In this study, we argue that similar to how an individual can experience a sense of ownership over a personal space (e.g., “This is MY space”), a sense of ownership can also manifest at the group level (e.g., “This is OUR space”) in VR. This proposition is anchored to the concept of Collective Psychological Ownership (CPO), which embodies the shared feeling of ownership that a group may hold over any material or immaterial target (Pierce and Jussila, 2010). Collective ownership feelings are fundamentally social in nature; they create boundaries between groups and organize the physical and symbolic world, as well as relationships and interactions between groups (Verkuyten and Martinović, 2017).

Group identity is a central aspect of CPO, as the concept collective ownership implies a shared sense of belonging to a group that has collectively invested into, has an intimate knowledge about, and controls the target of possession (Pierce et al., 2001; 2003; Pierce and Jussila, 2010). Grounded in the principles of the social identity approach (Tajfel and Turner, 1979; Turner et al., 1987), the sense of collective ownership needs to be underpinned by the social identity of the group (Pierce and Jussila, 2010) for the owned targets to become a part of the “extended self” of the owners (Dittmar, 1992). In intergroup research, CPO has primarily been applied to territorial ownership inferred from a group’s claims of primo-occupancy, investment, and formation (Martinović and Verkuyten, 2024). Extending this concept to VR interactions suggests that not only virtual objects but also co-constructed and shared spaces may foster a collective sense of ownership, thereby shaping intergroup dynamics.

Territorial CPO has often been linked with positive intragroup processes and negative intergroup outcomes (e.g., Martinović and Verkuyten, 2024). Specifically, CPO has been found to foster a sense of collective responsibility toward the target of ownership and encourage intragroup cooperation, solidarity, and stewardship behaviors, while also leading to defensive behaviors driven by perceived exclusive determination rights to control the target of possession (Pierce and Jussila, 2010; Nijs et al., 2022b). For example, CPO has been linked with the exclusion of certain social groups (e.g., immigrants and refugees) by the majority group members (Nijs et al., 2021). This is because ownership principles dictate what a group believes it can do with its possessions and, crucially, what non-owners are prohibited from doing with what is considered “ours” (Verkuyten and Martinović, 2017). Thus, in national contexts, especially when majority group members feel threats to their ownership, CPO may evoke a feeling of “losing what is ours” (Nijs et al., 2022a). As Nijs et al. (2022b) have empirically shown, this predicts exclusionary attitudes and behaviors (e.g., anti-immigration).

However, CPO can also support positive intergroup attitudes and outcomes, pending it is perceived as inclusive, i.e., shared among the larger superordinate ingroup. Aligned with the Common Ingroup Identity model (Gaertner et al., 1993), perceiving an outgroup as part (i.e., sub-group) of a larger “us”, particularly through positive contact, may elevate the sense of shared ownership and social identity, reducing prejudice. Thus, CPO holds promising potential to be explored as an important VR-specific mechanism to achieve positive intergroup outcomes. Interaction with other avatars in an immersive virtual space may promote the development of a collective sense of ownership over the virtual space shared among virtual agents, including those representing an outgroup based on other social characteristics (e.g., skin color). This, in turn, may enhance the feelings of unity, cooperation, and solidarity with stigmatized minority groups following such experiences. Moreover, investigating CPO as a potential facilitator of intergroup contact in VR offers valuable insights into how ownership-related processes can be leveraged to improve intergroup relations.

In the context of VR, the concept of ownership has been studied at the level of individual psychological ownership (IPO), namely, an individual’s feeling of possession over objects in the virtual space (e.g., Lee and Chen, 2011). Studies indicate that virtual targets can elicit ownership feelings much like physical ones (e.g., Buchem, 2012; Poretski et al., 2019), also in socially situated interactions (Lee and Chen, 2011; Poretski et al., 2021). A strong sense of ownership has been found to be important for motivating a person to spend time, engage actively, and return to a virtual space (Lee and Chen, 2011).

