Emotional States Encountered by Law Enforcement Officers: an Initial Test of the Schema Mode Concept in Police Actions

Working with extreme mood states is a regular challenge for law enforcement officers. In their encounters during police actions, they very often face manifestations of anger and aggression, as well as vulnerable states such as ones involving fear, helplessness, or humiliation. In this study, we investigated whether officers could reliably identify emotional states in the persons they encountered, using the Schema Therapy concept of “schema modes.” Although the mode concept has been extensively validated in the field of forensic mental health, this is its first application to police work. Twenty-seven police officers working for two different German police headquarters were asked to report about a recent operation involving a person showing intense emotional states. After receiving a short training in using the iModes, cards that facilitate working with modes, they retrospectively rated the presence and intensity of the schema modes of the persons whom they encountered. Two experts in Schema Therapy independently rated the same scenarios, using the Mode Observation Scale (MOS). Intraclass correlations showed moderate to excellent levels of agreement between the officers’ and experts’ mode-ratings. Modes involving anger (70%), aggression (70%), indirect hostility (44%), and arrogance/grandiosity (41%) were the most prevalent, while modes involving vulnerable emotions were less common. It can be concluded that police officers can reliably assess different modes in encountered persons, even without extensive training. This initial study suggests that the mode concept has applicability to police work and merits additional study for its potential for de-escalation and conflict management in police contexts.


