Rumination, especially brooding rumination, defined as the tendency to focus on negative aspects of one’s self or negative interpretations of one’s life, is a well-known risk and key factor of depression in midlife (Whitmer and Gotlib, 2013). It predicts the onset of new episodes (Nolen-Hoeksema, 2000; Abela and Hankin, 2011) and the severity and duration of existing depressive episodes (Nolen-Hoeksema, 1991). Rumination is suggested to consist of the two components brooding and reflection, whereas brooding is thought to be the more maladaptive response style (Treynor et al., 2003; Armey et al., 2009; Gibb et al., 2012). Neurobiologically, brooding rumination has been related to dysfunctions in fronto-limbic circuits, mostly reflected by an increased, “hyperresponsive” amygdala activation (Siegle et al., 2002; Cooney et al., 2010; Mandell et al., 2014) while dorsolateral prefrontal brain regions were frequently found to be decreased in activity in response to emotional stimuli (Ray et al., 2005; Brassen et al., 2012). Both findings have been interpreted in terms of an impaired cognitive control of heightened emotional reactivity (Ray et al., 2005; Woody et al., 2015). Moreover, the consistently increased engagement of the subgenual anterior cingulate cortex (sgACC) during rest in ruminators (Bratman et al., 2015; Hamilton et al., 2015) as well as during rumination induction in depressives (Cooney et al., 2010) has been related to the maladaptive pattern of aggravated self-referential processes typically observable in ruminators (Hamilton et al., 2015).

Autobiographical memory and self-referential processing are intrinsically related (Conway and Pleydell-Pearce, 2000), and negatively biased self-schemes are thought to hinder autobiographical memory retrieval in depression (Dalgleish and Werner-Seidler, 2014) and at-risk persons in younger age (Sperduti et al., 2013; Young et al., 2016). Reduced autobiographical memory ability and specificity have been related to trait rumination in the current (Schoofs et al., 2013; Hamlat et al., 2015) and previously depressed young patients (Raes et al., 2012). Rumination induction in healthy young participants during the presentation of autobiographical memories was associated with self-reports of negative affect and increased activity in regions known to be involved in self-referential processing, including the sgACC (Kross et al., 2009). Altogether, it could be hypothesized that ruminative thinking interferes with the search or construction of specific memories. In addition, activation of current negative self-schemes might bias memory search towards negative and/or less positive memories (Conway and Pleydell-Pearce, 2000), which in turn might trigger preservative maladaptive thoughts.

Only little is known about the influence of brooding rumination on the occurrence of depression in late-life. Overall, there is evidence that rumination tendency decreases with age (Nolen-Hoeksema and Aldao, 2011; Sütterlin et al., 2012) and that compared to young adults emotionally healthy elderly are better able to disengage from negative information while focusing on positive information (Carstensen, 2006; Reed et al., 2014; Sasse et al., 2014). Intriguingly, such so-called “positivity-effect in aging” is thought to increase resilience against depression in the context of typical age-related challenges (Reed et al., 2014; Mather, 2016). On the other hand, there is also evidence that older people tend to review their past (Ingersoll-Dayton et al., 2010), so one could speculate that if older individuals have an increased brooding tendency they might carry a particularly high risk for becoming depressed when being confronted with age-specific limitations. Indeed, recent data in large samples across the life-span indicate brooding rumination to predict lower life-satisfaction especially in late life (Sütterlin et al., 2012). Other studies in this context have focused on the investigation of regret responsiveness, which is thought to be closely linked to brooding rumination (Savitsky et al., 1997; Stewart and Vandewater, 1999). They could demonstrate a particularly negative influence of regret responsiveness on current mood state even in non-depressed elderly (Lecci et al., 1994; Wrosch et al., 2005). In the same vein, we previously observed neurobehavioral correlates of regret-responsiveness to be more crucial for mood state in old compared to young age (Brassen et al., 2012). In sum, these findings suggest an existing ruminative response style in late life to represent a high-risk factor for depression.

