Sage Journals: Discover world-class research

Abstract

Perceived social isolation (PSI) predicts adverse health and behaviour problems better than objective social isolation (OSI). PSI is related to higher sensitivity to negative emotional states and emotional dysregulation. Some studies suggest that the more socially isolated people feel, the more sensitive they are to the suffering of others, yet the less empathic they are. However, no study has tested the association between social isolation and empathy while distinguishing their state and trait forms. This study examined how individuals with high- and low-PSI trait perceived and empathized with virtual agents displaying facial expressions. Sixty-six healthy young adults (38 women) were equally divided into high- or low-trait PSI subgroups. All participants reported their current state of PSI and then watched short videos of virtual agents showing facial expressions of physical pain or sadness. For each video, participants rated the perceived intensity of affective states and empathy felt (state empathy). They also completed questionnaires on trait empathy, OSI, moods, and depressive symptomatology. The results showed no effect of group (high- or low-trait PSI), facial expression (pain, sadness), or their interaction on state empathy and intensity mean ratings. However, regression analyses revealed that OSI specifically predicted perspective-taking and empathic concern aspects of trait empathy, whereas trait PSI predicted the personal distress aspect. These findings suggest that social skills may play a role in maintaining and developing relationships but are not linked to one’s perception of their quality. This work highlights the importance of differentiating objective from perceived social isolation, given their distinct associations with specific social processes and skills, such as empathy. This distinction could further help develop more targeted interventions to alleviate social isolation.

Keywords

perceived social isolation loneliness objective social isolation empathy facial expressions

In 2024, one in three American adults reported suffering from social isolation at least once a week over the past year (American Psychiatric Association, 2024). Social isolation poses serious health threats, as it is associated with heightened stress response (Zilioli & Jiang, 2021), higher rates of depression (Mushtaq et al., 2014), poorer sleep quality (McLay et al., 2021), and even increased mortality (Holt-Lunstad et al., 2015). Social isolation is thus a major public issue that warrants appropriate attention in research.

Social isolation can be either objective or subjective (Gierveld et al., 2018). Objective social isolation (OSI) is a lack of social interactions between a person and their social network. It differs from perceived social isolation (PSI; or loneliness), an aversive psychological experience prompted by the subjective perception of being socially isolated. The experience of PSI stems from the perceived gap between the desired and the actual quality and quantity of a person’s interpersonal relationships. Therefore, one person might not feel lonely, even though they have few social contacts and are, objectively speaking, socially isolated. Inversely, another person may have many social interactions and not be isolated, yet still feel alone, separated, or apart from others and suffer from a perceived lack of social connectedness. PSI can be further described as situational (i.e., state PSI), such as when it is experienced as an aversive acute feeling after a breakup or a relocation (Cacioppo & Patrick, 2008). However, it is most typically defined in studies as a relatively stable trait (i.e., trait PSI) when this feeling occurs over an extended period rather than following a transient situation (Mund et al., 2020).

Trait PSI and OSI are moderately associated (Taylor et al., 2023), yet they have distinct adverse impacts (Cacioppo & Cacioppo, 2018; Gierveld et al., 2018). Trait PSI more strongly predicts health and behavioural changes than OSI (Cacioppo et al., 2011; Rico-Uribe et al., 2016), and PSI (as assessed by one question) is the item of a neuroticism questionnaire most strongly associated with premature mortality, including deaths from intentional self-harm (Deason et al., 2025). Trait PSI is also associated (if not confused) with negative psychological experiences that are closely related but conceptually distinct, such as social exclusion (Cacioppo & Patrick, 2008) and depression. Some researchers have even questioned whether trait PSI and depression are truly distinct (Mushtaq et al., 2014), given their moderate correlation (Russell et al., 1980) and that trait PSI predicts depression levels in longitudinal studies (Cacioppo et al., 2010). Nonetheless, these two constructs remain statistically distinguishable (Cacioppo et al., 2006; Russell et al., 1980).

Despite its negative consequences, short-term PSI is considered an adaptive response that helps individuals, according to cognitive and evolutionary theories, detect social isolation and its associated survival threat, and restore social bonds (Cacioppo & Cacioppo, 2018). A large body of data shows that loneliness increases sensitivity to negative social cues, which contrast with the sparsity of support for the more context-dependent positive cues (for a summary of theoretical approaches on loneliness and social information processing, see Blandl & Eisenberger, 2025). Trait PSI has also been linked to sensitivity toward negatively experienced affective states, including physical (i.e., pain; Wolf & Davis, 2014; Yamada et al., 2021) and emotional (e.g., sadness; Yang et al., 2023) suffering. An individual with high trait or state PSI tends to experience physical pain more frequently and intensely (Allen et al., 2020; Wolf & Davis, 2014; Yamada et al., 2021). PSI thus seems to be frequently associated with self-related negativity.

However, less is known about the relationship between PSI and perceived negativity in others. In other words, it remains unclear whether individuals who feel isolated are generally negative or whether this negativity is specific to their own feelings. Some studies have examined the influence of PSI on emotion perception and recognition in others, which are low-level social cognition processes (Eikelboom et al., 2025). The results indicate that trait PSI is associated with a heightened sensitivity to negative affective states in others, either through a greater likelihood of misidentifying affective facial expressions as painful or angry (Smith et al., 2022), or through potentially lower response time to identify sadness and fear depicted on faces (Vanhalst et al., 2017). Taken together, these findings suggest a possible relationship between trait PSI and greater sensitivity to negative affective states, both felt and perceived in others.

Trait PSI may also affect higher-level social cognition processes, such as empathy. Empathy refers to the ability to understand, feel, and share another person’s experiences, while maintaining a clear distinction between oneself and others (Håkansson Eklund & Summer Meranius, 2021). It can be divided into three interrelated components (Decety & Jackson, 2004). The cognitive component allows inferring a mental or affective state (e.g., intention, belief, emotion). The affective component enables sharing the target’s emotional state. The emotional regulation component distinguishes between what is experienced by the self and what is experienced by the other. This component also helps to balance the affective and cognitive components so that the target’s emotion does not cause distress to the observer. Distinguishing between these components is essential, as one may be more affected than others in certain situations or within specific clinical populations. For instance, individuals with autistic spectrum disorder mainly experience difficulties with the cognitive component of empathy (Fatima & Babu, 2024). Much like PSI, empathy can also be described as situational (i.e., state empathy) or dispositional (trait empathy; Nezlek et al., 2001). State empathy is the feeling of empathy that varies according to context and person (e.g., empathy felt during a behavioural task). In contrast, trait empathy refers to a stable characteristic that persists over time. According to Davis (1983), trait empathy can be divided into distinct aspects linked to specific components of empathy. For instance, the perspective-taking aspect (i.e., the tendency to spontaneously adopt the point of view of others) relates to the cognitive component, while the empathic concern aspect (i.e., “other-oriented” feelings of sympathy and concern for unfortunate others) aligns with the affective component (Wang et al., 2020).

Some correlational studies have shown that trait empathy is significantly and negatively associated with trait PSI (e.g., Beadle et al., 2012; Davis, 1983). People with low empathy skills may struggle to accurately identify the emotions and thoughts of others. Such difficulties may lead them to judge their social interactions negatively, thereby contributing to their feelings of PSI. However, a recent study revealed a negative but nonsignificant association between these concepts after controlling for confounding factors, including OSI (Mwilambwe-Tshilobo et al., 2023). Taking into account OSI and other confounding factors, like state PSI and depressive symptomatology, is thus essential when examining PSI and its relationship with empathy.

To our knowledge, only the study of Hu et al. (2020) has yet explicitly explored PSI and its influence on empathy in a situational setting. State PSI was experimentally manipulated using a recall task during which participants wrote about a time when they felt lonely and isolated. A control group of participants wrote about their daily routine instead. Participants in whom feelings of PSI had been elicited were less likely to empathize with static pictures depicting facial expressions of worry and anger than the participants in the control group. However, the generalizability of these results is specific to state PSI. Both state and trait PSI should be evaluated in parallel to understand their influence on empathy, given the clear distinction between the two. Other negative affective states that can trigger empathy for physical and emotional suffering, such as pain and sadness, were also not studied. In addition, static pictures of facial expressions depicting negative emotions also lacked external validity due to the absence of dynamism. Using computer-generated faces (“virtual agents”) that express various affective states offers a valuable alternative for measuring empathy. This approach enables more precise manipulation of parameters such as dynamism and better control of facial expressions, surpassing the limitations typically associated with databases of real individuals (Dildine & Atlas, 2019; Dildine et al., 2023). It also has a better balance between internal and external validity. Recent studies have demonstrated that virtual agents can elicit reactions similar to those observed in real-life human interactions (e.g., Bélanger et al., 2025; Glikson & Woolley, 2020).

Present Research

Despite the numerous and well-documented negative impacts of PSI on physical and mental health, few studies have investigated the relationship between PSI and essential social skills for maintaining and developing relationships, such as perceiving suffering in others and empathizing with them. Identifying potential factors associated with PSI (e.g., empathy deficits) is thus crucial and the first step in developing more targeted interventions. Furthermore, published studies on these relationships have relied predominantly on self-report questionnaires, often neglecting the multifaceted nature of social isolation and failing to adequately control for confounding variables. The present study aimed to address these gaps by using a multi-method approach that combined questionnaires with measures obtained during a behavioural task.