However, to the best of our knowledge, collective feelings of possession in VR in general and of a virtual space in particular have not yet been explored within VR. The most relevant research for this study comes from studies on shared Augmented Reality (AR). Carrozzi et al. (2019), for example, showed that shared experiences such as using shared devices, satisfying social identity needs, and enabling customization can foster collective psychological ownership of AR holograms.

Lee et al. (2024) showed that technological and social factors in the metaverse can elevate psychological ownership from the individual to the collective level. Although they did not isolate VR-specific effects, the metaverse includes multiple virtual environments, including social VR. Their findings indicated that platform controllability and interactivity, together with social factors (e.g., social support, social self-efficacy, and community support), shape social presence and social identity within metaverse communities; these determinants closely map onto core features of many VR experiences. Greater social presence and identity were, in turn, associated with higher engagement and stronger collective psychological ownership (Lee et al., 2024). Taken together, the combination of control, customization, and interactivity with the inherently social nature of many VR settings suggests that collective psychological ownership is likely to emerge in social VR experiences in which participants can move freely and interact with others.

Poretski et al. (2021) investigated IPO in AR and its impact on interpersonal tensions and ownership claims when ownership was made unclear. Their research emphasized that ownership conflicts arise when multiple users interact with the same virtual and physical objects or spaces, especially when these objects are modified. They identified five key dimensions of psychological ownership: possession, control, identity, responsibility, and territoriality, as critical to understanding how individuals in shared virtual spaces perceive their relationship to both physical and virtual objects. While these dimensions pertain to IPO, they may contribute to conceptual understanding of how shared environments might foster or disrupt a collective sense of ownership in VR, especially when it comes to collective settings where shared environments necessitate collaboration.

Building on Poretski et al. (2021) work, Seo (2024) highlighted the interplay between IPO and CPO in shared AR workspaces, emphasizing how collaborative functionalities influence these dynamics. In Seo’s (2024) study, exclusive control fostered IPO by enhancing personal agency and attachment, while shared control facilitated CPO by fostering mutual investment and collective responsibility. These mechanisms are directly relevant to VR, where intergroup interactions require balancing individual agency with shared ownership to achieve a sense of “ours.”

In conclusion, this study advances emerging work on (collective) psychological ownership in VR and argues that, alongside established VR-specific mechanisms such as co-presence and body ownership, CPO-VS is a key construct to examine when evaluating the prejudice-reduction potential of intergroup contact in VR.

The aim of this research is to expand the focus of VR studies to better acknowledge the role of CPO as an additional VR specific mechanism that emerges during virtual collective experiences and may contribute to attitudinal change following intergroup contact in VR. Specifically, we build our study on previous CPO research focusing on territorial ownership disputes in the real-life intergroup contexts (for a review, see Martinović and Verkuyten, 2024) as well as on existing studies on psychological ownership in immersive contexts (Lee et al., 2024; Poretski et al., 2021; Seo, 2024). We suggest that while feelings of co-presence imply being in the same virtual place together, CPO-VS implies perceiving shared ownership of the virtual place (“This virtual space is ours!”). Specifically, we examine whether even a brief interactive task in VR is sufficient to elicit CPO-VS, and whether CPO-VS is associated with and facilitates the effect of positive intergroup contact in VR on outgroup attitudes following the VR experience. This inquiry is critical for understanding the accessibility and scalability of CPO-VS as a mechanism in VR-based interventions.

This study is a part of a larger intervention study conducted in Italy and Finland that aimed at testing the direct and secondary transfer effects of positive and negative virtual intergroup contact on outgroup attitudes. The study was pre-registered prior to data collection. The pre-registration can be found at https://osf.io/dgqj9/overview. As only the Italian data set included the measure of CPO-VS, the pre-registered hypothesis concerning the potential effect of this factor has not been analysed nor reported in our previous studies on the intervention effects (Tassinari et al., 2024).

