A little past midnight of Midsummer, 21 June 2016, Muhammad Rafat Badran, a 15-year-old Palestinian was traveling by car with his cousins in the Central West Bank. The boys were returning from an outing to a swimming pool in the village of Beit Sira, not far from Ramallah, when their car was mistakenly identified by the Israeli Defense Forces (IDF) to be involved in throwing rocks and firebombs. The soldiers shot multiple times at the suspect vehicle: Muhammad was killed immediately, and his cousins were wounded (Harel et al., 2016). This incident is but one example of how uninvolved civilian casualties can result from errors and misjudgments, with tragic results.

Civilian casualties often occur during severe intergroup conflicts, despite rules of conduct such of the International Humanitarian Law that are designed to minimize them. While it is difficult to gauge actual numbers of civilian casualties (e.g., Ryan, 2018; Crawford and Lutz, 2019), evidence has accumulated that collateral damage—unintentional or incidental injury or damage to persons or objects that would not be lawful military targets in the circumstances ruling at the time (U.S. Joint Chiefs of Staff, 2013)—can have a deleterious impact on intergroup conflicts beyond the obvious human suffering and financial burden. The incidental death of non-combatants can contribute to the perpetuation and even escalation of conflict through increased radicalization and more negative attitudes toward the rival outgroup (e.g., Condra and Shapiro, 2012; Deri, 2012; Lyall et al., 2013; Shaver and Shapiro, 2015; Farooq et al., 2020; though see Shah, 2018).

Existing research focused on public attitudes toward enemy civilian casualties, yielding conflict results: Some opinion polls and lab experiments revealed an aversion to civilian casualties (e.g., Friedrich and Dood, 2009; Kreps, 2014; Pew Research Center, 2015, 2017; Walsh, 2015; Johns and Davies, 2019), whereas others found a more lenient view on the issue (Larson and Savych, 2007; Gallup, 2011; Sagan and Valentino, 2017; Carpenter and Montgomery, 2020; Slovic et al., 2020). In the present research, we explore a different angle of ingroup members’ answer to the question of potential outgroup civilian casualties. We go beyond mere attitudes regarding collateral casualties or support for macro-level policies and focus on group members’ decisions in situations, in which they would be required to make an immediate, life-or-death response that might result in the incidental killing of uninvolved civilians. In situations of violent conflicts that involve terror attacks on civilians, such occurrences can become real all too often. We focus on two elements that may affect the decision-making process and reduce tolerance of collateral casualties: accountability and forecast group-based moral emotions, while exploring possible connection between them.

Accountability refers to an implicit or explicit expectation that one may be called upon to justify one’s beliefs, feelings, and actions to others (Scott and Lyman, 1968; Semin and Manstead, 1983; Tetlock, 1992). Accountability can be conceptualized as a primarily social phenomenon in which individuals seek to maintain prestige and avoid “losing face” to any potential observers of their actions. It usually implies that individuals who do not provide an adequate justification for their actions will suffer negative consequences, ranging from scornful stares to loss of livelihood, liberty, or even life (Stenning, 1995). Conversely, people who provide satisfactory justifications will experience positive consequences, ranging from mitigation of punishment to lavish rewards, like a prestigious position in political office (Lerner and Tetlock, 1999). This conception of accountability expands its relevance to nearly every action or belief of an individual that could be witnessed or observed by others, especially on socially or politically relevant issues (Wayne et al., 2016).

The effect of accountability on decision-making has been tested in multiple fields, including medicine (Han et al., 2009), business negotiations (Lerner and Shonk, 2006), tax audits (Buchman et al., 1996), postwar aid (Skitka et al., 1991), education (e.g., Miller, 1995; Fuhrman and Elmore, 2004; Burke, 2005), the criminal justice system (Stenning, 1995), and representative government (e.g., Przeworski et al., 1999; Behn, 2001; Grant and Keohane, 2005). Accountability that meets certain criteria can promote thoughtful and careful consideration of the merits of a specific attitude or preference (Lerner and Tetlock, 1999). This form of accountability, which leads to “preemptive self-criticism” (Tetlock, 1983, p. 81), may therefore induce people to choose a more cautious course of action when confronted with potential aggressors, thus reducing the risk of civilian casualties. Thus, we expect that when individuals experience a sense of accountability, they would be less tolerant of enemy collateral casualties compared with situations, in which no perception of accountability is evoked.

