional cues on the part of either the adolescent or the mother may contribute to feelings of rejection or to development or maintenance of cognitive biases regarding self-worth, perceived social threat, or parenting incompetence (for mothers), in turn contributing to exacerbation of depressive symptoms. Regarding Aims 1 and 2, two unexpected findings emerged. First, in contrast to hypotheses, higher levels of maternal depression were associated with greater perception of NA in our task designed to elicit positive emotion. Although mothers’ depression seemed to only interfere for recognition of boys’, and not girls’, NA in this positive condition, caution should be taken interpreting this moderation as it was the only significant interaction that emerged across all analyses. Considering the main effects, this was the only relation of depression to recognition in a task that possessed a strong emotional signal. Potentially, individuals greater in depressive symptoms assume that positive events must necessarily come with a cost and attribute this cost to others’ emotions as well. Second, only at baseline was adolescent depression associated with perceptions of less PA for their mothers. In contrast to hypotheses, this underreporting bias for positive emotions did not emerge in the discussion task for adolescents or in any context for mothers. This lack of effect was somewhat surprising given that depressive symptoms are associated with (a) reduced attention to positive faces in adults (e.g., Cavanagh & Geisler, 2006; Fritzsche et al., 2010), (b) underreporting of maternal positive affect by adolescents in family discussions (Ehrmantrout et al., 2011), and (c) needing greater intensity of facial expressions to correctly identify positive faces in both adults (Joormann & Gotlib, 2006) and adolescents (Joormann et al., 2010). Most studies, like those cited above, find PA-related recognition deficits using clinical or at-risk samples, and such deficits may not be as readily apparent in community-based samples like the current study. Our results do add to a considerably mixed literature regarding identification and attention to positive faces, and it may be that the bias toward socially rejecting stimuli (e.g., perceiving more sadness in others) is stronger than the bias away from affirming stimuli (e.g., perceiving less happiness).

Aim 3: The Influence of an Individual’s Own Affective Response on Emotion Recognition Biases Our final contribution was to test whether two factors may help to explain recognition biases in dyadic interactions associated with either adolescent or maternal depression. It is possible that the expresser’s depression might influence expression of emotions which, in turn, influences the receiver’s perception. Likewise, it is possible that one’s own affective response in a given situation greatly influences his or her perception of the other’s emotions. That is, “what I feel is what others feel.” To the extent that depression influences “what I feel,” recognition biases in “what others feel” may emerge. Using APIMs allowed for parceling such effects while accounting for partners’ own symptoms. With two exceptions, across both adolescents and mothers as well as regardless of context, the expresser’s depressive symptoms did not influence how the dyad partner perceived the expresser’s emotion. The two exceptions were that adolescents who were

1169

higher in depression had mothers who rated them as being lower in PA in the failure task and higher in NA at baseline. Compared with many other effects in the model described below, these effects were relatively small in size. Although emotion expression in response to provocative stimuli differs in terms of duration and intensity for depressed individuals compared with controls (Berenbaum & Oltmanns, 1992; Ekman, Matsumoto, & Friesen, 1997), our findings generally suggest that depressive symptomatology-based differences in expression do not appear to influence recognition. Coupled with findings from Aims 1 and 2, depressive symptoms appear to influence reading others rather than being read by others. Indeed, one finding of our study was that an adolescent’s own affective reactivity in a given context appeared to mediate the relation of depression to emotion recognition biases. This finding was consistent for adolescents across affective valence and contexts. In general, adolescents appeared to attribute emotions to their mothers based primarily on their own feelings. Recognition biases, therefore, seem to emerge to the extent that depressive symptoms influenced an adolescents’ own affective reactivity. A developmental difference, however, emerged for this phenomenon. For mothers, the indirect effect of depression on recognition was mediated via a mother’s own affect in only one of the eight models tested (and in an unexpected direction). A closer examination of the model pathways indicated that although a mother’s own affect was consistently highly related to perceptions of her adolescent’s affect (i.e., “b” paths), depressive symptomatology (with a few exceptions) did not appear to relate to mothers’ own reactivity (i.e., “a” paths). Overall, for both mothers and adolescents, results concerning the “b” paths were consistent with the emotion congruent effect found in past work (e.g., Niedenthal et al., 2000; Schiffenbauer, 1974) such that one’s own emotions greatly influence perceptions of others’ emotions. Why, exactly, differences in the “a” path emerged is unclear. Potentially, in this laboratory context, mothers higher in depressive symptomatology were less sensitive to the emotional manipulations of the contexts than were youth higher in symptoms, a finding that would be consistent with Rottenberg, Gross, and Gotlib (2005) emotional insensitivity to context theory of depression (for a meta-analysis see Bylsma, Morris, & Rottenberg, 2008). Given the developmental difference found in our study, this would suggest that context insensitivity is somehow more applicable to depression occurring in later development. Indeed, support for context insensitivity in adolescents with depression is mixed, with a study using ecological momentary assessment finding greater variability in emotional experience for depressed adolescents compared with controls (Silk et al., 2011), whereas another study found greater emotional inertia (i.e., context insensitivity) for depressed adolescents in parent– child interactions (Kuppens, Allen, & Sheeber, 2010). A similar possibility may have to do with the chronicity, rather than current state, of depressive symptoms. Although chronicity was not explicitly measured in the current study, research with clinically depressed adults suggests lower emotional reactivity predicts longer course of depression (Peeters, Berkhof, Rottenberg, & Nicolson, 2010). Flipping this relation, it may be that compared with adolescents, the mothers in our sample have been experiencing depressive symptoms for a longer time, thereby blunting emotional reactivity. That said, emotion recognition biases have been found in never-

