The Persistence of Hedonically-based Mood Repair Among Young Offspring at High- and Low-Risk for Depression

Abstract

The aim of the present study was to examine whether offspring at high and low familial risk for depression differ in the immediate and more lasting behavioral and physiological effects of hedonically-based mood repair. Participants (9- to 22-year olds) included never-depressed offspring at high familial depression risk (high-risk, n = 64), offspring with similar familial background and personal depression histories (high-risk/DEP, n = 25), and never-depressed offspring at low familial risk (controls, n = 62). Offspring provided affect ratings at baseline, after sad mood induction, immediately following hedonically-based mood repair, and at subsequent, post-repair epochs. Physiological reactivity, indexed via respiratory sinus arrhythmia (RSA), was assessed during the protocol. Following mood induction and mood repair, high- and low-risk (control) offspring reported comparable changes in levels of sadness and RSA. However, sadness increased among high-risk offspring following the post-repair epoch, whereas low-risk offspring maintained mood repair benefits. High-risk/DEP offspring also reported higher levels of sadness following the post-repair epoch than did low-risk offspring. Change in RSA did not differ across the three offspring groups. Self-ratings confirm that one source of difficulty associated with depression risk is diminished ability to maintain hedonically-based mood repair gains, which were not apparent at the physiological level.

Mood repair, which refers to the process of recovering from sad, dysphoric affect, is an important facet of emotion-regulation and a cornerstone of adaptive functioning (Josephson, Singer, & Salovey, 1996). The successful regulation of sadness, along with other emotions, also is an important aspect of normative development (Cole, Martin, & Dennis, 2004; Eisenberg, Champion, & Ma, 2004; John & Gross, 2004). In turn, problems in repairing one’s sadness and dysphoria are associated with psychopathology and characterize individuals prone to, or actually suffering from depression (Aldao, Nolen-Hoeksema, & Schweizer, 2010; Gross & Muñoz, 1995; Kring & Werner, 2004). It has been proposed that the mood repair problems of depressed and depression-prone individuals may reflect the absence of the requisite skills needed to implement such responses (for reviews, see Joormann & Vanderlind, 2014; Kovacs & Yaroslavsky, 2014). But based on self-ratings of mood, the findings generally indicate that depressed and depression-prone individuals can attenuate sadness in the laboratory just as well as controls can when instructed to implement strategies such as cognitive reappraisal (Beauregard, Paquette, & Levesque, 2006; Dillon & Pizzagalli, 2013; Erk et al., 2010; Quigley & Dobson, 2014, but see D’Avanzato, 2013) or attention refocusing (Joormann, Siemer, & Gotlib, 2007).

On the other hand, depressed and depression-prone individuals do appear to have problems when instructed to recall positive autobiographical memories (PAM) to repair sad mood. For example, after sadness induction, recalling PAM resulted in mood repair benefits for control subjects, but exacerbated the sad mood of depressed adults (Joormann et al., 2007). Kovacs et al. (2015) likewise found that PAM for mood repair was less effective among currently depressed adolescents than among emotionally healthy peers. There also is evidence that the use of PAM for mood repair yields less than optimal results among individuals who have recovered from depression (Joormann et al., 2007; Kovacs et al., 2015). Notably, PAM appears to be compromised among never depressed individuals at high risk for depression, suggesting that it may serve as a marker of risk (Joormann & Siemer, 2004).

One important way in which PAM for mood repair differs from other regulatory responses that have been examined (e.g., reappraisal, distraction, expressive suppression) is that it leverages hedonic capacity, or positive affect, which appears to be limited among depressed and depression-prone individuals (for a review see Gilbert, 2012; for a meta-analysis see Bylsma, Morris, & Rottenberg, 2008). If it is the hedonic tone of PAM that poses a challenge to depression-prone individuals, one would also expect impaired performance in implementing other hedonic toned strategies, such as turning to up-beat entertainment, seeking out humorous material, or eating pleasurable foods such as chocolate (Parkinson & Totterdell, 1999; Scholey & Owen, 2013; Thayer, Newmann, & McClain, 1994). While there is evidence, for example, that comedy can ameliorate the negative physiological consequences of distress (Fredrickson & Levenson, 1998) and that consuming chocolate can reduce subjects’ negative mood (Macht & Muller, 2007), research has focused mostly on PAM in spite of the fact that it is infrequently used to repair mood in naturalistic settings (Heiy & Cheavens, 2014). Thus, one goal of the current study was to extend the array of hedonically toned mood repair responses to be examined and assess if their implementation poses a challenge to depression-prone individuals.

