Introduction

Seeing is believing, or so is the expression. People can see the same things but interpret them differently. In social situations, this can lead to misunderstandings and may, in the case of bullying, lead to unwanted negative spirals. Social cognition refers to how individuals construct their (subjective) social reality. It involves psychological and cognitive processes of how a person processes, stores, and applies information about the social world (Ostrom, 1984). When children are the victim of repeated, intentionally aggressive, or hurtful behavior from a powerful perpetrator, such as in bullying (Olweus, 1993), it is likely to affect how they experience their social world and may affect their general social cognitive tendencies. This may have long lasting consequences, even into adulthood, as victims report more mental health problems, lower levels of academic achievement, and poorer social relationships (Arseneault, 2018). A better understanding of the effects of victimization on social cognition of victims is needed, as social cognition likely influences (the perception of) new social interactions, and thus play a role in the victimization process. To date, it is unclear how victimization relates to different aspects of social cognition, as many studies only examine a subcomponent of this highly complex construct. Therefore, this systematic review provides an overview of the literature on the relation between victimization and different aspects of social cognition.

Bullying is a social process (Salmivalli, 2010). Both bullies and victims influence social interactions by sending and interpreting social messages. The past experiences and the social-cognitive thinking style of children influence how these social messages are received. According to the social information processing (SIP) model (Crick & Dodge, 1994), the social-cognitive processes can be organized in six subsequent steps or phases. The first two phases refer to the selection and interpretation of social cues: encoding (which cues are registered) and interpretation (interpreting the combined meaning of all registered cues). For example, perceiving someone’s facial expression without any interpretation is part of the encoding phase, whereas thinking about why someone is laughing contains a subjective component and is therefore an interpretation. A feedback loop exists between encoding and interpretation, such that interpretation can influence encoding of subsequent social cues. The last four SIP-phases refer to cognitive processes which enable an individual to form a behavioral response: goal selection, construction of possible responses, response choice, and the execution of the chosen response (Crick & Dodge, 1994). Finally, each person has a database which includes memories, schemas, and social knowledge that influence each SIP-phase (Crick & Dodge, 1994). For example, if people get to know each other (knowledge and memories), the interpretation of the other’s intentions will improve. While the database concurrently influences the SIP-phases, it is also simultaneously updated with new experiences. The updated database, in turn, influences future encoding, interpretation, and behavioral responses, possibly leading to the development of negative spirals. Victims are likely to have difficulties in sending and interpreting social information, as they seem to have poorer social cognition and social skills (Fox & Boulton, 2005).

No matter whether it is a precursor or consequence of victimization, having difficulties in social cognition (or social intelligence, see Kaukiainen et al., 2002) may relate to being victimized. Based on the SIP-model, prior victimization experiences likely affect how victims encode, interpret, and respond to social cues as their past experiences color the interpretation of new social situations. First, victims may have a positive social-cognitive style and focus on positive social cues and interpretations to facilitate reaffiliation (Bernstein, 2003; Pickett & Gardner, 2005), called the reaffiliation hypothesis in this review. In line with this hypothesis, socially excluded people tend to focus more on positive social cues (Buckner et al., 2010) and have better memory recall for positive events (DeWall et al., 2011). Second, instead of focusing on positive social cues, victims may develop a negative social-cognitive style and focus more on negative social cues or threat, and detect such stimuli more rapidly, to prevent subsequent negative interactions (Rapee & Heimberg, 1997), the prevention hypothesis. In line with this, previous studies have found that anxious individuals tend to focus on negative information (Bar-Haim et al., 2007), lonely individuals expect more rejection by others (Spithoven et al., 2017), and victims of all types of events (e.g., natural disasters and crime) are prone to interpret hostile intent (van Reemst et al., 2016). Third, victims may become increasingly insensitive to social information, the desensitization hypothesis, as repeated victimization might lead to an insensitivity to all kinds of social information (Bernstein, 2003). However, to date, it is unclear what the nature of this relation is.

Current Study

The aim of the present study was to examine the relation between victimization and social cognition (encoding and interpretation) in children and adolescents through a systematic literature review. The results were interpreted in light of hypotheses that were formed on the relation between victimization and social cognition: the reaffiliation (focus on positive social cues to reaffiliate), prevention (focus on negative social cues to avoid future victimization), and desensitization (numbing or insensitivity to social cues) hypotheses.

Methods

Search Strategy

The systematic review was carried out in line with the PRISMA guidelines (Moher et al., 2009). Search terms were included for victimization and social cognition in three databases (PsycInfo, Web of Science, ERIC), see the identification block in Fig. 1 and Online Source 1. The literature search was executed in July 2018 and updated in February 2021.

Fig. 1
figure 1

Flow diagram of the selection process of the articles. In the first box you read the utilized search term and the hits per database. In the following blocks you can see the other article selection phases (title deselection, abstract reading, diagonal reading & full-text reading)

Full size image

Study Selection

The presented results are of the combined initial search and update. The search resulted in 7628 hits (PsycInfo: 3477, Web of Science (WoS): 3232, ERIC: 919). 1890 hits were excluded, if they had titles unrelated to the topic (e.g., other victimization types, such as gang violence), were not peer-reviewed publications (to warrant study quality), or were not written in English. After removal of 1334 duplicates, 4404 articles were left (PsycInfo: 2611, WoS: 2521, ERIC: 706). In round two, the abstracts were read and articles were selected only when they referred to victimization of bullying or social cognition. This round resulted in 1227 articles (PsycInfo: 772, WoS: 767, ERIC: 213). In round three, diagonal reading, the articles were scanned for relevant information and selected articles that examined the relation between social cognition and victimization and had a particular focus on either encoding or interpretation. Articles were excluded if they included participants older than 18 years, or if they examined coping, attribution style (internal/external, stable/unstable, controllable/uncontrollable), general social skills, or if they examined the last four phases of the SIP-model (behavioral response). During the first search in 2018, two researchers independently did round three, the diagonal reading. Inter-rater reliability for selection in this round was 89.9% (79.3% for the first 100 and 91.5% for the remaining articles). In case of disagreement, the researchers discussed the article and respective arguments, to come to an agreement. During the search update in 2021 one researcher read the articles because of the high inter-rater reliability obtained during the first search. Diagonal reading resulted in 198 included articles (PsycInfo: 135, WoS: 151, ERIC: 37).

