Don’t Look Back in Anger! Responsiveness to Missed Chances in Successful and Nonsuccessful Aging

See allHide authors and affiliations

Science  04 May 2012:
Vol. 336, Issue 6081, pp. 612-614
DOI: 10.1126/science.1217516


Life-span theories explain successful aging with an adaptive management of emotional experiences like regret. As opportunities to undo regrettable situations decline with age, a reduced engagement into these situations represents a potentially protective strategy to maintain well-being in older age. Yet, little is known about the underlying neurobiological mechanisms supporting this claim. We used a multimodal psychophysiological approach in combination with a sequential risk-taking task that induces the feeling of regret and investigated young as well as emotionally successfully and unsuccessfully (i.e., late-life depressed) aged participants. Responsiveness to regret was specifically reduced in successful aging paralleled by autonomic and frontostriatal characteristics indicating adaptive shifts in emotion regulation. Our results suggest that disengagement from regret reflects a critical resilience factor for emotional health in older age.

There is an emerging literature showing that aging is accompanied by an increase in positive affect while negative emotions like anger decline (13). Motivational life-span theories propose that such emotional changes depend on adaptive shifts in emotional goals and strategies (2, 4, 5). One powerful predictor for these shifts is our subjective sense of remaining lifetime. When people perceive their time as “running out,” they increasingly focus on regulating emotional states to optimize well-being (2). Lifelong experience with emotional situations probably facilitates the use of emotion-regulation strategies (6). Shifts toward emotional control strategies may also compensate for declining resources and opportunities in old age (4, 5). The life-span theory of control (4, 5) holds that these compensatory mechanisms lead to a reduced emotional engagement in regrets in aging. Because opportunities to undo consequences of regrettable behavior decline with age, responsiveness to regret becomes a critical factor for life satisfaction in older adults (7, 8). At a young age, active attempts to overcome regretful situations may help to optimize future behavior. As people age, however, a more successful strategy may be disengagement (5, 79). Regret fundamentally depends on the feeling of responsibility (10). Therefore, disengagement should be facilitated by an attenuated attribution of internal control. Indeed, compared to young adults, healthy elderly report lower levels of internal-control attributions, which is associated with a reduced intensity of long- and short-term regrets (9).

What are the psychophysiological modulators of an age-adapted management of regretful experiences in emotionally healthy aging? The experience of regret is elicited by counterfactual thoughts (11), that is, comparing “what is” with “what could have been.” Thus, regret is typically induced by presenting people not only with the actual outcomes of their choices but also with missed opportunities (i.e., alternative outcomes that could have been obtained had the choice been different) (1215). The influence of missed opportunities on future choices can be regarded as an indicator of regret responsiveness. A neural structure involved in regret processing is the ventral striatum, which codes the value not only of experienced rewards but also of missed opportunities (12, 1517) and has been suggested to modulate age-related changes in decision-making (18). It is densely connected to the anterior cingulate cortex (ACC) (19), a key player at the interface between emotion and cognition (20) that may regulate emotional experiences in successful aging (21).

We investigated the behavioral and neuronal responses to regretful events in emotionally successful aging using a sequential decision task that included feedback about alternative outcomes (Fig. 1A). On each trial, an array of eight boxes was presented, where seven boxes contained a gain (“gold”) and one contained a loss (“devil”). Boxes could be opened from left to right. At any stage, volunteers could either open the next box or stop and collect the gains acquired so far in this round. Exposing the randomly distributed devil ended the trial, and all gains from this round were lost. If volunteers decided to stop and collect their gains, the position of the devil was revealed, indicating how far they could have safely continued (“missed chance”).

Fig. 1

(A) Outline of the possible choice outcomes. (B) Fisher z scores of the correlation between devil position and risk-taking in the next trial. Only in young and old depressed volunteers, the position of the devil in win trials (missed chance) predicted the risk taken in the next round (P < 0.007). No such effect was observed for the devil position in loss trials (P > 0.83), suggesting a selective influence of missed opportunities on future choice behavior. *P < 0.05 (two-tailed).

In a functional magnetic resonance imaging (fMRI) study, 21 healthy young (mean age 25.4 years) and 20 older (65.8 years), as well as 20 late-life depressed (65.6 years) participants without cognitive deficits were scanned while playing 80 rounds of the task. Elderly participants diagnosed with late-life depression (first episode after the age of 55) were included to investigate whether the responsiveness to regret in older adults represents a resilience factor against emotionally unsuccessful aging.

We tested several predictions: (i) compared with the two other groups, emotionally healthy older adults will demonstrate a reduced behavioral response to missed opportunities; (ii) ventral striatal brain responses will differ between optimal and nonoptimal outcomes in young and depressed older subjects but not in the healthy elderly; and (iii) only emotionally healthy older adults will employ regret-limiting strategies involving cortical structures previously implicated in emotion regulation [i.e., ACC (21)].