The impaired hedonic capacity associated with depression-risk may also make it difficult to maintain the benefits of mood repair responses that leverage positive affect. Indeed, there is evidence that depression is associated with deficits in sustaining positive affect (Tomarken & Keener, 1998). For example, compared to healthy peers, depressed adolescents report shorter duration of positive emotion (Sheeber et al., 2009). In the laboratory, depressed and remitted individuals exhibit difficulty sustaining positive emotion over a short period of time following the presentation of hedonic stimuli (Admon & Pizzagalli, 2015; Horner et al., 2014; McMakin, Santiago, & Shirk, 2009), which has been confirmed by neuroimaging findings (Admon & Pizzagalli, 2015; Heller et al., 2009). However, to the best of our knowledge, foreshortened duration of positive affect as part of mood repair has not been examined among depression-prone individuals.

There is little question that the experience and regulation of emotion and mood is supported by a complex array of physiological processes (Gross, 1998). One physiological process that has been specifically suggested as an index of emotion regulation is respiratory sinus arrhythmia (RSA; Porges, 2007). RSA mirrors the effect of the vagal nerve on the cardiac interbeat interval and is believed to reflect flexibility in responding to environmental stimuli or demands, including those in the affective domain (Appelhans & Luecken, 2006; Thayer & Lane, 2000, 2009). During a resting state, the effect of the vagal nerve is evidenced in longer interbeat intervals (IBI) during inspiration relative to the expiration phase of the respiration cycle, which is normative (and is associated with relaxation); in response to most stimuli, vagal input to the heart is decreased (vagal withdrawal), thereby reducing IBI asynchrony within the respiration cycle; upon termination of the external stimulus, vagal input to the heart is expected to be restored (vagal augmentation; Stange, Hamilton, Fresco, & Alloy, 2016). A growing literature suggests that diminished RSA withdrawal in response to sadness induction predicts an overall worse course of depression across time and poorer response to treatment (Fraguas et al., 2007; Panaite et al., 2016; Rottenberg et al., 2005).

Overall, however, surprisingly little is known about how RSA and mood repair strategies are associated in the context of depression. And the few studies that examined RSA reactivity during the implementation of mood repair have reported inconsistent findings. For example, while one study reported that RSA reactivity did not differ between depressed versus non-depressed samples (D’Avanzato, 2013), another study found greater RSA withdrawal among depressed individuals relative to controls (LeMoult, Yoon, & Joormann, 2016). While there appear to be no studies of RSA reactivity specifically during hedonically toned mood repair, there are indications that depressed individuals do not differ from controls in RSA reactivity to positive emotional stimuli or recall of PAMs (Jin, Steding, & Webb, 2015; Kang & Gruber, 2013). Furthermore, RSA reactivity to an amusing film clip did not predict prospective depressive symptoms in a non-selected sample of undergraduates (Stange, Hamiton, Olino, Fresco, & Alloy, 2017) nor did it predict response to treatment among depressed individuals (Fraguas et al., 2007). However, the impaired hedonic capacity associated with depression-risk may come into play in not being able to sustain the RSA response to hedonically toned mood repair.

In the present study, we therefore examined the link between various extents of depression risk and problems in mood repair via responses that leverage positive affect, and the ability to sustain the effects of mood repair, using behavioral (self-report) and physiological (RSA) indices. After mood induction, subjects were instructed to implement two hedonically toned mood repair strategies, namely: 1) watching a comic film clip, and 2) savoring chocolate. We hypothesized that high-risk (compared to low-risk) offspring will be: (a) less successful at alleviating sad mood (high-risk offspring with a personal history of depression having the most difficulty) and (b) less able to maintain the salubrious effects of mood repair (high-risk offspring with a personal history of depression having the most difficulty). We operationalized success as reduced self-ratings of sadness and dysphoria after mood repair along with a return of RSA values to baseline and the maintenance of such behavioral and physiological gains.

Our sample included high-risk offspring, whose parents had early-onset mood disorders, including depression (first onset by age 14), and healthy control offspring, whose parents never had depression or mood disorders (low-risk). High-risk offspring are at an increased risk for many affective disorders, including depression, throughout their lives as compared to their low-risk peers (for a review see Merikangas & Avenevoli, 2002; Weissman et al., 1984 Goodman, 2007). A subset of our high-risk offspring had already experienced a depressive episode, thereby increasing their risk for another depressive episode. If the subjective benefits of hedonically-based mood repair are less enduring among offspring at high risk for depression than among emotionally healthy controls, this may contribute to problems in attenuating sadness and thereby become a risk factor for eventual mood disorder. Thus, characterizing the mood repair problems of high-risk individuals has practical implications for the design of interventions to prevent depressive disorders.

Method

Participants

The current study enrolled 210 offspring of parents who, in a previous investigation, were ascertained to have had either childhood-onset mood disorder or no history of any major psychiatric disorder: the parents were recruited between the years 1996 and 2004 for a longitudinal program project that examined the correlates and eventual outcome of childhood-onset mood disorder (Forbes, Miller, Cohn, Fox, & Kovacs, 2005; Miller et al., 2002).

