The Neurobiology of Offensive Aggressionq

SF de Boer and JM Koolhaas, University of Groningen, Groningen, The Netherlands

Ó 2017 Elsevier Inc. All rights reserved.

The Various Forms of Aggression 1

Violence is the Pathology of Functional Aggressive Behavior 2
Neurobiological Correlates of Aggression and Violence 2
A Highly Interconnected Network of Brain Regions Controls Aggression 2
Network Components 3
Hypothalamus 3
Amygdala 4
Septum 4
Prefrontal Cortex 4
Periaqueductal Gray (PAG) 5
Neurochemical Modulation of the Aggressive Neural Network 5
Serotonin 5
Dopamine 6
GABA 7
Hormonal Modulation of the Network 7
Corticosterone 8
Vasopressin and Oxytocin 8
Neurogenetics of Aggression 9
Synthesis and Outlook 10
References 10

The Various Forms of Aggression

Across the animal kingdom, aggression is the behavioral weapon of choice for individuals to defend themselves, to compete for
resources like food, mates and territory, and to maintain social hierarchies. From a biological point of view, aggressive behavior
is considered a highly functional form of social communication leading to active control of resources and the social environment.
It is characterized by a set of species-typical behaviors performed in close interaction with another individual. The existence of
different kinds of aggression has long been recognized mainly on the basis of animal research. There are two types of agonistic
behaviors in both males and females related to attacks: offense and defense. These differ in motor patterns, bite or attack targets,
adaptive function and proximate neurobiological mechanisms. While offensive aggression is a form of agonistic behavior initiated
by an aggressor and displayed in the context of competition for resources, defensive aggression is elicited in response to threat or
attack by an offensive conspecific. For example, an offensive male may compete with other males for food, a territory or females. An
animal that is attacked by either a dominant male or a predator performs defensive aggression. For offense, the motor patterns are
chase, offensive upright posture, offensive sideways posture, and attacks (simple bites or bite and kick). The bite targets are primarily
the hindquarters of the flanks, back and base of the tail (non-vulnerable body regions). The function is to obtain and retain
resources such as space, food, and mates. For defense, the motor patterns are flight, defensive upright posture, defensive sideways
posture, and attacks (lunge and bite). The bite targets are primarily the face, neck, shoulders and belly (vulnerable body regions).
The function is to defend one’s self, mates, and progeny from attacks of another animal of the same or different species. The
offense-defense distinction plays a prominent role in understanding the biology and physiology of animal aggression. Besides
this distinction in offense and defense, other forms of aggressive behavior can be distinguished as well, such as infanticide, pred-
atory aggression, irritable aggression and maternal aggression. The latter can be observed in females during the late stages of preg-
nancy and the early phases of nursing. Predatory aggression is known as quiet-biting attack observed in cats killing its prey and as the
swift killing of a mouse or a cricket by a rat.
Different forms of aggression are also recognized in humans. Offensive aggression in animals generally relates to reactive, hostile
or impulsive aggression in humans. This form of aggression is also called affective aggression and has its strong initiative engage-
ment and autonomic arousal in common with offensive aggression in animals. Moreover, both in animals and humans, this form of
aggressive behavior is usually initiated in response to a perceived stress such as the intrusion of an unfamiliar conspecific into the
territory or in response to fear and frustration. Affective, emotional and hostile forms of human aggression are summarized under
the term reactive aggression whereas premeditated, instrumental aggression is called proactive aggression.

q
Change History: April 2016. Sietse F de Boer added abstract and keywords, made some changes to the text, replaced Figure 1, added Figures 2 and 3, updated
Table 1 and References section.

Reference Module in Neuroscience and Biobehavioral Psychology http://dx.doi.org/10.1016/B978-0-12-809324-5.00382-5 1

2 The Neurobiology of Offensive Aggression

Violence is the Pathology of Functional Aggressive Behavior

Although aggression is basically highly functional, it can be potentially harmful. Therefore, throughout the animal kingdom,
strong negative feedback mechanisms have developed such as taboos, ritualization, submission, reconciliation and appeasement
to keep aggression in control and to prevent its potentially adverse consequences. Much of the scientific and public interest in
aggression is however motivated by the violent, hostile, and presumably less adaptive forms of aggression observed in our human
society and clinically across a wide spectrum of psychiatric and neurological disorders. Several recent studies in rodents show that
violent-like forms of aggressive behavior can be observed in animals as well. Violence in rats and mice is characterized by its indis-
criminative nature and the complete absence of any introductory social behavior. Violent mice and rats attack any kind of oppo-
nent, including females, target their bites at vulnerable body parts and do not show social exploration and threatening behavior,
preceding an overt attack. In this view, violence can be defined as a pathological form of aggressive behavior that is not subjected
to inhibitory control mechanisms and that has lost its function in social communication (ie, out of control and out of context).
The relationship between the functional and maladaptive extremes of the offensive aggression spectrum is still far from clear and
forms a major obstacle in the integration of animal research with data on human aggression. Fortunately, recent preclinical
research start to make a clear distinction between aggression and violence as defined above, and novel experimental laboratory
models of violent-like aggression in rodents have appeared (Miczeket al., 2007; 2013).

Neurobiological Correlates of Aggression and Violence

Research on the neurobiology of aggression started with a classic approach of surgical lesioning or electrically stimulating specific
brain areas. These experiments were initiated more than 100 years ago with the taming effects of temporal lobectomies in rhesus
monkey by Brown and Schafer (1888) and extremely aggressive dogs by Friederich Golz (1884), followed approximately 30 years
later by Philip Bard’s (1928) demonstration that rage-like aggressive behaviors were absent in posterior hypothalamic knife-cut
transectioned cats. Around that same time period, Hess and Bruegger (1943) started his Nobel Prize-winning intracranial stim-
ulation (ICS) experiments in cats, demonstrating aggressive responses evoked by electrical stimulation of the hypothalamic brain
region. Since then, neuroscientists have sought to understand the neural basis of aggression and violence by perturbing and
monitoring brain activity through a variety of methods and in a wide variety of animals such as monkeys, dogs, cats, rats,
mice, voles and hamsters. By employing numerous increasingly sophisticated tools of functional neuroanatomy (ie, from the
classic electric/chemical lesion and stimulation techniques to neurochemical mapping and molecular-genetic manipulations),
many important strides have been made in understanding the functional brain circuit organization of different social (aggres-
sion, sex, parental care) behaviors, ie, the structurally and functionally highly interconnected “social behavior neural network”
(SBN) (Newman, 1999; Nelson and Trainor, 2007).

