1 Introduction

People constantly choose sides. Side-taking can be based on prior commitments and alignments, such as viewing oneself as Democrat or Republican or as a loyal friend or spouse regardless of circumstances. However, in a wide range of cases people take sides spontaneously where they have no previous commitment to either side. We call this phenomenon spontaneous side-taking (SST) and suggest SST can occur when people observe a disagreement or conflict. While SST may lead to intervention, we argue SST also occurs when no direct intervention is possible, such as in observing films, fiction, legal trials, sports, advertisements, and debates. We propose that SST, though quick, can lead to long-term side-taking in the form of choice reinforcement, polarization, and coalition formation. These phenomena can lead people to assume identities that orient actions in social conflicts. Although there is much literature on side-taking in situations where allegiances already exist, there is little work on what initially drives side-taking. In this exploratory study, we aim to address this gap in the literature by examining the possible mechanisms of SST, particularly in regard to general empathy, attention, and memory for characters in conflict.

To investigate the mechanisms of SST, we rely on narratives about ongoing or escalating conflict between two characters. Narratives provide an excellent opportunity to study the effects of side-taking because events can be manipulated and reactions to events can be recorded. Moreover, narratives provide a natural means for readers to track the relatability, understanding, morality, and likeability of characters. Finally, using victim narratives provides a powerful cue for spontaneously taking the side of one character over another. Our use of narratives provides not only a method for the study of side-taking generally but also a novel avenue for investigating SST at its onset. We propose SST as a potential framework to observe the emergence and development of preferences between parties, polarization, and conflicts. We suggest that SST manifests in the emergence of lopsided attention for one side and not the other, as well as increasing markers of sustained empathy with one side.

We hypothesized people within a conflict will spontaneously take the side that they perceive to be victimized and consequently more deserving of an ally. We also conjectured that participants would (1) encode more information about the victim, (2) take their perspective, and (3) be unlikely to switch sides after the initial side-taking. These three aspects are elements of the initial model we propose below. Consequently, if victimization does indeed drive SST, SST within a conflict will also be reflected in outcomes that affect (1) empathy, (2) memory, and (3) side-taking flexibility. Ultimately, our initial studies indicate that victimization drives SST and indeed results in (1) greater empathy for the victim, (2) greater memory of the victim, and (3) diminished flexibility to switch sides.

Our overall model suggests that when someone observes a conflict with a victim and a perpetrator, the observer is more likely to initially side with the victim and to remain loyal to this side in the long run. In part, this side-taking and loyalty is mediated by increased empathy and perspective-taking with the victim. However, we propose SST manifests itself in several cognitive operations related to attention, particularly aimed at tracking the state and well-being of the victim. This raised attention toward the victim increases memory of the victim and their circumstances. Attention and memory also make the “story” of the victim more concrete for the observer. Hence, the observer is in the position to tell the story and perspective of the victim This attention and storytelling do not require a first-person perspective but makes the perspective of the victim more available (our measure 2 below: perspective-taking; viewing the victim as storyteller). Lopsided attention and perspective-taking continue to make the perspective and side of the victim a likely target for continued side-taking. Once a side has been taken, this side is more likely to either stick and/or to be selected again (our measure 3 below: confirmation of side-taking after the initial side-taking).

There are many reasons people may take sides, including morality, coalition support [47], blameworthiness [20], perceived similarity [50], and out-party animus [15]. Perceived victimization may also be a powerful indicator in the process of SST. The Dyad Model [58] offers a framework to portray conflicts in terms of two sides: the perpetrator and the victim (with a positive bias toward the latter). Aggression may initiate SST, as it makes side-taking and intervention an overt option for observers to assist the victim. However, victims should fit a particular schema in how they look or act for observers to prioritize their safety [24]. Aggression and violence are linked to the moral foundation of care and harm [27] and thus are likely to trigger a protective impulse. Wegner and Gray [25] further propose that portraying oneself as a victim is actually more effective than demonstrating virtues (the hero strategy) when trying to avoid blame.

A different way to account for SST would be that in siding with perceived victims, one may be more likely to track the perpetrator to direct blame at them and those who perform similar actions [17, 20]. While our findings do not support this possibility, there may be cases where side-taking may lead to higher attention to and memory of the perpetrator and perpetrator perspective-taking (which could include a third party who punishes the perpetrator).