Introduction
Working with extreme mood states is a regular challenge for all police officers, from inexperienced to experienced ones (Schmalzl and Hermanutz 2012;Szymenderski 2012;Lorey and Fegert 2021). Police are often called when people discard their rule-or norm-compliant behavior and become "conspicuous, unpredictable and dangerous" (Schmalzl 2012a, p. 348). In this regard, special attention is placed on dealing with angry and aggressive individuals (Lorei et al. 2017;Oliva et al. 2010).  found that 80% of the officers faced verbal abuse in the past year. Two thirds of them experienced threats in their job. Also, one out of every three officers reported mild physical violence, and one out of five reported to already having been hit or kicked. Police officers routinely experience conflicts with a high escalation potential (Lorei et al. 2017), e.g., in cases of domestic violence (Ellrich et al. 2011) or protests, with mentally ill persons (Schmalzl and Hermanutz 2012;Lorey and Fegert 2021), or when substance abuse is involved (Ellrich and Zietlow 2012), increasing the risk of physical and psychological violence. As a result, police officers should have appropriate skills and strategies to appraise and manage critical or conflictual situations.
According to Lorey and Fegert (2021), scientifically validated models that are transferable to police work and relate to de-escalating deployment methods are lacking in police training. Operational training on police tactics and how to respond to common provocations are mostly given separately from courses on psychological issues in policing. Hermanutz and Spöcker (2012), however, suggest combined and integrative training focusing on communicative and operational, tactical skills. Officers do not have to make a differential diagnosis in encountered persons (i.e., the people whom police encounter) during their work. However, they should be able to generate a feeling for the individual's level of danger or threat through a process of precise and differentiated observation and perception (Schmalzl 2012a, see also Hahn and Kraimer, 2014). In this respect, Schmalzl (2012a) suggests considering "a strong change in the degree of activation, extreme moods and fluctuations in these extreme moods" (p. 352) as warning signals for the acute danger of a person. Equipping officers with appropriate skills and strategies to appraise and manage extremely fluctuating emotional states in conflicts with a high escalation potential may decrease the risk of injuries to officers and citizens (Olivia et al. 2010, Lorey andFegert 2021).
According to Szymenderski (2012), the focus on difficult emotional reactions and behaviors has not been sufficiently taken into account in research in police contexts. However, it has been part of considerable research in forensic psychiatric clinics to develop new methods to reduce aggression (van Wijk-Herbrink et al. 2021;Wong and Gordon 2013;Schröter et al. 2021). Within the field of forensic mental health care, practitioners are confronted with similar challenging moods and norm-violating, aggressive behavior as police officers routinely are. Although the two fields differ in their objectives, communication and de-escalation strategies are defined as key working tools for both of them (Lorei et al. 2017;Hoffmann and Wondrak 2009;Olivia et al. 2010).
One particular promising concept for working with extreme fluctuating emotional states in forensic mental health care is "schema modes" (Keulen-de Vos et al. 2017a, b;Chakhssi et al. 2014;Bernstein et al. 2012). Schema modes are a central focus of Schema Therapy, evidence-based therapy for personality disorders (Bamelis et al. 2014;Bernstein et al. 2021;van Vreeswijk et al. 2012;Giesen-Bloo et al. 2006;Nadort et al. 2009) which has proven effective with violent offenders (Bernstein et al. 2021). Schema modes reflect powerful fluctuating emotions, cognitions, and behavioral coping patterns activated at a given moment by situational triggers (Rafaeli et al. 2011). Bernstein and colleagues (2007) adapted the concept of schema modes for the forensic field, describing a variety of emotional states that are often seen in offenders, such as ones involving open displays of anger (Angry Child mode), hostile, indirect anger (Angry Protector mode), impulsivity (Impulsive Child mode), deception and manipulation (Conning Manipulative mode), reactive "hot" aggression (Bully and Attack mode), and cold, ruthless, instrumental aggression (Predator mode). The different schema modes are arranged in five mode domains. A description of the schema mode domains and the individual schema modes can be found in Table 1.
Keulen de Vos and colleagues (2016) investigated the role of schema modes and criminal behavior in a sample of 95 hospitalized violent offenders with personality disorders. Based on retrospective descriptions of crimes obtained from police reports and other sources, they assessed and compared the schema modes observed in events leading up to crimes and during the crimes, using the Mode Observation Scale (Bernstein et al. 2009). Modes involving vulnerable emotions, such as shame, abandonment, fear, or loneliness were present in 70 to 80% of situations preceding the crime. Modes involving impulsivity, anger, and aggression and threats characterized the crime itself. Furthermore, these retrospectively assessed modes were significantly correlated with scores on the Psychopathy Checklist-Revised (PCL-R; Hare et al. 1990) and predicted future acts of violence in institutional settings. These findings support the idea that schema modes play a role in criminal behavior and can be understood as an escalating sequence of emotional states, culminating in violent offenses. Bernstein (van Wijk et al. 2021;Bernstein et al. 2022a) developed a team-based form of Schema Therapy, Safe-Path, to teach personnel in forensic and psychiatric wards to respond effectively to extreme emotional states. Personnel is trained to recognize and differentiate among modes, when they occur in real-time-for example, in situations involving conflicts and provocations-and employ suitable interventions. For example, several modes involve the emotion of anger, but the appropriate intervention depends on the specific mode and its motivation or function. The Angry Child mode, for example, involves open expressions of anger, and functions as a kind of "protest" over perceived injustices or unfairness. When in this mode, individuals need to be given room to vent their anger and feel that they are taken seriously. In contrast, the Bully and Attack mode involves threats or aggression that are meant to intimidate or retaliate. When someone is in a Bully and Attack mode, giving him the chance to ventilate is likely to "throw fuel on the fire," increasing his anger, rather than reducing it. Instead, the Bully and Attack mode needs to be met with firm, clear, and consequential limits. SafePath incorporates an image-based system, the iModes (Bernstein et al. 2022a, b), with cartoon images of 23 different schema modes, to support the ability of ward staff to identify and intervene with modes in their patients.
In a recent randomized clinical trial of youth in residential care, van Wijk-Hebrink et al. (2021) found that teams trained in SafePath functioned better and created a more positive and supportive ward climate, compared to teams providing usual care in the same institution. Of particular relevance to police work, the SafePath teams made less use of severe, physical interventions (e.g., physical restraint) than the care-as-usual teams, presumably because they had learned more effective means of de-escalating situations. These findings support the idea that recognizing modes in real-time facilitates interventions to de-escalate situations.
Using the schema mode concept within police contexts is possible because of several special features that characterize this model. First, it is not limited to a therapeutic interaction with patients or people who are mentally ill. Rafaeli and colleagues (2011) clarify that all individuals, healthy persons included, show fluctuations between different schema modes. The difference between health and pathology is in the severity of emotional experiences, the transition between the modes (gradually versus abruptly), and the capacity to recognize and acknowledge one's own emotional states (Rafaeli et al. 2011). There are, in fact, many people who are not treated as patients but still have emotional and behavioral disorders and often exhibit severe psychopathology. However, not all of those who show challenging emotional states and behavior in police interventions are per se mentally ill. People who are in a crisis are likely to show modes because crises are usually characterized by highly emotional situations.
Moreover, Schmalzl (2012b) states that it is indispensable to understand the motives of encountered persons during police work, to counteract an escalation and to defuse a crisis or conflict situation. Transferring the schema mode concept to police contexts thus might help police officers to accurately appraise emotional states in encountered persons, leading to more effective interventions.

