This line of research was initiated by the seminal work of Hess performed in rats and cats (1928). Subsequent studies showed that the phenomenon is general: biting attacks can be elicited by hypothalamic stimulation in virtually all the species studied so far, suggesting that this mechanism of aggression control is evolutionarily conserved (e.g., cats: Romaniuk 1965; chicken: Putkonen 1966; cichlid fish: Demski 1973; guinea pig: Martin 1976; humans: Bejjani et al. 2002; lizards: Sugerman and Demski 1978; monkeys: Lipp 1978; mice: Lin et al. 2011; opossum: Roberts et al. 1967; rats: Vergnes and Karli 1969; Kruk et al. 1979). In lizards at one extreme, aggression can be induced from several brain areas while at the other extreme the hypothalamus is the only area from where attacks can reliably be elicited in rats (Kruk 1991). Hypothalamic structures subserving attack were best described in cats and rats, and substantial evidence is available in mice, hamsters, and humans; therefore, we will focus on these species.

The stimulation of the hypothalamus evokes two types of aggression in cats, namely defensive rage and predatory attacks. The former consists of threatening vocalizations and postures and strikes upon provocation. Predatory attack also called quiet biting involves the killing of pray (e.g., a rat) by biting the back of its neck without vocalizations or threat postures (Siegel et al. 1999). Hypothalamic structures controlling defensive rage and predatory aggression are topographically and functionally separated: the relevant hypothalamic structures (ventral/medial and lateral to the fornix, respectively) are not only behaviorally specific, but the two areas also inhibit each other (Cheu and Siegel 1998) (Fig. 4.1). Topographic organization is also revealed by the distinct neural connections and pharmacology of the two regions (Brutus and Siegel 1989; Shaikh et al. 1984; Siegel et al. 1999, 2007). The interpretation of hypothalamically evoked behaviors needs some additional interpretation especially in conjunction with behaviors elicited from analogous hypothalamic structures of rodents. “Rage” induced from ventral/medial areas of the hypothalamus is usually considered a defensive response, the main component of which is hissing that is executed by, and in parallel with, the opening of the jaws. This interpretation involves that the cat hypothalamus contains no mechanisms for attacks on conspecifics, in sharp contrast to all other mammals where such attacks can readily be evoked. We believe, however, that this strange conclusion derives from the specific conditions under which this type of studies is performed. In rats, stimulation is always performed in the presence of a potential opponent; in cats, this element is missing probably because of ethical reasons; carnivores can inflict heavy damage on opponents. Nevertheless, one can reasonably assume that attacks were performed if an opponent was present, and the jaws opened for hissing were sooner or later closed for biting. This is supported by the behavioral similarity of the initial phases of intraspecific contests and “defensive rage.” Consequently we assume that areas involved in “defensive rage” are in fact components of the system that controls “affective aggression.” In the following we will prefer the latter over the former phrase.

The study of hypothalamic structures involved in attack followed a different trajectory in the rat. The precise location and pharmacological responsiveness of the mediobasal hypothalamus (called “hypothalamic attack area”) was described in detail, while brain areas involved in predatory aggression received less attention. Similar to cats, attacks on conspecifics can be induced in rats by the stimulation of the mediobasal hypothalamus, which is located ventral and medial to the fornix and includes the posterior part of the anterior hypothalamus, the relatively cell-sparse region extending from the fornix to the third ventricle, and the ventrolateral aspects of the ventromedial hypothalamus (Kruk et al. 1979; Kruk 1991; Lammers et al. 1988). Later studies demonstrated that spontaneous aggression induces c-Fos activation in the very same region (Halasz et al. 2002a); moreover, the unilateral electric stimulation of the mediobasal hypothalamus induces attacks only when the contralateral side is also activated in a manner, and to the extent, similar to that seen in spontaneous aggression (Halasz et al. 2002b). The presence of predatory attack sites was also demonstrated in rats. For example, the electrical and neurochemical stimulation of the lateral hypothalamic area induces frog and/or mouse killing (Vergnes and Karli 1969; Smith et al. 1970; Bandler 1970; Woodworth 1971), and recent findings from our laboratory show that the lateral hypothalamus is strongly activated by mouse killing (Tulogdi et al. 2014). However, the topographic organization of the two types of aggression may not be as distinct as in cats. Some of the early studies revealed that the stimulation of the lateral hypothalamus does induce attacks on conspecifics albeit this response required considerably larger current intensities than muricide (Woodworth 1971). Although no attacks on conspecifics were induced from the lateral hypothalamus in other labs (Kruk 1991; Siegel et al. 1999), it was shown that the stimulation of the lateral hypothalamus promotes spontaneous attacks under certain circumstances (Koolhaas 1978). The assumption that the lateral hypothalamus is involved in intraspecific aggression is also supported by c-Fos and brain imaging studies showing that the lateral hypothalamic area is activated during resident-intruder conflicts (Clinton et al. 2011; Ferris et al. 2008; Tulogdi et al. 2010). Taken together, these findings show that intraspecific aggression and predatory aggression are controlled by hypothalamic areas medial and lateral to the fornix, respectively, but the two mechanisms’ area not as distinct as in the cat.