A Highly Interconnected Network of Brain Regions Controls Aggression

To more comprehensively identify this SBN, and particularly the specific neural circuitry involved in aggressive behaviors, deter-
mining the pattern of activation of immediate early gene expression has been employed successfully within the last two decades.
Fos is the protein product of an immediate early gene (IEG), c-fos, which is expressed in neurons shortly after their depolarization
(activation), and then induces expression of downstream genes. Fos-expression can be visualized using immuno-histochemical
staining techniques and the number of Fos-positive neurons in each brain area is used to quantify the activation the area. Appli-
cation of this technique in offensive aggression paradigms in rats, mice, and hamsters reveals a neuronal network that includes
(but is not limited to) the intimately interconnected forebrain (limbic) structures like cortico-medial amygdala (MeA and CoA),
bed nucleus of the stria terminalis (BNST), lateral septal area (LS), mediodorsal and anterior thalamus, several hypothalamic
nuclei including the anterior hypothalamus (AHA), ventromedial hypothalamus (VMH), lateral hypothalamus (LH), the paraven-
tricular nucleus (PVN), the medial prefrontal cortex (mPFC), the midbrain periaqueductal gray (PAG), dorsal raphe nucleus
(DRN), locus coeruleus (LC) and ventral tegmental area (VTA) (see Fig. 1 and de Boer et al., 2015 for a more detailed review
on the neurobiology of offensive aggression). Comparative research indicates that this highly interconnected neuronal network
for offensive aggression is remarkably similar in many vertebrate species including humans, indicating that it is evolutionary
ancient and very well conserved (O’Connell and Hofmann, 2012). Indeed, this interconnected brain aggression circuitry is gener-
ally confirmed in humans by modern brain-imaging techniques such as Functional Magnetic Resonance Imaging (fMRI) and Posi-
tron Emission Tomography (PET) that allow the in-vivo analysis of entire neuronal networks involved in certain types of
aggressive behavior. However, it is quite surprising that in most of the human neuroimaging studies, the hypothalamic limbic
brain structures involved in the direct control of animal fighting and attack usually do not show up in their region of interest
analyses. Rather, these studies predominantly focus on the higher cortical (ie, prefrontal, cingulate) and temporal lobe (amygdala)
brain structures.
The function of the “SBN” brain areas in the expression and control of aggressive behavior ranges from sensory processing and
perception up to the generation of somatomotor output patterns, the autonomic and neuroendocrine physiological support of
behavior, and all organizational processes in between. However, although the neural circuitry of aggressive social behavior is
 The Neurobiology of Offensive Aggression 3

ex
neo cort

PFC
cp hpc
ob
LS thal PAG
ao
b nacc
vp HAA
olfactory
input vta
AMYG
pit

neuroendocrine behavioural autonomic

output output output

Figure 1 Neuronal network involved in aggressive behavior and the organization of the accompanying neuroendocrine and autonomic activation.
(aob, Accessory olfactory bulb; AMYG, amygdala; cp, caudate putamen; HAA, hypothalamic attack area; hpc, hippocampus; LS, lateral septum; nacc,
nucleus accumbens; ob, olfactory bulb; PAG, Periaqueductal gray; PFC, Prefrontal cortex; pit, pituitary; thal, thalamus; vp, ventral pallidum; vta, ventral
tegmental area).

mapped relatively well in terms of brain (sub) nuclei and its interconnections, it still remains a challenging task to decipher how the
activity of distinct sets of neurons within the various nodes of this basic SBN circuitry give rise to different phases (initiation, execu-
tion and termination), levels and/or forms of aggressive behavior.

Network Components
Hypothalamus
Classic studies in rats, cats and monkeys using electrical/chemical stimulation of small populations of neurons, already revealed
a hypothalamic attack area in which offensive attack can be reliably elicited. This area consists of the intermediate hypothalamic
area and the ventrolateral pole of ventromedial hypothalamic nucleus. Upon weak electrical stimulation of this area in male rats,
they almost immediately start a fierce aggressive attack that stops at the moment the current is switched off (see Kruk, 2014 for
detailed review). However, despite its anatomical precision, electrodes still affect a rather ill-defined population of neurons and
fibers of passage that do not allow definite conclusions on the precise neuronal and circuit-level mechanisms underlying offensive
attack. The brain packs roughly 100 000 neurons and a billion synaptic connections in every cubic millimeter of tissue, and elec-
trically stimulating or lesioning even a tiny location in the brain will excite/silence a very large number of intermeshed cells of
different kinds. Recently, newly emerging techniques for mapping, measuring, and manipulating neural activity based on genetic
targeting of specific neuron subtypes has solved many of these problems. In particular, optogenetics and pharmacogenetics have
recently made it possible to rapidly and reversibly activate or inhibit small molecularly distinct populations of neurons (anatom-
ical and genetic precision) at any moment in time (temporal precision) (Anderson, 2012; Deisseroth, 2014). These revolutionary
techniques offer the ability to selectively manipulate individual neural circuit elements that underlie aggression-relevant
behaviors.
The first experiments investigating the role of the hypothalamus in the regulation of aggression using optogenetic stimulation
focused on the already characterized ventrolateral subdivision of the VMH. Following virally-delivered expression of the light-
sensitive protein channelrhodopsin-2 (ChR2) in this VMHvl region of mice, light pulses delivered through an implanted optic fiber
produced robust offensive attacks directed toward male mice, castrated male mice, female mice, and inanimate objects (Lin et al.,
2011). Accordingly, inhibiting these neurons using virally-expressed C elegansivermectin-gated chloride channel, which prevents the
initiation of action potentials by hyperpolarizing the neurons upon ligand binding (ie, a pharmacogenetic approach), potently sup-
pressed normal attacks. Subsequent studies have capitalized on the fact that the neurons of the VMHvl are primarily glutaminergic
and are enriched with estrogen receptors of the alpha subtype (ER1). Both ER1-knockout mice and RNAi knockdown of ER1 in the
VMHvl resulted in a dramatic decrease of natural inter-male aggression (Sano et al., 2013). Recently, optogenetic stimulation of
ER1VMHvl neurons triggered attack behavior whereas optogenetic inhibition suppressed fighting, suggesting that ER1-neurons in
this small hypothalamic area are necessary and sufficient to initiate and terminate bouts of attacks (Lee et al., 2014). Beside ER1,
neurons in the VMHvl also express a variety of other neuromodulator receptors, including serotonin, dopamine and oxytocin recep-
tors. Since many neuromodulators such as serotonin, dopamine and oxytocin change their levels dynamically during the course of
aggressive behaviors, they may influence VMHvl neuron excitability and hence aggressive attack.
Neuroanatomical track-tracing studies showed that the hypothalamic attack area forms a crossroad between input from the
medial nucleus of the amygdala and the prefrontal cortex and output to the periaqueductal gray, the medio-dorsal thalamus
and the lateral septal area (Fig. 1 and see de Boer et al., 2015 for details). Consistent with the general function of the hypothalamus
4 The Neurobiology of Offensive Aggression

