Traditional theories of corruption often make assumptions about motivations that may not necessarily be valid. We explored the power of an alternative theoretical paradigm to explain corrupt behaviour: cognitive psychology. We found evidence in the existing literature on the cognitive psychology of corruption about the psychological influence of power, personal gain and self-control, loss aversion and risk acceptance, rationalisation, and emotion on the propensity to act corruptly.
- Individuals holding power are more likely to act corruptly.
- Individuals are more likely to act corruptly when they stand to gain personally, have lower self-control, perceive that corruption will only cause indirect harm, and when they work in organisations where unethical behaviour is not punished.
- Individuals are likely to be more risk-acceptant to offset losses, and risk-averse to preserve gains. Uncertainty is likely to increase the likelihood of acting corruptly.
- Rationalisation narratives seem to make corruption more acceptable.
- Emotions such as guilt may make it less likely for individuals to act corruptly.
- To mitigate these cognitive influences, practitioners should support measures that improve information flows about the costs of corruption, that reward ethical behaviour and set basic integrity standards, and that improve organisational decision-making.
- More research is required on how, which, and when particular cognitive psychological mechanisms make corruption more or less likely; the social psychology of corruption, and how social and cognitive psychologies interplay; the psychological effects of corruption on individuals; as well as case studies of political elites.
Rational choice-inspired anti-corruption policy has failed
The basic question “what explains corrupt behaviour” has long plagued scholars and practitioners (Nye 1967; Bardhan 1997). It has recently received renewed attention in policy and academic circles because, despite the rise and spread of the global anti-corruption movement, many highly corrupt countries have made little progress on reducing corruption (Heywood 2017; Rose-Ackermann and Palifka 2016; Marquette and Pfeiffer 2015; Persson, Rothstein, and Teorell 2013; Mccoy 2001).93882daa1f59
Millions of dollars of foreign aid have been channelled to aid-receiving countries to implement a variety of institutional measures to reduce corruption, with less to show for these efforts than hoped.7e7802d6730f 1acc8fc5875b Reducing corruption in developing countries is an urgent task, especially given its negative impacts on economic growth, trust in government, the quality of public service provision, environmental outcomes, foreign investment and trade, and levels of inequality (Menocal et al 2015; Dimant and Tosato 2017).
Rational choice theory has long dominated the academic study of corruption, anti-corruption practice and policy approaches. It explains corruption as the function of calculating, strategic, self-interested behaviour. In this view, corruption is particularly likely to occur in situations of power asymmetry, where some individuals (agents) hold power over others (principals). Yet, rational choice explanations make assumptions about motivations that may not be valid. Psychology, political psychology, and behavioural economics in particular, have posed serious challenges to rational choice theories of human behaviour.
The political psychologist Jon Mercer (2005) argues that rational choice theories “explain how one should reason, not how one actually reasons” (80), and they do not adequately explain how people “make decisions to reach an outcome” (81). In Thinking, Fast and Slow, Daniel Kahneman (2011) writes, “[t]he definition of rationality as [logical] coherence is impossibly restrictive; it demands adherence to rules of logic that a finite mind is not able to implement” (411). Instead, as Kahneman’s work shows, human beings are susceptible to a variety of cognitive biases that shape their decision-making and behaviour in ways that do not conform to the predictions of rational choice and functionalist approaches. (Explore functionalist and rational choice theories of corruption.)
The failure of rational choice-inspired anti-corruption policy and programme measures has inspired corruption studies scholars and practitioners to look for alternative approaches to understand the individual-level drivers of corrupt behaviour. In particular, there has been a turn towards exploring the fields of psychology and behavioural economics.
To date, corruption scholars and practitioners working on the psychology of corruption have focused primarily on the social psychological determinants of unethical behaviour, such as the influence of group norms, interactions, and dynamics (c.f. Zaloznaya 2017; Hoffmann and Patel 2017; Camargo 2017; Köbis et al 2015; Weisel and Shalvi 2015 Bicchieri and Ganegonda 2016). This is well justified, given the social, interactive nature of corruption, but it neglects the the individual-level mental processes, such as decision-making and information processing that drive choices to act unethically, or the psychological determinants of unethical behaviour.
