Aging entails many changes in bodily and mental health. Although physical performances and cognitive abilities decline, emotional functions seem maintained or even enhanced in older relative to younger adults1,2. Older adults regulate their emotional states well, a crucial capacity for affective well-being and healthy aging3. Unlike younger adults, they often prioritize social and emotional interactions over other goals and show a ‘positivity bias’ in emotion perception4. In contrast, maladaptive emotional reactivity and impaired emotion regulation are related to psychopathologies such as anxiety, depression, worry and rumination throughout the lifespan5,6, including in aging7. Maladaptive affective styles may represent an important risk factor for dementia8,9,10, a major health burden11. However, the neural substrates underpinning proficient socio-affective processing and emotional resilience in older adults remain unresolved.

Maladaptive affective style may lead to ‘emotional inertia’, whereby emotions tend to carry over from one moment to the next12. Emotional inertia may reflect unsuccessful recovery mechanisms following affective events and low resilience to stress, associated with higher depression risks13,14 and higher trait anxiety and rumination tendencies15. Recent neuroimaging studies investigated emotional inertia using ‘task–rest’ paradigms16,17,18,19,20,21,22 where brain activity is probed not only during active stimulus processing but also in spontaneous post-task resting periods while the brain returns to homeostatic balance23,24. Positive or negative emotions evoked by images or videos were found to induce lasting carryover effects on activity and/or connectivity of brain networks during subsequent rest17,21. These effects may occur at different timescales (from seconds25 to minutes20), following different tasks (from passive viewing to active regulation of emotions16) and across different conditions of emotional valence and intensity18,19.

At the brain level, emotion carryover effects generally affect the functional dynamics of the default mode network (DMN) in the form of either increased19,25 or decreased17,18 activity patterns in regions comprising the medial prefrontal cortex (MPFC), posterior cingulate cortex (PCC), precuneus and inferior parietal cortex. These DMN regions are usually active when individuals are free to let their minds wander in undisturbed conditions26. Similar effects have been observed in the insula and amygdala17—two regions involved in emotional and social processing27,28,29. Slower recovery of amygdala activity after emotional videos correlated with higher anxiety and ruminations15. Subcortical limbic regions in amygdala and striatum also display sustained changes in their functional connectivity with cortical areas in medial PFC and PCC during rest after negative emotions17 and reward21. These findings converge with studies showing that disturbances in functional connectivity of the amygdala with medial parts of the DMN at rest are associated with anxiety (for example, decreased connectivity with MPFC30) and mood disorders (for example, increased connectivity with PCC31). These long-lasting carryover effects of emotions on activity and connectivity of limbic networks may provide an important neural marker of emotional regulation style and affective resilience.

However, previous work concerning emotional carryover focused on young healthy participants. It remains unknown whether emotional inertia occurs in older adults, how it is modified given the well-known ‘positivity effect’ observed in this population relative to younger adults1,32, and how age impacts DMN functional dynamics in affective contexts. Unlike young adults, older people fail to deactivate the DMN during externally directed cognitive tasks33 and show increased DMN connectivity with cognitive-related prefrontal regions34. Yet, little is known about how aging affects DMN responses to emotion and how this relates to other cognitive or socio-affective abilities.

In addition, individual differences in empathy may strongly influence responses to negative socio-affective events15,17, and possibly recovery from these events20. Because social competencies are relatively preserved in older adults2, socially significant emotional events offer an optimal window to probe affective reactivity and recovery in this population. Empathy is only scarcely studied in older people2,35,36,37, with some evidence for a decline of cognitive empathy but intact or even enhanced affective empathy and altruistic behaviors2,37,38. However, brain responses to seeing others’ pain are reduced in anterior insula (AI) and anterior cingulate cortex (ACC)35, two regions implicated in pain processing, negative affect and salience detection28,39. In contrast, empathy-related responses may increase in superior temporal sulcus (STS) and temporoparietal junction (TPJ)36, brain regions frequently associated with the Theory of Mind and perspective taking40. Yet, despite the importance of social interactions and emotional resilience for healthy aging41,42, neural substrates underlying the recovery from negative events, as well as their link with empathic skills, personality and psycho-affective traits, have not been investigated during aging.

To address these issues, we designed a ‘task–rest’ paradigm combining two lines of research: short (10–18 s) empathy-inducing videos from the socio-affective video task (SoVT)43 were shown interspersed with rest periods of 90 s (similarly to Eryilmaz and colleagues17) while participants underwent functional magnetic resonance imaging (fMRI) of brain activity (Extended Data Fig. 1). Videos consisted of short silent scenes depicting suffering people (high emotion, HE) or people in everyday life situations (low emotion, LE). By adding resting-state periods after blocks of videos of each kind, our paradigm allowed probing how the aging brain reacts both during and after exposure to emotionally challenging social information.

Using this paradigm, we conducted two experiments to assess emotion-related carryover effects in large samples of healthy older and young participants (N = 182). First (experiment 1), we compared the neural substrates of emotional recovery between old (n = 26; mean age, 68.7 years) and young (n = 29; mean age, 24.5 years) participants, allowing us to validate our paradigm and identify age-related effects. Next (experiment 2), we replicated this experiment in a large sample of older adults (n = 127; mean age, 68.8 years) and specifically investigated whether emotional inertia in brain networks is modulated by empathy and individual traits relevant for healthy aging, including rumination and anxiety. We hypothesized that exposure to others’ suffering (relative to neutral social situations) should: (1) engage brain regions implicated in emotional saliency and empathy (that is, insula and anterior medial cingulate cortex (aMCC)), but with lower responses in older than young adults;35 (2) induce subsequent carryover in functional connectivity at rest between emotion-related regions and DMN, with differential age-dependent patterns; and (3) unveil neural substrates of emotional inertia reflecting individual variability in anxiety, ruminative thinking and negative emotions, and thus point to functional biomarkers of affective risk factors for pathological aging8,9,10. We also expected to (4) observe a ‘positivity effect’ as often reported in older adults4, characterized by higher positive affect during the task in older compared to younger adults, and elucidate any relationship to affective empathy.


Participant characteristics

Demographical data, psycho-affective traits, cognitive abilities and socio-emotional questionnaires are reported in Table 1 and Fig. 1.

Table 1 Participant characteristics
Fig. 1: Participants’ characteristics in terms of psycho-affective traits and socio-emotional competencies for both experiment 1 and experiment 2.
figure 1

Age-related differences (experiment 1) were tested with t-tests (two-sided), and P values for significant results are displayed. Older versus younger participants (experiment 1) did not differ in trait anxiety, affective empathy and emotion regulation scores. However, older adults reported lower scores of cognitive empathy in the perspective taking (t53 = 4.2, P < 0.001, d = 1.13, two-tailed) and the fantasy subscales of the IRI (t52.3 = 3, P = 0.004, d = 0.81, two-tailed). Older adults also had lower scores in reflective rumination (t52.7 = 2.62, P = 0.01, d = 0.7, two-tailed). The two independent samples of older adults, that is, experiment 1 (n = 29 younger and n = 26 older adults) and experiment 2 (N = 127 older adults), did not differ in any of the scores (all t ≤ 1.6, all P ≥ 0.09, two-tailed). Gray diamonds denote younger adults; white dots denote older adults. GDS (for older adults only); BDI, Beck Depression Inventory (for younger adults only); STAI, State-Trait Anxiety Inventory; ERQ, Emotion Regulation Questionnaire (See Table 1 for further statistical details).

