The neurobiology of trust

(1)

________________________________________________________________

The Neurobiology of Trust

_________________________________________________________________________

Literature Thesis submitted for the degree of

MASTER OF SCIENCE of

BRAIN AND COGNITIVE SCIENCES, Behavioural Neuroscience Track,

University of Amsterdam

(2)

Name of student: Ursa Bernardic, 11123990 Name of supervisor: Dr. Uma Karmarkar Name of co-assessor: Dr. Jan Engelmann

(3)

Abstract

Section 1 discusses the definition of trust and why there is a clear need to understand trust as a phenomenon. Different frameworks from biology, psychology and economy, as well as measurements of trust are also reviewed. Section 2 outlines the neurobiology of trust: the role of genetics, different hormones and different brain activations. Next, the findings from neurobiological studies investigating trust were used as a platform to integrate the role of genetics, emotions, gender and age. Section 3 puts forth a conceptual framework of trust and further research questions are discussed. Lastly, Section 4 summarizes the findings and discusses challenges of trust and trustworthiness, propose the differences between trust and risk and put forward several hypotheses on the different roles of oxytocin and vasopressin. In each of the sections we discuss and suggest potential research questions for further studies which should try to combine methodologies and frameworks (such as multi-analysis and hormonal administration with different games from economics) and different theories (such as: moral, norm, and emotional theories) in order to better understand trust and distrust.

(4)

Acknowledgements

Many thanks to Dr. Uma Karmarkar for her great understanding, support and willingness to always offer an invaluable advice and comments on the previous versions of this manuscript. I would also like to thank Dr. Jan Engelmann for his willingness to co-assess and give feedback on this literature thesis.

Thanks to Ad Futura Scholarship for giving me the opportunity to study abroad.

Many thanks to people that increase my oxytocin levels, and activate my reward circuitry the most – in other words, to the people I trust the most, my family and my boyfriend, friends all around the globe, my professors, and colleagues.

(5)

Section 1: Introduction

In this thesis, we combine findings from social psychology, management, economy, neuroscience, and biology to better understand the concept of trust on an individual and organizational level. First, a review of different definitions of trust and different methods (i.e. trust game and questionnaires) used in economics and social sciences are discussed. Next, we examine whether trust is an individual or a societal concept and what are the conceptual differences in framing. Trust is among the most important factors in many situations in the human life. We discuss how trust research might give us better insights on other phenomena in all domains of society: ranging from individual (love, friendship), organizational (economic collaborations), medical and national domains.

Secondly, we reviewed the literature on the neurobiology of trust and divided discussions into three sections: the role of 1) genetics, 2) hormones and, 3) activation of different brain areas.

Thirdly, the economic framework evaluates monetary and material beliefs as well as costs that result from actions of participants studied mostly in the Trust Game. We review and integrate existing knowledge from the literature on the role of hormones and activation of different brain areas involved in the trust. Our goal was to build an interdisciplinary framework cantered on the concept of trust which recognizes the important role of emotions while being consistent with the broader range of neurobiological and psychological evidence in this field.

In the critical synthesis, we discuss two leading challenges of trust and trustworthiness. The first involves the measurement of these constructs. The second strives to untangle complex correlations between hormones, brain, and behaviour. We also reviewed interesting findings from the literature studying the role of uncertainty and betrayal aversion. These sparked new ideas and research questions for inconsistent literature on the two hormones studied in trust behaviour, oxytocin, and vasopressin.

(7)

What is ‘trust’?

Imagine initiating a new business relation with a close partner or a new start-up, or simply buying a new product from a well-known or unknown brand. How quickly do you respond to an email of a co-worker and would you respond at all to the email from a stranger? Would you save a credit card information online? Trust is among the most important factors in human societies that drives how we make many everyday decisions (Riedl & Javor, 2012; Baumgartner et al., 2008).

Trust is often described as a “social glue”, or “social lubricant” in relationships, groups or societies, which drives thoughts, motives and behaviours that enhance collective goals and connects people (Van Lange, 2015; Arrow, 1974, p. 23).

There are many definitions of trust, which mainly emphasize the definition of trust as:

1) Intention to accept vulnerability based on positive expectations or beliefs regarding the intentions or behaviour of another person or other people in general (Rousseau, Sitkin, Burt & Camerer, 1998, p. 395).

2) Expectations, predictability, and confidence in others’ behaviour (Zucker, 1986) 3) Expectations of other’s benevolent motives in situations that involve a conflict of

interest (Holmes & Rempel, 1989)

According to Willinger et al. (2003), trust is ‘based on a moral consensus between economic partners, which enables them to stick to their commitments without being obliged to do so by the sole means of formal contracts and legal obligations’. In some studies, trust is inherently regarded as a primitive code of behavioural conduct—like social norms (see Fukuyama, 1995). However, it is important to note that although there are many definitions of trust there are two critical elements all definitions share: vulnerability and expectation (Evans & Krueger, 2009; McKnight & Chervany, 2000). Whenever there is trust, there is also a possibility for betrayal, loss and regret. Therefore, it is important that a person who trusts has a confidence and positive expectations. In other words, when we trust, we accept risk because we believe we can avoid a negative outcome (Evans & Krueger, 2009).

(8)

It is essential that trust contains various suggestions regarding the determinants of trust and trustworthiness. More recent literature (Van Lange, rockenbach & Yamagishi, 2017) focus on micro factors (such as beliefs and perceptions, individual characteristics and preferences) and macro factors (context and structure society). Therefore, is trust a personal or societal trait?

Is trust a personal or a societal trait?

Delhey and Newton (2003) propose that there are two broad frameworks regarding trust, which defers whether trust is described as a personal or a societal trait. First, personality theory describes trust as an individual property, associating trust with personal characteristics. This is measured by variables such as optimism, a belief in co-operation, and the capacity to control the world (Erikson, 1950; Allport, 1961; Cattell, 1965). Uslaner (1999, 2000) suggests, these characteristics are learned at young age and tend to persist later in life.

On the other side, societal theories approach trust as a product of adult experience with society, in which an individual participates, contributes, and benefits in a trusting culture. Consistent with the social framework, four different theories (for details see Delhey & Newton, 2003, p. 101) describe relevant variables, such as community size, safety, wealth, and freedom.

Two different theories, personality and societal proposed various roles of trust in different phenomena. A full treatise is beyond the scope of this thesis, but we will next try to focus on how low or high levels of trust influence individuals, medicine, organizations and nations.

The role of trust

On the individual level

Growing evidence (Tov & Diener, 2008) suggests that high levels of trust in individuals are connected to more cooperative behaviour when facing uncertain and conflicting interests. Therefore, it is important to know the individual characteristics which drive these differences. Individuals with high trust in other people are better educated (Putnam, 2000; Paxton, 2007),

(9)

have higher income (Li et al., 2005) and more professional occupations (Alesina & Ferrera, 2000), they are more engaged in volunteering activities, and as report higher life satisfaction, increased physical health and life spans (Balliet & Van Lange, 2013).Lower levels of trust have been found among those who are divorced (Patterson, 1999), unemployed (Brehm & Rahn, 1997), have poorer health (Kawachi, 1997) or are a member of discriminated minorities (Alesina & Ferrera, 2000). A recent study by Guiso, Sapienza and Zingales (2008) suggests that less trusting individuals are less likely to buy stocks. And even if they decide to buy them, they buy less amount of the stocks.