Another powerful stimulus of side-taking is morality. In the framework of the Bystander-Coordination Model, morality particularly offers a public signal by which to coordinate the opinion and action of a larger group during side-taking in a conflict [13, 14]. In this process, their choices will be coordinated so long as they have a shared perspective through which to interpret the actions of the disputants. Consequently, the role of morality in side-taking may be a shared perspective since bystanders coordinate their side-taking based on the observable actions of each disputant.

The Bystander-Coordination model suggests higher flexibility when it comes to making judgments in a conflict. Behavior, identity, and perceived mind shape moral character judgments [29], and these can be ongoing processes that depend on emerging information. If members of a group simultaneously make coordinated judgments based on continuing information, then disseminating correct information is key to eliminating injustice. Such moral judgments in side-taking exist also in narrative entertainment at ongoing segments [41].

A different way to account for SST and ongoing support for side-taking is the Three-Person Model of Empathy [7, 8]. This model predicts that once SST occurs, it tends to stick as people take the perspective of the chosen side. Once they take one perspective, they observe the following social interactions asymmetrically. This perception further leads to the development of positive emotions congruent with their chosen side and seeing themselves as negatively impacted by the other side. Empathy can create a more radical sense of otherness for the non-chosen side [23, 42]. Perspective-taking and empathy then lead to continuous confirmation and enforcement of the spontaneous act of side-taking and thus increased polarization. As such, emotional engagement with a narrative empowers an audience to offer moral support for one side while empathy blocks support for a side deemed as morally dissimilar [3]. The Three-Person Model of Empathy suggests effects of empathy are increased in side-taking.

In most contexts, empathy is linked to beneficial sociality; there is ample evidence about the prosocial aspects of empathy, including its role in conflict de-escalation ([5, 32, 33]). Empathy or empathic concern is focused on the well-being of others and is a powerful effect to share the emotions of the unfortunate [4, 34]. Empathy intensifies pain reception and is modified by perceived closeness [26, 59] found that people who do not hold a belief that power distance or certain social hierarchies are justified react with higher empathy toward victims of company abuse. Empathetic narratives may then play a large role in business culture in informing hiring, layoffs, company culture, and competition.

However, this increased pain reception for some can fuel misdirected empathy in side-taking [6,7,8]. When people are already ingrained within sides or groups, they tend to focus their empathy only on those particular people, increasing polarization (see the interesting findings by Simas, Clifford, and Kirkland [54]) and unfair favoritism [6, 49]. Additionally, there is strong evidence that people experience more empathy toward ingroup over outgroup members [55]. Social media has been shown to amplify this polarization as well [44]. Empathy-like effects also occur in narrative fiction and are sometimes called narrative empathy [31]. After siding with a character in a narrative, readers can develop a strong sense of empathy for their chosen character, potentially inhibiting their empathy for other characters [7, 37].

2 Overview of the present studies

Our questions center on implications and outcomes important for social conflicts and examine some of the theoretical implications around dualisms, morality, and empathy in side-taking. We ask the following in our studies to further investigate the process of SST:

  1. 1.

    Do people consistently and spontaneously side with characters who have been victimized?

  2. 2.

    Will people pay more attention to victims or perpetrators? In particular, how does an event of aggression impact memory about later events concerning the victim-character and the aggressor-character?

  3. 3.

    How does victimization affect empathy for both the victim-character and the aggressor-character? We examine relatability, understanding, perception of moral goodness, and attribution of authorship as indicators of empathy.

  4. 4.

    Will people continue to be loyal to one side when receiving information about an ongoing conflict in which both characters are equally aggressive? (We measure loyalty positively by conviction about a selected side and negatively by switching sides).

2.1 Study 1

In this study, we construct a victim-perpetrator dynamic through a three-part story in which one character commits an aggressive action against the other in the second part of the story. We tested for empathy and perspective-taking with questions of which character participants sided with, which was thought to be the author, which they found more relatable, and which they understood more. Similarly, we asked participants to give numeric ratings for how relatable, understandable, and morally good each character was, with higher ratings being interpreted as indicative of higher overall empathy for the character. In addition to empathy, we also targeted memory by comparing how victimization influences the encoding of details related to the characters (victim and perpetrator). We hypothesized that the victim character would have a higher overall empathetic appeal as well as be better remembered.

