Introduction
Mounting evidence suggests age-related changes in social-cognitive capacities including perception of eye gaze (Slessor et al. 2010; Ziaei et al.
2016), emotional facial expression recognition (Ruffman et al.
2008), and theory of mind (Henry et al.
2013). Compared to other social-cognitive components, empathy has received considerably less attention in aging research. Furthermore, the few existing studies on age-related differences in empathic response have almost exclusively used self-report and have captured more trait-like aspects of the construct (e.g., using the Interpersonal Reactivity Index; Davis
1983). These previous studies largely agree that aging is associated with decline in cognitive empathy, the ability to decode and understand another’s perspective (Beadle & de la Vega
2019; Henry et al.
2013). However, evidence is less clear for affective empathy, i.e., the affective sharing of another’s emotional state (Singer & Lamm
2009), for which some studies suggest no age-related differences (Bailey et al.
2018; Beadle et al.
2015) while other studies support an increase with age (Grühn et al.
2008; O’Brien et al.
2012).
Rather sparse are studies measuring state empathy, specifically in aging, for example by using experimentally induced alterations in the state of affective empathy. In particular, Sze et al. (
2012) manipulated empathic response by showing uplifting or distressing films and reported an age-related linear increase in empathic concern and personal distress in response to both types of films. Similarly, Bailey et al. (
2018) found age-related increased emotional distress and reactivity towards another’s pain; and this enhanced affective empathy predicted prosociality (i.e., willingness to help). Finally, Beadle et al. (
2015) did not find evidence for age-group differences in affective empathy or personal distress in response to cancer patients describing their experiences with the disease. This currently limited and mixed knowledge base on age effects in cognitive and affective empathy warrants additional research (see also Bailey et al.
2021; Ebner et al.
2017, for overviews).
Over the past two decades robust evidence suggests prioritized processing of positive over negative information among older (compared to younger) adults. This effect is reflected in greater attention to, and memory for, positive over negative stimuli and has been termed the “positivity effect” in aging (Mather & Carstensen
2003; Reed & Carstensen
2012; Ziaei & Fischer
2016; Ziaei, Salami, et al.
2017a,
b; Ziaei et al.
2015). The positivity effect has been interpreted as a motivational, goal-oriented shift with age. Closely related to this notion of a positivity effect in aging is the motivational empathy account, also referred to as ‘empathic concern’ (Weisz & Zaki
2018). According to this account, an observers’ motivation/social goals can manifest in the reduction or intensification of empathy towards a target.
To date, understanding of the impact of valence on empathy in aging is still very limited. First findings in younger adults support the distinction of empathic response to positive vs. negative stimuli, known as positive vs. negative empathy, respectively (Morelli et al.
2015). Previous work suggests that people use the emotion expressed by others as a social signal to interpret what others are feeling (Van Kleef
2009). Thus, based on robust evidence of an age-related shift in processing positive over negative information (i.e., positivity effect), and in line with the motivational account of empathy, it is reasonable to assume that older adults’ empathic response is particularly impacted by valence of to-be-processed information.
The anterior insula, mid and dorsal anterior cingulate gyrus, and temporo-parietal junction have been identified as key brain regions involved in empathy (Bernhardt & Singer
2012; Bzdok et al.
2012; Decety & Jackson
2006; Lamm et al.
2011). However, to date, only two studies have investigated the neural substrates underlying empathy in older adults. In particular, Chen and colleagues (
2014) asked younger and older adults to rate their feelings towards another’s pain and found an age-related decrease in activation of insula and anterior cingulate during this task. Additionally, Riva et al. (
2018) reported reduced insula activity during both pleasant and unpleasant touch using a visuotactile stimulation paradigm among older female participants.
While meta-analyses support the importance of insula and anterior cingulate regions with a core “empathy network”, a distinction between affective-perceptual vs. cognitive-evaluative empathy networks has been proposed (Bellucci et al.
2020; Fan et al.
2011). In this context, the anterior insula has been suggested to be involved in affective-perceptual empathy and the anterior cingulate in cognitive-evaluative empathy. Furthermore, as empathy is a complex and multidimensional process, it is likely that empathic response activates large-scale brain networks and not just individual regions, but a multivariate approach to the study of empathy has not been undertaken yet.
White matter tracts, such as the cingulum bundle, which links the frontal lobe with the precuneus, posterior cingulate cortex, hippocampus, and parahippocampus (Wakana et al.
2004), are believed to be critical in attention, memory, executive functioning, and emotional processing (Keedwell et al.
2016; van den Heuvel et al.
2008; Wu et al.
