Detecting age differences in prosociality using a newly developed picture-based measure

Abstract

Previous research has indicated that older adults are more likely than younger adults to share money, suggesting more prosociality. However, the extent to which older adults exhibit prosociality in other forms, such as helping and comforting, remains unclear due to the absence of a comprehensive measure. This study addresses this gap by validating a novel picture-based measure of prosociality (PB-Prosocial) in older adults, which showed good psychometric properties. Using this measure, we examined age-related differences in prosociality and underlying mediators. Older adults were found to be more prosocial in sharing, but not in helping or comforting, compared with younger adults. The age difference in sharing appeared to result from older adults being more familiar with the situations and perceiving a lower cost of sharing than younger adults. Taken together, our study provides a validated measure of prosociality and demonstrates that older adults’ increased prosociality is specific to sharing.

Introduction

Prosociality describes voluntary behaviors intended to benefit others¹. According to a comprehensive model of prosociality, interpersonal prosocial behavior can be subdivided into the categories of helping, sharing, and comforting, depending on the needs of the recipients². Helping intends to meet instrumental needs, sharing intends to meet material needs, and comforting intends to meet emotional needs. Prosocial behavior is associated with profound benefits in both mental and physical health. Specifically, engaging in prosocial behavior has been associated with reduced symptoms of depression and anxiety^3,4, lower blood pressure and chronic disease risk⁵, thereby potentially extending lifespan. Therefore, it is important to investigate prosocial behavior, especially in older adults^6,7. Beyond these general health benefits, prosociality plays a unique role in older adulthood due to age-related changes in social roles and networks. For example, older adults often experience transitions such as retirement or bereavement, which can alter their social responsibilities and reduce their social circles⁸. Engaging in prosocial acts can help maintain social connections and foster a sense of meaning. According to the socioemotional selectivity theory, older adults tend to prioritize emotionally meaningful goals and relationships, making prosocial behavior particularly consequential in this stage of life⁹.

A growing body of research has focused on prosociality in older adults and found that older adults tend to act more prosocially than younger adults^10–13. For instance, when asked to donate money for a good cause (viz., Doctors Without Borders), older adults donated more money than younger adults¹⁴. According to Erikson’s eight stages of psychosocial development¹⁵, when people reach their old age, they develop a sense of generativity and a desire to contribute to the well-being of future generations, so they are more likely to engage in prosocial behavior to pass on their knowledge and experience. However, increased prosociality in older adults is not consistent across all situations. For instance, when asked to assist with an additional task after completing an initial experiment, older adults were as likely to help as younger adults^16,17. A recent large-scale study of 202,898 individuals across 22 countries found that helping behavior declines with age¹⁸. Similarly, Chi et al.’s study¹⁹ found negative associations between providing emotional support (comforting) with aging. These inconsistent findings may arise because different studies focused on different types of prosociality. According to a recent meta-analysis, older adults were only more prosocial in sharing, but not helping or comforting, than younger adults²⁰.

Researchers have proposed various explanations for the age differences observed in prosociality. According to the value-based decision-making framework²¹, engaging in prosocial behavior involves a decision-making process where individuals evaluate the costs and benefits of performing the behavior. Importantly, different types of prosociality are characterized by distinct cost structures. Sharing typically involves monetary cost, helping often requires time and physical effort, and comforting needs emotional engagement. Aging is associated with changes in the availability and salience of these resources. Older adults often possess greater financial stability compared with younger adults, which may reduce the perceived cost of monetary sharing²². In contrast, helping and comforting may rely more on time, physical energy, and emotional resources, which may be perceived as more limited in later life²³. Thus, age group differences may be associated with distinct perceived cost structures across prosocial types. This type-specific cost account provides a potential explanation for prior meta-analytic findings that showed age-related increases in sharing but not in helping or comforting.