These 198 articles were read thoroughly by the first author. An additional 56 articles (15 first search, 42 search update) were excluded (e.g., focus on rejection instead of victimization or because of participant age either being younger than 4 or older than 18). This led to an inclusion of 142 articles in the review (PsycInfo: 101, WoS: 110, ERIC: 25), published between 1998 and 2021. Studies included on average about 1600 participants (Median = 390, range: 14–25,684), who were on average 11.4 years old (range: 4.1–17.0). Figure 1 provides the selection flow diagram. Online Source 1 contains details on the selection process.

Quality Assessment

The commonly used Newcastle–Ottawa Scale for study quality assessment (Wells et al., 2000) was adjusted for the purposes of this systematic review. Studies were evaluated on sample, measures, and method in relation to our aim: examining the relation between social cognition and victimization of bullying. Sample referred to the representativeness of the sample, the appropriateness of the sample size, the non-response, and the participant characteristics. Measures referred to the measurement of victimization as well as social cognition, and the use of control variables. Method referred to the type of study (longitudinal/cross-sectional), pre-registration, the use of descriptives, and the use of statistics. The first author assessed the quality, which was replicated by the third author for 10% of the studies. The interrater reliability for this subset was satisfactory (r = 0.84, p < 0.001; ICC = 0.89, p < 0.001).The included studies received between 4 and 13 points out of the possible 16 (M = 8.50, SD = 1.82), with a notable exception of one receiving 1 point (see Online Source 2a Table S3 for the quality assessment per study and Online Source 2b for details). Studies most often lost points on not including effect sizes and response rates, using non-validated measures, not having a longitudinal design, and not pre-registering the study. Generally, the overall quality of the studies seems unsatisfactory, which demonstrates possible improvements for research designs.

Results

The results are organized based on the two aspects of social cognition: encoding and interpretation (first two SIP-phases) in relation to victimization of bullying. Within each SIP-phase, results are categorized in subdomains, based on topics addressed in the articles. Results were interpreted in light of the reaffiliation, prevention, and desensitization hypotheses.

Victimization and the Encoding of Social Information

The articles on encoding, the first SIP-phase, examined which cues were attended to (selection) and whether cues were registered as what they objectively were (accurate registration).

Selection of Social Cues

Selection of social cues, or attention, refers to how perceptual, motor, and cognitive systems are allocated to potentially competing information-processing demands. It involves both conscious and unconscious selection to different social cues (Anderson, 2015).

Only four studies examined victimization in relation to attention (see Table 1); therefore, conclusions are preliminary. Based on the reaffiliation hypothesis, it was expected that victims would attend predominantly to positive social cues, in order to identify possibilities to restore social relationships. However, most studies focused exclusively on negative social information. One study examined differences in the attention to emotions between victims and non-victims, but neither specified valence of emotions nor found support for victims differing in attention to others’ emotions compared with peers (Hussein, 2013). Therefore, no conclusions could be drawn for the reaffiliation hypothesis.

Table 1 Study results on selection of cues (attention)
Full size table

Based on the prevention hypothesis, it was expected that victims would attend predominantly to negative social cues, to prevent further negative interactions. There was support for this hypothesis, as frequent victimization was associated with a higher likelihood to notice bullying events in real life (Jenkins & Nickerson, 2017). Another study examined visual attention to bullying scenarios, and showed that all children spent most time looking at bullies and second most time looking at victims (Troop-Gordon et al., 2019). Thus, children focused attention mostly on the immediate threat, the bully, and not, for example, on bystanders who could possibly help. For victims, the attended social cues might affect behavior, as heightened attention for bullies by victimized children related to more aggressive behavior (Troop-Gordon et al., 2019). These findings provided indirect support for the role of threat prevention in attention to social cues in general.

A focus on threat not only became apparent in studies employing tasks with visual social cues, but also in a study examining interference of word valence on the ability to name the color of the word (emotional Stroop effect: Williams et al., 1996). Victimization experiences related to faster responses to victimization related words (Rosen et al., 2007). Faster responses to victimization related words indicate faster attendance to these social cues, which could imply that victims focus on threatening social cues. Alternatively, these results can also be interpreted in line with the desensitization hypothesis, from which it was expected that victims would attend less to social cues in general. Frequently victimized children may be more insensitive toward negative valence words, so that the valence of these words interferes less, resulting in faster reaction times. As such, whether this study provides more support for the prevention or the desensitization hypothesis remains unclear.

Taken together, the studies on attention and victimization suggest that victims may have a heightened focus on negative or threatening social cues, as two studies gave indirect support and one study out of four clearly indicated that victims might notice social threatening events more often. Across time, this focus on negative social cues may result in numbness for negative social information, but the conclusion regarding the desensitization hypothesis is tentative, as no study investigated prolonged (persistent) victimization. So far, studies have not examined heightened attention to positive information, which means that there is not enough evidence to assess the reaffiliation hypothesis properly. Although the (scarce) evidence so far seems most in line with the prevention hypothesis, it is possible that victims use different social-cognitive styles depending on the situation and that the situations invoking reaffiliation have been omitted. For instance, situations with unknown peers might invoke reaffiliation strategies, whereas situations with familiar peers might invoke prevention or desensitization strategies.

