The Vicissitudes of Positive Autobiographical Recollection as an Emotion Regulation Strategy in Depression

Abstract

Recalling positive autobiographical memories is a powerful way to regulate emotion and repair low mood. However, depressed individuals often derive limited emotional benefit from such positive recollection. To identify the conditions under which positive memory recall is emotionally beneficial, we investigated whether memories perceived as discrepant from one’s current sense of self have a differential impact on low mood compared to memories perceived as concordant with the current self. Depressed participants (n = 39) recalled either a self-concordant or self-discrepant positive memory to alleviate sad mood and reported reduced sadness only when the memory was concordant with, but not discrepant from, their current sense of self. In contrast, never-depressed participants (n = 40) experienced mood improvement regardless of the type of memory recalled. These results suggest that the degree to which depressed individuals can identify with the self from a past positive event is an important determinate of the emotional consequences that follow.

Keywords

depression autobiographical memory cognitive behavior therapy self-discrepancy emotion regulation

The good times of today are the sad thoughts of tomorrow.

Bob Marley

Fluctuations in mood as a response to everyday events are a ubiquitous feature of day-to-day life. Over time, people build up a set of core emotion regulation strategies that they can use to moderate emotional experiences in ways that prevent relatively minor downturns in mood developing into something more prolonged and severe (e.g., Totterdell & Parkinson, 1999). For the most part, a repertoire of effective mood regulatory strategies develops naturally, and skills are often deployed automatically without conscious effort (Gross & Thompson, 2007). However, there is substantial variability in how well individuals develop these skills and the effectiveness with which they are used. For example, individuals who experience Major Depressive Disorder (MDD) have relatively greater difficulty in regulating emotion and recovering from low mood, and this is a putative maintaining factor (for a review, see Aldao, Nolen-Hoeksema, & Schweizer, 2010). Accordingly, treatment techniques that aim to improve and develop emotion regulatory ability show promise in protecting against continued mood deterioration and depression.

Difficulties with emotion regulation can manifest in two ways: The first is in the strategy that is used, and the second is the effectiveness with which a given strategy is deployed. Research suggests that depressed individuals tend to use less adaptive strategies (e.g., rumination, catastrophizing), compared to their never-depressed peers (Ehring, Fischer, Schnulle, Bosterling, & Tuschen-Caffier, 2008; Garnefski & Kraaij, 2006; Kovacs, Rottenberg, & George, 2009). The adoption of these specific strategies is notable, both because it relates to current symptoms and because it predicts the recurrence of depression prospectively (Kovacs et al., 2009). Furthermore, even when depressed individuals employ what are regarded as more “adaptive” emotion regulation strategies, it does not necessarily follow that these strategies will confer emotional benefits for all individuals under all circumstances (Gotlib & Joorman, 2010). For example, cognitive reappraisal is considered an adaptive emotion regulation strategy, but studies have found that it is more effective in situations involving uncontrollable stressors, relative to when stressors are controllable (Troy, Shallcross, & Mauss, 2013). Therefore, the identification of circumstances under which specific emotion regulatory strategies are more helpful is critical if these strategies are to be harnessed within clinical practice in order to improve the repertoire and effectiveness of emotion regulation skills in depression.

There has been increasing interest in memory-based therapeutic approaches as a way to target depression (e.g., Dalgleish et al., 2013; Neshat-Doost et al., 2013; Raes, Williams, & Hermans, 2009). This interest has largely been due to advances in our understanding of the critical role that maladaptive autobiographical memory processes play in the onset and maintenance of depression (see Dalgleish & Werner-Seidler, 2014, for review), which has primarily emerged from the cognitive tradition of patients providing autobiographical information to provide material for therapy (Beck, 1972; Beck, Rush, Shaw, & Emery, 1979). What is novel about memory-focused approaches (“memory therapeutics”) is that they target underlying memory deficits, which, if reversed, have substantial clinical value. One strategy in particular that has received some empirical attention is the use of positive autobiographical memories to improve mood. It has been established that healthy individuals habitually use positive memories to regulate and repair sad mood (Parrott & Sabini, 1990; Rusting & DeHart, 2000). However, for individuals with depressive symptoms, recalling positive autobiographical material is more difficult (e.g., Clark & Teasdale, 1982; Lloyd & Lishman, 1975), and even when successfully recalled, the impact of such memories on mood is complex. The research investigating factors involved in the emotional consequences that follow positive memory recall in depression has produced mixed results. Some studies have found beneficial mood effects following positive recollection (Dalgleish et al., 2013; Werner-Seidler & Dalgleish, 2016), whereas in other studies this is not the case (Joormann & Siemer, 2004). Further complicating the picture is the finding that under some conditions, positive recollection can lead depressed individuals to experience worsened mood (Dalgleish & Werner-Seidler, 2014; Joormann, Siemer, & Gotlib, 2007).

One potential reason for this complex picture is that recollection of autobiographical memories can impact on current mood in multiple, potentially contradictory ways. For example, imagine remembering a happy day spent with a former romantic partner, where the relationship subsequently ended and the repercussions are an ongoing source of distress. The specific, concrete details of the memory are likely to elicit positive affect as they are decontextualized from the wider details of what eventually happened. However, reflecting on this bigger picture and on how the memory relates to one’s current sense of self could potentially be upsetting. A powerful example of this is portrayed in the film The Eternal Sunshine of the Spotless Mind. Here, a machine has been invented that can selectively erase unwanted memories from the past. The central character books a session to erase memories of a past relationship, choosing not to wipe memories of traumatic or upsetting incidents but to delete the memories of the happiest times that he spent in the relationship. Because the relationship has ended, these memories are now deeply painful and it is almost unbearable to reflect upon them.

