Introduction

The response styles theory (Nolen-Hoeksema, 1991) explains the process involved in the maintenance of a depressive mood. It divides behavior related to a depressive mood into two response styles: ruminative response, wherein thoughts and behaviors focus on one's depressive symptoms and their meanings (Nolen-Hoeksema, 1991), and distractive response, which involves turning one’s attention away from their depressive mood and its possible causes and consequences to more pleasant or neutral activities. According to the theory, individuals’ response to their initial depressed mood affects their psychological health. Thus, a ruminative response intensifies an existing depressed mood. In contrast, a distractive response leads to rapid mood improvement.

Experimental studies showed that a ruminative response style prolongs and exacerbates a depressed mood, and a distractive response style improves (or stabilizes) mood (Blagden & Craske, 1996; Huffziger & Kuehner, 2009; Morrow & Nolen-Hoeksema, 1990; Park et al., 2004; Wahl et al., 2021). At the trait level, a ruminative response has been almost consistently associated with higher levels of depression (e.g., Carnevali et al., 2018; Nolen-Hoeksema et al., 2008; Watkins, 2009; Whisman et al., 2020). However, several studies have emphasized that ruminative responses contribute to solving problems which cause one’s depressive symptoms (Andrews & Thomson, 2009; Sevcikova et al., 2020). The association between distractive response and depression is more obscure. Several studies indicate that higher distractive tendency predicted lower depression (e.g., Roelofs et al., 2009). However, a meta-analysis showed that the level of distractive response is not associated with depression in youth (Rood et al., 2009).

In this regard, both ruminative and distractive response styles have been suggested to have adaptive and maladaptive forms. The multidimensional nature of rumination has been discussed widely. Treynor et al. (2003) reported that the ruminative response style includes two subcomponents, namely brooding, which is passive and maladaptive rumination, and reflection, which is an adaptive form of rumination that aims at problem solving, and that the former is associated with more depression whereas the latter predicts less depression. Other studies also suggest that the adaptive form of ruminative response style is neutrally valenced, characterized by concrete thinking, and is aiming at problem solving whereas the maladaptive form is negatively valenced, characterized by abstract thinking, and is self-critical (e.g., Hjartarson et al., 2020; Shimazu & Koshikawa, 2014; Trapnell & Campbell, 1999; Watkins, 2008; Watkins & Nolen-Hoeksema, 2014).

The distractive response style has been considered as an adaptive coping mechanism in response to a depressed mood. However, it has been reported that it can lead to maladaptive consequences in some cases (Waugh et al., 2020). Wolgast and Lundh (2017) showed that distractive responses in combination with acceptance to aversive emotion is associated with a higher degree of well-being whereas distractive responses combined with avoidance is associated with a lower degree of well-being. Distractive responses for avoiding negative thoughts were suggested to maintain dysfunctional meanings of the thought contents by hindering awareness of one’s thoughts or feelings and developing of alternative interpretations of the situation (Steil & Ehlers, 2000; Watkins & Teasdale, 2004). Shimazu and Koshikawa (2014) reported that the adaptiveness of a distractive response style is affected by whether the response is a temporal diversion or aims at avoiding the problem for as long as possible. Whereas the former is negatively associated with depression, the latter is positively associated (Ishikawa, 2020; Ishikawa et al., 2018).

Considering that both focusing on and diverting from depressive mood are ambivalent coping, some factors may promote their adaptive use. One potential factor that affects adaptiveness is decentering, which is described as the capacity to take a present-focused, non-judgmental stance with respect to thoughts and feelings and to accept them (Fresco et al., 2007). Decentering can contribute toward decreasing excessive emotional responses to initial depressed mood, such as uncontrollable negative thinking about a problem or experiential avoidance. Several studies have suggested that decentering influences the adaptiveness of ruminative and distractive response styles. For example, Watkins et al. (Rimes & Watkins, 2005; Watkins & Moulds, 2005) showed that experiential self-focus, in which one focuses on concrete emotional experiences without any evaluation or judgment, fosters problem solving and reduces negative self-evaluation, as opposed to focusing on the causes, meanings, and consequences of one’s emotions. Garland et al. (2015) posited that decentering plays a key role in producing the positive effects of mindfulness, thereby promoting positive reappraisal of negative events and conscious reflection rather than negative rumination. Wu et al. (2022) demonstrated a weak association between negative life events and brooding among individuals with higher decentering levels than their counterparts. Although a few studies have focused on the mechanism underlying the association between decentering and adaptiveness of distraction, a non-avoidant distractive response has been suggested to share some components with decentering (Wolgast & Lundh, 2017).