1170

LUEBBE, FUSSNER, KIEL, EARLY, AND BELL

This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

depressed offspring of depressed mothers (Joormann et al., 2010). Clearly, more work is needed to determine if such recognition biases, acting in concert with emotion reactivity, represent endophenotypic risk for depression or are a function of some aspect of depression. A final possibility may have more to do with role differences in the parent– child dyad rather than development per se. In interactions involving their children, even mothers experiencing more depressive symptoms may be highly motivated to better regulate emotions in those contexts (e.g., to model emotional control by not allowing negative experiences to increase their NA). Exploring these developmental or role differences warrants more attention in future studies.

Limitations and Future Directions Our findings should be interpreted in light of several limitations. First, because of the need for deception (i.e., winning entries before losing them), the experimental contexts could not be counterbalanced. As such, the influence of carry-over effects is unknown (e.g., emotional inertia), which may be influential given data suggesting that depressed youth display greater emotional inertia for both negative and positive emotions (Kuppens et al., 2010). However, given the current study’s focus on individual differences, counterbalancing would not be recommended. Second, although our use of a community sample allowed us to test for the effects of depression across a range of symptoms in both dyad partners, testing these processes in clinical or at-risk samples is needed. Given that findings emerged in the current study without leveraging an extreme-groups approach, it is possible that effects of depression on recognition may actually be stronger in clinical samples. Indeed, across studies using varied samples, there appears to be a relation between depression severity and greater facial affect identification errors (Bistricky et al., 2011). The relatively low levels of depressive symptomatology, especially for mothers, may help explain the lack of relation between depression and self-reported affect as well as recognition biases. Third, the crosssectional nature of our data also precludes interpretation of causality. Affect recognition biases may place individuals at risk for overperceiving social threat or underperceiving support, elevating risk for depressive symptoms (e.g., Allen & Badcock, 2003), or and as suggested by studies that find biases only during activated negative mood states (e.g., Joormann et al., 2010) and by our findings regarding one’s own affect in the moment, depressive symptomatology might activate biases that could reinforce or maintain symptoms. Finally, our study relied only on self-report of own affect as the “gold standard” to which perceptions were compared, despite evidence that self-reported felt affect and emotional behavior may be only moderately correlated at best (though perception of affect tends to correlate higher with emotion behavior than other measures of emotional experience such as physiological markers; e.g., Mauss, Levenson, McCarter, Wilhelm, & Gross, 2005). The choice to use self-report did replicate past methodology (e.g., Ehrmantrout et al., 2011), and in all but one instance, perceptions of the dyad partners’ affect were significantly correlated with the other partners’ actual self-reported affect. Yet, this approach did not consider known individual differences in display rules or dispositional expressivity (Gross, John, & Richards, 2000). A fruitful area for future research is undoubtedly using observed affective behavior as the “gold standard” to which per-

ceptions from a dyad partner are compared. Similarly, discrepancy in felt versus expressed emotion may be an important moderator of the relation between depression and emotion recognition of others.

