Introduction

Imagine that you are sitting in a waiting room and strike up a conversation with the person next to you. During the course of the conversation, you end up sharing a painful memory. What determines how your new friend might respond? Decades of research suggest that empathy influences how people respond to others’ pain. Yet, empathy is a multifaceted phenomenon, one that unfolds across loosely coupled domains of cognitive, affective, and motivational changes1,2,3. When witnessing someone in pain, an empathic response may be to have warm feelings of compassion (Empathic Concern), try to understand the suffering individual’s point of view (Perspective Taking), or experience vicarious suffering (Personal Distress). Empathic responses to another’s suffering, in turn, are generally considered to motivate prosocial actions4,5. To make general claims about the social benefits of empathy, however, a more detailed understanding is needed as to how different subcomponents of empathy may uniquely contribute to prosocial outcomes.

Various components of empathy, including Empathic Concern, Perspective Taking, and Personal Distress, tend to be distinct from one another in terms of their neurological bases6,7,8,9 and behavioral consequences10,11,12,13. The divergent outcomes of these types of empathic responses are often discussed in the context of helping. Notably, Empathic Concern that involves feelings of compassion and care for suffering others is thought to be foundational to prosociality7,14. Empathic Concern has been associated with helping behaviors such as making altruistic moves in economic games15,16 and offering help to strangers encountered in-person17, online11,18, and via pre-recorded messages19.

To increase empathy, researchers sometimes use a perspective taking paradigm that involves asking participants to take viewpoints of a suffering individual. Such perspective taking manipulations improve attitudes toward stigmatized and suffering outgroup members20 and increase help in the form of monetary donations21. However, overall results on the direct relationship between perspective taking and helping tend to be inconsistent19. One challenge relates to the fact that Perspective Taking is highly correlated with Empathic Concern in most people across cultures22, and therefore it is difficult to tease out whether changes in prosocial outcomes are solely attributable to Perspective Taking independent of Empathic Concern. Furthermore, imagining another’s perspectives may not always result in accurate understanding of the target’s plight23. For example, having experienced similar life events of another, which can increase the ease of perspective taking24, has been ironically associated with less accurate understanding of the target’s experiences25 and less helpfulness26. In contrast with Empathic Concern and Perspective Taking, where the focus remains on the target in pain, Personal Distress is self-centered and focuses on the empathizer’s own aversive internal states that mirror the suffering target’s states27,28. Some research suggests that feeling distressed upon witnessing another’s suffering may motivate helping behaviors that would in turn reduce one’s own vicarious suffering5,29,30,31. Other research suggests that the affective mirroring in Personal Distress is neither a necessary nor sufficient catalyst for prosociality32,33,34. Personal Distress can also be stressful for the experiencer and thus unsustainable27, making it a precarious foundation for prosocial behaviors. For example, feeling distressed in response to witnessing human suffering may make people feel paralyzed, leading to inaction or fleeing from a helping situation28. Consistent with this view, individuals who tended to experience Personal Distress more were less likely to offer help when witnessing cyberbullying18 and in economic games11.

Why might there be such divergence in terms of prosocial outcomes across different components of empathy? Compared to the rich body of theoretical discussions29,34,35, less empirical work exists that examines the mechanisms linking different facets of empathy to helping outcomes. Here, we explore one possible account that focuses on affective responses to others’ suffering. Individuals who vary across different empathic tendencies may experience different feelings in response to another’s suffering. For example, people who tend to feel Empathic Concern, either directly in response to witnessing a suffering person or as a result of taking the perspective of the target individual, might experience negative feelings such as worry. At the same time, they may also experience warm, tender feelings of compassion. In a study that used compassion training designed to grow Empathic Concern by adopting the viewpoints of suffering individuals, researchers found increases in positive, but no changes in negative affect, among those who completed the compassion training36, suggesting that the other-directed positive affect might be a defining feature of compassion and Empathic Concern. By contrast, Personal Distress is described as an aversive state of anxiety that mirrors the negative emotions of the target individual in pain.

Differences in affective experiences, in turn, may lead to different helping outcomes. The relationship between positive affect and helping is well established37; feeling good increases the likelihood of contributing to charity38, helping coworkers39, and helping strangers in need40,41,42. If positive affect promotes helping, then when Empathic Concern and Perspective Taking result in positive affect, this may relate to more helping in daily life. By contrast, evidence is mixed about the relationship between negative affect and helping43. Although negative emotions like guilt44,45,46 can increase helpfulness, they may also have the opposite47 or no effect48. In the context of Personal Distress, the tendency to feel negative in response to another’s suffering may lead to egoistic help that reduces one’s own pain5,29,30,31, or it may relate to avoidance12.

