The roles of serotonin and dopamine in reactive and proactive aggression

(1)

The roles of serotonin and

dopamine in reactive and

proactive aggression

A literature thesis in partial fulfilment of the

requirements for the degree of Master of

Science

Jonathan Krikeb, BSc, 10065180

12 June 2015

Supervisor:

Co-assessor:

(2)

Abstract:

Aggression is often linked to violence but this is not a necessary connection. Aggression could also be motivating choices for economic-decision making. The question of what leads to aggression is what this paper will address as it discusses the bi-modal classification of aggression: proactive and reactive. These two classes will be linked to a new predator-prey research paradigm that separates the greed and its proactive tendencies, from the fear and its reactive actions. This, as well as a few other economic games, will be linked to the wide scope of research into aggressive violent

behaviour, that is mostly based on clinical cases, as well as decision-making research that is founded on the idea that focuses on impulsive behaviour as it has been linked to aggression in the past. These past findings have also found correlations between serotonin hypoactivity, and also dopamine hyperactivity, in cases of irregular aggressive behaviour. We will attempt to establish how activities of the serotonergic and the dopaminergic circuitries parallel aggression in predator and prey type of interactions.

(3)

1. Introduction

An attack to gain more resources by someone who has an abundance of them does not have the same motivation as the defensive reaction of the prey, in this same scenario, that defends its limited resources. In this sort of predator against prey interaction, two types of aggression come into play: proactive and reactive. Reactive aggression has been widely studied, often in context of impulsive aggression, while the literature on proactive aggression is more scarce. In light of the research into the motives behind these behaviours, and how this is reflected in terms of neurotransmitters, this paper will follow past studies and look at how the key neurotransmitters serotonin and dopamine interact in these two different types of aggression, namely predator and prey.

1.1. Aggressive behaviour

Aggression traditionally requires a conflict between at least two parties that may compete for the same object, physical or otherwise (Nelson & Trainor, 2007).Since aggression and violence often go hand-in-hand, the results of aggressive behaviour often lead to damage, which is often physical, and therefore aggression is dangerous and not always the best choice in conflict circumstances. It does have an evolutionary role in food, or mate, acquisition, or protection, as well as inner

motivations such as fear, greed, anger or even pleasure. When these motives lead to aggressive acts that hurt, or injure others, then we consider aggression as unaccepted in our current human society – war or criminal acts such as robbery or battery (de Almeida, Ferrari, Parmigiani, & Miczek, 2005). There is a wide selection of literature available on aggression, much of it is composed of

psychological research focusing on clinical and criminal cases such as: workplace aggression (Hills & Joyce, 2013; Piquero, Piquero, Craig, & Clipper, 2013; S. F. Smith & Lilienfeld, 2013), domestic aggression (George et al., 2001; Soler, Vinayak, & Quadagno, 2000), alcohol and drug related (Anholt

(6)

& Mackay, 2012; Badawy, 2003; de Almeida et al., 2005; George et al., 2001; Gowin, Swann, Moeller, & Lane, 2010; Skara et al., 2008), and arson (Linnoila, Virkkunen, & Scheinin, 1983). All this research shows how ingrained in human society aggression is and how destructive it could get, thus

necessitating a deeper understanding of the motives behind it.

In order to better understand aggression and its motives, different classification systems were established. In this paper, as the introduction suggests, we will focus on a binary classification system. Two other systems were suggested in previous studies as well. First, as described both by Siegel and Victoroff (2009) and Umukoro, Aladeokin, and Eduviere (2013) where they divide

aggression into seven separate motivations it may originate from: fear -induced, maternal, irritable, inter-male, sex-related, predatory, and territorial. However, in the case of humans, these are not as clear-cut cases as they are in the animal world.

A third classification is offered by separating the motives into four categories: stress and fear-induced, anger and frustration-fear-induced, instrumental offence, and pleasure motivated. However, in both cases of the alternative classifications, we could collapse them into a bimodal classification system that is easier to apply for humans, where the study of aggression is complex as it is. An example used by Siegel and Victoroff (2009) is a gang fight where the mix of planning, inter-male dominance, and emotions of anger or fear interact to spark the fighting. In such a bimodal classification, one side would be predatory, goal-oriented, calculated, instrumental, proactive, premeditated aggression. This would be contrasted with a reactive, impulsive, protective, defensive, hostile aggression (Malone et al., 1998; McEllistrem, 2004; Nelson & Trainor, 2007; Siegel &

Victoroff, 2009; Umukoro et al., 2013; Weinshenker & Siegel, 2002). Whereas the former has very little literature looking into it, the latter sort is more related to pathological aggression, and often associated in research with reduced levels of the neurotransmitter 5-hydroxytryptamine (5-HT or

(7)

serotonin) (Anholt & Mackay, 2012; Badawy, 2003; Berman, Tracy, & Coccaro, 1997; Booij et al., 2010; Brown et al., 1982; Brown, Goodwin, & Ballenger, 1979; Crockett, Clark, Lieberman, Tabibnia, & Robbins, 2010; Crockett, Clark, & Robbins, 2009; Crockett, Clark, Tabibnia, Lieberman, & Robbins, 2008; Daw, Kakade, & Dayan, 2002; de Boer & Koolhaas, 2005; Linnoila et al., 1983; Perez, 2012; Wetzler, Kahn, Asnis, Korn, & van Praag, 1991). These studies relate many pathologies related to impulse control including drug and alcohol addiction, pyromancy, suicidal tendencies, as well as repeated violent crimes, to observed low levels of serotonin. While it is important to notice that these studies find correlation, and not causation, we will examine throughout this paper the role of serotonin in aggression and impulsive behaviour.

Fig. 1. Bi-modal classification of aggression and where neurotransmitter research has been involved so far.

1.2. Bi-modal classification of aggression

As suggested above, this paper will focus on a bimodal classification of aggression, namely: predatory, and reactive. This bimodal classification is discussed in in depth in a few articles

(8)

(Chichinadze, Chichinadze, & Lazarashvili, 2011; Malone et al., 1998; McEllistrem, 2004; Siegel & Victoroff, 2009; Umukoro et al., 2013; Weinshenker & Siegel, 2002). Reactive aggression is closely related in the literature to affective defence, or aggression (McEllistrem, 2004; Weinshenker & Siegel, 2002). This type of aggression is much more studied in comparison to the other, predatory type (Siegel & Victoroff, 2009). In animal models, it is easily classified since it is expected in cases of invasion to personal space or ingression on food reserves. This aggression is also easily measured in terms of strong sympathetic nervous system activation (Weinshenker & Siegel, 2002). This suggests that this type of aggression is instinctive, and therefore impulsive – there is no consideration of the long-term results, only immediate elimination of the current threat. Implicitly this links this

aggression to emotions: anger, anxiety, and fear, and thus to the limbic system. The impulsive aspect will be discussed further since it is a core concept in criminal and clinical studies of aggression.

The other type of aggression of interest in this paper is offensive, predatory aggression. Amongst animals this is usually the manner in which a predatory animal hunts and consumes a prey animal. However, our focus should be, in order to compare to human cases, on intraspecies aggression such as climbing up the social hierarchy amongst groups of monkeys where one aims to dominate

(Chichinadze et al., 2011). A human example could be, for instance, a burglary – ingression into another’s property in order to obtain gain illegal possession of property. On a larger scale this could be the preying of a strong nation on a weaker, less resourceful one. This type is under-studied and while it may interact with the reactive aggression, does not rely on the same mechanisms

(McEllistrem, 2004; Umukoro et al., 2013; Weinshenker & Siegel, 2002). For starters, there is a lack of sympathetic arousal in the predatory aggressors. Additionally, feelings, if they play a role at all, lead to pleasure or satisfaction, as opposed to fear or anger involved in defensive acts. In this manner, it is possible to observe the most striking difference between the types of aggression: reactive defence is unmeasured, it lashes out (Weinshenker & Siegel, 2002). In economic terms, the

(9)

investment would be un-proportional to the risk. In contrast, predatory, calculated, aggression, as the latter term suggests, involves very thought-out allocation of resources – both manner and magnitude become significant, whereas in defence they are not.