Accurate Registration

Accurate registration of social cues is crucial for effective social interactions. Inaccurate registration of important information can lead to less adequate responses, which can come across as odd, rendering one vulnerable for victimization. Below, the general accurate registration of social cues (6 studies) and emotion recognition (12 studies) are discussed separately.

Table 2 displays the research on accurate registration of social cues and emotion recognition. Some evidence for general accurate registration points toward the desensitization hypothesis (1 of 6 studies), which predicts that victims would focus less on social events in general, and therefore have lower accuracy in perceiving social cues, regardless of their valence. One of the five studies indicated that victimization related to a less accurate registration of social cues in general. Children who, according to their teachers, had more victimization experiences were less accurate in telling what they saw in a picture of a social situation (Ogelman & Seven, 2012).

Table 2 Study results on registration accuracy
Full size table

Other research was in line with the prevention hypothesis (3 out of 6 studies), which predicts that victims would focus on negative social cues and therefore would register more cues as negative, leading to a lower registration accuracy of positive social cues. These studies showed that victims register cues with a negative bias. They underestimated the number of balls they received from other players in a digital ball tossing game (Cyberball) and overestimated the times they were excluded from the ball tossing interaction, regardless of being in- or excluded in the game (Lansu et al., 2017). Overestimation also occurred in rejection perceptions. Overt victimization, and not relational victimization, related to increased perceptions (overestimation) of peer rejection (Zimmer-Gembeck et al., 2013). Moreover, being victimized related to lower overestimation of one’s own likeability (Garandeau & Lansu, 2019). These general tendencies pointed to a negative social-cognitive style, in line with the prevention hypothesis.

One study provided some support for the reaffiliation hypothesis, which predicts that victims, in pursuit of reaffiliation, would register ambiguous social cues more positively, leading to lower registration accuracy of negative social cues. In a study where children had to identify children as being part of their own or another team, victims tended to be better at recalling their own rather than opposite team members, when they had seen pictures of the others only once before (Telzer et al., 2020). By remembering team member’s identities better, they may optimize social opportunities, as they have something (although minimal) in common. In a last study peer-reported victims were compared with bullies, followers, defenders, outsiders, and non-involved children. This study provided no support for any of the hypotheses, as victims did not differ from others in their ability to recall and retell vignettes of social provocation and ambiguous scenarios (Camodeca & Goossens, 2005). The differences between these studies might be because of the assessment of registration accuracy, as the type of task (e.g., recalling or describing) and the level of language skills needed to complete it, differed between the studies, possibly leading to diverse results.

In sum, results on accurate registration are limited and mixed, and much remains unknown, but there are first indications that victims differ in registration accuracy of social cues.

Emotions

Emotion recognition is the ability to label facial expressions of others correctly. Twelve studies on accurate registration were found, of which two studies examined recognition of specific emotions, eight studies overall emotion recognition, and two studies both overall and specific emotion recognition. Most evidence indicated no significant relation between victimization and overall emotion recognition (8 out of 10 studies, of which 1 had mixed results depending on the task used), see Table 2. There was no evidence for tasks with basic emotion recognition that victimization among young children (4–6 years old) related significantly to overall emotion recognition: neither for matching emotional pictures (Heinze et al., 2015; Laurent et al., 2018) nor when choosing a facial picture of how a person in a story feels (Belacchi & Farina, 2010; Garner & Lemerise, 2007; Heinze et al., 2015; Laurent et al., 2018). These results contradict the desensitization hypothesis, which predicts that victims are worse overall in emotion recognition, because of desensitization to others’ emotions. However, the results are also not in line with the prevention and reaffiliation hypothesis, which similarly predict worse overall emotion recognition: victims focusing more on negative social cues could be more open to register more expressions as negative (prevention) and vice versa, victims focusing more on positive social cues could be more open to register more expressions as positive (reaffiliation). However, the non-significant results might be due to the use of tasks that were too simple and thus unable to catch subtle differences between children or show ceiling effects. Indeed, the use of more complex measures provided different outcomes (study quality did not clearly differ between the simple and more complex tasks studies). Results on a more intricate task (labeling emotions of puppets giving visual and audio cues), indicated a positive relation between victimization and accurate labeling of emotions (Heinze et al., 2015). Additionally, physical victimization (but not relational) related to better overall emotion recognition (Garner & Lemerise, 2007). However, victimization was related to making more mistakes in labeling ten different emotions in pictures (Miller et al., 2005). Thus, the findings were mixed and also often based on relatively small sample sizes.

Among the studies on older children (10–15 years old), there were a few with larger sample sizes. The study with the largest sample size (N = 6233) provided evidence that victimization related negatively to self-reports of emotion recognition abilities (Hsieh et al., 2019), whereas the second study (N = 757) using a task (reading the mind in the eyes) did not (Guy et al., 2017). Both studies had higher quality levels in the quality assessment (9 and 11 points respectively). Furthermore, most other studies did not provide evidence that victimization related to overall emotion recognition (Baird et al., 2010; Baroncelli & Ciucci, 2014; Ciucci et al., 2014), except for cyber-victimization which related to lower general emotion recognition abilities (Ciucci et al., 2014). Overall, the relation between victimization and overall emotion recognition seems not significant.

Although overall differences in emotion recognition do not seem apparent, victims may have worse abilities in emotion recognition of specific emotions. For example, emotions like fear or sadness may be recognized as anger to prevent (possible) assault (prevention hypothesis), or contrarily, emotions such as contempt may be recognized as positive and social opportunities (reaffiliation hypothesis). Four articles examined separate emotions. One study focused on younger children (4–6 years old) and found that physical and not relational victimization, related to worse recognition of angry expressions (Garner & Lemerise, 2007). Another study on victimization type found that cyber-victimization related to making more mistakes on recognizing anger and fear (Ciucci et al., 2014), yet found no evidence for a relation with traditional victimization. A small study (N = 43 girls), found that female victims were less accurate in labeling negative emotions, especially if they were sensitive to rejection, but no significant effects were found for labeling positive emotions (Rudolph et al., 2021). There might be gender differences, as a last study found that boys’ victimization related to making more errors on recognizing fear and sadness, whereas for girls, victimization related to errors on recognizing sadness, disgust, and surprise and to less mistakes on recognizing happiness (Pozzoli et al., 2017). These studies indicate that victims might have trouble recognizing negative rather than positive emotions.