The association between adaptiveness of response styles and decentering is reported not only in Western samples but also in Eastern adolescents. Mori and Tanno (2015) examined the mechanism of how ruminative responses affect psychological adjustment in Japanese undergraduates and found that the adaptive form of ruminative response style was associated with higher levels of decentering and lower levels of depression, whereas the maladaptive form was associated with low decentering and high depression. Lo et al. (2014) showed that decentering was decreased by inducing negative rumination whereas increased by experiential self-focus in dysphoric undergraduate students or depressed patients in Hong Kong. Furthermore, a cross-sectional study in Japanese university students (Ishikawa et al., 2018) reported that adaptive forms of ruminative and distractive response styles were negatively associated with depression, whereas maladaptive forms were positively associated, and the level of decentering mediated all these associations. These findings suggest that both ruminative and distractive response styles play adaptive roles when they are associated with higher levels of decentering and play maladaptive roles when associated with lower levels, regardless of the difference between Western and Eastern cultures.

However, the direction of influence between decentering and adaptive/maladaptive response styles has not yet been clarified. Some studies have assumed that decentering decreases maladaptive rumination (Garland et al., 2015; Parmentier et al., 2019; Wu et al., 2022), whereas others have suggested that rumination influences decentering (Ben Salem & Karlin, 2023; Lo et al., 2014). Although decentering has been suggested to mediate the effects of an individual’s response styles on their psychological health in several studies (Ishikawa et al., 2018; Mori & Tanno, 2015), these findings are based on cross-sectional data. Thus, it is unclear whether decentering leads to the adaptive use of rumination and distraction or is produced by response styles. Therefore, the successive impacts of decentering on response styles and of response styles on decentering should be examined simultaneously using longitudinal data. If the decentering level predicts the subsequent frequencies of adaptive/maladaptive response styles, it indicates that decentering prevents the exacerbation of depressive symptoms by adjusting responses to an initial depressive mood. Thus, examining the successive relationship between decentering and response styles in a non-clinical sample is appropriate from a clinical perspective.

Considering the dichotomous characteristics of each response style is also crucial. In particular, no study, except Ishikawa et al. (2018), has directly examined the relationship between decentering and adaptive/maladaptive forms of distractive responses. For example, Jain et al. (2007) showed that one month of mindfulness meditation, which cultivates decentering, lessened the use of a distractive response style. However, the change did not mediate the improvement of psychological distress. These results may be evidence of the contradictory effects of decentering on adaptive/maladaptive forms of the distractive response. Measuring these forms separately would enable to reveal more detailed associations between decentering and distractive responses.

This study aimed to examine the successive influence of decentering on the frequencies of adaptive and maladaptive forms of both response styles through a short-term longitudinal survey in Japanese adolescents. The mediating effects of response styles on the influence of decentering on depressive symptoms were also examined. We hypothesized that the level of decentering at an earlier point in time is negatively associated with the maladaptive forms and positively with the adaptive forms of response styles at a later point in time, and it reduces depressive symptoms through response styles at a later point in time.

Method

Procedure and participants

We selected a sample of general undergraduates to represent typical Japanese adolescents. A survey was conducted among students in a psychology class at a Japanese university in Tokyo. When the questionnaires were distributed, the participants were informed that their participation was voluntary and that answering the questionnaire was seen as them giving their informed consent to participate in the study. The data were collected at two points in time (Time1 and Time2) that were four weeks apart. Only those who had completed the surveys twice were included in the analysis. A total of 126 Japanese undergraduate students (35 men and 91 women) completed both surveys. Their mean age was 19.8 years (SD = 0.8). The participants did not receive any compensation for participation.

Measures

Response styles scale (Shimazu et al., 2008)

This self-reported questionnaire was developed to measure the frequencies of adaptive/maladaptive forms of response styles when one “feels down, sad, or depressed.” The questionnaire consisted of four subscales: Negative Rumination Response (NR), which assessed the negative valenced ruminative response (e.g., “Once I have started to think about unpleasant things, I tend to keep doing it”); Rumination Response for Problem Solving (RP), which assessed the neutral valenced goal-directed ruminative response (e.g., “I consider the cause and think of what to do next”); Distraction Response for Avoidance (DA), which assessed the distractive response aimed at avoiding thought about the unpleasant problem (e.g., “I make every effort to avoid disagreeable things”); and Distraction Response for Mood Changing (DM), which assessed the temporal distractive response as a form of emotion regulation (e.g., “I do something that makes my mood good for a while”). NR and DA are postulated as maladaptive forms of each response style, while RP and DM are conceptualized as adaptive. Each subscale comprised 7 items. Items were rated on a Likert-type scale from 1 (not at all) to 4 (always). In this study, all subscales exhibited good internal consistency reliability. Cronbach’s alpha coefficients for Time1 were 0.86 (NR), 0.86 (RP), 0.88 (DA), and 0.85 (DM). Time2 alphas were 0.89 (NR), 0.91 (RP), 0.90 (DA), and 0.91 (DM).

