Introduction

Ostracism, excluding or ignoring an individual or group, is rooted in human and social dynamics. Its impact on individuals’ social and psychological well-being has been the target of extensive research in contemporary psychology. Socrates, a key figure in ancient thought, stressed the importance of social connection for personal well-being, prefiguring modern concepts of ostracism. While he did not use the term “need for belonging,” his ideas laid the groundwork.

Psychological research has corroborated Socrates’ insights, demonstrating that ostracism can have negative effects on individuals’ mental health. In this regard, studies have shown that being ostracized activates similar neural pathways as physical pain, underscoring the severe emotional distress caused by ostracism (Eisenberger et al. 2003). It can lead to a cascade of adverse outcomes, including depression, anxiety, and decreased self-esteem (Williams 2009; Paolini 2019; for a meta-analysis, see Hartgerink et al. 2015). Furthermore, ostracism can disrupt psychological needs according to Self-Determination Theory (Deci and Ryan 2000). The threat to these needs can impair cognitive functioning, diminish motivation, and reduce overall life satisfaction (Baumeister and Leary 1995). In addition to the psychological impact, ostracism has significant social consequences. People respond to ostracism with behaviors that can be classified as prosocial, antisocial, or withdrawal (for a review, see Kip et al. 2025). Ostracized people cognitively evaluate the event, which leads them to undertake behaviors aimed at fortifying the basic needs threatened during ostracism (Needs Fortification Hypothesis, Williams 2009). Therefore, they may increase aggression and antisocial behavior (i.e., punishing the sources) as individuals attempt to regain a sense of control and significance (Carter-Sowell et al. 2008; Lakin et al. 2008; Riva et al. 2014; Twenge et al. 2001). At the same time, to enhance their self-image and promote social acceptance, ostracized individuals may also display prosocial behaviors such as compliance, conformity, and obedience (Chow et al. 2008; Lelieveld et al. 2013; Ren et al. 2018; Tedeschi 2001). Finally, the experience of ostracism can initiate a self-perpetuating cycle, whereby ostracized individuals become more likely to withdraw socially, further reinforcing their isolation (i.e., solitude-seeking; Ren et al. 2016).

Most of the studies mentioned above—particularly those employing physiological or neural measures—support the existence of a “primitive” system for detecting ostracism (Haselton and Buss 2000; Kerr and Levine 2008; Spoor and Williams 2007; Williams and Zadro 2005), characterized by immediate negative effects. This system corresponds to the Reflexive Stage of the Temporal Need-Threat Model (TNTM; Williams 2009), which is driven by threats to the satisfaction of basic needs—namely, belongingness, self-esteem, control, and meaningful existence (Williams 2009; Zadro et al. 2004). This initial reaction reverberates at the neurophysiological level, activating pain-related neural pathways such as the dorsal anterior cingulate cortex, the anterior insula, and the right ventral prefrontal cortex (Eisenberger et al. 2003). It also involves autonomic nervous system (ANS) activity (Ijzerman and Semin 2009; Ijzerman et al. 2012; Paolini et al. 2016; Sleegers et al. 2015), further confirming a state of intense distress. According to Williams (2009), responses at this stage are automatic and universal, unaffected by moderating variables or individual differences such as gender or social categorization (van Beest and Williams 2006). However, more recent findings challenge this assumption. A meta-analysis by Hartgerink et al. (2015) suggests that certain moderators may, in fact, influence individuals’ reflexive reactions to ostracism. For instance, individuals less prone to experiencing social pain—such as those with schizotypal personality traits—or participants under the influence of particular medications may show reduced susceptibility to ostracism, thereby attenuating its initial impact. Nonetheless, Hartgerink and colleagues caution that this conclusion is limited by methodological considerations: many measures classified as “first reactions” in the meta-analysis may not fully capture reflexive responses as defined by Williams (i.e., physiological, online, or immediate reports). Moreover, Williams’s assumption specifically concerns fundamental needs, and indeed, analyses restricted to immediate and delayed fundamental need satisfaction were consistent with the model’s predictions.