in behavior, it seems likely that this brain area is also involved in the neuroendocrine and autonomic support of aggressive behavior.
The same hypothalamic area seems to be involved in other forms of aggressive behavior as well. However, it is not clear to what
extent this is due to closely overlapping dedicated networks for each form of aggression. Alternatively, one and the same network
of hypothalamic neurons might control all forms of aggressive behavior and/or a common aspect of all types of aggression (ie,
attack-bite).
A second important hypothalamic structure involved in aggression is the medial preoptic area. This hypothalamic structure is
sexually dimorphic and its function has mainly been explored in offensive aggression, maternal aggression and sexual behavior.
Its high density of gonadal steroid receptors and gonadal steroid converting enzymes such as aromatase and 5a-reductase charac-
terizes the medial preoptic area. It is one of the prime brain area’s involved in the modulation of social behavior by circulating
gonadal steroids. For example, the modulation of aggressive behavior by seasonal factors in seasonal breeding species is partly
mediated by the sensitivity of gonadal steroid receptors in this brain area.

Amygdala
Animal experiments at the end of the 19th century showed that electrolytic lesions of the amygdala had a strong taming effect in
feral animals. Even the most aggressive feral animal became tamer and calmer by a bilateral amygdala lesion. These observations
formed the basis of a wave of psychosurgery (lobotomies) in human patients in the first half of the 20th century. Despite the
increasing accuracy of the surgical techniques and the clear reduction of aggressive behavior in many of these patients, this type
of surgery was gradually abandoned, not only for ethical reasons, but also mainly because of the many behavioral side effects
due to lack of behavioral specificity.
The amygdala consists of a range of interconnected nuclei having a common output through the central nucleus and the bed
nucleus of the stria terminalis (Bnst). In particular the medial amygdala (MeA) and the Bnst have been implicated in offensive
aggression. MeA neurons are particularly active during fighting and in response to male/female conspecific chemosensory cues,
as evidenced by the induction of c-Fos and electrophysiology. In addition, lesion studies have implicated the MeA in both mating,
aggression and rage-like behavior. The medial amygdala receives direct input from the accessory olfactory bulb, which in turn is the
main relay station of olfactory information originating from the vomeronasal organ (Fig. 1). This part of the olfactory system is
specialized in the detection of species-specific chemosensory signals. Hence, olfactory stimuli that are relevant for social behavior
in general have a dedicated entrance into the brain, reaching the medial amygdala almost directly. The medial amygdala has an
important function in the modulation of social behavior on the basis of social experience and social recognition.
Recently, it was demonstrated that the majority of c-fos positive MeA neurons induced by attack are GABA-ergic, and optogenetic
photostimulation of these MeA GABA-ergic neurons triggered intense attack towards both gonadally-intact male intruders, castrated
males, females and even towards a toy mouse, indicating the independence of olfactory sensory cues. Interestingly, optokinetic
activation of neighboring glutaminergic MeA neurons effectively suppressed social behaviors and instead promoted non-social
self-grooming. Although the MeA is characterized by a high density of both ER1 and androgen receptors(AR), it is not known
whether these are located on these GABA-ergic cells. Likewise, whether and how these distinct sets of MeA neurons are precisely
interconnected with the “attack” neurons in the VMHvl is not known yet, but it seems highly feasible that they may form
a functional micro neuro-circuit for the initiation and execution of the final consummatory aspects of aggressive behavior.
Interestingly, ER1 and AR receptors are also located on neurons that produce the neuropeptide vasopressin (AVP). Interestingly,
the synthesis of AVP in these neurons is enhanced by testosterone, the male gonadal steroid hormone intimately linked to
aggressiveness. This testosterone dependent vasopressinergic system is sexually dimorphic and projects to the lateral septal area.

Septum
The lateral septum is reciprocally connected to the medial amygdala. A sexual dimorphic density of vasopressinergic fibers origi-
nating from the medial amygdala characterizes it. Males have a higher AVP fiber density than females and within the male gender
the density is negatively correlated with offensive aggression. Both in rats and mice, highly aggressive males have a less dense vaso-
pressinergic innervation of the lateral septum than low aggressive males. Early studies showed that lesions of the septal area in rats
induce the septal rage syndrome, which is in fact an extremely defensive reaction not only to conspecifics but also to humans or any
other threatening stimulus. This suggests that the contribution of the septum in aggressive behavior might be more anxiety-related.
Indeed, recent studies using microdialysis of vasopressin during offensive aggression in rats show that males characterized by low
anxiety and high aggression reduce AVP release in the lateral septum during the aggressive interaction. It seems that septal AVP is
involved in the acute modulation of non-aggressive social and anxiety related behaviors.