It was recently demonstrated that the optogenetic stimulation of the lateral aspects of the ventromedial nucleus rapidly induces coordinated and directed biting in mice (Lin et al. 2011). Although bites were elicited from a very restricted hypothalamic area in the mouse, a wider array of hypothalamic mechanisms are activated by territorial aggression in this species, suggesting that many hypothalamic structures analogous to the rat hypothalamic attack area are also involved in mouse aggression (e.g., the posterior part of the anterior hypothalamus, the ventrolateral nucleus of the hypothalamus, and the tuber cinereum; Duncan et al. 2009; Haller et al. 2006). It was also shown that individual variation in estrogen receptor α immunoreactivity in the anterior hypothalamus (part of “hypothalamic attack area” in rats) is positively correlated with aggressive behavior in mice (Trainor et al. 2006). Moreover, some findings suggest that the lateral hypothalamus also plays a role in mouse aggression (Duncan et al. 2009; Hasen and Gammie 2006). Taken together, these findings suggest that mouse aggression is controlled by hypothalamic structures analogous to those controlling aggression in rats.

In hamsters, the location and extent of hypothalamic structures involved in aggressiveness were not precisely delimited, but it is believed that the key regions are the anterior and lateral hypothalamic areas which control intraspecific attacks (Ferris and Delville 1994; Knyshevski et al. 2005).

Evidence on the involvement of the human hypothalamus in aggression came from the study by Sano et al. (1970) who lesioned posteromedial regions of the hypothalamus, which reduced or abolished aggressiveness in violent patients (Fig. 4.2). The lesioned region included medial parts of the hypothalamus and the medial subdivision of the lateral hypothalamus. These findings were replicated by many other studies (Dieckmann et al. 1988; Laitinen 2001; Pedrosa-Sanchez and Sola 2003; Ramamurthi 1988). A case study suggested that the electrical stimulation of the very same area induces aggressiveness (Bejjani et al. 2002). The behavioral response was similar to that seen in animals as it occurred within seconds, and had no post-stimulation effects on aggression. Importantly, the patient although clearly remembered it was not able to explain his/her behavior, suggesting that the hypothalamically evoked aggressiveness had no obvious cognitive aspect. The role of the human hypothalamus in aggression control was supported by recent studies showing that the continuous stimulation of the “triangle of Sano” for month and years suppresses aggression in patients with intractable aggression (Franzini et al. 2005, 2010; Hernando et al. 2008). At the first sight, reduced violence obtained by the Sano and the Franzini approaches is intriguing, as lesions and stimulations were targeted to very similar hypothalamic sites. It is believed, however, that deep brain stimulation is in fact a functional lesion (Bennazzous and Hallett 2000); high-frequency stimulation can rapidly render neurons inactive by a phenomenon called stimulation-induced inhibition of depolarization. High-frequency stimulations may even lesion the area if performed over prolonged periods. Thus, the Sano and Franzini procedures are very similar in their ultimate neural outcome. An interesting example of hypothalamic aggression control is the hypothalamic hamartoma. A subgroup of patients developing this malformation show increased aggression, while the removal of the affected hypothalamic parts abolishes aggressiveness (de Almeida et al. 2008; Weissenberger et al. 2001). The impact of hamartomas on hypothalamic function is poorly known, but the findings with this malformation show that hypothalamic functions are tightly linked to aggression in humans. This assumption is further supported by a brain imaging study performed in females exposed to threatening stimuli; a significant activation of the hypothalamus was noticed (Hermans et al. 2008).