Child Modes
Involves feeling, thinking, and acting in a 'child-like' manner Vulnerable Child Feels vulnerable, overwhelmed with painful feelings (e.g., shame, humiliation). Involves Abandoned, Abused and Humiliated Child Angry Child Feels and expresses anger in response to perceived or real abandonment, or humiliation Impulsive Child Acts impulsively to get needs met Lonely Child Feels lonely and empty

Maladaptive Parent Modes
Involves internalized dysfunctional parent "voices" Punitive Parent Internalized, critical or punishing parent voice; directs harsh criticism towards the self Demanding Parent Directs impossibly high demands toward the self; never be satisfied with oneself

Dysfunctional Coping Modes
Involves attempts to protect oneself from pain through maladaptive forms of coping Avoidant Coping Modes Detached Protector Uses emotional detachment to protect one from painful feelings; is unaware of his feelings, feels "nothing," appears emotionally distant, flat, or robotic Detached Self-Soother Uses repetitive, addictive, or compulsive behaviors to calm and sooth oneself Detached Self-Stimulator Uses pleasurable or exciting sensations to distance oneself from painful feelings Angry Protector Uses a "controlled/passive wall of anger" to protect oneself from perceived threats Surrendering Complaining Protector Complains, whines, and demands in a victimized manner Compliant Surrender Gives in to the real or perceived demands of other people in a anxious attempt to avoid pain Over-Compensatory Modes Over-compensation: Involve extreme attempts to compensate for feelings of shame, loneliness, or vulnerability Self-Aggrandizer Feels superior, looks down on others; acts in a self-important manner Bully and Attack Uses threats, intimidation, aggression, or coercion to get what he wants Conning & Manipulative Cons, lies, or manipulates in a manner designed to achieve a specific goal Predator Focuses on eliminating a threat/rival in a cold, ruthless, and calculating manner Paranoid Overcontroller Attempts to protect oneself from a perceived or real threat by focusing attention, ruminating, or exercising extreme control Obsessive Compulsive Attempts to protect oneself from a perceived or real threat by using order, repetition, or ritual

Healthy Modes
Involves healthy forms of emotional expression and adaptation Healthy Adult Reflects on himself and his situation in a balanced, realistic manner. Is aware of his needs and feelings Happy Child Acts in a playful, fun-loving, free and spontaneous manner In this study, we investigated whether police officers called to crisis scenes can accurately identify schema modes in the individuals whom they encounter. To do so, we assessed the agreement between officers and experts in their ratings of the presence and intensity of schema modes observed in encountered persons. We used the iModes to facilitate officers' ability to identify modes, given the evidence of its validity and clinical utility in previous studies in community and forensic settings (Bernstein et al. 2022a, b;van Wijk-Herbrink et al. 2021). We also investigated the prevalence of the various schema modes observed by police officers during their operations. We hypothesized that (1) police officers and expert raters will show greater than chance agreement in terms of emotional states and their intensity observed in the persons encountered; and (2), consistent with the findings of Keulen de Vos and colleagues (2016), police officers will mostly report observing vulnerable and overcompensating schema modes in encountered individuals.

Sample
The sample consisted of a total of 27 police officers working for two different police headquarters (one from the state police and one from the riot police) in Germany. The riot police support the state police in their regular field of responsibility (local police services and patrol duties), but also in the accomplishment of exceptional police actions (e.g., major events, like demonstrations). Twenty males and seven females participated in this study. The age ranged from 24 to 58 years, with a mean age of 35 years (SD = 9.5). The participants' level of experience ranged from approximately 2 to 35 years.