Japanese version of the experience questionnaire (Kurihara et al., 2010)

This questionnaire was established to assess the level of decentering and rumination. We used only the decentering subscale which comprised 10 items (e.g., “I can observe unpleasant feelings without being drawn into them”). Each item is rated from 1 (not at all) to 5 (always). In this study, good internal consistency was exhibited: Time1 α = 0.88, and Time2 α = 0.86.

Japanese version of the center for epidemiologic studies depression scale (Shima et al., 1985)

This questionnaire was developed to assess depressive symptoms. It comprised 20 items focusing on depressive symptoms over the past week (e.g. “I was bothered by things that usually do not bother me”). Each item was rated from 0 (rarely or none of the time) to 3 (most or all of the time). In the present study, the internal consistency reliability was good: Time1 α = 0.86, and Time2 α = 0.89.

All measures used in the present study have demonstrated adequate reliability and validity.

Data analysis

First, Pearson’s correlation analyses were conducted to confirm whether depression was associated negatively with adaptive forms and positively with maladaptive forms of response styles and to examine the relationships between response styles and decentering. Next, we conducted structural equation modeling (SEM). The cross-lagged panel analysis was used to test the successive association between each response style and decentering. This model has been used widely to examine causal or directional influences of variables over time (Kearney, 2017), because it can examine the level and direction of influences among multiple variables controlling for each variable’s own change over time. We confirmed whether the frequency of each response style at Time1 predicted the level of decentering at Time2 or decentering at Time1 predicted response styles at Time2. Finally, mediation analyses (nonparametric bias-corrected bootstrapping method with 2000 resamples) were conducted to examine the 95% confidence interval for the indirect effects of decentering (Time1) on depression (Time2) through each response style (Time2). The correlation analyses and mediation analyses were conducted using statistical software “HAD.” The cross-lagged panel analyses were conducted using “R” statistical software with the “lavaan” package.

Results

Correlation analysis

The zero-order correlation coefficients and descriptive data at Time1 and Time2 are provided in Table 1. At each point in time, both NR and DA were significantly positively correlated with depression and negatively correlated with decentering. In contrast, both RP and DM were significantly negatively correlated with depression and positively correlated with decentering. The zero-order correlation coefficients between variables at Time1 and Time2 were also calculated (Table 2). There were significant or marginally significant correlations between response styles except for DM at one point in time and decentering at another point in time.

Table 1 Zero-order correlation coefficients and descriptive data at each point in time
Full size table
Table 2 Zero-order correlation coefficients between Time1 and Time2
Full size table

Structural equation modeling

Cross-lagged panel analyses were conducted to test which response styles and decentering precedingly influenced the other. The model included cross-lagged paths from each response style at Time1 to decentering at Time2 and from decentering at Time1 to each response style at Time2. It also included paths from each variable at Time1 to the same one at Time2 to control for their stability through the inclusion of autoregressive relationships (Hamaker et al., 2015). Response styles and decentering at Time2 were connected by an error covariance because there were significant correlations between them. The model estimated in this study was saturated and had perfect fit because of zero degrees of freedom (Fig. 1).

Fig. 1
figure 1

Cross-lagged model of response styles and decentering

Full size image

NR (Time2) was negatively influenced by decentering (Time1) at a marginally significant level, whereas the cross-lagged path from NR (Time1) to decentering (Time2) was not significant (Table 3 Model A). The R2 for NR (Time2) and decentering (Time2) was 0.38 and 0.34, respectively. There was significant error covariance between variables at Time2 (r = -0.23, p < 0.01). Similarly, DA (Time2) was negatively influenced by decentering (Time1) at a significant level, while the cross-lagged path from DA (Time1) to decentering (Time2) was not significant (Table 3 Model B). The R2 for DA (Time2) and decentering (Time2) was 0.32 and 0.34, respectively. The error covariance at Time2 was not significant (r = -0.08, n.s.). RP (Time1) showed a significant positive effect on decentering (Time2), and decentering (Time1) also positively influenced RP at Time2 (Table 3 Model C). The R2 for RP (Time2) and decentering (Time2) was 0.54 and 0.36, respectively. The error covariance between the variables at Time2 was significant (r = 0.18, p < 0.05). In contrast, there was no significant cross-lagged effect between DM and decentering (Table 3 Model D). The R2 for DM (Time2) and decentering (Time2) was 0.45 and 0.33, respectively. At Time2, the error covariance was marginally significant (r = 0.14, p < 0.10).