While perceived cost is central to prosocial decision-making, prosocial choices are also shaped by how individuals interpret social situations. Situational familiarity is conceptualized here as a theoretically distinct construct from perceived cost, and may shape prosociality through social affective pathways. Familiar situations are typically processed more fluently, evoke clearer expectations, and reduce uncertainty, whereas unfamiliar situations may require greater cognitive effort to interpret²⁴. Importantly, familiarity is proposed to influence prosociality by facilitating autobiographical memory retrieval, perspective taking, and emotional resonance with others’ needs. Empirical evidence supports the role of situational familiarity in prosocial behavior. According to a recent study, situational similarity significantly influenced helping behavior, particularly in older adults²⁴. People are generally more inclined to help others when they perceive the situation as familiar. Given that older adults typically possess more extensive life experience than younger adults^25,26, they are likely to encounter more situations they recognize. Such situations may activate autobiographical memories that strengthen empathic understanding and motivation to assist, and may facilitate perspective taking and emotional concern for others. However, it should be noted that while previous study provided insights into the influence of situational familiarity on helping behavior, their focus was limited to this type of prosociality. The effects of familiarity on comforting and sharing remain underexplored. For example, while helping involves direct assistance, comforting may require a recognition of emotional needs, and sharing may require considerations of fairness and resource allocation². These types of prosociality could be influenced by familiarity in ways that differ from helping. Therefore, there remains a gap in the literature regarding the relationships between situational familiarity and different types of prosociality.

Another factor that needs to be taken into consideration when investigating prosociality is the social desirability bias, which refers to the tendency of individuals to respond in a socially acceptable manner, rather than expressing their true thoughts or behaviors²⁷. Since individuals may overreport their prosocial intentions to present themselves as kind and generous, this bias may contaminate the assessment of prosociality²⁸. In the current study, it is important to control for this bias given that older adults have been found to exhibit more social desirability bias compared to younger adults^29,30. Therefore, we included social desirability as a covariate in the statistical analyses of the age differences in prosociality.

Existing measures of prosociality vary widely in format and focus. Many studies rely on hypothetical text-based vignettes that require participants to imagine social scenarios^31,32, potentially introducing variability due to language comprehension. Other paradigms focus on a single type, such as monetary donations³³, which only capture sharing, or one-shot naturalistic helping tasks³⁴, which may lack test-retest reliability. These approaches make it difficult to directly compare different types of prosociality within a unified methodological framework. The picture-based measure of prosociality (PB-Prosocial)³⁵ was developed to address these limitations. By presenting naturalistic photos, this measure reduces reliance on linguistic understanding and increases ecological validity. It also comprehensively assesses helping, sharing, and comforting using parallel structures, allowing direct type comparisons within the same paradigm. This measure has demonstrated good psychometric properties in younger adults aged 18 to 35 years³⁵.

To summarize, previous studies have presented mixed findings regarding age differences in prosociality and proposed different mechanisms by which to explain these age differences. Therefore, the current study aimed to compare various types of prosociality between older and younger adults and investigate the potential mediators underlying age group differences in prosociality. To facilitate age group comparisons, we first validated the PB-Prosocialin older adults. We expected that this measure would demonstrate comparable properties in older adults aged over 60 years. After validation, we asked older adults and younger adults to complete an online survey, measuring their prosocial intentions, familiarity with different situations, the perceived cost of prosocial behavior, social desirability, and demographic factors. As a recent meta-analysis indicated that there was significant age difference in sharing, but not helping or comforting²⁰, we expected that older adults would be more prosocial in sharing, but not in helping or comforting. We expected that perceived cost and situational familiarity would mediate the relationship between age and sharing. Because prior meta-analytic findings do not indicate reliable age differences in helping or comforting, mediation was not hypothesized for these types.

Methods

Participants

Two independent samples of older adults were recruited online. Data collection was facilitated by an international data collection firm (Kantar), which maintains an established participant panel in Hong Kong. Each panel member is assigned a unique identification number and can complete a given survey only once, thereby preventing duplicate participation. Participants from Hong Kong were recruited to complete the survey online via Qualtrics, and Kantar incentivized participation through reward points. This study was performed in accordance with the Declaration of Helsinki. The study protocol was approved by The Hong Kong Polytechnic University’s Institutional Review Board (approval number HSEARS20230726005). Informed consent was obtained from all participants prior to their participation in the survey.