Empirical studies therefore suggest that trust is a key to health, wealth and happiness, or vice versa, that health, wealth and happiness are the factors which plays a role in trust behaviour. It is important to understand whether trust also co-determine the evolution of these same factors. For example, people with higher incomes might behave more trusting and accept greater vulnerability to the actions of others, but it might also be that trusting behaviour is simultaneously rewarded by rising incomes (Hofman, Bulte & Voors, 2017). To the best of our knowledge not many studies tried to untangle the causal effects and context-dependent factors. This opens the questions such as: could changing the context (e.g. policies) in which people make decisions improve trust and trustworthiness? Can this improve people lives? And importantly, on which level does the change happen, is it related to psychology and/or biology? These real world situations (for example a drastic change in the policy) could be interesting to address this questions in the longitudinal, neuroeconomics study.

Medicine

The research on trust has also influenced the field of medicine. Many studies focused on the role that trust plays in the patient-physicians’ relationship and found that health outcomes of different diseases (Lee & Lin, 2009), patient satisfaction, therapy adherence and continuity of care, were associated with the patient’s trust in physicians (Thom & Campbell, 1997; Safran et al., 1998).

Further, there are several neurobiological diseases which are known to lead to an impairment of trust-relevant brain areas and associated hormones. Recently, economically-oriented scientists have focused on different psychiatric disorders and neurological diseases. For example, borderline personality disorder is characterized by the incapacity to trust actions and

(10)

motives of others appropriately. Neuroeconomics researchers investigated this effect by (King-Casa et al., 2008) using a game-theoretic approach to study how behavioural and neuronal data differ between healthy subjects and patients with borderline personality disorder. Their findings suggest that individuals with borderline personality disorder are incapable to maintain cooperation and don’t have the ability to repair broken cooperation. Neuronal data suggests this was correlated with the graded activation of the anterior insula cortex, which is associated with awareness of social gestures in healthy participants (King-Casa et al., 2008).

Other studies involving individuals with psychopathic disorders additionally found subjects general show less cooperative responses during the trust games, and found reduced activity in the amygdala, an area correlated with uncooperativeness (Beck et al., 2015). Fett et al. (2012) similarly reported that patients with psychosis and their relatives have lower levels of trust when playing the trust game when compared to healthy participants. Lastly, low levels of trust in trust games were reported in patients with borderline personality (Unoka et al., 2009).

Understanding the psychological and neuroscientific side of trust as a phenomenon could help us provide insight into the biological foundations of psychiatric disorders (Javor et al., 2013) and offer ways to test the hypothesized mechanisms of action for some cognitive or medical treatments (Beck, Freeman & Davis, 2015). On the other side additional research could identify diseases which are leading to lower trust and provide guidelines for improving the patient’s trust (Javor et al., 2013).

On the organizational level:

The Nobel laureate Kenneth Arrow suggests that, “every commercial transaction has within itself an element of trust” (Arrow, 1972, p. 357) - especially in today’s international trade and financial market (Zak, 2017). Therefore, understanding the role of trust at an organizational level is a critical question for business and organizations.

There are three broad strands in literature of trust on the organizational level which focus on: intra-organizational trust (i.e. within organizations: between managers, among co-workers), inter-organizational trust (i.e. trust between organizations) and third, between organizations and their consumers (i.e. marketing concerns) (Seppänen, Blomqvist & Sundqvist, 2007).

(11)

On the intra-organizational trust, Six and Sorge (2008) compared two consulting organizations with different trust policies but similar characteristics. They studied the generation of trust and strategies the two company used when trust was threatened, with surveys and half-open interviews. Their findings suggest that higher state of trust can be reached by inter-related set of policies (i.e. promoting the relationship-oriented culture, creating opportunities for meeting informally, day-to-day management of competencies, etc.).

Zaher & Haris proposed that the outcomes of inter-organizational trust are visible on many levels: from direct economic outcomes (i.e. lower transaction costs), relational outcomes and indirect effects (such as mediating roles). Chen and colleagues (2014) also found that inter-organizational trust influence positevly the collaboration and knowledge sharing. In order to develop inter-organizational trust, they suggest that organizations should embrace shared goals, and social relationship.

Trust is important also in marketing research. For example, studies (Lau & Lee, 1999; Chaudhuri et al., 2001) suggest that trustworthiness of brand is one of the main factors for brand loyality.

On the national levelr

On the national level, higher social trust is positively correlated with GDP growth across countries (Knac & Keefer, 1997; LaPorta et al., 1997; Zak & Knack, 2001), reduced corruption (LaPorta et al., 1997), lower levels of state corruption (La Porta et al, 1999; Rothstein & Uslaner, 2006), lower levels of income equality (Uslaner, 2002), and lower levels of criminality (Halpern, 2001). Delhey and Newton (2003) found that social trust is high among the citizens who believe there are fewer social conflicts and where the sense of public safety is high. With regards to which framework, individual or social theories is used, Delhey and Newton (2003) suggest that the individual theories work better with societies where trust levels are higher. Contrary, they found that societal theories might give us more insights in the nations/countries, where levels of trust are low (for details see Delhey & Newton, 2003). Hence, the trust level of individuals may be translated to that of society, or even to a national level, and vice versa (Barr et al., 2015; Guillen & Ji, 2011).

(12)

Trust has also been shown to play an important role in international relations. Higher bilateral trust between two countries is correlated with more trade between the countries (Guiso et al., 2009).

To conclude, trust is a powerful construct and a key concept on multiple levels of social phenomena: high trust improves our individual life, makes relationships fare better, as well as increases the performance of groups and organizations. It makes our societies more functional and nations more flourishing. In other words, trust should develop and spread around the globe (e.g. Platteau, 1994; Fafchamps, 2011; Tu and Bulte, 2011, Hofman et al., 2017).

To do so, we must understand trust from interdisciplinary approach and integrate neurobiological findings that might shed light into current frameworks. Let’s first define how trust is assessed in behavioural, as well experimental economic studies.

How trust is measured?

There are two main frameworks in which trust is measured: survey instruments and the trust game. Most of the studies using survey instruments are focusing on the two levels of trust:

· Trust-in-others: which measures the belief, attitude or intention. Such

questionnaires and surveys, includes questions which are focused on the characteristics of the person, which we should trust, named trustee and their abilities, benevolence and integrity (for details see McKnight et al., 2002). Most papers in literature use the questions from World Values Survey or General Social Survey. These questions ask about beliefs in others, such as: “Generally speaking, would you say that most people can be trusted or that you cannot be too careful in dealing with people?” There are controversial results and opinions whether this question is a measure of trust or trustworthiness. Sapienza, Toldra-Simats and Zingales (2013) suggest that this question mostly measures the belief component of a trust game.