Taken together, these findings suggest that there is a region in the human hypothalamus that controls aggressiveness, and disparate findings suggest that this area responds to stimulation in a manner similar to that seen in laboratory animals.

In the cat, the affective aggression-bound medial hypothalamus is stimulated from upstream by substance P-ergic medial amygdala neurons; its “execution-bound” downstream outputs to the periaqueductal gray are glutamatergic (Siegel et al. 1999, 2007). Inhibitory outputs to the lateral hypothalamus are GABAergic. The mediobasal hypothalamus also receives facilitatory noradrenergic, dopaminergic, and cholinergic inputs and inhibitory serotonergic inputs. The role of serotonin appears complex as effects exerted via different receptors (e.g., 5-HT_1A and 5HT_2C) affect aggression in an opposite fashion (Hassanain et al. 2003; Siegel et al. 1999). The neurochemical nature of lateral hypothalamic neurons is known in less detail.

The rat mediobasal hypothalamus (involved in attacks on conspecifics) contains both glutamatergic and GABAergic neurons that have a distinct localization within the area (Hrabovszky et al. 2005). The same study showed that a subgroup of glutamatergic neurons also expresses thyrotropin-releasing hormone, the role of which remains unclear at present. From upstream, the system is activated by substance P neurons located in the medial amygdala, a mechanism that seems very similar to that seen in cats (Halasz et al. 2008, 2009). Local glutamatergic neurons are responsible for the induction of attacks, whereas GABAergic mechanisms locally inhibit them (Adams et al. 1993; Haller et al. 1998a, b, c; Roeling et al. 1993). Hypothalamic attack is under the control of several other factors, namely serotonin (local inhibition; Kruk 1991) and vasopressin (stimulation; Caffrey et al. 2010; Ferris et al. 2008). The latter mechanism is sensitive to testosterone (Trainor et al. 2006).

In hamsters, the anterior hypothalamus (i.e., the “hypothalamic attack area” in this species) is under the control of vasopressinergic inputs (Albers et al. 2006; Ferris and Delville 1994; Ferris and Potegal 1988) that seem to integrate other neurochemical and hormonal influences, namely those of serotonin and testosterone (Delville et al. 1996; Ferris et al. 1997). Attack-related hypothalamic mechanisms are also sensitive to the local application of glucocorticoids (Hayden-Hixson and Ferris 1991a, b).

Findings on the hypothalamic neurochemistry of aggression are sparse in humans. Nevertheless, a study in females suggests that the testosterone/cortisol ratio has a role (Hermans et al. 2008); a role for serotonin was also shown (Koch et al. 2007).

The hypothalamus hosts cell groups that are critically involved in the induction of attack behavior. Two regions of interest were identified, each being involved in a specific form of attack. The medial parts of the hypothalamus induce attacks on conspecifics, while lateral parts induce attacks on prey. While the two mechanisms are distinct in cats, the lateral hypothalamus seems to have a somewhat mixed role in rodents. Hypothalamic mechanisms are also involved in the control of human aggression.

Hypothalamic mechanisms appear to be controlled by substance P from upstream (particularly from the medial amygdala) and send glutamatergic projections downstream. The functioning of attack-bound hypothalamic mechanisms is controlled by inputs that utilize acetylcholine, dopamine, GABA, noradrenaline, serotonin, and vasopressin as neurotransmitters. Notably, these are the neurochemical mechanisms that are believed to control aggression overall; the studies reviewed above suggest that they have local impacts at the level of aggression-bound hypothalamic structures. Testosterone and glucocorticoids also have a role at this level. Neurochemical and neuroendocrine data in humans are sparse, but the available findings are in line with those obtained in animals.

One approach to the study of brain control is the elucidation of overall activities seen in particular brain areas. “Overall” means here that neuron types, neurochemical agents, receptors, etc., are not distinguished; the technique provides indistinct information on the level of activation. This is one of the main approaches in human brain imaging techniques; their rodent analogs are c-Fos studies or other studies of this kind.²

One may have expected that hypothalamic structures involved in attack were overactivated in abnormal aggression if the “hyperactivation theory” was right, i.e., if abnormal aggression resulted from the exacerbated activation of aggression-controlling brain mechanisms. However, hypothalamic mechanisms controlling attack undergo interesting, sometimes surprising, changes in rodent models of abnormal aggression.