Table 3 The cross-lagged panel analyses of each response style and decentering
Full size table

Mediation analysis

Mediation analyses were conducted to test the indirect effects of decentering (Time1) on depression (Time2) through NR, DA, and RP (Time2). We used the bootstrapping method in which the significance of indirect effects is determined by whether 95% confidence intervals contain zero. The indirect effect through NR was significant (B = -0.22, CIbootstrap = -0.39 to -0.09, β = -0.16, SE = 0.08) and the direct effect of decentering after controlling for the indirect effect was marginally significant (B = -0.19, β = -0.14, SE = 0.10, p < 0.10). The indirect effect through DA was not significant (B = -0.05, CIbootstrap = -0.16 to 0.01, β = -0.04, SE = 0.04), while the direct effect was significant (B = -0.36, β = -0.26, SE = 0.12, p < 0.01). RP significantly mediated the effect of decentering on depression (B = -0.14, CIbootstrap = -0.28 to -0.02, β = -0.10, SE = 0.06) and the direct effect of decentering after controlling for the indirect effect was also significant (B = -0.28, β = -0.20, SE = 0.13, p < 0.05). As the cross-lagged path from RP to decentering was also significant, an additional mediation analysis was conducted to estimate the indirect effect of RP (Time1) on depression (Time2) through decentering (Time2). Decentering fully mediated the effect of RP (B = -0.33, CIbootstrap = -0.57 to -0.16, β = -0.15, SE = 0.10).

Discussion

We examined the successive relationship between adaptive/maladaptive response styles and decentering in Japanese adolescents. As hypothesized, the level of decentering predicted a lower frequency of maladaptive rumination and a higher frequency of adaptive rumination. These response styles mediated the successive influence of decentering on depression. Present results support the finding that the adaptiveness of ruminative response style is associated with decentering (Ishikawa et al., 2018; Mori & Tanno, 2015) and further suggest that an individual’s level of decentering partly determines which type of rumination is likely to be used when one experiences initial depressive mood.

Since negative rumination is considered an automatic behavior triggered by unpleasant emotions, decentering contributes toward mitigating this passive routine. Maladaptive rumination associated with depression is described as a habitual pattern of mind caused by initial depressed mood (Segal et al., 2002; Watkins & Roberts, 2020). Decentering would be efficient to be aware of the occurrence of such negative thinking patterns and reduce their effects on one’s emotions and cognition (Bernstein et al., 2015). Further, some distorted metacognitive beliefs about the benefit or uncontrollability of rumination are seen as maintaining maladaptive rumination (Cano-López et al., 2021; Papageorgiou & Wells, 2001, 2003). Decentering can soften such beliefs by allowing one to observe whether they reflect reality. Having a decentered perspective can prevent one from engaging in maladaptive ruminative responses and unpleasant experiences and instead focus on what happens in reality and what to do to solve the problem.

However, the path coefficient of the successive effect from decentering to negative rumination was no more than marginally significant. An appropriate interpretation would be that decentering is one of several factors that influence maladaptive ruminative response styles rather than a decisive one.

Another finding is that rumination for problem-solving also showed successive influence on decentering, which further mediated its effect on depression. It suggests that the use of the adaptive form of ruminative response improves depression by cultivating decentering skills. The result is consistent with Lo et al. (2014), which found that an induction of experiential self-focus increased the decentering state. A general characteristic of adaptive ruminative response styles includes neutral valence and present-moment-based processing of unpleasant information, which is closely associated with decentering. Decentering includes the process of mentally simulating oneself using a distanced perspective (Bennett et al., 2021). This would be associated with problem-solving processes such as observing the surrounding environment, specifying the cause of the problem, and identifying ways to improve the situation. Thus, adaptive form of ruminative response style and decentering can reciprocally reinforce each other.

Distraction for avoidance was found to be negatively influenced by decentering over time. However, it did not mediate the effect of decentering on depression. On the other hand, temporal distraction to refresh one’s mood was not associated with decentering successively. These findings were opposed to our hypothesis.