· Trust-in-self: which measures the level of trust of other people in ourselves.

Usually they study this kind of trust with questions, such as: “I think that most other people trust me” (Van Lange, 2015, p. 72).

(13)

On the other side, Trust Games have been widely used in economics to measure both trust and trustworthiness as actual behaviour in interpersonal settings. In the framework used in experimental settings of economic games, the trust is described as a reflective expectation about the behaviour of others (e.g. the probability that someone will perform an action that is beneficial), which impact our considerations (e.g. engage us in some sort of cooperation) (Gambetta, 2000). This is irrespective of the ability to control the actions of the other party (Mayer, Davis & Schoorman 1995).

The classic version of the Trust Game was originally called the Investment Game (Berg, Dickhaut & McCabe, 1995). It is a simplified version of a game in Camerer and Weigelt (1988), where two players - called investor/trustor (first decision maker) and trustee (second decision maker) - interact nonverbally by sending money to each other in a two-step sequence (see Fig. 1).

In the mainstream version, a trustor and trustee have no face-to-face contact and are therefore informed about the rules upfront. Each player, trustor and trustee are endowed with some fixed level of endowment (let’s assume 10 units of money - MU). In the first step the investor has to decide about the amount (s)he would like to send, or invest i ∈ {0,1,2…10} to the trustor, via experimenter. The experimenter multiplies the amount by a number (in many experiments multiplier m = 3), and sends this amount (i x m) to the trustee (DM2). In the second step the trustee returns an amount r ∈ {0,1,…mi} back to the trustor. The monetary payoff of the investor and the responder at the end of the game are therefore, for trustor: 10 - i + r and for trustee: 10 + mi – r. In economic settings, the numbers i and r are the respective measures of trust and trustworthiness.

From the economic perspective, the trustee does not have a reason to reciprocate and therefore the Nash equilibrium occur and the rational way for the trustor to keep everything is to not invest at all. Many empirical studies (for a review see Declerck, Boone & Kiyonari, 2013; Johnson & Mislin, 2011) nonetheless reported that subjects do invest variable amounts, and that many trustee’s do payback the higher amount they receive.

Although the economic framework focus on monetary and material beliefs and costs that result from actions of participants studied mostly in the Trust Game, we suggest that studies from

(14)

neuroscience and biological findings could add to the whole picture of understanding trust behaviour. Therefore, we would like to discuss the important role of genetics, hormones and activations in brain regions as a result of different outcomes in trust game.

Section 2: Neurobiology and development of trust

There are three main strands of research on neurobiology and trust. We will first discuss the contribution from behavioural genetics. Secondly, we will discuss how different levels of various hormones can influence the behavioural measures of trust. Thirdly, the findings from neuroimaging techniques (such as fMRI and EEG), which investigate relationships between trust and neural activity in different brain regions will be discussed. Lastly, we will try to evaluate how these three main strands can be integrated and what is the role of uncertainty, emotions and gender in such trust-decisions.

The role of genetics

A survey-based research by Dohmen and colleagues (2006) indicated that willingness to trust is similar between parents and children. This begs the question whether trust is influenced by environmental (socialization, culture) or genetic factors. Most genetic studies (Cesarini et al., 2008; Sturgis et al., 2010; Van Lange et al., 2014) investigate trust by studying monozygotic or dizygotic twins. Monozygotic twins are developed from single fertilized egg and therefore share the same genes as well as their environment. Meanwhile dizygotic twins are developed from two fertilized eggs and therefore share on average 50% of the same genes, the same as non-twin sibling pair and the environment. Cesarini and colleagues (2008) addressed heritability of trust and trustworthiness in two large, independent samples of twins in either United States (72 dizygotic and 258 monozygotic same-sex twin pairs) or Sweden (75 dizygotic and 278 monozygotic same-sex twin pairs). They measured behavioural trust in the trust game, and found that heritability estimates for trust were 20% in the Swedish experiment and 10% in the US. Moreover, heritability estimates for trustworthiness were 18% in the Swedish experiment and 17% in American.

(15)

Sturgis and colleagues (2010, p. 212), studied the variance of trust with a survey, asking people to give estimates about: “My first reaction is to trust people”, “I think that most of the people I deal with are honest and trustworthy” in the sample of monozygotic and dizygotic twins. They conclude that the variance of trust is accounted for by an “additive genetic factor” and that environmental influences have no discernible effect. Although trust was again assessed with a three item measurement of trust-in-others survey (e.g. I completely trust most other people.) and trust-in-self (e.g. I think that most other people trust me.), Van Lange and colleagues (2014) found that in the sample of 1,012 twins and relatives of different ages and groups, heritability estimates for trust-in-others were 5% and heritage for trust-in self were 13%. Therefore, Van Lange and colleagues (2014) suggested that genetic influences are virtually absent for trust and concludes that “general trust is more a matter of culture than genetics” (for a discussion see: Van Lange, 2015, p. 72).

The role of hormones

The effect of gene expression is very important on the development of different brain structures, especially via different hormones (Harris et al., 1998; Rushton & Ankney, 2007). In this section we discuss how the studies from neuroendocrinology could help us better understand the framework and provide some possible connections between genetic heritability and trust. Based on empirical evidence, oxytocin (OXT) and arginine vasopressin (AVP) has been one of the first and most studied hormones which influence trust and social behaviour.

The important behavioural model of the role of OXT in trust behaviour, was first investigated in animal models of prairie voles (Insel & Young, 2001) and rats (Fleming & Rosenblatt, 1974). Their findings suggested that OXT is crucial for bonding, monogamy and willingness to provide maternal care. Based on the animal literature, Kosfeld et al. (2005) conducted a version of the trust game in which participants were randomly assigned to two groups: test group where subjects inhaled OXT spray and control group where subjects inhaled a placebo spray. They found that participants who were administrated with OXT and played a role of trustor (DM1) invested much more. Moreover, such participants more often showed maximum trust (with sending all the money to trustee) than participants in control condition. Importantly,

(16)

the OXT did not raise trustor belief about the back transfer from trustees. On the second step of the trust game, trustors who were administrated with OXT were not more willing to send money back to trustor, which suggests that OXT does not make people indiscriminately more prosocial. Although, one could hypothesize that OXT could influence risk preferences and make people more risk-seeking, further analysis of a non-social risk game (binary lotteries which where context was perfectly matched) found no significant difference between the placebo and oxytocin group in risk-taking or ambiguity preferences. Therefore, the neurophysiological mechanism of OXT from this study suggests that subjects’ preferences in the trust game are beyond those needed to account for risk and ambiguity.