The term “preemptive self-criticism” can also depict another factor that may impact ethical decision-making: forecast group-based moral emotions. Moral emotions influence the link between moral standards and moral behavior, driving people to behave in moral, socially appropriate ways in their social interactions and intimate relationships (Retzinger, 1987; Baumeister et al., 1994; Leith and Baumeister, 1998). Self-reflection evokes emotions that provide immediate reinforcement, either positive or negative, of behavior—not only behavior that already took place, but also planned behavior and its expected outcomes (Tangney et al., 2007). By providing immediate and salient feedback on our social and moral acceptability, moral emotions are a significant force in regulating behavior.

In the context of intergroup relations and intergroup conflict, shame and guilt are arguably some of the most relevant moral emotions. Group-based guilt is associated with appraised responsibility of one’s ingroup for moral violations (Branscombe, 2004); it is evoked when one feels personally or collectively complicit in other group members’ transgressions (Lickel et al., 2011), or actions that are seen as illegitimate (Branscombe et al., 2002) and can motivate group members to rectify the wrongdoing and make reparations to the victims (e.g., Doosje et al., 1998; Iyer et al., 2003; Čehajić et al., 2011). Group-based shame, on the other hand, is associated with appraisals implying that a wrongdoing tarnishes the moral image of the group (Lickel et al., 2011). Group-based shame was shown to induce a desire to distance the ingroup from the shame-invoking situation (Lickel et al., 2004; Iyer et al., 2007), yet there are inconsistent findings regarding its association with pro-social motivations (Rees et al., 2013; Berndsen and Gausel, 2015) and pro-reconciliation actions, such as expressing contrition over wrongdoings and compensating the outgroup (Brown and Čehajić, 2008; Allpress et al., 2010; Gausel et al., 2012).

The majority of research on group-based shame and guilt focuses on the emotions evoked either by past transgressions or present moral violations. However, self-conscious emotions can provide feedback regarding future actions, and whether those actions comply with one’s moral standards and values (Tangney et al., 2007). Forecast group-based moral emotions—that is, expected shame and guilt over future wrongdoings—can also impact on group members’ future actions. This notion received empirical support from a series of studies conducted by Shepherd et al. (2013a), in which expected group-based shame was associated with less ingroup favoritism, more egalitarian intergroup behavior (Shepherd et al., 2013b), and increased support for collective action and willingness to make reparations (Shepherd et al., 2013c). We therefore expect that forecasting higher levels of group-based guilt and shame over expected collateral casualties would be associated with lower support for actions that would lead to such casualties.

Furthermore, shame is driven by the damaged reputation or loss of respect and honor in the eyes of others (Crozier, 1998; Mosquera et al., 2002; Smith et al., 2002), which resonates with the concept of accountability. This suggests that forecast moral emotions and accountability may operate in tandem to impact the decision whether to take an action that may result in the death of outgroup civilians or to refrain from action at the possible cost of ingroup lives. Shame, in particular, can be construed as a form of internalized accountability system. We therefore expected that forecast group-based moral emotions would moderate the effect of the accountability manipulation. One possible outcome was additive: The most cautious decisions would be made by individuals who expected to feel high levels of guilt and shame over outgroup civilian casualties and were exposed to an accountability manipulation. Another potential outcome is derived from the notion that shame, in particular, can be construed as a form of internalized accountability system; therefore, it is possible that individuals who already forecast high levels of shame would be less affected by accountability manipulations. Both possible directions of the moderation were explored. Finally, we tested the possibility that forecast group-based guilt would also moderate the effect of the accountability manipulation, either making participants even more reluctant to put outgroup civilians at risk when primed with accountability, or making the manipulation redundant.

To test our hypotheses, we conducted two studies in the context of the Israeli–Palestinian conflict.¹ In Study 1, which was conducted in the lab, Jewish–Israeli students were first asked to forecast their level of group-based emotions in case of incidental deaths of enemy civilians, then underwent a traditional accountability manipulation, and responded to a set of vignettes presenting potential attackers/bystanders. In Study 2, Israeli Defense Force (IDF) soldiers and officers were recruited to respond to a modified series of vignettes adapted to their role as likely first responders on the scene; accountability was primed by requesting participants to describe military debriefings they had undergone during their military service. We expected that high levels of forecast moral emotions would lead to more cautious responses that decrease the chance of incidental civilian casualties. We also expected that priming a sense of accountability would lead to more cautious decision-making compared with no intervention. Finally, we examined the possibility that the accountability manipulation would have a different effect on participants with lower levels of forecast moral emotions than on those with higher levels of forecast moral emotions, which represent internalized accountability.

