1 Investigating Adult Age Differences in

2 real-life Empathy, Prosociality, and Well-

3 being using Experience Sampling

4 Lena Pollerhoff1*, Julia Stietz2, Gregory J. Depow3, Michael Inzlicht3,4,

5 Philipp Kanske2,5, Shu-Chen Li1,6, Andrea M.F. Reiter1,7 6 7 1 Lifespan Developmental Neuroscience, Faculty of Psychology, Technische Universität 8 Dresden, Dresden, Germany 9 2 Clinical Psychology and Behavioral Neuroscience, Faculty of Psychology, Technische 10 Universität Dresden, Dresden, Germany 11 3 Department of Psychology, University of Toronto, Ontario, Canada 12 4 Rotman School of Management, University of Toronto, Ontario, Canada 13 5 Max Planck Institute for Human Cognitive and Brain Sciences, Leipzig, Germany 14 6 Centre for Tactile Internet with Human-in-the-Loop, Technische Universität Dresden, 15 Dresden, Germany 16 7 Department of Child and Adolescent Psychiatry, Psychosomatics and Psychotherapy, 17 University Hospital Würzburg, Würzburg, Germany 18 * Corresponding author: Lena Pollerhoff ([email protected]) 19 20 Keywords: Empathy, Prosocial Behavior, Well-being, Adult Lifespan, Adult Development 21 Abstract 22 While the importance of social affect and cognition is indisputable throughout the adult lifespan, 23 findings of how empathy and prosociality develop and interact across adulthood are mixed, 24 and real-life data are scarce. Research using ecological momentary assessment recently 25 demonstrated that adults commonly experience empathy in daily life. Furthermore, predictors 26 of empathy were linked to higher prosocial behavior and subjective well-being. However, to 27 date, it is not clear whether there are adult age differences in daily empathy and daily 28 prosociality and whether age moderates the relationship between empathy and prosociality 29 across adulthood. Here we analyzed experience-sampling data collected from participants 30 across the adult lifespan to study age effects on empathy, prosocial behavior, and well-being 31 under real-life circumstances. Linear and quadratic age effects were found for the experience 32 of empathy, with increased empathy across the three younger age groups (18 to 45 years) and 33 a slight decrease in the oldest group (55 years and older). Neither prosocial behavior nor well- 34 being showed significant age-related differences. We discuss these findings with respect to 35 (partially discrepant) results derived from lab-based or traditional survey studies. We conclude 36 that studies linking in-lab experiments with real-life experience-sampling might be a promising 37 venue for future lifespan studies.

38

39 Introduction 40 Throughout the lifespan, satisfying social interactions are key for well-being (e.g., [1]) as well 41 as for mental and physical health (e.g., [2]). Experiencing empathy in response to someone’s 42 suffering, a feeling that is believed to trigger prosocial behavior, is an important ingredient for 43 establishing and maintaining relationships with other people [3,4]. Whilst it is certainly 44 indisputable that social functioning remains important throughout the adult lifespan [5,6], findings 45 on how empathy and prosociality develop and interact over the course of adulthood are mixed 46 to date. 47 Research on the lifespan development of empathy has often built on the distinction of 48 affective components (affect sharing, empathic concern, compassion) from cognitive 49 components (perspective-taking) of empathy [7,8]. In two recent laboratory studies, using a 50 naturalistic paradigm to dissociate both affective and cognitive aspects of understanding 51 others [9,10], we did not observe age-related differences in affect sharing and enhanced 52 empathic concern. However, perspective taking ability decreased in older compared to 53 younger adults. Whilst this pattern of findings contributes to an emerging prevalent view of 54 reduced cognitive but preserved or increased affective empathy in older vs. younger adults ([8] 55 for a recent review), there is substantial heterogeneity in the literature, particularly when 56 considering adult age effects across the lifespan: 57 While some cross-sectional lab studies reported age-related increase [11,12] in self- 58 reported empathic concern, others found an inverted U-shaped effect of adult age [12]. Further, 59 age-related deficits in self-reported affect sharing [13,14] were found, whereas no age-related 60 differences were observed when using a behavioral task [15]. The same inconsistencies were 61 found in terms of perspective taking. While, overall, meta-analytic evidence points towards a 62 decrease in perspective-taking ability as found in cross-sectional studies using behavioral 63 tasks [16] a longitudinal study showed no evidence for long-term change across adulthood with 64 respect to self-reported trait perspective-taking[4]. There are also studies reporting no age 65 difference [11], whilst others describe an age-related increase [12], or an inverted U-shaped 66 pattern in trait perspective taking [12]. 67 Similar to this heterogeneity in findings in the domain of empathy, previous research 68 on adult age differences in prosocial behavior has yielded mixed results. Experimental age- 69 comparative laboratory studies have often revealed a higher degree of prosocial behavior in 70 older compared to younger adults [17–21], and have shown a linear increase in prosociality, when 71 examining prosociality across the whole adult lifespan [11,22,23]. There are however also studies 72 which did not find such age-related increase in prosocial behaviors, neither in experimental 73 tasks or self-reported prosocial measures when comparing younger and older adults [24,25], nor 74 when comparing middle- and older age groups [26], or when looking at age correlations across 75 adulthood [27,28]. 76 Social psychology and neuroscience research suggests that feeling empathy with a 77 person usually results in greater prosocial behavior (for reviews see [29–32]) and some previous 78 studies have asked whether this link might be moderated by adult age. Experimental studies 79 in the laboratory demonstrated enhanced prosocial behavior after an empathy induction in 80 older adults (compared to younger adults) [19] as well as age-related linear increases in 81 prosocial behavior across younger, middle aged, and older adults that were partially mediated 82 by empathic concern [11]. Age was also found to positively moderate the association of self- 83 reported prosocial behavior and empathic concern, but only in participants younger than 75 84 years [26]. However another study did not find age-related differences regarding the link 85 between empathy and prosocial behavior when comparing younger and older adults [18]. 86 The association of well-being with age is a well-established research topic. An 87 influential lifespan theory of affective development (‘socio-emotional selectivity theory’ [33]) 88 postulates higher socio-emotional well-being in older adults. Large-scale international surveys 89 across several countries often revealed a U-shaped association between age and well-being, 90 with higher well-being in younger and older adult age (e.g., [34–36]). Studies using ecological 91 momentary assessment (EMA) replicated the U-shaped adult age profile in global well-being 92 [35] and demonstrated a decline in the experience of negative emotions until the age of around 93 60 [37]. Notwithstanding, there is also a current debate about this putative U-shaped pattern, 94 including critique with respect to its robustness and generalization [38–40]. Prosocial behavior 95 and well-being have been suggested to be associated with each other, and some studies have 96 suggested that their link is moderated by age: In a recently published large-scale daily diary 97 study [41], younger adults’ prosocial behavior was linked to both greater negative affect and 98 positive experiences, while these associations were attenuated in older adults. This was 99 interpreted as a decreasing influence of prosociality on well-being over the course of the life 100 span. In contrast, earlier age-comparative studies have pointed towards more beneficial 101 effects of volunteering on life satisfaction in older compared to younger adults [42]. 102 Considering the lifespan developmental findings (and their considerable heterogeneity) 103 reviewed above, an intriguing open research question is, whether there are adult age 104 differences in empathy, pro-sociality, and wellbeing as they occur in daily life. Ecological 105 Momentary Assessment (EMA) is a method which is well-suited to capture these processes in 106 real life. EMA is a relatively new method, with advantages like increased ecological validity, 107 and decreased recall bias, while measuring variability and change over a short time period [43]. 108 These advantages may be particularly important for lifespan developmental studies, as recall 109 bias[44] and intra-individual variability [45] have been shown to be subject to age effects in 110 different behavioral contexts. Unfortunately, and to the cost of external validity, to date only a 111 small list of studies [4,46,47] examined empathy and prosociality using EMA, with most of them 112 not focusing on adult age differences. With the current study we aimed to fill this gap, by 113 leveraging the advantages of EMA, and investigating daily empathy, prosociality, and well- 114 being under real-life circumstances, repeatedly per day within person. To this end we analyzed 115 smartphone based experience-sampling data recently published by Depow and colleagues [47], 116 to examine age effects with regards to daily empathy, prosocial behavior, and well-being, 117 which were not part of their initial study. Separating contributions of within- from between- 118 subject variability, we aimed to investigate whether age influences daily empathy, prosocial 119 behavior and well-being, as well as their interactions. Of note, in the current study, empathy 120 was defined as an umbrella term, subsuming the three different subcomponents emotion 121 sharing, compassion, and perspective taking. Prosocial behavior was defined as as an 122 opportunity to help someone. 123 Based on the heterogeneity of age-related findings in the current literature regarding 124 the constructs of empathy and prosocial behavior as reviewed above, and considering 125 inconsistencies in the definition of empathy, we had undirected hypotheses in terms of age 126 differences in daily empathy and prosocial behavior. In line with previous results [34–36], we 127 expected a U-shaped relationship of adult age with well-being (measured in terms of happiness 128 and sense of purpose), with higher well-being in younger and older adults. Based on a positivity 129 bias postulated for older adults in lifespan theories of emotional aging [48], we predicted that 130 older adults might experience more empathy in contexts with positive valence. Based on a 131 previous lab-based study [26], we hypothesized an attenuated relationship of empathy and 132 prosocial behavior in higher age. Furthermore, based on the literature[41] we assumed a greater 133 relationship of empathy and well-being with increasing age and the same age-related pattern 134 with respect to the link between prosocial behavior and well-being. 135