Although authors (Kosfeld et al., 2005) claim that the risk task bring them a control which allows to measure social interaction and betrayal aversion, we would like to be critical, as other social preferences like altruism or inequity aversion could be involved. As other researchers (Nave, Camerer & McCullough, 2015) pointed out, this is a drawback of many researches in OXT, because many theories are potential explanations of change in behaviour, but their capability to generate falsifiable predictions of how OXT will affect complex behaviours might be limited. Secondly, the findings from other studies are not consistent. One meta-analysis (Nave et al., 2015) of similar intranasal OXT games revealed that the combining the effect size of intranasal OXT and trust was not different from zero. Other researchers (Bartz et al., 2011) have proposed that exogenous oxytocin can be viewed as altering the salience of interpersonal cues, producing a wide variety of behaviour effects depending on the context. Also, the researchers (Kosfeld et al., 2005) proposes application of OXT treatment in people with autism. Nevertheless, it would be imprudent, given that some researches have shown that OXT reduces trust in subjects with borderline personality disorder (Ebert et al., 2013). Another minor concern is that participants in the study by Kosfeld and colleagues (2005) interacted with other humans while waiting for the task. That interaction could have affected the mood of the participants.

In addition, recent replication studies on OXT and trust behavior, which use slightly different experimental designs, are largely inconclusive (Barraza, McCullough, Ahmadi, & Zak, 2011; Baumgartner, Heinrichs, Vonlanthen, Fischbacher, & Fehr, 2008; Ebert et al., 2013; Klackl, Pfundmair, Agroskin, & Jonas, 2013; Mikolajczak et al., 2010; Yao et al., 2014). Although the causal effect and role of OXT on trust is intriguing, these findings need to be independently replicated with different methodologies in order to become scientific evidence. A

(17)

recent meta-analysis by Leppanen and colleagues (2017) which pooled the result from 33 studies found that intranasal oxytocin improves the recognition of basic emotions, particularly anger and fear. It is worth noting that this finding is limited to a healthy population.

Besides oxytocin, also arginine vasopressin (AVP), was associated with a regulation of various social behaviours. Importantly, they are both coded in a precursor form on chromosome 20 and consist of nine amino acids in a cyclic structure which differ only by two amino acids at position 3 and 8 (Gimpl & Fahrenholz, 2010). Anatomically, OXT and AVP are nonapeptides synthesized in the hypothalamic paraventricular and supraoptic nuclei, with AVP also synthesized in suprachiasmatic nucleus (Insel, 2010). Although OXT has a single receptor OXTR, vasopressin has three types of receptors, AVPR1 which is primarily present in neurons, liver and vascular smooth muscle, AVPR1b found in anterior pituitary and V2 in kidneys (De Keyzer et al., 1994; Thibonnier et al., 2002).  Bluhe and colleagues (1990) studied found that in rodents AVP is related with social recognition, and also with the capacity for recognize distinct conspecifics. Later studies on mammals showed that AVP plays a central role in memory and regulates many male social behaviours, such as aggressive behaviour, territorial marking, pair-bonding, monogamy and paternal behaviours (Choleris et al., 2013; Winslow et al., 1993; Le Moal et al., 1987; Landgraf et al., 1995, Everts & Koolhaas, 1997; Bielsky et al., 2004; Wersinger et al., 2002). Initial behavioural experiments indicate that AVP seems to be a potent modulator of reciprocal and un-reciprocal cooperation in social situations in humans (Rilling et al., 2012; Brunnnlieb et al., 2013; Chen et al., 2015).

In the studies where OXT and AVP were administered nasally, there are many critical concerns about methodological challenges, underpowered studies, confirmation and publication bias (Leng & Ludwig, 2016; Leng & Sabatier, 2016). Almost all the studies described before were based on three assumptions: that intranasal OXT crosses the brain-blood barrier, that 24-40 IU is a sufficient dose to produce behavioural changes and that between-subject design of small samples (of 30 subjects) are adequate.

To the best of our knowledge, it remains unclear whether such administration is sufficient for passing the blood brain barrier. There are studies demonstrating that intranasal OXT increase the OXT concentration in cerebrospinal fluid of animals (Neumann et al., Freeman et al., 2016)

(18)

and humans (Striepens et al, 2013). However, recent studies remain critical about measuring methods (bioassays and Elisa) for OXT and AVP (for a review see Leng & Sabatier, 2016).

An additional concern is the temporal span of the intranasal administration. Most of the studies started the tasks 40-45 minutes after the intranasal administration of oxytocin. However, it was proposed that the significant elevation takes place 75 minutes in humans (Lane et al., 2016).

Most of the studies also used the single blind design, where the subject is blind to the treatment condition but the experimenter is not – this might unconsciously also indluence the subject.

Researchers (Evens et al., 2014; Leng & Ludwig, 2016; Merkus & van den Berg) therefore suggest that oxytocin-based studies cannot be generalized and that there needs to be confirmation that nasal drug delivery bypasses the blood brain barrier in humans (Evens et al., 2014; Leng & Ludwig, 2015; Merkus & van den Berg).

We believe that the research agenda of studies on OXT and AVP should first review all the unpublished studies (for example as in Lane et al., 2016), and further work closely with neurobiologist to improve the standard procedures in hormonal studies.

Combining insight, procedures and methodologies from psychology, biology, neuroscience, and economics will hopefully not only highlight the modulation of neural circuits, but also the temporal span of administered treatments. This will provide additional insights into the treatments for social anxiety and attention deficit disorders. Importantly, we would like to raise critical concerns regarding the popular work by Zak (2008, 2017) which proposed that the effect of OXY is universal, although it remains unknown whether these effects are generalized to specific subjects population.

Lastly, trust is not a uniform phenomenon nor is its neurobiology. Therefore, other hormones, such as cortisol, dopamine, estrogen, and serotonin should be investigated as they likely play an important role in modulating trust behaviour.

(19)

Neural correlates of trust

The positive role of OXT on trust was further investigated in the fMRI study by Baumgartner and colleagues (2008), which examined the differences in neural activity between trust and risk-taking (Trust game and matched Risk game) in male subjects with nasal administered OXT or placebo. In contrast to Kosfeld and colleagues (2005), participants in this study (Baumgartner et al., 2008) received feedback after half the trials (pre-feedback) on low back transfers in the trust game or low investment success in the risk game. After learning the feedback of limited investment success (that decision to trust or take a risk were not returned in half of the trials), they played another half of the trail (post-feedback). Participants playing the trust-game, which was administered with a placebo, decreased the transfer of post-feedback investment, compared with pre-feedback. Meanwhile participants administered with OXT increased the transfer of post-feedback investment, compared with pre-feedback. Interestingly, differences in behavioural adaptation were not observed in the risk game. Neuroimaging findings demonstrate that activity in amygdala, the midbrain regions and dorsal striatum increases during trust taking in the post-feedback (compared to pre-feedback) in the placebo, but not in the group administered with OXT. This suggest that social risk taking (trust) goes beyond what is needed for taking a social risk and that OXT has an important role in regulating this processes.