The first objective of the present study was to investigate the differences in trait and state empathy for physical and emotional suffering (i.e., empathy for others’ distress; Morelli et al., 2015) between two groups of healthy adult individuals differing in their levels of trait PSI (i.e., low and high). Based on the same group comparison, the second objective was to examine the differences in the sensitivity to physical and emotional suffering, assessed by the perceived intensity of facial expressions depicting pain and sadness. To meet these two objectives, statistical control of the contribution of confounding factors, such as state PSI, OSI, and depressive symptomatology, was planned. The hypotheses were that individuals scoring high on trait PSI levels, compared to individuals scoring low, would report lower levels of state and trait empathy (regardless of the specific aspect; Hyp. 1); and perceive affective states as more intense even after accounting for levels of state PSI, OSI, or depressive symptomatology (Hyp. 2).

Method

Participants

The number of participants needed was based on previous studies with a similar design (Coll et al., 2017; Gingras et al., 2021; Tremblay et al., 2020) and an a priori power analysis. The analysis was conducted with G*POWER for the between-subjects effect of a mixed multivariate analysis of variance (MANOVA). A multivariate analysis was planned, as the two main dependent variables (i.e., state empathy and intensity of the affective state) are strongly associated (Coll et al., 2017; Tremblay et al., 2020). The following parameters were entered: effect size f = .30, α = .05, β = .80, number of groups = 2, number of measurements = 4, and correlation among repeated measures = .50. The analysis resulted in a minimum total sample size of 58 participants.

Interested individuals were recruited through ads on social networks, mailing lists of the university, and posters on the university campus. 105 participants agreed to participate. Of this sample, 39 participants completed only the remote online questionnaires, while 28 participants completed the questionnaires and later took part in an in-person experiment at the research team laboratory. 38 participants completed both the questionnaires and the in-person experiment in a single setting (see details in the Procedure section). All individuals who participated in the in-person experiment were combined into one sample of 66 participants. A $50 gift card was randomly awarded to one of the 67 participants who completed the remote online questionnaires, whether they decided to take part in the in-person experiment later. All participants in the in-person experiment received a $15 CAD compensation. Participants were required to have a proficient comprehension of written and spoken French. Individuals reporting neurological or psychiatric disorders, chronic pain, frequent exposure to individuals expressing pain (e.g., medical professionals), uncorrected vision problems, or those currently taking psychotropic medications were excluded from participating. The study was approved by an ethics committee (#2022-483). All participants voluntarily gave their written informed consent for their participation. The final sample (N = 66) included only participants who completed all questionnaires and computer task trials, ensuring the use of both state and trait measures while accounting for confounding variables (see Table 1 in the Results section for descriptive statistics).

Table 1.

Participants’ Frequencies and Mean Responses to the Questionnaires Divided Into Subgroups (Low and High UCLA-LS)

Variables	UCLA-LS_High (n = 33)	UCLA-LS_Low (n = 33)	t or χ² tests
Age (years) - M (SD)	26.45 (6.05)	25.06 (3.84)	t (64) = −1.12, p = .269
Education (years) - M (SD)	16.76 (2.46)	17.39 (1.98)	t (64) = 1.16, p = .252
Gender - n (%)			χ² (1) = 0.248, p = .618
Men	15 (45.50)	13 (39.40)
Women	18 (50.50)	20 (60.60)
Occupation - n (%)			χ² (1) = 0.00, p = 1.000
Studying	27 (81.80)	27 (81.80)
Working	6 (12.20)	6 (12.20)
UCLA-LS - M (SD)	47.42 (7.37)	30.48 (4.41)	t (64) = −11.33, p < .001**
IRI-PT - M (SD)	18.18 (5.00)	20.85 (4.33)	t (64) = 2.32, p = .024*
IRI-F - M (SD)	18.09 (5.52)	19.82 (5.70)	t (64) = 1.25, p = .216
IRI-EC - M (SD)	19.33 (4.65)	20.42 (4.72)	t (64) = 0.95, p = .348
IRI-PD - M (SD)	12.67 (5.21)	9.33 (4.78)	t (64) = −2.71, p = .009**
PHQ-9 - M (SD)	7.94 (5.12)	6.36 (3.56)	t (64) = −1.45, p = .152
PANAS-P - M (SD)	30.27 (5.29)	30.58 (6.77)	t (64) = 0.20, p = .840
PANAS-N - M (SD)	14.06 (4.34)	13.00 (3.39)	t (64) = −1.11, p = .273
LSNS - M (SD)	39.64 (9.98)	45.39 (9.77)	t (64) = 2.37, p = .021*

Note. UCLA-LS: UCLA Loneliness Scale; IRI-PT: Interpersonal Reactivity Index – Perspective-taking subscale; IRI-F: Interpersonal Reactivity Index – Fantasy subscale; IRI-EC: Interpersonal Reactivity Index – Empathic concern subscale; IRI-PD: Interpersonal Reactivity Index – Personal distress subscale; Patient Health Questionnaire; LSNS: Lubben Social Network Scale; PANAS-P: Positive and Negative Affects Schedule – Positive affects; PANAS-N: Positive and Negative Affects Schedule – Negative affects; Bold: Significant; *: <.05; **: <.01.

Material

Stimuli

The stimuli were 2000 ms videos of virtual agents displaying various facial expressions. These expressions, along with the agents’ appearances, were previously validated for their perceived affective state and intensity (Trieu et al., 2025). Facial expressions, especially those conveying pain, are an effective way to elicit empathy, as demonstrated by neuroscience research (e.g., Jauniaux et al., 2019). As key nonverbal cues, they can communicate emotions and elicit responses from observers, such as empathic ones (Frith, 2009). Such emotional expressions are perceived rapidly and automatically (Matt et al., 2021). They are among the most effective methods for inducing affective states, outperforming methods, such as autobiographical recall or story reading (Joseph et al., 2020). All virtual agents were generated using the MetaHuman Creator. They were animated in Unreal Engine software (Epic Games, Inc.), in which 161 facial controls can reproduce facial movements. Facial controls can be associated with Action Units (AUs; smallest visually discriminable facial movements) of the Facial Action Coding System (Ekman et al., 2002). According to this system, facial expressions can be broken down into AUs. Certain configurations of AUs are associated with specific affective states (e.g., pain = AUs 4, 6-7, 9-10 and 25-26-27, and sadness = AUs 1,4, 6, 15 and 17; Ekman et al., 2002; Kunz et al., 2019; Simon et al., 2008). Eight virtual agents (four representing women) with physical features stereotypical of four different phenotypes (i.e., Black, White, Latin American, or Asian; one per gender) were used. Each virtual agent displayed a moderate-to-intense (as perceived in average by participants in the validation study) facial expression of pain or sadness, or a neutral expression. A total of 24 stimuli were thus presented, corresponding to eight virtual agents each displaying three distinct expressions. For the pain and sadness stimuli, the videos showed a facial expression gradually changing from a neutral state to the target expression. A neutral expression (i.e., static image without facial movement) was used to ensure an adequate understanding of the task.

Questionnaires

Sociodemographic information. A self-reported questionnaire included open-ended questions about age and years of education, as well as multiple-choice questions on sex, gender, and occupation. Since responses to sex and gender were identical, only gender is reported.

Trait PSI. Trait PSI was assessed using the French version of the UCLA Loneliness Scale (UCLA-LS; Russell et al., 1980). This self-reported questionnaire has been validated and translated among undergraduate university students (De Grace et al., 1993). The French version has good psychometric properties (e.g., Cronbach’s α = .88; test-rest reliability: r = .85), comparable to the original. It comprises 20 items corresponding to statements such as “There is no one I can turn to” and “I fell on the same wavelength as people around me”. A 4-point Likert-type scale (“1 = Never”, “2 = Rarely”, “3 = Sometimes”, and “4 = Often”) is used to respond to each statement, with some items’ scoring inverted. The total scores range between 20 and 80. Higher scores indicate higher levels of trait PSI.

Trait Empathy. The French version of the Interpersonal Reactivity Index (IRI) was used to assess trait empathy (Achim et al., 2011; Davis, 1983), as self-report measures are one of the most comprehensive and efficient tools for assessing the subjective nature of this concept (Neumann et al., 2015; Vieten et al., 2024). This well-validated self-reported questionnaire consists of 28 items for which participants are asked to indicate how well a statement describes them on a 5-point Likert-type scale, with some items’ scoring inverted. Response options range from “Does not describe me at all” (0) to “Describes me very well” (4). The items are grouped into four subscales according to different aspects of empathy: Perspective-taking (PT), Fantasy (F), Empathic concern (EC), and Personal distress (PD). The first two subscales are often considered to measure empathy’s cognitive component, while the last two measure its affective component (Wang et al., 2020). Perspective-taking reflects the tendency to adopt the point of view of others spontaneously. Fantasy assesses the tendency to transpose oneself imaginatively into the feelings and actions of fictitious characters in books, movies, and plays. Empathic concern draws on “other-oriented” feelings of sympathy and concern for unfortunate others. Personal distress taps into “self-oriented” feelings of personal anxiety and unease in tense interpersonal settings. Each subscale comprises seven items, and the scores per subscale range from 0 to 28. For each subscale, higher scores indicate higher levels of the measured aspect of empathy.