Sample 1 Sample 1 included 333 older adults aged 60 to 80 years, which was recruited to first validate the PB-Prosocial. Sample 1 was then compared to another dataset of 307 younger adults (aged 18–35 years) from a previous study³⁵ to examine the age differences in prosociality and the potential mediators. The data of 307 younger adults had been analyzed to establish the psychometric properties of the PB-Prosocial, while in the current study, the data were analyzed in terms of age differences. This sample size enabled a power of 0.80 for a close fit of the proposed mediation model³⁶. The data collection was done from August to September 2023.

Sample 2 Sample 2 included 253 older adults aged 60 to 80 years, which was recruited for confirmatory factor analysis (CFA) of the PB-Prosocial only to further validate this measure. This sample size enabled a power of 0.80 for a close fit of the three-factor model with six items loading on each factor³⁶. The data collection was done from December 2023 to January 2024.

Measures

PB-prosocial

The PB-Prosocial³⁵ includes 6 items for each type of prosociality (viz., helping, sharing, comforting). In each item, a photograph showing a person in need is presented and paired with a question asking about the likelihood of helping (see Fig. 1 for examples). The photographs depict specific needs: instrumental needs (e.g., reaching an object) for helping, material needs (e.g., lacking money) for sharing, and emotional needs (e.g., experiencing a breakup) for comforting. The photographs were either sourced from existing image datasets for socioemotional research³⁷ or taken specifically for this measure. The protagonists depicted in the photographs were all East Asian and included children, young adults, and older adults. However, the distribution of protagonist age was not fully balanced across prosocial types. The photographs were taken in various settings, such as schools, supermarkets, roadsides, and workplaces. Participants rate on a 7-point Likert scale about their likelihood to help (1 = definitely will not help, 7 = definitely will help). The 18 items of PB-Prosocial were presented in a randomized order. The PB-Prosocial has a Cronbach’s α of 0.93 and a test-retest reliability of 0.82 in younger adults. Convergent validity of this measure was demonstrated by significant correlations with altruism, empathic concern, perspective taking, and uncaring traits. Exploratory factor analysis (EFA) and CFA supported a three-factor structure in younger adults.

Fig. 1.

Illustrative examples of PB-Prosocial. Note. The images shown are illustrative examples and not the actual stimuli used in the PB-Prosocial. All faces shown in this figure were generated using artificial intelligence and do not correspond to real individuals.

Hong Kong altruism index (A-index)

The A-Index includes 10 items³⁸. Participants indicated whether or not they have exhibited particular altruistic behaviors within the past month, within the past year, or in a lifetime. Altruism and prosociality both include behaviors aimed at benefiting others, which makes the A-Index an appropriate measure for establishing the convergent validity of the PB-Prosocial.

Inventory of callous-unemotional traits (ICU)

The Chinese ICU includes 11 items³⁹ and can be subdivided into two dimensions: callousness and uncaring. Higher callousness and uncaring traits are associated with less prosocial behavior according to a meta-analysis⁴⁰. Participants rated on a 4-point Likert scale ranging from 0 (Not at all) to 3 (Definitely true).

Interpersonal reactivity index (IRI)

The 28-item Chinese version of IRI⁴¹ was used to measure four dimensions of empathy, i.e., empathic concern, personal distress, perspective taking, and fantasy. Results of previous studies have indicated positive correlations between empathic concern as well as perspective taking and prosociality⁴², thus these two subscales were selected to examine the convergent validity of the PB-Prosocial. Participants rated on a 4-point Likert scale ranging from 0 (does not describe me well) to 4 (describes me very well).

