In principle, there are two types of definitions for “aggression”: one focuses on external appearance (i.e., the “form” of behavior), while the other focuses on its functions. “Behavioral form” or “phenomenological” definitions—highly preferred in the 1960s and 1970s—are usually restrictive but have the advantage of being simple and easily applicable to both animals and humans. In plain terms, aggressiveness can be defined as a behavior that delivers noxious stimuli to another organism (Buss 1961) or somewhat more explicitly: any behavior directed towards the goal of harm and injury (Baron 1977; Zillman 1979). These clear-cut definitions immediately raise a simple but relevant question: why should one organism deliver “noxious stimuli” to another? The explanation is provided by the other type of basic definitions, which attribute functions to this behavior. Functional definitions are likely based on the common observation that aggressiveness is present in any species that is endowed with the capacity of performing it, from invertebrates to humans.² One cannot reasonably assume that aggressive behavior survived evolutionary pressures for hundreds of millions of years in so many species without providing some functional advantages. It is generally believed that aggression is the most efficient form of competition that bridges the gap between the limitedness of resources on one side and the individual needs of survival and the more general “evolutionary need” of leaving as many descendants as possible on the other side. Thus, the proximate function of aggression is resource competition that ensures survival, while the ultimate function is reproduction that ensures contribution to the gene pool of the next generation. This dual goal is achieved by fights over food, territory, mating partners, etc., and by protecting offspring by aggressive means. This utilitarian view on aggression was formulated very early (Darwin 1871) and was maintained ever since (see Archer 2009; Lindenfors and Tullberg 2011; Weiger and Bear 1988 as examples).

Thus, individuals harm each other to get access to resources and to ensure gene transmission over generations in a world where resources are limited and the delivery of harm is more efficient than negotiation.³ At the first sight, the functional definition applies to animals only, because survival and the ability to leave progeny rarely depend on aggression in humans. Human societies consist of highly interdependent individuals, which are expected to negotiate rather than fight for resources. Still, human aggression is often gain oriented; this type of human behavior is commonly called instrumental aggression (Feshbach 1971). Moreover, particular forms of aggression (usually considered legitimate) serve survival in the form of self-defense, the defense of offspring, or the defense of others (Archer 2009; Björkqvist et al. 2000). Furthermore, intricate forms, e.g., relational and indirect aggressions, are means of social competition that are employed at a wide scale in human societies. Because human aggression may become a matter of survival and inclusive fitness under certain conditions, one should attribute functionality to human aggression as well. Taken together, these considerations show that (1) aggression is omnipresent in arthropods and vertebrates and is shown by all lower animals that are endowed with the capacity of performing it and (2) it confers competitive advantage in animals and has similar roles in people under particular conditions.

A final clarification that intensional definitions need refers to the word “harm.” In his influential work on aggression, Lorez (1963) notes that “crude” forms of aggression (i.e., the delivery of injuries) gradually progress into “symbolic” actions that fulfill the function of communication. The reason is that aggressiveness involves considerable risks for both the recipient and perpetrator. Either of the fighting parties can suffer injuries, and both lose energy reserves that may become vital in a subsequent challenge (e.g., predator attack) (Haller 1995). Therefore, injurious behaviors not only are gradually replaced with but are also preceded by less risky behaviors, e.g., threat signals that offer the chance of withdrawal before the actual fight is started. Such forms of behavior are integrant parts of the aggressive repertoire of animals (see van Staaden et al. 2011 for a review). Human aggression also entails threats, e.g., threatening facial expressions and body language, verbal warnings, etc. It is generally considered that threatening with physical harm is an aggressive act per se. On the other hand, harm should not always be understood in a physical sense. Manipulating social environments (e.g., by malicious gossiping, ignoring, and coercing) may hurt a person in a nondirect manner and may ultimately disrupt his/her social status (Bjorkqvist and Niemela 1992; Coccaro et al. 1997a, b; Galen and Underwood 1997; Gomes 2007; Paquette and Underwood 1999). Social manipulation is not uncommon in animals; the main behavioral tool is social exclusion in their case (see Sasaki and Uchida (2012) for a review and Goodall (1986) for studies in chimpanzees). Therefore, a more complete definition should make reference not only to harm and injury but also to threatening with these; in addition, it should consider “harm” in both physical and psychological terms (Björkqvist et al. 2000; Brower and Price 2001). Further specifications of the word “harm” include intentionality (to exclude accidental events) and exclude benefits for the victim (Berkowitz 1993, 1994; Björkqvist et al. 2000; Bushman and Anderson 2001; Feshbach 1971). As such, aggression includes neither accidental harm nor harm inflicted for positive ends (e.g., to discourage social parasites, to discipline offspring, or to maintain cooperative behavior; Clutton-Brock and Parker 1995). To avoid an impossibly complex sentence, we define aggression below by pointing out its essence and adding modifiers.