136 137 Figure 1. Daily survey design, visualizing the different survey levels and related questions. 138 Only the questions, relevant for the current study, are depicted here. For further details, and a 139 full study protocol, see Depow and colleagues [47]. Note that empathy (level 2) was assessed 140 as an umbrella term, thus, questions about the subcomponents emotion share, perspective 141 take, and compassion (level 3) were only rated if the participant indicated to have experienced 142 empathy on level 2 of the survey. 143 144 Results

145 Adult Age Differences in Everyday Empathy 146 Empathy Opportunities. Regarding the first level of the survey (see Fig 1), i.e., the frequency 147 of empathy opportunities, and the frequency of being the target of empathy, no age-related 148 differences were found (empathy opportunities: b = 0.01, SE = 0.08, z = 0.17, p = 0.867, adj. 149 p = 0.867, r = 0.00; target of empathy: b = 0.02, SE = 0.10, z = 0.18, p = 0.856, adj. p = 0.867, 150 r = 0.01). Reassuringly, this null effect of age suggests that potential age differences in all 151 subsequent analyses are unlikely to be affected by baseline differences in how often different 152 age groups experienced an opportunity to empathize in the first place. 153 154 Experiencing Empathy. With respect to the actual experience of empathy (2nd level of the 155 survey), significant linear as well as quadratic effects of age on feelings of empathy were 156 observed (linear: b = 0.38, SE = 0.15, z = 2.47, p = 0.013, r = 0.10; quadratic: b = -0.41, SE = 157 0.18, z = -2.26, p = 0.024, r = -0.11). Daily feelings of empathy increased across the first three 158 age groups, from 18 to 44 years, but, as the significant quadratic trend suggests, show a 159 tendency to decrease beyond these ages, in the age group of 55 years and older (see Figure 160 2A). 161 In lifespan developmental theories, a positivity bias for older as compared to younger 162 adults has been suggested [48]. Here, we tested this hypothesis with respect to the valence of 163 the empathy opportunity experienced before participants indicated their actual feelings of 164 empathy. That is, we tested for an interaction of age group with valence of the empathy 165 opportunity. As reported in Depow and colleagues [47], subjects generally showed higher 166 empathy after positive empathy opportunities; however, contrary to our expectation this effect 167 was not moderated by age (Chi2 (2) = 0.25, p = 0.881, see Figure 2B). 168 169 170 171 172 173 174

Figure 2. (A) Adult age differences regarding the actual experience of empathy. Significant quadratic association between actual experience of empathy and age. (B): Interaction between valence of the empathy opportunity and age on daily empathy feelings. Age did not moderate the link between the valence of the situation and actual feelings of empathy. 175 Subcategories of empathy. Regarding the subcomponents of empathy (3rd level of the 176 survey, Fig 1), no significant age-related differences (neither linear, nor quadratic) were found 177 for emotion sharing (b = -0.18, SE = 0.13, z = -1.36, p = 0.173, adj. p = 0.519, r = -0.05), 178 perspective taking (b = 0.02, SE = 0.14, z = 0.14, p = 0.887, adj. p = 0.887, r = 0.01), and 179 compassion (b = 0.19, SE = 0.36, z = 0.54, p = 0.592, adj. p = 0.887, r = 0.05). Note that ratings 180 of the subcomponents of empathy were only provided by those participants who indicated 181 empathy on the second level of the survey. Thus, we observe age-related differences in the 182 tendency to experience empathy in general (2nd level). Among the proportion of participants 183 who experience ‘general’ empathy however, no age differences with regard to the 184 subcomponents were observed. Further, after adjusting p-values, no age-related differences 185 were observed in extent, difficulty, and confidence (4th level) with respect to the three different 186 subcomponents emotion sharing, perspective taking, and compassion (all bs < 0.04, all SEs < 187 0.08, all ts < 0.98, all adj. ps > 0.090, all rs < 0.20; see supplementary table S1). 188 Depow and colleagues [47] showed that the co-occurrence of the three subcomponents 189 of empathy (i.e., emotion sharing, perspective taking, and compassion) was very high. Thus, 190 the different components of empathy seem to mainly co-occur in everyday life. Analyzing age 191 effects on the co-occurrence of the empathy subcomponents using a chi2-tests, we did not find 192 any significant effect of age (all chi²s < 5.76, all ps > 0.; see supplementary figures S1-4). 193 194

195 Adult Age Differences in Prosociality 196 Contrary to our expectations, we did not find significant age-related differences with respect to 197 daily prosocial behavior (b = 0.009, SE = 0.08, z = 0.10, p = 0.916, r = 0.00, see figure 3A). 198 Depow and colleagues [47] found significant associations between different empathy predictors 199 (e.g., empathy opportunity, actual empathy, subcomponents) and prosocial behavior. In the 200 current analysis most of those effects were not significantly moderated by participants’ age (all 201 adj. ps > .229, see Table 1). Only the interaction term of reported empathy opportunities and 202 age as quadratic and linear term, respectively, significantly predicted prosocial behavior as a 203 within-subject effect (linear: b = -0.13, SE = 0.05, z = -2.90, p = 0.004, adj. p = 0.024, r = -0.04, 204 quadratic: b = 0.15, SE = 0.06, z = 2.73, p = 0.006, adj. p = 0.024, r = 0.04). Across all age 205 groups the experience of an empathy opportunity was associated with higher prosocial 206 behavior, but this effect was more pronounced in the middle-aged groups than in the youngest 207 and oldest groups (see Figure 3B). 208