In sum, findings on registration accuracy suggest that across different types of social cues there are indications that negative social cues might be registered less accurately, which would be in line with the prevention hypothesis. If true, such misrepresentations of negative social cues would set victims at a disadvantage for appropriate responding, possibly leading to misunderstanding by others and yielding them even more vulnerable for rejection or victimization. However, the conclusions are tentative because of the small number of studies examining registration of social cues and lack of longitudinal studies investigating these links.

Interpretation of Social Information

After the selection and registration of social cues, the next step in the SIP model is the interpretation of social cues. Interpretation comprises assigning meaning to the (combination of) registered cues. For example, when you attend to and accurately register a face as happy, interpretation involves attributing a reason to why this person is smiling given the other social cues in that situation. The topics related to interpretation in the selected articles involve attribution of situations, peer perception, and the ability to understand others, which includes empathy and theory of mind (ToM).

Attribution of Situations

Attribution of situations was a popular topic in research on social cognition of victims (49 studies). Examples were interpretation of others’ intentions and expectations of what would happen. Most evidence favored the prevention hypothesis, which predicts that victims have more negative interpretations of social cues because of their focus on threatening information (33 studies, of which two found it only for a specific victimization type and one only for a certain nationality), see Table 3. With respect to attributions about bullying, victimization related to thinking that bullying is a serious problem at one’s school (Choi & Park, 2018), interpreting ambiguous scenarios more often as bullying (Calleja & Rapee, 2020) and thinking they were being bullied more often because of their personality instead of their behavior (Morrow et al., 2019), but not to a more negative attitude toward the concept of bullying (Caravita et al., 2019; Pouwels et al., 2017).

Table 3 Study results on attribution of situations
Full size table

Other research, on non-bullying situations, found that victimization related to expecting threat (Balan et al., 2018; González-Díez et al., 2017; Hunter et al., 2010; Röder & Müller, 2020), being rejection sensitive (Ding et al., 2020; Rowe et al., 2015; Zimmer-Gembeck, 2015; Zimmer-Gembeck & Duffy, 2014; Zimmer-Gembeck et al., 2013, 2014), worrying about negative evaluations (Estévez et al., 2019; Giannotta et al., 2012; Gómez-Ortiz et al., 2018; Taylor et al., 2013; Zimmer-Gembeck & Duffy, 2014), feeling rejected when receiving few likes in a social media task (Lee et al., 2020), and being more sensitive to aversive stimuli (Rudolph et al., 2016). Moreover, a longitudinal study found that adolescents with victimization experiences adopted rejection schemas more often. These maladaptive schemas, in turn, predicted later victimization (Calvete et al., 2018). Not only perceived threats of rejection, but also of hostility have been examined. Victimization experiences related to hostile intent attributions in ambiguous situations (Camodeca & Goossens, 2005; Camodeca et al., 2003; Guy et al., 2017; Hoglund & Leadbeater, 2007; Hung et al., 2017; Kokkinos & Voulgaridou, 2018; Ogelman & Seven, 2012; Perren et al., 2013; Pornari & Wood, 2010; Schwartz et al., 1998; Yeung & Leadbeater, 2007), as well as in provocative situations (Ziv et al., 2013). In other words, children with more victimization experiences seemed to interpret situations more often as threatening, consistent with the prevention hypothesis. In more positive situations, victims did not interpret more hostile intent (Smorti & Ciucci, 2000). Thus, the situation (positive, ambiguous, or negative) might influence the adopted social-cognitive style.

Findings from three other studies painted a more complex picture. The specific bully-role might be important, as one study indicated that only bully-victims, but not pure victims, interpreted video fragments of ambiguous social interactions (abstract figures, animals, and humans) more often as bullying compared to non-involved children. Additionally, bully-victims interpreted even positive interactions of apes in a zoo more often as bullying compared to non-involved children (Pouwels 2016). A second study found that only relational victimization, and not physical victimization, related to interpreting situations as more negative and social threatening (Calleja & Rapee, 2020).

However, there were several contradictory findings (14 studies). First, some studies did not find significant associations between victimization and hostile interpretation tendencies (Leff et al., 2014; Mathieson et al., 2011; Prinstein et al., 2005; Smalley & Banerjee, 2014; Smorti & Ciucci, 2000; van Dijk et al., 2017; Warden & Mackinnon, 2003) or being more upset after hypothetical, ambiguous provocations (McQuade et al., 2019), or only found evidence of this association for Japanese but not for American children (Kawabata et al., 2013). Other studies found no association between victimization and fearing negative evaluation (Kiekens et al., 2020; Pabian, 2019) or expecting and interpreting rejection (Rudolph et al., 2021; Stubbs-Richardson & May, 2020). In addition, one study found no evidence that victims interpret situations more often as emergencies when they were also asked whether they would intervene that particular situation (Jenkins & Nickerson, 2017). Notably, the average study quality of these studies (M = 7.21 points, SD = 1.42) is lower than the study quality of all studies on situation attribution (M = 8.69, SD = 2.02), slightly suggesting that the results that were more common and in line with the prevention hypothesis might be more valid.