Prefrontal Cortex
The prefrontal cortex consists of a number of sub-regions defined on the basis of their connections with thalamic nuclei and
neuronal cyto-architecture. Although the degree of complexity increases in higher vertebrates, there is a clear homology of prefrontal
structures across a wide variety of vertebrate species. In particular the ventromedial infralimbic (PFCvm) and orbitofrontal (OFC)
cortical areas have been associated with the inhibitory control of offensive aggression in a number of species. This brain area is more
generally involved in behavioral inhibition or impulsivity and the planning of motor output. Indeed, measures of impulsivity in
 The Neurobiology of Offensive Aggression 5

male hamsters are positively correlated with offensive aggression and prefrontal cortex activity measured by c-fos expression.
Moreover, reduced prefrontal cortex serotonergic input and functioning has been associated with impulsive and violent forms of
aggression in animals and humans. Recently, Takahashi et al. (2014) demonstrated by employing optogenetic techniques that
photo-stimulation of the principal pyramidal excitatory neurons in the mPFC, but not in the OFC, potently suppressed the initi-
ation and execution of intermale aggression in mice, while optogenetic silencing of mPFC neurons caused an escalation of aggres-
sive behavior. Hence, it is very plausible that the mPFC inhibits activity of a neural circuit that is tightly controlling the execution of
aggressive attacks (ie, VMHvl or MeA), although the VMHvl does not seem to receive direct input from this area. Indeed, the
prefrontal cortex is closely connected to both the amygdala and the brain structures involved in motor output such as the motor
cortex and the basal ganglia (see Fig. 1).

Periaqueductal Gray (PAG)

C-fos studies in several species consistently show that the periaqueductal gray is activated during offensive aggression. It is therefore
generally considered to be part of the neuronal network of offensive aggression. However electrical and neurochemical
manipulation of this brainstem structure mainly affects defensive behavior and many of the accompanying sympathetic and
parasympathetic responses rather than true offense. The PAG has a columnar structure. The dorsolateral and lateral column is
more generally associated with an active behavioral and physiological response to serious environmental challenges, whereas
the ventrolateral PAG is associated with a quiescent response to stress and the accompanying physiology. These columns have
extensive and differential connections to brainstem areas involved in the control of the autonomic nervous system and
analgesia. Hence, it is conceivable that the periaqueductal gray has a role in the network of aggression mainly by controlling
autonomic arousal and anti-nociception that accompany offensive and defensive aggression.

Neurochemical Modulation of the Aggressive Neural Network

Obviously, the functional activity of this entire social behavior neural network, and thereby the selection of the appropriate
behavioral response to social challenges and opportunities, is determined by a wide variety of molecular substrates (ie, neuro-
transmitters, hormones, cytokines, and their respective metabolic enzymes, receptors, and intraneuronal signaling molecules).
Undisputedly, among the neurochemical systems that are considered key signaling molecules in this neurocircuitry controlling
aggression are the monoamines serotonin (5-HT) and dopamine (DA), the “social” neuropeptides oxytocin (OXT), and vaso-
pressin (AVP), the “stress” neuropeptide corticotropin releasing factor (CRF), the “stress” HPA- and “sex” HPG-axis’s steroid
hormones (corticosterone, testosterone, estrogen), and their cognate receptors.

Serotonin
All components of the neuronal network for offensive aggression are substantially innervated by serotonergic (5-HT) neurons
originating in the dorsal and medianraphé nuclei in the brain stem (Fig. 2). More than any other neurochemical system, this
evolutionary ancient and very well conserved neurotransmitter system is considered the primary molecular modulator of
aggression in a wide variety of animal species, including man (Siever, 2008; Nelson and Trainor, 2007). However, the direction
and exact causal linkage of this association is very complex and it has proven notoriously difficult to unravel the precise role of
this amine (and every facet of its synthetic and metabolic pathways, uptake and storage processes, and dynamic receptor
signaling mechanisms) in the predisposition for and execution of aggressive behavior in both its normal and pathological
forms. For decades, high levels of aggressive behavior are believed to be associated with low brain 5-HT neurotransmission
activity. This frequently reiterated serotonin deficiency hypothesis seems consistent with the fact that serotonergic receptor
agonists used to mimic higher serotonergic activity, generally reduce aggressive behavior. However, recent studies of the func-
tional status of the 5-HT system before, during, and after the execution of normal adaptive and abnormal pathological forms of
aggression have led to a somewhat different view. Display of normal adaptive offensive aggressive behavior aimed at territorial
control and social dominance is associated with a higher 5-HT neuronal activity (see de Boer et al., 2015 for relevant refer-
ences). A negative correlation between aggression and 5-HT as captured in the deficiency hypothesis seems to be a trait-like
characteristic of pathological forms of aggression (eg, violence). For example, a clear positive correlation was found between
the level of normal adaptive expressions of offensive aggression and basal cerebrospinal fluid (csf) concentrations of 5-HT and/
or its metabolite 5-HIAA. A significant negative correlation between aggression and 5-HT levels was found only upon inclusion
of samples from abnormallydand excessively aggressive trained fighter animals. A critical evaluation of the csf 5-HIAA data in
aggressive humans confirms this idea that the serotonergic deficiency appears to hold in particular for specific groups of indi-
viduals who persistently engage in more aberrant, impulsive and violent forms of aggressive behavior rather than in individuals
with instrumental (functional) forms of offensive aggression.
Treatment with 5-HT1A or 5-HT1B receptor agonists is one of the most potent pharmacological methods to selectively
suppress aggressive behavior in a variety of animal species and experimental paradigms. Apart from acting on receptors at post-
synaptic sites, these two receptor agonists also affect the two main serotonergic auto-receptors involved in the negative feedback
control of the 5-HT neuron at the level of the synapse (5-HT1B) and at the level of the cell soma (5-HT1A) (see Fig. 2).
6 The Neurobiology of Offensive Aggression

Figure 2 2A: Serotonergic control of the social behavior neuronal network. See legend of Fig. 1. DRN, dorsal raphé nucleus; MRN, medial raphé
nucleus. 2B: More detailed neuromolecular characteristics of part of the DRN-prefrontal-HAA microcircuitry involved in the control of aggressive
behavior. Glu, glutamate; AVP, arginine vasopressin; 5-HT, serotonin; 5-HTT, serotonin transporter.