We formulate our exploratory hypotheses as follows:

H1

Higher sense of CPO-VS is associated with less prejudice towards the outgroup following positive intergroup contact in VR.

H2

Among participants exposed to positive intergroup contact in VR, greater CPO-VS is associated with reduced prejudice toward the outgroup.

We control for the sense of co-presence and body ownership in our analyses, as they are key components that may significantly influence the VR experience (Tassinari et al., 2024), as well as demographics (age and gender) associated with prejudice in previous research.

Attitudes are a manifold construct, encompassing both deliberate components and more automatic processes. To account for this complexity, our study incorporates a combination of measures that assess different dimensions of attitudes. Specifically, we include two explicit measures: a semantic differential scale (the General Evaluation Scale by Wright et al., 1997), and a visual analog scale (the feeling thermometer). To account for the automatic, less deliberate aspects of attitudes, we also employ the Implicit Association Test (IAT, Greenwald et al., 1998), which measures implicit attitudes. While we recognize that the IAT has limitations, particularly in terms of construct validity (Schimmack, 2021), it provides a critical advantage in bypassing social desirability bias–a common issue with self-report measures. Additionally, the IAT has been shown to have stronger predictive validity for discriminating behavior compared to explicit measures (Greenwald et al., 2009), making it a valuable tool for a comprehensive assessment of attitudes. By integrating these methods, we aim to provide a well-rounded analysis of how explicit and implicit processes shape intergroup attitudes in the context of this study.

Participants were recruited via social media outreach and snowball sampling, excluding those with educational or psychological academic backgrounds to avoid identification of the purpose of the study. Of the initial 160 participants, 28 were excluded based on pre-registered criteria (e.g., failing to identify avatar ethnicity, mixed ethnic backgrounds, VR issues, or IAT completed too fast). No additional exclusion criteria were applied. The final sample included 132 participants: 66 males, 65 females, and one identifying as “other,” aged 18–88 years with mean age 31 (SD = 14.25).

The experiment was conducted at a university located in Northern Italy. A confederate, who was a White Italian female, guided participants through the process. In the laboratory, participants were first instructed on how to properly wear and operate the Oculus headset. Following this, they completed the Altspace VR tutorial to familiarize themselves with the virtual environment and practice navigating their avatars. Participants were then asked to create avatars that closely resembled their own appearance. This step was designed to enhance their sense of body ownership over their avatars and to encourage the belief that other participants they encountered in VR had also modeled their avatars to resemble themselves. The avatars in the virtual environment were gender-matched to the participants, except for the experimenter’s avatar, which was represented as a White female.

Participants were then transported to the virtual room depicted in Figure 1. The experimenter was present in the virtual space, embodied by the avatar shown in Figure 2. Participants were randomly assigned to four conditions: intergroup cooperation, intragroup cooperation, intergroup competition, and intragroup competition. They received instructions to play a virtual ball-toss game, either cooperating or competing with another participant, who was actually a confederate. They were told to score points by throwing bean bags through a hole on a board, taking 10 alternating turns each. The game’s framing emphasized team alignment (same or different team) and was designed to simulate intergroup or outgroup interactions. Participants began the game on cue, with the confederate’s skill level matched to theirs to maintain a sense of fairness. After receiving these instructions, participants and the experimenter were joined by a third avatar, representing a teammate controlled remotely by a confederate. The teammate’s avatar represented a Black person in the intergroup contact condition, and a White person in the ingroup contact condition.

Considering the limited sample size and our focus on the effect of specifically positive (i.e., cooperative) intergroup contact on attitudes towards a specific outgroup (i.e., people of African descent), in this study, we restricted our main analyses to contrasting the effect of positive intergroup contact with an avatar representing a Black person (experimental condition, n = 36) with that to all other assessed conditions (n = 96).