Behavioral responses in the SoVT–rest task and age effects

Subjective ratings (empathy, positive and negative affect) showed that HE videos induced higher levels of negative affect, lower positive affect and higher empathy scores than LE videos in both younger and older participants in both experiments, validating successful emotional elicitation with the SoVT–rest task (Fig. 2a). All video sets elicited similar emotional responses (Supplementary Fig. 1) and differences between HE and LE were not affected by sex (Supplementary Fig. 6).

Fig. 2: Empathy, positive affect, and negative affect during high and low emotion videos across experiments and age groups.
figure 2

a, Self-reported scores of empathy, positive affect and negative affect for HE and LE videos. HE and LE videos were compared using pairwise t-tests for each of these ratings. Results from experiment 1 were fully replicated in experiment 2. Participants reported higher levels of empathy (Exp 1: t54 = 14.35, P < 0.001, d = 1.67, two-tailed; Exp 2: t126 = 14.5, P < 0.001, d = 1.31, two-tailed), higher negative affect (Exp 1: t54 = 23.35, P < 0.001, d = 3.77, two-tailed; Exp 2: t126 = 26.9, P < 0.001, d = 2.89, two-tailed) and lower positive affect (Exp 1: t54 = −16.85, P < 0.001, d = −2.31, two-tailed; Exp 2: t126 = −18.9, P < 0.001, d = −2.31, two-tailed), when presented with HE as compared to LE videos. The box plots show the interquartile range (25th percentile, median and 75th percentile), the whiskers (indicating variability outside the interquartile range) and the individual data points. Significant differences between age groups or video conditions are marked by ***P < 0.001, uncorrected for multiple comparisons. b, Scatterplots illustrate Spearman correlations between age and scores of empathy, positive affect and negative affect. c, Scatterplots illustrate Spearman correlations between scores of empathy and affective ratings. For b and c, correlation coefficients were obtained using two-sided tests, and analyses were computed together; therefore, P values are corrected for multiple comparisons using the FDR method. For b and c, significant P values are marked in bold. Dots represent averaged values for each participant per condition; dots/solid lines indicate older adults, and diamonds/dashed lines indicate younger adults; nExp1 = 55 (29 younger and 26 older adults), nExp2 = 127 older adults. Red indicates HE videos; gray indicates LE videos.

In experiment 1, we also examined age-dependent differences in ratings. Independent analysis of variance (ANOVA) showed main effects of age on empathy (F(1,53) = 10.8, P = 0.002) and positive affect (F(1,53) = 24, P < 0.001), not on negative affect (F(1,53) = 1.01, P = 0.3). Follow-up two-sample t-tests revealed higher empathy levels in older than younger adults only for LE videos (t51.5 = 4.45, P < 0.001, d = 1.19, two-tailed) as well as higher positive emotions for both HE (t36.5 = 4.63, P < 0.001, d = 1.29, two-tailed) and LE (t50.1 = 3.68, P < 0.001, d = 0.98, two-tailed) videos. Our two older samples (experiments 1 and 2) did not differ in any of the scores (all t ≤ 1.8, P ≥ 0.07; Fig. 2a) except for even higher empathy ratings for LE videos in older adults from experiment 2 than those from experiment 1 (t36 = 2.20, P = 0.03, d = 0.47, two-tailed).

We further tested for age-dependent effects within each age group separately (young and older) by computing Spearman correlations between age (as a continuous variable) and ratings (collapsing both older samples from experiments 1 and 2). Age correlated negatively with negative affect (rho = −0.2, false discovery rate (FDR)-adjusted P value (PFDR) = 0.03) and positively with positive affect (rho = 0.25, PFDR = 0.006) during HE video for older individuals but not younger adults. In addition, age correlated positively with empathy for LE videos in the young adults (rho = 0.44, PFDR = 0.03) but not older adults (Fig. 2b). The same results were obtained after excluding eight older adults who reported ‘moderate’ depression levels (Geriatric Depression Scale, GDS > 5; ref. 44; Supplementary Table 5). These participants were therefore kept for subsequent analyses.

To test whether empathy influenced positive and negative affect, we computed Spearman correlations between these rating scales. For both younger and older adults, higher empathy predicted increased negative affect during HE videos (YOUNG: rho = 0.86, PFDR < 0.001; OLD: rho = 0.63, PFDR < 0.001) and higher positive affect during LE videos (YOUNG: rho = 0.75, PFDR < 0.001; OLD: rho = 0.65, PFDR < 0.001). Interestingly, during HE videos, empathy correlated negatively with positive affect for older (rho = −0.35, PFDR < 0.001) but not younger (rho = 0.27, PFDR = 0.18) adults; whereas during LE videos, empathy correlated positively with negative affect for the younger (rho = 0.63, PFDR < 0.001) but not the older (rho = 0.13, PFDR = 0.14) adults (Fig. 2c).

Brain activity during the SoVT–rest task

Age-related differences in response to others’ suffering

After confirming differential brain activity during videos and rest periods (Supplementary Fig. 2), we determined the effect of the emotional condition in each age group, as well as age-related differences (experiment 1). In both groups, comparing HE and LE videos (voxel-wise P < 0.05 using family-wise error (FWE) correction; and P < 0.001 uncorrected, k = 20) demonstrated increases in left TPJ, right inferior-frontal gyrus (IFG), as well as temporal and occipital cortices (Fig. 3a and Supplementary Table 3a). Older adults showed larger activations in PCC and dorsomedial prefrontal cortex (dMPFC), whereas younger adults showed additional increases in AI and periaqueductal gray (PAG) (Fig. 3a). A direct between-group comparison (2 × 2 ANOVA) revealed stronger cortical responses for the older adults in bilateral angular gyrus (TPJ/IPL) and dorsolateral PFC. The young adults showed stronger activation in subcortical areas of ventral striatum and PAG, as well as in sensory areas in parietal and occipitotemporal cortices (Fig. 3b).

Fig. 3: Brain regions with greater activation during high emotion videos compared to low emotion videos across experiments and age groups.
figure 3

a, Brain maps for younger (n = 29) and older (n = 26) adults in experiment 1. b, Between-age-groups difference in experiment 1. For display purposes, results are thresholded at uncorrected P < 0.001, with a minimum cluster size of (k = 20). Clusters surviving correction for multiple comparisons (P FWE < 0.05 at the cluster level) are surrounded in white-dotted circles. c, Brain maps for older adults (N = 127) in experiment 2. Results survived FWE correction at the voxel level (P < 0.05 FWE corrected). For ac, t-statistics were obtained from the corresponding contrasts of interest (t-tests, one-sided). Overall, HE > LE videos activated regions previously reported as part of the Empathy Network (bilateral AI, aMCC), and regions in the Theory of Mind network (left TPJ; dMPFC) and the Compassion network (VS, ventral striatum)46,50. Full statistics on each cluster and related brain areas are in Supplementary Table 3a,b. Given the age-related positivity in affective ratings observed behaviorally in older adults (Fig. 2), we also computed regression analyses to examine whether whole-brain activity during emotional videos (HE–LE) was modulated by individual differences in self-reported positive affect scores, but this revealed no significant effects across experiments or age groups (Supplementary Fig. 7).