2016). Also, more integrity in the anterior subdivision of the cingulum bundle was associated with better cognitive control in older adults (Metzler-Baddeley et al.
2012). However, structural pathways that subserve empathic responding have not been well investigated yet, and currently unknown is the extent to which microstructure of the cingulum bundle may facilitate higher-order social-cognitive processes, such as empathy, among older adults.
To address the above-identified research gaps and to integrate previously parallel lines of work, the present study had three major aims: to examine the extent to which functional activation involved in cognitive and affective empathy: (1) differed between younger and older adults; (2) in interaction with stimulus valence; and (3) was related to cingulum bundle microstructure. Here, younger and older adults completed a modified version of the Multifaceted Empathy Test (MET) (Dziobek et al.
2011) that comprised both cognitive and affective empathy components (as well as a neutral age perception control condition) and included positive, negative, and neutral images to allow for a systematic investigation of valence effects on empathic response (Mazza et al., 2015). Both structural and functional brain images were acquired to determine the association between functional network activation during the empathy task and white matter tract integrity known to play a role in emotional processing and aging.
We hypothesized that older relative to younger adults would show poorer performance, and that the age groups would display differential recruitment of brain networks (e.g., the limbic system; Yu & Chou
2018), during cognitive empathy (
Hypothesis 1a; Beadle & de la Vega
2019). In contrast, we predicted that performance during affective empathy would be comparable between younger and older adults (Beadle & de la Vega
2019) and that engagement of affective empathy-related brain regions (such as the insula and the anterior cingulate cortex; Singer & Lamm
2009) would be comparable between the two age groups (
Hypothesis 1b).
Furthermore, we expected that the age groups would differ in their recruitment of brain networks in response to negative and positive stimuli, with reduced activity of the salience network for negative (
Hypothesis 2a) but enhanced or equal activity of the default mode network for positive (
Hypothesis 2b) stimuli among older (relative to younger) participants. These predictions were based on evidence that positive and negative empathy selectively activate regions associated with positive (e.g., ventromedial prefrontal cortex) and negative (e.g., anterior insula, dorsal anterior cingulate cortex) affect, respectively (Morelli et al.
2015; Ziaei et al.
2019). We have previously demonstrated that positive affect was associated with activation in regions within the default mode network in older adults (Ziaei, Ebner et al.
2017a,
b). To our knowledge, however, no studies have directly compared neural correlates of positive vs. negative stimuli between younger and older adults during the empathy task. Our predictions built on research in emotion recognition that suggests that regions such as the ventromedial prefrontal cortex and posterior cingulate cortex are involved in processing positive emotions, while regions such as the insula and anterior cingulate are involved in processing negative emotions. Here we also explored age-related differences in the cingulum bundle in its association with functional activation during both cognitive and affective empathy. We anticipated that higher fractional anisotropy of the anterior cingulum bundle would be related to functional activation during cognitive empathy and posterior cingulum would be related to functional activation during affective empathy among older adults.
Discussion
Age-related differences in empathy have been researched across multiple studies, and overall a pattern has emerged suggesting decrease in cognitive empathy with age, while effects on affective empathy are more mixed (see also Bailey et al.
2021; Ebner et al.
2017, for overviews). Going beyond previous research by examining not only behavior but also brain structure and function in cognitive and affective empathy among younger and older adults, this study generated several novel insights.
Behaviorally, we found that older adults’ empathic responding was slower than their responses to the age perception control condition; and their response times were affected by the valence of the stimulus during cognitive, but not affective, empathy. At the neural level, we did not find any evidence for age-related reduced activity during cognitive empathy, thus not supporting Hypothesis 1a. Supporting Hypothesis 1b, however, we found that for affective empathy both age groups recruited a similar brain network. We also found that older, but not younger, adults engaged regions of the salience network in response to negative emotions during cognitive empathy, supporting Hypothesis 2a. In contrast, with regards to affective empathy, both age groups, and not only older adults, engaged a similar pattern of brain regions that contained nodes of the default mode network in response to positive emotions, partially supporting Hypothesis 2b. Our structure–function analyses, finally, revealed that microstructure of the posterior, but not the anterior, cingulum bundle was related to engagement of major nodes of the default mode network during affective empathy to positive stimuli in both age groups. These central findings from this work will be discussed in more detail next.
The results of our whole-brain analysis (the first LV) indicated that regions such as bilateral insula, right superior frontal gyrus, left superior temporal gyrus, right anterior cingulate, left medial frontal gyrus, bilateral precuneus, and bilateral posterior cingulate were engaged across all experimental conditions among older adults, perhaps reflecting general cognitive effort needed for all conditions. Some of these regions have been implicated in empathic responding (e.g., insula and anterior cingulate; Bellucci et al.