The current study examined affective responses to naturalistic narratives and their relationships with different components of empathy and helping. Participants self-reported their tendencies to experience Empathic Concern, Perspective Taking, and Personal Distress, as well as the amount of help they provided to others in the past month. Building upon the theoretical and empirical evidence supporting the role of empathy in affective experiences to others’ suffering36 and altruism5,29, as well as the influence of affect in helping outcomes37, we explored whether individuals who vary across these dimensions of empathy would report different affective responses to someone else’s suffering expressed in emotional real-life stories, and whether these affective responses might then relate to their tendency to help others. We tested these ideas in the context of real-life emotional displays to complement findings from previous studies that relied on economic game paradigms and hypothetical scenarios.

Methods

Data, code, and protocol availability

Data, protocol, and analysis scripts are available at https://github.com/cnlab/empathy/. The current study reports a subset of data from a larger parent study, and the methodological details on participant recruitment and data collection have also been described in a study that examined the accuracy of interpersonal communication25 that is not the focus of the current report. The current study’s design and analyses were not pre-registered.

Participants

Participants were recruited via online advertisement and flyers for a study about storytelling and listening (N = 77; Mage = 21.16 years, SDage = 1.91; 71 White, 1 Asian, 1 Hispanic, 2 Mixed, 2 Other). Self-identifying heterosexual women were invited based on previous studies on sex and gender differences in empathy49 in dyadic interactions50 to avoid difference in responses to a story narrated by a woman speaker as potential confounds12. Other eligibility criteria included 18–25 years of age (to match the speaker’s age group), native English speaking, no serious medical/psychiatric history, and no current/recent use of controlled drugs or psychotropic medications. Additional eligibility criteria unrelated to the current report, as part of a parent study that investigated compassion training effects on neural response, included standard functional near infrared spectroscopy (fNIRS) neuroimaging eligibility (i.e., hair conditions favorable for the use of fNIRS and right-handedness) and no prior experience with compassion training. We note that some methodologies have been proposed to promote inclusion of more diverse samples in neuroimaging studies51,52 and discuss the limited representativeness of our sample due to eligibility constraints in the Discussion section. Eligible participants were identified via self-reports in response to an online prescreen survey.

Participants with usable data were included in analyses that tested relationships between empathic tendencies and helping outcomes (N = 77) and affect while listening to a speaker’s story (N = 72; 5 did not complete the visit). The sample size was determined by power analysis for the preregistered parent study, but the sensitivity analysis using G*Power53 suggested that the available sample size of N = 77 would allow us to detect a medium effect size of f2 = 0.15 with 85% power (alpha = 0.05, two tailed uncorrected) for the main regression analyses. This study was approved by the University of Pennsylvania Institutional Review Board. All experiments were performed in accordance with the Declaration of Helsinki and with relevant guidelines/regulations. All participants provided informed consent and were paid for their participation.

Procedure

Participants visited the laboratory and completed a 2–3 h study protocol (Fig. 1). Participants completed surveys that assessed their usual levels of empathic tendencies, the amount of help they provided to others in the past month as a measure of helping tendency, and demographic information. Participants then watched a pre-recorded video of a speaker who shared emotional real-life events. Next, participants reported how negative or positive they felt while listening to the story. All measures were embedded among other tasks and surveys, and participants completed some of the tasks while their brain data were collected using fNIRS (brain data are not reported here).

Fig. 1
Fig. 1
Full size image

An overview of the experimental procedure.

Measures

Empathy

Components of empathy most relevant to social outcomes were measured using the Interpersonal Reactivity Index1. The IRI consists of four subscales, including Personal Distress, Empathic Concern, Perspective Taking, and Fantasy. The current report focuses on the first three subscales given their relevance to prosociality11,54, and results using the fantasy subscale are included in the Supplementary Information 1 [SI1]. The Empathic Concern subscale measures emotional components of empathy that involve having compassionate and caring feelings for suffering others (e.g., “I often have tender, concerned feelings for people less fortunate than me”). The Perspective Taking subscale measures the cognitive aspect of empathy that allows individuals to infer mental states of others by imagining their viewpoints (e.g., “I sometimes try to understand my friends better by imagining how things look from their perspective.”). The Personal Distress subscale measures the tendency to feel anxious and distressed in response to others’ suffering (e.g., “When I see someone who badly needs help in an emergency, I go to pieces”). Scores were rated on a 1 = Does not describe me well to 5 = Describes me very well scale and coded such that higher values indicate higher empathy. Internal consistency for Empathic Concern (α = 0.85), Perspective Taking (α = 0.85), and Personal Distress (α = 0.80) items in the current study were high.