These activation patterns across both experiments overlap with brain networks classically associated with empathy28,39, compassion43,45,46, as well as cognitive and affective theory of mind40,47.

Carryover effects during post-emotion rest periods

To test for carryover effects of emotional videos on brain activity17, we compared rest periods after HE videos to those after LE videos. In experiment 1, this contrast (voxel-wise P < 0.001 uncorrected, k = 20) revealed greater activations, mostly in the older group, involving MPFC, left AI, right IFG, several temporoparietal cortices and right hippocampus (Fig. 4a and Supplementary Table 3a). The younger adults showed more limited increases, predominating in MCC. A direct between-group comparison (2 × 2 ANOVA) confirmed that older adults engaged these regions (AI, IFG and dMPFC) more strongly, with further significant effects in left MTG and left amygdala, whereas the younger adults showed higher activity in left hippocampus and precentral motor regions (Fig. 4b and Supplementary Table 3a).

Fig. 4: Carryover effects on brain activity at rest subsequent to high (post-HE) versus low emotion (post-LE) videos.
figure 4

a, Brain maps for younger (n = 29, blue clusters) and older adults (n = 26, violet clusters) in experiment 1. b, Direct comparisons of brain maps representing significant age-related differences in experiment 1, with threshold uncorrected P < 0.001 and minimum cluster size k = 20. Clusters surviving correction for multiple comparison (P FWE < 0.05 at the cluster level) are surrounded in white-dotted circles. c, Brain maps for older adults (N = 127) in experiment 2. Violet clusters show significant increases in post-HE > post-LE rest periods. Green clusters show the overlap of these activations with emotional responses to HE > LE videos (gray). Results are thresholded at P < 0.05 corrected for multiple comparisons using FWE correction at the voxel level. dh, Magnitude and time course of brain activity (parameter estimates) for relevant regions during the different task periods in older adults from experiment 2. d, Example of a region (in MCC) responding to HE versus LE videos, but showing no significant difference during rest after HE versus LE videos. e,f, Example of regions (PCC/Prec and dMPFC) responding to HE > LE videos and showing significant carryover with sustained activity during subsequent rest. g,h, The r.AMYG and the right ventral AI did not reliably respond to HE versus LE videos but showed significant increases in activations during post-HE rest. Pink and gray lines track activity time courses during HE and LE conditions, respectively. Pink and gray bars indicate activity during videos (three in each block, ~45 s), and white bars indicate activity during subsequent rest (3 bins of 30 s). ***P < 0.05 FWE corrected. t-statistics were obtained from the corresponding contrasts of interest (t-tests, one-sided). Data for younger adults are not plotted given the lack of significant increases in brain activity at rest in this group (see a). Prec, precuneus; MTG, middle temporal gyrus; Hipp, hippocampus.

In experiment 2, similar regions were found, surviving a more stringent statistical threshold and replicating our results in older adults from experiment 1. Post-HE > post-LE rest periods (voxel-wise P < 0.05 FWE corrected) disclosed higher activity among midline DMN nodes (ACC/dMPFC, and precuneus/PCC), as well as in the right amygdala and ventral right anterior insula (Fig. 4c and Supplementary Table 3b).

The larger sample size in experiment 2 allowed us to conduct additional analyses to assess whether carryover effects at rest directly resulted from higher activity in the same regions during videos periods. We identified voxels with the strongest emotional response (HE > LE) across the two periods (videos and rest) by applying an inclusive mask from one contrast (for example, videos: HE > LE) to the other contrast (rest: post-HE > post-LE), with a strict threshold for both (P < 0.00001). This overlap analysis identified brain areas showing persistent emotion-related increases across both video and rest conditions (Fig. 4c). Results highlighted a restricted set of regions, mainly dMPFC and PCC, where greater emotional activation during videos was followed by emotional carryover effects during rest after videos (Fig. 4e,f). Other regions activated during emotional videos did not display differential carryover effects in subsequent rest, including visual cortex and MCC (Fig. 4d). Interestingly the right amygdala and the right ventral AI did not show significant differences for HE > LE during videos but were robustly activated in post-HE > post-LE rest periods (Fig. 4g,h). Figure 4 illustrates this dissociation by plotting brain activity (contrasts estimates) over successive time bins of the video and rest periods across emotion conditions (Fig. 4d–h).

Differential effect on network connectivity in older adults

To further assess emotional carryover in brain activity dynamics, we examined changes in functional connectivity within and between the DMN, the empathy network, and the bilateral amygdala, during post-HE versus post-LE rest periods. Connectivity was computed using Pearson correlation matrices for each condition and each participant, and then compared between post-HE and post-LE using permutation tests (Methods and Extended Data Fig. 2). Consistent with functionally coherent activity within each network, we observed general patterns of intra-network connectivity (Empa-Empa, Amy-Amy, DMN-DMN) for both conditions and both age groups (Supplementary Fig. 3)

In experiment 1, differences in functional connections of the DMN were found only in older adults: relative to post-LE rest, post-HE periods rest showed stronger coupling of PCC with right amygdala (t = 2.52, P = 0.008, Z = 2.4, one-tailed) and left amygdala (t = 2.1, P = 0.02, Z = 1.97, one-tailed), as well as between anterior medial prefrontal cortex (aMPFC) and right amygdala (t = 2.02, P = 0.03, Z = 1.95, one-tailed), and between aMPFC and left amygdala (t = 2.24, P = 0.02, Z = 2.04, one-tailed; Fig. 5a). No such differences were found for young adults (Fig. 5a).

Fig. 5: Functional connectivity between pairs of regions of interest for the different rest conditions in experiment 1 and experiment 2.
figure 5

a, Correlation matrices computed on the difference between the two rest conditions, showing post-emotion increases (green) and decreases (orange) in functional connectivity (FC) between regions, for each age group in experiment 1 (N = 55). Left and right halves of the matrix with respect to the diagonal depict values for opposite contrasts (upper part, post-HE > post-LE rest; lower part, post-LE > post-HE rest). Significant emotion effects (Z > 1.64) are marked by an asterisk (*P < 0.05, one-tailed). b, Age-related differences between emotion effects on FC (old versus young, two-sample t-tests) showed increases predominating in older (violet) or younger (blue) adults. Significant differences were observed only for older relative to younger adults (*P < 0.05, two-tailed, uncorrected). c, Correlation matrix showing significant differences in FC between rest conditions in older adults (N = 127) from experiment 2. For a and c, differences between correlation matrices (post-HE versus post-LE rest) were determined by non-parametric permutation statistics (Methods and Extended data Fig. 2), with a significance threshold of Z > 1.64 (equivalent to P < 0.05, one-tailed, given the observed increases without decreases in general linear model (GLM) analysis, uncorrected). The upper-right schematic illustrates a priori selected ROIs for this analysis, including nodes from the DMN (aMPFC), empathy network (Empa, left and right AI; MCC) and bilateral amygdalae (Amy, left and right AMYG).

These age-related differences were confirmed by t-tests for each connectivity node between younger and older adults at rest (post-HE > post-LE). Compared to younger adults, the older adults showed enhanced connectivity between left amygdala and PCC (t = 2.12, P = 0.03, two-tailed), and between left amygdala and aMPFC (t = 2.08, P = 0.04, two-tailed; Fig. 5b).