According to TNTM, after the immediate reflexive response, individuals quickly begin to interpret the experience of ostracism. They consider whether the episode warrants attention and the investment of cognitive resources (e.g., Paolini et al. 2020; Xu et al. 2017, 2020) and reflect on how to act to restore threatened needs. This phase, termed the Reflective Stage, can be modulated by stable individual differences that shape the selection of coping strategies (Williams 2009). For instance, Oaten et al. (2008) examined the role of social anxiety in moderating self-regulatory abilities following inclusion or ostracism. They found that while ostracism initially impaired self-regulation for all participants, only those high in social anxiety continued to show deficits over time. Beyond individual differences, situational contexts also play a role in recovery from ostracism. Evidence suggests, for example, that recovery is quicker and more complete when exclusion comes from out-group rather than in-group members (Goodwin et al. 2010; Wirth and Williams 2009). Taken together, these findings indicate that during the Reflective Stage, individuals actively appraise the motives, meaning, and relevance of the ostracism experience. Within this evaluative process, impression formation may also emerge as a way of making sense of the initial affective reaction. Indeed, ostracized individuals may cope by attributing intentionally negative traits to the source, which can lead them to devalue it and thereby mitigate the emotional impact of the experience (Williams 2007; Wesselmann et al. 2010). Supporting this reasoning, Zadro and colleagues (2006) found that ostracized people, compared to included ones, evaluated the sources as less physically attractive and possessing more negative personality traits. Similarly, Sloan and colleagues (2011) reported that the sources of ostracism are perceived as less trustworthy, more prejudiced, and more arrogant than the sources of inclusion. Importantly, Paolini and colleagues (2016) also showed that bystanders to ostracism hold a negative impression of the sources of ostracism.

To sum up, scientific evidence highlights that the immediate response to ostracism is mainly affective and linked to the threat of basic needs. In addition, some research also shows that victims form an impression of the sources, trying to mentalize their psychological state to understand whether they can be a resource or a threat (e.g., Fiske and Neuberg 1990). This suggests that cognitive processes follow the initial affective reaction. Neuroscientific literature supports this view, showing that impression formation involves cortical regions associated with complex high-level cognitive functions, such as the dorsomedial prefrontal cortex, the superior temporal sulcus, and the orbitofrontal cortex (Mitchell et al. 2006).

In the present contribution, we aimed to verify the comprehensive role of affective and cognitive responses as antecedents of antisocial behaviors toward the sources of ostracism, relying upon the TNTM. As a preliminary step, we sought to confirm previous findings on the differences between ostracism and inclusion in terms of need, mood, impression, and antisocial behaviors. First, we predicted that ostracized people (vs included) would report a decrease in both need satisfaction (i.e., belongingness, self-esteem, control, and meaningful existence) and mood as an immediate affective reaction to ostracism (HP1). Furthermore, consistent with cognitive findings (i.e., Zadro et al. 2006), we hypothesized that ostracized people (vs included) would form a negative (vs positive) impression of the sources of ostracism (HP2).

In terms of behavior, Lelieveld and colleagues (2013, experiment 3) investigated whether contextual factors (i.e., financial compensation) influenced the behavioral responses of ostracized participants. During the Cyberball game, ostracized participants either achieved money for each ball pass they did not receive (financial compensation condition) or did not (neutral condition). Afterward, they were invited to play Dictator Games (Forsythe et al. 1994) in which they had to divide €10 between themselves and the sources of Cyberball. The authors revealed that ostracized (vs included) participants allocated less money to the sources. Furthermore, participants allocated more money to sources that had ostracized them when they were in the financial compensation condition than when they were in the neutral condition. Thus, it seems that the valence of ostracism’s experience guides the implementation of prosocial or antisocial behavior. Since ostracism is a negative social phenomenon that undermines individuals’ psychological well-being, it is reasonable to expect that being ostracized leads to antisocial behaviors. Using the Dictator Game, we therefore anticipated that ostracized participants (vs included) would allocate less money toward the sources of ostracism (HP3a). At the same time, we acknowledge that prior research has shown that ostracism can sometimes elicit displaced aggression toward neutral or unrelated targets (Twenge et al. 2001; Ren et al. 2016). However, we expected that participants in our study would still allocate less money to the sources than to unrelated others (HP3b). This expectation is grounded in the idea that the primary target of negative responses is typically the source of ostracism, as these individuals are most directly responsible for the exclusion experience (Lelieveld et al. 2013; Williams 2007). Finally, we predicted that ostracism would not significantly affect allocations to unrelated others (HP3c). This expectation reflects the methodological context of our study: in the Dictator Game, the ostracizing sources were salient and directly linked to the exclusion episode, whereas unrelated others are neutral and not implicated in the episode. Under these circumstances, antisocial responses were expected to be concentrated on the ostracizers as the most salient norm violators, reducing the likelihood of displaced aggression toward neutral targets (cf. Twenge et al. 2001; Ren et al. 2016).

