The study of helping behavior has attracted the scholars’ interest since the early works of Darley and Latanè in the 60s (Darley and Latanè, 1968) and has been recently transposed to virtual reality (VR; e.g., Slater et al., 2013). Indeed, classic phenomena such as the bystander effect (Kozlov and Johansen, 2010) or outgroup discrimination (Slater et al., 2013) in helping behavior have been successfully replicated in VR.

The reasons for using VR when studying the response to a help request are manifold. First, since information about facts or states might be inaccessible through introspection (Nosek et al., 2011) or affected by memory biases (Hyman and Loftus, 1998), self-reporting is increasingly replaced or complemented by behavioral measures (e.g., Monaro et al., 2018; D’Errico et al., 2020; Papapicco et al., 2021). VR allows the accurate recording of behavioral measures for subsequent inspection and analysis (Pan and Hamilton, 2018). Second, VR provides a perceptually vivid and responsive setting when a study in situ is not safe; indeed, these qualities motivate its use not only in the study of emergency behavior but also in its training (e.g., the commercial FLAIM Trainer^TM1 or the TEP platform by the United States navy²). Finally, a virtual environment can implement the experimental design of a study in a controlled yet affordable way, hardly achievable with a physical setting (Blascovich et al., 2002).

The studies carried out so far with or without VR have shown that situational, socio-cultural, and personal factors affect the decision to comply with a help request. Situational factors include concurrent tasks and time pressure (Darley and Batson, 1973), the presence of other potential helpers (Darley and Latanè, 1968), and the ambiguity and the seriousness of the emergency (Schwartz and Clausen, 1970; Clark and Word, 1974; Fischer et al., 2006; Lovreglio et al., 2015). Socio-cultural factors include the ethnicity of the victim (Dasgupta, 2004), the gender of the potential helper (Senneker and Hendrick, 1983), the gender of the bystander (Schwartz and Clausen, 1970), and the anonymity of the potential helper (Schwartz and Gottlieb, 1980).

One aspect that is usually not controlled in the literature on helping behavior is the relative spatial position of the potential helper and the help seeker (henceforth, the victim). We have examined the procedure of classic studies of helping behavior as well as the more recent studies using virtual reality; we found that the victim needing help and the potential helper are often in the same, delimited space and mutual sight (Darley and Latanè, 1968; Darley and Batson, 1973; Darley et al., 1973; Clark and Word, 1974; Gaertner, 1975; Gaertner and Dovidio, 1977; Senneker and Hendrick, 1983; Harari et al., 1985; Shotland and Heinold, 1985; Levine et al., 2005; Gillath et al., 2008; Kunstman and Plant, 2008; van den Bos et al., 2009; Slater et al., 2013; Zanon et al., 2014; Gamberini et al., 2015). Sometimes, they even share some danger or collaborate on the same task (Darley and Latanè, 1968; Gaertner, 1975; Gaertner and Dovidio, 1977; Senneker and Hendrick, 1983; Kunstman and Plant, 2008; Zanon et al., 2014).

We wondered whether letting the victim and potential helper share the same space might introduce a confound in the procedure. The reasons are offered by the social categorization framework (SC, Tajfel et al., 1979), showing that people tend to be more pro-social when the person in need is part of their ingroup (e.g., Dasgupta, 2004; Kunstman and Plant, 2008) and that the participants’ mutual position can split them into an ingroup and an outgroup (seatings’ spatial arrangement, Gaertner et al., 1993; neighborhood, Bernardo and Palma-Oliveira, 2016). Therefore, we hypothesized that people might be more pro-social when the victim is in the same space as the helper. The present study explores this hypothesis by manipulating the victim’s inclusion in the potential helper’s space and observing the help rate. Moreover, we wondered whether the effect of space—if any—would persist in front of a more blatant emergency; therefore, we varied the type of emergency, which could be either a vocal request for help or a request for help during a fire. The study follows a between-participant, 2 × 2 design with four experimental conditions (indoor without fire; outdoor without fire; indoor with fire; outdoor with fire). We hypothesized a higher help rate when the participants are indoor (i.e., in the same space as the victim) and that the effect of space persisted regardless of the type of emergency. With this study, we aim to contribute to the research area of pro-social behavior by highlighting a factor that can be worth methodological attention.

To examine the effect of the two main variables, we used a series of Pearson’s chi-square tests because the number of events per variable available in our study was not adequate for logistic regressions (van der Ploeg et al., 2014).

The help rates observed in the four experimental conditions of our study are displayed in Figure 2. An inspection of the figure suggests that the participants’ location during the help request consistently affected the help rates, which doubled when the participants were indoor. To test the statistical significance of these differences, we conducted a Pearson’s chi-square test comparing indoor (N = 32) and outdoor (N = 30) conditions; the test returned a statistical significance, χ2(1, N = 62) = 6.40, p = 0.01, confirmed after applying the Yates’s continuity correction, χ2(1, N = 62) = 5.18, p = 0.02. The effect size for this finding was moderate, φ = 0.32.