Figure 3. (A) Adult age differences in prosocial behavior. No significant association between age and prosocial behavior was found. (B) Within-subject effect of empathy opportunity x age on prosocial behavior. All age groups showed more prosocial behavior after an empathy opportunity. A quadratic effect of age group indicated that this effect was more pronounced in the middle-aged groups than in the younger and older age group. 209 Table 1: Within- and between person effects of different interactions regarding daily empathy 210 measures and age predicting daily prosocial behavior. Prosocial behavior was included in all 211 models as binary outcome variable. Statistics obtained from mixed models, nested within 212 participant and survey day. Each interaction ran in a separate model, with age as linear and 213 quadratic term separately. Model selection was conducted based on a loglikelihood ratio test. 214 P-values were adjusted to control the false discovery rate. * p < 0.05. 215 αreduced random effect structure due to convergence warnings, only nested within participant. Within-subject effects Between-subject effects Interaction term t or z- Adj. Estimate Effect t or z- Adj. Estimate Effect predicting score p- (SE) size score p- (SE) size prosocial value (r) value (r) behavior Empathy -2.90 0.024* 0.05 -0.04 -1.81 0.490 0.20 -0.10 opportunity * age group Empathy 2.73 0.024* 0.06 0.04 opportunity *age group2 Target of empathy 1.00 0.423 0.04 0.01 -0.93 0.744 0.32 -0.08 * age group Empathy * age 1.17 0.390 0.09 0.03 0.06 0.954 0.31 0.00 group Emotion share * -1.43 0.306 0.11 -0.04 0.80 0.744 0.33 0.07 age group Perspective take * -1.72 0.229 0.11 -0.05 0.46 0.899 0.32 0.04 age group Compassion * age 0.39 0.695 0.15 0.02 0.24 0.947 0.42 0.03 group Interaction term Chi2 Adj. Chi2 Adj. predicting p- p- prosocial value value behavior Valence * age 0.43 0.584 0.83 0.744 groupα 216

217 218 Adult Age Differences in Subjective Well-Being 219 Overall, and contrary to our a-priori hypothesis, we did not find significant adult age effects on 220 daily subjective well-being (b = 0.11, SE = 0.08, t(233) = 1.48, p = 0.141, r = 0.10, see figure 221 4A). In Depow and colleagues’ [47] analyses, daily empathy predictors were associated with 222 subjective well-being, e.g., participants showed increased well-being in case of an empathy 223 opportunity, when empathy was indeed reported. Thus, in a next step, we analyzed whether 224 such an association of empathy outcomes and well-being, as reported on the group-level, was 225 moderated by age. No significant interaction with age was revealed (all adj. ps > 0.144, see 226 Table 2). We found a significant interaction effect between age and prosocial behavior on 227 subjective well-being, but only as a within-subject effect (see Table 2). As can be seen from 228 figure 4B, within-person, across all age groups, higher well-being was associated with acting 229 prosocially before. However, the difference in well-being as a function of an opportunity to act 230 prosocially was slightly more pronounced in the younger age groups (see figure 4B). 231

Figure 4. (A) Adult age differences in subjective well-being. No significant association between age and well-being. (B) Within-subject effect of prosocial behavior x age on well-being. positive association between well-being and acting prosocially before, irrespective of participants age. Stronger influence of the preceding prosocial act in younger adults.

232 233 Table 2. Within- and between person effects of different interactions regarding daily empathy 234 measures and age predicting daily well-being. Well-being was included in all models as 235 continuous outcome variable. Statistics obtained from mixed models, nested within participant 236 and survey day. Each interaction ran in a separate model, with age as linear and quadratic 237 term separately. Model selection was conducted based on a loglikelihood ratio test. P-values 238 were adjusted to control the false discovery rate. ** p < 0.01 239 α Between-subjects effect model include religiosity as covariate. Within-subject effects Between-subject effects Interaction term t or z- Adj. p- Estima Effect t or z- Adj. p- Estimat Effect predicting well- score value te (SE) size score value e (SE) size being (r) (r) Empathy opportunity * 0.64 0.532 0.01 0.01 -2.06 0.144 0.35 0.14 age group α Target of empathy * 0.63 0.532 0.01 0.01 -2.27 0.144 0.40 0.15 age group α Empathy * age group 0.81 0.532 0.03 0.02 1.31 0.338 0.30 0.09 α Emotion share * age -1.07 0.532 0.04 0.03 -0.14 0.938 0.30 0.01 group Perspective take * -1.22 0.532 0.03 0.04 1.05 0.412 0.31 0.07 age group Compassion * age -0.96 0.532 0.05 0.03 -0.08 0.938 0.39 0.01 group Interaction term F- Adj. p- F- Adj. p- predicting well- value value value value being Valence * age group 1.64 .532 2.69 .236 Interaction term t or z- p- Estima Effect t or z- p- Estimat Effect predicting well- score value te (SE) size score value e (SE) size being (r) (r) Prosocial Act * age 2.91 0.004* 0.01 0.04 -1.34 0.181 0.34 0.09 group *