With regards to the reaffiliation hypothesis, in the seven studies that examined positive attributions separately from negative attributions, no evidence was found that victims interpret situations more positively or attribute less hostile intent (Garner & Lemerise, 2007; Hung et al., 2017; Smorti & Ciucci, 2000; Warden & Mackinnon, 2003), nor that victims perceive more social opportunities amid school transitions (Röder & Müller, 2020), nor judge bullying conceptually right (Caravita et al., 2019). Victims might even interpret intentions less positive compared with non-involved children (Camodeca et al., 2003; Ziv et al., 2013).

In sum, strongest evidence was found in line with the prevention hypothesis (33 out of 49 studies): victims interpreted situations and intentions of others more negatively, hostile, or dangerous, and were more sensitive to rejection. Moreover, this negative attribution style seems to include both more negative and less positive interpretations.

Peer Perception

There were many studies that examined how victims generally perceived their peers, see Table 4 (35 studies). In line with the prevention hypothesis, most research (26 studies, of which three with partial support) indicated that victimization related to negative peer perception (Hong et al., 2019, 2021; Ladd & Troop-Gordon, 2003; Purcell et al., 2021; Rudolph et al., 2009; Sainio et al., 2013; Salmivalli & Isaacs, 2005; Schacter & Juvonen, 2018; Troop-Gordon & Ladd, 2005), less trust in peers (Betts et al., 2017), negative expectations of peer interactions (Audley et al., 2020), and negative perceptions of the social climate and interpersonal relationships (Barzeva et al., 2020; Berg & Aber, 2015; Elsaesser et al., 2013; Gini, 2008; Harks & Hannover, 2020; Holfeld & Baitz, 2020; Lázaro-Visa et al., 2019; Leadbeater et al., 2015; Mertens et al., 2021; Moyano et al., 2019). Some longitudinal studies also provided evidence in this direction: Increased victimization led to increases in negative peer perception (Elsaesser et al., 2013; Troop-Gordon & Ladd, 2005), and negative peer perception predicted subsequent victimization for girls (Sainio et al., 2013). However, cyber-victimization did not explain negative peer perception over and above traditional victimization (DePaolis & Williford, 2019). In addition, a second study also did not find a concurrent relation between victimization and peer connectedness perceptions (Aldridge et al., 2020).

Table 4 Study results on peer perception
Full size table

Several studies indicated that it is important to examine to which peers the perception refers. Victims attributed less favorable characteristics to non-friends than they did to friends (van Noorden et al., 2014), even more strongly than non-involved peers (van Noorden et al., 2016). This might be partially based on sympathy, as victims judged non-involved peers more positively than they judged bullies and bully-victims (van Noorden et al., 2016), again in line with the prevention hypothesis. Though victims themselves might think less of bullies, they do not necessarily think that others have similar thoughts. Being victimized was not related to thinking that hypothetical victims and classmates would dislike a hypothetical bully (Garandeau & Lansu, 2019). However, victims themselves judged known bullies and followers more positively than unknown bullies and followers (reaffiliation hypothesis), whereas known defenders are judged less positively than unknown defenders (prevention hypothesis) (Pouwels et al., 2017). Possibly, judgements occur in reference to standards that are hold for specific peers.

Additionally, two studies indicated that the general peer perception of victims might be rigid. They found that victimization related positively to holding an entity rather than an incremental perspective on peer relationships (Rudolph, 2010) and on peers’ personalities (Kaufman et al., 2020). In other words, victims seem to think that peers and relationships are set and cannot change. As victims also hold more negative perceptions of peers, it is likely that such rigid perceptions might influence how they engage in future interactions with peers.

Three studies compared victimization types. Cyber-victims were less positive about peers than non-involved peers, but cyber-bully-victims were not (Guo et al., 2021). This finding was corroborated by Ding et al. (2020)), who found no differences on perceived social climate for non-involved, traditional bully-victims and cyber-bully-victims. A last study found that bully-victims, but not pure victims, perceived their school social climate less positive than non-involved children, whereas both victims and bully-victims perceived their peers less positive compared with bullies and non-involved peers (Bayar & Uçanok, 2012).

Only one study was fully consistent with the reaffiliation hypothesis, and found that victimization related to believing that others keep secrets and keep promises (Rotenberg & Boulton, 2013), contrasting the study of Betts and colleagues (2017). Although trust is usually a strength in social interactions, trust might lead to more opportunities to being bullied, for example when a victim tells others their secrets. These opportunities might be especially utilized if there is a motive, such as being perceived as untrustworthy (which was how others perceived victims, Rotenberg & Boulton, 2013). Contrasting the reaffiliation hypothesis, it seems unlikely that victims more readily like peers for superficial reasons, because participants did not like previously unknown teammates better than non-teammates (Telzer et al., 2020).

One study seemed in line with the desensitization hypothesis, as persistent victimization did not relate to peer perception (Ladd & Troop-Gordon, 2003). A finding that did not fit any of the hypotheses was that victimization, especially for girls, related to thinking one’s friends are susceptible to peer influence (Goldstein et al., 2020). Speculatively, this perception might relate to negative peer experiences where friends do not stand up for victimized youth and join in on negative behaviors of others in the peer group.

In conclusion, victims adopt more negative views of both their peers and relationships in general (26 studies in favor, 5 studies with nonsignificant results), though this tendency is less pronounced for friends and defenders. In addition, victims possibly think that peers cannot change (2 studies).

Understanding others

For successful interpretation, the ability to understand others is crucial. An important aspect is empathy, the ability to experience (affective empathy) and understand (cognitive empathy) feelings of others (Decety et al., 2012; Mehrabian & Epstein, 1972). Related to cognitive empathy is theory of mind (ToM). ToM involves the technical ability to ascribe mental states to others, such as feelings, intentions and thoughts, and specifically the understanding that people can have different information available and thus differ in mental states (Astington, 2001). Although victims might be trying hard to understand others (to anticipate social opportunities or threats), based on the reaffiliation and the prevention hypothesis, they actually might have less abilities to experience and understand emotions and (mental) states of others. Furthermore, based on the desensitization hypothesis, persistent victimization might lead to an even lower ability to experience and understand emotions and (mental) states of others, because of desensitization. Even more, it was expected that these victims would have a deficit in the ability to understand the mental states of others, as similar to rejected peers, they might have had fewer social opportunities to develop these abilities (Banerjee et al., 2011).