In the case of depression, if people reflect on their current mood state with reference to a past positive experience that is discrepant from how they currently view themselves or their life, the focus on the schism in both mood and psychological state between the two time points may worsen current feelings of negative affect rather than ameliorate them. Further, reflection involving high levels of such “self-discrepancy” may prompt rumination, which is likely to further exacerbate low mood (for a review, see Watkins, 2008).

Self-discrepancy has emerged as an important psychological concept in understanding the self and in particular how discrepancies between an “actual” and an “ideal” self can produce emotional discomfort (Higgins, 1987). Consistent with this approach, discrepancy between actual and ideal selves is characterized by vulnerability to dejection-related emotions such as disappointment and dissatisfaction (Higgins, 1987; Higgins, Bond, Klein, & Strauman, 1986). This vulnerability can be hypothesized to arise from a psychological situation in which hopes and desires are unmet, thereby triggering emotional discomfort. Although this theory has been applied universally, depression has been conceptualized as arising from chronic and marked perceptions of a discrepancy between “actual” and “ideal” models of the self (Higgins, 1987; Strauman, 1989). This proposition has garnered empirical support, with depressed individuals reporting greater levels of discrepancy between their perceived actual and ideal selves relative to nondepressed individuals (Strauman, 1989), and with greater levels of self-discrepancy being linked to maladaptive, depressogenic cognitive processes, such as the retrieval of overgeneral memories (Crane, Barnhofer, & Williams, 2007; Schoofs, Hermans, & Raes, 2011) and rumination (Papadakis, Prince, Jones, & Strauman, 2006).

We can extrapolate from self-discrepancy theory and posit that it is not only the valence of the contents of a memory for a past event (i.e., whether the original event was experienced as positive or negative) that determines impact on mood but also the degree to which that memory taps into any self-discrepancy. Accordingly, a prediction that would be in line with this is that memories of ostensibly positive events might have a reduced positive impact on mood to the extent that the “self” who experienced the original event is discrepant from the current sense of self.

The aim of the current study was therefore to investigate the differential impact for depressed individuals of using self-discrepant versus “self-concordant” memories (memories that are not discrepant from the current sense of self) of positive events to enhance mood. Specifically, depressed participants were asked to retrieve an autobiographical memory of a past event that had been experienced as positive at the time of its occurrence. Participants were randomly allocated to one of two conditions where they were asked to either recall an event where they feel they are currently very much the same person relative to who they were when the event occurred (self-concordant) or an event where they currently feel disconnected from who they were at the time the event occurred (self-discrepant). Participants were asked to reflect on the ways in which they feel similar to or different from these past selves as they generated and described their memory, before rating their mood on dimensions of sadness and happiness. Although our primary focus was on sad mood and the ability to repair or reduce it, we wanted to examine negative and positive affective dimensions separately because of the potentially complex impact of autobiographical recall on components of mood discussed earlier.

We hypothesized that the recall of positive memories that were concordant with the current self would lead to the desired reduction in sad mood but that memories of self-discrepant positive events would produce either no change or a counterproductive worsening of sad mood.

As a comparison for the depressed sample, we examined the impact of recollecting these different types of positive memories in a never-depressed control group where we induced a sad mood in the laboratory. We predicted that mood would improve for participants who recalled a memory concordant with their current self. For participants recalling a self-discrepant positive memory, we did not have a specific prediction for the direction of the effect. On the one hand, it is conceivable that mood may remain unchanged or even worsen because perceptions of discrepancy produce emotional discomfort universally (Higgins, 1987). On the other hand, such discrepancy may be interpreted by the participant as reflecting natural change over time, rather than as a failure in the self that triggers intense negative emotion, and so the discrepancy may not prevent participants receiving an emotional benefit from recalling such positive memories.

Method

Participants

Participants (N = 79) composed two groups—depressed and never-depressed British adults. Participants with a primary diagnosis of MDD (n = 39) were recruited from our community-based patient panel, located at the Cognition & Brain Sciences Unit in Cambridge. Participants on this panel had previously responded to advertisements placed in local newspapers and online requesting volunteers who had a history with depression to assist with research. MDD diagnosis and history and other Axis I psychiatric diagnoses were determined by the Structured Clinical Interview for the DSM–IV (SCID-IV; First, Spitzer, Gibbons, & Williams, 1996).¹ For the current study, participants were administered the Mood Module of the SCID-IV over the phone prior to the first session. Individuals who met criteria for a current Major Depressive Episode in this phone screen were invited to participate in the study and attend a face-to-face session.

An age-matched never-depressed control group (n = 40) was recruited from the Cognition and Brain Sciences Unit volunteer panel in Cambridge, which comprises volunteers willing to assist with psychological research. Although these participants had previously indicated no history of depression, this was nonetheless confirmed using the Mood Module of the SCID-IV administered by phone prior to the session. These individuals also reported no history of any mental health problems including anxiety (as verified by the Anxiety Disorder Module of the SCID-IV). All participants were over the age of 18 and fluent in English. The remaining descriptive group data are presented in Table 1.

Table 1.

Sample Characteristics

Characteristic	Depressed (n = 39)	Control (n = 40)
Age, in years	45.15 (16.92)	38.25 (17.19)
Gender, male:female	10:29	16:24
Education^a	6:14:7:7:5	3:15:3:10:9
Employment, employed:unemployed	20:19	16:24
Ethnicity, Caucasian:Other	36:3	39:1
Percentage taking antidepressant medication	54	0
Percentage receiving psychological therapy	31	0
BDI-II	28.36 (10.34)	4.70 (4.50)
BAI	16.87 (10.42)	5.00 (5.57)
Baseline sad mood	4.76 (2.05)	1.50 (.82)
Baseline happy mood	3.74 (1.74)	6.63 (1.48)
Postinduction sad mood	4.03 (1.89)	4.50 (2.11)
Postinduction happy mood	3.85 (1.58)	4.63 (1.86)

Note. Data refer to Means (and SDs) unless specified. BAI = Beck Anxiety Inventory; BDI-II = Beck Depression Inventory II.