Interview
An interview protocol was designed specifically for this study. The semi-structured questions focused on obtaining a description of a recent challenging intervention the police officer faced during work and on how he/she experienced the emotions, cognitions, and behavior of the encountered persons. Each interview started with explaining the goal, use, and scope of the study. The participants were informed that everything they said will remain anonymous and be used for research purposes only. Furthermore, the participants were asked to preserve the confidentiality of the individuals involved in the situation they will be talking about in the interview. After the informational part, the participants were instructed to choose a challenging situation as described above, which they remembered in good detail and which happened shortly before the interview, to limit the loss of valuable memories. The police officer's report was guided from the beginning of the situation of interest to the outcome of the intervention by open-ended and follow-up questions. The first question was designed to elicit basic information about the events that were rated later on (e.g., the reason for the police call). The following questions were designed to assess the behavior and emotional states they could observe in the person they were dealing with. This report was recorded and transcribed verbatim. When transcribing the audio recordings, statements about speech, noises, or sounds were removed, because they did not appear to be relevant for the schema mode ratings.
In the next part, the participants were asked to identify the schema modes they could observe in the person they were dealing with in the situation they previously described by using the iModes (Bernstein 2014). The iModes (Bernstein, 2014) are cards illustrating schema modes through visual depictions and short descriptions of the modes at the backside of the card. The officers were given these cards and asked to carefully go through all of them and pick out the modes that they observed in the encountered persons. The interviewer then recorded the answers on the interview protocol (see Appendix 1). They were then asked to arrange the cards in a sequence starting with the first mode they observed in the situation and going through it, mode by mode, until the end of the situation. The participants were invited to rate the intensity of the respective schema modes on a five-point Likert-Scale (1 = absent, 2 = mild; 5 = extremely intense).

iModes
During the interview, the iModes (Bernstein 2014) were used to support the officers' description of the sequence of schema modes observed by them. Since the concept of schema modes is a new one within police contexts, the officers were naive to this system. Therefore, the iModes were shortly introduced to them during the interview and before using them to identify the modes (see Appendix 1). iModes is a validated tool for assessing schema modes (Bernstein et al. 2022a, b) and consists of cards illustrating 23 schema modes in a way that laypeople can easily use them. A description of the modes can be found in Table 1.
The iModes were developed to facilitate the recognition of schema modes through their representation by cartoon images (see Fig. 1). The cards include a broad representation of the modes described in the clinical and research literature. The images include several different types of schema modes, including "child modes" involving the expression of basic emotions, such as sadness, fear, shame, and anger; "inner critic modes" involving self-directed criticism or demands; "maladaptive coping modes" involving dysfunctional forms of coping such as avoidance, submission, and overcompensation; and "healthy modes" involving the capacity for healthy, adaptive functioning, joy, and playfulness.
In a recent study of 366 students and 437 community participants seeking help for personality disorders (Bernstein et al. 2022a, b), participants chose the correct adjective describing the iModes images at significantly above chance levels with 52 to 98% accuracy, depending on the image. All of the iModes images showed convergent validity with a self-report measure of the same modes, the Schema Mode Inventory (Young et al. 2007), with convergent correlations ranging from r = 0.16 to r = 0.87 for male iModes images and from r = 0.23 to r = 0.87 for female iModes images, with all correlations being significant (Bernstein et al. 2022a, b). The iModes are used in clinical practice in a broad range of settings, including forensic and psychiatric hospitals, addiction centers, prisons, and ambulant treatment centers, for assessment, psychoeducation, and treatment purposes. However, to our knowledge, they have never been used for police practices.
The iModes cards include male and female versions, with the images of the modes depicted by male and female figures, respectively. The officers were free to decide whether they wanted to use the female or the male version of the iModes. After the participants identified the schema modes involved in the situation they described, they were asked to rate their intensity on a five-point Likert-Scale (1 = absent; 5 = extremely intense). This Likert Scale is the same as used in the Mode Observation Scale (Bernstein et al. 2009). The iModes were chosen for the participants to identify the schema modes, because they are widely used by workers in forensic settings, such as forensic hospitals, and are Fig. 1 Examples of iModes images, copyrighted material, David P. Bernstein easy to handle. The use of more complex assessment tools, such as the Mode Observation Scale (Bernstein et al. 2009), appeared less realistic for this concrete situation.