Empathy

In contrast with all three hypotheses, reaffiliation, prevention and desensitization, Table 5 reveals that most studies (29 out of 44, of which 8 studies found some significant results for either affective, cognitive or general empathy) provide no evidence for an association between victimization and empathy (e.g., Berg & Aber, 2015, see Online Source 3 for the 29 references). Moreover, a longitudinal study found no effect of within-person changes in empathy over time on victimization (Espelage et al., 2018b). Another study examined empathy (cognitive, affective, and somatic) separately for negative and positive feelings of others, and found mixed results (Chen et al., 2021). It must be noted, however, that these significant results (7 of 11 correlations), were only very small in size (r’s all < 0.07).

Table 5 Study results on empathy
Full size table

Nonetheless, some studies reported consistent associations between victimization and empathy. In studies examining specific bully-roles, traditional victims and not cyber-victims might have slightly higher levels of affective empathy, whereas only cyber-victims might have slightly higher levels of cognitive empathy when compared to outsiders or non-involved peers (Arató et al., 2020; Martínez et al., 2020). A third study, however, found that victims, compared to non-involved peers, had lower self-reported empathy (Mendoza-González et al., 2020). Some studies (9 out of 44), not comparing bullying roles, found that victimization related to less affective empathy (Chan & Wong, 2015), less cognitive empathy (Nasaescu et al., 2018; Williford et al., 2016) and less general empathy (Farrell et al., 2018; Jenkins et al., 2016, 2017; Malti et al., 2010; Pistella et al., 2020; Yudes et al., 2020). Nevertheless, other research reported a positive association of victimization (13 out of 44, of which 5 also reported some ns associations with (subtypes of) empathy), with affective empathy (Hood & Duffy, 2018; Kokkinos & Kipritsi, 2012; Nasaescu et al., 2018; Rodríguez-Hidalgo et al., 2018, 2019; van Noorden et al., 2016), cognitive empathy (Arató et al., 2020; Lázaro-Visa et al., 2019; Rodríguez-Hidalgo et al., 2018; van der Ploeg et al., 2017), and general empathy (Donat et al., 2020; Yang et al., 2020). Furthermore, one study found a between-person effect for general empathy and victimization (Espelage et al., 2018b). It is possible that the contradicting findings happened because of capitalization of chance. Studies with significant associations mostly utilized small samples or found small to very small effect sizes. The quality of studies with non-significant and significant findings was quite similar (Mns = 8.11, SD = 1.89 versus Msig = 8.91, SD = 1.54), providing support for the expectation that these results were a capitalization of chance. Furthermore, the studies with positive associations also did not differ from the studies finding negative associations, with respect to quality assessment (Mpos = 8.85, SD = 1.46; Mneg = 8.90, SD = 1.66). The studies with more power provided no or little evidence that victimization related to empathic abilities. However, there is a difference between empathic abilities and motivation to use empathic abilities. Only one study investigated motivation-related aspects and found that, although victims did not differ from non-victims in empathic abilities, they showed less cognitive and affective empathy for bullies and bully-victims than for non-involved peers (van Noorden et al., 2017). In other words, victims might be able to empathize with others, but not willing to do so with everyone.

Theory of Mind

Table 6 shows that the few studies (9) examining victimization in relation to ToM skills (i.e., the ability to ascribe mental states to others) showed diverse results. Some studies found no evidence for an association between victimization and ToM skills (4 out of 9), either first or second order false belief (Caravita et al., 2010; Espelage et al., 2018; Monks et al., 2005; Renouf et al., 2010), which is in line with the findings on empathy. Two studies found mixed results, showing evidence that victims score worse than bullies and non-involved children on cognitive, but not emotional ToM aspects (Gasser & Keller, 2009; Gini, 2006). Furthermore, three other studies indicated that victimization related to ToM skills negatively (Hsieh et al., 2019; van Dijk et al., 2017), with some evidence that a relative ToM deficit leads to future victimization (Shakoor et al., 2012). Overall, the association between ToM skills and victimization seems either absent or (negative and) small in magnitude.

Table 6 Study results on ToM
Full size table
Table 7 Abbreviations and characteristics of measurements
Full size table

Discussion

Social cognition, and hence how victims experience their social world, is likely affected by bullying, but the nature of this relation was unclear. Victims may become more sensitive to threat attempting to prevent or counteract new bullying situations (prevention hypothesis). However, children may also become more sensitive to positive cues while they try to restore their social situation (reaffiliation hypothesis). Finally, persistent victimization may cause a numbing effect on social cue processing to prevent the negative experience of victimization (desensitization hypothesis). The results of this systematic review on the literature on social cognition of victims across childhood and adolescence, mostly align with the prevention hypothesis. The review suggests that victims might have heightened attention to negative or threatening social cues, a less accurate registration of cues, and worse negative emotion recognition (encoding). Furthermore, victims perceived peers more negatively, attributed more hostile intent, and were more sensitive for rejection (interpretation). However, they did not differ in empathic skills. These results indicate that, regardless of whether the social-cognitive style is a precursor or consequence of victimization (or both), there might be a risk for negative spirals, in which negative interpretations may affect future experiences. According to SIP theory, encoding and interpretation of information and previous experiences have consequences for the displayed behavior (Table 7). As victims seem to have a negative social-cognitive style, they likely align their behavior accordingly. Indeed, victims’ social-cognitive style seems related to later aggressive behavior (Troop-Gordon et al., 2019; Yeung & Leadbeater, 2007). If victims display more negative and perhaps unexpected behavior, it could lead to a downward spiral of victimization, as such behaviors may evoke bullying (Thornberg, 2010).