2.1.1 Methods

2.1.1.1 Participants

Participants were collected through Amazon’s Mechanical Turk. A total of 127 participants took this survey, with 51 identifying as female. The average age was 40. We paid all participants at an approximate rate of $6/hour. Sample size was determined before any data analysis.

2.1.1.2 Materials

Participants read a three-part story about two interacting characters. The first part of the story introduces each character, one male and one female, describing each character and a general friendly relationship between them. The second part of the story contains an unexpected episode of aggression from one character toward the other. For example, one character pushes the other character aggressively, hurting them in the process, or yells loudly at them in a public setting. We alternated which character committed the aggressive act in this middle section. The third part of the narrative continues the story for each of the characters, who no longer interact, and how they move forward following the aggressive action.

2.1.1.3 Procedure

For an overview of the study, see Fig. 1. After each part of the story, participants answered the following questions:

  • Who do you think is telling this story? On a scale from 1–7, how confident are you that this character is telling this story?

  • In a conflict, with whom would you side (after Part 1)/Whose side are you on? (Parts 2 and 3)

  • Whom do you understand better?

  • On a scale from 1–7, how well do you understand [Aggressor's name]/[Victim’s name]?

  • Who is more relatable?

  • On a scale from 1–7 how morally good is [Aggressor's name]/[Victim’s name]?

Fig. 1
figure 1

Study design for Study 1. In Study 1, participants read 3 paragraphs about 2 characters in a conflict. After each paragraph, they answered questions about each character. After the third paragraph, they also completed a memory task for each character

Full size image

Numerical questions were presented with a slider that recorded and displayed integers from 1 to 7. Two anchor labels were provided for each question; the number 1 had labels such as “not at all” and the number 7 “very much.” The questions about understandability and moral goodness were asked for each character separately. In all choice questions, the names of characters were given in random order. Following the final story part, participants received an additional set of 6 multiple-choice questions testing their memory for non-conflicted related details about each character, such as what color shirt each character was noted as wearing. Each question had 1 correct choice and 3 incorrect choices.

Our intention in asking about authorship, side-taking, understanding, relatability, and moral goodness was to attain measures of different aspects of empathy for the characters within the story. Understandability and relatability target empathetic connection to the character through how the participant connects to the character, while side-taking then indicates with which character participants empathetically align themselves. The question about authorship indirectly targets perspective-taking, an aspect of empathy, since it reflects to which character the participants might attribute the third-person narrative. Moral goodness, while not a direct measure of an aspect of empathy, targets the potential empathetic appeal of the characters.

The different parts of the story and the questions for each were on separate pages, so participants could not refer back to the story. We used three base narratives for this study and alternated which character was aggressive in the second part of the story to randomize for character gender and the effect of particular character traits. Overall, this created a total of six different storylines. All stories are given in Appendix S1, as well as the memory questions. In these studies, we report all measures, manipulations, and exclusions.

3 Results

Overall, 20 comparisons were made in total for our different measures. The novel nature of our design in investigating empathy through narratives meant that there was little precedence for relying on previous findings for guidance. Moreover, since we aimed to distinguish possible effects for future study, we wanted to avoid falsely noting findings as statistically significant. Therefore, we used a Bonferroni correction, though severe, to constrict the likelihood of a Type 1 error. This resulted in an adjusted α = 0.0025.

Binomial tests were used to see if there was a significant difference between the participant groups (i.e. the proportion of responses being significantly different than 50%) regarding who they thought was the author, whose side they took, and who was more relatable. The question about relatability was mistakenly not asked after Part 3. Given our fixed n = 127, α = 0.0025, desired power of 0.80, and constant proportion we are testing against, we used the GPower software to perform a Sensitivity Power Analysis and determined a minimum detectable effect size of g = 0.177. No differences between groups were found for these measures after Part 1 (prior to the act of aggression). However, as expected in Parts 2 and 3, significantly more participants were found to have selected the victim character for each measure (Table 1).