Helping others

To measure participants’ tendencies to help others, the amount of help participants provided to others in the past month was assessed using a four-item Social Support questionnaire taken from the Wisconsin Longitudinal Study55. Questions assessed different types of help given to someone other than one’s spouse or young child in domains of instrumental and emotional help. The answer options included Yes or No, scored to receive 1 for each Yes and 0 for each No. Scores were summed across items such that higher values indicate greater amounts of help given to others.

Affective response to others’ suffering

A video task was used to measure individual differences in the tendency to experience positive or negative affect in response to others’ suffering. Participants first watched a ~ 5 min video of a person sharing an emotional real-life story, followed by providing verbal support to the speaker in response. The story included some positive elements (e.g., beginning of a romantic relationship, vacation) but was overall negative and concluded with a distressing event (e.g., stressful work environments, an unexpected breakup, feelings of confusion and sadness). Please see [SI2] for more details of how the seed story video was created, and additional data indicating within-individual consistency in how people respond to different types of emotional narratives from suffering others. Next, participants answered questions about how they felt while listening to the story. A modified version of the Perceived Understanding Instrument56 was used to assess 11 negative (nervous, anxious, annoyed, uncomfortable, dissatisfied, stressed, insecure, sad, a sense of failure, incomplete, uninterested) and 15 positive (connected, understanding, supportive, compassionate, loving, trusting, interested, satisfied, relaxed, pleasant, good, accepting, comfortable, happy, important) affect items. Scores were rated on a 1 = Not at all to 5 = Very much scale and were averaged to indicate mean levels of negative and positive affect experienced in response to another’s suffering. Internal consistency for the negative (α = 0.86) and positive (α = 0.88) affect items in the current study were high.

In addition, baseline mood was assessed using the modified Differential Emotion Scale (mDES)57 before participants heard about the emotional narratives to compare how different components of empathy may relate to more general moods as opposed to affective responses specific to human suffering [SI3].

Demographics

Participants self-reported their age and race/ethnicity. All participants self-identified as heterosexual women. The majority (71 out of 77; 92%) of the sample identified as White individuals due to the eligibility criteria that called for light hair conditions favorable for fNIRS scanning. Please see the Discussion section for limitations related to sample characteristics and generalizability of our findings. The race/ethnicity variable did not co-vary with any of the primary outcomes (p > 0.30), but participants’ age was positively associated with one of the primary outcome measures (positive affect in response to another’s suffering; b = 0.097, p = 0.042). Therefore, all analyses controlled for age as a covariate.

Analysis plan

We conducted a series of multiple regression analyses to explore the relationships among different components of empathy (Empathic Concern, Perspective Taking, and Personal Distress), affective (negative and positive) responses to others’ suffering, and help-giving. We first tested whether different components of empathy were uniquely related to people’s tendency to help others, operationalized as the amount of help given to others in the past month. We then explored how each component of empathy was associated with negative and positive affective responses to another’s suffering expressed in emotional narratives. These analyses revealed positive associations between (1) Empathic Concern and positive affect in response to the story, as well as (2) Perspective Taking and positive affect in response to the story. In addition, higher Personal Distress was associated with (3) lower positive affect and (4) higher negative affect in response to the story. To further understand the relationships between the components of empathy, people’s emotional responses, and their tendency to help others, we conducted four indirect effect analyses. The first and second indirect path analyses tested the idea that individuals with greater tendencies to experience Empathic Concern, or adopt others’ perspectives (Perspective Taking), may experience more constructive positive affect when responding to others’ suffering, which might then relate to a greater tendency to help others in need. These models included Empathic Concern or Perspective Taking, in separate models, and positive affect as predictors of self-reported amount of help provided. The third and fourth indirect effect analyses explored whether individuals who tended to experience Personal Distress in response to others’ suffering reported less positive affect and more negative affect, which might then relate to lower tendencies to help others. These models included Personal Distress as well as positive affect or negative affect, in separate models, as predictors of self-reported amount of help provided. These indirect effect analyses estimated the indirect effects of empathic tendencies on helping through affective response using the product-of-coefficients approach70. A non-parametric bootstrapping with 5000 iterations was performed to estimate bias-corrected and accelerated (BCa) confidence intervals58 using the boot.ci function of the boot package in R59. All analyses controlled for age by including it as potential covariates in the model. Analyses did not control for the other facets of empathy beyond the one being tested. Results remained consistent using data winsorized to + /− 3 standard deviations from the means [SI4]. All significant results survived false discovery rate correction [SI5]. The current study used a subset of data from a larger parent study that included a randomization to a compassion or control condition; we did not find any significant interaction between our primary predictors and the intervention condition (ps > 0.06). Unstandardized beta coefficients (b) and 95% confidence intervals (CI) are reported. All reported p-values are two-tailed. Analyses were performed in R (v4.4.2, www.r-project.org) using the R-studio interface (v2024.4.2). Please see https://github.com/cnlab/empathy/ for data and analysis scripts.