In the present study, we investigated whether brooding tendency in non-depressed older adults is associated with neurobehavioral dysfunctions during autobiographical memory retrieval typically observed in younger depressives. Due to the discussed crucial role of positive memories for well-being in aging, we were particularly interested in brooding effects on memories in response to neutral to positive cues. To this end, we investigated an extensively characterized sample of older, non-depressed participants with varying degrees of brooding tendency using functional magnetic resonance imaging (fMRI) combined with a slightly modified version of an established Autobiographical Memory Cue Word Paradigm (Holland et al., 2011). This paradigm allows for the distinction between: (i) a construction phase, in which participants have to mentally search for specific life episodes based on an initial cue specification; and (ii) an elaboration phase, in which the constructed episode is being re-lived by accessing episodic elements (Conway et al., 2003; Conway, 2009). The expected underlying neural autobiographical memory network comprises primarily left-lateralized medial and lateral prefrontal brain regions which are critical for initiating and monitoring the retrieval process (Svoboda et al., 2006; Cabeza and St Jacques, 2007). The hippocampus (HC) is thought to coordinate specific sensory and perceptual details stored in cortical regions including precuneus, cuneus and other regions of the occipital and parietal lobes (Moscovitch et al., 2006; McCormick et al., 2015). The contributions of emotion have been linked to the amygdala. Self-referential processes, which are crucial in the present study, have been related to sgACC engagement (Cabeza and St Jacques, 2007).

Based on aforementioned findings in younger adults, we expected older participants with brooding tendency to retrieve autobiographical memories with less detailedness and less positive valence. On the neural level, we used factorial and functional coupling analyses to place a special focus on networks related to maladaptive self-referential thinking including the sgACC.

In our study, healthy older brooders demonstrated a neurobehavioral response pattern during autobiographical memory retrieval that strikingly overlaps with previously reported dysfunctions observed in young depressive adults (e.g., Dalgleish and Werner-Seidler, 2014; Young et al., 2016). Specifically, brooders needed more time to construct an autobiographical memory event and rated their memories as less detailed, possibly due to interference by ruminative thinking. Moreover, older participants with high brooding tendency perceived their retrieved memories as less positive, which might be a consequence of negatively biased self-schemes. This is supported by our neural findings of: (i) stronger engagement and coupling of a network consisting of key regions related to maladaptive self-focused thinking; and (ii) reduced activation in cortical autobiographical memory circuits.

To our knowledge, this is the first study investigating the neural relation of brooding with both autobiographical memory construction and elaboration. Our paradigm resulted in a strong activation of the autobiographical memory network, highlighting the validity of our task. During the construction phase, we found left hippocampal activity to be related to the reported detailedness and bilateral HC activity to code the valence of the memories. This is in line with previous research highlighting the role of the HC in paving the way to perceptual details of autobiographical memories (Nadel and Moscovitch, 1997; Addis et al., 2004) and in emotional processing (Robinson et al., 2015).

Participants with higher brooding tendency reported their memories as less positive and demonstrated an increased activity in the right amygdala during memory construction compared to participants with lower brooding tendency. Interestingly, recent findings indicate altered amygdala activity during autobiographical memory recall to represent both state and trait markers of depression in younger age (Young et al., 2016). More specifically, our results also fit with research indicating that brooding and associated negative mood is related to a difficulty to find specific positive memories (Williams et al., 2007). In addition, sgACC-amygdala coupling directly predicted less positive memories in brooders. It has been speculated that maladaptive self-referential processes underlying rumination hinder adequate memory retrieval and embed memories in a more negative context (Sperduti et al., 2013; Dalgleish and Werner-Seidler, 2014). Support for this assumption comes from findings of increased sgACC activation during autobiographical memory elaboration after rumination induction in healthy young adults (Kross et al., 2009). In a similar vein, sgACC-amygdala coupling during resting state fMRI in adolescence was found to be related to negative affectivity (Davey et al., 2015) and was elevated in younger depressives compared to controls (Connolly et al., 2013). As far as we know, the current study is the first that extended these findings to participants with an existing brooding tendency and to the active construction of autobiographical memories. Since we used neutral to positive cues in our paradigm, such processes rather ended in less positive than more negative memories. Given large evidence that a focus on positive memories is critical for emotional well-being in late-life (Brassen et al., 2011; Mather, 2016) and findings that younger individuals at risk for depression demonstrate neural changes during positive autobiographical memory retrieval (Young et al., 2016), reduced positivity in older brooders might put them at particularly high risk for depression.