Weinshenker and Siegel (2002) discuss the advantage, for the sake of research, of distinguishing the psychopaths. These individuals, who have difficulty relating to emotions, comprise a large part of prison populations and their motivations are more alike to predatory in nature, as opposed to impulsive violent offenders. Interestingly, instrumental aggression is also linked to dominance (Chichinadze et al., 2011). The brain mechanisms involved in planning and executive function, are the ones that would also be instrumental in predatory action. This coincides with data linking advantage of serotonin enhanced performance leading to domination among monkeys whereas depletion of the neurotransmitter, associated with impulsivity, does not have the same effect (Berman et al., 1997).

In the following sections we will follow the research paradigms implemented in impulsive behaviour and aggression research in economic decision-making. This will lead to the predator-prey paradigm, devised by de Dreu, Scholte, van Winden, and Ridderinkhof (2014), that establishes, using an asymmetrical game, the roles of proactive and reactive aggression.

1.3. Aggression networks

Aggressive behaviour is closely linked to social behaviour, both because of the context in which aggression takes place, and also due to the brain network involved in aggressive behaviour. This circuitry involves: the amygdala, and the rest of the limbic system; the periaqueductal gray (PAG); the hypothalamus; and sections of the prefrontal cortex (PFC) (Nelson & Trainor, 2007). These regions are also innervated by the raphe nuclei where serotonin is produced in the brain

(10)

drugs that interact with the serotonergic system are used to treat psychiatric cases that affect social behaviour. The social context is important to consider whenever research using animal models is interpreted into a human environment. An example to consider is the discovery of the “sham-rage” phenomenon in a cat upon the stimulation of the hypothalamus (McEllistrem, 2004) - this does not translate so neatly into a human equivalent situation.

In later parts of this paper, we will establish links between these networks and aggressive behaviour in various experimental settings.

Fig. 2. Networks of neurotransmitters in the brain. The serotonergic system is in green and the dopaminergic system is in red. The origins of molecule production are seen in the table and their interaction can be seen in both amygdala and the

(11)

2. Experimental paradigms

2.1. Aggressive behaviour in economic games

In economic settings, an aggressive investor puts more money into play and takes bigger risks (Afza & Nazir, 2007; Nazir & Afza, 2009). Investments of this nature seem more impulsive. However, aggressive behaviour could also translate into offensive behaviour; initiating purchases and trying to increase assets. This behaviour is a type of calculated aggression. This highlights the difficulty in separating the two types of aggression in economic decision-making.

In economic experiments, using different games to model decision-making, defection or punishment is often conceptualised as aggression (Crockett, 2009). For instance, in the prisoner’s

dilemma game, the peaceful solution would be cooperation where both sides gain together

maximally, however, the aggressive solution, where both defect, is the Nash Equilibrium. These two behaviour choices alternate depending on the type and frequency of interaction between the parties (Kassinove, Roth, Owens, & Fuller, 2002; Martin, Juvina, Lebiere, & Gonzalez, 2013). This applies similarly to the ultimatum game. In this game, one side decides how to share an initial sum and the other side can decide whether to reject the offer, thereby leaving both participants with nothing, or accept it, and share as agreed. If one is aggressive, he would reject an unfair (or perhaps even fair) offer. If the player is cooperative, he would accept the offer, as long as he gets something (Mehta & Beer, 2010). Rejecting an offer in the ultimatum game can be perceived as a type of punishment. This however, was challenged by Crockett et al. (2008), as we will further discuss later. Finally, it would be of the biggest interest to separate instrumental (predatory) aggression, of the form where one side wishes to break the status-quo and increase its winnings, from reactive (prey) aggression, where one side fends off the predator, at cost to itself, in order to maintain the current equilibrium. We can relate both of these behaviours to market, and social, behaviours such as a buy-out of a

(12)

company or going bankrupt so as not to be bought-off. We have found two experimental paradigms that separate these different aggressive behaviours and elucidate their underlying mechanisms (Crockett, Clark, and Robbins, 2009; and de Dreu, Scholte, van Winden, and Ridderinkhof, 2014).

2.2. The predator prey game

Ideally we would like a game that creates a distinction between the two groups of aggressors: goal-oriented, predatory aggressors, versus the prey-like, reactive (defensive) aggressors. This in addition to manipulations of subjects in terms of neurotransmitters and stress, as well as tasks involving impulse control and aggression questionnaires; all of which will be addressed in a following section.

We need to first question the motives for being aggressive as either predator, or prey. We will start with the prey. We can look at an individual, or a group, as belonging to a prey category when they wish to maintain a certain status-quo – wishing to keep a job, or ownership of a company or a piece of land. Therefore, for someone in that situation to turn to aggression, to attack or invest money and resources, implies that they are compelled by fear of someone altering the current state of things. In lab conditions there are various games to notice this reaction. Looking at a prisoners’ dilemma game as an example, this fear of defection of the other party could motivate one to defect as well. This is a reactive defence behaviour. In an alternative version of the prisoners’ dilemma, a withdrawal option is added to the game (Insko, Schopler, Hoyle, Dardis, & Graetz, 1990). This option can distinguish between two fear reactions: those who turn to impulsivity and defect, or those who are calculated, maintain their calm, and withdraw.

(13)

Table 1. Prisoner's dilemma standard payoff matrix (adapted from Wood et al., 2006).

Table 2. Prisoner's dilemma alternative payoff matrix (adapted from Insko et al., 1990).

In contrast, the predator, a greedy, goal-driven, individual or group, would wish to revisit the current situation. This party would like to earn more than it currently does, or own another’s land or property. It is important when comparing to animal models to consider only intraspecies preying, as it is in humans. For that purpose of increasing winnings the predator would invest, and risk, its resources. Therefore seemingly it is greed that motivates the predator in its choices when it decides to aggress against a docile counterpart. Similar to the case described for the prey, the predator would also defect in the prisoners’ dilemma game. Unlike the prey, the predator would do so in the hope that the other party cooperates, and thus the predator will maximise its personal gain.

The two psychological motives have been taken apart in different paradigms in order to study only fear, or only greed, and what decisions they would lead to. The chicken game uses a similar payoff scheme to the prisoners’ dilemma, only in this scenario a tie is a loss thus only defection would lead to a win and fear is dominant (Bornstein & Gilula, 2003). The assurance game, in contrast, encourages cooperation since that would lead to the biggest payoff to both parties and therefore is meant to exemplify only greed (Bornstein & Gilula, 2003).

A\B Coop Defect Coop 20\20 0\30 Defect 30\0 10\10

A\B Coop Withdraw Defect Coop 57\57 48\48 33\66 Withdraw 48\48 48\48 48\48 Defect 66\33 48\48 39\39

(14)

Table 3. The assurance game group payoff matrix with the independent variable being the number of investors in every group (adapted from Bornstein & Gilula, 2003).

Table 4. The chicken game group payoff matrix with the independent variable being the number of investors in every group (adapted from Bornstein & Gilula, 2003).