The lines of work described above, on first impressions, imply that impression formation can play a crucial role in reactions to ostracism. In the present study, we aimed to deepen this line of inquiry by integrating, for the first time, impression formation into the TNTM. Although both ostracism and impression formation concern interpersonal responses, they have not yet been integrated into a comprehensive model. Different hypotheses can be formulated regarding the interplay of ostracism, need satisfaction, mood, and impression formation. We reasoned that, when facing ostracism, people require time to mentalize their experience (Williams 2009) and understand why a stranger is excluding them. In this sequence, we expected that impression formation would intervene immediately after need satisfaction as a cognitive reaction to ostracism. Empirical evidence indicates that targets of ostracism tend to form negative impressions of ostracizers. Such evaluations may arise both as a direct response to the ostracizers’ norm-violating behavior (i.e., exclusion) and as a coping strategy to protect self-esteem and restore control after a socially threatening experience (Bastian et al. 2013; Williams and Nida 2011). Importantly, these two processes are not mutually exclusive but may operate simultaneously. In our study, participants evaluated ostracizers whom they did not personally know, suggesting that negative impressions may reflect both recognition of unfair behavior and psychological regulation to cope with ostracism. Consistently, recent work on first impressions has shown that even when targets are expected to elicit strong negativity, average evaluations remain moderate (Aquino et al. 2024). Thus, negative evaluations in ostracism may reflect a combination of social judgment and psychological regulation. Based on this reasoning, we hypothesized that ostracism would exert a direct effect on need satisfaction and mood, which in turn would mediate its impact on impressions of the sources. Finally, these impressions were expected to predict antisocial behavior, namely, the allocation of money toward the sources (see Fig. 1, HP4a).

Fig. 1
figure 1

Graphical summary of the study’s conceptual model and hypotheses.

Full size image

Considering the complex interplay between affect and cognition in predicting behavior (e.g., Rocklage and Luttrell 2021), an alternative temporal hypothesis suggests that impression formation occurs with mood shifts, rather than as a subsequent, independent step. To determine whether impression formation occurs in a distinct second phase—as opposed to simultaneously with the initial affective response—we employed structural equation modeling. This method allows for the comparison of alternative models to identify the best fit to the data. Accordingly, we also tested an alternative model (HP4b) in which need satisfaction, mood, and impression formation operate simultaneously rather than sequentially.

Method

Participants and design

Hair and colleagues (2010) suggested that the sample size should be five times the number of items in the questionnaire. As applied to our research, a sample size of minimum 130 participants was required. However, considering potential drop-out, questionnaires were administered to a sample of 168 respondents (83 females and 85 males; mean age = 22.75; SD = 3.41). Participants were randomly assigned to the conditions (Cyberball experience: inclusion vs ostracism) in a between-subjects design.

Procedure

Participants were individually escorted to the social psychology laboratory. Before starting the research, according to the ethical standards Declaration of Helsinki (2001), participants were informed about all relevant aspects of the study (e.g., methods, institutional affiliations of the researcher). They were apprised of their right to anonymity, to refuse to participate in the study, or to withdraw their consent to participate at any time during the study without fear of reprisal. Participants then confirmed that they had understood the instructions correctly, agreed to participate in the study, and provided their demographic information. In the first section, participants were introduced to a three-player online virtual ball-tossing game—namely Cyberball (Williams et al. 2000). After the manipulation of ostracism, participants completed the Need-Satisfaction and the mood Scale (Zadro et al. 2004, 2006). Subsequently, participants rated the Cyberball manipulation check, and then, they completed the measure about their impression of the source. Finally, we assessed the behavior towards the source through a set of dictator games. At the end, participants were thanked and fully debriefed.

Measures

Ostracism manipulation

Participants were informed that they would take part in a three-player online virtual ball-tossing game—namely Cyberball (Williams et al. 2000). They were told that they were participating in research about their mental visualization ability and were led to believe they would play with two other players connected via the campus intranet. Actually, the other two players were pre-programmed avatars. Participants were asked to exercise mental visualization skills while playing the game; they were asked to imagine the context of the game and what the other players were like. The two fictitious players were labeled as Player A and Player B. These players were the sources and placed at the top of the screen on the right and on the left, respectively. Half of the participants were ostracized during the Cyberball game (i.e., they received one toss at the beginning and then never received another toss), and the other half of the participants were equally included by other players (i.e., they received one-third of the tosses). The game proceeded for 3 min, for a total amount of 30 throws (Williams and Jarvis 2006).