Familiarity

Participants rated their familiarity with each situation on a 7-point Likert scale ranging from 1 (extremely not familiar/have not experienced at all) to 7 (extremely familiar/have experienced exactly). Familiarity was conceptualized at the prosocial type level rather than at the level of individual situations. Familiarity with each type of prosociality was calculated by averaging the six items within each type. Aggregating across multiple items within the same type reduces random measurement error and provides a more reliable estimate of type-level familiarity than any single item alone⁴³.

Cost

Participants rated their perceived cost of each behavior on a 7-point Likert scale ranging from 1 (no cost at all) to 7 (extremely high cost). For each situation, participants were asked: “How much would it cost you to help the person in the picture?” The term “cost” was intentionally left broad to allow participants to consider the type of cost relevant to the specific situation (e.g., time, effort, emotion, or money). Perceived cost of each type of prosociality was calculated by averaging the six items within each type.

Marlowe-crowne social desirability scale (MCSD)

The 6-item Chinese version of MCSD⁴⁴ was used to measure social desirability bias. Participants indicated whether the statements were applicable to themselves or not and a total score was calculated.

Demographic variables

The following demographic information were collected: age, gender, education level, employment status, and religious belief, and socioeconomic status (SES). Specifically, subjective SES was assessed by using the MacArthur’s ladder⁴⁵.

Procedure

Sample 1 completed all the aforementioned measures in the first test. A random subsample of Sample 1 was invited to complete a retest of the PB-Prosocial three weeks later to assess the test-retest reliability and 48 older adults completed the retest. A three-week interval is a commonly used interval to balance between minimizing recall bias and ensuring the construct remain stable⁴⁶. Sample 2 completed the PB-Prosocial and the demographic questions. To ensure data quality, a question asking participants to count the number of items in the picture was added at the start of the survey, and those who failed this question were screened out. In addition, there were two attention check items embedded in the survey, and participants who failed any one of the attention check questions were removed from subsequent analyses.

Data analysis

Sample 1

Data cleaning: In the first test, 58 older adults were excluded due to failing attention check, and 15 older adults were removed because of multivariate outliers. Thus, the final dataset for EFA, convergent validity, and reliability analyses included 260 older adults. In the retest, after excluding five older adults who failed attention check, data from 43 older adults were used in the test-retest reliability analysis. In the analysis of age differences in prosociality, data of 333 older adults (i.e., Sample 1) and 307 younger adults were combined for data cleaning. Participants who failed attention check were removed (58 older adults and 54 younger adults). Of the excluded participants, 79 failed one attention check (41 younger adults and 38 older adults) and 33 failed both (13 younger adults and 20 older adults). Within the younger adult sample, participants who failed attention checks (n = 54) did not differ from those retained (n = 253) in SES or social desirability. Within the older adult sample, those who failed (n = 58) did not differ in SES but reported lower social desirability than those retained (n = 275). Social desirability was controlled in the analyses of age differences in prosociality. Univariate outliers were defined as a z score above + 3 or below − 3⁴⁷ and were then removed (six older adults and 11 younger adults). After removing eleven multivariate outliers (four older adults and seven younger adults), data from 500 participants (265 older adults and 235 younger adults) were used in the analyses of age differences in prosociality and mediation effects.

Validity and reliability: EFA was conducted to establish the factor structure of the PB-Prosocial. A principal component analysis with oblique rotation (Promax) was used. The Kaiser-Meyer-Olkin (KMO) and Bartlett’s test of sphericity were used to test the fitness of the data. The inclusion criteria of the factors are eigenvalue greater than 1. Convergent validity was calculated using Pearson correlation coefficient with A-Index, empathic concern, perspective taking, and ICU. Internal consistency was calculated using Cronbach’s α. The test-retest reliability was calculated using the Pearson correlation coefficient between the first test and the retest.

Age differences: To examine the age differences in various types of prosociality, a 2 (Age: younger vs. older) × 3 (Type: helping vs. sharing vs. comforting) ANCOVA was conducted, using social desirability as a covariate. Given that SES is a known correlate of monetary sharing⁴⁸, it was also included as a covariate to control for its potential confounding effects. To examine the age differences in the other measures, i.e., familiarity, cost, and social desirability, independent-sample t- tests were conducted.