To contrast these games, a predator-prey game captured both motives in an asymmetrical game and thus allowed both types of aggressive behaviours to come into play (De Dreu et al., 2014). In this game, one party, the predator, wins by investing more than the prey, thereby taking all the prey’s leftover sum after investments. During the same investment, the other party, the prey, can only keep their leftover sum if they amassed a sufficient defence – they invested equal to or more than the predator (De Dreu et al., 2014). This design, in its intra-group version (De Dreu, Giffin, Krikeb, Prochazkova, & Columbus, 2015), is easy to equate to warfare between nations over territory or resources. One side, with predatory motives, is aggressing by taking action and investing in order to win what resources the prey party has. In reaction to this aggression, the other side defends itself, investing of its own resources to protect the remaining resources. As we will discuss later, this defence could be entirely impulsive and therefore likely to be exaggerated, or it could be also calculated, with an added long-term effect view. Accordingly, brain regions involved in greed and fear are different and so it has been observed that the two parties have different brain activations:

A\B 0 1 2 3 0 135\135 45\120 45\105 45\90 1 120\45 120\120 30\105 30\90 2 105\45 105\30 105\105 15\90 3 90\45 90\30 90\15 90\90 A\B 0 1 2 3 0 45\45 45\120 45\105 45\90 1 120\45 30\30 30\105 30\90 2 105\45 105\30 15\15 15\90 3 90\45 90\30 90\15 0\0

(15)

the amygdala and superior frontal gyrus (anatomically part of the PFC) activations in the predators were dampened by oxytocin whereas the prey’s amygdala was active regardless of condition (De Dreu et al., 2014). Following this, we would expect a different pattern of neurotransmitter activity to support these activations.

2.2.1. Greed and calculated aggression in the predator

In a predator-prey game, any predator that plays the game, instead of simply keeping their initial endowment, becomes an aggressor. It is this type of aggression that is under-studied and we would like to better highlight what sort of neural mechanism is potentially motivating it, as opposed to the impulsive aggression that is under the main study focus in relation to reduced levels of 5-HT (de Almeida et al., 2005; Takahashi, Quadros, de Almeida, & Miczek, 2011). In a cat, predatory behaviour has been initiated by stimulation of the perifornical lateral hypothalamus, the ventral part of the PAG, and the ventral-tegmental area (VTA) where dopamine is produced (Siegel & Victoroff, 2009).

The goal of this aggression, being something with a benefit past the act itself, is in contrast to the reactive defence which lacks foresight. As Siegel and Victoroff (2009) point out, this planning leads to the base assumption that the cerebral cortex is involved in the predatory aggression, whereas it is not necessary for prey defence.

2.2.2. PFC and goal-oriented behaviour

To go further in depth into the decision-making of the predator, we need to examine some brain regions that are associated with planning. First, the prefrontal cortex (PFC) is mainly associated with top-down cognitive function, goal-oriented behaviour, temporal control, and attention control (Agnoli & Carli, 2012; Narayanan, Land, Solder, Deisseroth, & DiLeone, 2012; Winstanley, Theobald, Dalley, Cardinal, & Robbins, 2006). It is therefore connected to impulse-control and conditions involving its control such as suicidal tendencies, attention deficit hyperactivity disorder, and

(16)

obsessive compulsive disorder (OCD). The PFC also has projections from the thalamus via the basal ganglia, where dopamine is a key neurotransmitter (Agnoli & Carli, 2012). This aspect has been studied using the 5 choice serial reaction time task (5-CSRT) with rats where it was shown that a bilateral lesion of the medial PFC or dorsomedial (including the superior frontal gyrus) lead to increased impulsivity. The aspect of motor inhibition, associated with action planning, was studied by Rubia et al. (2005) using a go/no-go task. In their paper, they found decreased activation in the right orbital and the inferior PFC following tryptophan depletion in humans using functional magnetic resonance imaging (fMRI). Their results also indicate no real change in mood nor any actual effect on the inhibitory control during the task (Rubia et al., 2005). Impulse-control is also related to loss-aversion, that has also been a proposed task of the PFC (Murphy et al., 2009).

Further properties of the PFC we would need to inspect are the type of neurons to be found and which neurotransmitters activate them. Pyramidal neurons, as well as GABAergic ones, in the medial PFC (mPFC) express 5-HT1A receptors. While serotonin endogenously inhibits mPFC pyramidal excitation, systemic 5-HT agonists administered to rats tend to excite the VTA projecting cells (Lladó-Pelfort, Santana, Ghisi, Artigas, & Celada, 2012). The VTA, in the midbrain, also projects back to the PFC using dopamine as input (Narayanan et al., 2012). Both D1 and D2 receptors are expressed in the PFC. Narayanan et al. (2012) showed that D1 disruption interferes with temporal control while its stimulation enhances efficiency in a fixed-interval timing task. These dopaminergic neurons could be innervated by serotonergic projections coming from the raphe nuclei to both VTA and PFC where in addition to 5-HT1A, also 5-HT2A receptors are present (Pehek, Nocjar, Roth, Byrd, & Mabrouk, 2006). Pehek et al. (2006) showed that 5-HT2A blocking in the PFC results in dopamine blocking thus indeed linking the dopaminergic to the serotonergic system. Both these neurotransmitter systems will come into further focus later on and it is important that it first understood that they play a significant role in the PFC’s activity.

(17)

2.2.3. Fear and reactive aggression in the prey

A prey in the game, similar to a mother defending her offspring, would resort to an affective defence behaviour (Siegel & Victoroff, 2009). This defence behaviour is observed despite the anonymity of the players (De Dreu et al., 2015). This fear of danger leads to a reaction whose purpose is eliminating the threat. It is for this reason that this action, often aggressive, is impulsive; it does not require any calculation beyond the threat and therefore likely to be out of proportion for the assurance of its success. Additionally, strong impulses can induce fear (Apter et al., 1990), showing us that also in this mechanism there may be a feedback loop. The fear reaction is strongly based in the limbic system (Post et al., 1998; Yoon, Fitzgerald, Angstadt, McCarron, & Phan, 2007). Moreover, research into defensive rage found that by direct stimulation of the medial hypothalamus or the dorsolateral region of the PAG, this defensive reaction could be initiated (Nelson & Trainor, 2007; Siegel & Victoroff, 2009).

In terms of neurotransmitters, impulsive behaviour is mainly linked to serotonergic hypoactivity (de Almeida et al., 2005; McEllistrem, 2004; Siegel & Victoroff, 2009; Takahashi et al., 2011;

Weinshenker & Siegel, 2002) as well as irregularities in the dopaminergic system (Rogers, 2011; E. S. Smith, Geissler, Schallert, & Lee, 2013). It is important to realise that the reactive aggression, largely based on sympathetic activation, could be accompanied by calculated aggression, which would mitigate the impulsive reaction and allow planning to take place beyond the immediate effect of removing the threat (McEllistrem, 2004; Weinshenker & Siegel, 2002). This could be compared to a territorial war where the defender has realised that conquest of some of the attacker’s territory would be beneficial.

(18)

The limbic system in general, and the amygdala specifically as part of it, include the brain structures classically associated with emotions (Post et al., 1998; Yoon et al., 2007). This system is also believed to be a part that belongs to the early stages of brain evolution. Of highest relevance to this paper, it is associated with fear and anxiety (De Dreu, 2012; Saha, Gamboa-Esteves, & Batten, 2010). Irregular activation of the amygdala is associated with many anxiety disorders, such as PTSD (Nardo et al., 2010; Pagani et al., 2012) and different phobias (see: Caseras et al., 2010; Klumpp, Angstadt, Nathan, & Phan, 2010; Yoon et al., 2007). Crucially for this paper is its relation to decisions instructed by fear as the dominant emotion. As de Dreu and colleagues (2014) found in their study using the predator-prey game, the amygdala was more strongly activated among the prey,

contrasted with the predator, when making their investment decisions. This, in addition to the faster reaction time, indicates the decision was more instinctive.