To address this gap, we reviewed literature on the cognitive psychology of corruption to synthesise and evaluate current knowledge. We also aimed to draw out recommendations for how practitioners might incorporate a cognitive psychology lens in their anti-corruption programmes. Using a pre-established list of search terms, we combed through literature published in the fields of psychology, political science, political psychology, economics, business, and organisational studies. We organised our literature review according to the topics that are most commonly featured in the source. (Explore the Methodology of this study).
The existing evidence base on the cognitive psychology of corruption
Cognitive psychology tells us that to understand individual decision-making processes (including decisions about acting corruptly) we must look at factors that influence information processing (such as time, mental capacity, and motivation). Further, we must examine how individuals mentally interpret and organise information by using schemata, the salience of emotions and the importance of social context. (Explore cognitive psychology concepts for understanding corrupt behaviour).
We found evidence about these concepts in the existing literature on the cognitive psychology of corruption. Specifically, we found several studies on the influence of social status (power) and risk perceptions; a smaller number of studies about the role of personal gain and self-control; and very few studies regarding the effects of emotions and rationalisation narratives.
Summary of the existing evidence base and possible implications for practitioners
Individuals holding power are more likely to act corruptly (13 studies).
Possible implications for practitioners:
- Support information and sanctioning mechanisms targeted at power holders to prevent and punish corrupt acts.
- Support integrity measures for power holders.
- Ensure transparent and accountable decision-making processes to check power.
- Provide clear definitions of corrupt actions to prevent ethical sliding.
Personal gain and self-control:
Individuals are more likely to act corruptly when they stand to gain personally, have lower self-control, perceive that corruption will only cause indirect harm, and when they work in organisations where unethical behaviour goes unpunished (6 studies).
Possible implications for practitioners:
- Support institutional mechanisms that limit the ability of power-holders to gain materially from their position.
- Support integrity measures for power holders, including reward systems.
Loss aversion and risk acceptance:
Individuals are likely to be more risk-acceptant to offset losses, and risk-averse to preserve gains. Uncertainty is likely to increase the likelihood of acting corruptly (8 studies)
Possible implications for practitioners:
- Support information provision about the negative outcomes of corrupt behaviour to reduce uncertainty and mitigate risk-acceptant behaviour designed to offset potential losses.
- Pay attention to how information and situations are framed, and avoid negative framing.
Rationalisation narratives seem to make corruption more acceptable (4 studies).
Possible implications for practitioners:
- Provide clear definitions of corrupt actions (and enforce them) to prevent rationalisations and normalisation.
Emotions such as guilt may make it less likely for individuals to act corruptly (2 studies).
Possible implications for practitioners:
- Support information provision about the negative outcomes of corrupt behaviour to encourage a sense of moral responsibility
Power is fundamental to the study of corruption. It is a necessary condition of Klitgaard’s widely adopted definition11489754be73 and of rational choice approaches to the study of corruption. In this context, power is defined as certain individuals holding accountable degrees of power or authority over decision-making processes, creating lucrative windows of opportunity for unethical behaviour. According to this understanding, without power, there is no corruption. As a result, power has become one of the main areas of focus in understanding individual decisions to act corruptly.
Starting with Kipnis (1972), psychologists have researched the effects of holding power on individuals’ moral behaviour, the likelihood of misusing power, and views of the self and others. In relation to corruption, several scholars have found evidence of a negative relationship between power and corruption. This may be because power-holders tend to:
- be more risk acceptant
- seek rewards
- experience less guilt and embarrassment
- feel less empathy for others
- act more out of self-interest
(Wang and Sun 2016; see also Lee-Chai and Bargh 2002).
Wang and Sun (2016) carried out experiments with students and working adults in China to examine how individuals’ view of the type of power they hold – either “personalised” or “socialised” – shapes attitudes towards corruption and corrupt behaviour. The authors define the power concept as “the belief about the goals one should accomplish while using one’s power” (78). People who view power as personalised believe that power should be used to pursue self-centred goals for one’s own benefit. A socialised power view means that the power holder believes power should be used to pursue other-focused goals. The authors find that having a personalised power view increases self-interested behaviour and tolerance towards corruption, especially the corrupt behaviour of high-ranking individuals.