As described in detail elsewhere (Forbes, Fox, Cohn, Galles, & Kovacs, 2006), parents of the high-risk group had to meet DSM criteria for a depressive disorder episode (45 with unipolar depressive disorder and 17 with bipolar disorder) with onset by age 14. In the current study, parents with depression histories and non-depressed control parents were assessed for newly emerging psychopathology via the Structured Clinical Interview for DSM-IV (SCID; First et al., 1995) administered by trained professional clinicians. Previously non-depressed control parents who have developed lifetime depression (n = 15) were removed from the current study.

From among the 100 high-risk offspring, 27 already had histories of major depressive disorder (high-risk/DEP) while 73 offspring were free of personal depression histories (high-risk). From among the 75 low-risk offspring, 4 had personal histories of major depressive disorder and thus were removed from the current study. Therefore, low-risk offspring (n = 71) were free of personal depression histories and had parents free of any lifetime major psychiatric diagnoses. Ethnic background of the 171 offspring who participated in the study was 67% Caucasian, 20% African American, and 13% multi-racial. The sample included 99 families, of which 50% contributed more than one offspring to the study.

Given that several participants failed mood induction (see Results section for details), the final sample of 9- to 22-year olds (M = 15.72 years, SD = 3.20) included 64 never-depressed offspring at high familial depression risk (high-risk, aged: 9–22 years), 25 offspring with similar familial background and personal histories of depression (high-risk/DEP, aged: 12–22 years), and 62 never-depressed offspring at low familial risk of depression (controls, aged: 10–21 years).

Participants in the high-risk/DEP group were, on average, 1.7 years older than their high-risk never depressed peers, t(148) = 2.20, p = .03, and 1.6 years older than their low risk peers, t(148) = 2.52, p = .01. Similarly, the three groups differed in the number of siblings who participated in the study, with those in the high-risk/DEP group contributing the fewest siblings (12% had more than one sibling) as compared to those in the high-risk (47% had more than one sibling) and low-risk groups (49% had more than one sibling), χ²(8) = 18.74, p = .02. However, the groups did not differ in sex, χ²(2) = 3.33, ethnic distribution, χ²(8) = 7.45, and maternal education level, χ²(4) = 7.09, (See Table 1). High-risk offspring with both personal and familial histories of depression evidenced higher rates of anxiety disorders, χ²(2) = 27.56, p < .01, but similar rates of behavioral disorders, including attention-deficit/hyperactivity disorder; ADHD, and oppositional defiant disorder than their low-risk peers, χ²(2) = 5.49, p = .06, (see Table 1).¹ Finally, 11 participants from the high-risk/DEP group, 7 from the high-risk group and 4 from the low-risk group were prescribed psychotropic medication at the time of the study, χ²(2) = 21.36, p < .001.

Table 1.

Selected Demographic and Mood Rating Variables of Final Sample

		Groups
Variable	low-risk n = 62	high-risk n = 64	high-risk/DEP n= 25
Age (SE)	15.42 (.40)	15.48 (.40)	17.09 (.63)
Sex (% Male)	50	53	32
Anxiety Dx (%)	7	19	56
Behavioral Dx (%)	10	25	24
Maternal education (I/II/III)	0/9/53	7/8/49	2/4/19
Mood BL (M, SE)	1.55 (.16)	1.59 (.15)	2.30 (.25)
Mood SF (M, SE)	3.60 (.21)	3.70 (.21)	4.46 (.33)
Mood MR (M, SE)	1.46 (.13)	1.55 (.13)	1.60 (.21)
Mood Post-repair (M, SE)	1.50 (.15)	1.73 (.15)	2.06 (.24)
RSA BL (M, SE)	6.48 (.15)	6.78 (.16)	6.07 (.34)
RSA SF (M, SE)	6.35 (.15)	6.61 (.14)	5.88 (.27)
RSA MR (M, SE)	6.56 (.14)	6.66 (.15)	6.04 (.27)
RSA Post-repair (M, SE)	6.53 (.13)	6.75 (.14)	6.09 (.24)

Note. Anxiety Dx = history of some type of an anxiety disorder; Behavioral Dx = history of some type of a behavioral disorder (e.g., ADHD, ODD). Maternal education = Hollingshead education code: I = did not complete high-school, II = High-school graduate; III = at least partial college education. Mood = Assessment points of sad mood; RSA = Respiratory Sinus Arrythmia. BL = Baseline, SF = Sad film, MR = Mood repair.