Need satisfaction and mood scale

Participants completed 20-items of the Need-Satisfaction Scale (Zadro et al. 2004; see also Williams et al. 2000) assessing the impact of Cyberball on belongness (e.g., ‘I felt excluded’; ‘I felt disconnected’), self-esteem (e.g., ‘I felt liked’; ‘I felt satisfied’), control (e.g., ‘I felt I had control’; ‘I felt to able to significantly modify the events’), meaningful existence (e.g., ‘I felt invisible’; ‘I felt useful’) and they also completed an 8-items scale about the mood they felt during the game on positive and negative level (on a Likert-type scale from 1 (Not at all) to 7 (Extremely). Considering that the four needs had sometimes impacted in different ways on the expected outcome (e.g., Williams 2009), we tested our model by considering both the separated and the combined scores of needs. Given that no differences emerged in the hypothesized pattern, we decided to compute a total score of need satisfaction, following previous literature (van Beest and Williams 2006; Williams et al. 2000; Zadro et al. 2006). After reverse-coding negative items, we averaged responses to the “Need Satisfaction Scale” into a need satisfaction index (α = 0.93) and the questions about emotions into a mood index (α = 0.81). Lower ratings signaled more need threat and negative mood, higher ratings indicated more need satisfaction and positive mood.

The differential analyses for each need are available at: https://osf.io/zmgt5/?view_only=9a49c6547c9e43f79beaafc4e2c28f9e

Manipulation check

Participants responded to two items, “I was ignored” and “I was excluded” on a scale ranging from 1 (not at all) to 7 (extremely). They also responded to the following open-ended question: “Assuming the ball should be thrown to each person equally (33%), what percentage of the throws did you receive?”.

Impression formation

Participants were asked to report their overall impression of each of the two players they had interacted with during the Cyberball game. They reported their response to the question “What is your general impression of Player [A; B]?” on a scale ranging from 1 (= extremely negative) to 7 (= extremely positive). Responses were averaged to create a single index of impression toward the sources (r = 0.65; p < 0.001). Lower ratings signaled a more negative impression, while higher ratings signaled a more positive impression.

Antisocial behavioral responses

Participants randomly played three different Dictator Games (Forsythe et al. 1994), in which they had to allocate ten euros between themselves and another individual. In the first two Dictator Game participants had to decide to share money between themselves and Player A and Player B, respectively. In the last one, they had to allocate money between themselves and new people with whom they did not have any contacts before (unrelated others). In each game, participants could choose between 6 combinations: 1 (8 euros for themselves, 2 for the others), 2 (7 euros for themselves, 3 for the others), 3 (6 euros for themselves, 4 for the others), 4 (5 euros for themselves, 5 for the others), 5 (4 euros for themselves, 6 for the others) and 6 (3 euros for themselves, 7 euros for the others). We averaged responses related to players A and B into a single index (r = 0.78; p < 0.001). In this way, lower ratings signaled less money allocation to the sources and more to themselves, higher ratings indicated more money allocation to the sources and less to themselves.

Results

Data analysis

Unless otherwise specified, we performed a one-way analysis of variance (ANOVA) for each dependent variable, with the Cyberball experience (ostracism vs inclusion) as a between-subjects factor. Path analysis was performed through the regression approach and the bootstrap estimation in AMOS 21 (SPSS Inc., Chicago, IL 60606, USA). We further run a 1000-simulation Power Analysis for Significant Parameter Estimation (Wang and Rhemtulla 2021).

Manipulation check

Firstly, we evaluated the effectiveness of the Cyberball manipulation. Ostracized participants felt more ignored (M = 5.92, SD = 1.12) and excluded (M = 5.69, SD = 1.32) than included participants (M = 2.12, SD = 1.45; M = 2.06, SD = 1.59, respectively), F (1, 152) = 331.66, p < 0.001, partial η2 = 0.69; F (1, 152) = 236.25, p < 0.001, partial η2 = 0.61, respectively. Moreover, participants assigned to the ostracism condition reported that they received fewer throws out of the total (M = 9.97, SD = 7.71) compared to participants assigned to the inclusion condition (M = 32.82, SD = 14.18, F (1, 152) = 108.40, p < 0.001, partial η2 = 0.41. Taken together, these findings confirmed the effectiveness of our manipulation.