Objective Social Isolation. The French translation of the self-reported Lubben Social Network Scale (LSNS; Lubben et al., 2006) was used to evaluate the extent of OSI. The research team carried out a homemade back-translation due to the absence of an official French version. The Cronbach’s alpha computed with the home translation was high and similar to the original version (α = .80). This instrument comprises 18 items on a 6-point Likert-type scale (0–5) assessing the size of the social network, the proximity with interpersonal contacts, and the frequency of the interpersonal contacts. Response options vary according to the type of question. For a question regarding social circle size, the options are: “None”, “One”, Two”, “Three or four”, “Five to eight”, and “Nine or more”. For a question regarding proximity: “Never”, “Seldom”, “Sometimes”, “Often”, “Very often”, and “Always”. For a question regarding the frequency of interpersonal contacts: “Less than monthly”, “Monthly”, “Few times a month”, “Weekly”, “Few times a week”, and “Daily”. The total scores range from 0 to 90. Lower scores indicate smaller social networks, including fewer social interactions, and therefore, higher levels of OSI.

Moods. The French version (Gaudreau et al., 2006) of the self-reported Positive and Negative Affects Schedule (PANAS; Watson et al., 1988) was used to assess participants’ moods through positive and negative affective states reported at the beginning of the experiment. The PANAS has good validity and fidelity and was validated in general and clinical populations (Watson et al., 1988). This instrument consists of 20 items, each representing adjectives that describe various feelings and emotions. Two subscales of 10 items divide the positive and negative affective states. The respondent is asked to indicate the extent to which each adjective describes how they are currently feeling (e.g., “To what extent do you feel this way right now, that is, at the present moment?”). A 5-point Likert-type scale is used: “1 = Very slightly or Not at all”, “2 = A little,” “3 = Moderately”, “4 = Quite a bit”, and “5 = Extremely”. Higher scores represent higher levels of positive or negative affect.

Depressive Symptomatology. The French translation of the 9-item version of the Patient Health Questionnaire (PHQ-9; Kroenke et al., 2001) was used to measure the severity of depressive symptoms. The original English version shows good psychometric properties (Kroenke et al., 2001), and the scores for the English and French versions are comparable (Arthurs et al., 2012). The participant reports how often they have been bothered by a specific problem over the last two weeks (e.g., little interest or pleasure in doing things, feeling down, depressed, or hopeless). A 4-point Likert-type scale is employed: “0 = Not at all”, “1 = Several days”, “2 = More than half the days”, and “3 = Nearly every day”. The total scores range from 0 to 27. Higher the scores, greater the depressive symptomatology.

Procedure

The questionnaires in the online survey were remotely administered using the Dexero web platform (Dexero, Inc.). They required around 20 minutes to complete, and included a consent form, a sociodemographic questionnaire, and the UCLA-LS and IRI scales.

The in-person experiment was conducted in the presence of a research assistant. Written consent was first obtained, followed by the completion of the PANAS questionnaire. Participants’ moods were assessed prior to the main task to account for their potential impact on the results. This allowed for a better interpretation of whether participants’ baseline emotional states may have influenced their responses during the experiment. This consideration is important, as positive and negative moods can influence empathy (Nezlek et al., 2001). Then, participants started a computer task created using the PsychoPy software (Peirce et al., 2019). The state PSI was first assessed by asking the participant to indicate their current level of loneliness (i.e., state PSI) with a VAS ranging from “No loneliness” to “Maximum loneliness” (1-100). The VAS was displayed until a response was confirmed by pressing the space bar. Previous studies used similar techniques to investigate the relationship between trait and state PSI (e.g., Tam & Chan, 2019).

Three experimental blocks followed (see Figure 1): Empathy, Intensity, and Recognition. The order of the first two blocks (Empathy and Intensity) was counterbalanced across participants to avoid potential carry-over effects. The third block (Recognition) was always administered last, as it was used as a manipulation check to ensure adequate recognition of the stimuli. Before each block, instructions were presented, followed by a training phase comprising eight trials with virtual agents different from those used in the study phase. In the Empathy block, the participant was asked to put themselves in the person’s shoes and report their empathy on a VAS. This behavioral measure was selected to assess state empathy under controlled conditions and to provide a complementary measure to the IRI questionnaire, thereby facilitating a more comprehensive understanding of this concept. The following definition was provided at the beginning of the block: “Empathy refers to the ability to understand and feel to a lesser degree what another is experiencing”. Each trial started with a fixation cross presented for 1000 ms. The 24 stimuli (i.e., videos of the eight virtual agents expressing three different affective states each) were presented three times for the Empathy and Intensity blocks (72 trials per block) and only once for the Recognition block (24 trials). After the presentation of each stimulus, participants rated their felt empathy for the virtual agent using a VAS ranging from “No empathy” to “Maximum empathy” (1-100) in the Empathy block, or the intensity level perceived on a VAS ranging from “No intensity” to “Maximum intensity” (1-100) in the Intensity block. In each block, the VAS was displayed until a response was confirmed by pressing the space bar. In the Recognition block, the participant identified the emotional states (i.e., affective states) perceived among the following three choices: “Neutral”, “Pain”, or “Sadness”. Response order was counterbalanced between participants. The choices were displayed until a response was confirmed by pressing the space bar.

Figure 1.

Computer Task Trial

After the task, participants filled out the LSNS and PHQ-9 questionnaires. Participants who had not previously responded to the sociodemographic questionnaire and the UCLA-LS and IRI scales completed these measures at that moment (i.e., participants who had not completed the online questionnaires prior to the in-person experiment). The in-person experiment lasted approximately between 45 and 60 minutes.

Analyses

Data Preprocessing and Manipulation Checks

All statistical analyses were performed using SPSS (IBM, Corp.) with a significance threshold set at α = .05. The online survey only included the trait questionnaires for the PSI and empathy. No significant differences in levels of trait PSI (t (64) = .196, p = .845) or any aspects of trait empathy (PT: t (64) = −1.211, p = .230; FS: t (64) = 1.346, p = .183; EC: t (64) = .021, p = .983; PD: t (64) = 1.241, p = .219) were found between groups based on data collection method. Accordingly, individuals (n = 28) who completed specific questionnaires remotely (sociodemographic, UCLA-LS, and IRI) and later participated in an in-person experiment did not differ from those (n = 38) who completed all questionnaires and the experiment in a single in-person setting.

No data were missing for the analyses, and no participants were excluded due to univariate or multivariate outliers. Neutral facial expressions were not included in the planned analyses, as they were used solely to verify participants’ understanding of the task (i.e., by examining whether neutral facial expressions were indeed perceived on average as more intense than painful or sad ones) and were not part of any objective or hypothesis. No participants were excluded based on this criterion, and descriptive statistics for the full sample (N = 66) revealed a clear distinction between average ratings of empathy (M = 13.70, SD = 15.90) and intensity (M = 7.07, SD = 10.70) for neutral expression, and non-neutral expressions (empathy for pain: M = 58.00, SD = 21.20; intensity for pain: M = 71.20, SD = 17.90; empathy for sadness: M = 58.30, SD = 23.00; intensity for sadness: M = 57.90, SD = 19.90). The normality of the data distribution for all analyses was verified by visual inspection of the residual distribution. No data set needed to be transformed to normalize the distribution.

Based on a median split of the score distribution, two subgroups were created to binarize the trait PSI variable and enable group comparisons: UCLA-LS_High (UCLA-LS >37) and UCLA-LS_Low (UCLA-LS ≤37). Although dichotomizing continuous variables is generally discouraged due to loss of statistical power (Cohen, 1983), it remains a common and acceptable practice in psychological research to facilitate group comparisons and for clarity in interpretation (DeCoster et al., 2011). A median split was used to create two equal groups, allowing for the inclusion of individuals across the full range of trait PSI scores. This approach yields a sample that is more representative of the general population than one composed solely of individuals with extreme scores. The absence of a clear consensus in the literature regarding cut-off values for high and low trait PSI supported this decision (Cacioppo & Patrick, 2008). Recent studies on trait PSI also used this method (e.g., Arnold & Winkielman, 2021; Baek et al., 2023). As expected, participants in the UCLA-LS_High subgroup (M = 47.42, SD = 7.37) showed significantly higher levels of trait PSI than those in the UCLA-LS_Low subgroup (M = 30.48, SD = 4.41), t (64) = −11.33, p < .001, d = 2.79). The score distribution (ranging from 21 to 63) and effect size were comparable to those reported in prior studies also using a dichotomized trait PSI variable with the same questionnaire (Arnold & Winkielman, 2021; Cacioppo et al., 2016).

Independent samples t-tests and χ² tests were performed on the sociodemographic variables with the binarized trait PSI variable as the group factor to assess equivalency. No significant differences were found between the subgroups for any sociodemographic characteristics (see Table 1; all ps > .05). Additionally, no significant differences were observed between subgroups in terms of empathy (t (64) = −1.30, p = .198) and intensity (t (64) = −1.40, p = .166) mean ratings for the neutral facial expression used to control participants’ understanding of the task.

