In relation to people, the question of trust is different and more complicated than reliance. Trust is generally regarded as something that develops within (or establishes) a relation between humans (usually called a trustor and a trustee) that has ethical aspects (or creates an ethical dimension).
Before I discuss the relation between trust and ethics, let me first say something about trust and how it is usually approached, which influences the interpretation of the definition just given.
I propose that we distinguish between a contractarian-individualist and a phenomenological-social approach to trust and its relation to ethics. In the former approach, there are ‘first’ individuals who ‘then’ create a social relation (in particular social expectations) and hence trust between them. In the latter approach, the social or the community is prior to the individual, which means that when we talk about trust in the context of a given relation between humans, it is presupposed rather than created. Here trust cannot be captured in a formula and is something given, not entirely within anyone’s control.
I take most standard accounts of trust to belong to the contractarian-individualist category. For example, in philosophy of information technology Taddeo’s overview of the debate on trust and e-trust, and indeed her own work on e-trust, is formulated within a contractarian-individualist framework (Taddeo 2009, 2010a, b).
In her overview (Taddeo 2009) she discusses Luhmann (1979) and Gambetta (1988). Luhmann’s analysis starts from the problem of co-ordination. Trust is seen as a response to that problem: it is needed to reduce complexity and uncertainty. Thus, the analysis starts from individuals and then it is questioned how the social can come into existence (if at all). Similarly, when Gambetta defines trust as ‘a particular level of the subjective probability with which an agent assesses that another agent or group of agents will perform a particular action’ (Gambetta 1988, p. 216), we find ourselves in a kind of game-theoretical setting typical of contractarian views of the social. Whether or not to trust someone else is a matter of decision and choice. The agent calculates (at least in Gambetta’s view) and cooperative relations are the result of these choices and calculations.
Taddeo’s account of e-trust differs from these views but does not question the contractarian-individualist paradigm itself. E-trust may arise between agents under certain conditions (Taddeo 2009; see also Turili et al. 2010 and Nissenbaum 2001), but the starting point of the analysis remains individual agents, who take a risk when they trust someone else, form beliefs about the other agent(s) before they decide to trust, assesses the other’s trustworthiness, etc. The starting point is the (Kantian regulative ideal of the) ‘rational agent, able to choose the best option for itself, given a specific scenario and a goal to achieve’ (Taddeo 2010a, p. 244). What matters is individual advantage and achievement of the individual goal, and then an ‘objective assessment’ is made in order to decide whether or not to enter into the relation.Footnote 3 Taddeo’s view belongs to what Ess calls ‘cognitive’ or ‘rational’ accounts of trust: trust is the product of a rational process; ‘we have reasons that justify our trusting others’ (Ess 2010, pp. 289–290). While she acknowledges that less rational agents may follow different, more emotional criteria and that the criteria must be specified case by case (Taddeo 2010a, p. 254), her main analysis takes place at a higher level of abstraction and remains centred around the goals agents want to achieve, the choices they make, and the gains and costs their choices incur.
Weckert’s view (2005) is different since it puts less emphasis on calculation and rational evaluation of beliefs and risks: he has criticized rational accounts for neglecting experiences of trust such as those of children in relation to their parents and those of friends. He has argued that we have the tendency to see other agents as if they were trustworthy and hence choose to act as if we trust and postpone evaluation. But regardless of (other) problems with his more ‘affective’ view (Ess 2010, p. 291), it stays within the lines of the contractarian-individualist approach: it is concerned with the agent’s attitude and a similar contractarian evaluation is made—only afterwards.
The social-phenomenological view I attempt to articulate here, by contrast, defines trust not as something that needs to be ‘produced’ but that is already there, in the social. It is close to Myskja’s phenomenological view (2008), which is based on a development of the pragmatic rather than the rational dimension of Kant’s philosophy, and which locates trust in the centre of our embodied and social human condition. But is this still about ethics? Is this still about trust?
This leads me to the question: How do the different approaches conceptualize the relation between trust and ethics?
Trust and ethics
In the contractarian-individualist approach, responsibility is the flip side of trust. When we say that we trust someone, we also ascribe responsibility to that person: the person must be answerable to us for what he or she does. For example, if someone asks me to keep an eye on her bag and I walk away, I must answer to that person why I did not do what she expected. Thus, trust ascription creates a deontic field: if someone trusts me, I feel under an obligation not to misuse that trust. The person relies on me. The expectation becomes normative (as opposed to merely predictive, as in the case of artefacts—although indirect trust also involves normative expectations since then we deal with people). In response to the expectation, I may make a promise Whether or not I explicitly communicate this promise,Footnote 4 trust-giving and trust-receiving involves the employment of a kind of ‘moral language’ (I say I trust someone, I make promises, I communicate my expectations, etc.,) which creates the trust relation and its deontic implications.Footnote 5
In the phenomenological-social approach, by contrast, responsibility is already built-in in the social, communitarian relation, which crucially has non-verbal and implicit aspects. Here morality is not something that is created but that is already embedded in the social. There are moral-social relations. There is a kind of basic confidence, in which reliance and reasoning are rooted and which they must presuppose.Footnote 6
Preconditions for trust
Several conditions for trust (or, more precise, trustworthiness) have been discussed in the literature on trust and e-trust, such as direct interactions and shared values (see for example Taddeo 2009). But I wish to make a different kind of claim. The previous discussion of trust in human–human relations gives us at least three conditions we must presuppose about persons that might trust one another, regardless of (other) conditions under which they are justified to trust each other:
Ability to use language, in particular the moral language of giving trust, promise making, expressing expectations etc. In the contractarian-constructivist view of trust, this is absolutely necessary in order to establish trust. But accounts of trustworthiness ‘assessment’ based on calculation miss this moral-linguistic dimension. A social-phenomenological account, by contrast, has the conceptual resources to point to both linguistic and non-linguistic preconditions of trust. Talking about trust presupposes a subject-talker, who does not talk and act as an abstract ‘rational agent’ but as an embodied and social being.