Aggression is a behavioral means of resource competition that is based on the delivery of or threatening with harm. Harm should be understood in both physical and psychological terms, should be intentional, and should benefit the victim neither directly nor indirectly.

The definition adopted above shows that aggression is not a unitary phenomenon and cannot be the object of studies in such general terms. Therefore, aggression was divided into several subtypes along a variety of classification principles. The form of aggression is one such principle; in humans, subtypes are called physical, verbal, and indirect aggressions (Bjorkqvist and Niemela 1992; Gomes 2007; Paquette and Underwood 1999; Vaillancourt and Sunderani 2011). Physical aggression involves physical harm; verbal aggression (including body language) covers threats that aim at intimidating partners, while indirect aggression consists of social manipulation that inflicts psychological harm in most cases. In some papers, indirect aggression was restricted to or included aggression towards objects (Coccaro et al. 1997a, b; Crowe 1974). This approach was considered outdated by papers published later (Gomes 2007). We suggest here that aggression against objects (e.g., slamming doors) should be considered part of “verbal” aggression when its aim is intimidation and displaced aggression when it serves to channel off emotions. This issue will be detailed below. Rodent analogs of these behavioral categories include biting attacks (physical aggression), offensive threats (verbal aggression), and social exclusion (indirect aggression). Phenomenologically delimited subtypes can be graded according to their precise forms and the severity of consequences. For example, fighting with bare hands is considered milder than weapon use; injuring is a lesser form of aggression than killing (Björkqvist et al. 2000; Holm 1982). Similarly, body language and verbal threats may be perceived as graded versions of “verbal” aggression. Finally, indirect aggression can be divided into direct (conspicuous) and hidden forms (Gomes 2007).

The reason and emotional background of aggression can also serve as criteria for typifying this behavior. Emotional and instrumental forms are the most widely used categories (Blair 2001; Feshbach 1971; Kempes et al. 2005; Lopez-Duran et al. 2009; Raine et al. 2006; van Bokhoven et al. 2005; van Honk et al. 2010). The former (often called reactive) consists of uncontrolled aggressive outbursts that are elicited by perceived threat or provocation. This type of aggression is characterized by high levels of bodily arousal. Instrumental aggression (also called proactive) is controlled and goal directed. It is usually associated with low levels of arousal. Emotional and instrumental aggressions are associated with different types of criminal aggression and aggression-related psychopathologies, as discussed below. There are several rodent models of abnormal aggression, where important aspects of emotional (reactive) and instrumental (proactive) aggression were mimicked (see below). A specific form of emotional responses is displaced aggression; this behavior is directed towards a third party under conditions of frustration and distress. While this behavior is commonly studied in people (Denson et al. 2006; Fitz 1976; Kaufmann and Feshbach 1963; Pedersen et al. 2000; Reijntjes et al. 2007), similar studies in animals are sparse although displaced aggression is readily shown by vertebrates from fish to rats (Overli et al. 2004; Nephew et al. 2003; Hoffmann et al. 1987).

Other frequently used categories are overt and covert aggressions. These may be perceived and defined in different ways. Popma et al. (2007a), for instance, use the phrase “overt aggression” as a synonym of physical and verbal aggressions combined, while “covert aggression” depicts emotional components in their terminology (e.g., hostility, angriness, jealousy, etc.). In other papers, “overt aggression” includes physical fights, weapon use, cruelty, robbery, and sexual abuse, while “covert aggression” is defined as housebreaking, steeling, fire setting, lying, truancy from home, etc. (van Bokhoven et al. 2005). The phrases “overt” and “covert” are sometimes perceived as synonyms of physical and indirect aggressions, respectively (Vaillancourt and Sunderani 2011).