Laboratory Procedures

The protocol included various stimuli/tasks, lasting about 35 minutes. The portions of the protocol reported herein include a baseline-3-minute paced breathing period, which provided the baseline RSA value. During this period, subjects listened to a soft tone and were instructed to breathe in when the tone was rising, breathe out when the tone was falling, and pause between breaths when there was no tone. The tone pattern was set to induce a respiratory frequency of 12 cycles per minute with a normal fractional inspiratory ratio of 40%. The paced breathing period was followed by sad mood induction, instructed mood repair, and a subsequent post-repair rest period, that served to assess maintenance of mood repair. E-Prime software (Psychology Software Tools, Inc.) was used to present protocol stimuli on a computer monitor and to collect affect ratings (described below). Sadness was induced via a 2.5-minute clip from The Champ (Gross & Levenson, 1995) with instructions to “Try to imagine how the people are feeling. Try to get into the feeling” in order to deepen the mood induction (Joormann, Siemer & Gotlib, 2007). The mood repair tasks, each of which lasted about 3 minutes, involved either viewing a film clip of a comedic scene from Mr. Bean (happy film) or mindfully savoring chocolate (two pieces). Both the Champ and Mr. Bean film clips have been used for mood induction in prior studies (Gross & Levenson, 1995; Rottenberg et al., 2002). During the following 3-minute, post-repair rest epoch, participants were asked to sit quietly and wait for the next set of instructions to appear on the monitor.

After the sadness induction, participants were randomly assigned to one of the mood repair conditions. Subjects who were assigned to Mr. Bean (n = 77) were informed that they were about to watch a short video and that “we hope that you will have fun watching it”. Subjects who were assigned to consume chocolate (n = 74), were handed two pieces of chocolate to place in the mouth, one at a time (to make the chocolate last the appropriate time period) and were instructed to “Gently move the chocolate around with your tongue; notice if the chocolate is smooth; notice if the chocolate is creamy; and notice how the chocolate tastes”. Each chocolate piece was shaped as a wafer, approximately 2.1cm in diameter, and contained 41% cocoa. The choice of the chocolate was based on pilot testing with a sample of convenience of 6 young adults: they tasted six different brands of chocolates and rated the taste, texture and melting rate on a scale of 1 (‘terrible’ or very bad) to 5 (‘excellent’ or very good). The selected chocolate was rated highest on taste and texture (Mtaste = 3.67, Mtexture = 3.83) and the lowest on melting rate (Mmelting = 1.83).

Assessment and Measures

Affect ratings.

Computerized self- ratings of affect were collected at baseline (prior to paced breathing), after sad mood induction, after the mood repair tasks, and after the post-repair rest epoch, using Likert scales from 1 (‘very little or not at all’) to 8 (‘very much or extremely’). Affects rated included sad, unhappy, cranky, scared, glad, and happy. Our dysphoria index was the average of the ratings of the “sad” and “unhappy” items, similar to the approach taken by others (e.g., Joormann et al., 2007).

Assessment of RSA.

Mindware BioLab software (MindWare Technologies, Ltd., Gahanna, OH) was used to record electrocardiogram (ECG) signals. Electrodes were placed on participants’ lower left and upper right rib cage and in accordance with published guidelines (Berntson et al., 1997), the ECG signals were acquired at 1000-Hz sampling rate. Respiration was collected via bands placed around the abdomen and calculated by the Mindware HRV 3.0.21 software (MindWare Technologies, Ltd., Gahanna, OH). R-wave markers in the ECG signal were processed with the MAD/MED artifact detection algorithm using Mindware HRV 3.0.21 software (MindWare Technologies, Ltd., Gahanna, OH). Signals were visually inspected, and suspected artifacts manually corrected (Berntson et al., 1997). The interbeat interval (IBI) series was resampled in equal 250 ms intervals, linearly detrended, and tapered using a Hanning window. Heart rate variability (HRV) was calculated using Fast Fourier transformation analysis of the IBI series, and RSA was defined as the log transformed high frequency (HF) power band of HRV (.12-.40 Hz range; see Berntson et al., 1997). Hereafter we refer to HF-HRV as RSA, since HF-HRV is the power band of HRV that occurs in the typical range of respiration. RSA was calculated for each epoch separately (i.e., baseline paced-breathing, after sad mood induction, after the mood repair tasks, and after the post-repair rest epoch.

Offspring’s symptoms and diagnostic status.

Offspring were assessed for current and life-time history of psychiatric disorders using the multi-informant Interview Schedule for Children and Adolescents-Diagnostic Version (ISCA-D), which is an extension and modification of the Interview Schedule for Children and Adolescents (ISCA; Sherrill & Kovacs, 2000) and has good inter-rater reliability (Kiss et al., 2007). Final DSM-IV based diagnoses were derived by clinical consensus among experienced clinicians, using the best-estimate framework (Maziade et al., 1992).

Of the offspring, 92 had their full diagnostic evaluation at an earlier time than the laboratory procedure (M = 1.49 years earlier, SD = .41). Thus, at the time of the laboratory procedure, depression status was confirmed using the Follow-Up Depression Scale (FDS), which was developed from the ISCA (Sherrill & Kovacs, 2000). It consists of 18 depressive symptoms representing various symptom of depression diagnoses (each rated on a 0- to 3-point scale) and has been shown to have good psychometric properties (internal consistency, α = .94; interrater reliability, intraclass correlation coefficient = 0.97; Yaroslavsky, Rottenberg, & Kovacs, 2013).