We then checked the effect of space separately from the effect of the type of emergency. We compared the conditions with fire, indoor (N = 15) and outdoor (N = 17); separately, we compared the conditions without fire, indoor (N = 14) and outdoor (N = 16). In both cases, the Pearson’s chi-square test returned a result close to significance, χ²(1, N = 32) = 3.03, p = 0.08, with a moderate effect size (φ = 0.31), and χ²(1, N = 30) = 3.45, p = 0.06, with a moderate effect size (φ = 0.34). The size of the sample of these two subsets might account for the failure in reaching full significance.

Regarding the effect of the type of emergency on help rates, the inspection of Figure 2 seems to suggest that the presence of fire reduced the help rate by a similar amount in all space conditions. However, the Pearson’s chi-square test returned no statistically significant difference in help rates between fire (N = 32) and no-fire (N = 30) conditions, χ2(1, N = 62) = 0.25, p = 0.62 (φ = −0.10). We also tested the effect of fire separately from the effect of space. We performed a Pearson’s chi-square test on the outdoor conditions with (N = 17) and without fire (N = 16), but the difference was not statistically significant, χ2(1, N = 33) = 0.24, p = 0.62 (φ = −0.09). Likewise, we compared the indoor conditions with (N = 15) and without fire (N = 16) using Fisher’s exact test because the expected frequency for the “no help” outcome in the no-fire group was less than 5 (Fe = 4.85). Again, the effect of fire was not statistically significant (p = 0.70, φ = −0.12).

A few controls were then run. We considered the effect on help rates of three variables, i.e., the participant’s gender (female or male), and expertise (non-players; infrequent players; frequent players), and the experimenter running the session (A or B). None showed a relationship with the help rates (Table 1), and the effect size tested with Cramer’s φ for gender and experimenter and Cramer’s V for expertise (df = 2) was small (φ < 0.30; V < 0.21).

Finally, we considered the anxiety scores across the four conditions: indoor with fire (N = 15, mean rank = 23.20), indoor without fire (N = 14, mean rank = 31.79), outdoor with fire (N = 17, mean rank = 34.12), and outdoor without fire (N = 16, mean rank = 36.25). A Shapiro–Wilk test revealed that the anxiety scores’ distribution was not normal in one condition (indoor without fire), W = 0.77, p = 0.002, violating the ANOVA normality assumption. Thus, we used a Kruskal–Wallis test to compare the anxiety scores between conditions; no significant differences were found, H(3) = 4.68, p = 0.20 (η² = 0.03). The post hoc test, performed with the “kruskalmc” function of the “pgirmess” package as proposed by Siegel and Castellan (1988), confirmed this result also with pairwise comparisons (Table 2). None of the observed differences was greater than the critical difference, and the effect size associated with the comparisons was small (Vargha and Delaney’s A < 0.56). Therefore, the possibility that the groups of participants assigned to the different conditions differed in their level of anxiety was discarded.

In conclusion, our hypothesis that sharing the space with the victim increased the help rate was confirmed on the whole sample and seemed unaffected by the type of emergency.

According to Latanè and Darley (1970), interpreting the situation as an emergency is necessary for providing help. Thus, a possible explanation for our results is that, in the indoor conditions, the participants recognized the emergency more easily. However, the help request and, when present, the fire effects were equally perceivable from any position, indoor or outdoor. Another possible explanation of our results could be that the route to the victim looked more difficult in the outdoor conditions, thereby increasing the perceived cost of providing help (Piliavin et al., 1981). However, we have no evidence that dexterity affected the participants’ decision to help. For instance, the participants’ expertise with games did not affect their decision to help. Also, they could practice the needed route in the previous phase of the experiment.

We think that the explanation that better fits our study is that the different spatial arrangements worked as a social categorization device; in other words, being inside the same building created a common ingroup identity for potential helper and victim (Gaertner et al., 1993; Dovidio et al., 2010), making the former more likely to act pro-socially. The effect of belonging to the same spatial formation on social identity is not explored directly in our study but is consistent with its findings and the tenets of social categorization theory, as explained in the introduction. The role of spatial boundaries in identifying an ingroup could also account for other reported phenomena, such as the sense of belonging to an environment and the related loyalty and affection to it (Kolesovs, 2021) or the pro-social behavior in people synchronizing their movements in space and time to a common rhythm (Cross et al., 2019).

In conclusion, our results have a methodological import, suggesting to control the victim and the helper’s mutual positions in studies of helping behavior. Moreover, they suggest that social identity can be implied in the effect of space on pro-social behavior, and this explanation is worth further investigation. Finally, by highlighting the role of space in pro-social behavior, our study has added one more reason why VR can be helpful to this line of research. VR allows to manipulate space easily, as testified by other lines of the psychology of space that have already adopted VR in their toolkit (e.g., spatial behavior, Durlach et al., 2000; spatial abilities, Dünser et al., 2006).

The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author/s.

Ethical review and approval was not required for the study on human participants in accordance with the local legislation and institutional requirements. The patients/participants provided their written informed consent to participate in this study.

AS and LG conceived and planned the work and the methodological approach. MMr developed the virtual environment. MF and MMs collected the data. MMs and PP analyzed the data. AS, LG, MF, and MMs wrote the manuscript. All authors contributed to the article and approved the submitted version.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.