Recognition rates in the computer task were calculated to check that stimuli were adequately recognized. Rates obtained were similar for painful (M = 98.87 %, SD = 4.63 %), sad (M = 98.77 %, SD = 4.80 %), and neutral facial expressions (M = 97.35 %, SD = 9.44 %). These recognition rates were high above the chance level (33.33%). Scores for all VAS (i.e., state PSI, empathy and intensity ratings) were obtained by transposing the cursor position into pixels on a scale of 1 to 100.

Hypotheses Testing

Group comparisons analyses were first computed using the binarized trait PSI variable as the group factor. A series of t-tests were carried out on all mean scores of the questionnaires assessing psychological characteristics, including trait empathy and state PSI mean ratings, to respond in part to the first objective. To respond to both the first and second objectives, the primary analysis conducted was a 2 (Facial Expression: Pain, Sadness) x 2 (Group: UCLA-LS_High, UCLA-LS_Low) mixed MANOVA, which was performed on empathy and intensity mean ratings. Univariate tests and pairwise comparisons with Bonferroni correction were performed to decompose the effects.

To address the limitations associated with dichotomization, complementary relationship analyses were also conducted using the non-binarized continuous trait PSI variable. Pearson’s correlation coefficients were computed to identify whether there were significant associations between the main dependent measures (i.e., state and trait empathy, and intensity of affective state) and potential predictors, including non-binarized trait PSI, and confounding variables (i.e., state PSI, OSI, and depressive symptomatology). If so, linear regressions (standard or stepwise, depending on the number of predictors) were carried out.

Results

Trait Empathy

Mean scores of the questionnaires for each subgroup (low and high on UCLA-LS) are shown in Table 1, including IRI subscales. There was a significant difference between subgroups in IRI-PT (t (64) = 2.32, p = .024, d = 0.57), and IRI-PD scores (t (64) = −2.71, p = .009, d = 0.67). Participants in the UCLA-LS_High subgroup showed lower levels of perspective-taking than those in the UCLA-LS_Low subgroup (M = 18.18, SD = 5.00 vs. M = 20.85, SD = 4.33, respectively) but higher levels of personal distress (M = 12.67, SD = 5.21 vs. M = 9.33, SD = 4.78, respectively). No significant differences were detected between subgroups for IRI-F and IRI-EC (all ps > .05). Other questionnaire results are discussed in the Confounding Influence of Situational Perceived Social Isolation PSI, Objective Social Isolation, and Depressive Symptomatology section.

State Empathy and Perceived Intensity of Affective State

Empathy and intensity mean ratings can be found in Figure 2. There was no significant effect of Group (Pillai’s Trace = .006, F (2, 63) = 0.196, p = .822) on empathy and intensity means ratings. There was a significant effect of Facial Expression (see Figure 2; Pillai’s Trace = .402, F (2, 63) = 21.188, p < .001, η_p² = .402). The effect was significant for intensity (F (1, 64) = 28.224, p < .001) but not empathy (F (1, 64) = 0.006, p = .936). The pain facial expression was perceived as more intense than the sadness facial expression (p > .001). There was no significant interaction between Group and Facial Expression (Pillai’s Trace = .011, F (2, 63) = 0.342, p = .712).

Figure 2.

Empathy and Intensity Mean Ratings according to the Affective State and Trait Perceived Social Isolation Subgroup

Confounding Influence of Situational Perceived Isolation, Objective Social Isolation, and Depressive Symptomatology

Social network levels, including frequency of interpersonal contacts, were significantly higher (t (64) = 2.37, p = .021, d = 0.58) in the UCLA-LS_Low subgroup (M = 45.39, SD = 9.77) compared to the UCLA-LS_High subgroup (M = 39.64, SD = 9.98), meaning less OSI in the UCLA-LS_Low subgroup and more OSI in the UCLA-LS_High subgroup. No significant differences were detected between subgroups for PHQ-9, PANAS-N, and PANAS-P scores (all ps > .05). Participants in the UCLA-LS_High subgroup (M = 50.90, SD = 24.34) reported significantly higher state PSI (t (64) = −3.15, p = .002, d = 0.78) compared to those in the UCLA-LS_Low subgroup (M = 31.68, SD = 25.24).

Significant associations were found between the mean scores of the questionnaires and computer task mean ratings (see Table 2). Significant but weak positive relationships were found between IRI-EC and state empathy ratings for pain (r (64) = .243, p = .049) and sadness (r (64) = .261, p = .034) facial expressions. No other significant associations (all ps > .05) were found between empathy or intensity ratings and potential predictors, including OSI, depressive symptomatology (PHQ-9), and subjective states (i.e., PANAS and State PSI) assessed prior to the computer task.

Table 2.

Correlations Between Mean Scores of the Questionnaires and Computer Task Mean Ratings

Variables	1	2	3	4	5	6	7	8	9	10	11	12	13	14
1. UCLA-LS
2. IRI-PT	−.291*
3. IRI-F	−.042	.019
4. IRI-EC	−.069	.343**	.256*
5. IRI-PD	.327**	−.174	.163	.106
6. LSNS	−.390**	.419**	−.101	.251*	−.215
7. PANAS-P	−.027	.204	−.056	−.074	−.025	.023
8. PANAS-N	.164	−.153	−.079	−.168	.135	−.031	.370**
9. PHQ-9	.368**	−.145	.018	.098	.271*	−.183	.113	.289*
10. State PSI	.498**	−.059	−.001	−.035	.151	−.288*	.056	.299*	.520**
11. Empathy for pain	−.096	.174	.239	.243*	−.019	−.017	.129	.061	−.092	.118
12. Empathy for sadness	.045	.125	.236	.261*	.016	.003	.164	.009	−.016	.001	.492**
13. Intensity for pain	−.101	.032	.218	.022	−.128	−.015	−.098	−.126	−.087	.094	.513**	.207
14. Intensity for sadness	.086	−.037	.194	.050	.075	−.139	−.045	−.146	.059	.039	.245*	.564*	.429**
M	38.95	19.52	18.95	19.88	11.00	42.52	30.42	13.53	7.15	41.29	58.03	58.25	71.23	57.92
SD	10.45	4.83	5.64	4.68	5.24	10.22	6.03	3.90	4.45	26.44	21.19	23.02	17.86	19.85

Notes. N = 66; UCLA-LS: UCLA Loneliness Scale; IRI-PT: Interpersonal Reactivity Index – Perspective-taking subscale; IRI-F: Interpersonal Reactivity Index – Fantasy subscale IRI-EC: Interpersonal Reactivity Index – Empathic concern subscale; IRI-PD: Interpersonal Reactivity Index – Personal distress subscale; LSNS: Lubben Social Network Scale; LSNS: Lubben Social Network Scale; PANAS-P: Positive and Negative Affects Schedule – Positive affects; PANAS-N: Positive and Negative Affects Schedule – Negative affects; PHQ-9: Patient Health Questionnaire-9 Items; Bold: Significant; *: <.05; **: <.01.

As for the IRI subscale scores, IRI-PT was significantly negatively associated with UCLA-LS (r (64) = −.291, p = .018; small effect size) but positively with LSNS (r (64) = .419, p < .001; moderate effect size) scores. IRI-EC was significantly positively associated with LSNS (r (64) = .251, p = .042; small effect size) scores, and IRI-PD was significantly positively associated with UCLA-LS (r (64) = .327, p = .007; small to moderate effect size) and PHQ-9 (r (64) = .271, p = .018; small effect size) scores. State PSI was not related to any IRI subscales scores (all ps >.05). No significant associations (all ps > .05) were found between IRI-F scores and potential predictors. As for other questionnaires, UCLA-LS scores were significantly negatively associated with LSNS (r (64) = −.390, p = .001; small to moderate effect size), and positively with PHQ-9 (r (64) = .368, p = .002; small to moderate effect size) and State PSI (r (64) = .498, p < .001; moderate effect size) scores. LSNS scores were significantly negatively related to State PSI scores (r (64) = −.288, p = .019; small effect size). Finally, PHQ-9 were significantly positively related to PANAS-N (r (64) = .289, p = .018; small effect size) and State PSI (r (64) = .520, p < .001; moderate effect size) scores.

Multiple linear regressions were then carried out to predict the three IRI subscale scores (i.e., IRI-PT, IRI-EC, and IRI-PD) associated with potential predictors found in the correlation analyses. A stepwise multiple regression was carried out to predict IRI-PT scores with UCLA-LS and LSNS scores as predictors. The model was statistically significant (R = .419, p < .001) with 16% variance explained (adjusted-R² = .163). UCLA-LS was excluded (p > .05) from the final model, which contains only LSNS as a predictor (β = .419, p < .001). A standard regression was carried out to predict IRI-EC scores because LSNS scores were the sole predictor. The model was statistically significant (R = .251, p < .042) with 4.8% variance explained (adjusted-R² = .048). A stepwise regression was carried out to predict IRI-PD scores with UCLA-LS and PHQ-9 as predictors. The model was statistically significant (R = .327, p = .007) with 9.3% variance explained (adjusted-R² = .093). PHQ-9 was excluded (p > .05) from the final model, which contains only UCLA-LS as a predictor (β = .327, p = .007). The final models of the three regressions are illustrated in Figure 3.

