Psychology by BogdanovaThe Encyclopedia of Psychology for the Artificial Era

RESEARCH DOSSIER

Aggression and Social Conflict Escalation

Abstract. Aggression and social conflict escalation concern intentional interpersonal harm and the temporal processes through which disagreement becomes reciprocal hostility, coercive pressure, or an increased risk of serious injury. This fundamental research monograph reconstructs the conceptual and historical foundations of aggression research, distinguishes anger, reactive and proactive aggression, conflict, violence, and escalation, and integrates social-learning, social-information-processing, General Aggression Model, I³, dyadic, negotiation, and emotion-regulation perspectives. A critical narrative synthesis examines original experiments, behavioral and longitudinal research, psychometric methods, and quantitative reviews of hostile intent attribution and anger-regulation interventions. Evidence supports context-sensitive contributions from threat appraisal, learned scripts, arousal, rumination, incentives, and reciprocal feedback, while exposing limitations in the causal interpretation of correlations and generalization from laboratory proxies. Separate analyses address developmental and cultural variation, asymmetric power, ethical constraints, communication and institutional prevention, digital moral-outrage reinforcement, and testable hypotheses for human–artificial interaction. The synthesis reports no new experiment, clinical trial, systematic-review protocol, or independently calculated meta-analysis.

Keywords: aggression; social conflict escalation; interpersonal aggression; reactive aggression; proactive aggression; hostile attribution bias; social information processing; General Aggression Model; I³ model; social learning; reciprocal hostility; retaliation; anger and rumination; emotion regulation; conflict de-escalation; dyadic interaction; conflict resolution; behavioral observation; longitudinal methods; online conflict; Artificial Era

Aggression as an Intentionally Harmful Social Act

Aggression is behavior directed toward another person with the proximal intent to cause harm, where the target is motivated to avoid that harm. The criterion concerns the actor’s goal and the target’s interests, rather than the observer’s moral approval. Physical assault, verbal intimidation, humiliating public communication, and deliberate relational exclusion can therefore share an aggressive function while differing in modality, severity, and legal significance. Anger is an affective state and may occur without aggression; conflict concerns incompatible goals, expectations, claims, or perceptions and may proceed through peaceful negotiation. Violence describes especially severe forms of harm and has distinct practical and legal thresholds. The integrative conceptual account of Anderson & Bushman (2002) emphasizes the behavioral definition of aggression, while Dodge & Coie (1987) differentiates functions within seemingly similar acts.

This research object is the emergence, maintenance, and interruption of aggressive exchanges between persons. The unit of analysis can be an action, a provocation–response pair, a reciprocal sequence, or a sustained relationship in which earlier events alter later expectations. Distinctions between instrumental or proactive aggression and reactive aggression concern function, not mutually exclusive categories of people. A planned act may also be angry; retaliation may have expressive, deterrent, and reputational goals at once. Verbal hostility can intensify without a subsequent physical assault, so operational definitions must state the outcome studied. Dodge & Coie (1987) demonstrated the value of separating social-information-processing correlates by reactive and proactive aggression in children's peer groups.

The scientific questions are how ambiguous provocation becomes interpreted as hostility, how emotional arousal and behavioral choice interact, how one person's response changes the other's perceived threats, how environmental contingencies support escalation, and which interventions can interrupt these processes. The evidence here is assembled as a critical narrative synthesis of identifiable experiments, longitudinal and dyadic studies, measurement research, theoretical articles, and meta-analyses. There was no preregistered search protocol, independent dataset, or new quantitative pooling. Accordingly, individual experiments support bounded causal claims about their manipulations; observational relationships and synthesized associations are described according to their designs. This discipline is especially important when a laboratory measure is used as an analogy for real-world interpersonal violence rather than a direct instance of it.

Escalation as a Sequence and a Change in Interaction State

Social conflict escalation denotes a temporal progression in which interactions increase in hostility, coercive pressure, willingness to retaliate, breadth of grievance, or probability of serious harm. It is not simply a high score on trait aggressiveness or the occurrence of one severe action. An escalation event may follow a chain: an ambiguous act is perceived as a slight, an accusation signals malicious intent, a defensive reply is heard as contempt, and each response supplies new evidence for the other's adverse interpretation. A record of the sequence is needed to distinguish the initial trigger from amplifying responses. Negotiation research shows that social motives and the perceived configuration of interests alter the course and result of disagreement De Dreu et al. (2000).

Escalation should be represented by observable state variables. These include the intensity and frequency of hostile acts, changes in threat appraisal, response latency, dominance moves, willingness to accommodate, inclusion of third parties, and whether either party repairs or exits the interaction. The appropriate measurement window varies by setting: seconds in an observed argument, days across workplace exchanges, or years in a persistent relational pattern. Researchers must also distinguish escalation from recurrence: repeated arguments of similar intensity constitute chronic conflict, while a shift from disputes about one issue to mutually generalized threats constitutes a different trajectory. The transition from dissatisfaction to interpersonal harm can be influenced by regulation and instigation processes identified in Finkel & Hall (2018).

A useful process model separates initiating conditions, appraisal, affective activation, action selection, reciprocal interpretation, and subsequent reinforcement. These components form feedback rather than a universal fixed stage sequence. A provocation can change later interpretations; a conciliatory gesture can alter arousal and incentives; institutional sanctions can make retaliation less attractive. The initial stimulus may remain constant while its meaning changes through prior history and relationship power. The model therefore requires tests of temporal ordering and moderators, not just a correlation between aggressive traits and reported conflict severity. It also leaves room for conflicts in which one actor uses sustained coercion and the other responds defensively; symmetric language about mutual escalation would obscure the causal and moral asymmetry of such cases.

Historical Foundations: Social Climate, Frustration, and Modeling

Experimental inquiry into aggressive behavior developed through diverse programs rather than one founding account. In 1939, Lewin, Lippitt, and White examined aggressive behavior under experimentally organized social climates, showing that leadership and group conditions were legitimate causal questions Lewin et al. (1939). Postwar frustration-based theories connected blocked goals with aggressive readiness, while subsequent analyses criticized the assumption of a necessary one-to-one relation between frustration and harm. Berkowitz's influential reformulation Berkowitz (1989) treated aversive events as sources of negative affect and aggressive associations, with cognitive and contextual processes shaping whether aggressive behavior follows.

Observational learning expanded the field beyond immediate frustration. In the original modeling experiment of Bandura, Ross, and Ross Bandura et al. (1961), children's behavior after exposure to aggressive adult models was studied under controlled conditions. The findings supported acquisition and performance of modeled aggressive actions within that experimental context. The experiment did not demonstrate that watching every aggressive scene produces later serious violence, nor did it identify the social conditions under which a learned script is selected. Later social-cognitive frameworks gave modeling a more differentiated role by studying expectancies, outcomes, beliefs, peer reinforcement, and inhibition. Learning supplies a repertoire of possible acts; present appraisals and anticipated consequences help govern which act occurs.