Power can result in overconfidence, greater risk acceptance, and a focus on rewards (Yap 2013). In five separate experiments, Fast et al (2012) show that power leads to overconfidence in decision-making, and ultimately results in poor decisions that create material losses for the power-holder. Rusch (2016) argues that the overconfidence effect may lead to corruption in organisations when subjective confidence in judgments overpowers objective accuracy. This can warp self-perception of ethical issues when there is a significant gap between how people believe they will behave in a given situation versus how they actually behave. In other words, the overconfidence that individuals have in their own morality may prevent those same individuals from realising or admitting that they are acting corruptly (Darley 2005).a01ee9c57016
The situation, rather than the person, is blamed for the bad behaviour in a form of cognitive bias known as “self-serving bias” in which decision-makers process information in ways that support pre-existing views and advance their self-interests (Prentice 2007). Confirmation bias is the inclination to disregard information that conflicts with people’s beliefs and views of the world, whereas, the halo effect is the tendency for initial impressions to influence overall perceptions and beliefs about a person or organisation (Kahneman 2011).
Confirmation bias, together with the halo effect, might also make it more likely that corrupt actions by individuals, who are otherwise considered good persons, will be overlooked or ignored.3628cfd5f4c0 Individuals may be tempted to engage in hindsight bias, viewing their past actions as better thought out and intentioned than they actually were (Darley 2005). Individuals may also assume more personal responsibility for a success and less for a failure, known as causal attribution bias, (Arkin et al 1980).
The actions of the powerless may motivate power holders to act corruptly, either for reasons of obedience or because of expected rewards. On the former, Tepper (2010) and Prentice (2007) both recount the infamous Stanley Milgram experiments, in which a large majority of study participants blindly obeyed an authority’s instructions to deliver increasingly painful shocks to individuals for failing to correctly answer a question. To explain this finding, Prentice (2007) notes that “pleasing authority usually leads to rewards; displeasing authority often gives rise to penalties, including loss of employment” (18). Therefore, it should be no surprise that the powerless fail to confront or stop corrupt behaviour by powerful individuals.
Darley (2005) offers alternative explanations for obedience. First, corrupt acts may be motivated by intuition rather than reason and therefore may initially be done unintentionally and not be viewed as corrupt. As Darley (2005) argues, “many of the actions that begin cycles of corruption are the products of the intuitive judgment system, which means that they are rapidly arrived at, less than consciously considered, and unintentional in their ethical dubiousness. Further, they are often the product of pressure to make fast decisions” (1183). Second, corrupt acts may stem from previous actions that were not viewed as unethical, or were in an ethical grey area. People may accept initial actions as being ethical, while the next action may be less ethical but hard to distinguish from the first action – a type of “slippery slope” (Köbis et al 2017). Past practices form the benchmark by which individuals judge future actions. Over time, organisations can experience a form of ethical sliding or fading that may be difficult to recognise and stop, particularly as people tend to prefer the status quo (Rusch 2016; Reckers and Samuelson 2016; Prentice 2007; Bazerman and Tenbrunsel 2012). However, Köbis et al 2017, suggest that people are willing to engage in severe corruption without first engaging in mild corruption. Moreover, perceptions can be misleading, especially if individuals trust that what they see must be true (see also Kahneman 2011).
Manzetti and Wilson’s (2007) study touches on the latter mechanism in the relationship between the powerless and powerholders. Using World Values survey data, they show that when the powerless expect rewards from authorities, they are likely to overlook political elites’ corrupt behaviour. In countries with weak government institutions and strong patron-client relationships, citizens support corrupt governments because they expect to receive tangible benefits from corrupt leaders.
Personal gain and self-control
Studies on the effects of power show that power holders pursue actions that benefit themselves. In a field experiment on lying, Djawadi and Fahr (2015) show that individuals are highly willing to cheat (lie) for personal gain. Extending the logic of their findings to professional organisations, the authors argue that people are willing to lie even when the gains from lying are uncertain. They also find that lying is more likely when any potential harm is indirect and found in written statements or actions, rather than direct (face-to-face). The willingness to lie in the face of uncertain gains is potentially even greater in organisations where unethical behaviour is encouraged, routine, or even rewarded (Ashforth and Anand 2003).