Figure 3.

Results of the Regression Models

Discussion

This study aimed to identify factors contributing to trait PSI by assessing its relationship with components essential to maintaining and developing interpersonal relationships, namely empathy and the perception of suffering in others. The study also considered various confounding factors, such as state PSI, OSI, and depressive symptomatology.

Objective and Subjective Social Isolation Predict Distinct Aspects of Trait Empathy

It was hypothesized in the first objective that trait empathy, regardless of its specific aspect, would be lower in individuals with high-trait PSI and that this difference would not be accountable by other variables. As expected, the tendency to spontaneously adopt the view of others (i.e., perspective-taking) was lower in individuals with high-trait PSI. Still, it was found that the objective facet of social isolation contributes the most to perspective-taking, even if a person feels isolated. The contribution of PSI trait levels was thus negligible when OSI was also considered, consistent with a recent study (Mwilambwe-Tshilobo et al., 2023). According to Davis (2017), perspective-taking is essential in interpersonal relationships and is frequently positively associated with relationship quality indicators (e.g., global relationship satisfaction) and different types of relationship support. Contrary to expectations, we did not find any difference in empathic concern (i.e., “other-oriented” feelings of sympathy and concern for unfortunate others) levels between the subgroups. Furthermore, a significant but weak relationship was found between empathic concern and OSI, but no association was found between empathic concern and trait PSI. Other authors also found that empathic concern shows only weak associations with measures of relationship quality (Davis, 2017). These first two results suggest that specific aspects of trait empathy (i.e., perspective-taking and empathic concern) may play an essential role in having an actual broader social circle but are unrelated to its subjective perception. According to Morelli et al. (2017), people with high trait empathy are central to social networks characterized by trust among members, suggesting that empathy supports close relationships and may ultimately lead to additional and stronger social bonds. Still, it seems that specific aspects of trait empathy, such as perspective-taking, may be more critical than others. Moreover, Fu et al. (2022) showed that social support mediated the relationship between prosocial behaviour and trait empathy. Thus, promoting social support could ultimately enhance prosocial actions, such as helping behaviour to alleviate the distress of others, which some authors argue would be the primary purpose of an empathic response (Batson et al., 2015).

As predicted, participants with high-trait PSI reported greater personal distress, reflecting “self-oriented” feelings of personal anxiety and unease in tense interpersonal settings, compared to those with low-trait PSI, even after controlling for depressive symptoms. This result could be explained by emotional dysregulation in individuals showing high-trait PSI, who may exhibit poor cognitive and behavioural strategies (e.g., rumination, catastrophizing, blame attribution, less use of cognitive reappraisal, and active rejection or withdrawal from others; Preece et al., 2021). Emotional dysregulation in individuals with high-trait PSI may increase discomfort in tense interpersonal settings and heighten focus on their own distress. One key aspect of emotional regulation is the ability to differentiate between self and others. Although participants did not report any psychiatric disorders, some may have experienced psychological symptoms linked to reduced self-other distinction, contributing to elevated personal distress. Beyond depression, conditions such as social anxiety (Sun et al., 2024) and schizotypy (Badcock et al., 2016) have been associated with trait PSI. Personal distress may also predict schizotypal traits, including excessive social anxiety, cognitive disorganization, and ideas of reference, possibly due to impaired self-other distinction (Eddy, 2025). This confusion may lead to difficulty identifying the origin of one’s own thoughts, emotions, and actions. However, this explanation seems less applicable to trait PSI, as neuroimaging evidence indicates that individuals high in trait PSI exhibit lower self-other overlap (Courtney & Meyer, 2020). More personal distress in high-trait PSI individuals could also, in part, be attributable to self-centeredness. According to Cacioppo et al. (2017), trait PSI increases implicit motivation for self-preservation and predicts self-centeredness in the subsequent year. This chronic tendency to self-centeredness could thus lead individuals with high-trait PSI to focus more on their distress rather than on that of others.

State Empathy and Sensitivity to Suffering Unrelated to Perceived Social Isolation

Contrary to the first hypothesis, individuals with high-trait PSI did not report diminished state empathy compared to individuals with low-trait PSI. These results are inconsistent with those reported in other studies manipulating feelings of PSI (Hu et al., 2020) or social exclusion (DeWall & Baumeister, 2006; Twenge et al., 2007), possibly due to design differences. For instance, trait and state PSI were measured in the present study rather than experimentally manipulated by a recall task, as in Hu et al. (2020)’s study. Thus, eliciting feelings of PSI may only temporarily blunt the empathic response, which would be more similar to the experience of social exclusion than to trait PSI (DeWall & Baumeister, 2006; Twenge et al., 2007), reinforcing the idea that social exclusion and PSI differ. The divergence in results could also be explained by the differentiation between empathic ability and the spontaneous tendency to express it (i.e., empathic propensity), as proposed by some authors (Keysers & Gazzola, 2014; Rameson et al., 2012; Tremblay et al., 2020). For instance, Tremblay et al. (2020) showed that an instruction to actively empathize compared to a passive observation could increase empathy ratings for individuals with lower empathic propensity (e.g., people with high autistic, alexithymic, or psychopathic traits). The paradigm used to probe empathy (i.e., the empathy selection task) by Hu et al. (2020) differed from the one used in the present study in which participants were instructed to empathize with the target actively. Thus, individuals with high-trait PSI may have a lower propensity for empathy that can be circumvented when asked to empathize with a target actively. Moreover, it is also possible that the relationship between trait PSI and state empathy is context dependent. For instance, individuals with high-trait PSI could be less likely to be empathic in a context where they feel emotionally overwhelmed. This would make sense, as one of our first results showed that individuals with high-trait PSI were more prone to personal distress. As Kim and Han (2018) proposed, personal distress could block empathic interaction instead of enhancing it.

It was also hypothesized that individuals with high-trait PSI would perceive higher intensity in facial expressions depicting physical and emotional suffering compared to individuals with low-trait PSI. This hypothesis was refuted, as participants perceived the same level of intensity regardless of the group to which they belonged. This result does not align with those reported in studies by Smith et al. (2022) and Vanhalst et al. (2017). This incongruity could once again be due to methodological discrepancies. The methods used by Smith et al. (2022) and Vanhalst et al. (2017) were implicit (i.e., time response and categorization) in comparison to the one used in the present study, in which intensity was operationalized as an explicit measure using a VAS ranging from “No intensity” to “Maximum intensity”. Trait PSI may be related to subtle differences in facial emotional decoding that need to be assessed using more implicit measures that may give higher sensitivity. Finally, social desirability could also explain the absence of a difference in perceived intensity between individuals presenting distinct levels of trait PSI. This interpretation can also be applied to empathy. Regardless of trait PSI, all participants could have been prone to social desirability, as the measures used were explicit, which may have been a problem, especially as society values empathy (Kämpfe et al., 2009).

Strengths, Limitations, and Future Perspectives

To our knowledge, this study is the first to examine in the same experiment the relationships between PSI, empathy, and sensitivity to the suffering of others. This study has the strength to distinguish PSI as a state or a trait and to explore their associations in a behavioural experiment without manipulating PSI feelings. PSI can be transitional or prolonged over time, and it is a problematic psychological construct to manipulate in an experimental setting without changing its nature. PSI is sometimes inferred to stem from the manipulation of social exclusion. However, this approach is problematic, since PSI can originate from various reasons beyond being excluded. Controlling confounding factors, such as OSI, allowed better targeting of the specific facet of social isolation at stake. Virtual agents also gave more significant control over the facial expressions employed, which can be easily reproduced to ensure replicability.

Notwithstanding these strengths, this study has several limitations, for which future directions are suggested. Its cross-sectional design first precludes inference of causal relationships. Longitudinal research could provide a more comprehensive assessment of PSI and empathy over time. Other key factors frequently associated with PSI, such as social anxiety or social avoidance (Sun et al., 2018), schizotypal traits (Badcock et al., 2016; Eddy, 2025) and autism spectrum disorder (Grace et al., 2022), age (Terán-Mendoza & Cancino, 2025), and culture (Barreto et al., 2021), were not considered and should be included in future work. For example, misperceptions in social anxiety can interact with those associated with PSI and worsen social difficulties (Fung et al., 2017). Clinical samples may also help clarify the link between loneliness and empathy by amplifying differences in loneliness levels. Moreover, self-report measures assessing empathy are prone to subjectivity and social desirability bias (Neumann et al., 2015). Including a social desirability questionnaire in future studies could mitigate such influence.

As previously described, virtual agents offer several advantages. Still, the results may not fully translate to dynamic real-world interactions. Conducting dyadic studies could help bridge this gap. The empathy task did not distinguish between cognitive, affective, and emotional regulation components, making it unclear which best predicts trait PSI. Evidence from specific populations (e.g., autism spectrum disorder; Fatima & Babu, 2024) suggests specific components may be more affected in individuals with high PSI (e.g., Preece et al., 2021). Using separate VAS could help clarify this. While facial expressions are effective in eliciting empathy, the results may also not generalize to other stimuli, such as voices. People may infer emotions more accurately from voices than from faces (Kraus, 2017). Whether the PSI-empathy relationship varies by stimulus type is yet to be determined. Finally, the overlap between social exclusion, OSI, and PSI (both trait and state) remains unclear. Comparing subgroups (e.g., individuals with experience of social exclusion, those with high OSI, and those with high PSI) could help distinguish these constructs.