240

241 242 Discussion 243 In this study, we used experience sampling data [47], to test adult age differences with respect 244 to empathy, prosociality, and well-being. We were able to yield several new insights into the 245 lifespan development of these important social functions, as well as their interactions in daily 246 life. 247 Interestingly, no adult age differences with respect to the frequency of the opportunities 248 to empathize with another person, and of being the target of empathy could be observed. This 249 might be surprising at first glance, in light of findings that social network size and social 250 contacts diminish in older age [49,50] and based on prior evidence demonstrating negative 251 attitudes towards older adults (‘ageism’) [51,52]. The latter has been argued to lead to reduced 252 empathy towards older adults [53,54]. However, in line with what we observed here, previous 253 cross-sectional and longitudinal studies have shown that older age was not associated with a 254 decline in social connectedness per se [55,56]. 255 With respect to daily empathy, we found a (small) linear effect of age on empathy 256 across the lifespan, which is in line with postulated stable or even increased socio-affective 257 processes in older adults, as measured in laboratory studies (e.g., [9–11]). Interestingly, 258 additionally a significant effect of age and empathy in an inverted U-shaped quadratic pattern 259 was found, suggesting that daily empathy increased from 18 years on, with a peak in midlife, 260 but decreased slightly in the oldest group (55 years and older). Notably, the empathy level of 261 the oldest group was higher than the one of the youngest. This pattern is in line with the findings 262 from O’Brien and colleagues [12], with respect to their quadratic associations in self-reported 263 trait empathic concern and perspective taking. Further, there is cross-sectional and 264 longitudinal evidence that moving from early to middle adulthood is associated with more 265 stable dispositional traits like agreeableness[57], and that the period of middle adulthood is also 266 linked to highest values of generativity[58], potentially rendering empathy a particularly relevant 267 skill in this period of life. 268 When comparing the results of this experience-sampling study to previous research on 269 the lifespan development of empathy, it is important to note that the definition of the construct 270 empathy varies across studies, which has been criticized recently [59] reminiscent to a 271 discussion about so-called “Jingle-Jangle fallacies” [60,61]. Such an inconsistency with respect 272 to the definition of empathy also affects studies investigating age differences in empathy. In 273 the current study, empathy was regarded as an umbrella term spanning the subcomponents 274 emotion sharing, compassion, and perspective taking; consequently, participants only 275 provided ratings of these subcomponents when they had indicated feelings of empathy before. 276 Based on this operationalization, a high co-occurrence between emotion sharing, perspective 277 taking, and compassion was reported, and no age-related differences regarding the three 278 subcomponents nor their co-occurrence were found. Previous ageing studies differentiated 279 (emotional) empathy in terms of affect sharing, from compassion, and perspective taking (e.g., 280 [9,10]), by assessing them independently from each other, which revealed differential age- 281 related findings for each construct ([7,8] for review). 282 The majority of studies on empathy thus far focused, by design, on empathy exclusively 283 in the context of negative stimuli and most often with strangers and in the lab (e.g., EmpaToM- 284 Paradigm [62], as used in [9]). The dataset used for the current analyses offered the opportunity 285 to examine age effects in interaction with the valence of the situation. Indeed, Depow and 286 colleagues [47] had demonstrated that empathy had been reported more frequently following 287 situations with positive valence. In contrast to a hypothesized positivity bias [33], this effect was 288 not enhanced with increasing age. However, the experience of affective empathy, particularly 289 compassion [63], be it in an emotionally positive or negative context, might be rewarding in itself, 290 or subsequently lead to satisfying social interactions (if, e.g., followed by prosocial behavior as 291 demonstrated in the current dataset). In this regard, experiencing enhanced empathy in both 292 positive and negative domains is in line with the lifespan goals suggested by emotional 293 selectivity theory[33], namely an enhanced pursuit of emotionally satisfying interactions in 294 midlife and older age. 295 In this experience-sampling study, no adult age differences in prosocial behavior were 296 observed. This is in contrast to most laboratory studies using economic decision-making tasks 297 (e.g., dictator game, public goods game), or experimental helping/donating paradigms, which 298 showed pronounced prosocial behavior in older adults (e.g., [17,19,64]). Also studies using self- 299 report measures in terms of validated questionnaires, showed greater prosociality with 300 advancing age, by comparing younger vs. older adults [65], across younger and middle-old 301 adults [66], or across the whole adult lifespan [67,68]. While these inconsistencies in results may 302 be argued to be due to these studies not measuring behavior in real life, discrepancies also 303 occurred when comparing the current results with studies that measured real-life prosociality: 304 Cavallini et al [26] reported a significant negative association between age and self-reported 305 real life prosocial activities, arguing that real-life prosocial behavior is more cognitively and 306 physically demanding. An important difference between the study by Cavallini and colleagues 307 [26] and the current study is the higher age on average, as they only included participants aged 308 55 years and older, but no younger age group. Thus, our comparably young sample here might 309 not be ideally suited to detect potential effects of physical and cognitive demands on an aging 310 effect in prosocial behavior. 311 We expected an attenuated relationship of empathy and prosocial behavior with higher 312 age[26] which was only partially confirmed. Age did not modulate the associations between 313 empathy per se, nor of any of its subcomponents, with prosocial behavior. This is different 314 compared to the results from Cavallini and colleagues [26], which showed that with increasing 315 age, daily prosocial behavior was less driven by empathic concern. In this context, it has again 316 to be noted that the participants in Cavallini and colleagues [26] were considerably older. In the 317 current study, the interaction between one important empathy predictor, namely the 318 opportunity to empathize, and age, both as a linear and quadratic term, on prosocial behavior 319 was statistically significant as a within-person effect. That is, whilst all age groups showed 320 more prosocial behavior when there was an empathy opportunity, this stronger tendency to 321 act prosocially in case there was a situation eliciting empathy (as compared to when there was 322 none) was found to be slightly pronounced in the middle-aged groups of the sample. 323 In the current study, we did not observe the often-found U-shaped age-related pattern 324 of self-reported well-being previously found in large studies (e.g., [34,35]). Subjective well-being 325 had been assessed by two different questions, merged into one universal well-being score, 326 covering hedonic well-being (i.e., experienced happiness) and eudaimonic well-being (i.e., 327 purpose of life). Even though eudaimonic well-being gets more attention in the current literature 328 [2,69], most studies that have found a U-shaped pattern of subjective well-being emphasized a 329 hedonic operationalization of well-being (e.g., [36,38]). However, beyond the measurement level, 330 it is noteworthy that there is a recent debate about the robustness and generalization of the 331 putative U-shaped pattern with regard to the association of adult age and well-being [38–40]. A 332 recently published study [38] challenged this often as typical assumed U-shaped pattern, by 333 revealing inconsistent findings with respect to the association of age and well-being, both in 334 cross-sectional and longitudinal studies. Additionally, it is argued that age effects on well-being 335 might be an epiphenomenon of other variables, reflecting the current life circumstances 336 associated with ageing (e.g., income, education), fixed effects like personality traits, or 337 selection effects which might differ as a function of age group [36,40,70]. 338 We confirmed a hypothesized moderation effect of age on the link between daily 339 prosocial behavior and subjective well-being. Across all age groups, subjective well-being 340 within a person was higher when a prosocial act was performed 15 minutes before. However, 341 this trend was attenuated with increasing age. This is in line with Chi and colleagues’ [41] 342 recently published observations about a decreased influence of prosocial behavior on well- 343 being in older compared to younger adults. Interestingly, an earlier study [42] observed the 344 opposite when looking at the long-term impact (i.e., over 3 years of time) of volunteering in a 345 longitudinal survey, namely enhanced well-being as a result of volunteering in older compared 346 to younger adults. Given shrinking time horizons for older versus younger adults, which have 347 been associated with different (socio-emotional) life goals [33], it would be an interesting venue 348 for future studies to differentiate short- and long-term consequences of prosocial behaviors 349 and age differences therein. 350 In sum, it is interesting to note that across many of the measured variables, 351 inconsistencies (and indeed, often times null effects of age) as compared to previously 352 reported age-related results were observed. More specifically, compared to published reports 353 on effects of age differences on prosocial behaviors, well-being, and empathy, we observed 354 null effects of age on prosociality and well-being, and a small quadratic effect on empathy. In 355 the following, we speculate about where these discrepancies might arise from. 356 First, such discrepancies may reflect different motivational aspects elicited by different 357 study designs (e.g., in-lab vs. real-life, experimental vs. self-reported [71]). On the one hand, 358 there is evidence for the external validity of commonly used tasks in the domain of prosociality 359 [72,73]. On the other hand, studies could show that lab-induced prosocial behavior differed 360 drastically from real-life prosocial behavior in a natural-field dictator game, which the authors 361 interpreted as inflated prosocial behavior elicited by the lab context [74]. Further, observability 362 has been shown to be associated with higher prosocial behavior [75], underpinning the 363 assumption that lab-induced prosocial behavior might be different in nature from prosocial 364 behavior in real life. Further, different studies could demonstrate differential drivers of empathy: 365 The likelihood to engage in empathy-eliciting situations increased with monetary incentives, 366 and also when the target of empathy was familiar[76]. Further, increased perceived closeness 367 leads to better perspective taking abilities in older adults [77,78], and both, our own age and the 368 social interaction partners age affect these capacities[79]. 369 Second, it has been shown that the sampling strategy influences the degree of 370 prosociality. Indeed, in many age-comparative studies, the younger age group might consist 371 predominantly of students (e.g., [9,10]), who have been demonstrated to display systematically 372 less prosocial behavior in in-lab studies, also when controlling for age [80]. A strength of the 373 current study in this regard is that the sample was quota-sampled on six key demographic 374 variables (i.e., sex, ethnicity, education, geographic region, income, and age), even though 375 representativeness might not be given with respect to other demographics (e.g., marital status, 376 current living situation), as the sample is not random. Moreover, there was a reduced number 377 of participants sampled from the youngest (18-24 years) and oldest (65 years and older) age 378 group, which is why the two youngest and two oldest groups were merged for all analyses. 379 Thus, conclusions about differences with respect to these groups, and comparability with 380 ageing studies, which typically include a wider range of older adults, are limited. 381 A methodological strength of the dataset analyzed here is the experience-sampling 382 method used to acquire data. EMA studies help to provide closer and deeper insights and to 383 broaden our understanding of real differences across the lifespan, by investigating the 384 frequency, intensity, and complexity of different measures. Advantages lie in increased 385 ecological validity, decreased recall bias, and the possibility to measure variability and change 386 over a short time. These advantages might be particularly relevant for lifespan studies, as there 387 is evidence for an increasing memory-experience gap for negative affect with higher age, and 388 a smaller memory-experience gap for positive affect for the oldest old [44]. Despite its many 389 advantages, experience sampling shares with other self-report measures its susceptibility to 390 biases, like social desirability. More objective, implicit measures of empathy (e.g., physiological 391 reactions towards others’ suffering) have been used in the lab and have also revealed age 392 differences [11,81]. Repeatedly responding to the survey could also have an influence in how 393 people notice empathy opportunities (for a more detailed discussion of representativeness, 394 potential training, and fatigue-effects see Depow et al [47]). 395 396 Conclusion 397 Factors that contribute to successful social interactions, like empathy, prosocial behavior, and 398 well-being are important throughout life. This analysis of cross-sectional experience-sampling 399 data suggests that most of these features show no age-related pattern when measured 400 repeatedly via self-report in daily life. One exception was empathy, which showed a weak 401 inverted U-shaped trend of age over the course of the lifespan. Future studies should take into 402 account methodological differences stemming from varying study-designs and construct 403 definitions, which might be one source of inconsistencies in age-related results in the literature. 404 Multivariate studies combining standardized in-lab experiments with real-life experience- 405 sampling data (e.g., [82]) are a promising venue for future lifespan studies on the social mind. 406 407