Once the game concluded, participants moved to a virtual waiting room where they encountered an avatar representing an East Asian individual. No interaction occurred during this time, and the seating distance chosen by the participant was recorded. After 3 min, participants were informed that their team had won the game, regardless of their assigned condition, marking the end of the virtual experience. Finally, participants completed a self-report questionnaire and an implicit association test through the online platform PsyToolkit (Stoet, 2017). Afterward, they were thanked, debriefed, and dismissed.

We assessed outgroup attitudes using three distinct measures, each capturing different facets of prejudice: the General Evaluation Scale (GES), the feeling thermometer, and the Implicit Association Test (IAT).

General Evaluation Scale (GES) was used to assess explicit attitudes towards the individuals of African ethnic background using six bipolar items (Wright et al., 1997). Participants rated outgroup members described with pairs of adjectives (e.g., “cold/warm,” “suspicious/trusting,” “positive/negative,” “friendly/hostile,” “respect/contempt,” and “admiration/disgust”) on a 7-point scale, with higher scores indicating more negative attitudes toward people with African ethnic background. Responses were averaged into a composite score, demonstrating strong reliability (α = 0.86).

The feeling thermometer was used to assess emotional prejudice. Participants were asked to rate their feelings toward individuals of African ethnic background on a scale from 0 (unfavorable) to 100 (favorable).

Implicit attitudes were assessed using the Implicit Association Test (IAT; Greenwald et al., 1998). The IAT is a reaction-time-based task designed to measure automatic biases by assessing the ease with which participants associate Black or White faces with positive or negative words. Faster reaction times when associating Black faces with negative words and White faces with positive words indicate stronger implicit biases against individuals of African ethnic background. A higher score indicates higher implicit bias.

Collective Psychological Ownership of a Virtual Space (CPO-VS), adapted from the concept of CPO by Pierce et al. (2018), assesses participants’ sense of shared ownership over a virtual space while embodying an avatar. The scale, comprising four items with a Cronbach alpha of 0.908, evaluates the extent to which individuals perceive the space as collectively owned by their and others’ avatars. Items were formulated as follows: “When I think about the virtual space and my experience in that space as an avatar, I feel that: the space was owned by us, the avatars; the space belonged to us, the avatars; the space was in our collective possession; we owned the space together.” Responses were rated on a 7-point scale, with a higher score being indicative of a higher feeling of virtual space ownership. A one-factor CFA for the measure was estimated with WLSMV using the four observed items (N = 132). Factor loadings were high (0.85–0.91), and reliability was excellent (ω ≈ 0.94).

Co-presence in VR was assessed using a four-item scale adapted from Biocca and Harms (2002). The scale measured participants’ awareness of others’ presence in the virtual environment. Responses were rated on a 1–5 scale, with a higher score indicating higher sense of co-presence. This measure had a Cronbach alpha of 0.83.

Body ownership was assessed utilising a six-item scale developed by Peck et al. (2013), which evaluates the extent to which participants experienced the virtual body they controlled in VR as their own. Responses were rated on a 1–5 scale, with a higher score indicating higher sense of body ownership. The Cronbach alpha for this scale was 0.78.

To test the hypotheses, we used moderated multiple regression analysis. Assumptions of linear regression for all the models were met except for normal distribution of residuals, confirmed by the Shapiro-Wilk normality test (p < 0.05). However, as our sample size (N = 132) is large enough for relaxing the issues of non-normality (Pek et al., 2018), we proceeded with the linear regression models.

The first step of the linear model included the experimental condition (0 = no positive intergroup contact vs. 1 = positive intergroup contact) variable, and in the second step, CPO-VS was added. The third step included the interaction between the experimental condition and CPO-VS. In the fourth and final step, the control variables (i.e., co-presence, body ownership, age, and gender) were added. This model was repeated with three dependent variables: GES, feeling thermometer and IAT. The multicollinearity of the included correlated predictors was assessed using the Variance Inflation Factor (VIF), which ranged from approximately 1.08 to 1.21, indicating no concerns.