By the late twentieth century, research increasingly linked individual interpretation with person–environment interaction. Dodge's early study Dodge (1980) associated patterns of social cognition with children's aggressive responding, and Crick and Dodge Crick & Dodge (1994) synthesized encoding, interpretation, goal clarification, response construction, and decision processes into a reformulated model of social adjustment. This line helped turn an apparently immediate aggressive response into a sequence of information-processing opportunities. Meanwhile, experimental and relationship research tracked reciprocal interaction and physiological states. The resulting history warrants a plural account: frustration, learned behavior, situational affordance, appraisal, and relationship feedback constitute complementary explanatory levels when their distinct predictions are tested.

Reactive and Proactive Functions of Aggressive Behavior

Reactive aggression is commonly studied as a response to perceived provocation or threat, accompanied by anger, hostile interpretation, or rapid defensive mobilization. Proactive aggression is commonly studied as a goal-directed means of obtaining resources, status, compliance, or another anticipated reward. These are functional constructs inferred from patterns of circumstances and motives. The classic peer-group investigations of Dodge & Coie (1987) found different social-information-processing correlates for the two forms, helping establish the distinction as a research hypothesis. Nonetheless, a particular aggressive act can advance instrumental goals and express retaliatory anger simultaneously. Surveys that classify a person permanently into one type can suppress within-person variation across contexts.

The reactive pathway makes the accuracy of intent attribution central. A perceived slight is often reconstructed against the background of earlier mistreatment, reputation, status concerns, or memories of betrayal. An actor may take aggressive action because the behavior seems justified or protective, even if an impartial observer sees an ambiguous provocation. In the proactive pathway, aggression may be expected to secure attention, dominance, or avoidance of costs. Learned norms and the outcomes of earlier encounters shape these expectancies. The General Aggression Model Anderson & Bushman (2002) allows personal characteristics and situational inputs to influence present cognition, affect, and arousal, which then affect appraisal and choice; the model by itself does not establish which pathway explains a particular empirical case.

Measurement requires recording both the precipitating situation and the actor's perceived aim, with triangulation where feasible. A child who pushes another after a collision, a partner who threatens to compel compliance, and an employee who undermines a rival may show overt harm with distinct intent structures. Observer ratings, reports from different actors, time-ordered interaction records, and motive-sensitive vignettes capture nonidentical features. Researchers should test whether reactive and proactive scores retain discriminant validity after accounting for overall aggression frequency, environmental opportunities, and common reporting method. A scientific taxonomy has value when it predicts which provocation, reinforcement, and intervention conditions change specific behaviors beyond what a global aggression score predicts.

Integrating Person, Situation, and Social Feedback

The General Aggression Model organizes aggression through person and situation inputs, present internal states, appraisal, and behavioral outcomes, followed by learning across episodes Anderson & Bushman (2002). The I³ framework specifies instigating triggers, impelling forces, and inhibiting capacities as separable conditions shaping action Finkel & Hall (2018). Social-information-processing models focus on encoding cues, assigning meaning, selecting goals, accessing responses, and evaluating consequences Crick & Dodge (1994). These formulations explain different portions of the causal chain, and their overlapping language must be matched to measurable variables. A provocation manipulation concerns instigation; hostile interpretation concerns an appraisal process; an aggressive behavioral choice concerns the outcome. Theories acquire comparative value when these components can be independently measured and experimentally altered.

Integrated causal accounts face a recurring problem of flexible explanation. If every aggressive act can be attributed after the fact to some mixture of threat, impulse, learned scripts, and insufficient inhibition, the model risks becoming descriptively comprehensive but weakly discriminating. Strong studies specify a contrast: when provocation is ambiguous, interpretation bias should matter more than when hostile intent is unambiguous; when social rewards for intimidation are high, instrumental motives should contribute above momentary anger; when exit is feasible, a decrease in perceived coercion may change the escalation pathway. Manipulating the predicted mechanism or showing time-specific mediation can distinguish competing claims. Trait covariates alone rarely do so.

A systematic study of social conflict must include the other participant's agency. Each actor supplies a changing environment for the other, and an action can become a new instigating condition within seconds. This makes escalation an interpersonal process with nested individual components. Static one-person experimental paradigms can test appraisal and arousal, but cannot fully explain mutual adaptation without interaction data. Analyses should therefore connect controlled causal tests of component processes to ecologically grounded dyadic evidence Gottman & Levenson (1992), negotiation evidence De Dreu et al. (2000), and developmental observations Dodge & Coie (1987). The present monograph treats these approaches as a program of cumulative tests rather than one totalizing psychological law.

Ambiguous Provocation and Hostile Intent Attribution

Hostile attribution bias is a tendency to interpret ambiguous social cues as intentionally threatening or harmful. Its psychological significance lies in the fact that many everyday encounters provide incomplete evidence about another person's motives. An accidental collision, delayed reply, disagreement about workload, or ambiguous facial expression can be interpreted as a deliberate slight when the observer expects hostility. Dodge's study of aggressive children Dodge (1980) helped establish this problem empirically, and the social-information-processing framework Crick & Dodge (1994) identifies cue encoding and interpretation as linked but distinguishable stages. In an escalating exchange, hostile interpretation can alter both the response selected and the next evidence available to the partner.

The quantitative evidence supports an association between aggressive behavior and hostile interpretation, while leaving substantial causal and contextual questions. The earlier meta-analysis by Orobio de Castro and colleagues Orobio de Castro et al. (2002) synthesized 41 studies of children's intent attributions and aggression. An expanded subsequent meta-analysis Verhoef et al. (2019) reviewed 111 studies and reported a positive association across heterogeneous populations and methods. These findings suggest a recurring psychological regularity; the reported studies differ in the ambiguity of stimuli, operational definitions of aggressive behavior, participant characteristics, and whether interpretation and behavior are measured at similar times. The association alone does not establish that biased attribution always precedes aggression, because repeated conflict may also train hostile expectations.

Research can advance by randomly varying information about the provocateur's intent while recording interpretation, immediate behavioral choice, and later reciprocal response. Conditions should distinguish clearly accidental harm, ambiguous harm, and demonstrably malicious conduct. A strong causal account would predict that corrections of inferred intent reduce reactive aggression most in ambiguous conditions, while effects are moderated by the perceived credibility of the explanation. Longitudinal observations can also test whether exposure to aggressive peers or institutions increases bias over time. The interpretive question is then dynamic: how does a person update beliefs about another actor following actual evidence of cooperation or hostility, rather than merely score high on one questionnaire?

Threatened Self, Reputation, and Honor Norms

Appraisals of respect, status, and identity can transform a low-cost disagreement into a high-stakes social threat. A public criticism may be interpreted as an attack on personal worth, competence, social standing, or group membership. Aggression can become attractive as an attempt to restore reputation or deter future disrespect. Experimental work by Bushman and Baumeister Bushman & Baumeister (1998) examined narcissism, self-esteem, ego threat, and aggressive responding under provocation. The relevant inference concerns specific interactions between characteristics and an insulting stimulus in a laboratory context. It is too broad to infer from these findings that every person with high self-regard is aggressive or that all violence results from wounded pride.