Conclusion

Overall, this study highlights the importance of distinguishing objective and perceived social isolation when examining their relationships with social skills, such as empathy. The results also reinforce the idea that social skills may play an essential role in maintaining and developing relationships but are not related to one’s perception of their quality. This study ultimately enhances the understanding of the distinct aspects of social isolation and trait empathy. These findings are an essential first step toward developing interventions to promote specific aspects of empathy and enhance social support, and thus, mitigate the negative impacts of social isolation within the population.

Footnotes

ORCID iDs

Elliot Gagner

Audrey Marcoux

Marie-Hélène Tessier

Philip L. Jackson

Ethical Considerations

The study was approved by Université Laval’s ethics committee (#2022-483) on February 6, 2023.

Consent to Participate

All participants voluntarily gave their written informed consent for their participation.

Author Contributions

EG: conceptualization, methodology, acquisition of data, analysis, and interpretation of data, and writing of the original draft. AM: interpretation of data. MHT: virtual agents’ creation and validation, and interpretation of data. PLJ: conceptualization, funding acquisition, supervision, and interpretation of data. All authors contributed to critically revising the paper and approved the final paper.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: EG, AM, and MHT received scholarships from the Natural Sciences and Engineering Research Council of Canada and the Fonds de Recherche du Québec—Nature et Technologies. This work was supported by grants from the Canada Foundation for Innovation and the Natural Sciences and Engineering Research Council of Canada awarded to PLJ.

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Data Availability Statement

Dataset collected and analyzed during the study is available online:

Author Biographies

Elliot Gagner recently completed a doctorate in psychology (D.Psy.), with a specialization in clinical neuropsychology, at Université Laval, Quebec City, Canada. His doctoral research examined the relationship between social isolation and social processes such as face perception and empathy.

Audrey Marcoux is currently completing her Ph.D. in psychology, specializing in clinical neuropsychology at Université Laval, Quebec City, Canada. Her works explore the interpersonal communication of empathy and the factors that can influence it, such as individual characteristics (e.g., gender) and communication mediums (e.g., videoconferencing).

Marie-Hélène Tessier is currently completing her Ph.D. in psychology at Université Laval, Quebec City, Canada. Her thesis focuses on the communication of pain and emotions. She is also interested in the use of technology to assess and simulate affective states in humans using artificial intelligence and virtual characters.

Philip L. Jackson, Ph.D., is a full professor at Université Laval, Quebec City, Canada. His research focuses on human empathy, its physiological and cerebral markers, and its behavioral manifestations, which are increasingly influenced by the technology that surronds us.

References

Achim

A. M.

Ouellet

Roy

M. A.

Jackson

P. L.

(2011). Assessment of empathy in first-episode psychosis and meta-analytic comparison with previous studies in schizophrenia. Psychiatry Research, 190(1), 3–8. https://doi.org/10.1016/j.psychres.2010.10.030

Allen

S. F.

Gilbody

Atkin

van der Feltz-Cornelis

(2020). The associations between loneliness, social exclusion and pain in the general population: A N=502,528 cross-sectional UK Biobank study. Journal of Psychiatric Research, 130, 68–74. https://doi.org/10.1016/j.jpsychires.2020.06.028

American Psychiatric Association . (2024). New APA poll: One in three Americans feels lonely every week. https://www.psychiatry.org/news-room/news-releases/new-apa-poll-one-in-three-americans-feels-lonely-e

Arnold

A. J.

Winkielman

(2021). Smile (but only deliberately) though your heart is aching: Loneliness is associated with impaired spontaneous smile mimicry. Social Neuroscience, 16(1), 26–38. https://doi.org/10.1080/17470919.2020.1809516

Arthurs

Steele

R. J.

Hudson

Baron

Thombs

B. D.

CSRG Canadian Scleroderma Research Group . (2012). Are scores on English and French versions of the PHQ-9 comparable? An assessment of differential item functioning. PLoS One, 7(12), Article e52028. https://doi.org/10.1371/journal.pone.0052028

Badcock

J. C.

Barkus

Cohen

A. S.

Bucks

Badcock

D. R.

(2016). Loneliness and schizotypy are distinct constructs, separate from general psychopathology. Frontiers in Psychology, 7, 1018. https://doi.org/10.3389/fpsyg.2016.01018

Baek

E. C.

Hyon

López

Porter

M. A.

Parkinson

(2023). Lonely individuals process the world in idiosyncratic ways. Psychological Science, 34(6), 683–695. https://doi.org/10.1177/09567976221145316

Barreto

Victor

Hammond

Eccles

Richins

M. T.

Qualter

(2021). Loneliness around the world: Age, gender, and cultural differences in loneliness. Personality and Individual Differences, 169, Article 110066. https://doi.org/10.1016/j.paid.2020.110066

Batson

C. D.

Lishner

D. A.

Stocks

E. L.

(2015). The empathy—Altruism hypothesis. In Schroeder

D. A.

Graziano

W. G.

(Eds.), The Oxford handbook of prosocial behavior (pp. 259–281). Oxford University Press. https://doi.org/10.1093/oxfordhb/9780195399813.013.023

10.

Beadle

J. N.

Brown

Keady

Tranel

Paradiso

(2012). Trait empathy as a predictor of individual differences in perceived loneliness. Psychological Reports, 110(1), 3–15. https://doi.org/10.2466/07.09.20.Pr0.110.1.3-15

11.

Bélanger

Pouliot

Tessier

M.-H.

Marcoux

Cousineau

Trieu

J. N. P.

Jackson

P. L.

(2025). Reproducing age and sex biases in perceived pain intensity and prosocial behaviors using digital characters. Computers in Human Behavior, 170, Article 108702. https://doi.org/10.1016/j.chb.2025.108702

12.

Blandl

Eisenberger

N. I.

(2025). The lonely brain: Aligning theories of loneliness with data from neuroimaging studies. Annals of the New York Academy of Sciences, 1547(1), 75–87. https://doi.org/10.1111/nyas.15330

13.

Cacioppo

J. T.

Cacioppo

(2018). Chapter three - Loneliness in the Modern Age: An evolutionary theory of loneliness (ETL). In Olson

J. M.

(Ed.), Advances in experimental social psychology (58, pp. 127–197). Academic Press. https://doi.org/10.1016/bs.aesp.2018.03.003

14.

Cacioppo

J. T.

Chen

H. Y.

Cacioppo

(2017). Reciprocal influences between loneliness and Self-centeredness: A cross-lagged panel analysis in a Population- based sample of African American, Hispanic, and caucasian adults. Personality and Social Psychology Bulletin, 43(8), 1125–1135. https://doi.org/10.1177/0146167217705120

15.

Cacioppo

J. T.

Hawkley

L. C.

Ernst

J. M.

Burleson

Berntson

G. G.

Nouriani

Spiegel

(2006). Loneliness within a nomological net: An evolutionary perspective. Journal of Research in Personality, 40(6), 1054–1085. https://doi.org/10.1016/j.jrp.2005.11.007

16.

Cacioppo

J. T.

Hawkley

L. C.

Norman

G. J.

Berntson

G. G.

(2011). Social isolation. Annals of the New York Academy of Sciences, 1231(1), 17–22. https://doi.org/10.1111/j.1749-6632.2011.06028.x

17.

Cacioppo

J. T.

Hawkley

L. C.

Thisted

R. A.

(2010). Perceived social isolation makes me sad: 5-year cross-lagged analyses of loneliness and depressive symptomatology in the Chicago health, aging, and social relations study. Psychology and Aging, 25(2), 453–463. https://doi.org/10.1037/a0017216

18.

Cacioppo

J. T.

Patrick

(2008). Loneliness: Human nature and the need for social connection. WW Norton & Company.

19.

Cacioppo

Bangee

Balogh

Cardenas-Iniguez

Qualter

Cacioppo

J. T.

(2016). Loneliness and implicit attention to social threat: A high-performance electrical neuroimaging study. Cognitive Neuroscience, 7(1-4), 138–159. https://doi.org/10.1080/17588928.2015.1070136

20.

Cohen

(1983). The cost of dichotomization. Applied Psychological Measurement, 7(3), 249–253. https://doi.org/10.1177/014662168300700301

21.

Coll

M. P.

Grégoire

Eugène

Jackson

P. L.

(2017). Neural correlates of prosocial behavior towards persons in pain in healthcare providers. Biological Psychology, 128, 1–10. https://doi.org/10.1016/j.biopsycho.2017.06.005

22.

Courtney

A. L.

Meyer

M. L.

(2020). Self-other representation in the social brain reflects social connection. Journal of Neuroscience: The Official Journal of the Society for Neuroscience, 40(29), 5616–5627. https://doi.org/10.1523/JNEUROSCI.2826-19.2020

23.

Davis

M. H.

(1983). Measuring individual differences in empathy: Evidence for a multidimensional approach. Journal of Personality and Social Psychology, 44(1), 113–126. https://doi.org/10.1037/0022-3514.44.1.113

24.