Correlational analysis: Correlations among variables were computed using Pearson’s correlation coefficients. Holm’s procedure⁴⁹ was used to correct for multiple correlations and the adjusted value of alpha was 0.003.

Mediation analysis: To examine potential mediators between age groups and prosociality, a bootstrap analysis was conducted in which bias-corrected confidence intervals (CI) were constructed by drawing 5,000 random samples from the full sample with replacement. If the CI does not include zero, the indirect effect is considered statistically significant, thus supporting the mediation effect. Familiarity and cost were tested as mediators.

Sample 2

Following the exclusion of 25 older adults who failed attention check, data from 228 older adults were used for CFA. A bifactor model with one global factor (i.e., prosociality) and three specific factors (helping, sharing, comforting) with each factor comprising six items was examined using maximum likelihood estimation. The model fit was considered to be acceptable if fit indices met the following criteria: Comparative Fit Index (CFI) above 0.9, Tucker Lewis Index (TLI) above 0.9, and Root mean square error of approximation (RMSEA) below 0.08⁵⁰.

Results

Descriptive statistics

Table 1 summarizes the characteristics of the older adults from Sample 1 and younger adults. The gender ratio was balanced between older and younger adults, with 45.5% of the younger adults and 50.6% of the older adults being male. The mean age of the older adults was 66.4 while the mean age of the younger adults was 29.6. The mean age of Sample 2 was 67.1 and 46.5% of Sample 2 were male.

Table 1.

Sample characteristics.

Variables	Younger adults	Older adults
	M (SD) or N (%)	M (SD) or N (%)
Age	29.6 (4.3)	66.4 (5.5)
Male	107 (45.5%)	134 (50.6%)
Having diploma level of education or above	199 (84.7%)	145 (54.7%)
In employment	204 (86.8%)	139 (52.5%)
Having no religious belief	156 (66.4%)	181 (68.3%)
Socioeconomic status	5.02 (1.86)	5.17 (1.94)
Helping	5.60 (0.88)	5.72 (0.84)
Sharing	3.96 (1.30)	4.31 (1.13)
Comforting	5.13 (0.92)	5.04 (0.93)
Helping familiarity	4.64 (0.96)	4.86 (0.95)
Sharing familiarity	3.84 (1.11)	4.08 (1.02)
Comforting familiarity	3.81 (1.22)	3.72 (1.23)
Helping cost	2.44 (1.11)	2.27 (1.03)
Sharing cost	3.80 (1.11)	3.52 (1.05)
Comforting cost	3.05 (1.08)	2.83 (1.14)
Social desirability	0.52 (0.23)	0.62 (0.25)

Validity and reliability

EFA indicated a clear three-factor structure of the PB-Prosocial in Sample 1 (KMO = 0.93, χ² = 2961.45, p < .001), which accounted for 66.15% of the total variance. Fit indices of CFA indicated the model derived from EFA fit the data of Sample 2 (CFI = 0.929, TLI = 0.911, RMSEA = 0.078). Thus, the three-factor structure was confirmed in older adults.

In Sample 1, the three subscales of PB-Prosocial were significantly correlated with A-Index (Helping: r = .36, Sharing: r = .50, Comforting: r = .41, ps < 0.001), empathic concern (Helping: r = .47, Sharing: r = .47, Comforting: r = .57, ps < 0.001), perspective taking (Helping: r = .44, Sharing: r = .44, Comforting: r = .52, ps < 0.001), and ICU (Helping: r = .42, Sharing: r = .27, Comforting: r = .40, ps < 0.001). Given a floor effect observed in the callousness, where most participants reported very low levels, the correlations between the uncaring traits and the PB-Prosocial were also examined. All the three subscales were negatively correlated with uncaring traits (Helping: r = .46, Sharing: r = .40, Comforting: r = .52, ps < 0.001). These results supported the convergent validity of the PB-Prosocial.