Mode Observation Scale (MOS)
The MOS was used to independently code the interview transcripts of the officers' reports by an expert rater in terms of the presence and intensity of the schema modes in the situations described by the officer. The expert rater had a master's degree in Forensic Psychology and had received extensive training in Schema Therapy, including assessing schema modes and using the iModes cards. The expert's ratings were made blind to the ratings made by the officers. Compared to the iModes (Bernstein, 2014), the Mode Observation Scale (MOS, Bernstein et al. 2009) offers a more detailed evaluation basis, since the descriptions of the schema modes are more extensive. The MOS includes 18 schema modes and each of these modes can easily be rated on a five-point-Likert scale (1 = absent; 5 = extremely intense). It has been developed for forensic assessment and, therefore, is specifically useful for the purpose of this study. Its inter-rater reliability has been supported and replicated in several studies. In a previous study, the inter-rater agreement (ICC, average raters) for individual schema modes ranged from 0.04 to 0.99 (median = 0.85) and for schema mode domains from 0.40 to 0.99 (median = 0.96; Keulen-de Vos et al. 2017b).
For reliability purposes, the records were rated by a second expert rater using the MOS, who also had a master's degree in Forensic Psychology and had received extensive training in Schema Therapy.

Procedure
Official approval for the study was obtained by the two police headquarters, as well as the Ethical Committee of the Maastricht University's Faculty of Psychology and Neuroscience. The study was presented by the authors at a joint meeting at one of the participating headquarters. One of the authors was already known to one of the headquarters, as a result of his work with them. Information about the study was then disseminated through a network of colleagues from different headquarters while emphasizing that participating in the study was purely voluntary. Time was made available during working hours to participate and there was no financial or other reward for doing so. The participants were again informed verbally and in writing of the voluntary nature of the study before each interview. For every participant, approximately 1 h of testing was scheduled. The interviews including the questionnaires were conducted on the premises of the police headquarters or police stations, depending on where the respective officer was working. The only inclusion criterion was the experience of a situation, in which the police counterpart reacted in a way that was challenging for the officer. These situations could have been very emotional and include extreme behavior (e.g., anger, sadness, aggressive behavior, refusal to cooperate). There was no restriction on a type of offense since not all police actions necessarily involve an offense in the legal sense. Each interview followed the same procedure.
Prior to the study, two pilot interviews were conducted to refine the interview schedule and to assure whether it was feasible to assess schema modes with police officers based on their reports and by using the iModes (Bernstein 2014). The methods proved to be useful and feasible.

Statistical Analysis
To assess the level of agreement between police officers and expert raters in terms of emotional states and their intensity observed in the police counterpart, the intraclass correlation coefficient (ICC; Koo and Li 2016;Shrout and Fleiss 1979) was calculated based on a mean-rating, absolute-agreement, two-way mixed-effects model using SPSS statistical package version 24.
Modes with high skewness and low variance were excluded from the analysis because the further the skewness is from zero, the less likely it is that the data are normally distributed and there was almost no variance to correlate (Field 2013). By using severe criteria suggested by Field (2013) to decide whether there is a significant skew (z-score greater than 1.96), too many schema modes would have to be removed from the analysis. Therefore, only modes with a skewness greater than 4 and a variance lower than 1 were excluded and a more robust measurement, Spearman's rho, was used to assess the correlations. For reliability purposes, the level of absolute agreement between two expert raters was assessed with the ICC using a two-way mixed-effects model (Koo and Li 2016;Shrout and Fleiss 1979).
To reduce complexity, neither the sequence of the schema modes nor the change in the intensity of the respective modes across the situation was considered when calculating the correlations or the ICC. For the analyses, the highest level of intensity of the mode was taken. That is, if a police counterpart showed the angry child mode once with an intensity of 5, switched into another mode, and then returned to the angry child mode showing an intensity of 3 or 4, the value 5 counted for the analyses.

Results
Both expert raters independently evaluated each of the 27 situations reported by the police officers in terms of the presence and intensity of schema modes in the police counterpart. The intraclass correlation coefficient showed a very good inter-rater agreement of 0.79 to 0.98 for the assessments of modes with the MOS. The ratings of both experts were averaged for the analyses in this study.
In the following, the results of the descriptive and correlative analyses are presented and described. First, the descriptive analysis of the schema modes observed in police counterparts will be presented, followed by the agreement between police officers' and experts' ratings.