Activation of these receptors by agonists will potently activate the negative feedback and thereby reduce 5-HT neurotransmis-
sion. It appears that the anti-aggressive effects of these compounds are largely expressed via their action on the inhibitory
auto-receptors located at the cell soma and the nerve terminal, presumably by attenuating intruder-activated 5-HT neurotrans-
mission. Interestingly, highly aggressive animals are characterized by upregulated somatodendritic 5-HT1A and terminal
5-HT1Bautoreceptor functionality. This considerably (approximately 20-fold) enhanced tonic inhibitory control of serotonergic
neurons in aggressive males may explain the negative correlation between baseline levels of 5-HT and escalated aggression
found in many species. Furthermore to signify the causality of this correlation, 5-HT1Aautoreceptor sensitivity increased or
decreased upon enhancing (by repeated victorious experiences) or attenuating (by repeated defeat experiences) aggressiveness,
respectively. Notably, animals that escalated their aggressiveness and started to engage in violent-like aggressive behavior
demonstrated 5-HT1Aautoreceptor super-sensitivity. More persuasively, recent molecular genetic studies have shown that trans-
genic mice with conditional (at adult age) overexpression of somatodendritic 5-HT1Aautoceptors demonstrate suppressed 5-HT
neural firing that was associated with a profound hyper aggressive behavioral phenotype (Audero et al., 2013). These data
confirm the causal role of tonic 5-HT activity in setting a trait-like threshold for executing overt aggressive behavior.
In addition to the 5-HT1A and 5-HT1B receptor, the intrinsic feedback control of the serotonin neuron is also mediated by the
serotonin transporter (SERT). In humans and non-human primates, a polymorphism in the promoter region of the gene coding for
SERT is associated with aggression. In particular the short allele variant is associated with increased levels of aggression both in
males and females, presumably due to a reduced serotonin reuptake. In contrast, however, genetic loss-of-function mutations of
the SERT gene in rats and mice is associated with diminished aggressiveness.

Dopamine
Dopamine has several important functions in the general control of behavior. The nigrostriatal dopaminergic system originating in
the substantia nigra (SNR) (see Fig. 3) has a central function in the control of motor output, which is obviously important for
 The Neurobiology of Offensive Aggression 7

Figure 3 Dopaminergic control of the social behavior neuronal network. See legend of Fig. 1. SNR, substantia nigra; VTA, ventral tegmental area.

offensive aggression as well. The involvement of the mesolimbic dopaminergic system originating in the ventral tegmental area
(VTA) (see Fig. 3) in behavior is only partially understood. It is involved in the preparation, execution and consequences of aggres-
sive acts. At the level of the prefrontal cortex, the balance between the dopaminergic and serotonergic innervation is an important
determining factor in prefrontal cortex functioning and hence in impulsivity and control of goal directed behavior including offen-
sive aggression. Most of the mesolimbic dopamine effects on aggression are mediated by the dopamine D2 receptor. In general,
dopamine D2 antagonists reduce aggressive behavior in a variety of animal species. However, there is a general lack of behavioral
specificity of this effect when the compounds are applied systemically. When administered at the level of the hypothalamus, dopa-
mine seems to selectively modulate androgen-enhanced aggression through a D2 receptor mediated inhibition of GABA-ergic
neurons.
The dopaminergic projection from the ventral tegmental area to the nucleus accumbens (Fig. 3) is generally considered to
be involved in reward and the development of addiction. Numerous studies in a wide variety of animal species have convinc-
ingly demonstrated that in addition to securing access to resources, the most intriguing consequence of winning an aggressive
conflict is the self-reinforcing effect of this type of behavior. Actually, individuals seek out the opportunity to fight and engaging
in aggressive behavior appears to be a source of pleasure. The most convincing evidence that successful aggression seems
rewarding to animals is that the opportunity to engage in aggressive behavior can reinforce operant responding, ie, animals
are willing to work (eg, bar pressing, nose poking) for aggression as a source of reward and satisfaction (see Miczek et al.,
2007 for review). Not surprisingly, just like other events that function as positive reinforcers such as food, drugs or sex, the
mesocorticolimbic dopamine system is associated with the incentive salience of the rewarding properties of winning fights.
Nucleus accumbens (NAcc) dopamine is strongly released during (anticipation of) aggressive episodes, and pharmacological
antagonism of dopamine D1/D2 receptors in the NAcc diminishes the seeking of the opportunity to fight (Couppis and
Kennedy, 2008). In addition, direct optogenetic activation of ventral tegmental area (VTA) dopamine neurons increases aggres-
sion (Yu et al., 2014), proving that dopamine function and aggression are causally linked. However, to what extent these types
of manipulations selectively affect the rewarding/motivational components of aggressive behavior or modulate other behav-
ioral control functions is not clear yet.

GABA
GABA is the main inhibitory neurotransmitter in the brain. It is not surprising therefore that many compounds that act as an agonist
at the GABA-benzodiazepine receptor complex potently reduce aggressive behavior. This is partially due to the general tranquilizing
and sedative effects of GABA-ergicmimetic compounds. However, at low doses, benzodiazepines such as diazepam may enhance
aggressive behavior, a phenomenon that is not well understood so far. Moreover, also the aggression enhancing effects of
alcohol are mediated by its action on the GABA-benzodiazepine receptor. Indeed, the aggression enhancing effects of alcohol
can be antagonized using non-selective GABA-benzodiazepine receptor antagonists. Recent evidence in mice has demonstrated
that the modulation of 5-HT impulse flow by GABA, acting via distinct receptor subtypes in the dorsal raphe nucleus, is of
critical importance in the suppression and escalation of aggressive behavior (Miczek et al., 2015b).

Hormonal Modulation of the Network

Testosterone is the gonadal steroid hormone that has traditionally been associated with male offensive aggression. Indeed, castra-
tion in adulthood may reduce aggression and testosterone suppletion in castrated males will enhance aggression. Winning of an
aggressive interaction leads to a transient testosterone spike that enhances the chance of winning subsequent interactions. Losing
a fight will reduce plasma testosterone levels for a long period of time. Despite the clear causal involvement of testosterone in
aggression, the correlation between individual levels of circulating testosterone and aggression is often low or even absent.
8 The Neurobiology of Offensive Aggression