Accepting one’s thoughts and feelings as mere temporal inner events, a component of decentering, can play an important role in reducing the maladaptive form of the distractive response style. Avoidance behavior tends to become an overgeneralized response to low-intensity stimuli, even if it was initially the response to high-intensity emotional stimuli (Sheppes et al., 2012). Engaging in distractive response styles to avoid negative thoughts can lead to an increased use of this maladaptive response by stabilizing negative meanings and interpretations of one’s thoughts (Watkins & Teasdale, 2004). As decentering is a stance involving the acceptance of one’s overall thoughts and emotions, including aversive ones nonjudgmentally (Fresco et al., 2007), it can function to reduce the maladaptive distractive response style associated with experiential avoidance. However, the distraction for avoidance score did not mediate the successive effect of decentering on depression, as shown above. Considering that decentering indicated a direct effect on depression, other variables (e.g., ruminative response styles) may mediate the successive association between decentering and depression more strongly than did the maladaptive distractive response style. In addition, it is shown that both decentering and experiential avoidance affect maladaptive symptoms as parallel mediators (Mendes et al., 2017).

We could not find any successive relationship between decentering and distraction for mood changing. The core process of decentering differs from that of distraction. For example, Van Dillen and Papies (2015) suggested that while the efficacy of distraction depends on the reduction of bottom-up attentional selection to negative objects by devoting one’s working memory resources to some distractive stimuli, mindfulness based on decentering improves the top-down process of attention control. Although the top-down regulation inhibits avoidance behavior, it may not bring a uniform increase in the adaptive distractive response. For instance, people with medium-level decentering would be capable of using adaptive form of distractive responses because they are aware of their emotional states immediately, whereas people with high-level decentering would not need to use distractive responses because they can just accept and observe their negative moods without responding to them.

In summary, the present study is the first attempt to examine the successive influences of decentering on the adaptiveness of overall response styles. Although the results did not indicate the central role of decentering in defining the adaptiveness of response styles, decentering was found to be one of the factors mitigating the maladaptive forms of ruminative and distractive response styles.

The effects of decentering on both the adaptive and maladaptive forms of ruminative response style was typically suggested to decrease depression. This finding could provide empirical evidence for Garland et al. (2015), who argued that decentering plays a crucial role in mindfulness which promotes adaptive reflection on negative experiences rather than maladaptive rumination. Thus, one implication of the present study is that mindfulness-based interventions would allow individuals to engage in intentional goal-directed rumination rather than inhibit rumination about negative experiences.

Another finding is that adaptive rumination for problem-solving predicts a decrease in depression through heightened decentering. This might highlight the process by which rumination, used to find the cause of a negative problem, results in improved depressive mood in some cases (Andrews & Thomson, 2009). A recent study reported that analytical rumination significantly predicted decreased depressive symptoms after five weeks but did not predict the subjective likelihood of resolving one’s problems (Sevcikova et al., 2020). Rumination about aversive experiences with a specific attitude contributes to cultivating decentering at the trait level, and thus might enable individuals to improve as well as accept difficult situations. Since decentering can be cultivated by various psychological interventions not restricted to mindfulness-based interventions or CBT (Bennett et al., 2021), engaging in an adaptive form of rumination, instead of distraction, could also be an effective decentering training.

Finally, we highlight the limitations of this study and discuss the directions for future research. First, the sample size was relatively small, and the participants were limited to undergraduates at one university. Therefore, the sample in this study might not be representative of Japanese adolescents in general. Future studies should examine the generalizability of present findings to various regions and communities in Japan. Furthermore, it should also be examined whether the results are reproducible in other populations and ages or people with health problems. Second, the levels of decentering and depression were not controlled for in present study. Response styles, decentering, and depression may show different associations between clinical and non-clinical samples. There is also a possibility that the effect of each response style on depression in people with lower levels decentering is different from that in people with higher levels of decentering. Third, the adaptiveness of response styles is associated not only with depressive symptoms, but also other variables related to psychological health such as anxiety and stress-related symptoms (Bastin et al., 2015; Lackner & Fresco, 2016). The mediation effects of adaptive and maladaptive response styles on the relationship between decentering and these variables should be examined in future studies. Finally, the coefficients of cross-lagged paths in our cross-lagged panel models were relatively weak when compared to those of the autoregressive paths. It would be ascribed to, in part, the short interval between surveys at Time1 and Time2. Considering that a questionnaire measuring ruminative responses is reported to have one-year stability (Nolen-Hoeksema, 2000), it would be preferable to use a longer time interval.