Davis

M. H.

(2017). Empathy, compassion, and social relationships. In Seppälä

E. M.

SimonThomas

Brown

S. L.

Worline

M. C.

Cameron

C. D.

Doty

J. R.

(Eds.), The Oxford handbook of compassion science (pp. 299–315). Oxford University Press. https://doi.org/10.1093/oxfordhb/9780190464684.01

25.

Deason

K. G.

Luchetti

Deason

K. G.

Luchetti

Karakose

Stephan

O'Súilleabháin

P. S.

Hajek

Sutin

A. R.

Terracciano

(2025). Neuroticism, loneliness, all-cause and cause-specific mortality: A 17-year study of nearly 500,000 individuals. Journal of Affective Disorders, 368, 274–281. https://doi.org/10.1016/j.jad.2024.09.077

26.

Decety

Jackson

P. L.

(2004). The functional architecture of human empathy. Behavioral and Cognitive Neuroscience Reviews, 3(2), 71–100. https://doi.org/10.1177/1534582304267187

27.

DeCoster

Gallucci

Iselin

A.-M. R.

(2011). Best practices for using median splits, artificial categorization, and their continuous alternatives. Journal of Experimental Psychopathology, 2(2), 197–209. https://doi.org/10.5127/jep.008310

28.

De Grace

G.-R.

Joshi

Pelletier

(1993). L'Échelle de solitude de l'Université Laval (ÉSUL): validation canadienne-française du UCLA Loneliness Scale. Canadian Journal of Behavioural Science / Revue Canadienne des Sciences du Comportement, 25(1), 12–27. https://doi.org/10.1037/h0078812

29.

DeWall

C. N.

Baumeister

R. F.

(2006). Alone but feeling no pain: Effects of social exclusion on physical pain tolerance and pain threshold, affective forecasting, and interpersonal empathy. Journal of Personality and Social Psychology, 91(1), 1–15. https://doi.org/10.1037/0022-3514.91.1.1

30.

Dildine

T. C.

Amir

C. M.

Parsons

Atlas

L. Y.

(2023). How pain-related facial expressions are evaluated in relation to gender, race, and emotion. Affective Science, 4(2), 350–369. https://doi.org/10.1007/s42761-023-00181-6

31.

Dildine

T. C.

Atlas

L. Y.

(2019). The need for diversity in research on facial expressions of pain. Pain, 160(8), 1901–1902. https://doi.org/10.1097/j.pain.0000000000001593

32.

Eddy

C. M.

(2025). Self–other distinction and schizotypy: Affect sharing and alexithymia in the prediction of socially anxious and avoidant traits. Personality disorders, 16(2), 137–147. https://doi.org/10.1037/per0000669

33.

Eikelboom

W. S.

van den Berg

Beauchamp

M. H.

Bray

K. O.

Kumfor

MacPherson

S. E.

McDonald

Spikman

J. M.

Kessels

R. P. C.

(2025). Providing a taxonomy for social cognition: How to bridge the gap between expert opinion, empirical data, and theoretical models. Journal of Psychiatry & Neuroscience: Journal of Psychiatry & Neuroscience, 50(3), E157–E161. https://doi.org/10.1503/jpn.240156

34.

Ekman

Friesen

W. V.

Hager

J. C.

(2002). Facial action coding system. Manual and investigator’s guide, Salt Lake City, UT: Research nexus.

35.

Fatima

Babu

(2024). Cognitive and affective empathy in autism spectrum disorders: A meta-analysis. Review Journal of Autism and Developmental Disorders, 11(4), 756–775. https://doi.org/10.1007/s40489-023-00364-8

36.

Frith

(2009). Role of facial expressions in social interactions. Philosophical Transactions of the Royal Society of London - Series B: Biological Sciences, 364(1535), 3453–3458. https://doi.org/10.1098/rstb.2009.0142

37.

Wang

Chai

Xue

(2022). Examining the relationship of empathy, social support, and prosocial behavior of adolescents in China: A structural equation modeling approach. Humanities and Social Sciences Communications, 9(1), 269. https://doi.org/10.1057/s41599-022-01296-0

38.

Fung

Paterson

Alden

L. E.

(2017). Are social anxiety and loneliness best conceptualized as a unitary trait? Journal of Social and Clinical Psychology, 36(4), 335–345. https://doi.org/10.1521/jscp.2017.36.4.335

39.

Gaudreau

Sanchez

Blondin

J.-P.

(2006). Positive and negative affective states in a performance-related setting: Testing the factorial structure of the PANAS across two samples of French-Canadian participants. European Journal of Psychological Assessment, 22(4), 240–249. https://doi.org/10.1027/1015-5759.22.4.240

40.

Gierveld

J. de J.

Tilburg

T. G. van

Dykstra

P. A.

(2018). New ways of theorizing and conducting research in the field of loneliness and social isolation. In Vangelisti

A. L.

Perlman

(Eds.), The Cambridge handbook of personal relationships (pp. 391–404): Cambridge University Press.

41.

Gingras

Coll

M. P.

Tessier

M. H.

Tremblay

Jackson

P. L.

(2021). Pain evaluation and prosocial behaviour are affected by age and sex. European Journal of Pain, 25(9), 1925–1937. https://doi.org/10.1002/ejp.1809

42.

Glikson

Woolley

A. W.

(2020). Human trust in artificial intelligence: Review of empirical research. The Academy of Management Annals, 14(2), 627–660. https://doi.org/10.5465/annals.2018.0057

43.

Grace

Remington

Lloyd-Evans

Davies

Crane

(2022). Loneliness in autistic adults: A systematic review. Autism: The International Journal of Research and Practice, 26(8), 2117–2135. https://doi.org/10.1177/13623613221077721

44.

Håkansson Eklund

Summer Meranius

(2021). Toward a consensus on the nature of empathy: A review of reviews. Patient Education and Counseling, 104(2), 300–307. https://doi.org/10.1016/j.pec.2020.08.022

45.

Holt-Lunstad

Smith

T. B.

Baker

Harris

Stephenson

(2015). Loneliness and social isolation as risk factors for mortality: A meta-analytic review. Perspectives on Psychological Science: A Journal of the Association for Psychological Science, 10(2), 227–237. https://doi.org/10.1177/1745691614568352

46.

Zheng

Huang

(2020). Absence and presence of human interaction: The relationship between loneliness and empathy. Frontiers in Psychology, 11, 768. https://doi.org/10.3389/fpsyg.2020.00768

47.

Joseph

D. L.

Chan

M. Y.

Heintzelman

S. J.

Tay

Diener

Scotney

V. S.

(2020). The manipulation of affect: A meta-analysis of affect induction procedures. Psychological Bulletin, 146(4), 355–375. https://doi.org/10.1037/bul0000224

48.

Kämpfe

Penzhorn

Schikora

Dünzl

Schneidenbach

(2009). Empathy and social desirability: A comparison of delinquent and non-delinquent participants using direct and indirect measures. Psychology, Crime and Law, 15(1), 1–17. https://doi.org/10.1080/10683160802010640

49.

Keysers

Gazzola

(2014). Dissociating the ability and propensity for empathy. Trends in Cognitive Sciences, 18(4), 163–166. https://doi.org/10.1016/j.tics.2013.12.011

50.

Kim

Han

(2018). Does personal distress enhance empathic interaction or block it? Personality and Individual Differences, 124, 7783. https://doi.org/10.1016/j.paid.2017.12.005

51.

Kroenke

Spitzer

R. L.

Williams

J. B.

(2001). The PHQ-9: Validity of a brief depression severity measure. Journal of General Internal Medicine, 16(9), 606–613. https://doi.org/10.1046/j.1525-1497.2001.016009606.x

52.

Kunz

Meixner

Lautenbacher

(2019). Facial muscle movements encoding pain-a systematic review. Pain, 160(3), 535–549. https://doi.org/10.1097/j.pain.0000000000001424

53.

Lubben

Blozik

Gillmann

Iliffe

von Renteln Kruse

Beck

J. C.

Stuck

A. E.

(2006). Performance of an abbreviated version of the lubben social network scale among three European community-dwelling older adult populations. The Gerontologist, 46(4), 503–513. https://doi.org/10.1093/geront/46.4.503

54.

Matt

Dzhelyova

Maillard

Lighezzolo-Alnot

Rossion

Caharel

(2021). The rapid and automatic categorization of facial expression changes in highly variable natural images. Cortex; a journal devoted to the study of the nervous system and behavior, 144, 168–184. https://doi.org/10.1016/j.cortex.2021.08.005

55.

McLay

Jamieson

H. A.

France

K. G.

Schluter

P. J.

(2021). Loneliness and social isolation is associated with sleep problems among older community dwelling women and men with complex needs. Scientific Reports, 11(1), 4877. https://doi.org/10.1038/s41598-021-83778-w

56.

Morelli

S. A.

Lieberman

M. D.

Zaki

(2015). The emerging study of positive empathy. Social and Personality Psychology Compass, 9(2), 57–68. https://doi.org/10.1111/spc3.12157

57.