The brief review of various classifications shows that the terms are not used consistently, and there are overlaps between various categories. For example, indirect aggression is instrumental in many cases; covert aggression is considered synonymous with hostility and anger according to one categorization and with nonviolent lawbreaking according to the other. Any combination is possible between emotional (reactive) and instrumental (proactive) aggressions on one side and physical and verbal aggressions on the other. In addition, such subcategories do not categorize perpetrators but actions: one and the same person may show emotional aggression under certain circumstances and instrumental aggression under others. Despite inconsistencies and overlaps, these concepts are very useful for differentiating and studying distinct classes of behaviors.

Another criterion that may be used to typify aggression relates to its normality. Given the importance of this criterion for the present work, it will be presented in a separate section.

As shown above, aggression is ubiquitous in animals where it plays important functions. For this reason, abnormality may be perceived unlikely as far as animals are concerned. On the other hand, the high level of socialization and the strong moral, legal, and social stigmatization of aggressiveness may render aggression per se abnormal in humans. In simple terms, one may be tempted to believe that aggression is always normal in animals and is always abnormal in people. Neither assumption seems true.

Conflicts in animals are settled according to certain rules, which result from evolutionary pressures against dangerous forms of competitiveness (Hinde 1974). Intraspecific aggression is not aimed at inflicting lethal injuries but at ensuring access to resources. Species would be rapidly eliminated if its individuals risked their lives to obtain food, mating partners, etc. Moreover, the near-emptying of biochemical energy stores by excessive fighting per se may put the life of animals at risk (Haller 1995). Evolution shaped aggression such that these risks are avoided. Aggressive intent is signaled in advance to allow withdrawal, the level of aggressiveness is kept between certain limits to minimize risks and excessive energy losses, attacks are targeted towards body parts that lack vital organs and are not excessively exposed to infections, aggressiveness is rapidly deferred by submission or flight by the opponent, and, finally, females (rate-limiting factors in reproduction) are largely protected from attacks by males (not considering sexual coercive acts). Deviations from these rules capsize the balance between costs and benefits, render aggression nonfunctional, and provide a handle for differentiating normal and abnormal aggressions in animals (Haller et al. 2001, 2005a; Haller and Kruk 2006). Thus, abnormal aggression is defined in terms of natural rule breaking and the lack of functionality in animals.

Human aggression is considered abnormal if it violates moral and legal rules or if is associated with psychopathologic conditions. These two principles (morality, law and mental health) are closely related. An interesting study by Björkqvist et al. (2000) demonstrated that people base their attribution of aggression on factors that are consistent with the severity of punishment in sentences for different crimes of violence; thus, moral and legal thinking is closely related as far as aggression is concerned. Significant overlaps were also found between the behavioral and legal correlates of aggression as well as between aggression-related psychopathological constructs and criminal behavior (Coccaro and McNamee 1998; Lee and Coccaro 2001; Gao and Raine 2010; Yang et al. 2008). Moreover, the characterization of aggression-related psychopathologies by DSM-IV-TR explicitly includes rule breaking among the symptoms of conduct, oppositional-defiant, and antisocial personality disorders and implicitly in the case of others (e.g., intermittent explosive disorder) (American Psychiatric Association 2000). Overall, this source considers psychological conditions pathological if they inflict suffering; cause significant impairment in social, occupational, or other important areas of functioning; and are persistent. Persistent violent criminal behavior fulfills these criteria.

In legal terms, the key factor that differentiates normal and abnormal aggressions is rule breaking. For example, violence was recently defined as the infliction of physical harm “in violation of social norms” (Brower and Price 2001). In psychopathological terms, aggression is abnormal if it makes the individual dysfunctional. Persistent aggressiveness leads to social exclusion, to the loss of jobs, to destruction of families, etc.—briefly, it imperils social, occupational, and other important areas of functioning. Thus, legal and psychopathological delimitations of human abnormal aggression, when taken together, replicate the definition of abnormal aggression proposed above for animals; human abnormal aggression can be defined in terms of rule breaking and dysfunctionality. Differences between animals and humans lie in details only. For example, rules are shaped by evolutionary pressures in animals, while human rules are formalized. The nature of punishment may also be dissimilar; abnormally aggressive animals may take punishment in the form of defeats when meeting opponents of superior fighting abilities, while humans risk the same and legal punishment in addition. Even the very long-term consequences appear comparable; excessive increases in the share of abnormally aggressive individuals imperil the success of animal species on an evolutionary scale; in the case of humans, the same jeopardizes the success of societies on a historical scale. Given these close parallels, we define below abnormal aggression by summarizing all the considerations outlined above.