Social norms determine the costs and perceived benefits of retaliation. Cohen and colleagues Cohen et al. (1996) used experimental ethnography to study responses to insult among samples associated with the southern United States and comparison regions. Their findings illustrated how culturally patterned meanings of insult could influence behavioral and physiological responses in specific settings. Cultural differences require careful design: regions are internally diverse, experiments sample particular ages and social strata, and contemporary institutions reshape norms. A mechanistic account should specify whether the relevant variable is expectation of reputational loss, public audience, threat credibility, or learned obligation to retaliate, then test these variables independently.

Reputational threats have interpersonal feedback effects. An initial actor can intend a harmless joke while a recipient infers humiliation before witnesses; an aggressive response then supplies the initiating actor with a genuine grievance. Each party may believe restraint would signal weakness, thereby narrowing perceived response options. Observational studies should record audience, hierarchy, existing relationship, and opportunities for private repair. Theoretical attention to power Keltner et al. (2003) also reminds us that actors differ in ability to sanction, exit, or impose costs. Escalation under unequal power may reflect deliberate coercion rather than two equivalent offended parties. Precision about the direction of threats is essential to responsible explanatory and intervention models.

Anger, Arousal, and Excitation Transfer

Anger is an emotional response that often involves perceived obstruction, blame, threat, or unfairness. Its expression and behavioral consequences depend on available response options, learning history, regulatory processes, and social constraints. The integrative cognitive account of trait anger and reactive aggression by Wilkowski and Robinson Wilkowski & Robinson (2010) connects attention to hostile cues, interpretation, and effortful control. Anger must be measured separately from harmful action because an angry person may negotiate, withdraw, report abuse, or use nonaggressive protest. Conversely, calculated aggression can occur with little observable anger. Failure to maintain this distinction produces circular explanations in which aggressive behavior is used as both evidence and cause of an allegedly aggressive emotion.

Physiological arousal can influence aggressive choices when participants attribute activation to the present social situation. Zillmann's excitation-transfer research Zillmann (1971) studied how residual excitation can affect responses in communication-mediated aggressive contexts. Arousal is neither a standalone aggressive intention nor uniformly harmful; it participates in perception, response readiness, and bodily regulation. An experimental design should distinguish the autonomic response, the participant's awareness of it, and its interpretation. Heart rate and skin conductance can help describe activation but are not unique biomarkers of aggression or violence, and their relation to actual injury is limited by sampling context and outcome definition.

Emotion regulation research offers a framework for intervening at different stages of escalation. Gross Gross (1998) distinguished regulatory strategies that alter situation, attention, appraisal, or the response itself. Applied to conflict, this implies that preventing repeated provocative contact, changing the interpretation of an ambiguous act, and lowering already elevated arousal are distinct interventions. A methodological limitation arises when a regulation strategy is measured only through self-report after the argument: perceived skill may reflect general self-efficacy or retrospective justification. Researchers should record temporal sequences and test concrete manipulations. The practical aim is to identify a feasible interruption before aggressive conduct becomes reinforced by the other's next reaction.

Rumination, Retaliation, and the Interpretation of Revenge

Rumination sustains attention on an injury, grievance, or anger-inducing interpretation after the immediate situation ends. It can maintain emotional activation, increase the accessibility of hostile explanations, and prepare retaliatory plans. In an experimental comparison of catharsis, rumination, and distraction, Bushman Bushman (2002) found that rumination accompanied by anger-oriented activity did not produce the reduction in aggressive responding predicted by a simple venting model. This matters because some common conflict practices treat repeated rehearsal of grievance as intrinsically therapeutic. The precise effects depend on the form of reflection, task, participant, and outcome; constructive processing of a grievance is conceptually distinct from repetitive hostile rehearsal.

Laboratory work on the cognitive processing of negative emotions suggests that perspective and mode of reflection matter. Kross, Ayduk, and Mischel Kross et al. (2005) compared kinds of reflection that differ from immersive rumination, helping motivate experimental distinctions between abstract analysis and repeated emotionally immersive reexperiencing. Retaliation also has communicative functions: an actor may seek acknowledgement, deterrence, punishment, or evidence that the offender has understood the wrong. Gollwitzer and Denzler Gollwitzer & Denzler (2009) examined conditions under which revenge feels satisfying, indicating that interpretations of the offender's understanding can matter to the experience of retaliation. Satisfaction with revenge does not establish safety or long-term resolution.

In recurrent conflict, memory of prior episodes can become selective evidence about character and future intent. Confirmation-bias research Nickerson (1998) provides a general cognitive reason to investigate selective attention to hostile cues, although that review is not itself a causal test of violent behavior. A participant who recalls every insult and discounts repairs may strengthen a forecast of inevitable attack; the resulting hostile response may then increase the other's threat expectations. Time-intensive diaries, sequential interviews and interaction recordings can test this feedback. Relevant outcomes include duration of rumination, probability of contact, intensity of the next exchange, and capacity to consider credible counterevidence. Those outcomes should not be collapsed into one unspecified measure of conflict.

Impulse Control, Inhibition, and Behavioral Choice

Aggressive impulses compete with restraint, alternative goals, concern for consequences, and the possibility of leaving the situation. The I³ framework Finkel & Hall (2018) describes aggression through instigation, impelling forces, and inhibition. A person may experience intense provocation and still avoid harm when inhibition is available or alternatives remain rewarding. Inhibition is not simply a stable capacity: alcohol exposure, cognitive load, time pressure, exhaustion, social norms, and opportunities to exit can alter response conditions. The broad review by Denson, DeWall, and Finkel Denson et al. (2012) connects self-control research to aggressive behavior, but it does not license a single-cause explanation for all interpersonal harm.

Testing inhibition requires distinguishing ability from motivation. In an experimental task, the subject may be able to stop a harmful response yet perceive stronger incentives to retaliate. A response-latency measure captures speed or inhibition under specific instructions; it cannot independently establish the intention, seriousness, or future likelihood of interpersonal violence. Researchers can manipulate delay, alternative response options, incentives, and attention, while measuring immediate aggressive choices and their social consequences. In real interactions, the relevant regulatory resource may be an institutionalized pause, a safe exit, or a mediator rather than purely individual self-control. The inference should connect the specific intervention to the behavior it alters.

An integrative account treats aggressive action as a choice produced by appraisals, emotion, learned options, anticipated outcomes, and constraints. This makes predictions conditional. A threat can elevate readiness but yield peaceful action when protection is available; a low-arousal actor can threaten another person when coercion appears profitable. Personality evidence also demonstrates context dependence: the meta-analysis of aggressive behavior under provoking and neutral conditions Bettencourt et al. (2006) examined how dispositions relate to behavior across conditions. The explanatory priority is to identify when particular individual differences alter the response to specific situations, and then determine whether those relations persist across observation methods and environments.

Reciprocal Hostility and Dyadic Feedback

A dyadic model begins with two persons who interpret each other's behavior, make consequential choices, and revise expectations. Hostile criticism can provoke defensive correction; a defensive correction may be heard as contempt; an attempted withdrawal may be construed as abandonment or tactical evasion. The scientifically informative object is the ordered sequence, including the time between moves and the preceding context. Aggregating each participant's aggression score across an encounter erases who acted first, who repaired, which action changed the other's subsequent behavior, and whether the recorded episode was typical. Research on marital interactions Gottman & Levenson (1992) demonstrates the value of observing behavior and physiological processes rather than relying solely on retrospective descriptions of disagreement.