Conclusions and Implications Despite these limitations, to our knowledge, this was the first study to investigate the effects of both maternal and adolescent symptomatology simultaneously as well as to measure emotion perception processes in “real-time” among mother–adolescent dyads in multiple contexts. Our study was also the first to control for an individual’s own emotional response (i.e., state affect) when examining the relation of depression to emotion recognition during an interactional task. Given the very high correlations between self-ratings of emotion and perceptions of others’ emotions in the current study, the lack of inclusion of this variable in past research limits our understanding of what factors influence emotion recognition in social interactions. Overall, depressive symptoms appear to disrupt recognition of negative emotions (and to a lesser extent positive emotions) for both adolescents and mothers in interactions characterized by ambiguous or low-levels of emotional signal. Although both mothers and adolescents higher in depression exhibit emotion recognition biases, the mechanisms may be different based on developmental period. In our study, adolescents higher in depressive symptoms indicated greater reactivity for negative emotions and less reactivity to positive emotions, and as such, attributed similar emotional states to their mothers accordingly. Although mothers’ own emotions also influenced their perceptions of their child, depressive symptoms did not appear to influence their own emotional reactivity as strongly as for adolescents. As such, when biases were present for mothers higher in depression, mothers’ own emotions did not explain those findings. Our findings have several applied implications. First, researchers studying emotion recognition may be well served to measure the perceivers’ own emotional state at the time. Second, from a clinical perspective, given our findings that one’s own internal state may change recognition— especially for youth higher in depression— clinical interventions designed to increase one’s emotional awareness and subsequent coping skills to regulate said emotions may improve recognition. In turn, improving recognition should help individuals to gain appropriate support from others. Likewise, given that recognition biases emerged most saliently in situations when the emotional signal was ambiguous, interventions aimed at increasing appropriate emotion expressiveness in the family may work to increase environmental cues and improve recognition. In this light, family based treatments with a focus on interactional patterns that surround emotional situations may be most beneficial for youth or mothers with depression, with the ultimate goal that gains made in the family generalize to other social situations.

References Abramson, L. Y., Alloy, L. B., Hankin, B. L., Haeffel, G. J., MacCoon, D. G., & Gibb, B. E. (2002). Cognitive vulnerability-stress models of depression in a self-regulatory and psychobiological context. In I. H. Gotlib & C. L. Hammen (Eds.), Handbook of depression (pp. 268 –294). New York, NY: Guilford Press. Allen, N. B., & Badcock, P. B. (2003). The social risk hypothesis of depressed mood: Evolutionary, psychosocial, and neurobiological per-