Schema Modes in the Encountered Persons
The reasons for each police call or police action reported by the officers were various: domestic violence (5 situations), threat situations (2), (attempted) suicide (4), theft (1), general person check (3), riots or/and property damage (3), fight or rumble (5), disturbance (1), ongoing reanimation (1), minor assault (1), traffic accident escape (1). In 74% of the reported cases, the police measures resulted in an arrest of the person. In almost half of the situations (44%), either the encountered persons or the officer was (at least slightly) injured. Table 2 presents the frequency, means, and standard deviations of the modes that were rated present (scores of 2 or higher) in at least one situation by the police officers. The most frequently reported modes by the officers were the Child Modes, Avoidant Modes, and Over-Compensatory Modes. Evidence for the Angry Child Mode and the Bully and Attack Mode was found in more than two thirds of the situations, and for the Angry Protector Mode and the Self-Aggrandizer Mode in almost half of them. The following modes were rarely reported: Detached Self-Stimulator Mode, Conning and Manipulative Mode, Maladaptive Parent Modes, and Healthy Modes. No officer reported the Undisciplined Child Mode, the Demanding Parent Mode, the Obsessive-Compulsive Mode, or the Happy Child Mode with regard to the encountered person in the particular police operation.

Agreement Between Police Officers' and Experts' Ratings
Schema modes that were rated absent in all 27 situations by police officers and by experts were excluded from the analysis because there was no variance to correlate. These are the Undisciplined Child Mode, Demanding Parent Mode, Obsessive-Compulsive Mode, and the Happy Child Mode. Among the remaining 19 schema modes, some modes appeared only in a few instances, resulting in high skewness (> 4) and low variance (< 1), too (see Table 2). Therefore, the following modes were additionally removed from the analysis: Punitive Parent, Detached Self-Stimulator, Complaining Protector, Compliant Surrender, Conning and Manipulative, Healthy Adult, Angry Protector, and Paranoid Overcontroller. Table 2 shows acceptable variances for the Angry Protector and the Paranoid Overcontroller Mode reported by the police officers (iModes), but the variances for these modes when rated by experts (MOS) were 0.48 and 0.24, respectively. Consistent with the first hypothesis, estimates of the intraclass correlation coefficient showed evidence for a good level of agreement between police officers and expert raters, all being significantly greater than chance. There was an excellent agreement for the Abandoned Child Mode with an ICC = 0.97 (p < 0.01) and 95% confidence intervals (CIs) = 0.94-0.99. As expected, the results for the Humiliated Child Mode, the Detached Protector, and the Bully and Attack Mode showed good agreement between the raters (see Table 3). Estimates of the ICC for the Predator Mode, the Detached Self-Soother Mode, and the Abused Child Mode were moderate to good. For the ratings of the following modes, officers and expert raters displayed a moderate level of agreement: Lonely Child Mode, Angry Child Mode, Impulsive Child Mode, and Self-Aggrandizer Mode (see Table 3).