In some models, the expected changes were seen. The mediobasal hypothalamus was overactivated by aggression in four models, namely in the postweaning social isolation, anabolic steroid, and cocaine models, as well as in rats selected for low anxiety (Toth et al. 2012; Ricci et al. 2007; Knyshevski et al. 2005; Beiderbeck et al. 2012). As it regards the lateral hypothalamus (implicated mainly in predatory attacks), increased activation was seen in three models (the glucocorticoid dysfunction and cocaine model, as well as in low-anxiety rats) (Beiderbeck et al. 2012; Knyshevski et al. 2005; Tulogdi et al. 2010).

In other models, however, no changes in mediobasal hypothalamic activation were seen, but the lateral hypothalamus was activated where this area was investigated. Such findings were observed in the glucocorticoid dysfunction model, in mice selected for high aggressiveness, and in rats selected for high anxiety (Beiderbeck et al. 2012; Haller et al. 2006; Halasz et al. 2002a; Tulogdi et al. 2010). Lateral hypothalamic activation was not increased in rats selected for high anxiety and those submitted to postweaning social isolation (Beiderbeck et al. 2012; Toth et al. 2012). In rats selected for high anxiety, neither the mediobasal nor the lateral hypothalamus was overactivated despite clearly abnormal attack patterns (Beiderbeck et al. 2012).

Interestingly, this pattern of activity and nonactivity overlaps with the emotional responses associated with these models of abnormal aggression. The mediobasal hypothalamus was overactivated mainly in models where aggression is associated with hyperarousal, e.g., the postweaning social isolation and cocaine models both being characterized by increased glucocorticoid and autonomic stress responses (Ansah et al. 1996; Moldow and Fischman 1987; Toth et al. 2011). In contrast, the mediobasal hypothalamus was not activated in models characterized either by low glucocorticoid and autonomic stress responses or by low basal glucocorticoid levels and heart rates i.e., by hypoarousal (rat: Haller et al. 2001, 2004; mice: Caramaschi et al. 2008a, b; Veenema et al. 2004). Although aggression-induced autonomic activation was not directly investigated in rats selected for anxiety, studies on glucocorticoids support the assumption that hypo- and hyperarousal during aggressive confrontations have an impact on the activation of hypothalamic centers involved in attack. In highly anxious rats that attacked vulnerable targets, the glucocorticoid response to resident-intruder conflicts was decreased and no aggression-induced hypothalamic activation was noticed (Neumann et al. 2010). In the low-anxiety strain that showed vulnerable attacks and also attacked females and anesthetized opponents, the glucocorticoid response to resident-intruder conflicts increased, and the hypothalamus was strongly activated by fights.

The findings briefly reviewed above suggest that hypoarousal-associated aggression does not, while hyperarousal-associated aggression does correlate with increased mediobasal hypothalamic activation. This relationship is illustrated in Fig. 4.3 by the comparison of the glucocorticoid deficiency postweaning social isolation models. The direct comparability of these models is ensured by the consistency of behavioral and neuroanatomical analysis, as the findings were obtained by the same group. In the hypoarousal model, normal attack counts but abnormal attack features are associated with a “predatory-like” hypothalamic activation pattern (normal mediobasal but increased lateral hypothalamic activation; Tulogdi et al. 2010). In contrast, exacerbated “rivalry-like” hypothalamic activation pattern was seen in the postweaning social isolation (hyperarousal) model (increased mediobasal but normal lateral hypothalamic activation; Toth et al. 2012). Unfortunately, evidence on predation-related hypothalamic control mechanisms is lacking in mice, while the arousal component of aggression is poorly known in other models. Nevertheless, the findings briefly reviewed above and the example summarized in Fig. 4.3 suggests that the emotional component of abnormal attacks has a large impact on the hypothalamic control of aggression.

The “serotonin hypothesis” is considered a cornerstone of aggression control (Nelson and Chiavegatto 2001). Theoretically, abnormal aggression may be driven by decreased serotonergic control, which would indirectly upregulate the function of brain mechanisms that promote aggression. However, the role of hypothalamic serotonin is unexpectedly complex in models of abnormal aggression.