What might deter individuals from lying, cheating, or engaging in other similar behaviour? In an experiment, Trevino and Youngblood (1990) show that individuals with a greater locus of self-control and better cognitive moral development were more likely to make ethical business decisions. Locus of control “measures an individual’s perception of how much control he or she exerts over events in life” (ibid, 379), with research showing that individuals with a greater sense of control are more likely to do the right thing. Cognitive moral development measures the degree to which an individual’s moral judgment depends on external influences, with less dependence leading to greater willingness to resist pressure from authorities and to confront wrong doing. In a laboratory experiment, Reckers and Samuelson (2016) also find evidence that individuals with a stronger locus of control, and a stronger sense of moral responsibility, are less likely to comply with unethical management requests over time. These findings support Tepper’s (2010) recommendation that people should be provided with skills to confront individuals behaving corruptly.
Dimant’s (2015) review of psychological theories and their application to explaining corruption discusses self-control, but reaches the opposite implication of Trevino and Youngblood (1990). The theory of planned behaviour points to the importance of individual perceptions of the degree to which they control certain behaviours. Personal confidence to behave in a certain way influences the likelihood that a behaviour (such as a corrupt act) will occur.
Loss aversion and risk acceptance
Corruption is a potentially risky behaviour, particularly if the likelihood of being caught and punished is high. Attempts to understand corruption have led scholars to investigate how risk perceptions influence an individual’s propensity to behave corruptly. There is some evidence that perceptions of risk may lead women to behave differently than men when it comes to corruption, but this is far from conclusive.
Menocal et al (2015) review two sets of studies on the nexus between gender, corruption, and psychology. First, psychological studies provide evidence that women are more risk-averse than men, and are less willing than men to accept bribes. Women may also have different moral standards to men, and there is evidence that they are more likely to punish selfish behaviour in laboratory experiments than male participants, although the evidence on this is mixed.
Guerrero and Rodríguez-Oreggia (2008) also cast doubt on a gendered psychological explanation for corruption, arguing that any gendered difference in corrupt behaviour is because men find themselves in more corruption prone interactions (such as with the police) than women. However, women more often interact with public services like education and health where bribery and other forms of corruption are prevalent (especially in very poor countries), and Guerrero and Rodríguez-Oreggia’s findings may thus be very context-dependent.
Yap’s (2013) discussion of the psychological literature on power shows that power-holders are generally more likely to take risks. Yet people are generally risk averse and prefer sure wins to uncertain outcomes. If corruption is viewed as risky, due to the likelihood of being caught and punished, then willingness to act corruptly should be low. Why, then, might people still choose to act corruptly, particularly when the consequences are potentially very negative? One explanation may be the desire to prevent or mitigate losses.
Kahneman and Tversky’s (1979; see also Kahneman 2011) work on prospect theory provides evidence that attitudes towards risk depend on the likelihood of gaining or losing in a particular situation. Prospect theory states that individuals are likely to be risk-averse in situations involving sure gains but risk-acceptant, or even seeking, in situations involving sure losses. Human beings are generally loss-averse (while also risk averse) as losses are much more painful than gains are pleasurable. To avoid the pain of loss, people will undertake risky actions, whereas to preserve gains, people will forgo taking risks. An extension of this logic is that when individuals are in difficult situations, they make unethical choices to prevent what they perceive as a bad outcome or loss, such as the loss of a job or position, a large financial loss for an organisation, or a government losing power. For instance, Søreide (2009) finds that risk acceptance can actually increase firms’ propensity to offer bribes, such as when not behaving corruptly will ruin a firm’s chance of operating in a given market (a possible domain of loss) or when a company can borrow money to pay bribes.
So far, there has been very little empirical work that tests the degree that prospect theory influences the likelihood of acting corruptly. More work is needed to better understand how individuals understand and act on perceptions of risk, and how this influences the likelihood of acting corruptly. One exception is Ron Everett Prescott’s (2012) PhD dissertation, which shows that moral decision-making under uncertain conditions is similar to that predicted by prospect theory for economic risk decisions. He finds that people are likely to make risk-averse ethical decisions in positively framed situations and risk-acceptant decisions in negatively framed situations. Further research is needed to better understand how individuals comprehend risk and in turn how risk perceptions shape behaviour.