Table 1 Results of binomial exact tests for author, siding with, and relatability after each story part
Full size table

Nonparametric Wilcoxon Signed-Rank tests were used to analyze the ratings for the understandability and perceived morality of each character at the end of each part of the narrative. We used these nonparametric tests because we used ordinal Likert scales for our measures. Given our fixed n = 127, α = 0.0025, and desired power of 0.80, we used the Gpower software to perform a Sensitivity Power Analysis to determine a minimal detectable effect size of r(125) = 0.37. After Part 1, no significant differences between characters were found for understandability or perceived morality. After Part 2, the victim (Md = 5) was found to be more understandable than the aggressor and more moral. Similarly, after part 3, the victim was rated as significantly more moral than the aggressor. In Part 3, there were no differences in understandability between characters. (Table 2).

Table 2 Results of Wilcoxon signed rank tests for the understandability and perceived morality of each character
Full size table

For the memory items (Table 3, Fig. 2), we ran a χ2-test to see if participants answered more questions correctly about the victim character than the aggressor character. Given our 762 total questions for a n = 762, α = 0.0025, and desired power of 0.80, we used the GPower software to run a Sensitivity Power Analysis to determine our minimal detectable effect size of r(760) = 0.10. Our χ2-test revealed participants answered detectably more answers correctly about the victim than the aggressor (Table 3, Fig. 2).

Table 3 Memory accuracy regarding number of correct items per character
Full size table
Fig. 2
figure 2

Memory accuracy regarding number of correct items per character. Participants correctly remembered more information about the victim (637 correct recalls and 125 incorrect) than about the aggressor (381 correct recalls and 381 incorrect recalls)

Full size image

We also performed Mann–Whitney U Tests for the commitment ratings given for the selection of author and side. These commitment ratings were compared based on which character participants had chosen, and again we used nonparametric tests because of our use of ordinal Likert scales. However, these comparisons were not found to show significant differences between groups.

4 Discussion

These results indicate that victimization leads to greater empathy for and encoding of the victim. It is certainly not a surprise that participants have more empathy for the victim, as suggested would be the case by frameworks such as Wegner and Gray’s mural dualisms, Kurzban et al.’s Bystander-Coordination Model, and Breithaupt’s Three-Person Model of Empathy; however the blatantly lopsided memory finding is striking (see Fig. 2). We propose that victimization triggers a shift in perceived morality of the characters, which directs greater empathy toward the more moral victim in side-taking. Empathy in turn triggers greater encoding of the victim. The shift in morality is indicated in that after the victimization of one character by another, the victim is perceived as more moral for the remainder of the narrative. This perceived increase in morality is accompanied also by indicators of increased empathy for the victim, specifically as participants side more with the victim, view the victim as more likely to be the author, and find the victim more relatable and understandable. We suggest that because the victim becomes more empathetically appealing, participants may pay more attention to them and encode more details about them. A heightened attention toward and encoding of the victim explains the better performance on the memory task for questions related to the victim.

This lopsided attention to the victim through empathy and perspective-taking is especially noteworthy and has implications not only for how people read stories but also how they retell and propagate them. It would have been a plausible expectation that people track the perpetrator, as this character should be blamed and avoided. This is not what we found. Rather, participants seem to have stopped encoding or recalling details of the perpetrator. As participants remembered more details about the victim, they themselves may be more likely to retell these stories to emphasize the victim’s perspective. They not only side with and relate to the victim, but also track and encode the victim’s attributes which may provide both the motivation and ability to pass on the story of the victim’s struggle. Not only do people relate to and side with victim characters more, but they view these characters as more likely to author the given story, a novel measure we introduce. This connection of being a perceived story author, being the victim, and being more relatable could suggest that the act of telling a story makes the teller more relatable, more like a victim, and thus more likely to be sided with (and then remembered). While in this study we primarily investigated the connections between victimhood, relatability, authorship, memory, and spontaneous side-taking, we did not look at how SST sticks and potentially diminishes side-taking flexibility. We conducted another study to examine some of the effects of entrenchment in SST.