408 Methods 409 The current study is based on the dataset from Depow and colleagues [47] who used ecological 410 momentary assessment to increase ecological validity and to explore within-person 411 differences. In the current study, we focused on analyzing adult age-related differences on 412 daily empathy predictors, daily prosocial behavior, and daily subjective well-being in this 413 publicly available dataset. In the following, we reiterate the methodological approach used by 414 Depow and colleagues [47] for completion. 415 416 Participants 417 Quota-sampling was used for a nearly representative U.S. sample, in cooperation with the 418 survey company Qualtrics (Qualtrics®, 2002; www.qualtrics.com). Overall, 3486 participants 419 filled in a demographic questionnaire, including informed consent about participating in a “Daily 420 Interactions’ study”. In a next step, 841 quota-sampled participants were invited via mail to 421 participate in the study. Altogether, 375 completed the baseline survey (trait questionnaire 422 measures, instruction to download the app, glossary of the terms), whereas 285 participants 423 completed the full experience sampling for one week, seven times per day. Participants were 424 excluded if they missed more than 7 surveys in total, which resulted in a sample of 246 425 participants. Further, three participants had to be excluded for the current analysis because of 426 missing information regarding their age. In the dataset from Depow and colleagues [47] age was 427 originally measured in six age groups. The groups were divided into (1) 18 to 24 years (n = 428 14), (2) 25 to 34 years (n = 57), (3) 35 to 44 years (n = 59), (4) 45 to 54 years (n = 51), (5) 55 429 to 64 years (n = 42), and (6) 65 years and older (n = 20). Due to comparably small sample 430 sizes in the youngest and oldest age group Depow and colleagues [47] had merged the two 431 youngest and the two oldest age groups for more homogenous sample sizes across the 432 groups, an approach which we adopted for the current analysis. Thus, the final sample 433 consisted of 243 participants (18-34: n = 71, 45 female, 3 “other”; 35-44: n = 59, 28 female; 434 45-54: n = 51, 34 female; 55 and older: n = 62, 29 female). The age groups differed significantly 435 with respect to their gender and education distribution, but not in surveys answered, income, 436 and religiosity (see Table 3 for descriptive and interferential statistics). Participants answered 437 a total of 7251 surveys, which we consider a dataset characterized by a high degree of 438 ecological validity. 439 440 Table 3. Descriptive and interferential statistics of the final sample characteristics. 18-34 years 35-44 45-55 years 55 + years Test years Statistic Surveys Answered 29.68 ± 11.82 29.80 ± 30.84 ± 12.49 29.24 ± 12.95 F = 0.003, 11.17 p = .96 Gender (m/f/o) 23/45/3 31/28/0 17/34/0 33/29/0 chi² = 16.46, p = .012 Education (Highschool/ 23/14/20/14/0 12/19/16/8/4 18/11/9/8/5 5/17/20/10/10 chi² = GESD or less/some 28.63, p = college/college .004 graduate/ graduate degree) Income (under 25000/ 22/24/21/4 13/23/16/7 18/16/15/2 15/23/16/8 chi² = 7.11, 25000-50000/ 50000- p = .626 100000/over 100000) Religiosity (not at 22/17/14/10/5 20/13/15/4/7 11/12/11/12/2 16/17/14/8/5 chi² = all/slightly/religious/ 10.40, p = strongly/ extremely) .581 441 442 443 Procedure 444 Baseline survey. Participants first underwent a baseline survey, to collect demographic 445 information, as well as trait measures (compare [47]). The baseline survey included a glossary 446 of the important terms (empathy opportunity, emotion share, perspective take, compassion, 447 prosocial behavior), to ensure that all participants would understand the concepts in the same 448 way. For detailed information about the materials see Depow and colleagues [47]. 449 450 Experience-sampling survey. After the baseline survey, participants underwent seven short 451 surveys per day, sent between 10am and 10 pm for one week, delivered by Metricwire 452 (MetricWire®, 2013). Surveys were sent semi randomly, within a 90-minutes window, with 453 minimum 15 minutes gaps, and expiring after 20 minutes. The daily survey had four levels that 454 were built on each other (see Figure 1). On the first level, participants were questioned about 455 their current subjective well-being, and further if they had an opportunity to empathize 456 (empathy opportunity), to be the target of empathy, or to act prosocially in the last 15 minutes. 457 If an empathy opportunity occurred in the last 15 minutes, further details related to this 458 opportunity were interviewed. On a next level, they were asked if they actually experienced 459 some form of empathy for the person or people involved. If they responded “yes”, they were 460 subsequently asked to indicate whether they experienced any (or several) of the subcategories 461 emotion share, perspective take, and compassion. Subsequently, extent, difficulty, and 462 confidence were probed for every subcategory they indicated. The length of the survey was 463 always the same, irrespective of the answer provided. In the original study, a multitude of 464 questions were asked, which are not relevant for the current research question. For further 465 details see Depow and colleagues [47]. 466 467 Statistical Analysis 468 All data were analyzed using R (Version 4.0.3) [83] with RStudio [84]. We adopted the statistical 469 approach described in Depow and colleagues [47], but additionally examined the effect of age 470 group and interactions with age group on our variables of interest. Age differences in sample 471 characteristics (compare table 3) were examined with Chi2-tests for categorical variables and 472 ANOVA for continuous variables using the R package “stats” [83] and “rstatix” [85]. 473 Age-differences regarding empathy outcomes, prosocial behavior, and well-being were 474 analyzed with mixed-effect models, using the mixed function from the “afex” package [86]. 475 Binary outcomes (experience of empathy (yes/no), engaging in prosocial behavior (yes/no)) 476 were analyzed with generalized mixed-effects models, and p-values calculated with likelihood- 477 ratio tests. Well-being as continuous outcome (global well-being score resulting from 478 happiness and sense of purpose) was analyzed by using mixed-effects models, with p-values 479 calculated based on the Satterthwaite method [86]. For each outcome variable, we constructed 480 two models as follows: i) including age as a linear predictor ii) including age both, as a linear 481 and a quadratic predictor. Model selection was conducted based on a loglikelihood ratio test 482 (R function anova). In the results section, the results of the better fitting model are reported, 483 respectively. All models were nested within participant and survey day as random intercepts. 484 In case of convergence warnings, the number of iterations were increased, and the optimizer 485 changed to “bobyqa”, followed by reducing the maximum random effect structure by survey 486 day. We controlled for gender, income, religiosity, and education by including them as 487 covariates, each in a separate model. Controlling for these four different covariates did not 488 change the significance level of our predictors of main interest (i.e., age), respectively, thus, in 489 the results section only the models without covariates are reported. 490 In order to analyze between- and within-subject effects regarding the influence of 491 empathy and age on prosocial behavior and well-being, and prosocial behavior and age on 492 subjective well-being, predictors were centered in different ways. Continuous variables were 493 participant-centered for within-subject effects, and grand-mean centered for between-subject 494 effects. Binary variables were dummy-coded (1 = yes, 0 = no) for within-subject effects and for 495 between-subject effects the grand-mean centered average proportion of yes responses of a 496 participant was used (for further details see [47]). Model statistics, including effect sizes “r” for 497 fixed effects in mixed-effect models derived from R2 [87] were calculated with the summaryh 498 function from the “hausekeep” package [88]. The p-adjust function from the “stats” package [83] 499 was used to correct p-values for multiple testing with the false discovery procedure [89]. Data 500 are publicly available at: https://osf.io/y3ud7/, and scripts are publicly available at: 501 https://osf.io/fdmtg/.