At this point, we need to explore the projections of the amygdala in order to understand what the roles of dopamine and serotonin, as neurotransmitters in that region, may be. First, the amygdala, as part of the limbic system, projects to the hypothalamus, and the PAG (Siegel & Victoroff, 2009). This links the amygdala to the aggressions that we know could be initiated by the hypothalamus.

Serotonin, produced in the raphe nuclei, influences the amygdala through many projections from the dorsal raphe nuclei. Past studies also showed that 5-HT1A and 5-HT1B receptors are associated with anxiety and depression (Saha et al., 2010). Saha et al. (2010) further elaborated on this by showing the large number of neurons marked for 5-HT1B receptors and relatively low number with 5-HT1A receptors in the amygdala. This will become more significant later when we discuss the receptors’ separate roles.

(19)

In addition to serotonergic function, also dopamine is instrumental in fear response. There are many efferent fibres to the amygdala from the VTA (Rezayof, Hosseini, & Zarrindast, 2009) as well as bidirectional interaction with the substantia-nigra (E. S. Smith et al., 2013). These two areas are both part of the dopamine-dominant reward-system. The effects of dopamine are quite substantial on the amygdala. For instance, dopamine depletion prevents memory formation related to fear through amygdala function, whereas specific restoration of dopamine to the VTA –amygdala pathway reverses the effect (Li, Dabrowska, Hazra, & Rainnie, 2011). Specifically D1, and not D2, receptors are crucial for this process. D1 receptor antagonist SCH23390 has been shown to eliminate learning entirely by blocking the long-term potentiation (LTP) process from occurring in

glutamatergic neurons (Li et al., 2011). Also the role of dopamine will be further explored later in this paper.

2.3. Aggressive behaviour in current research on neurotransmitters

While this paper is focused on the decisions that derive from both reactive, as well as predatory aggression, there is little research that focuses on the neurotransmitters that are involved in both of these decisions in humans. In several studies, based mostly on the ultimatum game, the type of aggression that matches punishment was correlated to serotonin levels using a tryptophan depletion experiment (Crockett et al., 2010, 2009, 2008; Crockett, 2009). In other research, aggressive

behaviour is measured in other ways. The lifetime history of aggression scale is used by health workers to establish an individual’s trait aggression based on interviews and clinical history (Nelson & Trainor, 2007). Better suited to this paper is the aggression questionnaire the was developed by Vitiello, Behar, Hunt, Stoff, and Ricciuti (1990) that separates the questions to predatory and affective classifications. In their unique study they found that almost every subject had a mixture of the two characteristics. The more predatory individuals tended to have a higher IQ score, while the

(20)

reactive children had a higher prevalence of schizophrenia. Malone et al. (1998) have used this same questionnaire, as well as the Overt Aggression Scale and the Global Clinical Judgements Scale and had similar findings, yet not with the same significance.

Alternatively, the choice of violent offenders as subjects eliminates the need for objectivity amongst the subjects (Linnoila et al., 1983). This research also led to the idea that it is not aggression that is indicated by lower serotonin levels but impulsivity. This has been the dominant dogma for a long period and has been the focus of many studies (Anholt & Mackay, 2012; Badawy, 2003; Berman et al., 1997; Booij et al., 2010; Brown et al., 1982, 1979; Crockett et al., 2010, 2009, 2008; Daw et al., 2002; de Boer & Koolhaas, 2005; Linnoila et al., 1983; Perez, 2012; Wetzler et al., 1991). Another major data source for the 5-HT impulsivity theory comes from suicide cases; both from autopsies, and from CSF of failed attempts. Low levels of serotonin, or its metabolite, 5-HIAA, in the latter case, have been discovered in past studies (Dalley & Roiser, 2012). People who attempted suicide also testify through questionnaires on possessing more impulsive tendencies.

In animal models there are different paradigms. Research focusing on the hypothalamus and the “sham rage” phenomenon in cats was pioneering the way for future research on violence and aggression (McEllistrem, 2004; Weinshenker & Siegel, 2002). Studies involved direct stimulation of brain areas believed to be involved in aggression, introduction of a trespasser, or inter-male violence in competition for dominance or a female. In these studies they either manipulate or measure levels of different neurotransmitters. These are mostly measures of violence, as well as being in lab settings. This conditions make it difficult to translate to human environment (Nelson & Trainor, 2007).

Also studied, albeit with a little less focus, are the increased levels of dopamine in aggression (Boureau & Dayan, 2011; Daw et al., 2002; de Almeida et al., 2005; Seo, Patrick, & Kennealy, 2008;

(21)

Vukhac, Sankoorikal, & Wang, 2001) and norepinephrine (NE, also called noradrenaline) (Anholt & Mackay, 2012; Chichinadze et al., 2011; Higley et al., 1996; Perez, 2012), and the effects of gamma-Aminobutyric acid (GABA) (Anholt & Mackay, 2012; de Almeida et al., 2005; Gowin et al., 2010; Seo et al., 2008). However, as previously written, this paper will restrict itself to serotonin and dopamine.

2.4. Impulsive behaviour in experiments

In a large part of available research on aggression, the focus is on criminal and clinical cases, and not on aggressive choices. These cases come under the impulsive sort of aggression in a large part of the cases (exception of psychopaths who show no emotion and therefore seem to be more

predatory (Perez, 2012)). For instance, Linnoila, Virkkunen, and Scheinin (1983) found a link between serotonin and impulsive aggression amongst violent criminals. They went on to suggest that what serotonin hypoactivity indicates is impulsive behaviour, and not aggression in general.

Dalley et al. (2011) define impulsive behaviour as “the tendency to act prematurely without foresight”. They also clearly distinguish between impulsive and compulsive behaviour, despite the oftentimes seen confusion of the terms. Yet since their definition for compulsive behaviour focuses on the undesired consequences of the actions (p. 680), we could bundle both behaviours as having unforeseen (negative) results. Therefore, we would follow from studies that examine impulsivity to conclusions concerning rash decisions that lead to negative results.

A majority of the clinical data on human impulsive behaviour comes from suicide cases, a case of extreme impulsivity (Dalley & Roiser, 2012). This link is also explored in many other studies (Crockett et al., 2010; Dalley, Everitt, & Robbins, 2011; Higley et al., 1996; Malone et al., 1998; Mehta & Beer, 2010; Umukoro et al., 2013). The problem with many such studies is that the focus is on clinical or criminal cases which pre-assigns the subjects and places them under a certain category – impulsive individuals. A better scenario is using healthy individuals, or animals, to test impulsivity under

(22)

certain conditions. The conditions could be induced stress, hormonal manipulations, or neurotransmitter manipulation, as some examples. The manner in which the impulsivity, or aggression, could be tested is using questionnaires, such as the lifetime history of aggression scale, or decisions made in tests such as the 5-choice serial reaction time task, delay discounting task, stop-signal task, or go/no-go task. In questionnaires, individuals must testify about their own behaviour. In the tasks, impulsivity is decided based on the ability to learn where waiting can lead to reward. Using these methods it is possible to observe how healthy subjects react to manipulations and therefore allow for determination of causation. We will review these tasks later as they will be used to draw some conclusions concerning the roles of serotonin and dopamine in impulsive behaviour among healthy subjects.