Experiment 2 revealed similar patterns of increased connectivity in our larger group of older adults. Significant differences were observed for highly selective functional connections of the DMN with limbic areas: compared to post-LE rest, the post-HE rest periods induced stronger coupling between the PCC and right amygdala (t = 1.82, P = 0.03, Z = 1.81 one-tailed), as well as between the aMPFC and left insula (t = 1.98, P = 0.02, Z = 2.02 one-tailed). There was also higher coupling among the bilateral amygdala during post-HE versus post-LE rest periods (right with left, t = 1.88, P = 0.02, Z = 1.95 one-tailed; Fig. 5c). Further t-tests revealed that the increased functional coupling induced by emotional videos was stronger in r.AMYG–PCC than in other pairs of nodes engaging either the right amygdala or the PCC such as l.AI–PCC, r.AI–PCC and r.AMYG–aMPFC (Fig. 6a).

Fig. 6: Between-network functional connectivity during rest periods after HE > LE videos.
figure 6

a, Pairs of nodes were ordered from left to right according to connectivity strength. The r.AMYG–PCC pair (red) was significantly more connected than other pairs involving either the PCC or the right amygdala (in bold). Significant comparisons from t-tests (one-tailed) are marked with corresponding P values; NS, not significant. b, Pearson (r) and Spearman (rho) correlations show that higher r.AMYG–PCC connectivity during rest periods after HE > LE videos was positively related to trait anxiety (STAI-trait) and rumination (RRS total). The shaded area represents the 95% confidence interval bands. c, Comparison of r.AMYG–PCC connectivity between the group who verbally reported negative content during thought probes (present) versus the group who did not (absent), for both HE and LE conditions. After HE videos, 59 (54%) participants reported negative content, 30 (28%) did not report negative content and 20 (18%) were ambiguous. After LE videos, only 41 (37%) reported negative content, 50 (45)% reported no negative thoughts and 19 (17%) were ambiguous. The mean r.AMYG–PCC connectivity difference was then compared between the two groups with versus without negative thoughts (observed difference = 0.08), relative to a null distribution built by permuting the labels 5,000 times. Participants reporting negative content in their thoughts (54%) versus those who did not (28%) showed higher r.AMYG–PCC connectivity in the HE condition (P = 0.02, one-tailed), but no difference in the LE condition (observed difference = 0.06; P = 0.07, one-tailed). Data for a and c depict mean values and error bars represent 95% confidence intervals. N = 127 older adults from experiment 2. Red indicates the HE condition; gray indicates the LE condition. r.AMYG–PCC indicates connectivity between the right amygdala and the PCC. Percentages in the text are rounded. Data are from experiment 2.

Neural emotional carryover and psycho-affective measures

Our fMRI connectivity analyses identified a selective impact of emotional videos on functional brain connectivity of the posterior DMN (PCC) with the amygdala, replicated across two independent experiments carried out at different sites. These results provide a plausible neural mechanism underlying emotional inertia12,17 that is specific to older adults and may thus offer a valuable biomarker of homeostatic emotion regulation processes in aging.

In experiment 2, we could further examine whether this connectivity pattern reflected individual differences in socio-emotional abilities and psycho-affective traits. We tested for a correlation between significant changes in connections between two regions of interest (ROIs; Z > 1.64 for difference post-HE versus post-LE rest) and scores on trait anxiety (STAI-trait), rumination (Rumination Response Scale, RRS) and empathy (Interpersonal Reactivity Index, IRI). This revealed a significant positive relationship between the magnitude of changes in r.AMYG–PCC connectivity (rest HE > rest LE) and individual levels of trait anxiety (r = 0.21, P < 0.01, two-tailed) and rumination (rho = 0.22, P < 0.01, two-tailed; Fig. 6b), but no correlation with empathy (r = 0.1, P = 0.25, two-tailed). On the other hand, the age-dependent positivity effect observed behaviorally in affective ratings (Fig. 2) was found not to be related to emotional carryover in brain connectivity patterns, for neither r.AMYG–PCC nor aMPFC–insula (Supplementary Fig. 8).

Neural emotional carryover and thought probes

Because rumination scores were positively associated with greater changes in functional coupling between r.AMYG and PCC during post-emotion rest (HE > LE), we reasoned that some participants (that is, with higher ruminative tendencies) may hold more negative affect-related content in their thoughts during rest periods after emotional videos. This was directly tested in experiment 2 by using an explicit thought probe given after different rest periods (Extended Data Fig. 1b).

Behavioral results (Methods) showed more frequent reports of negative (54%) than non-negative (28%) thoughts in response to probes after emotional rest periods (X2 (1, N = 109) = 45.88, P < 0.001, two-tailed). Conversely, negative thoughts were less frequent (37%) than non-negative (45%) thoughts after LE video (X2 (1, N = 110) = 51.59, P < 0.001, two-tailed; Supplementary Table 4 and Fig. 6c).

We then compared the r.AMYG–PCC connectivity between a subgroup of participants who reported negative content in spontaneous thoughts in response to the probe (present) versus those who did not (absent), for both the HE and LE conditions. A non-parametric permutation analysis was applied where the r.AMYG–PCC connectivity difference between these two subgroups was compared to a null distribution built by permuting labels 5,000 times. As hypothesized, those 54% of participants reporting negative thought contents (versus 28% not reporting) showed increased r.AMYG–PCC connectivity during the rest periods after HE videos (observed difference = 0.08; P = 0.02, one-tailed). The same difference between the two subgroups during rest periods after LE videos showed only a trend (observed difference = 0.06; P = 0.07, one-tailed; Fig. 6c). These findings further unveil a direct relation between r.AMYG and PCC connectivity changes after negative emotions and individual reactivity to aversive or stressful socio-emotional stimuli.


Neural markers of emotional resilience and empathy in aging are increasingly recognized as important protective factors against mental illness and cognitive decline in this population48. Here we investigated both reactivity and recovery of brain networks to negative socio-affective situations (that is, during and after videos) in two independent experiments, including a large number of younger and older adults (N = 182).

Age-related behavioral characteristics

Older adults in experiments 1 and 2 did not differ in demographic or questionnaire data assessing affective or cognitive traits. However, they exhibited lower scores than younger adults in cognitive-related processes, including reflective rumination and cognitive empathy. This accords with previous work showing a decline of cognitive abilities in aging, including cognitive components of social functions2, while socio-affective abilities may remain stable. Indeed, measures of affect and empathy showed preserved patterns in older adults, extending previous findings in younger adults43. Seeing videos of others’ suffering induced higher levels of negative affect, lower positive affect and higher empathy scores than mundane scenes of daily life, regardless of age group.

Nonetheless, age differences were observed, with older adults reporting more positive emotions for both LE and HE videos. Moreover, the older the age, the lower the negative and the higher the positive emotions were rated when watching videos of suffering (Fig. 2b). This relationship between age and affect was not present for young participants. These results confirm a ‘positivity effect’ in older relative to younger adults4, which may reflect a motivation to upregulate positive and downregulate negative information4. In contrast, young and older adults reported similar levels of negative affect in response to videos. This suggests that the positivity effect in older adults does not impair their capacity to feel negative emotions when seeing others’ suffering. This underlines the importance of separately assessing negative and positive emotions, as done here and highlighted in previous research49

Finally, we found higher empathy correlated with increased negative affect during HE videos and with increased positive affect during LE videos, for both older and younger adults (Fig. 2c). However, positive emotions were reduced with higher empathy during HE videos only in older adults, suggesting that empathic effects on positive emotions in older people depend on context: the higher the empathy, the higher the positive emotions when facing social scenes without overt emotional content, but the lower the positive emotions when facing social scenes of others’ distress. These data offer a perspective on how empathy may impact the ‘positivity effect’ usually observed in older adults.