Reciprocity must be defined at an appropriate timescale. Immediate matching of hostility during a conversation, escalation over weeks of unresolved disputes, and enduring patterns across years are related but distinguishable phenomena. A person may produce a conciliatory move that is initially ignored but alters the longer trajectory. In a time-ordered model, an actor's behavior at one moment predicts a partner's next response conditional on prior behavior, individual tendencies, and contextual events. Such an association may still reflect a common provocation, unseen communication, or measurement effects. Video-coded sequences strengthen temporal description, whereas causal tests of a particular reply require designs that vary that reply or its interpretation.

The dyadic structure can be represented as interacting feedback loops. One loop increases hostility through adverse interpretation and retaliation; another reduces it through credible clarification, mutual problem solving, and regained predictability. The loops are often unequal because participants have different resources, safety, and exit options. A model that assumes balanced contribution is unsuitable when one individual persistently threatens, controls, or attacks another. Researchers should separate reciprocal high-conflict disagreement from coercive patterns and record the direction, seriousness, and consequences of each act. The central inferential challenge is to observe actual adaptive exchange without mistaking descriptive symmetry in a statistical model for symmetry of responsibility or power.

Close Relationships, Conflict Trajectories, and Recovery

Close relationships expose participants repeatedly to one another under conditions of emotional investment, shared obligations, and accumulated memories. Small incidents can acquire disproportionate meaning when perceived as confirming earlier grievances. The prospective study by Gottman and Levenson Gottman & Levenson (1992) linked coded marital interaction and physiological measures to later relationship outcomes, showing how interaction patterns can be investigated prospectively. Such evidence supports scientific interest in behavioral trajectories, but associations with dissolution do not establish that a single observed exchange caused a relationship to end. Sampling, shared stressors, economic conditions, and prior relationship history shape the observed relation.

Research on change in marital interaction Gottman & Levenson (1999) also highlights a crucial distinction between the average level of conflict and its development across time. A relationship may sustain frequent disagreement while maintaining respectful engagement, whereas rare encounters can contain intense intimidation. Studies should capture repair attempts, expressions of affection, contempt, interruptions, withdrawal, response delays, and the physiological states associated with different phases. These indicators require clear coding manuals and interrater reliability. Generalizing from heterosexual couples in historic U.S. samples to every kind of partnership, culture, or period requires independent replication and measurement invariance.

Conflict intervention depends on whether the partners share a safe basis for negotiation. In ordinary disagreements, structured listening, accurate restatement of concerns, and a planned pause may make cooperative alternatives available. Where there is credible fear, threats, stalking, or coercive control, safety, independent support, and accountability take priority over techniques premised on equal bargaining power. A scientific monograph must not treat victim compliance or forced reconciliation as evidence of de-escalation. Observed outward calm can conceal heightened risk. Accordingly, research outcomes should include safety, freedom to exit, recurrence, and the perspective of the threatened participant, alongside reductions in openly hostile exchanges.

Negotiation, Social Motives, and the Structure of Disagreement

Many social conflicts begin with conflicting preferences over money, work, autonomy, recognition, space, or responsibility. The choice between competitive and integrative behavior depends partly on whether actors expect their interests to be incompatible. The meta-analytic analysis of social motives in integrative negotiation De Dreu et al. (2000) demonstrates that social motivation is a relevant variable in negotiated outcomes, although studies of bargaining do not directly measure serious aggression. Negotiation and aggression overlap as research objects when perceived threats, intimidation, retaliation, and withdrawal alter the available bargaining space. A productive conceptual distinction is between a disagreement's underlying distributional problem and the hostile behaviors that may be used within it.

Escalation can result from mistaken assumptions about the other's demands, incentives to signal toughness, or an institutional arrangement that rewards public confrontation. When participants believe every concession is a defeat, even a mutually beneficial solution may be neglected. Researchers can distinguish these processes by manipulating information about counterpart preferences, the observability of concessions, repeated-interaction incentives, and the presence of an impartial facilitator. A credible compromise requires the capacity to verify commitments, especially when trust has been damaged. Measurement should track whether participants uncover joint gains, whether threats occur, and how outcomes affect future interactions, rather than simply asking whether a participant felt satisfied.

Conflict can be maintained by structural conditions even when immediate anger is low. Competition over scarce resources, asymmetric enforcement, chronic workload, and procedural unfairness can create repeated provocations. These causes may operate through appraisal and emotion, but institutional redesign may be more effective than repeatedly training individuals to regulate a situation that remains unchanged. A multilevel account records actor-specific motives, relationship histories, organizational rules, and material incentives. It makes a testable prediction: changing the rules of interaction can lower hostile behavior even without a large measured change in dispositional anger. This line of inquiry connects social-psychological mechanism tests with field interventions and negotiation design.

Developmental Pathways, Social Learning, and Peer Contexts

Childhood aggression offers a setting in which cognitive development, peer relationships, learned scripts, and situational affordances can be studied together. Children learn social responses through observation and consequences, while abilities to represent others' intentions and regulate frustration change with development. Experimental modeling work Bandura et al. (1961) demonstrated that the social environment can shape the immediate expression of specific aggressive actions. Dodge's social-cognitive investigations Dodge (1980) and later reactive–proactive analyses Dodge & Coie (1987) add hypotheses about interpretation and function. These traditions must be integrated without treating childhood laboratory behaviors as fixed forecasts of adult criminality or interpersonal violence.

Peer relationships can reinforce aggressive repertoires when intimidation brings attention, resources, or status, and can discourage them when cooperative conduct is reliably rewarded. The same visible action may have different consequences depending on classroom norms, adult intervention, and the target's response. Longitudinal research should measure changes in friendships, exposure to aggression, disciplinary practices, and opportunities for nonaggressive conflict resolution. A child reporting hostile interpretations may have experienced repeated actual hostility, so interventions should address the social environment alongside individual appraisal. The earlier and expanded meta-analyses on hostile attribution Orobio de Castro et al. (2002) Verhoef et al. (2019) show why repeated, developmentally informed tests of interpretation are warranted.

Measurement changes with age. Young children may respond to stories, pictures, or observed play; adolescents may face reputation-sensitive peer situations and digital exchanges; adults may navigate institutions, intimate partnerships, and legal consequences. Applying a single scale across these populations can mistake developmental differences in comprehension or opportunity for genuine differences in aggression. Studies should report participants' age, social context, recruitment, and the reliability of the measurement used. Strong developmental inference requires repeated assessments and appropriate separation of within-person change from stable between-person differences. Cross-age claims should rest on designs that actually include comparable measures, rather than a collage of studies using incompatible operationalizations.