Discussion
The present study is the first to explore the reliability of the schema mode concept by applying it to understand emotional states and behavior in police actions. With regard to the first hypothesis, the results showed moderate to excellent levels of agreement between police officers and expert raters, all being significantly greater than chance. It can be concluded that police officers can accurately detect different modes in encountered persons, even after only a brief training in using the iModes. Although it remains unclear to what extent this is due to the instrument or the police officers' capabilities to recognize emotional states, it seems plausible that both play a role. The positive feedback obtained from police officers during the interviews supported the benefit of the iModes for police contexts. The instrument appeared to facilitate officers' abilities to consciously define and structure what they already unconsciously perceive. As such, the iModes may enhance officers' ability to recognize and act effectively when faced with challenging emotional states.
Consistent with the second hypothesis, schema modes in the vulnerable and over-compensatory domain were often rated present by police officers when being asked to report about a recent police operation, where people had reacted in ways that were challenging for them. The Angry Child Mode and the Bully and Attack Mode were the most frequently observed schema modes in encountered persons, followed by the Self-Aggrandizer, Angry Protector, and the Impulsive Child Mode. Modes involving vulnerable emotions, such as shame, fear, or sadness, were also identified by police officers, but at lower rates. That is, even though there were some identifications of vulnerable modes, their prevalences were lower.
Studies using factor analysis in adult forensic patients and youth with behavioral disorders have found that modes tend to group into externalizing and internalizing dimensions (Keulen de Vos et al. 2017a, Van Wijk-Herbrink et al. 2018. Our findings suggest that police officers appear to be more often dealing with externalizing emotional states involving anger, aggression, self-aggrandizement, and impulsivity while internalizing emotional states were less prevalent. The situations encountered by police officers appeared to be characterized by similar patterns of escalation from anger and impulsivity to states of aggression (e.g., Bully and Attack Mode) as were found in the study by Keulen de Vos and colleagues (2016). An example of how schema modes developed during a police operation and how these modes triggered violence or crimes is displayed in Fig. 2. In Schema Therapy theory, aggression often serves an overcompensatory function (Bernstein et al. 2007;Keulen de Vos et al. 2013). For example, it could be that the schema modes involving overcompensation (e.g., Bully and Attack Mode, Self-Aggrandizer Mode) observed by the officers were used to cope with intense feelings of anger, impulsivity, weakness, or helplessness (Child Modes) in the situations that had triggered them.
In their retrospective study of crimes, Keulen de Vos and colleagues (2016) found that high prevalence of modes involving vulnerable emotions (e.g., Abandoned Child mode, Lonely Child mode) during the events leading up to the crimes, but not at all while a crime was being committed. In our study, the police often arrived on the scene after crimes had already occurred. This might explain the presence of vulnerable modes involving weakness, fear, and humiliation, though at considerably lower rates than reported by Keulen de Vos and colleagues (2016), in the present study.
Our findings suggest that training in the identification of emotional states with the iModes might facilitate officers' ability to appraise and effectively manage extreme fluctuating emotional states of individuals encountered at crime scenes. This may enhance their immediate and appropriate interventions to de-escalate challenging (e.g., threatening) behavior, as well as responding effectively to other behaviors, such as those involving vulnerable emotions.
There is no doubt that the situations in question can be extremely difficult for police agents. This also includes decision-making under stress and with incomplete information (Henriksen and Kruke 2020). The results of this study raise the possibility that SafePath principles, which have been successfully implemented in forensic institutions to manage challenging situations, could also be used to enhance, and improve the officer's decision-making process. Adjusting interventions according to the emotional states of encountered persons could reduce the necessity of severe, physical interventions, and the risk of injuries on all sides. Further studies will be needed to determine whether SafePath can be adapted to the needs of police officers and the challenging situations that they encounter.

Limitations of the Present Study
The findings of the present study should be considered in the context of certain limitations. First, the assessment of the emotional states in the police counterpart is based on a retrospective design and on the police officers' subjective perception of the situation and the person. Although  Fig. 2 Example of how schema modes developed through a police operation reported in this study and how these modes triggered violence much information could be gained by using interviews, no direct access to the encountered persons' emotional states in the situations reported by the officer was possible. Second, several interviewed officers reported that they did not face a situation involving challenging and intensive emotions shortly before the interview, but rather some months ago. The passage of time may affect recollection. However, the officers were able to recall earlier situations in sufficient detail for the purpose of this study. Third, the analysis of the schema modes was only based on one source of data, namely, the reports of the officers. Other measures for the assessment of the modes could be self-reports by encountered persons, police files, or bodycams, 1 allowing for more than one source of data and for direct access to these data. Fourth, the study was based on an n of 27 participants leading to skewed data and low variance for some of the mode variables, which had to be excluded from the analyses. Furthermore, the scale level of the evaluations with the iModes and MOS (1 = absent to 5 = extremely intense) was not necessarily interval scaled, since the difference/the distance between values 1 and 2 is not necessarily the same as between values 2 and 3. Sixth, to reduce complexity, neither the sequence of the schema modes nor the change in the intensity of the respective modes across the situation was considered in the analysis. In the future, the sequence and change in intensity of schema modes in such situations of police action should be assessed. Also, the interactive dynamics between the schemas and schema modes of police officers and encountered persons should be considered in future studies.

Conclusions
This study is the first to investigate the reliability of the schema mode approach in applying it to recognize and describe the emotional states and behavior of encountered persons in police actions. The present findings suggest that police officers with only minimal training using the iModes (Bernstein, 2014) can accurately identify the schema modes in the persons they encountered on police calls, as demonstrated by high levels of agreement with expert raters. These findings need replication in larger samples, and the validity of the mode concept in police situations should be investigated, for example, by determining whether modes observed during police calls predict the likelihood of injuries to officers or others. Nevertheless, the current study suggests that police officers can be trained to accurately identify emotional states, such as ones involving anger, aggression, impulsivity, and vulnerability, that often pose challenges during police work. The schema mode approach may provide the basis for new practices for de-escalation and conflict management in police contexts, similar to those that have already demonstrated effectiveness in forensic mental health settings (van Wijk-Herbrink et al. 2021).