Neural markers of negative social emotions and empathy

Negative socio-affective videos engaged brain regions overlapping with networks previously associated with social cognition and emotion46,50. These encompassed aMCC and AI, both implicated in empathy for pain28,39, saliency detection51,52 and negative affect53, as well as TPJ and dMPFC, implicated in cognitive aspects of empathy and theory of mind28,40. In addition, in younger adults in experiment 1 and the larger sample of older adults in experiment 2, negative videos also activated the ventral striatum, an area associated with positive affect and reward54 and recruited during compassion for other’s suffering45,46.

Remarkably, despite its prominent role in emotional processing, there was no significant activation in the amygdala during the HE > LE videos in either group. This null result might reflect a broader role in encoding social or self-relevant information rather than just negative valence55, with the amygdala already activating to the content of LE videos and therefore showing similar increases during both video conditions in experiment 2 (Fig. 4h and Supplementary Fig. 2).

More critically, age affected brain responses to emotional scenes. Older adults activated less regions typically related to empathy (AI and PAG) and more those related to social cognition and emotion regulation (dMPFC, PCC and IFG). This accords with previous studies that examined age effects on empathy for pain35,36 and empathy for negative and positive emotions56,. Lower activity in affect-related regions, along with higher activity in cognition-related frontal regions might reflect better emotion regulation, possibly mediating the positivity effect of older people1. Increased activity in frontal regions may also reflect compensatory brain mechanisms acting to overcome cognitive deterioration in older adults57. In the current paradigm, these neural responses were not modulated by individual positive or negative affect scores, neither during videos nor during rest. Further research is needed to identify precise cognitive factors influencing emotional activation patterns in older adults, an issue beyond the purpose of the present study.

In sum, we find brain regions mediating empathy and Theory of Mind exhibit globally normal responses to negative social situations in healthy older adults. We also find positive affective biases in aging, seen in both behavioral and neural responses, indicating preserved socio-affective functions and empathy skills.

Emotional inertia after exposure to others’ suffering

Beyond transient responses to negative stimuli, regulating the impact of emotions over time is crucial to cope with stressful events58. Emotional inertia denotes a persistence of emotional states reflecting inefficient recovery and greater risk for psychological maladjustment12, but underlying neural substrates remain largely unexplored, especially in old populations. To uncover its neural underpinnings, we probed for emotional carryover effects in brain activity and assessed age-related differences. Across our two experiments, we observed selective increases during rest periods after HE (versus LE) videos. These comprised midline areas (ACC/MPFC and precuneus/PCC), involving core parts of the DMN typically active at rest59, together with increases in amygdala and insula, critically implicated in emotional processing29. Importantly, these effects occurred only in older adults, suggesting an important modulation of emotion regulation mechanisms during aging. The DMN mediates self-related internally oriented processes, including memory, interoception and value-based decision-making26. Previous studies reported divergent findings on DMN reactivity to emotion, including decreases17,18 or increases19,25, similar to our results. Interestingly, the duration of DMN activation to negative stimuli correlates with subjective emotional intensity, better than response magnitude19. Here (experiment 2), we found that two midline nodes of DMN (precuneus/PCC and dMPFC) not only activated during the HE (versus LE) videos, but also continued their activity during subsequent rest (post-HE > post-LE), providing direct evidence for ‘emotional inertia’ in the aging brain. These findings resonate with previous work showing that older, but not younger, adults fail to deactivate DMN regions during cognitive tasks33,34. To our knowledge, these data reveal for the first time that increased DMN activation in older adults may persist over time after exposure to negative socio-emotional contexts.

Sustained changes were also observed in limbic regions in experiment 2. The AI showed increased activity during both (HE > LE) videos and (post-HE > post-LE) rest periods, although voxel-wise patterns did not fully overlap: videos produced a more dorsal engagement during videos responses, while the more ventral AI was active after emotional events. Given findings in younger adults60 that dorsal AI may be recruited during adaptive behavior mechanisms, while ventral AI is recruited during internal homeostatic regulation, our result may reflect a shift from controlled/explicit adaptation to more spontaneous/implicit homeostatic regulation. On the other hand, the amygdala did not differentially respond during HE videos, but showed a lower return to baseline levels during post-HE (versus post-LE) rest. Accordingly, prolonged amygdala activity after negative images correlates with trait neuroticism61, and enhanced amygdala response after negative emotion elicitation is amplified in high-anxiety individuals15.

Altogether, our data highlight the temporal dynamics of brain responses to emotion and its relevance for individual affective styles and psychopathology17,19,22.

Brain connectivity patterns related to emotional inertia

Our functional connectivity analysis revealed that post-emotional carryover implicated different circuits, linking core parts of the DMN (PCC and MPFC) with limbic regions (amygdala and AI). These connections were selectively enhanced in post-HE versus post-LE rest, exclusively in older adults, and across both experiments (Fig. 5). Further, PCC–amygdala connectivity was not only stronger for post-HE than post-LE conditions, but also selectively stronger than other between-network connectivity patterns involving either the PCC or the amygdala (Fig. 6a). Interestingly, the strength of PCC–amygdala connectivity predicted individual anxiety and rumination in older adults (experiment 2). Moreover, explicit verbal reports revealed more frequent negative thought contents during rest periods following HE videos, and participants with more frequent negative thoughts also had higher PCC–amygdala connectivity than those who reported no negative thoughts. Such a relationship between negative thoughts and PCC–amygdala connectivity was significantly weaker during post-LE rest. These findings suggest that increased functional connectivity between PCC and amygdala may support the persistence of negative contents in spontaneous thoughts.

PCC is involved in internally directed cognition, rumination and memory26,62 especially when people retrieve affective autobiographical information63. As the amygdala also plays a central role in affective memory27,55,64, we speculate that PCC–amygdala communication may contribute to emotional inertia and recovery from negative social situations, possibly through associations with personal affective memories, and especially in older adults with higher levels of anxiety and rumination. These data unveil age-related effects on neural processes associated with rumination and repetitive negative thinking, mental states associated not only with maladaptive emotion regulation but also with increased risk of cognitive decline and Alzheimer’s disease9,10. As neurodegenerative anomalies in PCC and medial brain regions are common in Alzheimer’s disease65, changes in PCC connectivity might constitute a possible neural marker for deficient affective resilience, which is in turn associated with higher risk for dementia.

In parallel, increased functional connectivity was also observed between AI and aMPFC in post-HE versus post-LE rest in older adults from experiment 2. These changes showed no correlation with anxiety or rumination but only a weak positive correlation with the empathic concern IRI subscale (Supplementary Fig. 4). This may reflect a more general role of AI in emotional awareness66 and empathy28, and of aMPFC in the representation of affective states in both the self and others47,67. These results extend prior work by showing that connectivity between these two regions not only is modulated during the appraisal of socio-emotional stimuli but also persists beyond emotional events.