Self-reported highest education level attained (to age 16:to age 18:vocational courses; undergraduate:postgraduate).

Materials

Self-report symptom measures

The Beck Depression Inventory II (BDI-II; Beck, Steer, & Brown, 1996) measures depressive symptoms and severity over the past 2 weeks and was included to verify depressive symptoms.

The Beck Anxiety Inventory (BAI; Beck & Steer, 1993) is a 21-item measure used for assessing anxiety over the past week and was included to measure state anxiety symptoms.

Mood rating scale

To assess mood at different points in the experiment, we employed a 9-point Likert-type scale as previously used in this field to examine mood change (e.g., Joormann et al., 2007). Participants were required to rate how they were feeling at the moment, where 1 = not at all and 9 = very on dimensions of “sad” and “happy,” with extra items such as “distracted” and “excited” interspersed throughout to mitigate response bias. In the current study, sad mood ratings provided the primary variable of interest, but we also present analyses for happy ratings.

Film clips

Sad mood was induced in the nondepressed group using a 10-min film clip from the film Dead Poets Society. This material is widely used in the field for this purpose (e.g., Joormann et al., 2007) and depicts the suicide of a young boy. Participants are instructed to allow themselves to be affected by the material they are watching. Depressed participants were not administered a mood induction, as they were already in a state of low mood, and so it was not necessary to administer an induction designed to worsen mood. Instead, a neutral film clip used previously for this purpose (Werner-Seidler & Moulds, 2013) depicting a do-it-yourself (or DIY) instructional video about how to lay a floating floor was presented so that the groups were matched on exposure to a video clip. Participants were asked to attend to the material they were watching as best they could for the duration of the clip. This method of inducing a sad mood in the nondepressed group and having a neutral clip for the depressed group was intended to bring participants to the same mood level (although it is acknowledged that the nature of the mood state varies between groups). Previous studies have used this method to produce comparable mood ratings between groups (e.g., Joorman et al., 2007; Werner-Seidler & Moulds, 2012), which allows for comparisons across studies to be made. Immediately before and after viewing their allocated film clip, participants completed the mood rating scale to enable assessment of mood change. Although the depressed group was not administered a mood induction, we shall refer to the presentation of the film clips to participants, generically, as the mood induction procedure for ease of reference.

Memory task

Following the induction, participants rated their mood and were then instructed to select a positive autobiographical memory and to provide a written description of the memory, before rating mood again and completing a Memory Rating Scale. This scale was adapted for the current study from previous work in the area (e.g., Werner-Seidler & Moulds, 2012) and required participants to rate their memory on a 9-point scale measuring positivity, vividness, intensity, control over emotion at the time the event occurred, and vantage perspective. Participants also reported memory age in years.

Prior to completing the memory task, participants had been randomly allocated to either the self-discrepant or self-concordant memory condition. In the concordant condition, participants were asked to identify a memory in which they felt they were very much the same person now as they did at the time the event occurred. The specific instructions were as follows:

Please provide a brief description of a positive event that has occurred within the past 5 years (but not in the past year), which made you feel very happy at the time. The memory that you identify must be for a specific event that lasted for less than 1 day. In coming up with your positive memory, we would like you choose a memory where the person who experienced the original event feels like the person who you are today. That is, you feel similar to who you were at the time the event occurred.

In the discrepant condition, participants were asked to select a memory that involved a version of themselves from which they currently felt disconnected or discrepant. The specific instructions were as follows:

Please provide a brief description of a positive event that has occurred within the past 5 years (but not in the past year) that made you feel very happy at the time. The memory that you identify must be for a specific event that lasted for less than 1 day. In coming up with your positive memory, we would like you choose a memory where the person who experienced the original event feels like a completely different person from whom you are today. That is, you feel different now compared to whom you were at the time the event occurred.

Participants who had difficulty with generating a memory were given several prompts in a scripted fashion, reiterating that the memory had to be specific and positive. If participants sought clarification on whether a memory they generated met these criteria, they were provided feedback. To preclude the possibility that these two sets of instructions would produce systematic differences in the temporal distance from the memory (because older memories might be expected to produce greater self-discrepancy), memory age was limited to within the past 5 years but not within the past year (to minimize the likelihood of recency effects). After the memory task was completed, a manipulation check question asking participants to indicate the degree to which they felt like the same person now as they had been at the time the event occurred (again scored on a 9-point Likert-type scale) was administered.

Procedure

Eligible participants were invited into the laboratory to take part in a study examining imagination and memory. Participants were tested individually by the experimenter in a quiet testing room. After providing written, informed consent, participants completed the BDI-II and BAI, were administered the mood ratings (baseline mood), and then viewed the film clip. Immediately after the film clip, participants completed a second set of mood ratings (postmood induction) and were randomly allocated to the concordant or discrepant memory recall condition. They were then instructed to identify either a concordant or discrepant specific positive event that had occurred within the past 5 years but not the past year. Participants had as much time as they needed to identify their memory and were assisted by the experimenter as required. Although the experimenter was not blind to condition or group, the support offered was scripted, structured, and consistent across participants. Participants then provided a brief written description of the memory and completed the third mood rating (postmemory mood), the Memory Rating Scale, and finally the manipulation check item. Participants were then thanked for their time and fully debriefed. During the debrief, any participants who reported significant distress as a consequence of the memory exercise had the option to watch a comic Fawlty Towers film clip (n = 3) and/or to spend time discussing their emotions with the experimenter (a clinical psychologist) until their mood returned to at least baseline levels (n = 0). The session lasted approximately 45 to 60 min, and all participants were reimbursed at a rate of £6 per hour for taking part.