Also studies on patients with brain lesions are informative to evaluating the neural correlates of trust. For example, Adolphs and colleagues studied patients with bilateral amygdala damage and compared them to normal viewers. They found that patients with bilateral amygdala demage judged others to look more trustworthy and approachable based on their facial appearance (Adolphs et al., 1998). Winston and colleagues (2002) studied which brain areas are activated during the evaluation of trustworthiness of faces or judgement of age (control task). They found that increased activity in the bilateral amygdala and right insula is correlated with responses of untrustworthy faces judgements. Unlike the amygdala, the right temporal sulcus showed enhanced signal change while judging explicit trustworthiness. However, it also showed

(20)

greater responses to untrustworthy individuals when judging age. This suggests responses might be task-dependent (Winston et al., 2002).

Brunnlieb and colleagues (2016) found that intranasally administered AVP increases male's’ willingness to engage in risky cooperation, while playing a Stag hunt. This is not mediated by general risk attitudes, a change in beliefs, social preferences or mood. Findings from functional imaging suggest that when making a risky choice AVP down-regulates the BOLD signal in the left dorsolateral prefrontal cortex (dlPFC), which is known as a risk-integration region. Moreover, results from functional connectivity suggest that AVP increases the connectivity between dlPFC and ventral pallidum (VP). VP is a region which is rich in AVP receptors and dopaminergic pathways. It is plays a role in social reward processing in mammals. Together, these findings suggest that AVP facilitates beneficial risky cooperation by reducing aversion to social risk, without changing general aversion to risky outcomes, nor social preference.

The role of emotions

For many years’ researchers examining trust in an economic framework did not consider or integrate psychological factors on trust behaviour into models. However, recent evidence suggests emotions play an important role in attention and social decision-making (Cohn et al., 2015; Harle & Sanfey, 2007; 2010; Lerner et al., 2015), and also specifically in trust research (Mislin et al., 2015; Engelmann & Fehr, 2016; Engelmann et al, 2017).

On one side, there are studies which shows that emotions and mood can increase the likelihood of trust behaviour. Dunn and Schweitzer (2005) found that happy emotional states increase a person’s intention to trust both to known and unknown others. This was also confirmed by Mislin et al. (2015) which showed that watching a clip that induced a happy mood, as well as writing about happy moments increased the likelihood of trust behaviours.

(21)

In contrast, many other behavioural studies focused on which emotions are involved in trust and how trust is differing in risky non-social decisions. Bohnet and Zeckhauser (2004) showed that when playing a modified version of the trust game and structurally identical risky dictator game, participants elicit higher minimum acceptable probabilities in the trust game than in risk game. This suggests that social settings influence trust taking beyond mere risk aversion. Emotions are one of the potential mechanisms involved in these differences given that they go above and beyond the monetary gain (Mislin et al., 2015) or loss (Bohnet et al., 2008; Kosfeld et al., 2005).

The role of gender

Many previous studies, especially on the role of hormones studied only males or females, and recent studies involving both, has shown gender differences in trust behaviours. Females are generally less likely to trust other parties (e.g. Glaserer et al., 2000). One explanation for this effect is that females are warier of uncertainty and risk (Byrnes et al., 1999; Alesina & La Ferrara, 2002; Buchan et al., 2008). One evolutionary argument for this has been that is a result of their responsibility to protect offspring (Delhey & Newton, 2003). However, while lower levels of trust in females was found in the USA (Patterson, 1999), there is little difference between levels of trust and gender in other western countries (Whiteley 1999: 41; Newton 2001), arguing against an evolutionary underpinning. Therefore, we propose that the education, cultural identity and trust on the societal levels could play an important role in gender equality and different trust levels across genders.

Interestingly, in the previously discussed study by Aimone and colleagues (2014) it was also found that women were less affected by betrayal aversion than men. This might be a result of gender differences in AVP levels. Previous findings from neuroimaging studies found that brain regions involved in vasopressin circuitry - such as the bed nucleus of the stria terminalis, lateral septum and stria terminals have lower BOLD signals in females (Rilling et al., 2012). Additional, evidence shows that AVP is more influential in males (Young et al., 2005). Indeed, subsequent studies using fMRI also found an important role of gender in response to AVP

(22)

administration; bilateral insula and right supramarginal gyrus activity increased in men, but decreased in women, during reciprocated cooperation (Feng et al., 2014).

fMRI studies which intranasally administered oxytocin found gender differences in brain activity. On study demonstrated that OXT decreases amygdala activity in response to emotional faces in males, but increase amygdala activity in females (Domes et al., 2007; Domes et al., 2010). However, using the same OXT paradigm did not appear to have any effect on females (Rilling et al., 2014). One promising way to evaluate this contradictive results are to review the studies. For example, a recent meta-analysis examining the role of online trust in business-to-consumer e-commerce suggests that there are no significant sex-differences (Wang et al., 2017).

Measuring trust with the Trust game has also provided mixed gender differences across both the trustor and trustee behaviour. Some studies have found that men are more trusting or risk-seeking and invest more when they are in the role of trustor (DM1) (Lemmers-Jansen et al., 2017; Bohet, 2007; Cox & Deck, 2006; Schwieren & Sutter, 2008, Takahashi et al., 2016). In contrast, others have found there is no gender difference in sending behaviour (Chaudhuri & Gangadharan, 2007; Buchan et al., 2008; Garbarino & Slonim, 2009). However, one study reported higher levels of trusting for women than men (Bellamare &Kröger, 2003).

Trustee (DM2) results are also inconsistent. Some report women are more reciprocal than man (Buchan et al., 2008; Chaudhuri & Gangadharan, 2007; Schwieren and Sutter, 2008; Snijders & Keren, 2001) while others report no differences between genders (Bohnet, 2007; Cox & Deck, 2006). In contrast, a baseline study by Bellamare and Kröger (2003) found that men were more reciprocal than women. One recent study showed that age and gender differences become increasingly apparent during unfair interactions (Lemmers-Jansen et al., 2017). Moreover, they found that gender differences in brain activation are only found during cooperative repayment, with increased activation of temporo-parietal junction (TPJ) in males, and increased activation of caudate in females.

(23)

Development of trust and the role of ages

There are three main strands of literature on trust and trustworthiness across different ages. The first strand concentrates on the economic behaviour of children (Harbaugh et al., 2003), the second on the behavioural differences in the adult population (Fehr et al., 2003; Bellemare & Kröger, 2007) and the third combines findings from developmental psychology and economics (Sutter & Kocher, 2007).

Harbaugh and colleagues (2003) used the experimental trust game with children ranging from 8 years-old to 17 years-old. They reported insignificant age differences in the influence of either trustors’ or trustees’ decisions. A potential explanation for this effect is that children might not fully understand their protocol, especially how their strategy vector was implemented.