This is due to the fact that social experience and contextual factors may strongly interfere. Moreover, its pulsatile secretion char-
acterizes testosterone plasma dynamics, which is a complicating factor to assess plasma levels on the basis of only a single or
a few samples.
The behavioral effects of testosterone are partly due to its action on peripheral secondary sex characteristics, changing the stim-
ulus characteristics of the animal. More important however is its action on the brain. During development, testosterone plays an
important role organizing brain and behavior into a more masculine direction. Characteristic of the neuronal network for offensive
behavior is that several of the forebrain structures are sensitive to gonadal steroids. The medial amygdala, bed nucleus of the stria
terminalis and preoptic area are characterized by a high density of estrogen and androgen receptors. Moreover, these structures
contain high amounts of aromatase and 5a-reductase, enzymes that convert testosterone into estradiol and the androgen di-
hydro-testosterone respectively. Both metabolites of testosterone play a distinct role in the modulation of offensive aggression
through their respective action on estrogen and androgen receptors. In particular the ER1 receptor has been implicated in the modu-
lation of offensive aggression. Deleting this receptor through genetic modification completely abolishes aggression in male mice.
Hence, individual variation in offensive aggression may depend on the density of ER1 and the amount of aromatase present in
various brain areas. Indeed, aggressive males show higher numbers of ER1 expressing cells in the lateral septal area, the Bnst and
the preoptic area. Moreover, the number of ER1 receptors increases with aggression in seasonal reproducing animals. However,
also the androgen receptor is involved. Evidence in mice suggests that the androgen receptor increases in various limbic brain areas
after winning experience. One can conclude that the effects of testosterone on aggressive behavior depend on the complex interac-
tion between the balance between the two converting enzymes, the density of ER a and androgen receptors and experiential and
contextual factors.
The prominent link between gonadal steroids and enhanced aggression is further clearly demonstrated in animals and humans
that are exposed to anabolic/androgenic steroids (AAS) during adolescence (Morrison et al., 2016). Several studies have shown that
AAS exposure during this developmental period consistently increased aggressive behavior via alterations in several neurotrans-
mitter systems (ie, 5-HT, DA and AVP) implicated in the control of aggression within the hypothalamic attack area.

Corticosterone
Corticosterone or cortisol is the main hormone of the pituitary adrenocortical axis secreted by the adrenal cortex in response to
environmental challenges. It is has an important function in metabolism and in stress and adaptation. It is beyond the scope of
this article to elaborate on its main function. However, recent evidence suggests that it is involved in aggression and the develop-
ment of violence as well; low to moderate acute changes are associated with increased aggression in rats whereas higher,
longer-lasting changes are associated with decreased aggression in mice. Recently, Kruk proposed an interesting concept that the
anticipation of an impending conflict rapidly activates the HPA axis, producing an adrenocortical response that promotes an
increased sensitivity for aggression releasing and directing stimuli by a rapid feedforward to appraisal mechanisms in the brain.
This stress-aggressive conflict feedforward mechanism seems to depend on the mineralocorticoid receptor (MR) as pretreatment
with the MR antagonist spironolactone robustly inhibits resident’s offensive aggression towards an intruder.
In particular, sustained glucocorticoid deficiency observed in a number of psychiatric disorders such as antisocial person-
ality disorder and posttraumatic stress disorder is associated with abnormal aggressive behavior observed in these patients.
The causal involvement of glucocorticoid deficiency in aggression and violence has been demonstrated in male rats (Haller
and Kruk, 2006). Glucocorticoid deficiency created by surgical removal of the adrenal gland and low corticosterone replace-
ment induced violent-like forms of aggressiveness in male rats. These animals direct their attacks to vulnerable parts of the
body an effect that is normalized by corticosterone injections. Further analysis of the abnormal aggressive behavior suggests
that the effects of chronic glucocorticoid deficiency on aggression and violence are somehow related to anxiety and/or physi-
ological hypo-arousal.

Vasopressin and Oxytocin

Besides their important peripheral physiological functions as neurohypophysial-released hormones, the neuropeptides arginine
vasopressin (AVP) and oxytocin (OXT) are also implicated in inter-neuronal communication within various nodes of the social
brain network to modulate emotional and social behavioral and physiological responding (Lee et al., 2009a). AVP is generally
known to increase anxiety-like behaviors, stress responsivity and aggressiveness; whereas OXT has the opposite effects and facil-
itates social attachment, care, and affiliation (Heinrichs et al., 2009). Existing data from early pioneering work on these neuro-
peptides convincingly demonstrated opposite roles for AVP and OXT in fear learning processes (Bohus and de Wied, 1998). More
recent studies in our wild-type rats and/or artificially selected aggressive and non-aggressive mice have demonstrated that high-
aggressive animals exhibit higher levels of AVP release when compared to their non-aggressive counterparts (Koolhaas et al.,
2010). In addition, there is abundant experimental evidence to support a causal function of vasopressin in proactive aggressive
behavior and OXT in passive affiliative behavior. Direct micro-infusion of AVP or OXT into the cerebral ventricles or in selected
brain regions facilitates or suppresses, respectively, offensive aggression (Calcagnoli et al., 2015). In addition, a positive corre-
lation between levels of CSF vasopressin and life history of general aggression as well as aggression towards individuals
(Lee et al., 2009) has been reported, whereas impaired brain OXT-ergic signaling has been implicated in several human
neuropsychiatric disorders associated with social deficits, impulsivity, and excessive aggression (Lee et al., 2009).
 The Neurobiology of Offensive Aggression 9

Furthermore, mutant mice with the vasopressin receptor V1A/B gene deleted showed virtually no offensive aggressive behavior
anymore, whereas elevated aggressiveness was found in mice with deletions of the OXT receptor gene. Consistent with the
aggression-promoting role of brain AVP, systemic as well as intra-hypothalamic administration of AVP V1A/B receptor
antagonists effectively block offensive aggressive behavior in male hamsters and WTG rats (Koolhaaset al., 2010). Basically,
an opposite picture seems to emerge for brain OXT signaling. Recent ethopharmacological studies have clearly demonstrated
that enhancement of brain OXT-ergic function, using both intraventricular, intra-amygdalar, and even intranasal
administration routes, produced marked anti-aggressive and pro-social affiliative effects that are dose-dependent, behavior-
and receptor-selective, and long-lasting (Calcagnoli et al., 2013, 2015).
Based on the findings outlined above, it can be hypothesized that an endogenous balance between vasopressin and oxytocin
signaling within (components of) the social behavioral neural circuit may gate the expression of either aggressive or affiliative
responses to salient social stimuli.