Although most evidence (56.3% of all articles, and 71.4% of articles only on peer perception and attribution of situations) was in favor of the prevention hypothesis, a considerable number of studies found insignificant or mixed results. Specifically, most studies indicated that victims did not differ from non-victims in their ability to recognize emotions in general, understand others’ mental states and empathize with others’ emotions. Some methodological issues in these studies should be considered. First, the utilized tasks might be improved. Emotion recognition tasks adopted by researchers are usually quite simple, for example, using basic emotions such as anger, happiness, sadness, and surprise that can be easily differentiated, or using long presentation and response times. Moreover, the questionnaires on empathy were often self-reported. Therefore, they required considerable self-reflection on one’s own empathic abilities and possibly assessed whether children try to empathize with others instead of the ability to empathize. Last on the used tasks, ToM tasks were often quite simple, measuring first and second order ToM, which skills are usually mastered by most 5-year-olds (Perner & Wimmer, 1985), whereas the studies examined 7- to 9-year-olds. If severe problems arise in basic emotion recognition, basic empathy, or basic ToM skills, this may affect social functioning, as, for example, has been found for people with autism spectrum disorders (Trevisan & Birmingham, 2016). When no differences are found in basic or easy tasks, it does not guarantee that victims have no trouble with emotion recognition of short-lived or complex emotions, when different cues are shown simultaneously, or when people try to hide their emotions, such as in real life. A second possible methodological issue involves that in real life registration of social cues happens automatically. Having to report and explicitly think about it, might lead to different results in the lab compared with real-life social situations. Simple designs might thus not capture relevant (ecological valid) differences in victims. A last methodological issue relates to assessments of having versus using skills. The literature mostly focusses on having skills to understand others and not on whether these skills are used in social interactions. One study reported that victims did not differ from non-victims in empathic abilities, but reported lower levels of empathy for bullies and bully-victims than for non-involved peers (van Noorden et al., 2017). This finding is important as it indicates a difference between motivation and ability. Therefore, the motivational aspect should be considered when examining empathy, ToM, or social skills in further research.

While most evidence favored the prevention hypothesis, there was little evidence in favor of the reaffiliation hypothesis. To some extent this may be a consequence of few studies examining more positive aspects of SIP. Researchers examining the interpretation SIP-phase focus mainly on negative interpretations, leaving the reaffiliation hypothesis understudied. Examining positive SIP aspects can help changing toward a more positive framework, with factors for personal growth to improve victims’ resiliency, instead of focusing on (SIP) tendencies that may be detrimental. Similar to the reaffiliation hypothesis, the desensitization hypothesis is also understudied. Only two studies explicitly examined persistency of victimization. One study found that differences in longevity of victimization over five years did not relate significantly to positive peer beliefs, whereas the most recent victimization level related to less positive peer beliefs (Ladd & Troop-Gordon, 2003). The second study found that persistent victims were more sensitive to aversive stimuli than non-victims (Rudolph et al., 2016). Given the (limited) results, it cannot be discerned whether victimization recency or persistency is most influential on SIP. Studies on differences between recency and persistency of victimization in relation to SIP can also help unraveling the timing that SIP-styles take to change. Furthermore, it can help unravel whether victims perhaps move through different social-cognitive-styles as victimization persists. Initially, victims may try to reaffiliate, but when victimization does not cease, they may switch to the prevention style and if victimization persists for even longer periods, they may yet switch to the desensitization style.

Theoretical and Practical Implications

The prevention hypothesis was best supported in this review and stated that victims focus more on negative events or interpretations to be able to better prevent future negative experiences. These negative interpretations may be a precursor of initial victimization or play a role in maintaining victimization when one responds defensive or aggressive after such interpretations. Alternatively, these interpretations may be a result of previous victimization and thus be very realistic. It can be beneficial in ambiguous social situations, as this allows victims to avoid the situation and thus prevent social exclusion. Longitudinal studies examining victimization and intent attribution may provide more clarity on the causality of this relation.

Anti-bullying interventions can benefit from these insights, as social-cognitive factors are not commonly addressed in interventions (Ttofi & Farrington, 2011). Raising awareness about social-cognitive styles may break the downward spiral of victimization and result in changes in behavioral responses, as making automatic thoughts explicit is a core principle of cognitive behavioral therapy (Leahy & Rego, 2012). Moreover, as bullying is a social process and multiple parties are involved (Salmivalli, 2010), it could be beneficial to teach children how to use perspective taking skills and to acknowledge differences in perspectives (Imuta et al., 2016). This might lead to a better understanding of and more sympathy for one another, and to more prosocial behavior (Allen & Kinsey, 2013). For victims this could help to reevaluate their focus of attention, interpretations, and behavioral responses. Such changes in SIP of victims might reduce vulnerabilities or risk factors for future victimization. For non-involved peers, it could lead to better understanding of victims and higher levels of empathy, which likely leads them to respond differently.

Limitations of the Literature and Future Directions

A first limitation of the current review involves the examined victims in the included studies. Most of the time victims of bullying were examined as a homogeneous group. However, the victim group is quite heterogeneous with respect to whether victims also bully others (Lereya et al., 2015) and whether the victimization is long- or short-term (Kaufman et al., 2018). The conclusions might not hold for all types of victims, as there might be different processes at play for the different types of victims. There is some research investigating the interpretation SIP-phase of bully-victims. Some studies found that bully-victims had more negative interpretations of ambiguous scenarios than victims (Camodeca et al., 2003; Pouwels et al., 2016), whereas other studies found no evidence for a difference between victims and bully-victims (Bayar & Uçanok, 2012; Guy et al., 2017), and a last study found that it depended on the situation whether bully-victims had more negative interpretations (Ziv et al., 2013). Such studies typically did not address victimization duration, nor individual differences over time. Long-term victimization merits more attention, as it is related to indicators of both biological (Giletta et al., 2018) and mental stress (Hong et al., 2018). Moreover, long-term victims in schools with low levels of bullying (healthy context) have increased levels of mental health problems compared with long-term victims in schools with high levels of bullying (Huitsing et al., 2019). With increasingly more schools employing anti-bullying interventions, there is a strong and urgent need to identify factors that will improve interventions for long-term victims.