502 503 References 504 1. Ishii-Kuntz, M. Social Interaction and Psychological Well-Being: Comparison across 505 Stages of Adulthood. Int. J. Aging Hum. Dev. 30, 15–36 (1990). 506 2. Steptoe, A., Deaton, A. & Stone, A. A. Subjective wellbeing, health, and ageing. The 507 Lancet 385, 640–648 (2015). 508 3. Batson, C. & Powell, A. Altruism and prosocial behavior. in Handbook of psychology: 509 Personality and social psychology vol. 5 463–484 (John Wiley & Sons, 2003). 510 4. Grühn, D., Rebucal, K., Diehl, M., Lumley, M. & Labouvie-Vief, G. Empathy across the 511 adult lifespan: Longitudinal and experience-sampling findings. Emotion 8, 753–765 (2008). 512 5. Holt-Lunstad, J., Smith, T. B. & Layton, J. B. Social Relationships and Mortality Risk: A 513 Meta-analytic Review. PLoS Med. 7, 20 (2010). 514 6. Cacioppo, J. T. & Cacioppo, S. The growing problem of loneliness. The Lancet 391, 515 426 (2018). 516 7. Stietz, J., Jauk, E., Krach, S. & Kanske, P. Dissociating Empathy From Perspective- 517 Taking: Evidence From Intra- and Inter-Individual Differences Research. Front. Psychiatry 10, 518 126 (2019). 519 8. Beadle, J. N. & de la Vega, C. E. Impact of Aging on Empathy: Review of Psychological 520 and Neural Mechanisms. Front. Psychiatry 10, 331 (2019). 521 9. Reiter, A. M. F., Kanske, P., Eppinger, B. & Li, S.-C. The Aging of the Social Mind - 522 Differential Effects on Components of Social Understanding. Sci. Rep. 7, 11046 (2017). 523 10. Stietz, J. et al. The Aging of the Social Mind: Replicating the preservation of socio- 524 affective and the decline of socio-cognitive processes with age. Accept. R Soc Open Sci 525 (2021). 526 11. Sze, J. A., Gyurak, A., Goodkind, M. S. & Levenson, R. W. Greater emotional empathy 527 and prosocial behavior in late life. Emotion 12, 1129–1140 (2012). 528 12. O’Brien, E., Konrath, S. H., Gruhn, D. & Hagen, A. L. Empathic Concern and 529 Perspective Taking: Linear and Quadratic Effects of Age Across the Adult Life Span. J. 530 Gerontol. B. Psychol. Sci. Soc. Sci. 68, 168–175 (2013). 531 13. Phillips, L. H., MacLean, R. D. J. & Allen, R. Age and the Understanding of Emotions: 532 Neuropsychological and Sociocognitive Perspectives. J. Gerontol. B. Psychol. Sci. Soc. Sci. 533 57, P526–P530 (2002). 534 14. Schieman, S. & Gundy, K. V. The Personal and Social Links between Age and Self- 535 Reported Empathy. Soc. Psychol. Q. 63, 152 (2000). 536 15. Wieck, C. & Kunzmann, U. Age differences in empathy: Multidirectional and context- 537 dependent. Psychol. Aging 30, 407–419 (2015). 538 16. Henry, J. D., Phillips, L. H., Ruffman, T. & Bailey, P. E. A meta-analytic review of age 539 differences in theory of mind. Psychol. Aging 28, 826–839 (2013). 540 17. Bailey, P. E., Ruffman, T. & Rendell, P. G. Age-Related Differences in Social Economic 541 Decision Making: The Ultimatum Game. J. Gerontol. Ser. B 68, 356–363 (2013). 542 18. Bailey, P. E., Brady, B., Ebner, N. C. & Ruffman, T. Effects of Age on Emotion 543 Regulation, Emotional Empathy, and Prosocial Behavior. J. Gerontol. B. Psychol. Sci. Soc. 544 Sci. 75, 802–810 (2020). 545 19. Beadle, J. N., Sheehan, A. H., Dahlben, B. & Gutchess, A. H. Aging, Empathy, and 546 Prosociality. J. Gerontol. B. Psychol. Sci. Soc. Sci. 70, 213–222 (2015). 547 20. Charness, G. & Villeval, M.-C. Cooperation, Competition, and Risk Attitudes: An 548 Intergenerational Field and Laboratory Experiment. (2007). 549 21. Lockwood, P. L. et al. Aging Increases Prosocial Motivation for Effort. Psychol. Sci. 32, 550 668–681 (2021). 551 22. Freund, A. M. & Blanchard-Fields, F. Age-related differences in altruism across 552 adulthood: Making personal financial gain versus contributing to the public good. Dev. Psychol. 553 50, 1125–1136 (2014). 554 23. Bjälkebring, P., Västfjäll, D., Dickert, S. & Slovic, P. Greater Emotional Gain from Giving 555 in Older Adults: Age-Related Positivity Bias in Charitable Giving. Front. Psychol. 7, 846 (2016). 556 24. Beadle, J. N. et al. Effects of age-related differences in empathy on social economic 557 decision-making. Int. Psychogeriatr. 24, 822–833 (2012). 558 25. Roalf, D. R., Mitchell, S. H., Harbaugh, W. T. & Janowsky, J. S. Risk, Reward, and 559 Economic Decision Making in Aging. J. Gerontol. Ser. B-Psychol. Sci. Soc. Sci. 67, 289–298 560 (2012). 561 26. Cavallini, E., Rosi, A., Ceccato, I., Ronchi, L. & Lecce, S. Prosociality in aging: The 562 contribution of traits and empathic concern. Personal. Individ. Differ. 176, 110735 (2021). 563 27. Bruine de Bruin, W. & Ulqinaku, A. Effect of mortality salience on charitable donations: 564 Evidence from a national sample. Psychol. Aging (2020) doi:10.1037/pag0000478. 565 28. Borges, A. P., Reis, A. & Anjos, J. Willingness to pay for other individuals’ healthcare 566 expenditures. Public Health 144, 64–69 (2017). 567 29. Batson, C. D. Empathy-induced altruistic motivation. in Prosocial motives, emotions, 568 and behavior: The better angels of our nature. (eds. Mikulincer, M. & Shaver, P. R.) 15–34 569 (American Psychological Association, 2010). doi:10.1037/12061-001. 570 30. Davis, M. H. Empathy and Prosocial Behavior. The Oxford Handbook of Prosocial 571 Behavior 572 https://www.oxfordhandbooks.com/view/10.1093/oxfordhb/9780195399813.001.0001/oxfordh 573 b-9780195399813-e-026 (2015) doi:10.1093/oxfordhb/9780195399813.013.026. 