Results

Participant characteristics

Demographic information and characteristics of the analysis sample are presented in Table 1. Chi-square and ANOVA analyses were carried out and established that, across the groups and conditions, participants did not differ in terms of age as a function of group (p = .08) or condition (p = .91), nor was there an interaction (p = .54). Gender was distributed equally across group (p = .23) and condition (p = .23), and there were no differences across groups and conditions for education level (p = .51 and p = .73; Fisher’s exact test), employment status (p = .37 and p = .82), or ethnicity (p = .36 and p = 1.00; Fisher’s exact test). There were the expected differences in percentage of participants taking antidepressant medication (χ² = 29.34, p < .001) or receiving psychological treatment (χ² = 14.50, p < .001), but this did not differ as a function of condition for either medication (χ² = 2.94, p = .13) or psychological treatment (χ² = 3.36, p = .12). Specifically, 54% of depressed participants were taking antidepressant medication, and 31% were receiving psychological therapy. ANOVAs were conducted on depression and anxiety symptom measures, and, as expected, there was no main effect of memory condition (p = .13 and p = .51, respectively) nor an interaction (p = .06 and p = .38). However, the depressed group reported greater symptom levels for both depression and anxiety relative to the never-depressed group—depression: F(1, 75) = 233.74, p < .001, η_p² = .76; anxiety: F(1, 75) = 39.47, p < .001, η_p² = .35. In terms of sad mood ratings at baseline, ANOVAs with group and condition as between-subject factors showed no main effect of condition (p = .59) and no interaction (p = .27), but there was a main effect of group, F(1, 75) = 85.94, p < .001, η_p² = .53, with participants in the depressed group reporting feeling more sad relative to their nondepressed peers. This pattern was inversed for happy mood ratings, with no main effect of condition (p = .82) and no interaction (p = .82) but a main effect of group, F(1, 75) = 61.48, p < .001, η_p² = .45.

Mood change following induction

To verify the effects of the induction, we first conducted a mixed-model ANOVA with group (depressed, never-depressed) and condition, (concordant, discrepant) as the between-subjects factors, time (baseline, postinduction) as the within-subjects factor, and sad mood ratings as the dependent variable. There was a main effect of time, F(1, 75) = 29.25, p < .001, η_p² = .28, a main effect of group, F (1, 75) = 21.38, p < .001, η_p² = .22, and a Time × Group interaction, F(1, 75) = 53.93, p < .001, η_p² = .42, but no main effect of condition (p = .92) and no other interactions (all ps > .27). To break down the significant Time × Group interaction, follow-up paired-samples t tests were conducted and showed that, as anticipated, participants in the control group reported increased sad mood in response to the sad mood induction, t(39) = 8.81, p < .001, d = 1.40, whereas those who were currently depressed showed no change in sad mood in response to the neutral film clip, t(38) = 1.43, p = .17. Importantly, as hoped, following the mood induction, the two groups did not significantly differ in their sad mood ratings, t(77) = .43, p = .67.²

Memory characteristics

To examine whether there were differences in the features of the memories recalled by participants, we conducted a MANOVA on the Memory Rating Scale data with group (depressed, never-depressed) and memory condition (concordant, discrepant) as between-subjects factors and memory positivity, vividness, intensity, control, vantage perspective, and age entered as univariate dependent variables (see Table 2). There were no significant main effects of group (Wilks’s λ = .92), F(6, 70) = .95, p = .46, η_p² = .08, or condition (Wilks’s λ = .89), F(6, 70) = 1.91, p = .09, η_p² = .14, and no significant Group × Condition interaction, although this did reach trend level (Wilks’s λ = .85), F(6, 70) = 2.07, p = .07, η_p² = .15. To ensure that reported memory features were not an artifact of current mood state, we repeated the analysis but this time entering sad mood and happy mood ratings (taken immediately after the mood induction) as covariates. The effects remained unchanged (all ps > .05, minimum p value = .07). Importantly, average memory positivity was high (M = 7.89, SD = 1.31) and memories were highly vivid (M = 7.46, SD = 1.31), were emotionally intense (M = 6.38, SD = 1.96), were controlled (M = 5.95, SD = 2.23), and were typically viewed from a first-person, field perspective (M = 6.96, SD = 2.19). On average, events from the recalled memories occurred 2.6 years ago.

Table 2.

Qualities of the Recalled Memories on the Memory Rating Scale

	Depressed (n = 39)		Never-Depressed (n = 40)
Quality	Concordant (n = 19)	Discrepant (n = 20)	Concordant (n = 20)	Discrepant (n = 20)
Positivity	7.80 (1.44)	7.55 (1.67)	8.30 (.86)	7.90 (1.11)
Vividness	7.32 (1.60)	7.30 (1.30)	7.45 (1.39)	7.75 (.97)
Intensity	6.57 (2.34)	5.65 (2.11)	6.40 (1.90)	6.90 (1.25)
Control	6.79 (2.18)	6.00 (2.41)	6.10 (1.89)	4.90 (2.15)
Field perspective	6.67 (2.23)	6.69 (2.39)	7.88 (1.45)	6.81 (1.91)
Memory age, years	2.71 (1.18)	2.98 (1.12)	2.64 (.99)	2.78 (1.13)

Note. Data are means (SD). Ratings presented out of 9, where a higher score reflects a greater level for that variable.