A second limitation of this review is that the role of development in social-cognitive abilities is hardly examined. As experience is used throughout all SIP-phases, it is expected that younger children have less experience in processing social information, and with more experience they will develop more advanced social-cognitive abilities. Consequences of victimization on the development of social-cognitive abilities can also be expected based on the desensitization hypothesis. With persistent victimization, social cognition of victims may become less focused on social cues, which may have consequences for the further development of social-cognitive abilities. The development of social-cognitive abilities is backed by research findings such as that three year-olds do not pass ToM tasks on false belief, whereas five to six year-olds do (Wellman & Peterson, 2013). In other words, perspective-taking abilities, as well as other social-cognitive abilities, such as language skills, develop with age. However, to date most research on social cognition and victims of bullying is cross-sectional, limiting the conclusions regarding causality or developmental trajectories of social-cognitive styles of (long-term) victims. Possibly, victims with a reaffiliation style can break negative spirals of sustained victimization, whereas victims with a prevention style may not be able to do so, and they might eventually develop a desensitization style over time. Without longitudinal studies such developmental patterns cannot be discerned. Similarly, without longitudinal studies it cannot be determined whether certain aspects of social cognition are precursors, consequences or both precursor and consequence of victimization. Knowledge on whether social-cognitive aspects adopted by victims result in subsequent (long-term) victimization is highly valuable for interventions that aim to increase resilience of victims. Whereas this systematic review provided evidence for a social-cognitive style that is in line with the prevention hypothesis, based on the desensitization hypothesis, long-term victims are expected to have a social-cognitive style marked by insensitivity to social cues.

A third limitation involves the examination of encoding. Studies on encoding often rely on recall and are therefore vulnerable for memory biases (alternations, enhancements, or impairments). Negative social events might be more emotionally impactful for victims than positive ones, and thus, have more leverage in recall (LaBar & Cabeza, 2006; Roozendaal & Mcgaugh, 2012). First, this might bias these results in favor of the prevention hypothesis. Second, the results might be more indicative of differences in the database and not of encoding.

A last limitation is related to the studies examining interpretation. Peer perception and attribution of situations are often examined in relation to school- or classmates, meaning that these studies examine the school-context and therefore the context in which victims are commonly bullied. Due to the limited number of studies on peer perception and situational attribution over different contexts (e.g., sport clubs, and new places), it is currently unknown whether victims also have a more negative social-cognitive style when they interact with unknown peers, are in other contexts, and whether it has consequences for future interactions (e.g., being able to trust others and invest in new relationships).

The underlying neurobiological processes of social cognition are a relevant aspect not examined in this review. Prior research shows that victims had blunted cortisol responses to stress compared with non-victims (Vaillancourt et al., 2013). Because of such altered stress responses, victims might be more sensitive to threats and therefore adopt a prevention style. Along similar lines, neuroimaging studies showed that victimized or chronically rejected children had heightened neural responses to social exclusion, compared with non-victimized (McIver et al., 2018) or non-rejected children (Will et al., 2016), suggesting stronger neural sensitivity to negative social interactions (see for an overview: Güroǧlu & Veenstra, 2021). Moreover, evidence exists that the strength of these neural responses is more strongly related to internalizing symptoms for long-term victims compared with non-victimized adolescents (Rudolph et al., 2016). In a recent large-scale study, it was found that bullying involvement related to brain structure. Victims differed in cortical thickness of the fusiform gyrus, a brain region involved in visual processing, from bullies, bully-victims and uninvolved children (Muetzel et al., 2019). Finally, it was shown that victimization experiences in childhood may modulate the relation between stress sensitivity and adolescent brain structure (Du Plessis et al., 2019), suggesting more complex relations between how early victimization experiences might influence future development. Despite evidence that victims have altered neural responses to social exclusion experiences, how these responses generalize across various processes of social cognition and differ for temporary versus long-term victims is largely unknown. Moreover, studies on neural processes of social cognition might inform on underlying motivational (Apps et al., 2016; Telzer, 2016) or regulatory processes (Sebastian et al., 2010), which might be difficult to assess using self-reports or observations of behavior. An example of possible underlying motivational processes is that victims were more empathetic toward non-involved children and friends compared with bullies and non-friends (van Noorden et al., 2017). Although motivation is hard to measure behaviorally, neural measurements might prove fruitful. Furthermore, given existing evidence on the role of neurobiological processes in atypical social cognition (e.g., seen in persons with autism spectrum disorder), a developmental neuroscience perspective is an important avenue for further research (Happé & Frith, 2014).

Conclusion

It had been unclear how victimization relates to social cognition, as many researchers only examined subcomponents of this complex construct. This systematic review reveals that victimization during childhood and adolescence relates to a stronger emphasis on negative events and a negative social-cognitive style, mostly in line with the prevention hypothesis. Victims appeared to have a heightened focus on negative or threatening social cues and represent social cues less accurately. Furthermore, they perceived peers more negatively, attributed more hostile intent and were more sensitive to rejection than non-victimized children. However, they did not differ in empathic or ToM skills. Further research is needed on the (neurobiological) development of social-cognitive styles and in particular on persistent victims and whether they, in line with the desensitization hypothesis, become insensitive to social cues because of the long period of victimization. For now, researchers and practitioners (e.g., who develop anti-bullying interventions) should realize that, in line with the prevention hypothesis, victims focus on negative social cues.