Results

Relationships among subcomponents of empathy

Empathic Concern and Perspective Taking were significantly correlated with each other (r = 0.605, 95%CI [0.441, 0.730], p < 0.001), consistent with previous findings22. Personal Distress did not correlate with Empathic Concern (r = -0.067, 95%CI [−0.287, 0.159], p = 0.562) or Perspective Taking (r = -0.107, 95%CI [−0.324, 0.120], p = 0.353), also consistent with prior discussions7,34,60 and data6,11 that distinguished them as separate constructs.

Empathy and helping

We examined the relationships between different components of empathy and people’s tendency to help others, operationalized as the amount of help participants reported providing to others in the past month. Empathic Concern (b = 0.325, 95%CI [0.024, 0.626], p = 0.034) and Perspective Taking (b = 0.325, 95%CI [0.010, 0.639], p = 0.043) were directly associated with helping, such that individuals who reported experiencing compassionate feelings toward others’ suffering and the tendency to take their viewpoints more also reported greater amounts of help given to others in the past month. We did not find any significant direct association between Personal Distress and help-giving (b = −0.267, 95%CI [−0.580, 0.047], p = 0.094).

Empathy and affective responses to others’ emotions

Next, we examined how each component of empathy related to affective experiences in response to another’s suffering. Individuals who tended to feel Empathic Concern or take others’ perspectives (Perspective Taking) felt more positive, while those prone to experiencing Personal Distress felt more negative while listening to someone’s emotional narratives (Fig. 2): Higher Empathic Concern (b = 0.454, 95%CI [0.216, 0.693], p < 0.001) and Perspective Taking (b = 0.409, 95%CI [0.159, 0.659], p = 0.002) scores were associated with more positive affect. By contrast, empathic tendencies to experience higher Personal Distress was associated with feeling less positive (b = −0.275, 95%CI [−0.542, −0.007], p = 0.044) and more negative (b = 0.344, 95%CI [0.065, 0.624], p = 0.016) in response to hearing about another’s suffering. Empathic Concern (b = -0.138, 95%CI [−0.413, 0.138], p = 0.323) and Perspective Taking (b = -0.202, 95%CI [−0.482, 0.079], p = 0.156) were not associated with negative affect experienced in response to another’s suffering.

Fig. 2
Fig. 2
Full size image

Subcomponents of empathy and affective responses to another’s suffering. Participants retrospectively reported how positively or negatively they felt while listening to a speaker sharing emotional real-life stories. While listening to another’s emotional narratives, empathic tendencies to experience higher (A) empathic concern and (B) perspective taking were associated with feeling more positive. (C) Personal Distress was associated with feeling less positive, (D) empathic concern and (E) perspective taking were not associated with negative affect, and (F) higher personal distress was associated with feeling more negative.