Abnormal aggression breaks natural or formalized rules and is dysfunctional on the long run in terms of negative consequences for the perpetrator. This definition implies that certain forms of aggression are not abnormal in humans. Indirect aggression, for instance, does not break the law and may reach its function if carried out skillfully (i.e., it may confer competitive advantage). The same holds true for verbal aggression in many instances. Even physical aggression may be normal if performed by authorized bodies (e.g., law enforcement) or if lawful (e.g., self-defense). On the other side, some forms of aggression may be abnormal in animals. In the laboratory, such abnormal forms can be induced by treatments that mimic etiological factors of human abnormal aggression. Although the issue was not investigated systematically, the conditions leading to abnormal aggression (e.g., excessive stressors suffered in certain periods of life; see below) are likely occurring under natural conditions, which makes it probable that abnormal aggression exists in free-ranging populations.

We suggest here a classification of aggressive acts based on the constructs overviewed above. The classification was visualized in Fig. 1.1. This figure will be used as a template for evaluating the correspondences between theoretical definitions and operational approaches. Animal aggression was not considered at this stage, but will be included in the system later on. We believe that there are three principles along which aggression should be categorized: behavioral form, reason, and abnormality.

The forms of behavior were listed in the rows of the figure. The three main categories mentioned above (physical, verbal, and indirect) were slightly changed. Particularly, the phrase “verbal aggression” was replaced with “intimidation” to accommodate nonverbal forms that have the same function: body language and object-directed aggression when the latter is displayed with the aim of intimidation (e.g., hammering the table, slamming doors, etc.). Other forms of object-directed aggression (e.g., breaking objects to channel emotions) are considered displaced aggression. Instead of the word “indirect aggression,” we use here “relational aggression” to avoid the inconsistencies mentioned above; the two phrases are often used as synonyms in the literature. The severity of the behaviors increases downwards; intimidation is considered the least severe form because it only threatens with harm; relational aggression involves harm in psychological meaning, while physical aggression in physical meaning. These major forms were also graded according to more specific behavioral aspects. Body language was considered a milder form of intimidation than verbal aggression or object-directed aggression. In the case of indirect aggression, overt forms were considered less severe because they allow for mitigation. The level of injury infliction differentiates physical forms of aggression.

The reasons of being aggressive were listed in column headings. The following reasons were differentiated: legitimate uses of aggression (e.g., the defense of others and self-defense), the inability to control emotions and behavior (uncontrolled outbursts of anger), and gain-motivated aggression. Legitimate uses of aggression were further subdivided into proactive and reactive forms. The former is typical to law-enforcement bodies, while the latter is usually shown by those defending themselves from attack. Uncontrolled outbursts of anger were further subdivided into target oriented (e.g., aimed at the source of threat or provocation) and displaced (channeled towards objects or a third party). The legitimacy of the three main reasons decreases from left to right. Any form of defense is considered legitimate provided that it is proportional; disproportional aggression belongs to one of the subsequent two categories. Uncontrolled outbursts of anger cover aggressive acts that are not justified by the situation (i.e., they are out of proportions). This form of aggression is usually called reactive/emotional and is characterized by strong emotional reactions. Such actions are not legitimate, even less so if the target is a third party (i.e., displaced aggression against people). Yet in legal and moral terms, acting under psychological distress is extenuative to a certain extent; therefore, reactive/emotional aggression was considered less rule breaking (somewhat more legitimate) than instrumental aggression. The least legitimate forms of aggression are those performed for gain.