Based on evidence from a laboratory experiment, Jacquemet et al (2008) show that the optimism bias skews people’s judgments about the probability of experiencing a positive event in the future. People assume that unpleasant things are more likely to happen to other people than to themselves. Applying this to corruption, the authors find that people who engage in illegal activities are also likely to think they will be able to avoid monitoring, detection and ultimately punishment. This could help to explain why people choose to act corruptly even when the objective probability of being caught is high.2f0d8d58349a
Berninghaus et al (2013) challenge the argument that attitudes towards risk shape corruption. Based on evidence from a set of game experiments, these authors find that attitudes about risk do not explain the choice between corrupt and non-corrupt behaviour. Rather, beliefs about probability do. In an experiment, risk-seeking participants were not necessarily more corrupt; rather, what motivated an individual to act corruptly was the subjective estimation (beliefs) of the probability of a bribe being accepted. A higher degree of uncertainty (less information) about probable actions reduced corruption. A real-world application of this is the spread of public information about convictions for corruption to reduce beliefs about the probability of bribe acceptance.
People rationalise or justify future or past behaviours to make them acceptable to themselves and others. This is also true of corrupt behaviour. Benson (2015) labels the study of rationalisation as “neutralisation theory.” Choo and Tan (2007) argue that in the context of the United States, the cultural narrative of the “American Dream” strongly influences the worldview of corporate executives, and makes them likely to engage in illegal behaviour. The “American Dream” emphasises monetary success as the ultimate symbol of individual success, therefore, when rules and procedures interfere with achieving this goal, executives are tempted to use illicit means to achieve success. This is made worse in organisational cultures that justify achieving success by any means and allow executives to disregard controls.
Ashforth and Anand (2003) argue that one of the ways that corruption is normalised in organisations is through rationalisation, “the process by which individuals who engage in corrupt acts use socially constructed accounts to legitimate the acts in their own eyes” (3) (see also Campbell and Göritz 2013). Rationalisations for corrupt behaviours attempt to account for breaking social norms against unethical behaviour and are a means of avoiding judgment for an ethical breach, a form of self-defence. Rationalisations can also lead to ethical fading (Reckers and Samuelson 2016). While rationalisations are useful for individuals to justify their behaviour towards themselves, in the context of organisational corruption, the authors argue that their acceptance within the larger group is most important .
Emotion is an important but under-studied component and driver of beliefs, perceptions, decision-making, and actions (Mercer 2010; Kahneman 2011). This includes corrupt behaviour (Köbis et al 2016). Abraham and Pane (2014) examine how so-called “moral” emotions influence corrupt behaviour by measuring the emotional responses of students in Indonesia to situational vignettes. The authors find that a strong sense of responsibility for the welfare of the group (collectivism) correlates with feelings of guilt. It is often assumed that guilt reduces the tendency to act corruptly, interestingly, the authors find no impact of collectivism on shame proneness. Guerrero and Rodríguez-Oreggia (2008) also find that guilt plays a strong role in decisions to act corruptly. In qualitative focus group interviews in Mexico City, interviewees did not report a sense of guilt related to bribing, since offering bribes was a practical means of getting things done with a low risk of punishment.
Implications for practitioners
Practitioners can take several steps to address the cognitive psychological mechanisms that may influence individuals to act corruptly. Some mechanisms have already been integrated into traditional approaches to anti-corruption programming, and they should be continued. These include supporting the implementation of measures within their own agencies, as well as in the organisations to which they provide assistance. Measures include information, integrity and accountability, rewards for ethical behaviour, and decision-making. While the suggested measures are intended to shift the cognitive psychological mechanisms that can motivate individuals to act corruptly, it should be recognised that they will be implemented in social environments, and that social interactions and dynamics are important dimensions of changing motivations.
Individuals are more likely to act corruptly when they lack good information about possible sanctions for bad behaviour. Thus, practitioners, such as donor agencies, should continue to support and expand measures that improve information flows about the individual and social costs of corruption, particularly about punishment for corruption. This can help to deter individuals from acting corruptly via a pure logic of consequence and can discourage the temptation to rationalise corrupt acts. To induce positive behavioural change and avoid enhanced risk acceptance, careful consideration of how to frame information is needed. Finally, clear definitions of corruption should be communicated repeatedly and consistently within organisations to avoid ethical backsliding and grey areas that facilitate corruption, as well as rationalisations and justifications for corrupt behaviour.
One of the strongest findings from our review is that holding power seems to change cognitive processes in ways that make people more likely to behave unethically. Thus, practitioners can support the creation and implementation of integrity measures like codes of conduct, which can help to remind power-holders of their duty to act cleanly and create clear standards for ethical behaviour.