4.1 Study 2

In Study 1, we tested narratives with two people where one side suddenly commits a strongly immoral transgression. This narrative design of lopsided morality allowed us to track the side-taking dynamics by readers and how they affected attention/memory and empathy toward each of the two characters. We then wanted to know how SST sticks over time as a potential consequence of increased attention and thus made a key change to the narrative design of Study 1. Instead of presenting a morally-lopsided conflict with one perpetrator and one victim, we investigated situations with ongoing and morally balanced conflicts. This design change allowed us to encourage defection since no side was clearly morally right. We created stories that presented an ongoing, balanced conflict with five acts of aggression committed by each side. In this paradigm, we had participants make a forced choice between the sides and provide a confidence rating. If people maintain cognitive flexibility, we would expect about equal rates of defection and loyal side-taking. If previous SST sticks, we would expect an imbalance.

4.2 Methods

We built four stories consisting of five paragraphs. Each of the five paragraphs described similarly aggressive actions (name-calling, tripping, punching, vandalism, etc.) committed between both characters (either both male or female); each character was both perpetrator and victim in each of the five segments. After each of the five paragraphs in which each character acts aggressively toward the other, participants answered the forced choice question, “Which character do you feel is more justified than the other in this narrative?” They were then asked to “Please rate how much you feel you support this character” (we refer to this as conviction) and rated their conviction on a slider that recorded and displayed integers from 0 to 10. Two anchor labels were provided; the number 0 had the label “weakly” and the number 10 “strongly.” Our analysis focused on comparing at each stage of the story whether participants repeated the same choice as the previous one or not and what their confidence was in making each decision.

4.2.1 Participants

We recruited new participants via Amazon Mechanical Turk. We included 161 participants. The average age of participants who participated was 40 years old, and 96 reported as male and 65 as female. We paid all participants at an approximate rate of $6/hour. Sample size was determined before any data analysis.

5 Results

Each of our 161 participants made five side-taking decisions. The choices of participants confirmed that the conflicts within the stories were well-balanced. For example, after the final paragraph, the most imbalanced of the stories had a rate of 12 to 26 between the characters, while the other stories had more balanced divisions.

We particularly looked at whether or not participants were likely to remain loyal to their past choice. In other words, after choices 2, 3, 4, and 5, we examined how many participants chose to side with the same character as they did in their directly preceding choice. Consequently, we divided participants based on whether or not they switched sides from that last choice.

To analyze these data, we performed 4 Binomial exact tests, one for the data collected after each of the last 4 paragraphs. To constrict the likelihood of a Type 1 error, we again used a Bonferroni Correction for an adjusted α = 0.0125. Because the narratives themselves did not favor a particular character, without any effect of past choices we would expect participants to side with and rate characters equal to chance, meaning half the time they would be loyal to their previous choice. Consequently, the constant proportion we tested against was 0.5. Given our fixed n = 161, α = 0.0125, desired power of 0.80, and constant proportion we are testing against, we used the GPower software to perform a Sensitivity Power Analysis and determined a minimum detectable effect size of g = 0.131.

From our binomial exact tests, we found significant and detectable effects after paragraphs 2 and 5. After both of these paragraphs, significantly more participants were loyal to their previous choice. Specifically, after paragraphs 2 and 5, we found that 68% of participants were loyal to their previous choice. While not significant, after paragraphs 3 and 4 more participants were also loyal to their previous choice than disloyal, with 55% remaining loyal after paragraph 3 and 58% after paragraph 4 (Table 4). Conviction was always higher if people stayed loyal to the previous choice but consistently declined as the narrative progressed. However, we did not test for any significant differences in conviction rating as this fell outside our scope of how SST sticks over time. Specifically, any differences in conviction could be attributed to either an effect of switching, or to a cause of switching. Therefore, we concluded that analyzing these data would not help answer our question of whether people stay loyal to their chosen side in SST.

Table 4 Results of binomial exact tests for side-taking after each paragraph
Full size table

6 Discussion

Our results indicate that previous side-taking influences SST by diminishing cognitive flexibility. Specifically, we find that overall people were more likely to stay loyal to their previous choice. Nevertheless, overall confidence declined in a narrative in which both characters continuously commit equally morally bad actions. This decline suggests that while the past side-taking choice made in a conflict influences subsequent choices, it still matters what actually happens in the conflict. In other words, while participants may have been influenced by their previous side-taking choice, they were not blind to new information. Instead, it appears that they continued to take in more information in making their decision and saw the conflict as more balanced over time. This growing inflexibility contrasts with the Bystander-Coordination Model though supports Breithaupt’s Three-Person Model of Empathy.