Study 1 was conducted at the height of the “Knife Intifada,” often referred to as “Lone Wolves Intifada” in the Israeli media, while Palestinian attacks against Jewish–Israeli civilians and security forces were almost daily occurrences. This period of conflict escalation began in September, 2015 and lasted several months, in which dozens of Palestinian attacks were attempted against Jewish–Israeli soldiers and civilians. According to Israeli reports, between October, 2015 and October 2016 there was a total of 166 stabbing attacks and 89 attempted stabbings; 108 shootings; 47 vehicular (ramming) attacks; and one vehicle bus bombing (Israel Ministry of Foreign Affairs, 2018). The attacks occurred in numerous towns and settlements in Israel and were directed at both civilians and soldiers, thus exposing much of the population to the threat of attack. Most Palestinian “lone wolves” were young (their median age was 20) and who did not seem to be backed by a broad Palestinian consensus, nor were the Palestinian political organizations much involved in the uprising (Chorev, 2019). During this period, nearly 50% of Palestinian attackers were killed; in some incidents, Israeli soldiers and members of other security forces allegedly used excessive force to “neutralize” attackers or suspect attackers (B’Tselem, 2015). This combination of circumstances made salient both the sense of imminent threat and the potential mistaken response that could cost civilian lives.

Against this backdrop we conducted Study 1. The goal of the study was to examine the effects of accountability and forecast moral emotions on the decision whether or not to choose an action that may lead to enemy civilian casualties. This decision was operationalized by presenting Jewish–Israeli participants with ambiguous vignettes describing possible terror attacks and requesting them to decide whether the potential attacker in each vignette should be shot.

Based on the results of Study 1, we developed more nuanced hypotheses regarding the influence of accountability and forecast group-based emotions on tolerance of enemy collateral casualties, while taking into account the difference between the two moral emotions. Specifically, we expected that shame, but not guilt, would moderate the effect of the accountability manipulation on our outcome measures, so that those with lower levels of forecast shame would make more cautious decisions (i.e., fewer “shoot” choices) compared with those in the control group. We tested these hypotheses among active duty soldiers, who are more likely than the civilian sample of Study 1 to encounter the dilemma of possible collateral casualties, while trying to stop a terror attack. This was particularly true for our respondents as many of the, 2015–2016 terror attacks in Israel involved IDF soldiers—as targets, as those on scene and trained to stop an attack, or both. In Study 2, we evoked a sense of accountability by asking respondents to recall military debriefings or post-action reflective analyses.

The process of debriefing is an important part of military learning after missions. In “after-action review” practiced in the US Army (2011) participants talk through what happened in a military incident. In the IDF, soldiers and officers sit down in groups after battle to reconstruct what happened, a process in which the soldiers’ thoughts and feelings are considered part of the reality of combat (Shalev et al., 1998). When applied to—and tested in—a civilian context (surgical teams in a large hospital), debriefings were found to foster accountability norms (Vashdi et al., 2007). We therefore assumed that reminding participants who are members of the IDF of situations, in which they were debriefed would put them in a mindset of explaining and justifying their actions—in other words, make them feel more accountable.

In two studies conducted among Jewish–Israeli civilians and IDF soldiers, we examined the effects of accountability, forecast group-based moral emotions, and the interaction between them on tolerance of enemy collateral casualties. The results, we obtained suggest an interaction between forecast shame and accountability on tolerance of enemy civilian casualties. When primed with accountability, group members who did not expect to feel much shame over the incidental killing of uninvolved enemy civilians were more likely to choose not to “shoot” suspects in ambiguous situations. In contrast, group members who forecast high levels of shame were unaffected by experimental manipulations of accountability. Group-based guilt did not predict TECC, nor did its interaction with the accountability prime.

The results of the studies provide some support for our hypotheses, but it is important to acknowledge their limitations. The samples are relatively small, and although the pattern of interaction between accountability and forecast group-based shame was similar in both studies, the interaction was found significant in Study 1 and only marginally significant in Study 2. Thus, while these results are suggestive of an interesting direction, future research in larger samples is needed to replicate and confirm them. Despite these limitations, we offer some thoughts on what our findings suggest.

The findings of the two studies presented are in line with the notion that in some situations shame can be an adaptive emotion. Often construed as a highly aversive emotion that entails an appraisal of the self as fundamentally flawed (Lewis, 1971), much of the previous literature associated shame with a tendency to avoid failure and its consequences for oneself and others (Gilbert and Andrews, 1998; Tangney and Dearing, 2002; Tracy and Robins, 2004). However, more recent evidence shows that shame can also be linked to constructive approach orientation (e.g., Gausel et al., 2018; Mashuri and van Leeuwen, 2020), particularly in situations in which one’s failure, or one’s social image, seem more reparable (Leach and Cidam, 2015). When the failure or social image appear difficult to repair, suggest Leach and Cidam (2015, p. 984) in their meta-analysis, “What better response can there be to shame […] than to lessen one’s approach of it?” Killing uninvolved civilians is an irreparable act; thus, forecast shame over such detrimental future action can lead to avoiding the undesirable outcomes altogether by choosing the more cautious, and less deadly, option.