Limitations and future directions

Our study is not without limitations. First, participants watched videos passively, and subsequent carryover effects on brain activity and connectivity could reflect a lack of spontaneous implicit emotion regulation. Future studies should examine whether instructing participants with explicit emotion regulation strategies would change emotional inertia. Second, fMRI engendered technical constraints and affective ratings of videos were obtained only outside the scanner. Although this may bias the ratings, we deliberately made this choice to: (1) avoid top-down cognitive influences during scanning, which may confound neural activity during emotional perception;68 and (2) maximize older adults’ comfort by reducing the time spent inside the scanner. Third, parts of basal forebrain voxels were excluded from analyses in experiment 2 due to magnetic field inhomogeneities (Supplementary Fig. 5), as frequently induced in brain regions near air-filled cavities69. This prevented us from reliably studying regions such as the orbitofrontal cortex, despite their role in positive emotions and reward45,54.


We show that empathy for suffering and affective resilience can reliably be investigated in older adults using the SoVT–rest, a paradigm with very low cognitive load and high ecological validity. Our results reveal neural and behavioral markers of the positivity bias in older adults and unveil sustained carryover effects (or emotional inertia) in corticolimbic brain circuits in healthy older adults. PCC–amygdala functional connectivity at rest was selectively increased following negative events, and such an increase was related to anxiety, rumination and negative thought content. This resting connectivity pattern constitutes a plausible neural substrate for emotional inertia. These findings provide an important cornerstone for better understanding empathy and mechanisms underlying affective resilience in the aging brain, and thus contribute to identifying potential risk markers for neurodegenerative diseases associated with poor social stress coping.



For experiment 1, a total of 58 healthy participants including 30 younger adults (aged between 19 and 30 years), and 28 older adults (aged between 65 and 78 years) with corrected-to-normal vision and no history of neurological, psychiatric disorder or alexithymia took part. Thirty individuals were expected to participate in each group; however, new research guidelines during the coronavirus disease 2019 pandemic prevented us from continuing with scanning. Recruitment was performed through social media and advertisement in various locations within the University of Geneva. Three participants were excluded due to a priori exclusion criteria, including artifacts in brain images and/or extreme head motion during scanning. The final sample for experiment 1 included 29 young participants (mean age, 24 years; 14 females) and 26 older participants (mean age, 68.7 years; 13 females), resulting in a total of N = 55 participants (see Table 1 for detailed participant characteristics). All participants provided written informed consent. This study was approved by the local Swiss ethics committee (Commission Cantonale d’Ethique de la Recherche CCRE, Geneva) under project number 2018–01980.

For experiment 2, a total of 137 healthy older adults participated, community-dwelling, with corrected-to-normal vision and no history of neurological or psychiatric disorders, aged between 65 and 83 years. This session was part of the baseline visit of the Age-Well randomized controlled trial (RCT) within the Medit-Ageing Project70, conducted in Caen (France). Detailed inclusion criteria of the Age-Well RCT are provided in Supplementary Table 1. Participants were recruited via advertising in media outlets, social media and flyers distributed at relevant local events and locations. Two participants were excluded for eligibility criteria and intervention allocation issues71. A total of 8 participants were excluded from the final data analysis due to a priori exclusion criteria, including abnormal brain morphology (n = 3), extreme head motion (n = 3) and presence of artifacts in brain images (n = 2). The final sample for this study included 127 participants (mean age, 68.8 years, s.d. 3.63 years; 79 females; see Table 1 for other characteristics). All participants provided written informed consent before participation. The Age-Well RCT was approved by the ethics committee (Comité de Protection des Personnes Nord-Ouest III, Caen, France; trial registration number: EudraCT, 2016-002441-36; IDRCB, 2016-A01767-44; identifier: NCT02977819).


To account for interindividual differences in psycho-emotional profile, all participants from both experiments answered several questionnaires assessing psycho-affective traits and cognitive functions, including empathy (IRI72), depression (GDS73 for older adults and BDI74 for younger adults), anxiety (State-Trait Anxiety Inventory, STAI75), emotion regulation capacities (ERQ76) and rumination levels (RRS77). A summary of these questionnaires is provided in Table 1 and Fig. 1. All scores were in the normative range. For a full list of tasks and measures in the Age-Well trial (experiment 2), please refer to Poisnel and colleagues70.

Socio-affective video task–rest

The emotion-elicitation task used in both experiments was adapted from the previously validated SoVT43,45. The SoVT aims to assess social emotions (for example, empathy) in response to short silent videos (10–18 s). During this task, participants watch 12 HE and 12 LE video clips grouped in blocks of three (see instructions in Supplementary Table 2). HE videos depict people suffering (for example, due to injuries or natural disasters), while LE videos depict people during everyday activities (for example, walking or talking). In this study, each block was followed by a resting-state period of 90 s (see instructions in Extended Data Fig. 1 and Supplementary Table 2) to assess the carryover effects of emotion elicitation on subsequent resting-state brain activity (similar to Eryilmaz and colleagues17). This combination of both paradigms (task and rest) was specifically designed to test for emotional inertia and its relation to empathy. The combined task (SoVT–rest) is illustrated in Extended Data Fig. 1.

Overall, three sets (V1, V2 and V3) of 24 videos each were created and randomized across participants. In experiment 1, the video sets V1, V2 and V3 were seen by n = 21, 18 and 16 participants, respectively. In experiment 2, these were seen by n = 42, 40 and 45 participants, respectively. In both experiments, these videos were presented in two separate runs, always followed by a rest period. In experiment 2, each run was followed by a thought probe to assess current mental content during the last rest period (after LE videos in one run and after HE videos in the other run). The order in which runs were presented was randomized so that half of the participants started the experiment with a HE block and the other half with an LE block. No thought probe was given in experiment 1 (as it primarily aims at determining age-related brain activity patterns at rest). Brief instructions (in French) were presented before each period within each block. These indicated ‘The task is about to start’ before the first period of videos, or ‘The next videos are going to start’ before those in the following periods; while the display ‘Rest: wait for the next videos’ appeared before each rest period (Extended Data Fig. 1a,b). The total duration of the SoVT–rest fMRI paradigm was approximately 21 min, consisting of 9.5 min for each run, plus 1 min on average for the thought probes.

After the fMRI session, participants watched all video clips again on a computer outside the scanner and provided ratings on their subjective experience of empathy (‘To what degree did you feel the emotions of the characters?’) as well as their subjective positive affect (‘Indicate the intensity of your positive emotions’) and negative affect state (‘Indicate the intensity of your negative emotions’; translated from French), for each of the 24 videos. Each scale offered 21 possible responses ranging from 0 (‘not at all’) to 10 (‘extremely’) with increments of 0.5. The order of questions was always the same: empathy, positive affect and negative affect. We chose to obtain ratings after fMRI not only to minimize the time older adults spent in the scanner, but also to avoid potential cognitive effects during scanning that may confound neural activity during emotional perception and spontaneous rest recovery periods68,78. The total time for post-scanning ratings was, on average, 10 min. Onset times and response times for both neuroimaging and behavioral tasks were collected via the Cogent toolbox (developed by Cogent 2000 and Cogent Graphics) implemented in MATLAB 2012 (MathWorks).