6.1 Overall discussion

Victimhood, empathetic engagement, and previous choices lead to strong and confident side-taking that is reflected in memory encoding, empathy, and diminished choice flexibility. By having participants read two different types of conflict narratives and in response to our main questions regarding SST, we noticed:

  1. 1.

    People have better memory of the perceived victim than the perpetrator.

  2. 2.

    People tend to side with the perceived victim in a conflict.

  3. 3.

    People attribute authorship to the victim, and rate the victim as more understandable, relatable, and of higher perceived morality, exhibiting higher empathy for this character.

  4. 4.

    People exhibit diminished cognitive flexibility after choosing a side in a conflict (i.e. in evenly matched cases, people tend to choose the same side they most previously chose).

These findings in narrative conflict complement real-world findings and suggest larger implications as to how side-taking more generally functions. As anticipated, people tended to side with the perceived victim of a situation and exhibited greater empathy for them. More surprisingly, when people choose the side of a victim, they also attributed authorship to the victim and remembered more details about them than they did about the perceived aggressor. This difference in memory was particularly dramatic as in our first study only half of the details about the aggressor were remembered correctly. Alternatively, participants remembered details of the victim correctly 84% of the time (637 of 762). We reason that this one-sidedness of memory indicates positive attention to the victim. The alternative possibility that people choose a side (the victim), but track (pay attention to) the perpetrator was consequently not confirmed in our study. Instead, participants showed poor memory recall of details concerning the perpetrator.

Our study validates potential positive outcomes of victimhood found in other psychological research. Presenting oneself or others as victims can be advantageous. Wegner and Gray’s model of moral dyads (as portrayed in Study 1) favors the victim and can lead parties in a conflict to present themselves as such, a process called competitive victimhood [46, 60]. Competitive victimhood strengthens one’s own sense of victimization and resistance to resolving conflict [28, 60], while also negatively impacting trust and empathy felt toward the outgroup [11, 45]. The Three-Person Model of Empathy [8] suggests similar favorable perceptions of victims and negative views of perpetrators. We add to this existing scholarship that victimhood may further negatively impact attention paid toward the outgroup.

Our findings also intersect with research on an affinity for the underdog, which may be associated with helplessness and humiliation [10, 22, 36, 43]. Relatedly, Plantinga [48] coined the term “sympathetic narrative” to explain why many people identify with characters who are introduced in the context of their misfortunes. These disadvantages make them more relatable and trigger an observer preference for the victim. Such coordinated siding with an innocent victim can be explained in existing side-taking frameworks, including the Bystander-Coordindation Model [13, 14]. The lopsided attention toward the victim, expressed by memory, we see in Study 1 is an important addition to scholarship on the effects of sympathy regarding side-taking. Stronger memory of perceived victims also has implications for social movements, potentially shaping moral attitudes on subjects such as the #MeToo movement [2, 9, 18, 57] and Black Lives Matter [56]. The influence of empathy on morality that we found in our studies also complements previous research around perceived agency and morality, which are correlated with competitive victimhood in real-life conflicts [30].

SST has implications for economic narratives as well [53]. We take sides in the workplace as people interact within larger groups while working toward a shared objective [38]. Marketing decisions between competing brands are often made based on emotional reactions [1] and SST may act as the initial step in establishing consumer loyalty [12].

We also found that SST is related to attributing authorship to the victim. In a forced choice, people are likely to assume that the victim would be the author (or narrator) of a conflict. What does it mean to be deemed the author of a story? To the best of our knowledge, this is not an aspect previous studies have measured. Author attribution may be akin to first-person narratives, but crucially it is based on a judgment of whom the audience believes is telling the story. A common scheme of narrative fiction is a first-person narrator as the protagonist and moral hero of the story with whom the audience identifies. Yet it is relatively uncommon to receive first-person insights about antagonists [17, 31]. Thus, judgments about authorship may go hand in hand with estimates of who will be the moral hero of the story [40], who should be blamed [21], and who deserves empathy. Seeger and Sellnow [52] have suggested that “stories told by victims are among the most compelling narratives following a crisis and are usually widely reported” (14). However, our studies do not show what about victimization leads to author attribution, or what impact this might have on how stories are remembered and retold. Given the novelty of these findings and their potential importance for how stories are propagated through society, we particularly suggest the role of author attribution in victimization narratives as a topic for future study.