Cronbach’s α for helping, sharing, and comforting was 0.86, 0.88, 0.92 respectively. The test-retest reliability coefficients for helping, sharing, and comforting were 0.82, 0.75, 0.72, respectively. To summarize, the PB-Prosocial is a valid and reliable measure of prosociality in older adults.

Age differences

The 2 × 3 ANCOVA yielded a nonsignificant main effect of group, F(1, 497) = 0.410, p = .522, η_p² < 0.001, a significant main effect of type, F(1.812, 900.726) = 119.598, p < .001, η_p² = 0.194, a significant interaction between group and type, F (1.812, 900.726) = 9.346, p < .001, η_p² = 0.018, and a significant effect of social desirability as a covariate, F(1, 497) = 23.439, p < .001, η_p² = 0.045. Simple main effect comparisons revealed that compared to younger adults, older adults showed more sharing, p = .028, Cohen’s d = 0.240, similar helping, p = .485, and similar comforting, p = .076 (Fig. 2). The age difference in sharing remained significant after controlling for SES, p = .047, Cohen’s d = 0.206. Simple main effect comparisons of the within-subject effect revealed that older adults showed more helping than comforting (p < .001, Cohen’s d = 0.677) and sharing (p < .001, Cohen’s d = 1.433), more comforting than sharing (p < .001, Cohen’s d = 0.756). Younger adults also showed more helping than comforting (p < .001, Cohen’s d = 0.474) and sharing (p < .001, Cohen’s d = 1.619), more comforting than sharing (p < .001, Cohen’s d = 1.145).

Fig. 2.

Age differences in prosociality.

Compared to younger adults, older adults were more familiar with the helping scenarios, t(498) = 2.590, p = .010, Cohen’s d = 0.232, more familiar with the sharing scenarios, t(498) = 2.510, p = .012, Cohen’s d = 0.225, similarly familiar with the comforting scenarios, t(498) = −0.834, p = .405. Compared to younger adults, older adults perceived a similar cost of helping, t(498) = −1.800, p = .072, less cost of sharing scenarios, t(498) = −2.891, p = .004, Cohen’s d = −0.259, less cost of comforting, t(498) = −2.118, p = .035, Cohen’s d = −0.190. Compared to younger adults, older adults had more social desirability, t(498) = 4.950, p < .001, Cohen’s d = 0.444 (Table 1).

Correlational analysis

Correlation coefficients among variables are presented in Table 2. Familiarity was positively correlated with helping r(498) = 0.50, p < .001, sharing r(498) = 0.60, p < .001, and comforting r(498) = 0.38, p < .001. Cost was negatively correlated with helping r(498) = 0.38, p < .001, sharing r(498) = 0.43, p < .001, and comforting r(498) = 0.22, p < .001.

Table 2.

Correlations among variables.

	1	2		3	4	5	6	7	8	9	10	11
1. Age groupa		0.22**		0.08	0.14**	− 0.05	0.12**	0.11*	− 0.04	− 0.08	− 0.13**	− 0.09*
2. Social desirability				0.21**	0.22**	0.12*	0.10*	0.08	0.01	− 0.23**	− 0.23**	− 0.22**
3. Helping					0.55**	0.58**	0.50**	0.35**	0.22**	− 0.38**	− 0.30**	− 0.28**
4. Sharing						0.49**	0.53**	0.60**	0.48**	− 0.11*	− 0.43**	− 0.14**
5. Comforting							0.35**	0.35**	0.38**	− 0.12**	− 0.20**	− 0.22**
6. Helping familiarity								0.73**	0.63**	− 0.17**	− 0.25**	− 0.09*
7. Sharing familiarity									0.76**	0.03	− 0.24**	− 0.03
8. Comforting familiarity										0.11*	− 0.15**	0.08
9. Helping cost											0.48**	0.69**
10. Sharing cost												0.55**
11. Comforting cost

^aYounger adults = 1, older adults = 2.

*p < .05, **p < .01.