Other researchers included a trust game in a representative survey and also found that age is an insignificant determinant for behaviour in the trust game (Fehr et al., 2003; Bellemare & Kröger, 2007, Hofmann et al., 2017). They found that participants over the age of sixty-five show less trust when being a trustor than participants in their thirties or forties. Additionally, the demonstrated that trustworthiness (the decision trustee makes) increases with age. However, Sutter & Kocher (2007) criticize this approach as the participants didn’t have any information on the age of their interaction partners.

Therefore, subjects (aged from 8 to 68 years old) interacted with members of the same age group. Their study brings two main findings. First, the transfers amount of trustors’ increased monotonically from 8-year-old children (2.0 units) to students (6.6 out of 10 units). However, the transfers are almost identical between students, professionals, and retired persons. Secondly, they report that trustworthiness (the amount of return) prevails in all age groups. Their findings confirmed that trustworthiness is found in young children and that the degree of trustworthiness increases with ages.

Lastly, age-related increases of activation of the TPJ and the right dorsolateral prefrontal cortex (dlPFC) have been found when a trustee receives trust from trustor. In younger

(24)

individuals the same phenomena increased the activation in anterior medial prefrontal cortex (amPFC), a region associated with self-oriented processing and mentalizing (van den Bos et al., 2011). These results suggest that different neural activation patterns in brain areas are involved with age-related changes in trust and trustworthiness of others.

Trust building in online settings

Internet and mobile technology has fuelled the sharing economy, including firms such as Airbnb, Amazon, BlablaCar and Uber, which have heavily impacted almost all the aspects of the consumer behaviour worldwide (Avital et al., 2015). Trust is an important factor in online interactions and have been labelled as a sharing economy’s currency (Botsman, 2012) between the collaborative peers-to-peers consumption (Botsman & Rogers, 2010).

An experimental framework for trust in the sharing economy is based on the renowned trust game, named sharing game (Hawlitschek et al., 2016). As in original game, also sharing game has two steps. In the first step a consumer’s trust in provider and in second, the provider trust in consumer (for a detailed review of the game see Hawlitschek et al., 2016, p. 6). Most importantly, sharing game mimic the real online situation and offers manipulation of many dependent focus variables, such as speed, rate, characteristics of matching, as well as trusting and reciprocating behaviour of participants.

Integrating the literature on distrust in the context of impersonal IT-enabled exchanges between buyers and sellers in online marketplaces, Dimoka (2010) showed that distrust is a distinct construct from low trust, associated with different neurological processes. While trust was associated with the brain area for reward, prediction and uncertainty, the distrust was associated with the brain’s intense emotions and fear areas.

(25)

Environmental factors: The role of socialization, culture

Individuals of different countries and cultures often base their behaviour on different values and norms (Netzer & Sutter, 2009). One possible explanation for this is that people in countries less polarized along class or ethnicity cooperate more. Additionally, individuals in high-trust societies (which promote property and rights) generally spend less to protect themselves (Knack & Keefer, 1995). This frees up time which can be devoted to innovation of new processes and products. Therefore, they give more credentials to education, than to personal attributes, such as blood ties or personal knowledge (Clague’s, 1993; Knack & Keefer, 1997). Inspired by Knack & Keefer (1995), many studies and cross-cultural experiments are investing trust using the similar experiments (e.g. trust game) to shed light on the role of culture in different countries (Guillen & Ji, 2011; Hennig-Schidt et al., 2007, Kim et al., 2013). Holm and Danielson (2004, 2005) found no differences between Sweden and Tanzania. Ashraf et al. (2006) reported similar levels of trust in South Africa, US and Russia. Similarly, Bohnet and colleagues (2008) found very similar behaviour in a trust game in six different countries (i.e. Brazil, China, Oman, Switzerland, Turkey and the US).

On the other side, intercultural studies have the aim to examine situations in which subjects from different cultures interact with each other. Studies investigating intercultural situations not only study how our decisions are affected by different value systems, but also the role of perception and reputation of a culture or country. When participants could identify the ethnic origin of their interaction partner (by identification of the first name), a large degree of mistrust towards male Jews of eastern origin was found (Fershtman & Gneezy, 2001). Willinger et al. (2003) found higher levels of trust in Germans compared to their French counterparts in student population and vice versa. Hennig-Schmidt and colleagues (2006) found highest levels of trust in Argentina, lower levels in China, and the lowest in Germany. Therefore, intercultural interactions in trust seem congruous. Causal psychological determinants, such as cultural characteristics and/or stereotypes of the interacting partners, have been proposed in the literatures of trust, positive reciprocity, fairness and also public good contribution (see Carpenter & Cardenas, 2011; Chuah et al., 2007).

(26)

Recently, the level of trust and reciprocity was investigated in the intercultural trust game between Austrian and Japanese compatriots (Netzer & Sutter, 2009). Japan and Austria are countries which differ in a number of cultural dimensions. Japanese culture is characterized as collectivist while Austria is individualist. The authors conducted the trust game in three different treatments: an intercultural treatment (where Austrian subjects were matched with subjects in Japan) and two intracultural treatments (subjects played among subjects of the same country). They found that Austrian participants trusted Japanese participants more than they do fellow countrymen; while, Japanese did not differentiate between Austrian and Japanese partners. A post-experimental questionnaire revealed that these differences are consistent with differences in general attitudes towards trust and reciprocity. Austrian subjects were more trusting and reported themselves as more trustworthy than Japanese subjects did. The authors suggest that the driving force for the higher levels of trust and trustworthiness might be in line with the widespread perception and characterization of Japan as a society with high trust and trustworthiness. Such national perceptions or stereotypes therefore influence behaviour and reveal the influence of stereotypes and culture on decision-making. On top of that, evolutionary psychologists have demonstrated that humans have the capacity to learn social norms. These norms help individuals identify with a certain social or identity group. Therefore, different intercultural interactions and trust levels between individuals’ behaviour may be determined by social norms (Barr et al., 2015; Kim et al., 2013), and religion (Gupta et al., 2013).

Stanley and colleagues (2012) provided findings on how social reputation in group membership (i.e. defined by different races) impact the brain areas involved in trust decisions and trustworthiness. Using the single-shot trust game they found that higher blood-oxygenation-level-dependent (BOLD) signals in striatum correlated with a partner’s decision to trust a partner from the race group which was perceived less trustworthy. In contrast, amygdala activation occurred when the emotionally relevant social group was represented.

(27)

Section 3: Integrated model of trust

Figure 1: An integration of findings from economic, neuroscientific and biological studies in the model of the trust game. The economic outcomes are presented with the dollar sign, the outcomes from studies investigating emotions and different brain regions involved are marked with question mark, the changes on the hormonal level are marked with the molecular structure of OXT.