Acquisition and preprocessing of magnetic resonance imaging data

Experiment 1

MRI scans were acquired at the Brain and Behavior Laboratory of the University of Geneva, using a 3T whole-body MRI scanner (Trio TIM, Siemens) with the 32-channel head coil. A high-resolution T1-weighted anatomical volume was first acquired using a magnetization-prepared rapid acquisition gradient echo (MPRAGE) sequence (repetition time, 1,900 ms; echo time, 2.27 ms; flip angle, 9°; slice thickness, 1 mm; field of view, 256 × 256 mm2; in-plane resolution, 1 × 1 mm2). Blood oxygen level-dependent (BOLD) images were acquired with a susceptibility-weighted EPI sequence (TR/TE, 2,000/30 ms; flip angle, 85°; voxel size, 3 × 3 mm; 35 slices, 3-mm-slice thickness, 20% slice gap; direction of acquisition, descending). Quality control and preprocessing were conducted using Statistical Parametric Mapping software (SPM12; Wellcome Trust Centre for Neuroimaging) on MATLAB 2017 (MathWorks). Before preprocessing, we manually centered all images to the AC-PC axis, aligned the functional and anatomical MRI images, and then realigned all images to the SPM anatomical template. Preprocessing included the following steps: (1) EPI data were realigned to the first volume and spatially smoothed with an 8-mm FWHM Gaussian kernel; (2) preprocessed fMRI data were denoised for secondary head motion and cerebrospinal fluid-related artifacts using automatic noise selection as implemented in ICA-AROMA, a method for distinguishing noise-related components based on ICA decomposition79. Additionally, components with high spatial overlap with white-matter regions were also removed by means of a linear regression using the fsl_regfilt function of FSL 6.0 (FMRIB’s Software Library;; (3) denoised EPI data were co-registered to the anatomical T1 volume; (4) the anatomical T1 volume was segmented and the extracted parameters were used to (5) normalize all EPI volumes into the Montreal Neurological Institute space. This procedure was performed using FSL and SPM12.

Experiment 2

MRI scans were acquired at the GIP Cyceron using a Philips Achieva 3T scanner with a 32-channel head coil. Participants were provided with earplugs to protect hearing, and their heads were stabilized with foam pads to minimize head motion. A high-resolution T1-weighted anatomical volume was first acquired using a three-dimensional fast field echo sequence (3D-T1-FFE sagittal; repetition time, 7.1 ms; echo time, 3.3 ms; flip angle, 6°; 180 slices with no gap; slice thickness, 1 mm; field of view, 256 × 256 mm2; in-plane resolution, 1 × 1 mm2). BOLD images were acquired during the SoVT–rest task with a T2*-weighted asymmetric spin-echo echo-planar sequence (each run ~10.5 min; TR, 2,000 ms; TE, 30 ms; flip angle, 85°; FOV, 240 × 240 mm2; matrix size, 80 × 68 × 33; voxel size, 3 × 3 × 3 mm3; slice gap, 0.6 mm) in the axial plane parallel to the anteroposterior commissure. During each functional run, about 310 contiguous axial images were acquired and the first two images were discarded because of saturation effects. Additionally, to improve the preprocessing and enhance the quality of the BOLD images80, T2 and T2* structural volumes were collected. Each functional and anatomical image was visually inspected to discard susceptibility artifacts and anatomical abnormalities.

Quality control and preprocessing were conducted using statistical parametric mapping software (SPM12; Wellcome Trust Centre for Neuroimaging) on MATLAB 2017 (MathWorks). Before preprocessing, we manually centered the images to the AC-PC axis, realigned the functional and anatomical MRI images and then realigned all images to the last version of the SPM anatomical template. The preprocessing procedure was done with SPM12 and followed a methodology designed to reduce geometric distortion effects induced by the magnetic field, described by Villain and colleagues80. This procedure included the following steps: (1) realignment of the EPI volumes to the first volume and creation of the mean EPI volume; (2) co-registration of the mean EPI volume and anatomical T1, T2 and T2* volumes; (3) warping of the mean EPI volume to match the anatomical T2* volume, and application of the deformation parameters to all the EPI volumes; (4) segmentation of the anatomical T1 volume; (5) normalization of all the EPIs, T1 and T2* volumes into the Montreal Neurological Institute space using the parameters obtained during the T1 segmentation; (6) 8-mm FWHM smoothing of the EPI volumes.

For each individual, frame-wise displacement (FD)81 was calculated. FD values greater than 0.5 mm were flagged to be temporally censored or ‘scrubbed’ during the first-level analysis (see description below). The average of FD volumes censored was 6.8 (s.d. 8.3, minimum of 1, maximum of 38) for both runs for a total of n = 65 participants. Three participants were excluded from further analysis because >10% of volumes showed FD > 0.5 mm within one run.

General linear model analysis

For both experiments, the MRI SoVT–rest data were analyzed using GLMs in SPM12 (implemented in MATLAB 2017). This comprised standard first-level analyses at the subject level, followed by random-effect (second-level) analyses to assess the effects of interest at the group level. For the first-level analysis, a design matrix consisting of two separate sessions was constructed for each participant. Experimental event regressors in each session included the fixation cross (10 s), instructions (8 s in experiment 1, 4 s in experiment 2), the three videos (~15 s each) modeled separately, and the rest periods following each block (90 s). Each rest period was divided into three equal parts (30 s time bins) to model different time intervals during which brain activity may gradually change after the end of the HE and LE video blocks (similar to Eryilmaz and colleagues17).

The different regressors were then convolved with a hemodynamic response function according to a block design for univariate regression analysis. To account for motion confounds, the six realignment parameters were added to the matrices, and low-frequency drifts were removed via a high-pass filter (cutoff frequency at 1/256 Hz). The final first-level matrix consisted of 2 sessions of 21 regressors each (1 fixation cross + 1 instruction for videos + 1 instructions for rest + 3 HE videos + 3 post-HE rest + 3 LE videos + 3 post-LE rest + 6 motion parameters). Additionally, we addressed the influence of remaining motion on BOLD data by performing data censoring as described by Power and colleagues81. Specifically, during the estimation of beta coefficients for each regressor of interest, volumes with FD > 0.5 mm were flagged in the design matrices and ignored during the estimation of the first levels.

For the second-level analyses, we used flexible factorial designs where the estimated parameters from first-level contrasts of interest were entered separately for each participant. The second-level design matrix was generated with SPM12 and included 12 regressors of interest (3 HE videos + 3 post-HE rest + 3 LE videos + 3 post-LE rest). This step allowed us to investigate the effect of each experimental condition on brain activity, including the main condition effects (video and rest), the specific emotional effects (HE and LE) during either the video or the subsequent rest periods as well as the age effect on the different conditions (young versus old; experiment 1).

Functional connectivity during rest periods

For both experiments, we conducted functional connectivity analyses between the most important brain ROIs associated with the empathy network and with the DMN. In addition, we also included the bilateral amygdalae among regions used for this analysis because previous studies assessing carryover effects in the brain have related sustained amygdala activity to anxiety traits15 and emotional reactivity61. For nodes of the DMN, we chose the PCC and the aMPFC, following Andrews-Hanna and colleagues59. Based on the results of a meta-analysis by Fan and colleagues39, the bilateral AI and aMCC were used as ROIs in the empathy network. Time series were extracted from 6-mm-radius spheres around the peak of each of these ROIs. The amygdala was defined anatomically using the current SPM anatomical template provided by Neuromorphometrics (

Functional connectivity analyses were performed using MATLAB 2017 and R studio (version 3.6.1). For each participant, time courses of activity (from each voxel of the brain) were high-pass filtered at 256 Hz, de-trended and standardized (z-score) before extracting specific time courses from the defined ROIs. In addition, white matter, cerebrospinal fluid signals, and realignment parameters were included as nuisance regressors in experiment 2. For each participant, time series from the instructions and videos periods were removed, and the remaining time series corresponding to the rest periods were concatenated. This procedure was previously proposed by Fair and colleagues82 and proved to be qualitatively and quantitatively very similar to continuous resting-state data. Additionally, to correct extreme head motion without affecting the autocorrelation of the time series, image volumes flagged with FD > 0.5 mm were removed and replaced by interpolation (every flagged volume X was replaced by the estimated mean of the X − 1 and X + 1 volumes). The final concatenated time series resulted in 184 frames (386 s) of resting-state data for each participant. We then correlated the time courses between the different ROIs using Pearson correlations, and the resulting coefficients were subjected to Fisher’s r-to-z transformation to improve normality in the data. Individual z-score maps (correlation matrices) were created for each participant (Extended Data Fig. 2a,b,c).