Memory recall

The manipulation check item was first examined to verify whether the concordant/discrepant memory manipulation was successful. We conducted an ANOVA on discrepancy ratings with group (depressed, never-depressed) and memory condition (concordant, discrepant) as between-subjects factors. Results indicated a main effect of memory condition, F(1, 75) = 17.36, p < .001, η_p² = .18, a main effect of group, F(1, 75) = 9.55, p < .003, η_p² = .11, but importantly, no significant Condition × Group interaction, F(1, 75) = .07, p = .80. Follow-up independent-samples t tests on discrepancy ratings were conducted and showed that, collapsed across condition, nondepressed participants were more able to identify with themselves in their memory than depressed participants were (Depressed: M = 4.36, SD = 2.75; Never-depressed: M = 6.05, SD = 2.48), t(77) = 2.87 p = .05, d = .65), and collapsed across group, those in the concordant condition reported being able to identify with the self from the memory to a greater degree than those in the discrepant condition (Concordant: M = 6.35; SD = 2.30; Discrepant: M = 4.10, SD = 2.69), t(77) = 4.01, p < .001, d = 1.17. This indicates that the memory instructions were successful in eliciting memories that were perceived as either concordant with, or discrepant from, the current self.

To test our core hypothesis about the effects of memory recall on mood, we first calculated standardized residual change scores for sad and happy mood by regressing postinduction mood ratings onto postmemory recall mood ratings. The use of standardized residuals to index change is optimal when examining change independent of baseline values, and this approach has been consistently adopted in studies of cognitive and mood change in depression (Kuyken et al., 2010; Raes et al., 2009; Segal et al., 2006). For ease of comprehension, however, the raw descriptive data are presented in Figure 1.

Fig. 1.

Mean (+1 SE) for sad mood ratings before and after the memory recall task. ND = never-depressed; DEP = currently depressed.

The standardized residuals indexing postinduction to postmemory recall sad mood change were then analyzed using ANOVA, with group (depressed, never-depressed) and memory condition (concordant, discrepant) as between-subjects factors. There was a main effect of group, F(1, 75) = 18.92, p < .001, η_p² = .20, a main effect of memory condition, F(1, 75) = 11.38, p = .001, η_p² = .13, and the predicted Group × Condition interaction, F(1, 75) = 8.51, p = .005, η_p² = .10. To deconstruct the interaction, independent samples t tests were conducted and showed that for never-depressed participants sad mood change did not differ as a function of memory condition (t < 1). However, for those in the depressed group, sad mood change was dependent on the type of memory recalled, t(37) = 3.87, p < .001, d = 1.20.³ Comparisons using paired-samples t tests of the raw data scores for the sad mood ratings at postinduction and postmemory indicated that depressed participants reported significantly reduced sadness only when the memory they retrieved was concordant with their sense of self, t(18) = 2.95, p = .009, d = 0.86, but not when it was discrepant from their sense of self, t(19) = 2.04, p = .06, when sad mood actually increased numerically (see Figure 1).

We repeated the above analysis using residual change indices of happy mood ratings (raw data presented in Figure 2). There was again a significant main effect of group, F(1, 75) = 11.93, p < .05, η_p² = .001, with happy mood change lower overall in the depressed group (this was consistent across both memory conditions, t < 1), but no significant main effect of condition, F(1, 75) = 2.12, p = .15, nor an interaction (F < 1). There was thus no support for an effect of memory condition on the degree of increased happiness following memory recall.⁴ We followed up the significant main effect of group by examining the postinduction and postmemory happy rating scores to elucidate the patterns in the data. Using paired-samples t tests for each group separately, collapsed over memory condition, we found that both the depressed, t(38) = 2.55, p = .02, d = .41, and never-depressed, t(39) = 6.64, p < .001, d = 1.05, groups reported significant improvements in happy mood.⁵

Fig. 2.

Mean (+1 SE) for happy mood ratings before and after the memory recall task. ND = never-depressed; DEP = currently depressed.

Comparing the effects of memory recall on happy and sad mood

Given the suggestion of a differential impact of memory recall on sad compared to happy mood, which differed between depressed and nondepressed participants, we compared statistically the impact of memory recall on sad versus happy mood ratings. A mixed-model ANOVA was conducted, with group (depressed, never-depressed) and memory condition (concordant, discrepant) as between-subjects factors, mood (sad, happy) as the within-subjects factor, and standardized residual mood change scores as the dependent variable. The analysis revealed significant Mood × Group, F(1, 75) = 20.92, p < .001, η_p² = .22, Mood × Condition, F(1, 75) = 7.74, p = .007, η_p² = .09, Condition × Group interactions, F(1, 75) = 4.14, p < .045, η_p² = .05, and a borderline-significant Mood × Condition × Group interaction, F(1, 75) = 3.79, p = .055, η_p² = .05.

This three-way borderline interaction suggests that the aforementioned differential effect on sadness versus happiness in the depressed sample may be distinct from the pattern shown by nondepressed individuals. To examine this further, we first compared the effects on sadness versus happiness in the depressed sample alone. A mixed-model ANOVA was conducted, with memory condition (concordant, discrepant) as the between-subjects factors, mood (sad, happy) as the within-subjects factor, and standardized residual mood change scores as the dependent variable. The analysis revealed a significant Mood × Condition interaction, F(1, 37) = 7.81, p = .007, η_p² = .17, suggesting that the differential impact of memory on sad versus happy mood was contingent on the memory type recalled, with sad and happy mood change aligned for self-concordant memories but somewhat divergent for self-discrepant memories. We repeated this analysis for the never-depressed controls and found no significant Mood × Condition interaction (F < 1). In sum, these analyses suggest that for those who had never experienced depression, mood change aligned (i.e., reduced sadness and increased happiness; Figs. 1–2) following memory recall, which was independent of memory type. However, this appeared to be less clearly the case for those in the depressed group, with mood change aligning for concordant memories but diverging to some extent for discrepant memories.