Statistical Analyses

All analyses were conducted in SAS v 9.3 software (SAS Institute Inc., 2013). Repeated measures analysis of covariance (rANCOVA) was used to examine group differences in mood ratings following mood induction, repair, and post-repair rest epochs, controlling for sex. We controlled for sex since it significantly predicted sad affect (see manipulation check section). Mood repair type (savoring chocolate vs. watching a happy film clip) was also entered as a between-subjects factor given that it is unknown whether the two strategies differ in efficacy. Significant interactions between group and mood repair outcomes were probed via planned contrasts. Similar analyses were used to examine group differences in RSA following mood induction, repair, and post-repair rest epochs controlling for age, sex and psychotropic medication that are known to influence RSA (Alvares, Quintana, Hickie, & Guastella, 2016; Gentzler, Rottenberg, Kovacs, George, & Morey, 2012; Rottenberg, 2007; Silveto, Drago, & Ragonese, 2001).

As the participants ranged in ages, the potential effect of age was examined as: (1) a continuous covariate and (2) a categorical covariate that reflected the key developmental stages of childhood (12 or younger, n = 17), adolescence (13–17, n = 94), and adulthood (18–22, n = 40). Multilevel modeling was employed to account for familial dependence among participants, using restricted maximum likelihood. Heterogeneous compound symmetry was assumed for within-subject correlation of repeated ratings, and empirical (a.k.a. “sandwich estimator”) standard errors were computed for model parameters and least squares means (Liang & Zeger, 1986).

Results²

Manipulation Check

Given that mood repair requires the presence of some degree of dysphoria, we examined the effectiveness of our manipulation. Altogether 12% (n = 20) of the offspring denied any feelings of dysphoria following sad mood induction (including 9 low-risk offspring). Offspring who failed mood induction were similarly distributed across the risk groups, χ²(2) = .56, p = .76 and tended to be 1.72 years younger (M = 13.99, SD = 2.81) relative to their peers, F(1, 169) = 7.54, p = .01; they were excluded from subsequent analyses.

In the remaining sample, the sad film clip induced significant dysphoria relative to baseline levels, F(1, 206) = 152.90, p < .001, which did not differ as a function of risk-group membership, F(2, 206) = .04, p = .96. High-risk offspring with depression histories tended to report higher baseline levels of dysphoria than did their low-risk peers, M_{high-risk/DEP} = 2.32 vs. M_low-risk = 1.54, F(1, 206) = 2.59, p = .08. But the three groups reported similar levels of dysphoria following sad mood induction, M_{high-risk/DEP} = 4.47, M_high-risk = 3.74, M_low-risk = 3.56, F(2, 206) = 2.10, p = .13. Of the demographic characteristics, only sex emerged as a significant predictor of sad mood induction outcome, F(9, 654) = 42.52, p < .001: girls reported higher levels of dysphoria than did boys (M = 4.21 vs. M = 3.35).

As expected, sad mood induction elicited RSA withdrawal relative to the paced breathing task, M_PB = 6.96 vs. M_Induction= 6.30, F(1, 201) = 9.04, p = .01, the extent of which was similar across the three subject groups, F(2, 201) = .04, p = .96. Though the three groups significantly differed in baseline RSA, F(2, 201) = 3.08, p = .05, and at a trend level following the sad mood induction, F(2, 201) = 2.51, p = .08, the differences were driven by the high rate of psychotropic medication use among those with depression histories (44% vs 15% among high-risk and 6% among low-risk offspring), and became non-significant when psychotropic medication use was controlled, Fs(2, 199) = 2.06–2.24, ps = .11 - .13.

Mood Repair Outcomes

Group (low-risk, high-risk, high-risk/DEP) by epoch (sad mood induction, mood repair) rANCOVAs tested our first hypothesis regarding the effect of risk status on change in sad affect and RSA levels from mood induction to mood repair. Mood repair type was added as a between-subjects factor, and age was entered as a covariate. The effect of sex was statistically controlled. In addition, we controlled for psychotropic medication use when modeling RSA levels.

Affect Ratings. Results showed a significant within-subject effect of epoch, (F(1,196) = 8.43, p = .004), which suggests that subjective levels of dysphoria changed across our study epochs. Contrary to expectation, the epoch-by-group interaction was not significant, F(2, 196) = 0.43, p = .65 (high-risk/DEP: M_{Mood repair - induction} = 2.56, SE = 0.41; high-risk: M_{Mood repair - induction} = 2.18, SE = .20; low-risk: M_{Mood repair - induction} = 2.15, SE = .18). Further, the groups did not differ in levels of dysphoria following mood repair, F(2,196) = .21, p = .81 (high-risk/DEP: M_{Mood repair} = 1.53, SE = 0.21; high-risk: M_{Mood repair} = 1.57, SE = 0.13; low-risk: M_{Mood repair} = 1.45, SE = 0.14), which suggests that they evidenced similar mood repair gains.