This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

DEPRESSION AND EMOTION RECOGNITION IN CONTEXT spectives. Psychological Bulletin, 129, 887–913. doi:10.1037/00332909.129.6.887 Aviezer, H., Hassin, R. R., Ryan, J., Grady, C., Susskind, J., Anderson, A., . . . Bentin, S. (2008). Angry, disgusted, or afraid? Studies on the malleability of emotion perception. Psychological Science, 19, 724 –732. doi:10.1111/j.1467-9280.2008.02148.x Barrett, L. F., Mesquita, B., & Gendron, M. (2011). Context in emotion perception. Current Directions in Psychological Science, 20, 286 –290. doi:10.1177/0963721411422522 Beck, A. T. (1987). Cognitive models of depression. Journal of Cognitive Psychotherapy, 1, 5–37. Beevers, C. G., Wells, T. T., Ellis, A. J., & Fischer, K. (2009). Identification of emotionally ambiguous interpersonal stimuli among dysphoric and nondysphoric individuals. Cognitive Therapy and Research, 33, 283–290. doi:10.1007/s10608-008-9198-6 Berenbaum, H., & Oltmanns, T. F. (1992). Emotional experience and expression in schizophrenia and depression. Journal of Abnormal Psychology, 101, 37– 44. doi:10.1037/0021-843X.101.1.37 Bistricky, S. L., Ingram, R. E., & Atchley, R. A. (2011). Facial affect processing and depression susceptibility: Cognitive biases and cognitive neuroscience. Psychological Bulletin, 137, 998 –1028. doi:10.1037/ a0025348 Bylsma, L. M., Morris, B. H., & Rottenberg, J. (2008). A meta-analysis of emotional reactivity in major depressive disorder. Clinical Psychology Review, 28, 676 – 691. doi:10.1016/j.cpr.2007.10.001 Carroll, J. M., & Russell, J. A. (1996). Do facial expressions signal specific emotions? Judging emotion from the face in context. Journal of Personality and Social Psychology, 70, 205–218. Cavanagh, J., & Geisler, M. W. (2006). Mood effects on the ERP processing of emotional intensity in faces: A P3 investigation with depressed students. International Journal of Psychophysiology, 60, 27–33. doi: 10.1016/j.ijpsycho.2005.04.005 Chaplin, T. M., & Aldao, A. (2013). Gender differences in emotion expression in children: A meta-analytic review. Psychological Bulletin, 139, 735–765. doi:10.1037/a0030737 Costello, E. J., Erkanli, A., & Angold, A. (2006). Is there an epidemic of child or adolescent depression? Journal of Child Psychology and Psychiatry, 47, 1263–1271. doi:10.1111/j.1469-7610.2006.01682.x de Gelder, B., & Hadjikhani, N. (2006). Non-conscious recognition of emotional body language. NeuroReport, 17, 583–586. doi:10.1097/ 00001756-200604240-00006 Derogatis, L. R. (1993). BSI brief symptom inventory. Administration, scoring, and procedures manual (4th ed.). Minneapolis, MN: National Computer Systems. Ehrmantrout, N., Allen, N. B., Leve, C., Davis, B., & Sheeber, L. (2011). Adolescent recognition of parental affect: Influence of depressive symptoms. Journal of Abnormal Psychology, 120, 628 – 634. doi:10.1037/ a0022500 Ekman, P., Matsumoto, D., & Friesen, W. V. (1997). Facial expression in affective disorders. In P. Ekman & E. L. Rosenberg (Eds.), What the face reveals (pp. 331–341). New York, NY: Oxford University Press. Feinberg, T. E., Rifkin, A., Schaffer, C., & Walker, E. (1986). Facial discrimination and emotional recognition in schizophrenia and affective disorders. Archives of General Psychiatry, 43, 276 –279. doi:10.1001/ archpsyc.1986.01800030094010 Frewen, P. A., & Dozois, D. J. A. (2005). Recognition and interpretation of facial expressions in dysphoric women. Journal of Psychopathology and Behavioral Assessment, 27, 305–315. doi:10.1007/s10862-0052410-z Fritzsche, A., Dahme, B., Gotlib, I. H., Joormann, J., Magnussen, H., Watz, H., . . . von Leupoldt, A. (2010). Specificity of cognitive biases in patients with current depression and remitted depression and in patients with asthma. Psychological Medicine, 40, 815– 826. doi:10.1017/ S0033291709990948

1171

Gotlib, I. H., & Hammen, C. L. (1992). Psychological aspects of depression: Toward a cognitive-interpersonal integration. Oxford, UK: Wiley. Gross, J. J., John, O. P., & Richards, J. M. (2000). The dissociation of emotion expression from emotion experience: A personality perspective. Personality and Social Psychology Bulletin, 26, 712–726. doi:10.1177/ 0146167200268006 Gur, R. C., Erwin, R. J., Gur, R. E., Zwil, A. S., Heimberg, C., & Kraemer, H. C. (1992). Facial emotion discrimination: II. Behavioral findings in depression. Psychiatry Research, 42, 241–251. doi:10.1016/01651781(92)90116-K Hodges, K. (1990). Depression and anxiety in children: A comparison of self-report questionnaires to clinical interview. Psychological Assessment, 2, 376 –381. doi:10.1037/1040-3590.2.4.376 Joormann, J., Gilbert, K., & Gotlib, I. H. (2010). Emotion identification in girls at risk for depression. Journal of Child Psychology and Psychiatry, 51, 575–582. doi:10.1111/j.1469-7610.2009.02175.x Joormann, J., & Gotlib, I. H. (2006). Is this happiness I see? Biases in the identification of emotional facial expressions in depression and social phobia. Journal of Abnormal Psychology, 115, 705–714. doi:10.1037/ 0021-843X.115.4.705 Keltner, D., & Kring, A. M. (1998). Emotion, social function, and psychopathology. Review of General Psychology, 2, 320 –342. doi:10.1037/ 1089-2680.2.3.320 Kenny, D. A., Kashy, D. A., & Cook, W. L. (2006). Dyadic data analysis. New York, NY: Cambridge University Press. Klimes-Dougan, B., Brand, A. E., Zahn-Waxler, C., Usher, B., Hastings, P. D., Kendziora, K., & Garside, R. B. (2007). Parental emotion socialization in adolescence: Differences in sex, age and problem status. Social Development, 16, 326 –342. doi:10.1111/j.1467-9507.2007 .00387.x Kovacs, M. (1992). Children’s depression inventory: Manual. North Tonawanda, NY: Multi-Health Systems. Kuppens, P., Allen, N. B., & Sheeber, L. B. (2010). Emotional inertia and psychological maladjustment. Psychological Science, 21, 984 –991. doi: 10.1177/0956797610372634 Laursen, B., Coy, K. C., & Collins, W. A. (1998). Reconsidering changes in parent-child conflict across adolescence: A meta-analysis. Child Development, 69, 817– 832. doi:10.1111/j.1467-8624.1998.tb06245.x Leppänen, J. M., Milders, M., Bell, J. S., Terriere, E., & Hietanen, J. K. (2004). Depression biases the recognition of emotionally neutral faces. Psychiatry Research, 128, 123–133. doi:10.1016/j.psychres.2004.05.020 MacKinnon, D. P., Lockwood, C. M., & Williams, J. (2004). Confidence limits for the indirect effect: Distribution of the product and resampling methods. Multivariate Behavioral Research, 39, 99 –128. doi:10.1207/ s15327906mbr3901_4 Mauss, I. B., Levenson, R. W., McCarter, L., Wilhelm, F. H., & Gross, J. J. (2005). The tie that binds? Coherence among emotion experience, behavior, and physiology. Emotion, 5, 175–190. doi:10.1037/1528-3542.5 .2.175 McClure, E. B. (2000). A meta-analytic review of sex differences in facial expression processing and their development in infants, children, and adolescents. Psychological Bulletin, 126, 424 – 453. doi:10.1037/00332909.126.3.424 Muthén, L. K., & Muthén, B. O. (1998 –2007). Mplus user’s guide (5th ed.). Los Angeles, CA: Muthén & Muthén. Niedenthal, P. M., Halberstadt, J. B., Margolin, J., & Innes-Ker, A. H. (2000). Emotional state and the detection of change in facial expression of emotion. European Journal of Social Psychology, 30, 211–222. doi:10.1002/(SICI)1099-0992(200003/04)30:2⬍211::AID-EJSP988⬎3 .0.CO;2-3 Peeters, F., Berkhof, J., Rottenberg, J., & Nicolson, N. A. (2010). Ambulatory emotional reactivity to negative daily life events predicts remission from major depressive disorder. Behaviour Research and Therapy, 48, 754 –760. doi:10.1016/j.brat.2010.04.008