Discussion

A key feature of depression is a difficulty in regulating emotion, and it was the aim of the current research to verify whether perceptions of self-discrepancy play a role in the effectiveness of a specific emotion regulation strategy—positive memory recall. Broadly consistent with the ideas discussed in self-discrepancy theory (Higgins, 1987; Higgins et al., 1986; Strauman, 1989), among depressed participants, we expected memories that were concordant with current self-views to reduce sad mood, whereas discrepant memories were predicted to have no such effect, leaving sad mood unchanged or counterproductively exacerbated. For individuals without a history of depression, we expected that participants would similarly be able to repair an induced sad mood by retrieving a positive autobiographical memory if it was perceived as concordant with their current sense of self, whereas we did not have a specific hypothesis with respect to the directional effect of discrepant positive recollections.

Results provided support for these hypotheses. First, never-depressed individuals reported a reduction in sad mood following the recall of positive personal memories regardless of whether they were perceived as concordant with, or discrepant from, their current sense of self. Although self-discrepancy theory posits that the gulf in perception between the actual and ideal self may produce disappointment or dissatisfaction, for never-depressed individuals, the discrepancy and associated emotions were not sufficient to mitigate the reparative effect of recalling a positive self-discrepant memory on sad mood. We expect this is because discrepancies are unlikely to be toxic and to trigger negative emotion in this group. Rather, it is likely that such self-discrepancy is interpreted as reflecting a natural change over time, thus allowing reflection on past positive events in these individuals to still trigger a positive emotional experience. This is also in line with previous studies establishing that healthy individuals frequently draw upon past positive memories to regulate and recover from experiences of sad mood (e.g., Rusting & DeHart, 2000).

The impact of memory recall on mood for depressed individuals was more complex, with participants reporting reduced sadness only after recalling a positive autobiographical memory perceived as concordant with, but not discrepant from, the self. This pattern was less evident for happy mood, as there was no significant difference in happy mood change as a function of memory type in the depressed group and supplementary analyses showed that this divergence in the trajectories of happy and sad mood change as a function of memory type for those with depression was significant.

The finding that self-discrepant memories did not alleviate sad mood (and numerically, at least, exacerbated sad mood) in the depressed group is consistent with the results reported by Joorman et al. (2007), who found that depressed individuals recalling a positive personal memory report worsened mood. Our data suggest that even though depressed individuals were able to access and recall positive autobiographical material, the affective consequence and therefore the overall effectiveness of this as an emotion regulation strategy seem to depend, at least in part, on the perceived discrepancy between current and past representations of self.

The current study has implications regarding how material from memory might be best used to regulate mood by depressed individuals, particularly in a therapeutic context. Retrieving any positive memories in the service of mood regulation can be difficult for those with depression. However, recent work on the use of mnemonic aids to scaffold autobiographical recollection in depression (Dalgleish et al., 2013; Werner-Seidler & Dalgleish, 2016) provides a platform for the use of positive memories in mood regulation, and the current data suggest that such endeavors will reap a greater therapeutic benefit when the focus is on specific positive memories that are concordant with the current self. It will be important for practitioners to ensure that positive memories elicited in the service of mood regulation be concordant with the self for optimal affective value.

It is of course important to note that, although our data suggest that low mood may be alleviated in the short-term by the recall of positive autobiographical memories that align with the view of the current self, we are not suggesting that this would be sufficient to treat the intense and persistent low mood experienced by clinically depressed individuals. Rather, we are proposing that this strategy might be used as one of a number of techniques that could be incorporated into one’s repertoire of emotion-regulation skills, deployed to offset minor downturns in mood in order to prevent the spiral toward further mood deterioration (Dalgleish & Werner-Seidler, 2014). Further, given that the identification of autobiographical counterexamples as a way to challenge common depressogenic beliefs, thoughts, and themes (e.g., hopelessness, failure) remains a core strategy employed by cognitive therapists, greater awareness of the emotional impact of these memories is needed. In this context, requesting examples of past positive events involving low levels of self-discrepancy may be needed if therapists are to successfully help those with depression to amend their beliefs in light of contradicting evidence. Finally, as is characteristic of depression, rumination or a longing for the past may be addressed by prompting depressed individuals to refocus their attention toward past times to which that they can still relate.

Several potential limitations to the current study deserve consideration. First, self-report mood ratings provided the key dependent variables of interest. We adopted this approach because participants’ experience of their mood is the critical, clinically relevant variable of interest. Using such indices is also consistent with methods previously used in this area (e.g., Joormann et al., 2007). It is also worth noting that the primary criticism leveled at self-report measures—namely, their susceptibility to demand effects—seems less likely to account for the present results, as we find a divergence of mood change trajectories between happy and sad mood in the depressed group. Nevertheless, future work may benefit from including a convergence measure of emotion (e.g., a behavioral task) to complement the self-report data.