In the mouse variant of the postweaning social isolation model, rats submitted to maternal separation as well as anabolic steroid-treated hamsters, findings support the notion that serotonin suppresses aggression. In these models, serotonin fiber density and/or the density of various serotonin receptors decreased in the mediobasal hypothalamus that is involved in the control of rivalry attacks (Bibancos et al. 2007; Grimes and Melloni 2002, 2005; Ricci et al. 2006; Veenema et al. 2006). Some contradictions do exist; e.g., 5HT_1A receptor density decreased in hamsters treated with anabolic steroids, but remained unchanged in mice submitted to postweaning social isolation (Bibancos et al. 2007; Ricci et al. 2006). Overall, however, a decrease in hypothalamic serotonin functions is evident in these models.

In sharp contrast with the aforementioned models, hypothalamic 5-HT_1B and 5-HT_2A receptor mRNA levels doubled in mice submitted to the alcohol model (Chiavegatto et al. 2010). Hypothalamic 5-HT₆ receptor expression also doubled in the anabolic steroid model (Ambar and Chiavegatto 2009), while the density of serotonin fibers and the number of serotonergic varicosities increased in the anterior hypothalamus of hamsters submitted to the subjugation model (Delville et al. 1998; Ferris and Potegal 1988). Similar findings were obtained with N-type Ca channel alpha_1B subunit-KO mice where attack counts increased over five times, yet the serotonin content of the hypothalamus doubled without parallel increases in hypothalamic noradrenaline and dopamine (Kim et al. 2009).

One has to conclude that while hypothalamic serotonin undergoes the expected changes in some models, certain forms of abnormal aggression are associated with increased serotonergic neurotransmission in the hypothalamus. Findings with the postweaning social isolation and alcohol models are especially intriguing; as the subjects were mice in both cases, the findings were obtained by an overlapping set of authors, and the immunocytochemical methodologies were highly similar (Fig. 4.4). One can argue that the intriguing changes seen at hypothalamic level were “overwritten” by reduced serotonergic neurotransmission in other brain areas, e.g., the prefrontal cortex (Chiavegatto et al. 2010). However, this argument does not hold true for two reasons. First, increased serotonergic neurotransmission in abnormally aggressive animals was seen in extra-hypothalamic sites as well. For example, Delville et al. (1998) observed increased number of septal 5-HT varicosities in the hamster “subjugation” model. Chiavegatto et al. (2010) showed that in addition to the hypothalamus, the expression of 5-HT_1B receptors was increased in the amygdala of mice that showed excessive increases in aggression after alcohol, while Bibancos et al. (2007) reported increased 5-HT₆ receptor expression in the hippocampus of mice submitted to the postweaning isolation model. Second, paradoxical associations were also noticed when neurotransmitter/behavior correlations were investigated. There was no significant correlation between raphe serotonergic activation and aggressiveness in the hypoarousal model which was in sharp contrast with data obtained in controls (Haller et al. 2005a, b). In addition, the 5-HT_1A partial agonist buspirone ameliorated normal (rivalry) aggression but dramatically increased attack counts in the glucocorticoid deficiency model (Haller et al. 2007). In a similar fashion, there was no obvious link between serotonergic neurotransmission and aggressiveness in mice selected for aggressiveness (Natarajan et al. 2009a, b, c; Wallinga et al. 2009). Moreover, 5-HT_1A and 5-HT_1B agonists decreased aggressiveness in highly aggressive feral rats by decreasing serotonin release via their presynaptic autoreceptor function (de Boer and Koolhaas 2005).

Taken together, these findings demonstrate that the anti-aggressive effects of serotonergic neurotransmission are compromised in certain models of abnormal aggression; moreover, serotonin may paradoxically promote aggressiveness in some models. Most importantly, these paradoxical interactions were seen in models where abnormal aggression was associated with hypoarousal (blunted cortisol response to aggression in subjugated hamsters: Ferris et al. 2005; low basal corticosterone after adolescent anabolic steroid treatment: Rejeski et al. 1990; low corticosterone stress responses and low basal heart rate in mice selected for high aggressiveness: Caramaschi et al. 2008a, b; Veenema et al. 2004; low basal CORT and low autonomic responses to aggression in the rat glucocorticoid dysfunction model: Haller et al. 2001, 2004).