Recently, many studies have tried to combine knowledge from multiple disciplines to study trust and trustworthiness on different levels. Researchers have proposed numerous explanations for why we trust or distrust. From different personality traits, group processes and social norms in psychology, to preferences and expectations in economics and different neural

(28)

circuits in neuroscience. A large amount of these studies used the same measurement, the trust game, although using different frameworks. We believe that an integrative model examining both trust and trustworthiness is needed. Therefore, we reviewed literature and provide summary of outcomes, both in economic terms and material benefits from the game as well as the emotional, neurological, and hormonal responses.

According to the original trust game, one player is called the trustor (first decision maker) and the other is assigned the role of trustee (second decision maker). The trustor is instructed to invest all, some or none of the money units to the player 2 (the trustee). If the trustor does not invest any of the started investment (in the Fig. 1, 10 units), the game is over and the amount is split equally between two of the players (e.g. 5 units to trustor and 5 units to trustee). Distrust has been viewed as a lack of confidence, fear of harm and lack of care about the trustor’s welfare (Govier, 1994). Recent studies (Aimone et al., 2014, Bohnet & Zeckhauser, 2004). Betrayal aversion is when people dislike the non-reciprocated trust and therefore don’t invest anything. Engelmann and Fehr (2017) proposed that aversive emotions play an important role in trust game, especially in betrayal aversion. The role of uncertainty and betrayal are because of it’s importance discussed in the next chapter of critical synthesis. It has been also shown that half of the trusting trustors (i.e. those who invest) believe that they will not make any profit out of trusting (Dunning, 2014). Interesting, they keep trusting also when they are informed about their non-maximizing option (Ortmann et al., 2000).

Another strategy for trustors it to invest some or all of the 10ECU to player 2. In this case, the units that are sent are tripled and the trustee therefore receives 30ECU. It has been also shown that half of the trusting trustors (i.e. those who invest) believe that they will not make any profit out of trusting (Dunning, 2014). Interesting, they keep trusting also when they are informed about their non-maximizing option (Ortmann et al., 2000).

The act of being trusted is inherently rewarding. Social preference theories suggest that at least some people treat the payoffs of others as positive utilities (Bolton & Ockenfels, 2000; Fehr & Schmitt, 1999). They also suggest that such benevolent preference for trust and cooperation makes them (i.e. trustor’s in the trust game) feel good.

(29)

Li and colleagues (2009) found that in trustees’ brains the VMPFC is more active when trustees receive money from the trustor without any sanction or threats. Next, the trustee decides how much he or she will return, or reciprocate to the trustor. If he or she decides to return and reciprocate to the trustor, in this case trust game measures the trustworthiness of the trustee. The studies show that learning about trustworthiness of the trustee is correlated with projections in anterior insula and amygdala and high levels of oxytocin, dopamine, estrogen and serotonin in trustor’s. These are the hormones that mediate reward and positive emotion circuits. Cooperation and trustworthiness increase the levels of dopamine and vasopressin in trustee. This suggests that trustworthiness is correlated with higher activations in prefrontal cortex and amygdala.

Although not many studies focused on the brain activity of trustees, Baumgartner and colleagues (2009) found that when trustees had to indicated how frequently they were intending to be trustworthy, the brain activity of those who break the promise is higher in the anterior cingulate cortex and frontoinsular cortex. This is in constrat to the activity for those who are honest and keep the promise. Combining these results with the results from different economic games, where cortex activity has been related to negative emotions triggered by unfair outcomes, we suggest that negative emotions play an important role in betrayal. However, further studies should focus on the brain regions and neurobiological mechanisms involved in such untrustworthiness.

A recent study by Espin, Exadaktylos and Neyse (2016) focused on trustful and trustworthy behaviours in a binary trust game. First, they found that individuals who are both trustful and trustworthy differ remarkably from those individuals who are trustful but not trustworthy. The main difference is that trust in the trustful and trustworthy individuals is driven more by efficiency concerns (i.e. a preference for maximizing the total surplus) than by strategic considerations. They found that egalitarianism or aversion to inequality impacts trustworthiness but not trust. We propose that such findings should be combined with the administration of different hormones and fMRI techniques, to get the unified view of trust and trustworthiness.

On a macro scale, cultural, educational, and societal differences are important factors in developing trust behaviour (Slonim & Fuillen, 2010). One of the theories which stands out is the

(30)

attachment theory from psychology. This suggests that the security of an environment provided by a primary caregiver is a strong determinant of later attachment styles and emotion regulation (Mikulincer, 1998). We agree with Van Lange (2015) suggestion that the comprehensive analysis of different types of influences on trust behaviour should be tested on the three levels. First, he proposed the role of personal social interaction experiences, since there are powerful experiences which may lower the trust behaviour (e.g. burglary, unexpected unemployment, mistreatment of authorities). Secondly, the influence of the experiences from close others, who have powerful social experiences and third, the societal experiences (e.g. through media, especially online social networks) are needed. Further studies should focus on the role of such such social-cognitive filters and take into consideration individual difference factors and different norm theories (Evans & Krueger, 2009).

To conclude, neuroimaging studies have implicated different brain regions and confirmed that trust is a complex construct. We tried to combine findings from betrayal aversion theory, social norm, and moral norm theory, as well as neurobiological findings on hormones. Future studies should investigate the integration of different frameworks, games and disciplines.

Section 4: A critical synthesis

The challenges of trust and trustworthiness?

Trust and betrayal are among the most important factors in human societies (Baumgartner et al., 2008) and an increasing number of scholars have begun to investigate the topic in social psychology (Simpson, 2007), economics (Zak & Fakhar, 2006; Zak & Knack, 2001; Jen et al., 2010), neuroscience (Insel, 2010; Rilling et al., 2012) and biology (Ebstein et al., 2010, Walum et al., 2008, Riedl et al., 2017).

Trust is a construct that has been examined in many scientific disciplines as trust seems to be of great importance to understanding individuals, groups, societies and even nations. Although researchers are beginning to address questions with an interdisciplinary approach, we

(31)

would like to discuss two leading challenges in recent literature. These involve 1) measurements of trust and trustworthiness, and 2) possible confounds of different factors.

First, this review presented many behavioural and neuroimaging results based on two dominate measurements of trust: the trust game and survey questions. Sometimes comparisons report inconsistent results. It is possible that the number of participants, general experimental procedure, design variations, and statistics used might be explain this inconsistency.

On the other side, participants in the trust game have real incentives (monetary as well as emotional) to act according to their true beliefs and expectations. It is well-known phenomenon in behavioural studies that people tend to self-enhance and give more (politically) corrected answers to survey questions. Although it was out of the scope of the current review to study how each of those changes affect behaviour, we suggest that experimental procedures (especially in the studies involving hormonal administration) should have the awareness of multiple sources of factors. Combining these results across different frameworks was puzzling and standardizing the procedure would give us the opportunity to combine different parameter, while all other things being equal. Next, we believe that it would be interesting to measure trust and trustworthy behaviour in a binary version of trust game, where a participant plays both roles, trustor and trustee. We believe such binary version of the trust game could miror real social exchanges (i.e. in many everyday situations we are trustors and trustees in no or short temporal span).