A second issue concerns the fact that the induced low mood in the never-depressed group is likely to be qualitatively distinct to the naturalistic low mood of the depressed group. This provides a likely advantage to the control participants when faced with the task of mood repair, as induced low mood is transient in nature. That is, for the control participants, mood repair may have simply occurred as a function of time. Taken in the context of our broad research question, this would have been more of a concern if the present data had only revealed a simple main effect of group with depressed participants unable to reduce sad mood with any type of memory and never-depressed participants able to do so. However, the fact that the depressed group participants were able to reduce their sad mood with self-concordant memories mitigates this concern. Indeed, it could be argued that successfully alleviating a more chronic sad mood is a greater challenge and thus the presence of chronic low mood in the depressed group represents a particularly conservative test of our hypothesis. It remains possible, however, that in the case of self-discrepant memories, mood repair would have been more difficult for the never-depressed group participants if they had been experiencing naturally occurring and tonic low mood. However, even in these circumstances we know that naturally low mood in those with no history of depression is qualitatively different than it is in those suffering from the disorder, and so the methodological challenge would remain.

In sum, the findings from the current study suggest a critical role for self-discrepancy in determining the effectiveness with which positive autobiographical memories can be used to regulate mood among depressed individuals. We have shown that the recall of a positive memory is most likely to be emotionally beneficial if the memory is perceived as concordant with the current self. This work contributes to the development of psychotherapeutic approaches that directly target positive autobiographical memory deficits as a way to regulate emotion in depression.

Footnotes

Declaration of Conflicting Interests

The authors declared that they had no conflicts of interest with respect to their authorship or the publication of this article.

1.

Participants were initially assessed with the SCID-IV for eligibility in our research within 2 years of the current study taking place. At that time, of the 39 participants, 17 met criteria for at least one additional current comorbid diagnosis in the past (six of the comorbidities involved generalized anxiety disorder, four social anxiety disorder, three panic with agoraphobia, two obsessive-compulsive disorder, two posttraumatic stress disorder, two specific phobia, two hypochondriasis, one anorexia, and one bulimia). In addition, 10 participants met criteria for a past diagnosis of an Axis I disorder (five for alcohol dependence, two for posttraumatic stress disorder, one for substance dependence, one for anorexia, and one for bulimia).

2.

When we repeated this analysis using happy mood ratings as the dependent variable, an identical pattern emerged with the critical Time × Group interaction, F(1, 77) = 24.56, p < .001, η_p² = .24, reflecting reduced levels of happy mood in the control group as a consequence of the induction, t(39) = 6.58, p < .001, d = 1.04, and no change in happy mood following the neutral mood induction for depressed participants, t(38) = .34, p = .73.

3.

We also repeated this analysis using simple (postmemory minus postinduction) mood change scores rather than standardized residuals and obtained the same pattern of results, with the critical Group × Condition interaction being significant, F(1, 75) = 4.50, p = .03, η_p² = .06, and with independent t tests showing no mood change as a function of condition for the never-depressed group (t < 1), which was not the case for the depressed group, t(37) = 3.62, p = .001, d = .57.

4.

We again repeated this analysis using simple (postmemory minus postinduction) mood change scores and obtained the same pattern of results, with a significant main effect of group, F(1, 75) = 4.09, p < .05, η_p² = .05, but no significant main effect of condition, F(1, 75) = 1.75, p > .05, nor an interaction (F < 1).

5.

For expository purposes, even though the interactive effects were nonsignificant, we examined the patterns of happiness change in the depressed group in the discrepant and concordant conditions. There was no significant change in happy ratings for the discrepant group, t(19) = −1.15, p = .264, whereas mood improved in the concordant group, t(18) = −2.52, p = .022, suggesting that there was some mirroring of the sadness findings in these conditions.

References

Aldao

Nolen-Hoeksema

Schweizer

(2010). Emotion-regulation strategies across psychopathology: A meta-analytic review. Clinical Psychology Review, 30, 217–237. doi:https://dx-doi-org.web.bisu.edu.cn/10.1016/j.cpr.2009.11.004

Beck

A. T.

(1972). Depression: Causes and treatment. Philadelphia, PA: University of Pennsylvania Press.

Beck

A. T.

Rush

Shaw

B. F.

Emery

(1979). Cognitive therapy of depression. New York, NY: Guilford Press.

Beck

A. T.

Steer

R. A.

(1993). Beck Anxiety Inventory manual. San Antonio, TX: Harcourt Brace and Company.

Beck

A. T.

Steer

R. A.

Brown

G. K.

(1996). Manual for the Beck Depression Inventory–II. San Antonio, TX: Psychological Corporation.

Clark

D. M.

Teasdale

J. D.

(1982). Diurnal variation in clinical depression and accessibility of memories of positive and negative experiences. Journal of Abnormal Psychology, 91, 87–95. doi:10.1037/0021-843x.91.2.87

Crane

Barnhofer

Williams

J. M. G.

(2007). Cue self-relevance affects autobiographical memory specificity in individuals with a history of major depression. Memory, 15, 312–323. doi:10.1080/09658210701256530

Dalgleish

Navrady

Bird

Hill

Dunn

B. D.

Golden

A. M.

(2013). Method-of-Loci as a mnemonic device to facilitate access to self-affirming personal memories for individuals with depression. Clinical Psychological Science, 1, 156–162. doi:10.1177/2167702612468111

Dalgleish

Werner-Seidler

(2014). Disruptions to autobiographical memory in depression and the emergence of memory therapeutics. Trends in Cognitive Sciences, 18, 596–604. doi:10.1016/j.tics.2014.06.010

10.

Ehring

Fischer

Schnulle

Bosterling

Tuschen-Caffier

(2008). Characteristics of emotion regulation in recovered depressed versus never depressed individuals. Personality and Individual Differences, 44, 1574–1584. doi:10.1016/j.paid.2008.01.013

11.

First

M. B.

Spitzer

R. L.

Gibbons

Williams

J. B. W.

(1996). Structured Clinical Interview for DSM-IV Axis I Disorders (SCID-I). Washington, DC: American Psychiatric Press.

12.