Culture, Institutions, Power, and Asymmetric Risk

Social conflict is shaped by cultural expectations about respect, face, hierarchy, gender roles, and the acceptability of retaliation. The experimental ethnographic study by Cohen and colleagues Cohen et al. (1996) remains important because it operationalized a status-related prediction under a controlled insult manipulation. Its historically and geographically situated samples also show the limits of inference. A culture label should not substitute for direct measures of relevant norms, perceived reputational stakes, or situational behavior. Comparative work benefits from sampling within multiple communities, documenting institutional conditions, and testing whether an aggression instrument has the same meaning across groups.

Power changes both the opportunity to aggress and the meaning of apparent compliance. The theoretical account of power, approach, and inhibition Keltner et al. (2003) motivates hypotheses about how control over resources may alter attention to costs and rewards. A supervisor can intimidate a subordinate under formal authority; one partner may control access to housing or money; a peer group may punish withdrawal as disloyalty. These cases differ from two equally empowered participants in a laboratory game. Analyses should measure the credible consequences of dissent, availability of third-party protection, and whether one actor can safely disengage. The aim is an explanation that locates harmful conduct in its actual relational structure.

Institutional context affects escalation by defining accountability, confidentiality, and the consequences of threats. An organization that rewards performance through public humiliation may repeatedly generate harmful interactions, while credible complaint procedures can change incentives and reduce perceived helplessness. However, reporting mechanisms are only protective when retaliation is controlled. Researchers should examine actual uptake and adverse outcomes, not only whether a formal policy exists. Culture and institutions enter the psychological process through learned expectations, appraisals of danger, and response opportunities, and they also impose material constraints that cannot be represented adequately by individual self-report. This makes multilevel field studies essential to generalization.

Measurement: Questionnaires, Observational Coding, and Construct Validity

Measurement begins by specifying whether aggression refers to intentional harmful action, anger, aggressive beliefs, trait propensity, or the history of violent incidents. The Aggression Questionnaire introduced by Buss and Perry Buss & Perry (1992) provides a prominent self-report approach to several aggression-related dimensions. Its value depends on the research question and validation in the target population; a high score does not prove that a particular person committed a specific act. Theoretical labels should be attached to observable criteria rather than inferred indiscriminately from one instrument. Studies of escalation require an additional dimension: time-ordered changes in an interaction, which a retrospective trait questionnaire was not designed to capture.

Behavioral observation can code the form and intensity of hostile statements, interruptions, accusations, threats, repair attempts, withdrawal, and responses to provocation. Coding systems need operational manuals, rater training, reliability estimates, and rules for context. Raising one's voice may signify threat in one situation and urgency or auditory difficulty in another; intentional harm cannot always be inferred from an acoustic signal. Video and transcript records preserve sequence but also change participant behavior when observation is known. Valid analysis can combine participant reports of intent, partner perceptions, independent coding, and contextual information while retaining disagreements among those sources as theoretically meaningful.

Construct validity must also address common-method variance, retrospective bias, and the risk of conflating aggressive behavior with the moral evaluation of a dispute. An actor's self-report can understate wrongdoing; a threatened participant may accurately report coercion that a third-party observer misses; institutional incident counts may reflect reporting access rather than true prevalence. Psychometric work should test factor structure, test–retest reliability where appropriate, within-person sensitivity, and measurement invariance across settings and groups. For escalation, sequential indicators may capture pace, breadth, severity, and recovery. No single composite should be presented as universally valid without evidence that its components represent a common construct.

Experimental Provocation and Causal Identification

Controlled experiments manipulate insults, frustration, social cues, available rewards, or the opportunity to respond, then observe a standardized outcome. These designs help isolate component mechanisms under specific conditions. The experimental insult work of Bushman and Baumeister Bushman & Baumeister (1998) and the culturally situated provocation studies of Cohen and colleagues Cohen et al. (1996) illustrate the possibility of testing how particular threats affect immediate responding. However, a laboratory punishment choice, a hostile rating, and serious interpersonal assault differ in severity and consequences. Researchers should describe the dependent variable exactly rather than using violence as an umbrella label for every proxy of aggression.

A causal experiment requires transparent assignment procedures, adequate power for its specific hypothesis, clear predefinition of outcomes, and checks that the intervention changed the intended perception. Manipulating insult can simultaneously increase embarrassment, threat, anger, and audience concern. It therefore tests the effect of an insult package until more targeted conditions separate components. Mechanism identification benefits from factorial comparisons that vary provocation ambiguity, availability of response options, and public versus private context. The ethics of eliciting aggression also constrain design, making naturalistic and lower-risk analogues useful while limiting the outcomes they can ethically observe.

External validity depends on whether experimentally available choices resemble the social constraints and stakes of real conflicts. Participants in a brief anonymous task do not necessarily anticipate future retaliation, workplace repercussions, or threats to housing and physical safety. Field experiments can test institution-level practices when ethical, but face treatment contamination, unequal implementation, and complex outcomes. An evidentiary synthesis should report which causal link each design supports and distinguish observed behavior from hypothetical intention. Claims of universal real-world de-escalation require replication across populations and settings, not one significant contrast in a small convenience sample.

Longitudinal, Dyadic, and Time-Series Inference

Repeated observations allow researchers to test whether earlier threat appraisals predict later aggressive behavior after accounting for prior behavior, and whether experienced aggression predicts subsequent hostile interpretation. A dyadic design measures both participants' experiences and actions, allowing actor and partner pathways to be estimated. Yet a conventional cross-lagged panel model may mix stable differences between relationships with fluctuations within a particular relationship. Designs that separate these sources, include enough assessment waves, and predefine an appropriate time interval can test more precise questions. The unit of inference must match the observation process: an hourly diary cannot resolve the ordering of acts occurring within seconds.

Time-series analyses can represent transition probabilities from disagreement to hostility, escalation conditional on a prior response, the onset of recovery, and the effect of interruption. Coding a discussion at a coarse time interval can conceal brief threatening exchanges, while extremely fine temporal coding can turn normal conversational dependence into misleading statistical significance. Researchers should handle autocorrelation, missing interaction segments, varying contact frequency, and the fact that a person may deliberately exit interaction. Disengagement is informative, and treating every missing exchange as random absence biases conclusions. Relationship histories and institutional constraints may function as both confounders and moderators.

Interpretation of prospective studies remains conditional on measured and unmeasured factors. Major stress, sudden financial loss, victimization, illness, or third-party involvement can influence two people simultaneously. Fixed effects and within-person models remove some stable confounding but cannot automatically correct time-varying biases or reverse causation. Experimental micro-interventions nested within repeated observation can strengthen causal understanding when participant safety permits. Relationship work Gottman & Levenson (1992) Gottman & Levenson (1999) provides historical precedents for studying observed exchange across time, while modern designs can add preregistration, diverse sampling, validated coding, and explicit causal diagrams.

Evidence Synthesis, Effect Heterogeneity, and Scientific Limits

Quantitative reviews are particularly important because aggression studies differ in population, operationalization, setting, and exposure. The meta-analysis of hostile intent attribution by Orobio de Castro and colleagues Orobio de Castro et al. (2002) and its larger successor Verhoef et al. (2019) support a recurring positive association in childhood research. The latter synthesized 111 studies and reported an association of approximately d = .33 with a 95% confidence interval from .28 to .39, while emphasizing variation among studies. This summary is a standardized association, not a claim that changing one person's attribution automatically lowers aggressive acts by one third. The effect's meaning is bounded by the review's sampling, constructs, and methods.