574 31. Eisenberg, N. & Miller, P. A. The relation of empathy to prosocial and related behaviors. 575 Psychol. Bull. 101, 91–119 (1987). 576 32. Telle, N.-T. & Pfister, H.-R. Positive Empathy and Prosocial Behavior: A Neglected 577 Link. Emot. Rev. 8, 154–163 (2016). 578 33. Carstensen, L. L., Isaacowitz, D. M. & Charles, S. T. Taking time seriously: A theory of 579 socioemotional se- lectivity theory. Am. Psychol. 54, 165–181 (1999). 580 34. Orben, A., Lucas, R. E., Fuhrmann, D. & Kievit, R. Trajectories of adolescent life 581 satisfaction. https://osf.io/y8ruw (2020) doi:10.31234/osf.io/y8ruw. 582 35. Stone, A. A., Schwartz, J. E., Broderick, J. E. & Deaton, A. A snapshot of the age 583 distribution of psychological well-being in the United States. Proc. Natl. Acad. Sci. 107, 9985– 584 9990 (2010). 585 36. Frijters, P. & Beatton, T. The mystery of the U-shaped relationship between happiness 586 and age. J. Econ. Behav. Organ. 82, 525–542 (2012). 587 37. Carstensen, L. L., Mayr, U., Pasupathi, M. & Nesselroade, J. R. Emotional Experience 588 in Everyday Life Across the Adult Life Span. 12 (2000). 589 38. Galambos, N. L., Krahn, H. J., Johnson, M. D. & Lachman, M. E. The U Shape of 590 Happiness Across the Life Course: Expanding the Discussion. Perspect. Psychol. Sci. 15, 591 898–912 (2020). 592 39. Laaksonen, S. A Research Note: Happiness by Age is More Complex than U-Shaped. 593 J. Happiness Stud. (2016) doi:10.1007/s10902-016-9830-1. 594 40. Li, N. Multidimensionality of Longitudinal Data: Unlocking the Age-Happiness Puzzle. 595 Soc. Indic. Res. 128, 305–320 (2016). 596 41. Chi, K., Almeida, D. M., Charles, S. T. & Sin, N. L. Daily prosocial activities and well- 597 being: Age moderation in two national studies. Psychol. Aging 36, 83–95 (2021). 598 42. Van Willigen, M. Differential Benefits of Volunteering Across the Life Course. J. 599 Gerontol. B. Psychol. Sci. Soc. Sci. 55, S308–S318 (2000). 600 43. Handbook of research methods for studying daily life. xxvii, 676 (The Guilford Press, 601 2012). 602 44. Neubauer, A. B., Scott, S. B., Sliwinski, M. J. & Smyth, J. M. How was your day? 603 Convergence of aggregated momentary and retrospective end-of-day affect ratings across the 604 adult life span. J. Pers. Soc. Psychol. 119, 185–203 (2020). 605 45. Bielak, A. A. M., Cherbuin, N., Bunce, D. & Anstey, K. J. Intraindividual variability is a 606 fundamental phenomenon of aging: Evidence from an 8-year longitudinal study across young, 607 middle, and older adulthood. Dev. Psychol. 50, 143–151 (2014). 608 46. Nezlek, J. B., Feist, G. J., Wilson, F. C. & Plesko, R. M. Day-to-Day Variability in 609 Empathy as a Function of Daily Events and Mood. J. Res. Personal. 35, 401–423 (2001). 610 47. Depow, G. J., Francis, Z. & Inzlicht, M. The Experience of Empathy in Everyday Life. 611 Psychol. Sci. 16 (2021) doi:https://doi.org/10.1177/0956797621995202. 612 48. Mather, M. & Carstensen, L. L. Aging and motivated cognition: the positivity effect in 613 attention and memory. Trends Cogn. Sci. 9, 496–502 (2005). 614 49. Cornwell, B. Age Trends in Daily Social Contact Patterns. Res. Aging 33, 598–631 615 (2011). 616 50. Lang, F. R. & Carstensen, L. L. Close emotional relationships in late life: Further 617 support for proactive aging in the social domain. Psychol. Aging 9, 315–324 (1994). 618 51. Cuddy, A. J. C., Norton, M. I. & Fiske, S. T. This Old Stereotype: The Pervasiveness 619 and Persistence of the Elderly Stereotype. J. Soc. Issues 61, 267–285 (2005). 620 52. North, M. S. & Fiske, S. T. Modern attitudes toward older adults in the aging world: A 621 cross-cultural meta-analysis. Psychol. Bull. 141, 993–1021 (2015). 622 53. Bergman, Y. S. & Bodner, E. Ageist attitudes block young adults’ ability for compassion 623 toward incapacitated older adults. Int. Psychogeriatr. 27, 1541–1550 (2015). 624 54. Laganá, L. et al. A Randomized Controlled Study on the Effects of a Documentary on 625 Students’ Empathy and Attitudes towards Older Adults. Psychol. Cogn. Sci. - Open J. 3, 79– 626 88 (2017). 627 55. Cornwell, B., Laumann, E. O. & Schumm, L. P. The Social Connectedness of Older 628 Adults: A National Profile. Am. Sociol. Rev. 73, 185–203 (2008). 629 56. Kohli, M., Hank, K. & Künemund, H. The social connectedness of older Europeans: 630 patterns, dynamics and contexts. J. Eur. Soc. Policy 19, 327–340 (2009). 631 57. McAdams, D. P. & Olson, B. D. Personality Development: Continuity and Change Over 632 the Life Course. Annu. Rev. Psychol. 61, 517–542 (2010). 633 58. Wojciechowska, L. Subjectivity and generativity in midlife. Pol. Psychol. Bull. 48, 38– 634 43 (2017). 635 59. Hall, J. A. & Schwartz, R. Empathy present and future. J. Soc. Psychol. 159, 225–243 636 (2019). 637 60. Kelley. Interpretation of educational measurements / by Truman Lee Kelley. 394 638 (1927). 639 61. Marsh, H. W. Sport motivation orientations: Beware of jingle-jangle fallacies. J. Sport 640 Exerc. Psychol. 16, 365–380 (1994). 641 62. Kanske, P., Böckler, A., Trautwein, F.-M. & Singer, T. Dissecting the social brain: 642 Introducing the EmpaToM to reveal distinct neural networks and brain–behavior relations for 643 empathy and Theory of Mind. NeuroImage 122, 6–19 (2015). 644 63. Klimecki, O. M., Leiberg, S., Ricard, M. & Singer, T. Differential pattern of functional 645 brain plasticity after compassion and empathy training. Soc. Cogn. Affect. Neurosci. 