Garnefski

Kraaij

(2006). Relationships between cognitive emotion regulation strategies and depressive symptoms: A comparative study of five specific samples. Personality and Individual Differences, 40, 1659–1669. doi:10.1016/j.paid.2005.12.009

13.

Gotlib

I. H.

Joorman

(2010). Cognition and depression: Current status and future directions. Annual Review of Clinical Psychology, 6, 285–312. doi:10.1146/annurev.clinpsy.121208.131305

14.

Gross

J. J.

Thompson

R. A.

(2007). Emotion regulation: Conceptual foundations. In Gross

J. J.

(Ed.), Handbook of emotion regulation (pp. 3–24). New York, NY: Guilford Press.

15.

Higgins

E. T.

(1987). Self-discrepancy: A theory relating self and affect. Psychological Bulletin, 94, 319–340.

16.

Higgins

E. T.

Bond

R. N.

Klein

Strauman

(1986). Self-discrepancies and emotional vulnerability: How magnitude, accessibility, and type of discrepancy influence affect. Journal of Personality and Social Psychology, 51, 5–15. doi:10.1037/0022-3514.51.1.5

17.

Joormann

Siemer

(2004). Memory accessibility, mood regulation, and dysphoria: Difficulties in repairing sad mood with happy memories? Journal of Abnormal Psychology, 113, 179–188. doi:10.1037/0021-843X.113.2.179

18.

Joormann

Siemer

Gotlib

I. H.

(2007). Mood regulation in depression: Differential effects of distraction and recall of happy memories on sad mood. Journal of Abnormal Psychology, 116, 484–490. doi:10.1037/0021-843x.116.3.484

19.

Kovacs

Rottenberg

George

(2009). Maladaptive mood repair responses distinguish young adults with early-onset depressive disorders and predict future depression outcomes. Psychological Medicine, 39, 1841–1854. doi:10.1017/s0033291709005789

20.

Kuyken

Watkins

Holden

White

Taylor

R. S.

Byford

. . . Dalgleish

(2010). How does mindfulness-based cognitive therapy work? Behaviour Research and Therapy, 48, 1105–1112. doi:10.1016/j.brat.2010.08.003

21.

Lloyd

G. G.

Lishman

W. A.

(1975). Effect of depression on the speed of recall of pleasant and unpleasant experiences. Psychological Medicine, 5, 173–180. doi:10.1017/S0033291700056440

22.

Neshat-Doost

H. T.

Dalgleish

Yule

Kalantari

Ahmadi

S. J.

Dyregrov

Jobson

(2013). Enhancing autobiographical memory specificity through cognitive training: An intervention for depression translated from basic science. Clinical Psychological Science, 1, 84–92. doi:10.1177/2167702612454613

23.

Papadakis

A. A.

Prince

R. P.

Jones

N. P.

Strauman

T. J.

(2006). Self-regulation, rumination, and vulnerability to depression in adolescent girls. Development and Psychopathology, 18, 815–829. doi:10.1017/S0954579406060408

24.

Parrott

W. G.

Sabini

(1990). Mood and memory under natural conditions: Evidence for mood incongruent recall. Journal of Personality and Social Psychology, 59, 321–336. doi:10.1037/0022-3514.59.2.321

25.

Raes

Williams

J. M. G.

Hermans

(2009). Reducing cognitive vulnerability to depression: A preliminary investigation of MEmory Specificity Training (MEST) in inpatients with depressive symptomatology. Journal of Behavior Therapy and Experimental Psychiatry, 40, 24–38.

26.

Rusting

C. L.

DeHart

(2000). Retrieving positive memories to regulate negative mood: Consequences for mood-congruent memory. Journal of Personality and Social Psychology, 78, 737–752. doi:10.1037/0022-3514.78.4.737

27.

Schoofs

Hermans

Raes

(2011). Effect of self-discrepancy on specificity of autobiographical memory retrieval. Memory, 20, 63–72. doi:10.1080/09658211.2011.637934

28.

Segal

Z. V.

Kennedy

Gemar

Hood

Pedersen

Buis

(2006). Cognitive reactivity to sad mood provocation and the prediction of depressive relapse. Archives of General Psychiatry, 63, 749–755. doi:10.1001/archpsyc.63.7.749

29.

Strauman

T. J.

(1989). Self-discrepancies in clinical depression and social phobia: Cognitive structures that underlie emotional disorders? Journal of Abnormal Psychology, 98, 14–22.

30.

Totterdell

Parkinson

(1999). Use and effectiveness of self-regulation strategies for improving mood in a group of trainee teachers. Journal of Occupational Health Psychology, 4, 219–232. doi:10.1037/1076-8998.4.3.219

31.

Troy

A. S.

Shallcross

A. J.

Mauss

I. B.

(2013). A person-by-situation approach to emotion regulation: Cognitive reappraisal can either help or hurt, depending on the context. Psychological Science, 24, 2505–2514. doi:10.1177/0956797613496434

32.

Watkins

(2008). Constructive and unconstructive repetitive thought. Psychological Bulletin, 134, 163–206.

33.

Werner-Seidler

Dalgleish

(2016). The Method of Loci improves longer-term retention of self-affirming memories and facilitates access to mood-repairing memories in recurrent depression. Clinical Psychological Science. Advance online publication. doi:10.1177/2167702615626693

34.

Werner-Seidler

Moulds

M. L.

(2012). Mood repair and processing mode in depression. Emotion, 12, 470–478. doi:10.1037/a0025984

35.

Werner-Seidler

Moulds

M. L.

(2013). Recalling positive self-defining memories in depression: The impact of processing mode. Memory, 22, 525–535. doi:10.1080/09658211.2013.801494