This document is copyrighted by the American Psychological Association or one of its allied publishers. This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.

1172

LUEBBE, FUSSNER, KIEL, EARLY, AND BELL

Rottenberg, J., Gross, J. J., & Gotlib, I. H. (2005). Emotion context insensitivity in major depressive disorder. Journal of Abnormal Psychology, 114, 627– 639. doi:10.1037/0021-843X.114.4.627 Sanders, M. R., Dadds, M. R., Johnston, B. M., & Cash, R. (1992). Childhood depression and conduct disorder: I. Behavioral, affective, and cognitive aspects of family problem-solving interactions. Journal of Abnormal Psychology, 101, 495–504. doi:10.1037/0021-843X.101.3 .495 Schiffenbauer, A. (1974). Effect of observer’s emotional state on judgments of the emotional state of others. Journal of Personality and Social Psychology, 30, 31–35. doi:10.1037/h0036643 Silk, J. S., Forbes, E. E., Whalen, D. J., Jakubcak, J. L., Thompson, W. K., Ryan, N. D., . . . Dahl, R. E. (2011). Daily emotional dynamics in depressed youth: A cell phone ecological momentary assessment study. Journal of Experimental Child Psychology, 110, 241–257. doi:10.1016/ j.jecp.2010.10.007

Smucker, M. R., Craighead, W. E., Craighead, L. W., & Green, B. J. (1986). Normative and reliability data for the Children’s Depression Inventory. Journal of Abnormal Child Psychology, 14, 25–39. doi: 10.1007/BF00917219 Steinberg, L., & Morris, A. S. (2001). Adolescent development. Annual Review of Psychology, 52, 83–110. doi:10.1146/annurev.psych.52.1.83 van Beek, Y., & Dubas, J. (2008). Decoding basic and non-basic facial expressions and depressive symptoms in late childhood and adolescence. Journal of Nonverbal Behavior, 32, 53– 64. doi:10.1007/s10919-0070041-7

Received January 4, 2013 Revision received May 29, 2013 Accepted June 27, 2013 䡲

Role of adolescent and maternal depressive symptoms on transactional emotion recognition: context and state affect matter.

Depressive symptomatology is associated with impaired recognition of emotion. Previous investigations have predominantly focused on emotion recognitio...
682KB Sizes 0 Downloads 0 Views