Mediation analysis

The pathway diagram of the mediation analyses is shown in Fig. 3. The direct effect from age group to sharing was not significant, β = 0.045, SE = 0.035, 95%CI [−0.025, 0.114], p = .198. The indirect effect from age group to sharing via familiarity with sharing was significant, β = 0.142, SE = 0.057, 95%CI [0.033, 0.251], p = .012. Age group positively predicted familiarity, β = 0.112, SE = 0.096, 95%CI [0.025, 0.200], p = .012, which in turn positively predicted sharing, β = 0.519, SE = 0.046, 95%CI [0.445, 0.586], p < .001. The indirect effect from age group to sharing via cost of sharing was significant, β = 0.094, SE = 0.035, 95%CI [0.029, 0.155], p = .008. Age group negatively predicted cost, β = −0.128, SE = 0.095, 95%CI [−0.213, −0.043], p = .003, which in turn negatively predicted sharing, β = −0.298, SE = 0.045, 95%CI [−0.379, −0.216], p < .001.

Fig. 3.

Mediating pathways.

Discussion

In the current study, we first validated the PB-Prosocial in older adults and found evidence to support its reliability and validity. Then, we used the validated PB-Prosocial to systematically compare different types of prosociality across a substantial sample of older adults and younger adults and tested the underlying mechanisms. The results revealed that older adults were more inclined to share than younger adults, while no significant age differences were observed in helping or comforting. This specific propensity for sharing among older adults can be attributed to a greater familiarity with the situations and a lower perceived cost of sharing.

Apart from replicating the finding that older adults are more prosocial in sharing than younger adults, we further clarified the potential mechanisms underlying this age difference. On the one hand, the more sharing observed in older adults might be explained by their greater familiarity with the sharing scenarios, e.g., poor people in earthquake areas. According to previous studies, older adults experience more critical life events as they grow older, and prior experience with difficult situations increases the accuracy of estimating others’ needs and empathy for others’ difficulties^25,51,52. Fraire et al. also provided evidence that individuals who had experienced more lifetime adversities were more likely to donate to charities and engage in volunteering⁵³. Similarly, Gaesser et al. also demonstrated that familiar contexts facilitated a greater willingness to help others when imagining helping⁵⁴. With more situational familiarity, more details may be imputed to others’ sufferings, thus increasing the willingness to help⁵⁵. On the other hand, the more sharing might also be explained by the fact that older adults perceived a lower cost of sharing, which could be related to them possessing more resources⁵⁶. To the best of our knowledge, this study is the first to find that perceived cost mediates the relationship between age-related differences and prosocial sharing. The considerations of cost can be broad and may include monetary cost, negative emotions, physical efforts, etc.^19,57. It should be noted that prior studies have operationalized cost in different ways. In the current study, perceived cost referred to the personal resources required to perform prosocial behavior (e.g., time, effort). In contrast, in a study⁵⁸ focusing on the prosocial COVID-19 behavior (e.g., keeping social distancing), cost was operationalized as the potential negative consequences of not engaging in the behavior, such as the risk of infection. Under such circumstances, more cost was associated with more prosocial behavior. Thus, the direction of the association between cost and prosociality may depend on whether cost reflects the personal sacrifice required to act or the potential consequences of not acting. Although the cost-benefit analysis has long been proposed as a framework to understand prosocial behavior⁵⁹, our study is one of the few that examined the association between cost and prosocial sharing.

Age differences were not significant in helping or comforting, which aligns with previous research^16,60,61. The familiarity results showed that older adults were less familiar with the situations that required comforting. Consequently, they may not know how to offer comfort, or they may perceive the situations differently (i.e., not believing that comforting is necessary in a particular circumstance). The similar levels of comforting exhibited by older adults could also be attributed to their emotion regulation strategies. Previous research suggests that older adults are more likely to take a passive approach or just wait to see if things improve when they encounter emotional problems compared to younger adults⁶². According to the strength and vulnerability integration model⁶³, older adults are more likely to avoid exposure to emotional distress because they have greater difficulties in returning to emotional homeostasis. It is noteworthy that emotional support is closely related to greater life satisfaction and less depression in older adults^64,65. Particularly, providing emotional support contributes more to well-being than receiving it⁶⁶, thus it is important to promote comforting behavior in older adults for their well-being.