Finally, with two aggressor characters in a narrative, Study 2 demonstrated diminished flexibility in participants to defect from their previous choice. In conflicts without a clear right or wrong where one might only reluctantly choose a side, the act of SST then affects one’s later positions of support for both sides of the conflict. While Kouchaki [35] has noted that perceived morality is based largely on past moral credentials, we would suggest that particularly one’s alignment in the most recent conflict shapes SST. However, Study 2 does not definitively illustrate how or to what extent side-taking flexibility diminishes, if side-taking flexibility diminished in all conflicts, or if there is anything that can be done to increase flexibility in side-taking. Consequently, we suggest that future work must be done to further understand flexibility in side-taking.

6.2 A model of spontaneous side-taking

Based on our results, we propose a model of side-taking by which perceived victimization triggers spontaneous side-taking, which then increases memory of victim characters. SST is characterized by greater attention to and encoding of the victim, which in turn leads to greater empathy with and perspective-taking of the victim. This enhanced attention toward, encoding of, and perspective-taking of the victim, indicators of or consequences of empathy, then leads to decreased flexibility of side-switching (Fig. 2). As this SST is entrenched through diminished choice flexibility, more attention continues to be directed toward the victim, resulting in greater encoding and perspective-taking of the victim. Overall, this process results in a cycle of positive reinforcement by which the original moment of SST is perpetuated (Fig. 3).

Fig. 3
figure 3

Model of spontaneous side-taking in conflict observation. This figure demonstrates the process of SST as we propose it for conflicts. Perceived victimization of one member in the conflict acts as a trigger for greater attention toward that victim and higher encoding of details related to them. This heightened attention and encoding of the victim leads to taking the perspective of the victim reflected in increased understandability, relatability, and assumption of authorship (Study 1). The effect of this dynamic of stronger encoding and perspective-taking is diminished flexibility to encode the other character (such as the perpetrator) and to take the perspective of this character. As a result, circular enforcement of the focus of the victim character emerges (higher encoding, increased perspective-taking, diminished encoding flexibility for other characters) that leads to the entrenchment of side-taking (Study 2)

Full size image

7 Conclusion

Overall, we find that victimization in narratives leads to enhanced memory and empathy for victim characters, an attribution of authorship to victims, and diminished flexibility in side-taking. Based on these findings, we suggest that future work on empathy in conflicts will particularly benefit from studying how people spontaneously choose sides and ultimately become engrained in their choices. While we suggest that memory and attention are important aspects of side-taking, we also suggest future work on the role and function of author attribution in conflict narratives. Although Study 1 identified that participants were more likely to view the victim as the narrator, future work must be done on why victimization makes one be perceived as more likely to be the narrator, and on how the author attribution may affect how people recall the narrative at later moments. Similarly, though we find tentative evidence that people show diminished flexibility in side-taking after making an initial choice of side, future work must be done to determine exactly what causes this decrease in flexibility and how it impacts functions such as empathy and moral judgment.

Although future work remains to be done on the dynamics of SST, our study has tentative implications for an increasingly polarized society in which many people believe they have the “right side.” Framing stories around victimhood leads to SST and may make SST more salient. While one may hope that SST trains flexibility of the mind not only to choose a side quickly in a conflict but also to escape entrenched polarization by imagining other’s perspectives [14, 19], our studies suggest that SST may stick and thus result in cemented side-taking and polarization. SST may be connected to people’s unwillingness to give voice to the other side of an issue or conflict, demonstrated by current conversations about “cancel culture” and “safety-ism” [39]. This restricted view of others’ morality is strengthened by filter bubble effects in media consumption [51] as well as reinforced by algorithms that depend on strong, often polarizing emotional reactions on social media platforms [16]. While side-taking can be a way to engage and cognitively reflect on current crises, side-taking can also then become a side effect of existing within an echo chamber. Most novel in our study is our finding around increased memory of victimized characters. Victim narratives not only evoke empathy but attention, leading to potentially long-lasting effects of SST.