2.5. Experimental paradigms of impulsivity

What we learn from past experiments on aggression is that it is necessary to extract the cognitive mechanism leading to it, which is, to a large extent impulsive behaviour (Dalley, Mar, Economidou, & Robbins, 2008; Homberg, 2012; Kiser, Steemers, Branchi, & Homberg, 2012). It is a type of behaviour that is more conspicuous in clinical cases related to gambling, alcoholism, drug abuse, depression, and schizophrenia (Badawy, 2003; Booij et al., 2010; Dalley et al., 2011; Daw et al., 2002; Doya, 2002; Linnoila et al., 1983; Okai, Samuel, Askey-Jones, David, & Brown, 2011; Rogers, 2011; Scholes et al., 2007; Umukoro et al., 2013) but while it does enlighten us as to some

particularities of the system, we are not interested in the pathological case, but in the healthy. It is healthy people who mostly interact in predator-prey situations.

Impulsivity could be examined using questionnaires or different tasks. Questionnaires include: the Barratt impulsiveness scale, the urgency, premeditation, perseverance, and sensation seeking impulsive behaviour scale, Behaviour Scale, the Impulsiveness Venturesomeness and Empathy

(23)

Questionnaire, and the Lifetime History of Impulsive Behaviours (Dalley et al., 2011; Dalley & Roiser, 2012). There are several tasks that are often used in the lab, some only fit for animals, and some for both animals and humans, that are used to examine impulsive behaviour. Even though action suppression is mostly the manner of study, and impulsive action does not use the exact same mechanism as impulsive choice, it is a useful study tool since the mechanisms interact along with the serotonergic and dopaminergic systems (Dalley et al., 2011). We will now briefly introduce some of the more common methods.

2.5.1. 5-CSRT

The 5-choice serial reaction time task involves the animal initiating the trial and then waiting for one of five cues to light up. A press before the light has turned on is deemed impulsive and results a 5 second timeout. A press on the button under the correct light, once it has turned on, results in a reward food-item. An incorrect press, or no choice at all, are incorrect choices and result in a timeout as well. The goal of this task is to learn to suppress a response in order to gain a reward (Dalley et al., 2011; Dalley & Roiser, 2012).

2.5.2. Go/no-go

A Go/no-go, or stop-signal, task can have a human or an animal version to it; in the human version the reward is often simply a smiley face while in the animal model version the reward is a food item. In this task the subject must press the button on ‘go’ trials upon the presentation of a cue. In ‘no-go’ trials the subject must avoid pressing the button when the cue is given (Dalley & Roiser, 2012). This task combines the effects of learning – subjects learn during the task itself how to respond to the cue, and relearning; thereby fighting the impulse to act on what has been learnt previously for the ‘go’ trials in order to succeed in the ‘no-go’ trials.

(24)

2.5.3. SSRT

The stop signal reaction time task models inhibition of motor control – the ability to stop an action that has already been initiated. It is similar to the go/no-go task only here all trials are ‘go’ trials until a ‘stop’ signal is given at which point subjects must stop responding (Dalley et al., 2011). Crucially, reaction times for the responses are taken so it can be compared between trials after a ‘stop’ signal has been given and thus observe readjustment. Despite the similarities in the tasks, different manipulations affect the results of SSRT and go/no-go differently (Dalley et al., 2011).

2.5.4. Reversal learning

In a reversal learning task subjects are conditioned to respond to a specific conditioned stimulus and ignore a second stimulus. Once this conditioning is established, the subjects are asked to switch the reactions between the stimuli – act on the second and ignore the first (Homberg, 2012). This task can be practised both with humans, as well as with animals, even though, according to Homberg (2012), it is not entirely clear whether the two employ the same brain mechanism for the task. This is another paradigm of learning and suppression of action.

2.5.5. Delayed reward

A delayed reward, or delayed discounting task, constructs a situation where the subjects face a choice between a small reward now and a larger reward later. The manipulations could be either on the size of the reward or the length of the delay (Dalley & Roiser, 2012). In animals, the delay is measured in seconds and the reward is food. In humans, the design involves a hypothetical choice of monetary reward that spans between minutes and years. This paradigm has a clear focus on

impulsive choice since the strictly logical choice is always the larger reward.

(25)

Serotonin, or 5-HT, is a well-studied neurotransmitter that has been associated with mood, survival, social behaviour, sleep, brain plasticity, impulsivity, and mental illness (Benningfield & Cowan, 2013; Blier & El Mansari, 2013; Dalley et al., 2011, 2008; Dalley & Roiser, 2012; Gellynck et al., 2013; Homberg, 2012; Jouvet, 1999; Kiser et al., 2012; Lovinger, 2010; Martinowich & Lu, 2008; Navailles & De Deurwaerdère, 2011; Roberts, 2011; Rogers, 2011; Scholes et al., 2007; Takahashi et al., 2011). Our focus in this paper is on impulsive (choice) and aggressive behaviour as they are linked to altered serotonin levels. Brown et al. (1979) are the first to find a negative correlation between serotonin, based on 5-HIAA concentrations in CSF, and aggressive behaviour in humans. In another study, both aggression and low levels of serotonin are correlated with suicide attempts (Brown et al., 1982). In another lane of research, Linnoila et al. (1983) suggest, based on their results that separate impulsive offenders and those who were conscious and calculated in their actions, that low levels of serotonin are an indication of impulsive behaviour. A problem with this conclusion is that they decided according to the offence itself who was premeditated and who was not (Berman et al., 1997). It seems however, that despite the problems this study may present, studies that examined the influence of alcohol on aggression, do find that certain individuals with a tendency for violence will be more prone to it following the ingestion of alcohol and this could stem from the alcohol’s effect on brain serotonin levels – i.e. it depletes them (Badawy, 2003). These findings correlate with the idea that the correct food supplementation could eliminate, or reduce, aggressive behaviour (Badawy, 2003).

3.1.1. Molecule

5-HT is produced in the raphe nuclei that sit in the mesencephalon (brainstem) (Martinowich & Lu, 2008). It is a simple monoamine molecule, much like many other neurotransmitters such as: dopamine, melatonin, epinephrine, and norepinephrine. 5-HT is derived from L-tryptophan, one of the basic amino acids (Folk & Long, 1988; Kiser et al., 2012; McEllistrem, 2004). 5-HT has a major

(26)

central metabolite, 5-hydroxyindoleactic acid (5-HIAA), and the most reliable method to examine the presence of serotonin in the human brain is looking for it in a sample of cerebrospinal fluid (CSF) (Brown et al., 1979).

3.1.2. Different receptors in different brain regions

Serotonin seems to have appeared early in our evolution and thus, being an important molecule, it presents the highest number of receptors in the brain. These divide into ionotropic, which allow ion transfer upon ligand binding, and metabotropic, which start a signalling chain within the cells when the ligand binds. Accordingly they have different functions (Gellynck et al., 2013; Kiser et al., 2012; Martinowich & Lu, 2008).

Serotonin is infamous for its effects following the ingestion of 3,4-

methylenedioxymethamphetamine (MDMA), which inhibits the reuptake of the neurotransmitter back into the presynaptic neuron from the synaptic cleft by competing with it, and other

monoamines, for their reuptake transporters (SERT – serotonin reuptake transporter, and DAT – dopamine reuptake transporter) (Benningfield & Cowan, 2013; Rubia et al., 2005). It also inserts itself into the presynaptic neuron and induces release of more serotonin into the synaptic cleft. This leads to social disinhibition and therefore MDMA is a very popular for recreational use. These effects are similar to those that form the basis of antidepressant drugs, that also have the same effect of saturating the synapses with serotonin, either through blocking reuptake or through inducing increased production (Blier & El Mansari, 2013; Martinowich & Lu, 2008). The effects on social behaviour are the important effects we need to keep in mind in order to understand the role of serotonin.