Consistent with our previous findings in an independent young sample (12), the degree of missed chances predicted subsequent risk-taking behavior in healthy young and depressed old volunteers (Fig. 1B). Specifically, the greater the number of missed chances in the current trial, the more risk these subjects took in the next trial, despite the fact that consecutive rounds were explicitly independent. This behavioral response to missed chances was not observed in healthy elderly [young > oldhealthy: t(39) = 2.08, P < 0.05; olddepressed > oldhealthy: t(38) = 2.62, P < 0.02]. To see whether these behavioral results were merely driven by the position of the devil, we compared the predictive value of devil position on subsequent choice behavior between gain and loss trials. Results show that our findings were limited to gain trials (Fig. 1B). Raw behavioral data are given in the supplementary materials (fig. S1A).

Enhanced responsiveness to missed chances in young and depressed old volunteers correlated with a decreased blood oxygen level–dependent (BOLD) signal in bilateral ventral striatum when a missed chance was presented (nonoptimal: missed chance > 0) compared with when there was no missed chance (optimal: missed chance = 0; tested contrast: optimal > nonoptimal, P < 0.05corrected) (Fig. 2A). As reflected by the significant group interaction in this contrast (P < 0.05corrected), healthy elderly showed a different pattern of ventral striatal response to missed opportunities. Similar signal increases were observed in all gain trials, regardless of whether there was a missed chance or not. Only during real loss trials (devil), the BOLD signal decreased significantly (tested contrast: nonoptimal > devil). This kind of neural differentiation between nonoptimal gains and actual losses was not observed in old depressed and young subjects (P < 0.05corrected). Inspection of striatal signals in depressed and young subjects indicates that the signal decrease in response to nonoptimal wins was greater relative to actual loss trials (Fig. 2B and supplementary materials). Even though missed chances represent only fictive losses, they are coded similar to real losses in subcortical dopaminoceptive brain structures (12, 16).

Fig. 2

Group differences in ventral striatal BOLD signal in response to missed opportunities. (A) As compared with healthy elderly, young and older depressed volunteers showed differential striatal responses to optimal and nonoptimal outcomes (i.e., a reduced BOLD signal during the presentation of a missed opportunity). Bar plots show contrast estimates of the peak voxel from the comparison olddepressed + young > oldhealthy (peak coordinates x, y, z: –10, 0, –6, Z = 4.66 ; 12, 2, –10, Z = 4.34). (B) Unlike old depressed and young subjects, healthy older adults showed an enhanced striatal signal during nonoptimal win trials relative to actual loss trials (i.e., trials in which the devil was unpacked). Bar plots show the contrast estimates for the peak voxel from the comparison oldhealthy > olddepressed + young (peak x, y, z: –10, 0, –6, Z = 3.59 ; 12, 4, –10, Z = 3.73). Activations are overlaid on the mean structural image of all participants (display threshold P < 0.001 uncorrected). *P < 0.05; +P < 0.10; small volume corrected for multiple comparisons.

Analysis of BOLD signals in the ACC revealed further group differences in response to missed chances: As compared with young and depressed participants, healthy older adults showed an increased ACC engagement during nonoptimal relative to optimal trials (P < 0.05corrected) (Fig. 3). This result most likely reflects the recruitment of cognitive control processes that support the regulation of regretful experiences.

Fig. 3

Anterior cingulate activation during the presentation of missed opportunities relative to win trials without a missed chance was observed in healthy elderly but not in old depressed and young subjects. The left plot depicts the contrast estimates for the peak voxel from the comparison oldhealthy > olddepressed + young (peak x, y, z: –20, 44, –4, Z = 4.32). Activations are overlaid on the mean structural image of all participants (display threshold P < 0.001 uncorrected). *P < 0.05; small volume corrected for multiple comparisons.

After showing that only healthy older adults have a reduced regret responsiveness, in our second study we investigated how this effect is related to the emotional valuation of regret and autonomic responses. For this purpose, an independent group of 15 emotionally healthy older adults was compared with 15 newly recruited adults with late-onset depression. Volunteers performed the devil task while we recorded autonomic responses (i.e., skin conductance and heart rate). Because our paradigm likely evokes “hot” emotions of regret like anger (22), subjects rated how satisfied/angry they were with the current outcome and how much they regretted their last decision leading to that outcome. These ratings were given after each trial.

A positive relation between missed chances and subsequent risk-taking was observed in depressed subjects but was lacking in healthy older adults [t(28) = 2.74, P < 0.01], thus replicating the behavioral findings of the fMRI study (raw data are given in fig. S1B). Regret and anger ratings were positively correlated with the magnitude of missed chances in both groups (see supplementary materials for details). Like depressed subjects, healthy elderly were generally happier and less regretful when the missed opportunity was small. However, only in depressed subjects did the intensity of regret predict subsequent choice behavior [t(28) = 2.42, P < 0.02] (Fig. 4A). Healthy elderly were obviously able to apply counterfactual thinking during conscious ratings. However, their emotions did not affect future decisions.