Analyses were performed using Rstudio (R Core Team, 2024). R-packages used in the analysis included apaTables (Stanley, 2021), car (Fox and Weisberg, 2019), gt (Iannone et al., 2024), lmtest (Zeileis and Hothorn, 2002), moments (Komsta and Novomestky, 2022), officer (Gohel and Skintzos, 2024), psych (Revelle, 2024), readxl (Wickham and Bryan, 2023), sjPlot (Lüdecke, 2024), table1 (Rich, 2023), and tidyverse (Wickham et al., 2019).

Regarding the VR-specific variables, participants generally showed moderate degrees of feelings of co-presence, body ownership, and CPO-VS across conditions. We next conducted independent sample t-tests to compare participants’ feelings of body ownership, co-presence, and CPO-VS between the two conditions. Given the disparity in sample sizes, Welch’s t-test was utilized to account for unequal variances. The results revealed a statistically significant difference in body ownership scores between the conditions, with the participants in the experimental condition (M = 2.95, SD = 1.02) reporting lower body ownership compared to participants in the control condition (M = 3.46, SD = 0.91); t (57.00) = 2.63, p = 0.011. The mean difference was 0.51, with a 95% confidence interval of [0.12, 0.89], indicating moderately lower levels of body ownership in the positive intergroup contact group. There were no significant differences for co-presence (t (57.65) = 0.74, p = 0.461) or CPO-VS (t (63.55) = −0.24, p = 0.814) between the conditions studied.

Next, we examined the correlations between the study’s predictor and dependent variables. Body ownership, co-presence, and CPO-VS were all significantly and negatively correlated with explicit attitudes measured by GES, and body ownership also showed a significant negative correlation with IAT (see Table 1).

The results of the step-by-step moderated multiple linear analyses, examining the direct and interaction effects of condition (cooperative intergroup contact vs. other conditions) and CPO-VS on outgroup attitudes, are presented in Tables 2–4 for each dependent variable (GES, Feeling Thermometer, and IAT) separately. The final step of the models also included control variables specific to the VR experience (co-presence and body ownership), which are associated with successful VR interventions and reduced prejudice, as well as demographic factors (age and gender). Three post hoc power analyses were conducted in G*Power (F tests; linear multiple regression: R² deviation from zero) for the final regression model including six predictors (four control variables and two predictors). Using the observed effect sizes (f² _GES = 0.129, f² _therm = 0.167, and f² _IAT = 0.112), with α = 0.05 and N = 132, the achieved power for detecting a non-zero overall R² were 0.87 for the GES, 0.95 for feeling thermometer, and 0.81 for the IAT.

The results partly supported our H1, as CPO-VS directly predicted lower explicit prejudice towards people of African ethnic background measured through GES (β = −0.23; p = 0.008). This effect lasted, when the interaction effect between the experimental condition and CPO-VS was added into the model in step 3 (β = −0.25; p = 0.012), but lost statistical significance after the control variables were added in step 4 (β = −0.15; p = 0.164).

We also hypothesised that the relationship between cooperative intergroup contact and outgroup attitudes is moderated by CPO-VS, such that higher levels of CPO-VS amplify the positive effect of intergroup contact on outgroup attitudes (H2) ¹ . The results showed no significant interaction between condition and CPO-VS on outgroup attitudes. However, consistent with Tassinari et al. (2024) study, there was a direct effect of experimental condition on IAT, with cooperative intergroup contact significantly decreasing the level of implicit prejudice (β = −0.40; p = 0.048) after controlling for co-presence, body ownership, age, and gender. However, the effects of the experimental condition on explicit attitude measures (GES and Feeling Thermometer) remained non-significant.

As regards the control variables, body ownership showed a significant negative effect on the IAT (β = −0.24; p = 0.017), and co-presence on GES (β = −0.19; p = 0.042), suggesting that these variables contribute to the decreasing prejudice following social interaction in VR. Out of the demographic controls, higher age was associated with more negative feelings toward the outgroup (Feeling thermometer: β = −0.34; p < 0.001).