Yet, we found that forecast shame tended to have a moderating effect on tolerance of enemy collateral casualties rather than a main effect. As suggested earlier, this consistent relationship indicates that individuals who expect to feel high levels of shame are unaffected by the accountability manipulation, possibly because they have already internalized the external agent or judge to whom they need to justify their actions. Therefore, they tend to choose the more cautious option, regardless of whether they are primed to think of accountability or not. For those who forecast little shame over their group’s actions, accountability mimics the effect of shame by inducing them to make more choices that would not result in the outcome that would taint their group’s social image.

The interpretation of shame as internalized accountability also explains, at least in part, why accountability did not have a main effect on participants’ decisions in Study 2. For respondents who already experienced “preemptive self-criticism” (Tetlock, 1983) in the form of forecast shame, the priming of accountability was redundant and did not impact their decisions. Another possible explanation is that the subtle priming manipulation of Study 2 was not sufficient to produce a main effect, though it resulted in a conditional effect. This also suggests that accountability has the potential to be a useful tool to induce thoughtful and conscientious decision-making process in challenging and complex situations within intergroup conflict. Future research that further examines the intersection between moral emotions and accountability in this context, both among decision-makers and members of the public, could shed additional light on the underlying mechanisms and assist in developing accountability-based interventions.

While the moderated effect of accountability and its potential benefits are rooted in previous findings, the finding that forecast guilt did not affect tolerance of enemy collateral casualties is more puzzling. Why did guilt, long considered to be the more adaptive of the two moral emotions (e.g., Dearing et al., 2005; Tangney et al., 2007; Stuewig et al., 2010), have no effect on group members’ choice of action? Our findings are consistent with research that demonstrated that shame, but not guilt, affected outcomes such as support for collective action (Shepherd et al., 2013c), reduced ingroup favoritism (Shepherd et al., 2013a), and reconciliation orientation in violent reciprocal conflicts (Gausel et al., 2018). Moreover, it is possible that in the context of the decisions our participants were presented with, the expectation of guilt was mostly detached from the actual action taken. In other words, group members care about their ingroup’s image in this context (i.e., their expectation of shame) rather than about the actual moral aspects of inadvertently killing uninvolved enemy civilians (i.e., forecast guilt).

Finally, let us consider the manner in which tolerance of enemy collateral casualties was assessed. This concept is more traditionally assessed by directly asking participants to indicate the number of uninvolved civilians they would deem acceptable “collateral casualties” (e.g., Sagan and Valentino, 2017; Schori-Eyal et al., 2019), a measure that may be susceptible to social desirability (Krumpal, 2013; though see Carpenter and Montgomery, 2020). Our approach tackled the issue simultaneously more directly and more circuitously, by asking respondents about concrete, less hypothetical scenarios which are closer to their own reality. The tool we devised was uniquely tailored to the situation in which ingroup civilians—not necessarily members of the armed forces—found themselves likely to address such decisions. We contend, however, that the merits of applying SDT-based measures of TECC goes beyond the specific situation and can be usefully adapted to other contexts.

The present research provides some evidence on forecast moral emotions, accountability, and their potential impact on decision-making in the context of enemy collateral casualties. Future research that replicates these findings, examines the underlying mechanisms in greater detail and expands the scope of the research to additional contexts, while addressing the limitations of the current set of studies, would support our suggestion that these psychological processes can act in tandem to induce more moral decision-making in intergroup conflicts.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by IRB, Interdisciplinary Center Herzliya. The patients/participants provided their written informed consent to participate in this study.

NS-E designed the research, collected, analyzed, and interpreted the data, and drafted and revised the manuscript. DS-S designed the research, collected and interpreted the data, and revised and approved the manuscript. ES took part in designing the research and the research tools, analyzed and interpreted the data, and revised and approved the manuscript. EH supervised the designing of the research, interpreted the data, and revised and approved the manuscript. All authors contributed to the article and approved the submitted version.

This work was supported by the European Research Council (grant number 335607) to EH and the Israel Science Foundation (grant number 1585-2016) to NS-E.

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.