Thought probes

For each participant in experiment 2, two thought probes were recorded after the last rest period of each run and subsequently analyzed to test for differences in spontaneous mind wandering after emotional videos. Participants freely described their thoughts, and these narratives were digitally recorded and transcribed for analyses by two independent raters (Supplementary Table 4). For each probe (post-HE rest and post-LE rest), the two raters attributed the presence (present) or the absence (absent) of specific thought contents according to a diverse set of predefined categories (Supplementary Table 4). These categories were selected according to a priori relevant affective or cognitive dimensions, and included the following: negative and positive emotions, directed attention to oneself and to others, emotion regulation (voluntary control of emotions), negative and positive social emotions, rumination and temporality (present or past/future). Categories with low variability (that is, the same thought content reported by more than 85% of participants) were not included in further analyses because this prevented reliable regression analysis (Supplementary Table 4). The final dimensions included negative and positive emotions, directed attention to oneself and to others, and positive social emotions. This final analysis of thought probes comprised data from 109 participants for rest periods after HE videos and 110 participants for the rest periods after LE videos. This was due to (1) missing thought probes for 9 participants and (2) exclusion of reports not referring directly to thoughts or feelings in the rest period (but rather to factual details in the videos) for both runs (n = 5), following LE rest (n = 3) or following HE rest (n = 4). Inter-rater agreement on the final dimensions ranged from 0.28 to 0.66 (Cohen’s kappa index; Supplementary Table 4 for details). The statistical analyses were performed with R studio (version 3.6.1) and the corresponding graphs were created with ggplot2 (version 3.2.1).

Statistics and reproducibility

Statistical analyses of behavioral data

We performed a repeated-measures multivariate analysis of variance (with Pillai’s trace statistics) with the within-subject factor ‘video type’ (HE and LE), the between-subject factor ‘video set’ (V1, V2 and V3) and three dependent variables: ratings of empathy, positive affect and negative affect. This was followed up by pairwise t-tests. We also computed Spearman’s rank correlations between these different scores. Additionally, we performed correlation analyses between ratings of empathy, positive affect and negative affect of videos and age (as a continuous variable), using non-parametric Spearman’s rank correlations because some of these variables were not normally distributed. All statistical analyses are reported with a significance level of P < 0.05, and when necessary, P values were corrected for multiple comparisons using the FDR method83. Normality in the data was tested, and non-parametrical tests (for example, Spearman’s rank correlations) were performed when estimated necessary.

Statistical analyses of brain activity

In both experiments, we conducted t-tests contrasts to compare the conditions of interest (videos versus rest periods and vice versa) and the specific emotional effects (videos, HE versus LE; rest, HE versus LE). In experiment 1, we additionally tested for age differences in these effects (OAs versus YAs (videos, HE versus LE); OAs versus YAs (rest, HE versus LE)). In experiment 1, results are reported at uncorrected P < 0.001, k > 20 because this dataset concerned a smaller sample size and aimed at defining a comprehensive set of brain regions with emotion-related modulations in either videos or rest period in either age group. These regions could then be further probed with higher reliability and related to relevant individual characteristics in the larger dataset of experiment 2. Moreover, a more permissive combination of voxel-based and cluster-based thresholds has been shown to be adequate and reliable for experiments assessing cognitive and affective processes with unprecise onsets84, as in our task–rest paradigm. In addition, we had strong predictions concerning relevant regions (for example, DMN) based on previous works17,24,43. Finally, clusters surviving whole-brain family-wise error correction at P < 0.05 at the cluster level (FWEc) in experiment 1 are indicated in figures and tables (Supplementary Table 3a). In experiment 2, all comparisons are reported with a whole-brain FWE correction at P < 0.05, at the voxel level (Supplementary Table 3b).

Statistical analyses of brain functional connectivity

To test for significant differences between the two correlation matrices (post-HE rest and post-LE rest), we used a non-parametric permutation test85. For each pair of nodes, the permutation test compared the true correlation difference (for example, HE − LE) to a null distribution built by randomly flipping the sign of the correlation coefficients and computing the difference many times (n = 5,000; Extended Data Fig. 2d). More precisely, for each pair of nodes (for example, HE − LE for ROI 1 and ROI 3), a vector of values of n = number of participants was obtained and a one-sample t-test was computed to obtain the real t value (treal). Then, the signs of the elements in the vector were randomly flipped (n = 5,000) and the model was fitted repeatedly once for every flipping. For each fit, a new realization of the t-statistic was computed to construct an empirical distribution of t under the null hypothesis (tpermuted). From this null distribution, a P value was computed by assessing the probability of the treal to be higher than 95% of the values on the empirical tpermuted distribution85. Finally, the obtained P values were converted into an equivalent z-score and significant changes (marked by an asterisk in matrices) were retained for z > 1.64 (equivalent to P < 0.05, one-tailed, given observed increases without decreases in GLM analysis, uncorrected).

The final sample for experiment 1 included N = 55 participants. Three participants were excluded due to a priori exclusion criteria, including artifacts in brain images and/or extreme head motion during scanning. No statistical methods were used to predetermine sample sizes but our sample sizes are similar to those reported in previous publications17,35. The final sample for experiment 2 included N = 127 participants. Two participants were excluded for eligibility criteria and intervention allocation issues71. Eight participants were excluded from the final data analysis due to a priori exclusion criteria: abnormal brain morphology (n = 3), extreme head motion (n = 3) and presence of artifacts in brain images (n = 2). For the primary outcome of the Age-well RCT, an effect size of 0.75 per comparison was targeted, with 80% power and a two-sided type I error of 1.25% (Bonferroni correction for test multiplicity), resulting in a total of 126 participants needing to be included. The final number of participants in the Age-Well RCT (n = 137) was higher than the required minimum of 126 participants70,71. Sensitivity analyses then indicated that the final sample included in experiment 2 (n = 127) was reliable and sensitive enough to detect small- to medium-sized effects at a voxel level (α = 0.001, effect size of dz = 0.37), given a power of 1 − ß = 0.8 (ref. 86). For both experiments, the video sets (V1, V2 and V3) and whether participants started the experiment with HE or LE videos first were randomly attributed to participants in an Excel table. While investigators were blinded to the training arm that participants were randomized to in the Age-Well trial, no blinding was performed for the SoVT–rest task at baseline (experiment 2). The statistical analyses were performed with R studio (version 3.6.1) and MATLAB 2017 (MathWorks), and the corresponding graphs were created with ggplot2 (version 3.2.1).

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.