Selective serotonin reuptake inhibitors (SSRIs) are the current antidepressants of choice among clinicians (Blier & El Mansari, 2013; Crockett et al., 2009; Dalley & Roiser, 2012; Kiser et al., 2012;

(27)

Martinowich & Lu, 2008). SSRI drugs are agonists of the 5-HT1 receptors family (Takahashi et al., 2011). The higher concentrations of 5-HT are believed to have beneficial effects on social behaviour generally even though that in the case of major depression, a common mental disorder, the low levels of serotonin are not necessarily the cause (Blier & El Mansari, 2013). However, even with only partial support linking the low levels of serotonin to aggression and major depression, increasing the levels of 5-HT is often beneficial.

To counteract the effects of aggression, similar to antidepressants, there are drugs that activate only a subset of the serotonin receptors. There are 16 different genes encoding 5-HT receptors (Kiser et al., 2012) and various drugs work with different specificity on these receptors. These drugs are tested mostly in experiments using animal models to examine their effects. De Boer and Koolhaas (2005) examine some compounds that act as agonists and antagonists to 5-HT1A and 5-HT1B family receptors. The former acts as an autoreceptor on 5-HT neurons, as well as a postsynaptic receptor on pyramidal and GABAergic (inter)neurons (Lladó-Pelfort et al., 2012), that reacts differently to different drugs according to its location. For example, S-15535, a benzodioxopiperazine, acts as an agonist to 5-HT1A when it is on the presynaptic neuron and as an antagonist when it is on the postsynaptic one (de Boer & Koolhaas, 2005). It has a similar effect as other pharmacological agents in reducing aggressive behaviour, for instance: repinotan, 8-OHDPAT, ipsapirone, and some others. Unlike these other compounds, S-15535 does not affect other aspects of non-aggressive motor behaviour (de Boer & Koolhaas, 2005). These other effects could occur due to several reasons: a higher dose is needed, or the non-selective activation of other receptors. This sort of study highlights the difficulty in assessing the specific role of a specific receptor. Despite this, there are many other studies doing so and we will look into some more of them.

(28)

Selective antagonists of 5-HT1A and 5-HT1B indeed demonstrate blocking of the attenuating aggression effects if given alongside S-15525, while having no effect by themselves. Specifically related to the amygdala function, the effects of 5-HT1A receptor deficit seems to lead to anxiety whereas its over-expression leads to aggressive behaviour (Saha et al., 2010). This is contrasted with 5-HT1B receptors whose over-expression restricts aggressive behaviour while their rat knockout models show increased aggression (Saha et al., 2010). To contrast the actions of 5-HT1A from 5-HT2 and 5-HT3, it was shown by Stein, Davidowa, and Albrecht (2000) that a 5-HT1A agonist decreases the firing rate of neurons in the amygdala while 5-HT2 and 5-HT3 agonists resulted an increase in firing rate. Moreover, research on the different 5-HT2 receptors leads to different results based on the specific location of action, as opposed to global infusion of a drug, and that is in addition to the effect of the drug being agonist or antagonist (Homberg, 2012). In that review, Homberg (2012) states that 5-HT6 “univocally contribute[s]” to impulsivity (p. 230) however it receives little mention elsewhere. In addition to the amygdala, also the presence of 5-HT1A receptors in the PFC has been addressed. Additionally, there are functional links between the OFC and the basolateral amygdala that influence decision making (Winstanley, Theobald, Cardinal, & Robbins, 2004).

Interestingly, there is no hypersensitivity of 5-HT receptors observed in aggressive individuals (Wetzler et al., 1991). However, aggressive individuals do have different ratios of 5-HT receptors expressed in the brain (de Almeida et al., 2005).

3.1.3. Tryptophan depletion

One manipulation that can be done in studies specifically oriented at the effects of 5-HT is using an acute tryptophan depletion paradigm. In the studies using this paradigm, participants drink a liquid that contains a concoction of different concentrations of amino acids besides tryptophan (control group drinks the same concoction that includes tryptophan as well). Based on previous

(29)

studies analysing rat brain tissue, this leads to a sharp reduction in serotonin production (Crockett et al., 2009). Five hours after the intake of said drink, participants normally are controlled for changes in mood, and then are given the task and/or commence the game (Crockett et al., 2009; Wood, Rilling, Sanfey, Bhagwagar, & Rogers, 2006). This form of research allows for a contrast between normal and decreased systemic concentrations of serotonin to exemplify its behavioural effects.

3.1.4. Tryptophan supplementation

Similar to depletions studies, there are also tryptophan supplementation studies performed. Tryptophan could be supplemented in an acute study (Scarnà, McTavish, Cowen, Goodwin, & Rogers, 2005) or over a duration of a couple of weeks (Murphy et al., 2009). In either case, the dietary supplement consists of a high level or tryptophan that should keep the enzyme tryptophan hydroxylase, the rate-limiting enzyme in the 5-HT production, close to saturation (Murphy et al., 2009). The increased levels of tryptophan should in turn lead to increased 5-HT levels that alter social behaviour (Murphy et al., 2009), similar to effects of other drugs that interact with the serotonergic system that have been previously touched upon. Other effects on mood, memory, or attention are then controlled for before the main task is initiated.

3.1.5. Serotonin knockouts

Another manipulation that can be performed in the lab using animal models is knockout a specific gene using viral agents and breeding. In this paper our highlight is the knockout of specific 5-HT receptor or enzyme genes. In such knockout models the role of a specific receptor in the entire mechanism can be assumed from the changes observed in behaviour in the knockout animals, as opposed to the wild-type healthy controls. Examples include knockout of the 5-HT1A or 5-HT1B receptor (Saha et al., 2010), 5-HT transporter (5-HTT) (Kiser et al., 2012), or monoamine oxidase A

(30)

(MAOA) which is a main serotonin catabolising enzyme (McDermott, Tingley, Cowden, Frazzetto, & Johnson, 2009).

3.1.6. Specific agonists/antagonists and neurotoxins

Other paradigms that involve mature animals are the use of neurotoxins or specific agonists and antagonists. For example, 5,7-dihydroxytryptamine (5,7-DHT) injected directly to the brain removes all serotonin (Dalley & Roiser, 2012).

Some examples of agonists and antagonists are: 8-OH-DPAT is a specific 5-HT1A agonist, (±)-1-(2,5-dimethoxy-4-iodo- phenyl)-2-aminopropan (DOI) is a 5-HT2A/2C agonist while SER082 is an antagonist for these receptors, SB242084 is a 5-HT2C antagonist and M100907 is an antagonist specific for 5-HT2A (Dalley & Roiser, 2012). These are just some examples of molecules used to examine the role of specific receptors or the interactions between them. Some of these molecules can be injected also very accurately to specific regions, while others are given globally.

3.2. Serotonin and impulsivity

Since we discussed serotonin as a key player in impulsive behaviour, we will now look at some evidence from the lab. Dalley and Roiser (2012) address some of the past results using various agonists and antagonists for different 5-HT receptors. For instance: a systematic administration of DOI increases impulsivity in a delay discounting task relying on 5-HT2A mechanism. A similar result was achieved in the 5-CSRTT using the 5-HT2C antagonist SB242084 (Dalley & Roiser, 2012). SER082, a 5-HT2A/2C antagonist, did not affect 5-CSRTT behaviour but did impact the delay discounting task by reducing impulsive response. 8-OH-DPAT, a 5-HT1A agonist had positive results for the choice reaction time task however it negatively affected delay discounting. 5-HT reuptake inhibitors yielded positive results for both tasks (Dalley & Roiser, 2012). M100907, a 5-HT2A antagonist injected to the

(31)

PFC, and the 5-HT1A receptor agonist 8-OHDPAT, both result in blocking impulsivity in the 5-CSRTT (Dalley et al., 2011).