Indirect effects of empathy on helping through affect

From the main analyses above, we observed positive associations between Empathic Concern and Perspective Taking and positive affect, a negative association between Personal Distress and positive affect, and a positive association between Personal Distress and negative affect in response to emotional narratives. To further explore the role of affective responses to another’s suffering in helping, we conducted four separate indirect effect analyses using a bootstrapped product-of-coefficients approach (Table 1). We found a significant indirect effect of Empathic Concern on helping via positive affect (a*b = 0.140, BCa 95%CI [0.001, 0.312]). The indirect path from Perspective Taking to helping via positive affect was also significant (a*b = 0.128, BCa 95%CI [0.009, 0.316]). In other words, individuals who reported a greater tendency to feel Empathic Concern and to take others’ perspectives were also more likely to feel positive while responding to someone else’s suffering, which, in turn, was related to their higher tendency to help others. The indirect paths from Personal Distress to helping via positive affect (a*b = -0.096, BCa 95%CI [−0.282, 0.005]) or negative affect were not significant (a*b = 0.045, BCa 95%CI [−0.040, 0.234]). The coefficients and statistics for all models are reported in Table 1.

Table 1 Indirect path analyses.
Full size table

Indirect effects of empathic tendencies (Empathic Concern, Perspective Taking, and Personal Distress) on tendencies to help via affective response to another’s suffering were tested. Notes: Models were bootstrapped with 5000 iterations to estimate bias-corrected and accelerated (BCa) confidence intervals and included age as a covariate. CI confidence interval, SE standard error.

Discussion

Empathy is multifaceted. Results from our study highlight that different components of empathy—Empathic Concern, Perspective Taking, and Personal Distress—are distinct processes that have unique relationships to how individuals process others’ suffering and their orientation towards helping others.

We found that Empathic Concern—feeling compassionate care and concern for suffering others—and Perspective Taking—taking viewpoints of suffering individuals—were associated with a greater tendency to help others in daily life. These results are consistent with previous theoretical claims7,14,27 and empirical evidence15,16,17 that consider the compassionate state of Empathic Concern a reliable and sustainable foundation for prosociality. Furthermore, the cognitive ability for Perspective Taking, previously theorized to elicit Empathic Concern20,21, may covary with individual drives to help others in need. By contrast, Personal Distress was not directly associated with helping, suggesting that the effect of Personal Distress61 on prosociality may likely be weak and/or context dependent35.

We found that Personal Distress in response to another’s suffering was associated with feeling negative while hearing about another’s suffering. Given that the overall sentiment of the story our participants responded to was negative, this likely represents the affective mirroring in which individuals experienced negative affect in response to the speaker’s negative emotional displays. By contrast, individuals with more Empathic Concern felt more positive, but not more or less negative, while listening to the same story. This adds to the argument that compassion does not necessarily involve affective mirroring7,34,60 or increases in negative affect36. Although a compassionate person may still experience worried concern for suffering others, positive feelings of love and connection seem to be the focal point in the experience of Empathic Concern.

A perhaps obvious and yet important point of clarification is that the types of other-focused positive affect commonly associated with Empathic Concern (e.g., warm tender feelings for suffering others) are distinct from schadenfreude where one rejoices in others’ suffering62. Rather, positive emotions experienced in Empathic Concern or compassion are characterized by self-transcendent drives to benefit others beyond the self8. Self-transcendent states are commonly experienced to be positive and intrinsically rewarding63,64. Interestingly, Empathic Concern was not associated with baseline positive mood in our data [SI3], suggesting that the positive other-directed affect experienced among individuals with high Empathic Concern may not be a function of being generally in a good mood; rather, it may be specific to situations that require compassionate responses to suffering others.

Perspective Taking—a cognitive aspect of empathy—was also associated with feeling more positive in response to listening to another person’s emotional life story. This is consistent with prior research focused on compassion training. Taking the perspective of others in terms of their needs and desires is one of the main goals of compassion training65, and as would be expected, compassion training increased self-reported amounts of perspective taking66 and engaged neural activity associated with mental state inference8,65. Together with these previous intervention efforts, our result further supports that targeting cognitive aspects of empathy may be an effective strategy to promote positive feelings relevant to prosociality.