Need satisfaction, mood, and impressions toward the sources

The ANOVA on need satisfaction showed that ostracized participants reported a lower level of need satisfaction (M = 2.93, SD = 0.72) than included participants (M = 4.77, SD = 0.79), F (1, 166) = 275.39, p < 0.001, partial η2 = 0.62. The ANOVA on mood showed that ostracized participants reported lower scores of mood (M = 4.41, SD = 0.87) than included participants (M = 5.49, SD = 0.85), F (1, 166) = 64.78, p < 0.001, partial η2 = 0.28. Finally, the ANOVA on impression towards the sources of ostracism showed that ostracized participants reported a more negative impression (M = 2.75, SD = 0.94) than the included participants (M = 4.61, SD = 0.98), F (1, 166) = 156.87, p < 0.001, partial η2 = 0.49.

Results confirmed our first two hypotheses, indicating that ostracized (vs included) participants felt more threatened (i.e., less need satisfaction and more negative mood; Hp1) and reported a more negative impression toward the sources (Hp2).

Anti-social behavioral responses

A 2 (Cyberball experience: inclusion vs ostracism) × 2 (Dictator game: sources vs unrelated others) R mixed-model ANOVA was conducted with dictator game varying within participants. The ANOVA showed a significant two-way interaction (F (1, 165) = 13.62, p < 0.001, η2 = 0.08).

A simple effects analysis confirmed our hypotheses regarding the impact of ostracism on monetary allocations in the Dictator Game. Supporting Hp3a, ostracized participants allocated significantly less money to the sources of ostracism (M = 3.57, SD = 1.26) than included participants (M = 4.19, SD = 1.15), F (1, 165) = 11.24, p = 0.001, η² = 0.06. Consistent with Hp3b, in the ostracism condition, participants allocated less money to the sources compared to the unrelated other person, F(1, 165) = 17.06, p = 0.001, η² = 0.09, indicating that the negative response was specifically directed toward the ostracizers rather than neutral recipients. In contrast, in the inclusion condition, no significant difference emerged between sources and unrelated others, F(1, 165) = 1.20, p = 0.27, η² = 0.01. Finally, in line with Hp3c, allocations to the unrelated other did not differ between ostracized (M = 4.19, SD = 1.25) and included participants (M = 4.02, SD = 1.36), F(1, 165) = 0.68, p = 0.41, η² = 0.004.Footnote 1

Structural equation modeling

Table 1 shows the mean, standard deviation, and correlation coefficients among all measures.

Table 1 Mean, standard deviation, and zero-order correlations among variables.
Full size table

In line with our HP4a, we drew a predictive model where ostracism had a direct effect on the impression towards the sources as well as an indirect effect mediated by both need satisfaction and mood. Given that need satisfaction and mood have been shown to be related to each other in previous literature (Zadro et al. 2004), we included a correlation between the error terms of these endogenous variables. We further expected that the impression towards the sources would predict the allocation of money toward the sources. To assess the fit of the proposed model, we report the chi-square goodness of fit test, the non-normed fit index (NNFI), the comparative fit index (CFI), and the root mean square error of approximation (RMSEA). A chi-square value of zero indicates optimal fit, whereas a higher chi-square value indicates worse fit. More specifically, a non-significant chi-square indicates that the difference between the observed and estimated variance/covariance matrices is not significantly different from zero. The NNFI and CFI indicated the extent to which the tested model improves upon the null model, which assumes no covariance among the measured variables.

The value of these indices can vary between 0 and 1, with higher values indicating a better fit between the observed and estimated covariance matrices. The RMSEA indicated how well the model, with unknown but optimally estimated parameters, would fit the population’s covariance matrix. SRMR is an absolute measure of fit, defined as the standardized difference between the observed correlation and the predicted correlation. Because the SRMR is an absolute measure of fit, a value of zero indicates a perfect fit. SRMR and RMSEA values < 0.08, suggest an acceptable model fit, while values < 0.05 indicate excellent model fit. CFI and TLI > 0.90 indicate good model fit, while values > 0.95 reflects an excellent model fit. To test the indirect effects, we used bootstrapping (with 1000 resamples) to compute 95% confidence intervals (CI). CIs that do not include 0 denote statistically significant effects. Since our focus was on the ostracism condition, we included it in the SEM model as a dummy variable, assigning a value of 1 to ostracism and 0 to inclusion. We assumed that the variances/covariances of the endogenous variables were equivalent between the conditions.