Vasopressin plays important roles in aggression control (see above); as such, one can hypothesize that increased vasopressin neurotransmission promotes abnormal forms of aggression. This, however, is not always the case.

Vasopressin neurotransmission appears to promote abnormal manifestations of aggression in several models. The interaction has many facets. In the anabolic steroid model, the expression of the V_1A vasopressin receptor increased in the ventrolateral but not in the anterior hypothalamus (DeLeon et al. 2002). Despite this, V_1A antagonists administered into the anterior hypothalamus decreased the anabolic steroid-induced aggression (Harrison et al. 2000). In the amphetamine model, V_1A expression increased in the anterior hypothalamus, and V_1A antagonists abolished amphetamine-induced aggression (Gobrogge et al. 2009). V_1A antagonists also abolished exacerbated aggression in the cocaine model, where an increased stimulation-induced vasopressin release was noticed without changes in vasopressin content or vasopressin fiber density (Jackson et al. 2005). In the maternal separation model (the only rat study addressing hypothalamic vasopressin), no change in vasopressin expression was noticed in the anterior hypothalamus, but vasopressin expression (both basal and aggression induced) increased in the lateral hypothalamus (Veenema et al. 2006). Despite this variability of findings, increased vasopressin functions are evident in these models.

Changes seen in other models were opposite to expectations. Decreased anterior hypothalamic vasopressin content and decreased vasopressin fiber density was noticed in the subjugation model (Delville et al. 2000). Gene polymorphism studies suggest similar changes in rats selected for anxiety. Low-anxiety rats express a vasopressin gene that has a reduced transcription rate, and consequently vasopressin production is lower in this than in the high-anxiety strain (Bunck et al. 2009; Kessler et al. 2007). The low-anxiety strain showed more disturbed attack patterns than the high-anxiety strain; consequently, higher abnormality of aggression was associated with a lower hypothalamic vasopressin levels.

Similar to overall activation levels and serotonin neurotransmission, the role of hypothalamic vasopressin in abnormal aggression seems model dependent. While it clearly promotes abnormal aggression in the anabolic steroid, amphetamine, cocaine, and maternal separation models, the opposite is true for the subjugation model and rats selected for low-anxiety behavior.

There are neurotransmitter systems that were studied in one or a few models only. This does not diminish the importance of the findings, or the relevance of the respective systems for abnormal aggression, but the various models of abnormal aggression cannot be compared in their case. For example, the impact of substance P neurotransmission (the effects of which are mediated by neurokinin 1 (NK₁) receptors) was investigated in the glucocorticoid dysfunction model only (Halasz et al. 2008, 2009). In this model, mediobasal hypothalamic neurons expressing NK₁ receptors were specifically overactivated by aggressive encounters; moreover, the systemic application of NK₁ blockers or the selective destruction of NK₁ receptor expressing mediobasal hypothalamic neurons by saporin-conjugated substance P dramatically reduced violent biting, without affecting milder forms of aggression. Thus, NK1-expressing mediobasal hypothalamic neurons appear specifically involved in abnormal features of aggression in this model.

In the anabolic steroid model, dopamine production by the circular and medial supraoptic nuclei of the hypothalamus was increased, and the expression of D₂ dopamine receptors was increased in the anterior hypothalamus (the attack area in the species studied; Ricci et al. 2009; Schwartzer et al. 2009). Thus, dopamine neurotransmission may be another mechanism that promotes abnormal aggression at hypothalamic level.

Finally, the role of glutamatergic and GABAergic neurotransmission was also investigated in a few models. In the hamster anabolic steroid model, glutamate synthesis increased in both the anterior and lateral hypothalamus, but equivocal changes were noticed at the level of their projections (Carrillo et al. 2009; Fischer et al. 2007). The role of GABA was investigated in the anabolic steroid model (Grimes et al. 2003; Schwartzer et al. 2009), in the cocaine model (Ricci et al. 2005), and in mice selected for high aggressiveness (Weerts et al. 1992). In the latter model, findings were in line with expectations: GABA binding was increased in low aggression genotype and was decreased in the highly aggressive genotype. In the former two models, however, increased GABA synthesis was noticed in the anterior hypothalamus despite the fact that GABA agonists inhibit rivalry aggression when infused into mediobasal hypothalamic sites (Adams et al. 1993; Roeling et al. 1993).