Personality research likewise benefits from situational qualification. Bettencourt and colleagues Bettencourt et al. (2006) examined relations between personality and aggressive behavior under provoking and neutral conditions, a design choice that addresses whether dispositional predictors retain the same meaning across contexts. For anger management, the meta-analysis by Kjærvik and Bushman Kjærvik & Bushman (2024) drew together 154 studies and found an average reduction in anger outcomes following arousal-decreasing activities, whereas arousal-increasing activities showed no comparable average benefit. The review addresses anger and related outcomes; it should not be treated as definitive proof of reductions in severe violence. Heterogeneity, intervention duration, comparator type, and follow-up windows remain integral to interpretation.

Evidence quality should be assessed claim by claim. Direct experiments can support causal inference within their manipulations; prospective studies establish temporal relationships under assumptions; meta-analyses summarize a selected literature and may inherit its measurement limitations. Reports of statistical significance without effect magnitude, uncertainty, and replication are inadequate to establish practical importance. Strong review practice includes sensitivity to publication bias, duplicate samples, risk of bias, outcome switching, and subgroup multiplicity. Because the present work is a critical narrative review rather than a systematic review, it does not claim complete coverage of every indexed study or compute a pooled effect independently. The priority is traceable attribution and transparent alignment of conclusion with design.

Digital Communication, Outrage Reinforcement, and Escalation

Digital environments alter the speed, audience, persistence, and visibility of disagreements. A public reply can become a reputational challenge before the initial participants have clarified meaning; repeated reposting can retain an accusation after the original context is lost. These affordances can increase opportunities for conflict escalation without making every strong disagreement aggressive. The distinction between moral outrage expression and intentional interpersonal harm is especially important. Brady and colleagues Brady et al. (2021) studied how social learning can amplify moral outrage expression in online networks, connecting interaction feedback to later expression. The study is relevant to reinforcement mechanisms, but outrage posts should not be automatically classified as violence or interpersonal aggression.

Online conflict is observable through messages, engagement metrics, moderation decisions, and changes in network participation. These records permit fine-grained temporal analysis but raise interpretive difficulties: anonymity, sarcasm, deleted content, bots, unequal visibility, and selective participation can distort apparent sequences. A post attracting high engagement may be emotionally vivid or morally relevant without being a direct threat. Researchers should code intentional targeting separately from negative sentiment, and distinguish public audience effects from interpersonal reciprocity. Platform algorithms and social norms act as contextual reinforcers, yet algorithmic causation requires exposure data or exogenous variation rather than a correlation between engagement and hostile content.

Digital research also opens an Artificial Era question about interactions involving artificial conversational systems. A system can mediate, mirror, or amplify a user's hostile interpretation through its generated responses, while the psychological outcome to study may be the human participant's belief, affect, choice, or subsequent action. Operational hypotheses include whether calibrated uncertainty about intent reduces escalation and whether agreeable validation of grievances intensifies perceived threat. These hypotheses require controlled tests with transparent outcome measures, harms assessment, and conditions that isolate model behavior from selection effects. Attributing human anger or moral intent to a software system is unnecessary to test whether its outputs modify interpersonal conflict.

Interrupting Escalation Through Arousal and Appraisal Regulation

De-escalation refers to a measurable transition away from threatening or harmful conduct and toward safer, more regulated interaction. The mechanism should be specified: reduced physiological activation, revised intent attribution, interruption of rumination, a less adversarial incentive structure, or safe separation may each change a trajectory through different pathways. The meta-analytic review by Kjærvik and Bushman Kjærvik & Bushman (2024) provides evidence that arousal-decreasing activities can reduce anger on average, while activities that increase arousal do not show a comparable pooled reduction. The evidence establishes the relevance of arousal regulation to anger; it does not alone demonstrate an equal effect on interpersonal assaults, coercion, or durable relationship safety.

Interventions aimed at hostile interpretation may help when an ambiguous act has been taken as intentional harm. Their justification depends on ambiguity: when a threat is real, asking a person to reinterpret it as benign can increase danger. Accurate appraisal involves gathering credible information, recognizing the possibility of mistake, and retaining sensitivity to actual evidence of harm. Research on hostile attribution Verhoef et al. (2019) supports the importance of this construct, but associations in children do not by themselves establish adult intervention efficacy. Controlled studies should specify target populations, compare credible alternatives, and measure both immediate responses and harmful events over sufficient follow-up.

Interruption strategies can operate at different moments. Before contact, actors can arrange a facilitator or clarify procedures. During an exchange, reducing the immediate audience, taking a mutually understood pause, or establishing rules against threats can alter the interaction's available responses. After an event, perspective-sensitive reflection may reduce repetitive anger-focused rehearsal; experiments on rumination and reflective processing Bushman (2002) Kross et al. (2005) motivate this distinction. These are research mechanisms rather than universal instructions: practical choice depends on safety, consent, and ability to exit. Outcome assessment should include recurrence, severity, participant safety, and effects on third parties, alongside self-reported tension.

Communication, Joint Problem Solving, and Institutional Prevention

Communication interventions target the flow of information that sustains conflict. Asking participants to identify the concrete issue, distinguish observation from intention, and state proposals can reduce ambiguity when both parties can safely participate. Joint problem solving addresses perceived incompatibility of outcomes; the negotiation evidence synthesized by De Dreu and colleagues De Dreu et al. (2000) offers a theoretical basis for investigating cooperation and social motives. A successful encounter requires more than a reduction in visible argument: it may involve an agreed allocation, reliable future rules, or a decision to disengage peacefully. Researchers should therefore measure implementation and longer-term recurrence.

At the institutional level, prevention can modify escalation opportunities through clear rules, protected reporting, prompt intervention for threats, meaningful review of grievances, and avoidance of public humiliation. Such measures influence both incentives and appraisals of legitimacy. But a nominal code of conduct does not establish effective enforcement, and individual emotion-regulation training cannot compensate for a setting where retaliation is rewarded. Cluster-randomized interventions, interrupted time series, and carefully matched longitudinal comparisons can evaluate whether changes in procedures affect recorded harms. Reporting rates may initially rise when safer disclosure becomes available, so an increase in reports should not automatically be treated as increased incidence.

The design of conflict-resolution programs should distinguish training outcomes from actual behavioral outcomes. Participants can learn negotiation vocabulary without changing how they behave under provocation. Laboratory improvements need replication under conditions that involve real stakes, repeated contact, and potentially asymmetric power. Meaningful endpoints include frequency of threatening acts, sustained access to safe exit, independent reports of coercion, and the durability of agreements. Studies should also evaluate adverse effects, including forced contact with an aggressor or retribution after a complaint. The defining practical standard is safer conduct and improved options, measured over time rather than inferred from short-term satisfaction.