Consistent with findings in younger depressed ruminators, brooding in healthy older adult was related to impaired autobiographical memory retrieval in terms of longer construction times and lower detail ratings (Schoofs et al., 2013; Hamlat et al., 2015). While longer construction phases might result from interference by negative, self-referential thinking, our finding of decreased activation patterns in autobiographical memory networks during elaboration might mirror reduced perception of detail richness. Especially the IFG plays an important role in appropriate information selection (Badre and Wagner, 2007), emotion regulation (Ochsner et al., 2012) and controlled retrieval, for instance by diminishing the impact of negative goal-irrelevant emotional distraction (Dolcos et al., 2013) or the voluntary up-regulation of positive emotions (Kim and Hamann, 2007). The fact that brooders showed a decreased activation of this region during memory elaboration might point to a less controlled, maybe inappropriate selection of memory details resulting rather from rigid self-schemes than appropriate and cohesive selection. The IFG has been shown to be sensitive to age-related changes especially during the elaboration of autobiographical memories (St Jacques et al., 2012). Thus, it could be speculated that such decline triggers potentially negative effects of brooding tendency in aging.

While sgACC-HC coupling was negatively related to detail ratings in non-brooders, the opposite pattern emerged for participants with high brooding tendency. Due to the small sample sizes of subgroups we can only speculate on this finding. One interpretation could be that non-brooders engaged cognitive control to suppress an impact of potentially interfering self-referential processes on detailed memory retrieval. Detailed memories in brooders, in contrast, might activate (negative) self-schemes which in turn might further promote persistency and re-encoding of distorted memories. Indeed, the sgACC and the medial temporal lobe (MTL) are strongly, reciprocally connected and their coupling is enhanced in younger patients with major depressive disorder (de Kwaasteniet et al., 2013). The testing of such causal connectivity hypotheses, however, requires larger subsamples. In any case, the association between the sgACC and regions of the MTL seems to be crucial for mood state and rumination. This might become especially relevant in the context of age-related changes in fronto-limbic network integrity (Brassen et al., 2008; Lehmbeck et al., 2008; Fjell et al., 2009). For example, the “vascular depression” hypothesis posits that white matter lesions, caused by ischemic changes, are the main cause for the frequent comorbidity between vascular disease and depression in aging (Taylor et al., 2013). Lesions are primarily located in target fronto-subcortico circuits such as the cingulum bundle (Taylor et al., 2014). Hence, one could speculate that disruptions in critical neural emotion-regulation circuits due to vascular disease hinder successful adaptation in aging. In our sample, four participants with varying degree of brooding tendency (individual scores: 7, 8, 12, 13) took low doses of antihypertensive medication which rather argues against a significant influence of vascular disease on our findings. However, since we did not use specific MRI sequences sensitive for hyperintensities (e.g., Fluid-attenuated inversion recovery [FLAIR] sequence), an impact of subtle micro-structural changes on our findings (e.g., Popa-Wagner et al., 2014) cannot be excluded.

It should be noted that this first study specifically focused on depression-like neurobehavioral patterns in older adults and not on age-effects in general, e.g., regarding autobiographical memory functions (St Jacques et al., 2012). The overlap of our findings with reports on neurobehavioral changes in younger ruminators is striking but, of course, exact common or specific patterns can only be investigated by including a young age group.

Moreover, in line with previous studies on the emotional evaluation of autobiographical memories (e.g., Greenberg et al., 2005; Bado et al., 2014) we only assessed the overall value participants assigned to their retrieved memory using well-established bipolar dimension scales (e.g., Bradley and Lang, 1994). In order to get a more complete picture of emotional and mood aspects, participants relate to previous events; in future studies scanning sessions could be followed by detailed autobiographical memory interviews in which memories have to be evaluated on different subscales related to both positive and negative affect (e.g., The Positive and Negative Affect Schedule, PANAS; Watson et al., 1988).

Altogether, our findings underline brooding tendency in aging as a critical indicator for maladaptive self-referential processes when it comes to autobiographical memory retrieval. Aging has been related to an increased tendency to review the past (Ingersoll-Dayton et al., 2010) and disengagement from unreached goals as well as a focus on the positive is crucial for well-being in late life (Heckhausen et al., 2010; Mather, 2016). Thus, ruminating upon previous decisions and counterfactual thinking might be particularly maladaptive in older age, especially in the context of age-related changes in fronto-limbic brain circuits and executive control. Recent data demonstrate beneficial intervention effects on autobiographical memory function and rumination in younger age (Raes et al., 2009, 2012; van Vugt et al., 2012; Bratman et al., 2015), e.g., by applying a memory specificity training (MEST; Raes et al., 2009) or nature experiences (Bratman et al., 2015). Targeting older peoples’ autobiographical memory ability in prevention and intervention studies might therefore be a promising approach to increase resilience against late-life depression.

SB developed the study design. SB and SS jointly designed the experiment. SS programmed the paradigm and she collected and preprocessed all data. SB conducted the final behavioral and neuroimaging analyses and prepared the manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.