The color code of Fig. 1.1 denotes abnormality based on the functionality of behavioral acts. As aggressiveness is believed to be a means of competition, functionality was evaluated from the point of view of competitive advantages. On the short term, any of the aggressive behaviors listed in Fig. 1.1 may be considered functional according to this criterion. However, certain types of aggressive behaviors are competitively disadvantageous on the long run, for which these were classified as dysfunctional. For example, excessive and uncontrolled outbursts of aggression disrupt families, social networks, and work places, which renders this behavior dysfunctional overall, even if such outbursts may serve momentary interests in terms of assertiveness under particular circumstances. Similarly, instrumental aggression may make the individual successful in certain competitive situations, but on the long run, they involve personal risks and social punitive reactions that make instrumental aggression dysfunctional overall. By contrast, defense-oriented aggression is advantageous on the short run (it averts dangers) and does not entail long-term risks in legal or other terms. Therefore, defensive aggression was considered functional. Social acceptability and the severity of aggressive acts were also taken into account to grade the dysfunctional nature of aggressive behavior. Intimidation and relational aggression are ubiquitously used by humans as means of competition. Expressing displeasure by body language, verbally, or by object-directed aggression (e.g., slamming doors, hammering tables) is quite common during conflict. Spreading rumors, excluding people from social networks, or setting confederates against competitors is similarly common within social groups. Regular involvement in such behaviors may disfavor the individual on the long run, but negative consequences are not as serious as with physical forms of aggression. Therefore, these behaviors were considered less dysfunctional.

In animals, aggression is usually studied “in action,” i.e., while aggressive acts are actually performed. This is in sharp contrast with the most frequently employed approaches in human aggression research (see below). Even in animals, however, the neurobiology of aggression is rarely studied under entirely naturalistic conditions, partly because the required equipment and the data sampling procedures need laboratory environments, partly because laboratory studies ensure a stability of conditions that are rarely met in nature. Therefore, aggression is usually studied by employing models. The most fruitful models used so far are brain-stimulation-evoked aggression models and the resident-intruder test (the latter including studies on maternal aggression). Other methodologies include defensive, fear-induced, shock-induced, and social aggressions. Models of predatory aggression also exist; these will be discussed later on in relation with intraspecific aggression only. The methodologies discussed here are assumed to address various aspects of normal aggressiveness, i.e., are supposed to model aggressiveness shown by animals under naturalistic conditions. We will briefly characterize the main models below.

The idea of studying abnormal forms of aggression in rodents is relatively new in aggression research: our lab was the first to propose a model (Haller et al. 2001). This does definitely not mean that abnormal aggression was not seen in rodents before this publication; moreover, models used earlier may readily be included into the large family of abnormal aggression models. There were three concepts that differentiated this publication from the earlier ones. First, we deliberately generated the model by making use of data on the etiological factors of antisocial aggressiveness. Second, we detached from the generally employed quantitative approach: instead of counting attacks, we focused on their form and behavioral context. Third, attention was not restricted to aggressive behavior per se; we sought to investigate other features related to antisocial aggressiveness (e.g., autonomic responses, anxiety, and social behaviors beyond aggressiveness) (Haller et al. 2004). Based on the principles underlying this experimental work, we developed a general concept for abnormal aggression models (Haller et al. 2005a; Haller and Kruk 2006). The same complex approach was employed when our second model of abnormal aggression was developed (Toth et al. 2008, 2011).

Models of abnormal aggression gained terrain in two directions: models used earlier were reevaluated in the light of this concept (see Miczek et al. (2013) for a review), and new models were developed to address other etiological factors of abnormal aggression. The criteria that differentiate normal from abnormal aggression were reformulated in several ways by the authors who undertook research in this relatively new field. Nevertheless, the essential points remained the same. Rats are submitted to procedures that mimic etiological factors of aggression-related psychopathologies and are tested for abnormal attack features in the resident-intruder test. These features can be of three types: mismatch between provocation and response, disregard of species-specific rules, and insensitivity towards the social signals of opponents.

The etiological factors considered in abnormal aggression models include the application of stressors at various ages, treatments with drugs, and genetic manipulations. We will review below the main models based on the type of the etiological factor that produced abnormal aggression. Two models will be discussed separately, because of their specific features. One model was based on the repeated exposure of rats to victories, while the other on the suppression of glucocorticoid production. In all cases, the emotional background of aggression will be evaluated, because this feature seems relevant to the type of aggression performed and the classification of the models.