Safety, Coercion, and Ethical Boundaries of Conflict Research

Some harmful encounters are better explained by organized coercion, persistent intimidation, or control over resources than by reciprocal emotional escalation. A participant's silence or apparent acquiescence may reflect threat rather than successful conflict management. The conceptual boundaries of this dossier therefore require differentiation between ordinary conflict, aggressive escalation, and patterns of coercive violence studied in adjacent clinical, legal, and family-psychology research. The current synthesis analyzes harmful behavior and its immediate social dynamics without providing individual diagnoses or substituting for specialized risk assessment. It places participant safety at the center of research design when real-world harms are possible.

Ethical experiments cannot recreate the full consequences of serious interpersonal victimization. Even when behavioral proxies are permitted, researchers must minimize distress, use transparent oversight, and explain the limits of external validity. Observational work with couples, children, online communities, or institutional disputes requires appropriate consent, confidentiality protection, and protocols for discovering credible danger. Joint interviews can be unsafe in coercive relationships, and disclosure may expose a participant to retaliation. The method itself can alter risk. Ethics consequently forms part of scientific validity: a sample containing only those able to participate safely may systematically exclude people facing the gravest forms of aggression.

A safety-centered interpretation distinguishes behavioral restraint, credible conflict resolution, protection of a threatened person, and accountability for harmful conduct. Each can change risk by a different mechanism. Evaluations should avoid defining success solely by increased reconciliation or reduced complaining, because both may occur under coercive pressure. Where threatened parties face limited options, credible separation, independent support, and formal protections may be crucial components of a safer trajectory. Research should report the circumstances under which findings can be transferred, especially when interventions were tested only in low-risk voluntary disagreements. Practical guidance must remain proportional to the evidence and stakes.

Scientific Disagreement, Replication, and Future Research

Major explanatory tensions concern the relative causal contribution of biased interpretation, anger, rewarding consequences, self-control, and structural constraints. Hostile attribution could precede reactive aggression, follow repeated victimization, or participate in a bidirectional process. A high correlation between measures collected from one informant at one time cannot resolve these alternatives. Similarly, success of a de-escalation intervention may reflect reduced arousal, increased perceived safety, reduced exposure to the provoking person, or demand characteristics. Research needs competing predictions and credible controls rather than retrospective assignment of every outcome to a broad integrative framework.

Replication must examine both a mechanism's stability and the transportability of its measures. Classic laboratory provocation studies often use narrowly selected samples and low-stakes outcomes; studies of relationship conflict may be restricted to one type of partnership; digital studies may depend on a platform's historically specific feedback design. Preregistered experiments, open coding manuals, independent replications, and longitudinal multisite field research can identify boundary conditions. Robust designs should report null findings, effect sizes, confidence intervals, missing data, and outcome definitions. Review practices also need to check whether the same participant sample appears in multiple publications and whether coding decisions change the pooled estimate.

An evidence-based program should prioritize at least five questions. First, when does correcting a misperceived threat change actual conduct, and when is threat perception accurate? Second, which transition points in interaction sequences best predict later harm? Third, how do power and safe exit moderate the apparent benefits of mediation? Fourth, how do social norms and reinforcement systems alter aggressive scripts across childhood, institutions, and digital environments? Fifth, how can researchers distinguish durable safety from short-term suppression of visible conflict? These questions invite experiments nested in longitudinal observation, validated behavioral sequences, and ethically grounded assessment of interventions in the populations for whom they are intended.

Artificial Era and the Scientific Synthesis of Escalation

The Artificial Era extends the environments in which interpersonal conflict develops. Human participants increasingly receive social interpretations, summaries, and suggested replies from artificial conversational systems. Such systems may change what is noticed, how confidently hostile intent is inferred, and which alternative responses become salient. The research object remains identifiable through observed human behavior, system outputs, exposure conditions, and measured downstream effects. The social-learning evidence on digital moral expression Brady et al. (2021) motivates hypotheses about reinforcement of contentious responses, but it does not directly establish the effect of a specific artificial mediator on aggression. Those claims require new experiments and robust safety monitoring.

An experimentally testable design could randomize conflict participants to carefully controlled assistance conditions: neutral summarization of both positions, uncertainty-calibrated interpretation, noncoercive procedural guidance, or ordinary unassisted drafting. Researchers would measure interpretation accuracy, behavioral hostility, misunderstanding repair, willingness to withdraw safely, and later recurrence. Safety screening should exclude or separately address situations involving credible threats, coercion, or acute risk. Another design could test whether a system's uncritical validation of an unverified grievance raises confidence in hostile intent; this is a hypothesis about the effects of outputs on human judgment, not an assumption that the system has human anger or personal motives. The relevant model may be evaluated as a participant in a socio-technical causal chain.

The scientific synthesis of this dossier is a multilevel dynamic account. Aggressive escalation emerges when harmful behavioral options become accessible and rewarding under appraisals of threat or obstruction, affective mobilization, learned norms, and constraints on inhibition; each act then changes the partner's information and subsequent incentives. Historical experiments support particular social-learning and provocation mechanisms Bandura et al. (1961) Bushman & Baumeister (1998), social-cognitive work clarifies interpretation Crick & Dodge (1994), quantitative syntheses establish recurring but heterogeneous associations Verhoef et al. (2019), and dyadic research demonstrates the need to analyze interaction across time Gottman & Levenson (1992). Effective inquiry requires explicit behavior definitions, attention to asymmetry and safety, temporally adequate designs, and independently verified outcomes. This is the conclusion of a critical synthesis of existing evidence, not a claim to an original experimental discovery.

Conclusion

Aggression and social conflict escalation are intelligible as temporally organized interactions between harmful behavioral choices and the conditions that make them appear necessary, rewarding, or justified. The evidence connects several mechanisms: observational acquisition of aggressive behavior Bandura et al. (1961), hostile appraisal under ambiguity Verhoef et al. (2019), emotional and cognitive readiness Anderson & Bushman (2002), self-control and inhibition Finkel & Hall (2018), and reciprocally changing relational circumstances Gottman & Levenson (1992). These findings become more informative when each construct is measured independently and the transition between stages is observed. A complete explanation therefore needs both the psychology of individual action and the changing ecology of the encounter.

Empirical confidence is strongest where the operational question and method align. Experiments identify some responses to provocation and social cues; quantitative reviews document recurring associations and intervention effects within defined outcomes; longitudinal and observational work establishes temporal regularities while facing confounding and sampling limits. The 2019 hostile-attribution meta-analysis Verhoef et al. (2019) and the 2024 arousal-regulation meta-analysis Kjærvik & Bushman (2024) provide important cumulative evidence with different causal reach. This literature supports careful differentiation among anger reduction, reduced retaliation, nonaggressive negotiation, and sustained freedom from harm. A successful safety outcome cannot be inferred solely from outward quiet, willingness to reconcile, or immediate satisfaction with a conversation.