These results exemplify the role serotonin has in different types of impulsive behaviour and would suggest therefore that it is a good candidate to maintain when examining human impulsive decision-making.

3.3. Link between serotonin research to predator and prey behaviours

Apter et al. (1990), in their paper, discuss separating impulsivity, as a character trait, from the different psychological diagnoses available, since it seems that this is what 5-HIAA measurements indicate. Berman et al. (1997) suggest in their discussion that a distinction between impulsive and non-impulsive aggression be made in further research and that potential mediating factors, such as environmental factors, be examined. This suits our bimodal classification system.

In brain circuitry terms, this goes in line with the fact that the raphe nuclei project to the orbitofrontal cortex (OFC), and also project back to it, thus creating a feedback loop with the serotonergic system (Roberts, 2011). The OFC is associated with depression and obsessive compulsive disorder but additionally, along with the PFC, it is associated with top-down control, planning, and decision-making (Dalley et al., 2011). This decision-making could be impulsive and more leaning towards risk, while it could be more risk-aversive. It has been found that tryptophan supplemented diet reduced risk-taking in the form of the reflection effect as well as influencing loss-aversion (Murphy et al., 2009, in Rogers 2011). Additionally, carriers of the ss (short) allele of the 5-HTTPLR were more sensitive to the framing effect (Tversky & Kahneman, 1981), and could be supposed to have been more emotional in their rationalisation since a stronger interaction between the amygdala and the PFC was observed (Roiser et al., 2009, in Rogers 2011).

(32)

A link between decision-making and impulsivity could be made based on the study that found that only delayed bad outcomes were retarded by tryptophan depletion (Blair et al., 2008, and Tanaka et al., 2009, in Rogers 2011). This delayed learning would invite further impulsive socially inappropriate behaviour. This is also discussed by Homberg (2012), detailing the discounting of future reward, as well as past punishment, following tryptophan depletion. The social implications of low levels of serotonin are also evident through the iterated prisoner’s dilemma game where tryptophan depleted subjects were less cooperative, even while playing against the quite forgiving tit-for-tat strategy (Wood et al., 2006).

Also according to Kiser and colleagues (2012), impulsive aggression correlates with low levels of serotonin and calculated aggression correlates with higher levels. This works well with the idea that higher 5-HT levels are expected in socially capable individuals and therefore they are able to

perceive hostile behaviour towards them and react accordingly. This would apply in economic terms as well when a person, or group, should feel threatened it would enact decisions to protect itself, however, in a measured manner.

An example for the effects of increased serotonin levels was found in monkeys. In some studies of monkeys it was shown that tryptophan supplementation led to both better social interaction, as well as strategic aggression the led to a rise in the hierarchy (Kiser et al., 2012).

As we have described, impulsive decisions, besides violence (a poor choice of behaviour in current society), lead to choices that have no or little foresight. Let as move from there to economic decisions. Thus, we would expect people with lower concentrations of 5-HT to act impulsively not only in the standard prisoners’ dilemma game, but also in the iterated version (Wood et al., 2006; Yi, Johnson, & Bickel, 2005). These people will choose to defect also when repeated interaction plays a role in the eventual outcome of the game and therefore we would term them as being selfish, or

(33)

aggressive players. The outcomes in Yi et al. (2005) show that delayed discounting correlated with (the successful) tit-for-tat strategy, and not with random play, in the iterated game. This behaviour indicates an inability to see the mutual benefits of cooperation as the model of reciprocal altruism of the iterated prisoners’ dilemma suggests, especially under the tit-for-tat strategy. Interestingly, in Wood and colleagues’ (2006) work, trait aggression is not a consequence of tryptophan depletion. Furthermore, the results were not replicated in the second day of the study; an effect related to the role serotonin plays in social learning – possibly after learning the game on the first day, being depleted on the second day did not induce as strong an of an effect.

Fig. 3. Ultimatum game decision-making (based on results of Corckett et al., 2008, 2010).

An additional point was made by Crockett et al. (2008) in their study of altruistic punishment as a tool for social behaviour. Altruistic punishment, being a punishment where one has to make an effort, or invest of their own funds, in order to inflict (financial) pain on others, has shown success in experimental conditions at maintaining a social norm and is believed to serve a reciprocal social system (Fehr & Rockenbach, 2004). According to other studies, that link serotonergic activity to it,

(34)

altruistic punishment negatively correlates to serotonin levels following tryptophan depletion, suggesting that it is impulsive behaviour rather than calculated (Crockett et al., 2010, 2008). Crockett and colleagues (2010, 2008), used the ultimatum game as a tool for measuring altruistic punishment. In this game, participants could choose whether to accept offers of splitting a sum or reject them. By accepting, both sides split the sum as agreed, while rejection means that neither side receives anything, thus in fact being an altruistic punishment. While it is commonly believed that rejection of unfair offers is perceived as a calculated “teaching” strategy on behalf of the responder, according to their findings, Crockett et al. (2010, 2008) conclude that it is in fact an impulsive choice. This is based on the significant increase in rejection of unfair offers following tryptophan depletion. They further support this by showing worse choices made in a delay discounting task (Crockett et al., 2010). This task demands people to choose between a small reward now or a bigger reward later (with the size of reward and length of time delay as variables) and is amongst the common tests used to measure impulsivity since it makes little sense to opt for the immediate reward. In Crockett and colleagues’ (2010) study, tryptophan depletion also resulted in increase in number of instant rewards chosen. Additionally, in their discussion, Crockett et al. (2010) connect the dorsal striatum, part of the reward system and mostly dopamine dependant, to the enjoyment of punishment (p.860).

Protective, or reactive, aggression, is most likely related to fear and therefore to the amygdala and the rest of the limbic system. This is based on de Dreu and colleagues’ (2014) work where they found that the subjects acting as prey made a quicker choice in their investment. This decision was accompanied by activation of the amygdala and therefore we can call this decision a more instinctive one. Since the amygdala carries serotonin projections as well (Homberg, 2012; Kiser et al., 2012; Saha et al., 2010), excitement in the amygdala is likely to involve, amongst other neurotransmitters, also serotonin action. If we can observe increased activation then we can also assume a higher firing rate of the 5-HT2 as well as 5-HT3 (Stein et al., 2000). This can lead also 5-HT1B activation but that

(35)

should reduce aggression (Saha et al., 2010). Therefore, our expectation would be 5-HT1A activation, that we would expect to correlate with anxiety, but that should decrease activity levels across the amygdala (Stein et al., 2000). Once again we are confronted with the intricacy of the system.

Siegel and Victoroff (2009) discuss the suppressing effect serotonin has on reactive aggression through 5-HT1A activation in the medial hypothalamus or in the PAG. However, they indicate that through the 5-HT2 receptor activation in the PAG, limbic system, and the hypothalamus, reactive aggression is initiated. This links fear and aggression through the serotonergic system.

Increased activity in the superior frontal gyrus is associated with impulse control as well as calculated behaviour (De Dreu et al., 2014). Decreased levels of activity follow tryptophan depletion in healthy humans (Rubia et al., 2005). These two findings interact neatly to form a picture of calculated decision-making in the PFC.