Freedom and uncertainty: the giver of trust (the trustor) must be free, since the trustor cannot be forced to trust someone. A true gift cannot be forced. Moreover, the receiver of trust (trustee) must be free, since we must suppose that the receiver has the possibility to misuse the given trust—if there is absolute certainty about what will happen then there is no point in trusting someone. This means that there has to be freedom in the sense of proper delegation and no (direct) supervision. As Turilli et al. write when summarizing Taddeo’s view: ‘The trustor does not supervise the trustee’s behaviour (…). Delegation and absence of supervision are then the defining characteristics of the occurrence of trust.’ (Turili et al. 2010, p. 340). In the contractarian-constructivist view, individual freedom is crucial. The social-phenomenological view, however, can point out that the game of giving and receiving trust is already part of a social context in which trust is less under the control of individuals than assumed by the individualist view, and is more an emergent and/or embedded property. This also allows us to take a different perspective on the uncertainty related to trust: it is not so much that I am uncertain about whether or not I (as a rational agent) will reach my goal by delegating a particular task to someone else; rather, if there is a problem of trust I am uncertain about the social relation itself. When trust is an issue, the social relation, and therefore I, am at stake as a vulnerable and embodied social being.
Social relations. From a contractarian-individualist point of view, social relations are constructed or produced by individuals and any concept or institution that is related to the social, such as trust, also has this status: it is a construction or product. From a phenomenological-social point of view, trust-talk and talk about individual freedom presupposes social relations (and embodiment). In other words, there is trust because there are already social relations. Trust is something basic that must be presupposed; it is not created but emerges from social relations.Footnote 7 Therefore, we must presuppose of persons that might trust one another that there is already a social relation, which the persons experience as embodied and vulnerable beings that stand-in-relation.
I take the latter perspective to be in line with the phenomenological and virtue ethics views of trust summarized by Ess, which pay more attention to embodied, affective and social dimensions of trust as opposed to the rational and the individual dimensions (Ess 2010).
Can the two approaches be reconciled?
One may object now that I over-emphasize the differences between the two approaches. Contractarian-individualism, or at least one version of it, could respond to the phenomenological social challenge by claiming that both approaches are not far apart since they could agree that trust is the ‘default’; only when there is a problem we switch to trust assessment. This objection rests on a particular version of the contractarian-individualist approach: the thesis is changed from ‘One trusts only when there is a good reason to’ to the different, modified thesis: ‘One trusts unless there is a good reason not to trust’. If this modification is made, then it seems that the gap between the two approaches is not as wide as I suggest, since it seems that both approaches could agree that we trust by default and that in the default mode no thought is given to trusting. However, I believe the two approaches still differ in how they describe this default trust and in how they understand mistrust.
For contractarian-individualists, default trust is a matter of individual attitude or stance, whereas the phenomenological-social approach understands trust as something that arises or emerges from the social relations in which we find ourselves. The very term ‘default’ still belongs to the contractarian-individualist vocabulary since it presupposes that there is always a choice situation. As in electronics and computing, ‘default’ refers to a pre-selected option that is followed except when changed. Used as a metaphor by contractarian-individualists, it means that the social ‘system’ may well pre-select the option ‘trust’ but that we individuals can change this if there is a good reason to do so (the thesis is that ‘one trusts unless there is a good reason not to trust’), which implies that we can always assess and re-assess, and then adapt our relation to others. But this presupposes that we have always a choice with regard to social relations and their form(ation). And if we are aware of this, then the seed of mistrust has already been planted. The phenomenological-social approach, by contrast, attends us to the possibility that we do not always and perhaps not usually have full control over giving trust or not giving trust (and indeed over our social relations and their form). Sometimes we trust in spite of good reasons not to trust, or sometimes we mistrust in spite of good reasons to trust. The phenomenological-social approach concedes that sometimes we live in the mode of ‘trust assessment’, but it stresses that the contractarian-individualist thesis ‘One trusts unless there is good reason not to trust’ describes only one way of shaping our social relations, one possibility of how we can look at human relations. Of course the new formula is already more ‘trustful’ than the initial ‘One trusts only when there is a good reason to’, but still gives the last word to human reasoning. The phenomenological-social approach accounts for the experience that sometimes we are drawn into trust or mistrust, that sometimes—and perhaps more often than we like—we cannot help (mis) trusting.
In any case, we now have an overview of different approaches to human trust and we have some preconditions for trust in humans. But what about trust in robots?