The research agenda is to identify which behavioral sequences make escalation probable, which contextual changes interrupt them, and for whom those changes are safe. High-resolution dyadic observation, ethically controlled component experiments, cross-cultural psychometric validation, and prospective evaluation of actual harmful outcomes are complementary methods. Digital and human–artificial interaction introduce additional hypotheses about amplification and correction of hostile interpretations, to be tested without assuming person-like motives in computational systems. The scientific horizon is a cumulative, reproducible account of how conflict becomes aggression, how it can be interrupted, and how power, history, and institutions condition both processes.

References

  • Anderson, C. A., & Bushman, B. J. (2002). Human aggression. Annual Review of Psychology, 53, 27–51. https://doi.org/10.1146/annurev.psych.53.100901.135231
  • Bandura, A., Ross, D., & Ross, S. A. (1961). Transmission of aggression through imitation of aggressive models. Journal of Abnormal and Social Psychology, 63(3), 575–582. https://doi.org/10.1037/h0045925
  • Berkowitz, L. (1989). Frustration-aggression hypothesis: Examination and reformulation. Psychological Bulletin, 106(1), 59–73. https://doi.org/10.1037/0033-2909.106.1.59
  • Bettencourt, B. A., Talley, A., Benjamin, A. J., & Valentine, J. (2006). Personality and aggressive behavior under provoking and neutral conditions: A meta-analytic review. Psychological Bulletin, 132(5), 751–777. https://doi.org/10.1037/0033-2909.132.5.751
  • Brady, W. J., McLoughlin, K., Doan, T. N., & Crockett, M. J. (2021). How social learning amplifies moral outrage expression in online social networks. Science Advances, 7(33), eabe5641. https://doi.org/10.1126/sciadv.abe5641
  • Bushman, B. J. (2002). Does venting anger feed or extinguish the flame? Catharsis, rumination, distraction, anger, and aggressive responding. Personality and Social Psychology Bulletin, 28(6), 724–731. https://doi.org/10.1177/0146167202289002
  • Bushman, B. J., & Baumeister, R. F. (1998). Threatened egotism, narcissism, self-esteem, and direct and displaced aggression: Does self-love or self-hate lead to violence? Journal of Personality and Social Psychology, 75(1), 219–229. https://doi.org/10.1037/0022-3514.75.1.219
  • Buss, A. H., & Perry, M. (1992). The aggression questionnaire. Journal of Personality and Social Psychology, 63(3), 452–459. https://doi.org/10.1037/0022-3514.63.3.452
  • Cohen, D., Nisbett, R. E., Bowdle, B. F., & Schwarz, N. (1996). Insult, aggression, and the southern culture of honor: An experimental ethnography. Journal of Personality and Social Psychology, 70(5), 945–960. https://doi.org/10.1037/0022-3514.70.5.945
  • Crick, N. R., & Dodge, K. A. (1994). A review and reformulation of social information-processing mechanisms in children's social adjustment. Psychological Bulletin, 115(1), 74–101. https://doi.org/10.1037/0033-2909.115.1.74
  • De Dreu, C. K. W., Weingart, L. R., & Kwon, S. (2000). Influence of social motives on integrative negotiation: A meta-analytic review and test of two theories. Journal of Personality and Social Psychology, 78(5), 889–905. https://doi.org/10.1037/0022-3514.78.5.889
  • Denson, T. F., DeWall, C. N., & Finkel, E. J. (2012). Self-control and aggression. Current Directions in Psychological Science, 21(1), 20–25. https://doi.org/10.1177/0963721411429451
  • Dodge, K. A. (1980). Social cognition and children's aggressive behavior. Child Development, 51(1), 162–170. https://doi.org/10.1111/j.1467-8624.1980.tb02522.x
  • Dodge, K. A., & Coie, J. D. (1987). Social-information-processing factors in reactive and proactive aggression in children's peer groups. Journal of Personality and Social Psychology, 53(6), 1146–1158. https://doi.org/10.1037/0022-3514.53.6.1146
  • Finkel, E. J., & Hall, A. N. (2018). The I³ model: A metatheoretical framework for understanding aggression. Current Opinion in Psychology, 19, 125–130. https://doi.org/10.1016/j.copsyc.2017.03.013
  • Gollwitzer, M., & Denzler, M. (2009). What makes revenge sweet: Seeing the offender suffer or delivering a message? Journal of Experimental Social Psychology, 45(4), 840–844. https://doi.org/10.1016/j.jesp.2009.03.001
  • Gottman, J. M., & Levenson, R. W. (1992). Marital processes predictive of later dissolution: Behavior, physiology, and health. Journal of Personality and Social Psychology, 63(2), 221–233. https://doi.org/10.1037/0022-3514.63.2.221
  • Gottman, J. M., & Levenson, R. W. (1999). What predicts change in marital interaction over time? A study of alternative models. Family Process, 38(2), 143–158. https://doi.org/10.1111/j.1545-5300.1999.00143.x
  • Gross, J. J. (1998). The emerging field of emotion regulation: An integrative review. Review of General Psychology, 2(3), 271–299. https://doi.org/10.1037/1089-2680.2.3.271
  • Keltner, D., Gruenfeld, D. H., & Anderson, C. (2003). Power, approach, and inhibition. Psychological Review, 110(2), 265–284. https://doi.org/10.1037/0033-295X.110.2.265
  • Kjærvik, S. L., & Bushman, B. J. (2024). A meta-analytic review of anger management activities that increase or decrease arousal: What fuels or douses rage? Clinical Psychology Review, 109, Article 102414. https://doi.org/10.1016/j.cpr.2024.102414
  • Kross, E., Ayduk, O., & Mischel, W. (2005). When asking “why” does not hurt: Distinguishing rumination from reflective processing of negative emotions. Psychological Science, 16(9), 709–715. https://doi.org/10.1111/j.1467-9280.2005.01600.x
  • Lewin, K., Lippitt, R., & White, R. K. (1939). Patterns of aggressive behavior in experimentally created “social climates.” The Journal of Social Psychology, 10(2), 269–299. https://doi.org/10.1080/00224545.1939.9713366
  • Nickerson, R. S. (1998). Confirmation bias: A ubiquitous phenomenon in many guises. Review of General Psychology, 2(2), 175–220. https://doi.org/10.1037/1089-2680.2.2.175
  • Orobio de Castro, B., Veerman, J. W., Koops, W., Bosch, J. D., & Monshouwer, H. J. (2002). Hostile attribution of intent and aggressive behavior: A meta-analysis. Child Development, 73(3), 916–934. https://doi.org/10.1111/1467-8624.00447
  • Verhoef, R. E. J., Alsem, S. C., Verhulp, E. E., & Orobio de Castro, B. (2019). Hostile intent attribution and aggressive behavior in children revisited: A meta-analysis. Child Development, 90(5), e525–e547. https://doi.org/10.1111/cdev.13255
  • Wilkowski, B. M., & Robinson, M. D. (2010). The anatomy of anger: An integrative cognitive model of trait anger and reactive aggression. Journal of Personality, 78(1), 9–38. https://doi.org/10.1111/j.1467-6494.2009.00607.x
  • Zillmann, D. (1971). Excitation transfer in communication-mediated aggressive behavior. Journal of Experimental Social Psychology, 7(4), 419–434. https://doi.org/10.1016/0022-1031(71)90075-8