The above findings reveal that GABAergic mechanisms may also differentiate certain models of abnormal aggression. Comparative data are not available for substance P, dopamine, and glutamate, but the available findings show that they are also relevant from the point of view of hypothalamic mechanisms involved in abnormal aggression.

The findings summarized above seem harmonious at the first sight, but a closer scrutiny reveals that they are fundamentally incongruent. Problems start already with the size of this brain area. While the human prefrontal cortex is delimited in various ways by various authors (for contrasting delimitations, see Critchley et al. 2000; Barrash et al. 2000; Raine et al. 1997, 2011; Davidson et al. 2000), it can be considered very large according to any of the delimitations. For the purposes of this review, we summed up the various views of the authors reviewed here and corroborated these with the brain atlas of Mai et al. (1998) (Fig. 4.5a). As seen on the figure, the brain area generally considered the “prefrontal cortex” makes up more than one-third of the neocortex. It is rather difficult to imagine how such a large brain area can as a whole project to particular subcortical structures and influence their functions as a whole.

The prefrontal cortex of rats is relatively smaller. The widespread consensus is that it is composed of the medial and orbitofrontal prefrontal cortices (Fig. 4.5b). Some authors also include the agranular insular and piriform cortices. Even if smaller relatively, it still comprises approximately one-fifth of the rat neocortex. Thus, it is still too large to assume that it is functionally homogeneous and affects subcortical structure as a whole.

These considerations show that the prefrontal cortex is not a unit that can be considered as such when mechanisms underlying aggressiveness are evaluated. As a phrase, it is just as general as “subcortical structures” or “limbic system.” Aggression cannot be controlled by the whole prefrontal cortex; projections that can be physically accommodated by lower structures must come from restricted regions of this vast area. The question arises whether the available data allow the delimitation of aggression-specific regions in the prefrontal cortex. A cross-checking of findings obtained with various technologies shows that they cannot.

One of the strong arguments supporting the “prefrontal deficit theory” of aggression is that lesions of the prefrontal cortex induce a state called “acquired psychopathy,” while psychopaths show structural and functional deficits exactly in the prefrontal cortex. However, lesions leading to “acquired psychopathy” are located elsewhere than the structural and functional alterations seen in aggression-related psychopathologies.

These studies do not support each other in the meaning that the deficits observed by different authors are located in anatomically distinct regions of the prefrontal cortex.

All the studies reviewed above belong to the “aggression history” type where brain imaging is performed while subjects lie still in the apparatus; aggressiveness is attributed to them based on their past actions. In the majority of such subjects, prefrontal deficits are seen, even if the location of these is highly variable. In manifest aggression studies, where subjects perform aggression-related tasks during brain imaging, prefrontal activations are seen; moreover, the magnitude of activation correlates positively with the history of violence. The two types of studies do not support each other in the meaning that while deficits are seen in the first type, activations are seen in the second.

The findings reviewed above can be summarized as follows: (1) prefrontal damages increase aggression only when they include orbitofrontal areas. The very same region is not necessarily dysfunctional in aggressive individuals. (2) Non-orbitofrontal damages do not increase aggression. Aggressiveness was often specifically associated with non-orbitofrontal dysfunctions. (3) Aggression-related prefrontal dysfunctions occur in patches. The location and extent of dysfunctional “patches” vary from study to study. Findings are discordant even when the characteristics of subjects and the methods of study are highly similar. (4) The execution of aggressive acts activates the prefrontal cortex. Activation is stronger in abnormally aggressive individuals. (5) Human and animal findings are highly consistent. Despite their rather simplistic formulation, the statements are essentially true when the totality of findings is considered.

The metaphor refers to the sensory and motor homunculi that visualize the organization of primary sensory and motor areas, which are topographically precise representations of the body. It occurs that the prefrontal cortex is organized in a somewhat similar way: it represents psychological faculties in a mosaic-like, topographically structured, and hierarchically organized fashion. This feature of the prefrontal cortex is highly relevant for aggression control and may explain a great deal of the discrepancies outlined in the previous sections. In the following we will briefly overview the proposed cognitive and emotional maps of the prefrontal cortex. The integration of these findings with those obtained in aggression research and the hypothesis per se will be described in separate sections.