Morelli

S. A.

Ong

D. C.

Makati

Jackson

M. O.

Zaki

(2017). Empathy and well-being correlate with centrality in different social networks. Proceedings of the National Academy of Sciences of the United States of America, 114(37), 9843–9847. https://doi.org/10.1073/pnas.1702155114

58.

Mund

Freuding

M. M.

Möbius

Horn

Neyer

F. J.

(2020). The stability and change of loneliness across the life span: A meta-analysis of longitudinal studies. Personality and Social Psychology Review: An Official Journal of the Society for Personality and Social Psychology, Inc, 24(1), 24–52. https://doi.org/10.1177/1088868319850738

59.

Mushtaq

Shoib

Shah

Mushtaq

(2014). Relationship between loneliness, psychiatric disorders and physical health? A review on the psychological aspects of loneliness. Journal of Clinical and Diagnostic Research: Journal of Clinical and Diagnostic Research, 8(9), WE01–WE04. https://doi.org/10.7860/JCDR/2014/10077.4828

60.

Mwilambwe-Tshilobo

Setton

Bzdok

Turner

G. R.

Spreng

R. N.

(2023). Age differences in functional brain networks associated with loneliness and empathy. Network Neuroscience, 7(2), 496–521. https://doi.org/10.1162/netn_a_00293

61.

Neumann

D. L.

Chan

R. C. K.

Boyle

G. J.

Wang

Westbury

H. R.

(2015). Measures of empathy: Self-report, behavioral, and neuroscientific approaches. In Boyle

G. J.

Saklofske

D. H.

Matthews

(Eds.), Measures of personality and social psychological constructs (pp. 257–289). Elsevier Academic Press. https://doi.org/10.1016/B978-0-12-386915-9.00010-3

62.

Nezlek

J. B.

Feist

G. J.

Wilson

F. C.

Plesko

R. M.

(2001). Day-to-Day variability in empathy as a function of daily events and mood. Journal of Research in Personality, 35(4), 401–423. https://doi.org/10.1006/jrpe.2001.2332

63.

Peirce

Gray

J. R.

Simpson

MacAskill

Höchenberger

Sogo

Kastman

Lindeløv

J. K.

(2019). PsychoPy2: Experiments in behavior made easy. Behavior Research Methods, 51(1), 195–203. https://doi.org/10.3758/s13428-018-01193y

64.

Preece

D. A.

Goldenberg

Becerra

Boyes

Hasking

Gross

J. J.

(2021). Loneliness and emotion regulation. Personality and Individual Differences, 180, Article 110974. https://doi.org/10.1016/j.paid.2021.110974

65.

Rameson

L. T.

Morelli

S. A.

Lieberman

M. D.

(2012). The neural correlates of empathy: Experience, automaticity, and prosocial behavior. Journal of Cognitive Neuroscience, 24(1), 235–245. https://doi.org/10.1162/jocn_a_00130

66.

Rico-Uribe

L. A.

Caballero

F. F.

Olaya

Tobiasz-Adamczyk

Koskinen

Leonardi

Haro

J. M.

Chatterji

Ayuso-Mateos

J. L.

Miret

(2016). Loneliness, social networks, and health: A cross-sectional study in three countries. PLoS One, 11(1), Article e0145264. https://doi.org/10.1371/journal.pone.0145264

67.

Russell

Peplau

L. A.

Cutrona

C. E.

(1980). The revised UCLA loneliness scale: Concurrent and discriminant validity evidence. Journal of Personality and Social Psychology, 39(3), 472–480. https://doi.org/10.1037//0022-3514.39.3.472

68.

Simon

Craig

K. D.

Gosselin

Belin

Rainville

(2008). Recognition and discrimination of prototypical dynamic expressions of pain and emotions. Pain, 135(1- 2), 55–64. https://doi.org/10.1016/j.pain.2007.05.008

69.

Smith

K. E.

Norman

G. J.

Decety

(2022). Increases in loneliness during medical school are associated with increases in individuals' likelihood of mislabeling emotions as negative. Emotion, 22(4), 740–750. https://doi.org/10.1037/emo0000773

70.

Sun

Liu

(2018). Attentional characteristics of high-empathy people in processing emotional stimuli and evidence of eye movements. Journal of Psychological Science, 41(5), 1084–1089.

71.

Sun

Wang

Zhu

Qian

Zhu

(2024). Social anxiety and loneliness among older adults: A moderated mediation model. BMC Public Health, 24(1), 483. https://doi.org/10.1186/s12889-024-17795-5

72.

Tam

K. Y. Y.

Chan

C. S.

(2019). The effects of lack of meaning on trait and state loneliness: Correlational and experience-sampling evidence. Personality and Individual Differences, 141, 76–80. https://doi.org/10.1016/j.paid.2018.12.023

73.

Taylor

H. O.

Cudjoe

T. K. M.

Lim

M. H.

(2023). The state of loneliness and social isolation research: Current knowledge and future directions. BMC Public Health, 23(1), 1049. https://doi.org/10.1186/s12889-023-15967-3

74.

Terán-Mendoza

Cancino

(2025). Ageism and loneliness in people over 50: Understanding the role of self-perception of aging and social isolation in a Chilean sample. Journal of Applied Gerontology: The Official Journal of the Southern Gerontological Society, Article 7334648251314283. https://doi.org/10.1177/07334648251314283

75.

Tremblay

M. B.

Marcoux

Turcotte

Woods

Rouleau

Grondin

Jackson

P. L.

(2020). I can but I shall not always be empathic. Psychological Reports, 124(4), 332941209451801672. https://doi.org/10.1177/0033294120945180

76.

Trieu

J. N. P.

Tessier

M. H.

Pouliot

Bélanger

Leanza

Jackson

P. L.

(2025). Physical characteristics of digital characters influence group categorization and recognition of affective states. Computers in Human Behavior, 168, Article 108638. https://doi.org/10.1016/j.chb.2025.108638

77.

Twenge

J. M.

Baumeister

R. F.

DeWall

C. N.

Ciarocco

N. J.

Bartels

J. M.

(2007). Social exclusion decreases prosocial behavior. Journal of Personality and Social Psychology, 92(1), 56–66. https://doi.org/10.1037/0022-3514.92.1.56

78.

Vanhalst

Gibb

B. E.

Prinstein

M. J.

(2017). Lonely adolescents exhibit heightened sensitivity for facial cues of emotion. Cognition & Emotion, 31(2), 377–383. https://doi.org/10.1080/02699931.2015.1092420

79.

Vieten

Rubanovich

C. K.

Khatib

Sprengel

Tanega

Polizzi

Vahidi

Malaktaris

Chu

Lang

A. J.

Tai-Seale

Eyler

Bloss

(2024). Measures of empathy and compassion: A scoping review. PLoS One, 19(1), Article e0297099. https://doi.org/10.1371/journal.pone.0297099

80.

Wang

Xiao

Jie

(2020). Investigation on the rationality of the extant ways of scoring the interpersonal reactivity index based on confirmatory factor analysis. Frontiers in Psychology, 11, 1086. https://doi.org/10.3389/fpsyg.2020.01086

81.

Watson

Clark

L. A.

Tellegen

(1988). Development and validation of brief measures of positive and negative affect: The PANAS scales. Journal of Personality and Social Psychology, 54(6), 1063–1070. https://doi.org/10.1037//0022-3514.54.6.1063

82.

Wolf

L. D.

Davis

M. C.

(2014). Loneliness, daily pain, and perceptions of interpersonal events in adults with fibromyalgia. Health Psychology: Official Journal of the Division of Health Psychology, American Psychological Association, 33(9), 929–937. https://doi.org/10.1037/hea0000059

83.

Yamada

Wakaizumi

Kubota

Murayama

Tabuchi

(2021). Loneliness, social isolation, and pain following the COVID-19 outbreak: Data from a nationwide internet survey in Japan. Scientific Reports, 11(1), Article 18643. https://doi.org/10.1038/s41598-021-97136-3

84.

Yang

Wei

Ren

Zhang

(2023). How loneliness linked to anxiety and depression: A network analysis based on Chinese university students. BMC Public Health, 23(1), 2499. https://doi.org/10.1186/s12889-023-17435-4

85.

Zilioli

Jiang

(2021). Endocrine and immunomodulatory effects of social isolation and loneliness across adulthood. Psychoneuroendocrinology, 128, Article 105194. https://doi.org/10.1016/j.psyneuen.2021.105194

Objective Social Isolation Predicts Perspective-Taking and Empathic Concern

Abstract

Keywords

Present Research

Method

Participants

Material

Stimuli

Questionnaires

Procedure

Analyses

Data Preprocessing and Manipulation Checks

Hypotheses Testing

Results

Trait Empathy

State Empathy and Perceived Intensity of Affective State

Confounding Influence of Situational Perceived Isolation, Objective Social Isolation, and Depressive Symptomatology

Discussion

Objective and Subjective Social Isolation Predict Distinct Aspects of Trait Empathy

State Empathy and Sensitivity to Suffering Unrelated to Perceived Social Isolation

Strengths, Limitations, and Future Perspectives

Conclusion

Footnotes

ORCID iDs

Ethical Considerations

Consent to Participate

Author Contributions

Funding

Declaration of Conflicting Interests

Data Availability Statement

Author Biographies

References