However, compared to levels of dysphoria at baseline, high-risk offspring with depression histories garnered more benefit from mood repair than did the other two groups, F(2, 196) = 3.98, p = .02 (high-risk/DEP: M_{Mood repair - baseline} = −.65, SE = .19; high-risk: M_{Mood repair - baseline} = .05, SE = .07; low-risk: M_{Mood repair - baseline} = .10, SE = .08). These results were independent of mood repair type, which did not significantly influence mood repair outcomes across, between-, or within-subjects, Fs = 0.01 – 2.02, ps = .16-.99. Age was unrelated to dysphoria levels, irrespective of whether it was examined as a continuous, Fs(1, 196) = .21-.62, p = .43-.65), or a categorical variable, F2(2, 194) = .16-.30, p = .74-.86.

RSA. Contrary to expectations, RSA levels remained essentially the same across sad mood induction and mood repair, F(1, 195) = .13, p = .72, nor was there evidence of a significant epoch-by-group interaction, F(1,195) = 1.37, p = .26 (high-risk/DEP: M_{Mood repair - Induction} = −.10, SE = .14; high-risk: M_{Mood repair - Induction} = −.05, SE = .08; low-risk: M_{Mood repair - Induction} = −.22, SE = .06). From among the between-subjects effects, age (as a continuous variable) and risk-group status reached significance, F(1–2, 195) = 3.43–6.15, p = .02-.03, older offspring evidencing lower RSA levels (B = −.07) and high-risk/DEP group (M = 6.01, SE = 0.23) evidencing lower average RSA levels than their high-risk peers (M = 6.61, SE = 0.13), t(195) = 2.58, p = .01. However, these differences were reduced to non-significance when psychotropic medication use was controlled, F (2, 193) = 2.83, p = .06. The three groups did not significantly differ in RSA levels during the mood repair epoch, F(2,193) = 2.35, p = .10 (high-risk/DEP: M_{Mood repair} = 5.92, SE = 0.26; high-risk: M_{Mood repair} = 6.57, SE = 0.20; low-risk: M_{Mood repair} = 6.42, SE = 0.20). Neither age (continuous or categorical), sex, nor mood repair type were related to changes in RSA levels across epochs, Fs(1–2, 191–195) = 0.13–2.94, p = .09-.72.

The Maintenance of Mood Repair Outcomes

A group (low-risk, high-risk, high-risk/DEP) by epoch (mood repair, post-repair) rANCOVA tested our second hypothesis regarding the maintenance of mood repair benefits by the end of the post-repair rest epoch. As with the prior models, mood repair type was entered as a between-subjects factor and age was entered as a covariate. The effect of sex was statistically controlled. In addition, we controlled for psychotropic medication use when modeling RSA levels.

Affect Ratings. Results showed a marginal within-subject effect of epoch, F(1,196) = 3.40, p = .07), and, as predicted, a notable epoch-by-group interaction, F(2, 196) = 3.85, p = .02 (see Figure 1). Specifically, both high-risk groups experienced increased dysphoria from mood repair to post-repair epochs, relative to the levels of dysphoria just after mood repair (high-risk/DEP: M_{Post-Repair – Repair} = 0.47, S.E. = 0.16, t(196) = 2.93, p = .004; high-risk: M_{Post-Repair – Repair} = 0.18, S.E. = 0.08, t(196) = 2.13, p = .04). However, dysphoria level of low-risk subjects did not change from the mood repair to the post-repair epoch, t(196) = 0.80, p = .43 (M_{Post-Repair – Repair} = 0.04, S.E.= 0.05).

Figure 1.

Group effects on dysphoria index following mood induction, mood repair, and the post-repair period. Error bars reflect +/−95% confidence intervals generated using standard errors of the mean of each group at each specific epoch.

*p < .05. Reflects group difference across the mood repair and post-repair epochs.

Planned contrasts revealed that the increase in dysphoria from mood repair to the post-repair epoch for high-risk/DEP subjects was greater than among their low-risk peers (t(196) = 2.58, p = .01). However, the change in dysphoria from mood repair to the post-repair epoch for the high-risk group did not significantly differ from the change among their high-risk/DEP peers, t(196) = 1.46, p = .15, or among the low-risk controls, t(196) = 1.64, p = .10. Further, the three groups did not significantly differ in their levels of dysphoria at the post-repair assessment point, F(2,196) = 1.62, p = .20 (high-risk/DEP: M_Post-repair = 1.99, SE = 0.29; high-risk: M_Post-repair = 1.79, SE = 0.15; low-risk: M_Post-repair = 1.49, SE = 0.13). Further, dysphoria levels following the post-repair rest period did not significantly differ from baseline levels, F(1, 196) = .70, p = .40, nor across the three risk groups, Fs (1, 196) = 1.45–1.55, ps = .22-.24. The effects were independent of mood repair type, which did not significantly influence mood repair outcomes across between- nor within-subject levels, Fs = 0.03–1.06, ps = .30–.97). Age was also unrelated to dysphoria levels, irrespective of whether it was examined as a continuous, F(1, 196) = .03–1.09, p = .30-.87) or a categorical variable, F(2, 194) = .03-.53, p = .59-.97.