Although empathic tendencies to experience Personal Distress was associated with elevated negative affect in response to another’s suffering, higher negative affect did not predict helping. This is in line with previous studies that showed inconsistent relationships between negative affect and helping outcomes43. By comparison, positive affect has been more reliably associated with helping outcomes37. We found that those who tended to experience positive affect in response to another’s suffering also tended to help others more. However, affect is unlikely to be the only path to helping; rather, researchers argue that Empathic Concern or compassion are multifaceted experiences that involve multiple—attentional, motivational, affective, and behavioral—components14 that may independently and/or interactively promote helping. Furthermore, taking the perspective of another often leads to other motivational and behavioral changes that relate to prosocial outcomes21. The complexities of Empathic Concern and Perspective Taking warrant future work that identify other mechanisms beyond affective changes reported here that may contribute to various types of prosocial outcomes. Indeed, previous studies examined the relationships among different kinds of empathy and broader landscape of emotions relevant to prosociality, including indices of aggression61,67. Various emotional styles beyond traditional measures of empathy, such as callous and unemotional traits, were strongly associated with aggression67,68, underscoring the importance of considering empathy in a broader emotional context.

Some limitations are worth noting. Mainly, this study entailed stringent eligibility criteria that included hair conditions favorable for the use of fNIRS as part of a larger study that tested questions unrelated to the current report. This resulted in an extremely unrepresentative sample, with the majority of the participants being self-identified White individuals. Although our findings are firmly grounded in and build upon decades of data and theories of empathy and compassion, many of these previous studies also suffer from the same issue of overreliance on White populations69. Therefore, it is unclear whether the observed relationships are generalizable to a broader population or specific to the characteristics of the current sample. Future research that includes more diverse and accurately representative samples are critically warranted. Another limitation relates to our use of self-reports to measure empathy, which may reflect, implicitly or explicitly, the desire to be seen as such. Researchers also have raised concerns about the possibility that the IRI scale used in this study may measure traits other than empathy11, such as more general dispositions toward being agreeable or anxious rather than specific responses to human suffering. Counter to this view, however, we did not find that baseline differences in affect were associated with helping [SI3]; rather it was people’s affective responses to hearing another person’s painful story that predicted tendencies to engage in helping. Future studies may take advantage of increasingly diverse and multimethod measurement tools14 to complement self-reports. We also note that of different types of helping-giving70, the Social Support questionnaire55 we used focused on only specific attributes of social support (instrumental and emotional). We also acknowledge that our cross-sectional data cannot rule out the possibility of reverse directionality between variables. For example, rather than empathic concern prompting a helping response, it is possible that the act of helping others makes people grow empathic concerns for others. However, theories of empathy4,5,29 and positive affect37 argue that these variables predict helping outcomes. Experimental data are needed to infer causality of the shown relationships. Finally, the current sample size was relatively small for testing indirect effects, an issue shared with many other psychological studies71. Future studies with fully powered samples are warranted to test the robustness of these relationships.

Together, the current findings support that empathy is a multidimensional experience—a phenomenon that can vary across different individuals and entail distinct affective and behavioral consequences. Our findings respond to the growing call for establishing more precise relationships among different components of empathy and the distinct processes with which they contribute to experiencing others’ suffering and orientations towards helping others. Notably, our results highlight the central role of affective responses to another’s suffering as one pathway to helping outcomes: namely, individuals who tend to feel compassionate concern for another and take others’ perspectives, but not those who tend to feel distressed in response to others’ suffering, may derive more positive feelings by helping, which may explain their proclivity to help others. These findings echo the growing call for broadening the construct of empathy to consider affective traits relevant to prosocial (e.g., helping) and antisocial (e.g., aggression) outcomes61,67,68. Future intervention and education programs may focus on selectively growing the facets of empathy and emotional traits that are helpful for, while discouraging others that may harm, healthy orientations for prosociality.

Positionality statement

Mindful that our identities can influence our approach to science, the authors wish to provide the reader with information about our backgrounds. With respect to gender, when the manuscript was drafted, three authors self-identified as women and one as a man. With respect to race/ethnicity, two authors self-identified as Asian, one as White, and one as Hispanic. None of the authors are current college students (i.e., the main population sampled).

Citation diversity statement

Recent work in several fields has identified a bias in citation practices such that papers from women and other minority scholars are under-cited relative to the number of such papers in the field72,73. Here we sought to consider choosing references that reflect the diversity of the field in thought, form of contribution, gender, and other factors. We obtained the predicted gender of the first and last author of each reference by using databases that store the probability of a first name being carried by a woman74,75. By this measure, our references contain 28.83% woman(first)/woman(last), 16.48% man/woman, 12.75% woman/man, and 41.95% man/man. This method is limited in that (a) names, pronouns, and social media profiles used to construct the databases may not, in every case, be indicative of gender identity and (b) it cannot account for intersex, non-binary, or transgender people.