Neurogenetics of Aggression

Similar to other behaviors, aggressiveness is influenced by both genetic and environmental factors. Heritability-estimates for aggres-
sive behavioral traits in both animals and humans range between 35 and 60%, indicating that a substantial genetic component is
responsible for this innate behavioral trait. Despite a decade of hunting for the responsible genes, employing a variety of molecular
genetic techniques, no clear picture of the candidate genes or gene-networks underlying trait aggressiveness has been unequivocally
established yet (Vassos et al., 2013). Similarly, genome-wide association studies (GWAS) in humans have largely failed to identify
common genetic variants that underpin aggressive behavior.
Nevertheless, the era of molecular genetics has generated a wealth of transgenic animals, some of which were specifically targeted
at genes related to offensive aggression. Others were developed for different purposes, but appeared to differ from their controls in
the level of aggressive behavior. Hence, analysis of the neurobiological systems affected in these transgenic animals may elucidate
additional components of the neurobiology of offensive aggression. Manipulation of genes directly involved in the molecular
signaling cascade of serotonergic, dopaminergic and vasopressinergic/oxytocinergic neurotransmission clearly affect offensive
behavior as predicted from the current knowledge of the involvement of these systems in offensive aggression.An example of
this is given in Table 1, which summarizes the effects on offensive aggression of genetic modification in various facets of the sero-
tonergic system.
Given the complexity of social aggressive behaviors, it should not be surprising that a large number of genes have been
associated with aggressiveness in both human and animal studies. Currently, there are over 100 genetically modified mouse
lines known that show increased or decreased levels of offensive aggression (see Freudenberg et al., 2015 for recent compre-
hensive review). Quite a number of these transgenic animals show changes in aggressive behavior that could not be predicted
from the existing knowledge of the molecular neurobiology of aggression. However, changes in serotonergic neurotransmission
through unknown molecular pathways seem to be a common factor in the altered levels of aggression in many transgenic
mouse lines.

Table 1 Genetic modification of serotonergic system components and its effects on offensive aggression

Effects on
Gene (chromosome) Coding protein product Method aggression

Pet-1(Fev) (1) Plasmacytoma expressed transcript; ETS domain Knockout [

transcription factor for 5-HT
Tph (11) Tryptophan hydroxylase Polymorphism [
Tph2 (10) Tryptophan hydroxylase 2 isoform Knockout [
R441H mutation Polymorphism [
Knockout
Knock-in
Htr1a (13) 5-HT1A receptor Knockout Y
Conditional overexpression [
In 5-HT neurons
Htr1b (9) 5-HT1B receptor Knockout [
HTr2a (14) 5-HT2A receptor Polymorphism [
HTtr2b (14) 5-HT2B receptor Knockout [
HTr6 (4) 5-HT6 receptor Polymorphism [
HTr3A (9) 5-HT3 receptor Polymorphism Y
Maoa (X) Monoamine oxidase A Knockout [
Slc6a4 5-HT transporter Knockout Y
SERT (11)
10 The Neurobiology of Offensive Aggression

Synthesis and Outlook

A large body of animal neurobehavioral research convincingly demonstrates that abnormal expressions of aggressive behavior
principally find its origin in a dysregulation of the deeply rooted neuronal circuits and/or neurochemical pathways in the brain
that mediate normal social affiliative-aggressive behaviors. This highly conserved neural and gene expression brain network
encompasses monoaminergic neurons in the mesencephalon projecting to hypothalamic nuclei, amygdaloidal, septal,
prefrontal, and hippocampal forebrain regions, striatal and thalamic loops with the frontal and prefrontal cortex, as well as
important feedback loops to limbic and mesencephalic nuclei. The structural and functional properties of this social aggressive
behavior brain network are established and constantly shaped by a dynamic interplay of genetic and environmental factors
(stress, maltreatment, vicarious experiences, substance abuse) in particular during certain sensitive (ie, perinatal and adoles-
cent) developmental periods. Undisputedly, among the neurochemical systems that are considered key signaling molecules
in this neurocircuitry controlling aggression are the canonical monoamines serotonin (5-HT) and dopamine (DA), the “social”
neuropeptides oxytocin (OXT) and vasopressin (AVP), the “stress” neuropeptide CRF, the “stress” HPA- and “sex” HPG-axis’s
steroid hormones (corticosterone, testosterone, estrogen) and their receptors. Evidently, recent genetic studies in both human
and animals have demonstrated that polymorphisms or mutations in a number of genes regulating the functional activity of
these important signaling molecules may confer risk factors, either alone but usually in co-action with (early) stressful life
conditions, for development of antisocial aggressive traits. Particularly, from the viewpoint of targeting novel molecular
sites for intervention, the intrinsic 5-HT autoregulatory mechanisms (ie, the presynaptic 5-HT1A/B autoreceptors and 5-HTT),
and extrinsic neuropeptidergic (ie, OXT, AVP and CRF) and steroid receptor (ie, MR and AR) modulatory influences of 5-
HT signaling are emerging as important molecular determinants of escalated aggression regulation. Although early efforts
during the 1950 and 1960s to translate preclinical neurobiological aggression research findings into clinical use have
a sordid history, the current emerging circuit-level knowledge of the neuromolecular underpinnings of aggression in both
its normal and excessive forms has great potential to guide the rational development of effective therapeutic interventions
for pathological social and aggressive behavior.