Our hypothesized model (Fig. 2, HP4a) fits the data extremely well (HP3a), as indicated by a non-significant chi value = 1,93, p = 0.588, high NFI and CFI values (1 and 0.99, respectively), a RMSEA of 0.01 (CIs = 0.00; 0.10), and SRMR of 0.02.

Fig. 2
figure 2

Results of the structural equation analyses. For each relationship, standardized coefficients are reported (95%). CI confidence interval. ***p < 0.001; **p < 0.05.

Full size image

As expected, the ostracism predicted both the level of both need satisfaction (β = −0.79; p < 0.001, 95% CIs: −0.88; −0.69, Power: 0.99) and the mood (β = −0.53.; p < 0.001, 95% CIs: −0.66; −0.39, Power: 0.98), which in turn predicted the impression towards the source of ostracism (β = 0.46; p < 0.001 for need; Power: 0.99; 95% CIs: 0.52; 0.81; β = 0.16; p = 0.023; 95% CIs: 0.01; 0.27 for mood, Power: 0.42). Need satisfaction and mood partially mediated the effect of ostracism on the impression towards the source of ostracism. Alongside a significant indirect effect (I.E. = −0.44; p = 0.001; 95% CIs: −0.54; −0.35), the direct effect remained significant (β = −0.25; p = 0.002; 95% CIs: −0.36; −0.14, Power: 0.43). As expected, impression towards the source of ostracism predicted the amount of money allocated to the source (and consequently to themselves), β = 0.38; p = 0.001; 95% CIs: 0.26; 0.49, Power: 0.99). The impression towards the source of ostracism fully mediated the effect of both need satisfaction and mood on the money allocation toward the sources. Neither the direct effect of participants’ level of need satisfaction (β = −0.10; p = 0.396; 95% CIs: −0.31; 0.10) nor that of mood (β = 0.14; p = 0.20; 95% CIs: −0.04; 0.33) on money allocation toward the sources was significant. Consistent with our hypothesis, we observed a significant indirect effect of both need satisfaction and mood on money allocation (I.E. = 0.17; p = 0.001; 95% CIs: 0.08; 0.28; I.E. = 0.06; p = 0.014; 95% CIs: 0.01; 0.12, respectively).

Of relevance for our purpose, the ostracism did not exert a direct effect on money allocation (β = 0.03; p = 0.747; 95% CIs: −0.14; 0.20), but only an indirect effect (I.E. = −0.28; p = 0.001; 95% CIs: −0.40; −0.16), giving support to our predictive model.

As a control process, we tested the same model considering the allocation toward the unrelated other person. In this case, no variables predicted the allocation toward the unrelated other, either directly or indirectly, including the impression (β = −0.04; p = 0.786; 95% CIs: −0.24; 0.65), supporting our hypotheses.

Finally, we also tested an alternative model in which impression, need satisfaction, and mood simultaneously predicted the money allocation toward the sources (HP4b). However, this model did not show good fit indices: χ2 (3): 52.75, p = 0.000; RMSEA: 0.31; SRMR: 09; CFI:.89; TLI: 0.65.

Taken together, the results confirmed our hypothesized model (HP4a), showing that the ostracism had a direct effect on the impression towards the sources, as well as an indirect effect mediated by both need satisfaction and mood. In turn, the impression predicted the allocation of money toward the sources. Our findings clearly demonstrated that impression toward the source fully mediated the impact of ostracism, need satisfaction, and mood on behavior (i.e., the allocation of money to the source of ostracism).

Discussion

The present study aimed to verify the comprehensive role of affective and cognitive consequences as antecedents of antisocial behavioral response toward the sources of ostracism, relying upon the TNTM. Notably, we introduced, for the first time, the concept of impression formation into the TNTM. To this end, we tested a sequential model that incorporated both need satisfaction and impression formation. Various hypotheses can be formulated regarding the interplay between need satisfaction, mood, and impression formation. We reasoned that in front of an ostracism experience, people need time to mentalize their experience (as suggested by Williams 2009) and to understand the motives behind a stranger’s exclusionary behavior. In this sequence, we expected that impression formation would intervene immediately after need satisfaction as a cognitive response to ostracism. At the same time, it is important to acknowledge that negative impressions of ostracizers may not only serve as a coping mechanism, but may also arise as a direct response to their norm-violating behavior. Excluding someone from a social interaction represents a violation of fundamental relational expectations, and such violations are often met with moral condemnation and negative evaluations (e.g., Lelieveld et al. 2013). Taken together, these considerations suggest that impression formation following ostracism may serve a dual role: it can help individuals cope with the immediate psychological threat of exclusion, while also functioning as a social evaluation of the norm-violating behavior of the ostracizers.