9, 873– 646 879 (2014). 647 64. Bailey, P. E., Brady, B., Ebner, N. C. & Ruffman, T. Effects of Age on Emotion 648 Regulation, Emotional Empathy, and Prosocial Behavior. J. Gerontol. Ser. B (2020) 649 doi:10.1093/geronb/gby084. 650 65. Gaesser, B., Dodds, H. & Schacter, D. L. Effects of aging on the relation between 651 episodic simulation and prosocial intentions. Mem. Hove Engl. 25, 1272–1278 (2017). 652 66. Gibson, K. L., McKelvie, S. J. & de Man, A. F. Personality and culture: A comparison 653 of Francophones and Anglophones in Québec. J. Soc. Psychol. 148, 133–165 (2008). 654 67. Li, T. & Siu, P.-M. Socioeconomic Status Moderates Age Differences in Empathic 655 Concern. J. Gerontol. B. Psychol. Sci. Soc. Sci. (2019) doi:10.1093/geronb/gbz079. 656 68. Philippe Rushton, J., Fulker, D. W., Neale, M. C., Nias, D. K. B. & Eysenck, H. J. Ageing 657 and the relation of aggression, altruism and assertiveness scales to the Eysenck personality 658 questionnaire. Personal. Individ. Differ. 10, 261–263 (1989). 659 69. Dolan, P., Kudrna, L. & Stone, A. The Measure Matters: An Investigation of Evaluative 660 and Experience-Based Measures of Wellbeing in Time Use Data. Soc. Indic. Res. 134, 57–73 661 (2017). 662 70. Lelkes, O. Happiness Across the Life Cycle: Exploring Age-Specific Preferences. 17 663 (2008). 664 71. Böckler, A., Tusche, A. & Singer, T. The Structure of Human Prosociality: Differentiating 665 Altruistically Motivated, Norm Motivated, Strategically Motivated, and Self-Reported Prosocial 666 Behavior. Soc. Psychol. Personal. Sci. 7, 530–541 (2016). 667 72. Franzen, A. & Pointner, S. The external validity of giving in the dictator game: A field 668 experiment using the misdirected letter technique. Exp. Econ. 16, 155–169 (2013). 669 73. Peysakhovich, A., Nowak, M. A. & Rand, D. G. Humans display a ‘cooperative 670 phenotype’ that is domain general and temporally stable. Nat. Commun. 5, 4939 (2014). 671 74. Winking, J. & Mizer, N. Natural-field dictator game shows no altruistic giving. Evol. 672 Hum. Behav. 34, 288–293 (2013). 673 75. Bradley, A., Lawrence, C. & Ferguson, E. Does observability affect prosociality? Proc. 674 R. Soc. B Biol. Sci. 285, 20180116 (2018). 675 76. Ferguson, A. M., Cameron, C. D. & Inzlicht, M. Motivational effects on empathic 676 choices. J. Exp. Soc. Psychol. 90, 104010 (2020). 677 77. Zhang, X., Fung, H. H., Stanley, J. T., Isaacowitz, D. M. & Ho, M. Y. Perspective taking 678 in older age revisited: A motivational perspective. Dev. Psychol. 49, 1848–1858 (2013). 679 78. Zhang, X. et al. Plasticity in older adults’ theory of mind performance: the impact of 680 motivation. Aging Ment. Health 22, 1592–1599 (2018). 681 79. Reiter, A. M. F., Diaconescu, A. O., Eppinger, B. & Li, S.-C. Human aging alters social 682 inference about others’ changing intentions. Neurobiol. Aging 103, 98–108 (2021). 683 80. Carpenter, J., Connolly, C. & Myers, C. K. Altruistic behavior in a representative dictator 684 experiment. Exp. Econ. 11, 282–298 (2008). 685 81. Chen, Y.-C., Chen, C.-C., Decety, J. & Cheng, Y. Aging is associated with changes in 686 the neural circuits underlying empathy. Neurobiol. Aging 35, 827–836 (2014). 687 82. Grosse Rueschkamp, J. M., Brose, A., Villringer, A. & Gaebler, M. Neural correlates of 688 up-regulating positive emotions in fMRI and their link to affect in daily life. Soc. Cogn. Affect. 689 Neurosci. 14, 1049–1059 (2019). 690 83. R Core Team. R: A language and environment for statistical computing. (R Foundation 691 for Statistical Computing, 2020). 692 84. RStudio Team. RStudio: Integrated Development for R. (RStudio, PBC, 2020). 693 85. Kassambara, A. rstatix: Pipe-Friendly Framework for Basic Statistical Tests. (2021). 694 86. Singmann, H., Bolker, B., Westfall, J., Aust, F. & Ben-Shachar, M. S. afex:Analysis of 695 Factorial Experiments. (2021). 696 87. Edwards, L. J., Muller, K. E., Wolfinger, R. D., Qaqish, B. F. & Schabenberger, O. An 697 R2 statistic for fixed effects in the linear mixed model. Stat. Med. 27, 6137–6157 (2008). 698 88. Lin, H. hausekeep: Miscellaneous helper and utility functions. (2019). 699 89. Benjamini, Y. & Hochberg, Y. Controlling the False Discovery Rate: A Practical and 700 Powerful Approach to Multiple Testing. J. R. Stat. Soc. Ser. B Methodol. 57, 289–300 (1995). 701

702 Acknowledgements 703 The study was supported by the German Research Foundation (DFG RE 4449/1-1). AMFR 704 acknowledges further support by a grant from the German Research Foundation (DFG) (SFB 705 940/3 B7) and by a 2020 NARSAD Young Investigator Grant from the Brain & Behavior 706 Research Foundation. 707

708 Author Contributions 709 L.P., and A.M.F.R. conceived the study. J.D., and M.I. developed the survey and collected 710 data. L.P. and A.M.F.R. analyzed the data., L.P., J.S., P.K., and S-C. L. and A.M.F.R. 711 interpreted the data. L.P. wrote the manuscript. All authors provided critical revisions and 712 approved the final version of the manuscript. 713

714 Additional information 715 Competing Interests: The authors declare that they have no competing interests. 716

717 Data availability 718 All data, and materials are publicly available at: https://osf.io/y3ud7/, and scripts are publicly 719 available at: https://osf.io/fdmtg/. 720 721