From a conscious logical perspective it is interesting to wonder why would the predator aggressor attempt an offence to begin with since generally speaking, the defender increases its levels of aggression often to the extent that the predator loses the conflict and therefore from the logical standpoint, the foresight should be that initiating an offensive is the wrong move. Rogers (2011) indicates that there is little research looking into risky choices that lead to bigger rewards and serotonin but part of it is related to delay discounting being disturbed by tryptophan depletion that has been previously discussed. Interestingly, probabilistic discounting is not diminished by the depletion of tryptophan. These are both examples of behaviour that could be paralleled to predator behaviour. To further support this effect of 5-HT deficiency, an experiment with macaques

demonstrated that switching from risky choice to safe choice demands a lower variance for the pay-offs following tryptophan depletion, further suggesting that calculated aggression is not as

(36)

All the evidence from the research on serotonin suggests that it is a necessary component of intelligent decision making. We can hypothesise that a potential prey will become one when they are with low levels of serotonin and one that is already a prey will invest excessively as it is the impulsive choice. We can further hypothesise that a predator with low serotonin levels does not exist.

4. Dopamine

4.1.1. Molecule

Dopamine, another monoamine neurotransmitter, is often studied for its effects on fine procedural memory, working memory, reward circuit, addiction, and motor skills in relation to Parkinson’s disease (Doya, 2002; Lovinger, 2010; Navailles & De Deurwaerdère, 2011; Okai et al., 2011; Segura-Aguilar et al., 2014; Wise, 2009). It is also the major focus of research into the reward circuitry in the brain and how it affects both addiction and reinforcement learning (Daw et al., 2002; Glimcher, 2011; Volkow, Wang, Fowler, Tomasi, & Telang, 2011). It is produced from tyrosine in dopaminergic neurons of the VTA and the substantia-nigra (SN) that project to the striatum mainly (Brichta, Greengard, & Flajolet, 2013; Dalley & Roiser, 2012; Ikemoto, 2010; Scholes et al., 2007; Wise, 2009). The former is related to the reward mesolimbic system and the latter the nigrostriatal projections that are key for motor control (Wise, 2009).

4.1.2. Different receptors in different brain regions

Dopaminergic input, besides the striatum, project from the VTA to the PFC (Doya, 2002). Its effect is mediated through five different metabotropic G protein-coupled receptors (Beaulieu & Gainetdinov, 2011) that are differentially functional as well as differentially spread out in the brain. The D1 and D2 receptors are significantly present in the amygdala whereas the D5 receptors are

(37)

present in a high concentration in the pyramidal neurons of the PFC (Beaulieu & Gainetdinov, 2011). They are all present in other regions, in addition to the other two receptors; however, it is less significant for this paper.

Some examples for the rewarding function of dopamine will be given here since they interaction of this system with social behaviour is obvious, even if it is not directly related to aggression or impulsivity. Dopamine is rewarding when there is a spike in its activity and certain dopamine receptors, such as D4, have different polymorphisms that allow levels of sensitivity and stronger or weaker good feeling that is very important for the sort of social behaviour humans present (Bachner-Melman et al., 2005). The prediction errors for reward, or lack thereof, have been correlated to spikes in dopamine activity in the midbrain following unexpected reward, or alternatively, a dip following a lack of an expected reward (Schultz, 2004, 2007, in Rogers, 2011).

Other dopamine receptors are studied in other contexts. Of relevance for our paper is the interventions performed on the dopaminergic systems in order to reduce aggression. For instance, haloperidol has been used for many years as a treatment for psychotic patients. Haloperidol acts as an antagonistic agent on the dopamine D2 receptors (de Almeida et al., 2005). In other research it has been found that D2 receptor antagonists also hinder people from detecting angry faces (Seo et al., 2008). D2 receptors also have two isoforms that have been studied in mouse models: D2L, long form, and D2S, the short one, and in the mouse brain the long form is more abundant, with an overall D2 density reduction with old age (Vukhac et al., 2001). It has been observed that older mice are significantly less aggressive, moreover, mice lacking the long isoform, D2L, are significantly less aggressive, without losing overall D2 density (Vukhac et al., 2001). D2 receptor’s role is deeper than aggression though since it could be characterised as working to regulate compulsions and impulsive

(38)

behaviour as has been shown with the effects on gambling behaviour with people who take D2 affecting drugs (Rogers, 2011).

4.1.3. Parkinson’s and L-dopa

Since dopamine is produced in the substantia-nigra that is the main target of Parkinson’s disease, a large amount of research is focused in that area. The loss of dopamine producing neurons leads to loss of fine motor control as well as impulsive behaviour such as compulsive gambling and binge eating (Brichta et al., 2013; Dalley & Roiser, 2012; Okai et al., 2011). The standard treatment is using L-dopa, an in-between molecule in the process of converting tyrosine to dopamine, in drug form (Brichta et al., 2013; Segura-Aguilar et al., 2014). Thanks to this treatment, tests can be performed comparing patients in their “on” and “off” states. For instance, in “off” state, patients show impatience in a delayed discounting task (Dalley & Roiser, 2012). This is implicating both L-dopa, thus dopamine in general, as well as D2/D3 specific agonists that are also sometimes used as drugs. The D2 receptor is also associated with learning as another Parkinson’s study shows us. Patients in the “off” mode learnt better from negative outcomes than from positive outcomes, while this was reversed in the “on” mode (Frank et al., 2004, in Rogers, 2011).

Further data on the D1 and D2 function in the amygdala is given by Smith, Geissler, Schallert, and Lee (2013) using a Parkinson’s disease animal model. The study focused on a cognitive attention switching behaviour that is also deficient in patients. The importance of the dopamine projections from the substantia-nigra to the amygdala are stressed in this study and D1 importance in a switching as well as a 5-CSRT task.

4.1.4. Other manipulations to dopamine

Administering L-dopa is a similar treatment to tryptophan supplementation in order to examine higher levels of serotonin. Also specific agonists can be used in the case of the different dopamine

(39)

receptors, as has been described for some of the Parkinson’s disease studies. Additionally, targeting neurons using toxin injections or knockout models is used in dopamine research as it is used in 5-HT research.

4.2. Dopamine and impulsivity

The D2 receptor is associated with impulsivity and aggression, and D1 is associated with decision-making (Rogers, 2011). A few studies described in his review indicate an interaction taking place in the anterior cingulate cortex (ACC) and the nucleus accumbens that seem to reflect cost-benefit calculations. This has been surmised from rats being observed willing to exert more effort for larger rewards. Rogers further describes the distribution of the D1 and D2 receptors as being instrumental in the balance between the direct and the indirect striatal pathways (2011). These two pathways are believed to dominate the Go and No-Go competing actions respectively. While there is empirical support of this hypothesis, affirming these assumptions with humans is at the moment tricky due to drug selectivity and location of action. D1 receptors, and not D2, are associated with fear learning as well as with mechanisms that involve the decisions relevant for the 5-CSRT (E. S. Smith et al., 2013).

Impulsive behaviour is often observed in substance abusers. Rogers (2011) discusses the deficiencies substance abusers have in probabilistic decision making that would be caused by damage to the fronto-striatal systems. However, as he further remarks, it could be that the drug seeking, impulsive, behaviour is the cause of an innate irregularity in the system that affects decision-making.

4.3. Link between dopamine research to predator and prey behaviours

How to link dopamine effects to predator and prey behaviour becomes a bit tricky, despite the effects it has on decision-making and the close link to the serotonin system. McEllistrem (2004) discusses some evidence found, most of it from research that is a little dated. In cats, dopamine

The roles of serotonin and dopamine in reactive and proactive aggression