Crucially, in line with our hypothesis, we successfully tested a comprehensive temporal model, showing that ostracism exerted a direct effect on the impression towards the sources and an indirect effect through the mediation of both need satisfaction and mood. Then, in turn, the impression predicted the allocation of money toward the sources. These findings significantly contribute to research on both ostracism and the evaluation field. From the ostracism perspective, the present study represents one of the first empirical confirmations of a processual model about the consequences of ostracism. In line with the temporal model hypothesized by Williams (2009), after a reflexive reaction (e.g., threat of basic need and lowering level of mood as a proxy of an affective reaction) people shape a negative impression towards the source of ostracism as a proxy of a socio-cognitive process in which they try to explain the ostracism experience (i.e., reflective stage). In other words, the first impression of the source intervenes immediately after the instinctive affective reaction (i.e., reflexive stage), representing the first step towards the second temporal stage (reflective stage), in which cognition sits alongside affectivity, gradually assuming an increasingly important role in the coping process. In line with the impression formation configurational model (Asch 1957), ostracized individuals use ostracism as a central diagnostic cue around which they build their global evaluation of the source. In terms of cue diagnosis, person-perception research provides evidence for a negativity bias (Skowronski and Carlston 1987), meaning that a person’s negative behavior (i.e., ostracizing the victim) acts as a central cue in impression formation (Baumeister et al. 2001; Rozin and Royzman 2001). Our findings confirm that impression formation is a high-level cognitive activity (Mitchell et al. 2006), which requires people to mentalize the other's psychological state to understand if the other person can be a resource or a threat for the perceiver (Fiske and Neuberg 1990).

It is important to note that the alternative model with a parallel (and not temporal) sequence of mediators did not fit the data well, as well as the model with behavioral response towards unrelated others. From both methodological and theoretical points of view, these control analyses increase the confidence that we have identified the process underlying the behavioral response to ostracism.

Moreover, the present study confirmed the relevant role of both cognitive and affective reactions in determining the behavioral responses (e.g., Di Plinio et al. 2023; Haddock and Maio 2019). Recently, Kiviniemi and colleagues (2017) argued that it is important not to focus on models that treat thoughts and feelings as distinct phenomena, but to consider the interplay of affect and cognition. Accordingly, the ostracism experience impact on need satisfaction and mood that serves an affective informational role, guiding the content and valence of subsequent formation of the impression towards the source of ostracism as cognitive processing of the negative affective experience of ostracism. The interplay of affect and cognition in ostracism leads to the behavioral reaction of ostracized people, punishing the source of ostracism by giving them less money. This behavior reaction leads to a spiral of negative action, which can have important consequences on long-term for both the ostracized and the ostracizing people (Paolini 2019; Riva et al. 2016). Furthermore, our findings underlined the importance of considering the first signals of ostracism, in order to avoid the above-mentioned negativity spiral. Ad hoc program intervention would address to envelop or enhance a group-based resilience to face and prevent the negative consequences of ostracism (Pagliaro et al. 2013).

It is worth noting that the observed pattern holds across different threatened needs, thereby strengthening the robustness of our findings. Although this consistency is not unexpected in the ostracism literature (e.g., van Beest and Williams 2006), TNTM suggests that distinct types of need threat may elicit different coping responses. A plausible explanation for the consistency of our results may lie in the relatively limited variance of the outcome measures—specifically, the reliance on a unidimensional valence measure and a single antisocial behavioral outcome—which could have constrained the emergence of differential effects. Future research employing multidimensional mediators or a broader range of outcomes may reveal distinct patterns for each threatened need (e.g., Sacino et al. 2024).

Our results also confirmed previous literature about the affective, cognitive, and behavioral consequences of ostracism. Indeed, we confirmed that ostracized people (vs included) reported a decrease in need satisfaction (i.e., belongingness, self-esteem, control, and meaningful existence) and negative mood, as an immediate affective reaction to ostracism (Williams 2009). Moreover, our findings confirmed that ostracized people (vs included) shaped a more negative impression towards the sources of ostracism (Zadro et al. 2006; Paolini et al. 2017; Twenge et al. 2001). Finally, moving to the behavioral consequences, our findings confirmed that ostracized participants allocated less money to the sources compared with included participants (Lelieveld et al. 2013).