2020) and understanding others’ mental states (e.g., superior temporal gyrus; Gallagher & Frith
2003; et al., 2020). In addition, however, they have also been involved in exerting cognitive effort during a task (e.g., superior frontal gyrus; Aron et al.
2014) as well as in self-referential processing (possibly in relation to others) more generally (e.g., precuneus and posterior cingulate; Buckner et al.
2008), in support of the notion that these areas reflect general domain processing rather than experimental condition specific processing.
Neither our brain nor our behavioral data supported age-related differences in affective empathy. Thus, in contrast to Chen et al. (
2014), we did not observe reduced engagement of core empathy regions including the anterior insula, mid-cingulate, and medial prefrontal cortex in older adults. Rather, both younger and older adults in our study recruited these regions similarly during the affective empathy task. Differences in the type of stimuli used in our study (i.e., non-pain stimuli) compared to the pain stimuli used in Chen et al. (
2014) may underlie these divergent findings. Speaking against this methodological explanation, a recent meta-analysis found similarities in empathic responses to pain and non-pain stimuli (Timmers et al.
2018); this comparability across stimulus types, however, has not been confirmed in research with older individuals yet. Additionally, it has to be noted that Chen et al. (
2014) investigated empathic responses in three age groups, younger, middle age, and older adults; and while the neural correlates of each age group were compared in their study, the use of different stimuli and a sample from quite a different population than the present study’s sample has to be noted when comparing Chen et al. (
2014)’s results with the results from the current study. Further, they conducted univariate, contrast-based imaging analyses, while the present study adopted a multivariate analysis method.
Our functional MRI findings for affective empathy do not support the notion of diminished internal bodily perception (interoceptive awareness) among older adults (Mendes
2010), but functional sparing for affective processing in aging, as observed here, is in line with the brain maintenance hypothesis discussed in the cognitive aging literature (Nyberg et al.
2012). One could also argue that the idea of functional reserve that protects older adults from decline does especially well apply as an explanation to our findings given that participants in this study were cognitively high functioning (see background measures in Table
1). Thus, they may have had high cognitive reserve, which could have resulted in performance on a level comparable with the younger participants—and especially so for the affective empathy task which may have been easier to perform than the cognitive empathy task. Future research will be able to test this interpretation.
We observed age-related differences in valence modulation during cognitive empathy, both on the behavioral and the neural level. In particular, we found that older adults recruited a neural network that comprised bilateral insula and anterior cingulate, core nodes of the salience/midcingulo-insular network network (Menon
2015; Menon & Uddin
2010), more in their cognitive empathic response to negative stimuli than younger adults did. This finding aligns with behavioral evidence that older compared to younger adults experience greater difficulty processing negative than positive emotions (Hayes et al.
2020; Ruffman et al.
2008). In line with evidence from these meta-analyses, processing of negative (relative to positive) emotions may be more effortful and considered more threatening for older adults, as it does not align with their primary goals and implicit motivations. As a result, enhanced prefrontal cortex and insula activation for processing negative relative to positive information among older adults (Ebner et al.
2012; Ziaei et al.
2016) may be reflective of greater cognitive control and/or emotional down-regulatory processes at work.
Another possibility for engagement of the salience network during cognitive empathy to negative emotions in older adults is that this network, and insula specifically, is involved in orienting attention towards relevant stimuli in the environment (Menon & Uddin
2010). Given the salience of negative emotions and their importance for survival, orienting attention towards negative emotions is crucial and, therefore, associated with insular response. Moreover, insular activity in this context may be reflective of a brain response that is commonly observed across various cognitive tasks to guide behavior in dynamic social contexts where recognition of negative emotions is crucial (Bernhardt & Singer
2012). Our result, however, contradicts suggestions of age-related reductions in insular activation subserving interoception and the simulation of emotions in others (Mather
2016). Certainly, further investigation is needed to determine the relationship between bodily response (such as heart rate variability and skin conductance) and insular activity during empathy, and social cognition more broadly, across adult development and in aging.
Our finding of no age-related difference in affective responding to positive stimuli is in line with socioemotional selectivity theory (SST) (Carstensen et al.
2003,
1999), which proposes that older adults preferentially process positive over negative stimuli. In other words, older adults’ bias towards positive emotions may have facilitated processing of positive emotions in the affective empathy task, resulting in comparable brain and behavioral activity patterns between the age groups under this condition. Thus, the present study’s results could be interpreted as suggesting that the positivity effect reported in the aging literature for various cognitive and social-cognitive processes also extends to affective empathy. In fact, this finding also aligns well with the motivational empathy account (Weisz & Zaki
2018). That is, older adults might be more motivated to process positive than negative emotions, as these emotions correspond with their social goals.