Secondly, we would like to be critical to the regression framework, where trust evolves as a response to many different factors. These factors could co-determine the evolution of these factors. In other words, trust might evolve both as a dependent and explanatory variable. Hofman and colleagues (2017) give an example of this. They found that people with higher incomes exhibit more trusting behaviour. However, it is unknown whether people with higher incomes really trust more, or they are just simultaneously rewarded by rising incomes. Not controlling for such factors might present the omitted variables problem.

Thirdly, we facilitate that triangulation between combining behaviour in multiple games (investigating trust, cooperation, altruism, inequality-aversion and risk preferences), with

(32)

different theories (social, behavioural, norm and emotion theories) is needed in order to better understand the cognitive mechanisms involved in trust decision making.

The role of uncertainty: the difference between risk and trust are due

betrayal aversion and deactivation of fair circuitry (OXT)

Imagine your manager, supervisor, colleague or relative ask you to cross the arms, relax and close the eyes for a fall backward into another’s arms. Although this is a well-known group therapy exercise –the interesting question is, whether it is revealing if we are willing to take risk or to trust someone. In this example there is some sort of interaction with different degrees of uncertainty about the consequences of one’s own action and impact of this action on other’s decision. A large body of work focus on trust (McCabe et al., 2001; Riedl & Javor, 2012) or risk (Huettel et al., 2006; Christopoulos et al., 2009; Wu et al., 2012), but most important there are behavioural and neural studies which suggest that this phenomenon are distinct (Kosfeld et al., 2005; De Dreu, 2011; Aimone et al., 2014).

Trust involves strategic uncertainty, and decisions are said to be ‘trusting’ when the outcome of the decision depends on the uncertain decision of another person (as in the beginning story) (Bohnet & Zeckhauser, 2004), which is connected to greater activity in lateral prefrontal cortex. Contrary, risk involve state uncertainty and decisions are said to be ‘risky’ when they are made in environments where uncertainty is caused by nature (Bohnet & Zeckhauser, 2004), and studies with fMRI suggest greater activity of posterior parietal cortex (Bach et al., 2009).

Although many works on trust from different disciplines, such as economics (Ben-Ner & Putterman, 2001), philosophy (Luhmann, 1979) or sociology (Cook & Cooper, 2003) associate trust with willingness to risk, the recent empirical evidence suggest a different mechanism.

Lauharatanarun and colleagues (2012) suggested that the source of uncertainty (social vs. non-social risk) have the role of different activation of amygdala, such as individuals who were more risk averse in social context exhibit decreased risk-related activity in the amygdala during non-social decisions, and vice versa. But not only the source of the uncertainty (non-social vs non-non-social risk), but also how this uncertainty is resolved impact our decisions whether to trust (Aimone et

(33)

al., 2014). They provided evidence from an fMRI study suggesting that the differences between trust and risk are due to betrayal aversion, which was correlated to BOLD signal changes in anterior insula. Betrayal aversion means that people dislike non-reciprocated trust and when making a decision that may result in betrayal, insular activity is higher and participants participate less in efficient social exchange. Authors (Aimone et al., 2014) suggest that betrayal aversion is due a desire to avoid negative emotions, which are learned when our trust is betrayed. With other words, the important gap between trusting and risky decisions is due emotion regulation (Engelmann & Fehr, 2017).

One of the important brain area which is correlated with emotion regulation is amygdala. Trust seems to be a complex construct that is correlated with many different brain areas and many hormones. However, studies showed (Dimoka, 2010; Winston et al., 2002) that distrust is mainly correlated with fair circuitry, especially amygdala. Hsu and colleagues also showed that more risk is also correlated with higher activation of amygdala. One of the hormones which was proposed to be correlated with distrust is arginine vasopressin (Riedl & Javor, 2012). Therefore, could we gain better understanding with studying AVP?

The different role of OXT and AVP

Recent research in the field of neuroeconomics, by Baumgartner and colleagues (2008) showed that oxytocin plays a significant role in trust adaptation, which has been associated with a specific reduction in activation in the amygdala, the midbrain regions and dorsal striatum in male subjects. To our knowledge no studies has investigated the effect of intranasal vasopressin in trust and trust-adaptation and neural circuitry remains unaddressed in both, males and females’ subjects (Insel, 2010; Baribeau & Anagnostou, 2015; Riedl & Javor, 2012).

Previous findings from neuroimaging studies (Rilling et al., 2012) in males led us to hypothesize that brain regions involved in vasopressin circuitry - such as BNST, lateral septum and stria terminals will have lower BOLD signal in females, as evidence also from knockout mice shows that AVP is more influential in males (Bielsky et al., 2004). Indeed, subsequent studies using fMRI also found the important role of sex in response to AVP administration;

(34)

bilateral insula and right supramarginal gyrus activity was increased in men, but decreased in women, during reciprocated cooperation (Feng et al., 2015). However, while vasopressin receptors have been detected in insula, amygdala and cerebral cortex in rodents, current data are not so consistent in primates and humans (Baribeau & Anagnostou, 2015). Far fewer studies have been conducted at looking at AVP administration than in OX intranasal administration, although specific alleles of AVPR1a gene have been linked to autism spectrum disorder (Wassink et al., 2004). Therefore, our research will provide important informations about AVP and its neuronal circuitry in trust and distrust for females and males. Finally, Uzefovsky and colleagues (2012) showed that AVP effect on men may be specific to male social stimuli and thus our participants will interact in trust game with same-sex partner.

The possible research questions are to the best of our knowledge unaddressed: 1) Are people with higher base levels of AVP less likely to trust and more likely to distrust? 2) Are people with higher base levels of AVP less likely to risk? 3) Will investors average transfer in the risk and trust game be different across the pre feedback and the post feedback phase? 4) Is the effect of AVP sexually different? 5) What are the drug effects of intranasal AVP on brain activation in men and in women? 6) Will estradiol have the effect on behaviour measures of AVP? 7) What are the concentration of AVP in blood, using radioimmunoassay technique, before and after the task? 8) Are there any nonspecific effects that might be associated with AVP administration, such as mood, calmness, and wakefulness before substance administration and after the end of the scanning?

AVP have anxiogenic effect (Heinrichs et al., 2009) and previous studies suggest AVP play a role in inter-male aggressive communication (Thomson et al., 2004; Thomson et al., 2006) as well it increase male-male agonism and decrease cooperation in player 2 in Prisoner dilemma game (Rilling et al., 2012). A data from syndrome of central diabetes insipidus, which involves a deficiency of AVP or abnormal V2 receptors shows that central diabetes insipidus is associated with social deficits and subtle memory problems (Bruins et al., 2006). In light of previous research, we predict that AVP administration would be associated with decreased rates of trust. In contrast to oxytocin, AVP is thought to exacerbate anxiety associated with defensive responses (Young & Wang, 2004; Thompson et al., 2006; Patel et al., 2015). Accordingly, AVP