We also replicated the finding that older adults exhibit a more socially desirable manner^29,30,67, possibly due to a stronger inclination to adhere to social norms. In the current study, we controlled for the potential influence of social desirability bias on prosociality by including it as a covariate. The age difference in sharing remained significant after accounting for social desirability bias, indicating that other factors may explain the higher propensity for sharing in older adults. The age difference in helping and comforting was not significant, regardless of whether social desirability was controlled for or not. Social desirability bias may influence helping, sharing, and comforting to different extents. Helping is often a strongly emphasized social norm and may therefore be more sensitive to impression-management concerns⁶⁸. In contrast, sharing might be less influenced by social desirability bias as the monetary cost of sharing is more apparent and easily quantified than the effort cost of helping and sharing, which may further diminish the influence of social desirability bias on sharing. This interpretation should be viewed cautiously, as the PB-Prosocial scenarios do not explicitly manipulate public versus private contexts. Our results underscore the importance of controlling for social desirability bias in studies of prosociality.

Several limitations of the current study should be noted. First, all the measures were based on self-report, which may be influenced by response bias, such as extreme and acquiescence biases. To mitigate this, future studies should incorporate objective measures of prosociality, e.g., actual monetary donations, to provide a more ecologically valid measure of prosociality. Second, although we sought to examine the cost associated with prosocial behavior, we did not differentiate between various types of costs, such as financial, effort, time, and emotional. Older adults may perceive financial support as less costly because they have accumulated more assets, but perceive emotional support as more costly because they have limited emotional resources⁶³. Future studies should explore cost in more detail by developing more accurate operationalizations of cost perception. Third, middle-aged adults were not included in the current study, so it remains unclear whether the changes in prosociality are gradual and continuous. Future research should include middle-aged adults to provide a more comprehensive understanding of how prosociality evolves throughout adulthood. Fourth, our study was cross-sectional, so causal inferences regarding age-related differences in prosociality cannot be made. Longitudinal studies should be conducted to investigate the developmental trajectory of prosociality. Fifth, the older adults who participated in this online study may represent a relatively healthy and technologically proficient subgroup, and thus may not be fully representative of the broader older population. In addition, the older sample ranged from 60 to 80 years (Mean = 66.4), primarily reflecting younger-old adults. Individuals over 80 were not included. These factors may limit the generalizability of the findings, and future research should replicate the results using more diverse samples and a broader age range in later life. Sixth, although children, young adults, and older adults were represented across prosocial types, their proportions were not strictly balanced. It is possible that recipient age may interact with participant age in shaping prosocial decisions. Future studies should systematically manipulate and balance recipient age to examine potential age-based in-group or out-group effects.

Conclusions

By adopting a newly developed and validated comprehensive measure of prosociality (PB-Prosocial), our study demonstrated that older adults are more prosocial only in sharing, a tendency that may result from a lower perceived cost of sharing and greater situational familiarity. These findings provide a valuable resource for assessing and promoting prosociality, which contributes to a more inclusive and supportive society. Future research should continue to investigate the developmental process of prosociality and develop strategies to promote it.

Author contributions

All authors designed the study. D.L. analyzed the data, interpreted the data and wrote the original manuscript. Y.C., B.P.H.H., and D.H.K.S. provided helpful comments on the manuscript. All authors reviewed the manuscript.

Funding

David H. K. Shum is supported by the Yeung Tsang Wing Yee and Tsang Wing Hing Endowed Professorship. Duo Li is supported by the Research Postgraduate Scholarship of The Hong Kong Polytechnic University.

Data availability

All data are available at https://osf.io/8enx4/.

Declarations

Competing interests

The authors declare no competing interests.