Another explanation for similar responses among the age groups towards positive affective empathy might be a general tendency in humans, at any age, to show empathic responses towards positive emotions. Relatedly, considering evidence that it is easier to show support for a partner’s positive than negative life events (Andreychik
2019; Gable et al.
2006), it is possible that the age groups do not differ in the neural network involved under this condition. Recent neuroimaging evidence furthermore supports the idea that positive empathy makes prosocial acts feel more rewarding, activating regions such as the ventromedial prefrontal cortex (Harbaugh et al.
2007; Hare et al.
2010). Thus, empathizing with positive emotions may be more rewarding and/or easier, generally, and comparably so for younger and older adults.
Taken together, the findings reported here well align with previous studies supporting salience network/ midcingulo-insular network (including the insula and the cingulate cortex) activation during cognitive empathy (Bernhardt & Singer
2012; Pasquini et al.
2020; Rankin et al.
2006). Positive empathy can engage areas related to the processing of positive emotions such as the default mode network/ medial frontoparietal network as well as the reward system (e.g., the ventral striatum). Our observation that regions associated with the default mode network are engaged in response to positive emotions during affective empathy is also in accordance with previous studies (Ziaei et al.
2019; for a review see Morelli et al.
2015). Based on our results, we propose that there might be a distinction in the brain networks recruited by older adults in response to positive vs. negative emotions, as a function of task context and task effort required; a hypothesis that needs to be followed up in future studies.
We furthermore provide first evidence here of posterior cingulate bundle involvement during affective empathy for positive emotions in both younger and older adults. In particular, we saw a high concordance between regions that were connected by the posterior cingulum and regions that were activated for positive empathy during affective empathy. A growing body of work now supports that the default mode network might play a role in the processing of positive emotions, possibly due to lower cognitive resources involved in processing of positive (relative to negative) stimuli, and greater salient features of positive cues (e.g., showing teeth; Ziaei & Fischer
2016). Additionally, research has previously shown that the regions connected with the cingulum bundle have a high overlap with regions within the default mode network (van den Heuvel et al.
2008). However, what had not been demonstrated before, and our results speak to this gap, is the role of these areas for affective empathy, and especially for affective empathy in response to positive emotions in aging. Older adults with higher FA values in the posterior cingulum bundle tract exhibited higher activity in this network for positive emotions. This result again aligns with the motivational account of empathy and provides first evidence of a structure–function link in affective empathic responding to positive emotions in aging.
The activation of the posterior cingulate during positive affective empathy may reflect self-referential (affective) processing in linking one’s own and another’s emotional state to enable an adequate empathic response. In support of this interpretation, the posterior cingulate cortex has been shown to play a role in a wide range of social-cognitive processes (Brewer et al.
2013; Sperduti et al.
2012). For example, posterior cingulate activation is involved in theory of mind and mentalizing (Frith & Frith
2006; Mitchell
2009; Molenberghs et al.
2016). A meta-analysis furthermore showed that the posterior cingulate cortex subserves empathy (Bzdok et al.
2012), specifically the evaluation of how “one relates to one’s experience” (Brewer et al. (
2013). Our findings importantly add to previous literature by demonstrating a role for the posterior cingulate in affective empathy in aging. Our findings are in line with the last-in-first-out hypothesis (Madden et al., 2019) that proposes that prefrontal cortex areas, relative to posterior parts of the brain, are the first affected by the aging process. Additionally, our results support the posterior-to-anterior shift in aging (Davis et al.
2008), suggesting that in older compared to younger adults, anterior brain regions (e.g., prefrontal cortex) are more activated than posterior brain regions (e.g., sensory and visual cortex) across a variety of tasks. The higher FA values in the posterior cingulum associated with affective empathy observed in the present study suggest that structural integrity of this region plays a role in affective empathic response, specifically for positive emotions.
We acknowledge that we considered reaction time in this study as the indicator for performance in cognitive and affective empathy. In real life, adequate empathic response may not be evaluated with how fast we respond, but how appropriate our response is given the nature of the stimuli we respond to, as well as how well we can express our empathic concern towards others. Future studies are needed to integrate behavioral responses such as response time with real-life physiological and behavioral measures to study empathy in aging. Additionally, future research is needed to investigate more ecologically valid ways to assess empathic response to emotional stimuli and to investigate the relationship between psychiatric symptoms, such as depression and anxiety, and empathic response among older adults.