Accountability mechanisms are needed to hold power-holders to account for abuses of power, and to help prevent abuses from happening at all. This should include measures for preventing power-holders from using their position to accumulate material wealth for their own personal benefit.
Rewards for ethical behaviour
Prospect theory teaches us important lessons about when and why individuals are likely to engage in risky behaviours, like corruption. Anti-corruption campaigns that seek to punish wrongdoers can paradoxically heighten incentives to act corruptly if individuals perceive that such campaigns will potentially place them in a domain of loss. This could heighten the likelihood of individuals acting corruptly to prevent further losses (such as loss of position or access to resources). Instead, the success of punitive anti-corruption campaigns may hinge on the sequencing of reforms (preparing elites for losses and allowing them time to cut their losses) and on rewarding people for behaving ethically rather than (only) punishing unethical behaviour. However, the design policies that intend to “prove” ethical behaviour must be carefully considered to avoid providing incentives for individuals to hide deviant behaviour (as has occurred with “zero-tolerance” policies towards corruption).
Timing of anti-corruption work
Practitioners should pay careful attention to the timing and sequencing of anti-corruption measures directed towards society in general and anti-corruption measures directed towards political elites. Research on basic cognitive psychology shows that there is a strong bi-directional relationship between cognition and behaviour, as well as a decisive interconnection between the individual and their environment. These dynamics imply that if not timed well, anti-corruption campaigns directed towards the society and campaigns directed towards political elites might cancel each other out. Society-wide and elite-focused anti-corruption campaigns are more likely to be successful if they are carried out in tandem rather than in isolation.
Finally, decision-making processes within organisations should be as transparent and accountable as possible to reduce the temptation for power-holders to abuse their power. If power-holders believe that they will not be held accountable for their actions, they are more likely to be risk-acceptant and act corruptly. Providing sufficient time for decision-making may also help to reduce the lure of cutting corners.
Need for stronger conclusions through research
The study of the psychological drivers of corruption is an emerging field, and as such, more research is required on several fronts. First, the existing evidence base about the cognitive psychology of corruption is generally quite thin, and more research is required to draw strong conclusions about how, which, and when particular cognitive psychological mechanisms make corruption more or less likely.
Second, the discussion in the previous section highlighted that individuals do not act in isolation; rather, their actions and cognitive psychology are shaped by the social world. More research should be conducted to better understand the social psychology of corruption, and how social and cognitive psychologies interplay. This should include a focus on the development of corruption norms and how they can be changed, as well as how group dynamics and interactions, social identity, trust, culture, and other dynamics influence individual propensity to act corruptly.
More research on how social norming and nudging (including anti-corruption messaging) can help to change beliefs and preferences would be useful, as would a better understanding of the influence of context in heightening the importance of certain messages and nudging mechanisms.
Finally, two additional avenues of research could provide fruitful insights into the psychology of corruption. First, understanding the psychological effects of corruption on individuals could help to illuminate why systemic corruption in particular can be hard to break; such a focus could draw on existing literatures about trust in government and on individuals’ ability to change their behaviours and update prior beliefs.
Second, case studies of political elites would be useful, as elites hold high amounts of power and their actions affect many people. More research is required to better understand how political elites perceive the costs and benefits of acting corruptly, how and why they rationalise such behaviour, how they process information to reach decisions about acting corruptly (including the role that analogies and heuristics play), how individuals learn from prior experiences, and how individuals factor time horizons into risk-acceptant behaviours.
- For instance, the bottom 40 countries listed on the 2016 Corruption Perceptions Index have made little progress on improving perceptions of corruption over the past five years, with perceptions of several countries even worsening.
- The World Bank estimates that businesses and individuals pay USD $1.5 trillion in bribes each year.
- According to the OECD International Development Statistics online databases, DAC, non-DAC, and multilateral agency donors have spent millions of dollars per year since 2002 on anti-corruption programmes.
- Klitgaard's (1988) corruption formula:
C = M + D – A
Corruption = monopoly + discretion – accountability
- See for example You’re not as virtuous as you think
- The opposite of confirmation bias is “attribution error”, wherein individual characteristics are blamed for negative actions rather than situations.
- See more in Kahneman 2011 on how cognitive biases influence judgment and assessment of probabilities.
This work is licenced under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International licence (CC BY-NC-ND 4.0)