RSA. Contrary to expectation, RSA levels remained stable across the two epochs (mood repair, post-repair, F(1, 195) = 1.33, p = .25, and were unrelated to risk groups status, F(2,195) = .85, p = .43 (high-risk/DEP: M_{Post-repair - Repair} = .09, SE = .17; high-risk: M_{Post-repair - Repair} = .08, SE = .07; low-risk: M_{Post-repair - Repair} = −.04, SE = .06). Further, the three groups did not significantly differ in their RSA levels during the post-repair rest epoch, when psychotropic medication use was controlled, F(2, 193) = 2.38, p = .10 (high-risk/DEP: M_Post-repair = 6.25, SE = 0.22; high-risk: M_Post-repair = 6.72, SE = 0.14; low-risk: M_Post-repair = 6.45, SE = 0.13). Age (as a continuous variable) was inversely related to RSA levels across both tasks, B = −.08, F(1, 193) = 7.23, p = .008. However, change in RSA levels also was unrelated to age (continuous or categorical), sex, or mood-repair type, Fs(1–2, 191–195) = 0.01–1.19, p = .28-.94.

Discussion

The present study tested the hypotheses that offspring at high risk for depression are less effective in repairing sad mood via strategies that leverage hedonic capacity and less able to maintain the resultant mood repair gains relative to their low risk peers. Contrary to our first hypothesis, all offspring at high familial risk for depression were just as able to repair experimentally induced sad mood as were control peers. However, in support of our second hypothesis, high-risk offspring, as a group, were not able to sustain the benefits of hedonically toned mood repair. Moreover, subjects with both a personal and familial history of depression evidenced notably greater increases in dysphoria from mood repair to the post-repair epoch than did their low-risk peers. While high-risk, never-depressed offspring did not significantly differ from the other two groups in the ability to sustain the benefits of hedonically toned mood repair, they did experience increased dysphoria at the post-repair point and at a level that was intermediary to the levels of dysphoria of the other two offspring groups. Surprisingly, RSA, our physiological marker of emotion regulation, did not vary as expected across the mood repair and maintenance epochs in any of the subject groups and thus failed to reflect the subjective changes in mood that were reported.

Our finding of comparable mood repair outcomes among high- and low-risk offspring adds to a growing literature that depression-prone individuals are able to attenuate sad mood in a variety of ways when instructed to do so in laboratory settings (Beauregard et al., 2006; Dillon & Pizzagalli, 2013; Ehring, Tuschen-Caffier, Schnülle, Fischer, & Gross, 2010; Erk et al., 2010; Greening, Osuch, Williamson, & Mitchell, 2014; Heller et al., 2009). In fact, our high-risk offspring with depression histories garnered even more benefit from mood repair than did the other two groups with respect to baseline levels. Therefore, skill deficits cannot account for the mood repair problems of depression-prone individuals.

On the other hand, our results appear to be inconsistent with reports that depression-prone individuals have difficulty in repairing their mood via recalling PAM, which is another regulatory strategy that leverages positive affect (e.g., Joormann & Seimer, 2004; Joormann et al., 2007; Kovacs et al., 2015). The apparently inconsistent findings may reflect that while our hedonic mood repair tasks involved the processing of relatively simple sensory (visual and gustatory) input, PAM use requires access to mood-incongruent memories and is therefore more complex (Witheridge, Cabral, & Rector, 2010). Relatedly, our mood repair tasks required minimal effort. In contrast, mood repair via PAM involves considerable cognitive effort that may overtax depression-prone individuals’ resources (Dalgleish et al., 2007; Werner-Seidler & Moulds, 2011; Williams et al., 2007).

While the foregoing explanations should be tested in future studies, there is evidence that dysphoric adults and high-risk children invest more cardiac and neural effort in order to perform as well as do healthy individuals (Brinkmann & Gendolla, 2007; Pérez-Edgar, Fox, Cohn, & Kovacs, 2006). The possibility that the cognitive, elaborative aspects of PAM may challenge the skills of depression-prone individuals is partly supported by our findings that the PAM of unaffected (high-risk) siblings of depressed youth were less positive and less elaborated than were the PAMs of controls (Begovic et al., 2017). High-risk individuals, relative to controls, also show differential brain activation in regions recruited during recall of PAM (Young et al., 2013). Therefore, future studies of individuals at high risk for depression should consider comparing the efficacy of PAM to other hedonically toned but less effortful mood repair strategies.