References

Anderson, D.J., 2012. Optogenetics, sex, and violence in the brain: implications for psychiatry. Biol. Psychiatry 71, 1081–1089.
Audero, E., Mlinar, B., Baccini, G., Skachokova, Z.K., Corradetti, R., Gross, C., 2013. Suppression of serotonin neuron firing increases aggression in mice. J. Neurosci. 33 (20),
8678–8688.
Bard, P., 1928. A diencephalic mechanism for the expression of rage with special reference to the sympathetic nervous system. Am. J. Physiol. 84, 490–515.
Bohus, B., de Wied, D., 1998. The vasopressin deficient Brattleboro rats: a natural knockout model used in the search for CNS effects of vasopressin. Prog. Brain Res. 119,
555–573.
Brain, P.F., 1979. Differentiating types of attack and defense in rodents. In: Brain, P.F., Benton, D. (Eds.), Multidiscriplinary Approaches to Aggression Research. Elsevier,
Amsterdam, pp. 53–77.
Brown, S., Schaefer, E.A., 1888. An investigation into the functions of the occipital and temporal lobe of the monkey’s brain. Phil. Trans. R. Soc. B 179, 303–327.
Calcagnoli, F., de Boer, S.F., Althaus, M., den Boer, J.A., Koolhaas, J.M., 2013. Antiaggressive activity of central oxytocin in male rats. Psychopharmacology 229, 639–651.
Calcagnoli, F., Stubbendorff, C., Meyer, N., de Boer, S.F., Althaus, M., Koolhaas, J.M., 2015. Oxytocin microinjected into the central amygdaloid nuclei exerts anti-aggressive effects
in male rats. Neuropharmacology 90, 74–81.
Couppis, M.H., Kennedy, C.H., 2008. The rewarding effect of aggression is reduced by nucleus accumbens dopamine receptor antagonism in mice. Psychopharmacology 197,
449–456.
De Boer, S.F., Caramaschi, D., Natarajan, D., Koolhaas, J.M., 2009. The vicious cycle towards violence: focus on the negative feedback mechanisms of brain serotonin
neurotransmission. Front. Behav. Neurosci. 3, 52.
De Boer, S.F., Olivier, B., Veening, J., Koolhaas, J.M., 2015. The neurobiology of aggression: revealing a modular view. Physiol. Behav. 146, 111–127.
Deisseroth, K., 2014. Circuit dynamics of adaptive and maladaptive behavior. Nature 505 (7483), 309–317.
Freudenberg, F., Gutierrez, H.G., Post, A.M., Reif, A., Norton, W.H.J., 2015. Aggression in non-human vertebrates: genetic mechanisms and molecular pathways. Am. J. Med.
Genet. part B 9999, 1–38.
Goltz, F.L., 1884. Ueber die verrichtungen des grosshirns:funfte abhandlung. Archiv fur die gesammte physiologiedes menschen und der thiere. 34, 450–505.
Goodson, J.L., 2005. The vertebrate social behavior network: evolutionary themes and variations. Horm. Behav. 48, 11–22.
Haller, J., Kruk, M.R., 2006. Normal and abnormal aggression: human disorders and novel laboratory models. Neurosci. Biobehav. Rev. 30, 292–303.
Heinrichs, M., von Dawans, B., Domes, G., 2009. Oxytocin, vasopressin, and human social behavior. Front. Neuroendocrinol 30 (4), 548–557.
Hess, W.R., Bruegger, M., 1943. Das subkortikale Zentrum der affectiven Abwehrreaktion. Helv. Physiol. Acta 1, 33.
Kruk, M., 2014. Hypothalamic attack: a wonderful artifact or a useful perspective on escalation and pathology of aggression? A viewpoint. Curr. Top. Behav. Neurosci. 313.
Koolhaas, J.M., de Boer, S.F., Buwalda, B., van Reenen, R.K., 2007. Individual variation in coping with stress: a multidimensional approach of ultimate and proximate mechanisms.
Brain Behav. Evol. 70, 218–226.
Koolhaas, J.M., De Boer, S.F., Coppens, C.M., Buwalda, B., 2010. Neuroendocrinology of coping styles: towards understanding the biology of individual variation. Front. Neu-
roendocrinol. 31, 307–321.
Lee, H.J., Macbeth, A.H., Pagani, J.H., Young, W.S., 2009a. Oxytocin: the great facilitator of life. Prog. Neurobiol. 88, 127–151.
Lee, R., Ferris, C., Van de Kar, L.D., Coccaro, E.F., 2009b. Cerebrospinal fluid oxytocin, life history of aggression, and personality disorder. Psychoneuroendocrinology 34,
1567–1573.
Lee, H., Kim, D.W., Remedios, R., Anthony, T.E., Chang, A., Madisen, L., Zeng, H., Anderson, D.J., 2014. Scalable control of mounting and attack by Esr1þ neurons in the
ventromedial hypothalamus. Nature 509, 627–632.
Lin, D., Boyle, M.P., Dollar, P., Lee, H., Lein, E.S., Perona, P., Anderson, D.J., 2011. Functional identification of an aggression locus in the mouse hypothalamus. Nature 470,
221–226.
Miczek, K.A., de Almeida, R.M., Kravitz, E.A., Rissman, E.F., de Boer, S.F., Raine, A., 2007. Neurobiology of escalated aggression and violence. J. Neurosci. 27 (44), 11803–11806.
Miczek, K.A., de Boer, S.F., Haller, J., 2013. Excessive aggression as model of violence: a critical evaluation of current preclinical methods. Psychopharmacology 226, 445–458.
 The Neurobiology of Offensive Aggression 11

Miczek, K.A., Takahashi, A., Gobrogge, K.L., Hwa, L.S., de Almeida, R.M., 2015a. Escalated aggression in animal models: shedding new light on mesocorticolimbic circuits. Curr.
Opin. Behav. Sci. 3, 90–95.
Miczek, K.A., DeBold, J.F., Hwa, L.S., Newman, E.L., DeAlmeida, R.M.M., 2015b. Alcohol and violence:neuropeptidergic modulation of monoamine systems. Ann. N.Y. Acad.
Sci. 1–23.
Morrison, T.R., Sikes, R.W., Melloni, R.H., 2016. Aanabolic steroids alter the physiological activity of aggression circuits in the lateral anterior hypothalamus. Neuroscience 315,
1–17.
Nelson, R.J., Trainor, B.C., 2007. Neural mechanisms of aggression. Nat. Rev. Neurosci. 8, 536–546.
Newman, S.W., 1999. The medial extended amygdala in male reproductive behavior. A node in the mammalian social behavior network. Ann. N.Y. Acad. Sci. 877, 242–257.
O’Connell, L.A., Hofmann, H.A., 2012. Evolution of a vertebrate social decision-making network. Science 336 (6085), 1154–1157.
Sano, K., Tsuda, M.C., Musatov, S., Sakamoto, T., Ogawa, S., 2013. Differential effects of site-specific knockdown of estrogen receptor alpha in the medial amygdala, medial pre-
optic area, and ventromedial nucleus of the hypothalamus on sexual and aggressive behavior of male mice. Eur. J. Neurosci. 37, 1308–1319.
Siever, L.J., 2008. Neurobiology of aggression and violence. Am. J. Psychiatry 165, 429–442.
Takahashi, A., Miczek, K.A., 2014. Neurogenetics of aggressive behavior: studies in rodents. Curr. Top. Behav. Neurosci. 17, 3–44.
Vassos, E., Collier, D.A., Fazel, S., 2013. Systematic meta-analyses and field synopsis of genetic association studies of violence and aggression. Mol. Psychiatry 1–7.
Yu, Q., Teixeira, C.M., Mahadevia, D., Huang, Y., Balsam, D., Mann, J.J., Gingrich, J.A., Ansorge, M.S., 2014. Optogenetic stimulation of DAergic VTA neurons increases
aggression. Mol. Psychiatry 19 (6), 688–698.