Fig. 4

(A) Only in depressed elderly, the regret intensity predicted subsequent choice behavior (Fisher z scores, *P < 0.05, two-tailed). (B) A differential skin conductance response to missed opportunities was limited to depressed older adults. (C) There was a significant interaction of trial outcome and group on heart rate responses, mainly due to an attenuated acceleration in response to missed opportunities in depressed subjects.

Skin conductance responses (SCRs) were modulated by the presence of a missed opportunity only in older depressed participants [t(14) = 2.43, P < 0.03]. Subjects from both groups were less satisfied with nonoptimal relative to optimal outcomes, but only in depressed participants was this accompanied by a reduction in SCRs (Fig. 4B). The overall reduced SCR reactivity in these subjects replicates previous findings in depression (23) and thus clearly underlines the validity of our group classification. Most importantly, SCR results confirmed a further lack of differential physiological reactivity in healthy elderly to the nonoptimal outcome. Along similar lines, there was a significant group by trial type interaction regarding the discriminatory heart-rate response to missed opportunities [F(1,28) = 4.37, P < 0.05]. Inspection of the 5 poststimulus seconds indicates that this effect was mainly due to a stronger relative heart-rate decrease after nonoptimal outcomes in the depressed group (Fig. 4C). Thus, like BOLD responses, autonomic nervous system reactivity to missed opportunities discriminated successfully aged from nonsuccessfully aged subjects.

As in the fMRI study, all subjects were extensively screened for cognitive and affective functioning (see supplementary materials). Furthermore, in this second study, we characterized experiences with life regrets using a standardized interview (8). Here, as compared with depressed subjects, successfully aged participants reported lower emotional intensity, as well as lower intrusive thoughts associated with their life regrets (see supplementary materials).

Using a multimodal approach, our data show that emotionally healthy aging is associated with a reduced responsiveness to regretful events. Humans tend to avoid the negative feeling of regret (11, 24) by changing future decision behavior. In the current study, these effects were restricted to healthy young and depressed older subjects. Importantly, the lack of such behavioral responses to fictive losses in healthy older adults had no negative effect on overall performance. Similar to other sequentially independent guessing tasks (25), the consideration of missed opportunities was irrelevant for performance in the independent rounds. Nonetheless, people process such fictive losses similar to real losses, reflected by a very similar neural response of dopaminoceptive brain structures like the ventral striatum (12, 16). Decision-making is modulated by emotions, as indicated by changes in the autonomic nervous system (26). Feelings of regret and disappointment are related to changes in skin conductance (13) and heart rate declines in “frustrating” situations (27). In our study, both markers responded differentially to missed opportunities in depressed elderly who clearly demonstrated behavioral consequences of regret. All these responses were attenuated in healthy older adults, in accordance with their lack of a behavioral effect of regret on subsequent choices. When explicitly confronted with the alternative outcome, healthy elderly are able to think counterfactually. Therefore, the lack of behavioral and psychophysiological responses to regret in healthy aging is not due simply to a deficit in comparing actual with counterfactual outcomes. Recent epidemiological findings on large data sets have documented an increase in emotional well-being from middle age onward, whereas the experience of anger declines (1, 3). Motivational life-span theories have assumed that focusing on positive and disengaging from negative experiences in late life depend on an active cognitive engagement (2, 4, 5). Our finding of ACC activation when healthy elderly are confronted with the negative experience of regret supports this notion and emphasizes that healthy older adults have changed compared with young and old depressed subjects. The ACC is known for its role in cognitive control of emotions and has been related to an attentional positivity bias in successful aging (21). We propose that healthy elderly use cognitive control strategies like external attribution (9) to successfully disengage from regretful experiences. For instance, it could be speculated that healthy older adults remind themselves that the devil is determined by factors they cannot influence (i.e., chance and the experimenters), whereas depressed elderly blame themselves for the outcome. Our results also indicate that a more youth-like processing of missed chances is evident in nonsuccessful aging such as late-life depression. Older depressives exhibit ACC dysfunction (28), and regret is related to reduced emotional well-being and depressive symptoms in aging (7, 8). The lack of an age-adapted management of regret experiences may thus represent a risk factor for highly prevalent late-life depression. Disengagement from regret experiences at a point of life where the opportunities to undo regrettable behavior are limited may be a protective strategy to maintain emotional well-being and thus can be seen as a resilience factor. The ability to adapt to changes in life circumstances likely depends on several factors, including response styles (e.g., rumination versus distraction), cognitive flexibility, and lifelong emotional experience. Training these abilities using age-specific cognitive-behavioral programs might therefore be a promising approach to preserve affective health in old age.

Supplementary Materials

Materials and Methods

Figs. S1 to S3

Tables S1 and S2

References (2936)

References and Notes

  1. Acknowledgments: S.B. is supported by the German Research Foundation (DFG, BR 2877/2-1). C.B. is supported by the Bundesministerium für Bildung und Forschung (Bernstein 01GQ0912). S.G. is supported by the DFG (GRK 1247, CINACS).
View Abstract

Navigate This Article