Our finding that depression-prone offspring were able to benefit from mood repair that leveraged positive affect but failed to sustain it converges with the empirical literature (Dietz et al., 2008; Durbin, Klein, Hayden, Buckley, & Moerk, 2005; Kovacs et al., 2015; Olino et al., 2011; Olino et al., 2014; Shaw et al., 2006). Indeed, neuroimaging studies found short-lived neural activity in brain reward regions among depression-prone individuals exposed to hedonic experiences (Admon & Pizzagalli, 2015). The fact that offspring with both familial and personal depression histories had the most attenuated maintenance periods of positive affect suggests that having a depressive episode amplifies the harmful effects of familial depression risk. Our findings, and those of others, also reinforce the importance of examining affective dynamics in the study of emotion (Davidson, 1998). Given the apparent difficulties of depression-prone offspring to sustain the benefits of mood repair that relies on hedonic capacity, it remains to be seen if interventions that aim to stimulate positive affect in order to reduce depressed mood (e.g., McMakin, Siegle, & Shirk, 2011) can serve as ways to forestall depressive disorders.

While we view our results as evidence that depression risk is associated with difficulty in maintaining mood repair gains that involve leveraging positive affect, the findings lend themselves to at least one alternative explanation: It is possible that during the post-repair rest period, depression-prone offspring engaged in some habitual maladaptive mood repair response such as rumination (Aldao, Nolen-Hoeksema, & Schweizer, 2010; Garnefski & Kraaij, 2006; Kovacs et al., 2009) that serves to upregulate negative affect. Future research should include de-briefing of subjects as one way to explore the just noted possibility.

The use of physiological indices of emotional processing, which are thought to be less subject to the biases associated with self-report, hold the possibility of a more comprehensive understanding of how individuals experience and regulate affect, along with clarifying mixed findings in the self-report literature. However, using RSA as our physiological index, we did not find evidence that hedonic experience following sadness induction is associated with physiological recovery (i.e., return to baseline levels), as has been previously suggested (Fredrickson & Levenson, 1998; Fredrickson, Mancuso, Branigan, & Tugade, 2000). It is possible that adolescents, who comprised a notable segment of our sample, experience positive and negative emotional states as equally physiologically challenging (Gilbert, Nolen-Hoeksema, & Gruber, 2016). Furthermore, consistent with findings that various features of emotions are associated with small but important differences in physiological outcomes (Kreibig, 2010), it could be that neither our mood induction nor our hedonic mood repair strategies were strong or specific enough to instigate measurable perturbations in physiological processes.

There were consistent group differences in RSA levels throughout the experimental protocol. However, the significant differences were greatly driven by the high rates of psychotropic medication use among offspring with depression. The latter finding is consistent with a body of work that ties psychotropic medication use to alterations in the autonomic nervous system (Lederbogen et al., 2001; Licht et al., 2008; Licht, de Geus, van Dyck, & Penninx, 2010).

It is not entirely surprising that we found a lack of coherence between behavioral and physiological measures of psychological processes. A meta-analysis showed that cognitive mood repair strategies such as reappraisal were associated with non-significant effects on physiology but significant effects on self-reported emotional experience (Webb, Miles, & Sheeran, 2012; see also Troy, Shallcross, Brunner, Friedman, & Jones, 2018). Similarly, Troy et al., (2018) found that different mood repair strategies exert differential effects on various response systems (e.g., emotional versus physiological reactivity).

Our study had several positive features. The rate of mood induction failure was similar to that reported by others (e.g., Singer & Dobson, 2007; 2009) and we excluded subjects who did not respond to the negative mood induction, mirroring recent recommendations for best practices in emotion research (Rottenberg, Kovacs, & Yaroslavsky, 2017). We extended the scope of hedonic regulatory responses that have been examined and we used mood repair strategies, which have face and ecological validity and are suitable to a wide age range. As well, we examined both self-report and physiological outcomes.

However, our study also had limitations. Importantly, because of sample size constraints, we did not have a non-hedonically-based mood repair response condition. Therefore, it remains unclear to what extent the results reflect deficits in hedonic capacity versus other processes that were activated during mood repair (e.g., attention). Furthermore, we cannot disentangle the extent to which the hedonic experience, or the mindfulness component of savoring chocolate was responsible for the mood repair outcomes in that condition. Another limitation of our study is the wide age range of our sample and the dissimilar number of cases across the various age groups. Although we did not find a relationship between age and mood repair, others have shown that adults evidence more enduring neural correlates of emotion regulation than do their adolescent counterparts (Silvers, Shu, Hubbard, Weber, & Ochsner, 2015). Future studies of offspring that extend the upper age range into adulthood would help to clarify the impact of age on mood repair. Finally, we note that the effect sizes we are reporting are small in magnitude. However, mood repair is known to be influenced by multiple physiological, behavioral, cognitive, and environmental factors that act in concert with one another (e.g., Aldao, 2013). Therefore, small effect sizes are expected for any set of predictors, as we previously noted (Kovacs et al., 2016).

Overall, however, our findings suggest that vulnerability to depression is linked to difficulties in maintaining the subjective benefits of mood repair that leverages positive affect, which may be a contributor to the widely reported and well documented mood repair problems of depression-prone people. If our findings are confirmed by future studies, it may be useful to examine the utility of leveraging intact mood repair ability and improving mood repair outcome maintenance in the treatment of depression-prone offspring.