Limitations and future directions

Our results do present limitations worth noting and raise questions for future investigation. First, we used an allocation game with money choice by considering only anti-social behaviors. Although this measure has an established tradition in ostracism studies (see Paolini et al. 2019, for a review), research revealed that people vary greatly in their responses to being ostracized, showing not only antisocial behavior but also prosocial and withdrawal behavior (Salvati et al. 2019; for a review, see Kip et al. 2025). Future research could be directed at understanding what factors can lead ostracized people to behave in different, as well as in opposite ways. From a methodological point of view, we did not randomize the order in which the various dependent variables were assessed. Participants first completed the need satisfaction and the impression formation, which were followed by the allocation game. We chose this order because we were interested in understanding the role of mood, need satisfaction, and impression formation on behavioral response, thus, the choice was guided by our theoretical model. Another limitation of the present study was the use of a single item to assess the impression of others. Although such measures are common within psychology (Allen et al. 2022), a multiple-dimensional measure of the impression could give interesting and innovative insight. In this regard, extensive research has shown that our first impressions of others are centered around two global social dimensions. While these dimensions have been referred to under varying names and have been defined somewhat differently, they share much common content (see Abele and Wojciszke 2007; Abele et al. 2021 for a review). Communion comprises characteristics that are related to forming and maintaining social connections (also referred to as warmth or nurturance, e.g., Abele and Wojciszke 2007; Fiske, Cuddy and Glick 2007), whereas agency comprises characteristics aimed at pursuing personal goals and manifesting skills and accomplishments (also referred to as competence or dominance). According to a functional interpretation of these classes of information, when an individual meets a new person, they want to know the person’s intentions—that is, whether they represent an opportunity or a threat (communion) and their capability—that is, whether they are able to put their intentions into action (agency). More recently, research has jettisoned the warmth dimension, distinguishing instead between sociability (e.g., friendliness and likeability) and morality (e.g., honesty and trustworthiness) and showing that morality plays a far greater role than sociability (and competence) in shaping evaluations of individuals and groups (see Brambilla et al. 2021 for a review). Considering the predominant role of morality in predicting behavioral responses (Brambilla et al. 2013), future studies could assess impression across multiple traits and examine potential differences among them in the ostracism reaction. Furthermore, in the present study, we reasoned that impression formation may serve a dual role, functioning both as a coping mechanism and as an evaluative response to norm-violating behavior. Future studies could help to disentangle these processes more clearly, for example, by directly assessing coping responses to exclusion in order to determine the extent to which negative impressions reflect psychological regulation vs immediate evaluative reactions.

Another promising direction for future research concerns the investigation of dehumanization, broadly defined as the denial of full humanness to others, often through the attribution of fewer uniquely human traits (Haslam 2006). Regarding ostracism, dehumanization has been hypothesized to serve a self-protective function, allowing individuals to reduce the emotional impact of negative social experiences (Bastian et al. 2013). In this sense, these findings align with our interpretation of impression formation as a functional coping strategy. At the same time, dehumanization may serve additional functions, such as legitimizing retaliatory behaviors against the source of exclusion (Bastian et al. 2013). Importantly, research on ostracism suggests that these processes are not limited to other-perceptions: victims may also engage in self-dehumanization, experiencing themselves as less human following exclusion (Bastian and Haslam 2010). Considering both self- and other-directed dehumanization could therefore provide a more comprehensive understanding of the divergent behavioral responses observed after ostracism. Future studies would benefit from integrating measures of self- and other-perceptions to clarify how these processes interact in shaping reactions to social exclusion.

Finally, it remains an open question whether antisocial responses generalize beyond the direct sources of exclusion. In our study, unrelated others were unaffected in the Dictator Game (Hp3c), consistent with the idea that participants targeted their reactions toward identifiable ostracizers. Yet prior work has documented displaced aggression against neutral targets (Twenge et al. 2001; Ren et al. 2016). This discrepancy suggests that contextual and individual factors—such as anonymity, salience of alternative targets, or opportunities for retaliation—may determine whether antisocial responses extend more broadly. Future investigations could explicitly manipulate these factors to clarify the conditions under which aggression spreads beyond the immediate perpetrators of exclusion.

Conclusion

To sum up, the present research revealed how ostracized people react to this unexpected negative experience and provided further evidence on an individual’s behavioral response to ostracism. We therefore provide initial evidence that the negative behavioral intention toward the sources are driven by a pattern of affective and cognitive consequences, thereby opening new and intriguing avenues for future research—for instance, on the interplay of affective and cognitive consequences to ostracism.