Trust: nature and dynamics

Trust is a complex notion -- with various components and dimensions-, and a multi-role relation: Trust (x y t G c); x trusts y as for action/task t useful for goal G, in context c.

It is an attitude, a disposition towards another agent (natural, technical, or social) on which our "welfare", that is, the realization of some goal of us, depends. This attitude makes us disposed to expose ourselves to failure or damage by relying on y for satisfying our goal.

This attitude towards y can be based just on feelings of safety and perceived benevolence, or on feelings due to the analogical evocation of previous or similar positive experiences; or it is more "rational", or better "reason-based", grounded on some specific beliefs, evaluations, and expectations about y, that justify our reliance. On the basis of this positive expectation and evaluation we decide to depend on y.

Thus, trust also is a decision and an act: the act of trusting y as for t, of exposing ourselves to dependence. And it also becomes a specific relation between x and y. Trust in y (on the basis of the strength of our beliefs or feelings) can be sufficient or insufficient for our decision to delegate; depending on the perceived risk and possible harm.

The evaluation of y, on which the expectation is based, has two basic components: (i) y's "competence", efficacy, expertise: "Is y really able and in condition to perform the expected 'action' and produce the desired outcome?" (ii) y's "willingness": "Will y actually perform the needed action?", "Is y predictable, reliable?", "Is y really willing to do the expected action?". Clearly these two kinds of evaluation are rather independent: y can be very well disposed but not really skilled; or y can be really able but not credible.

Moreover, trust as judgment implies the 'internal attribution' to y of skills, qualities, 'virtues'; but it also imply some evaluation about the 'external' favorable or adverse contextual conditions for y's action. This is why not necessarily y's failure entails a decreasing of y's trustworthiness; it might not be his fault, but just due to 'external' interferences.

Trust in not only 'social', addressed towards other persons; it can be also towards some process or mechanism (I can trust or not a given elevator), and technology: how much effective and good is it as for its service; how much reliable and predictable; how much accessible and friendly;...? The opposite (but complementary) side of trust is the perceived risk and the perceived unreliability or unmanageability of the technology.

Trust dynamics is a very important and complex issue, with many aspects.

On the one side, there is the problem of trust transitivity: if x trust y, and y trust z, will x trust z? Not automatically: it depends on the specific object of those trust relations. If x trust y "as good evaluator of t performances" and y trust z as for t, then x will trust z as for t.

On the other side, there is the general problem of trust transfer:

(a) If x trust y as for t, will x trust y also for another task t'? It depends: do the qualities, skills, needed for successfully performing t overlapping with the quality needed for t'? If "Yes", the trustworthiness of y as for t is a good predictor of y's trustworthiness also for t'.

(b) If x trusts y as for t, will x trust z for t? It depends on the similarity between y and z: does z have the same qualities of y necessary for t?

Another important dynamics is trust as self-fulfilling prophecy. Trust is an expectation, but this expectation can affect the expected outcome, both its probability and quality. In fact, on the one side x's positive evaluation of y can increase y's commitment, effort, self-esteem, etc. and influence the quality of y's performance. On the other side, the fact that x is or becomes dependent on y can increase y's 'benevolence' or responsibility towards x. In general, it is well known that trust can induce trust and reciprocation, while diffidence elicits diffidence.

Finally, those dynamics can be taken into account even in x's evaluation and decision to trust: perhaps x's trust in y would not be sufficient, but x predicts that his act of trusting y will increase y's reliability and performance, thus trust becomes enough and x decides to trust/rely on y.

Trust technology. Trust is a very dialectic and dynamic phenomenon, and it should acquire the same level of quality with technology. There are two different (but not independent) perspectives on that:

(i) A technology really able to support social trust relations and to create new trust dimensions among humans.

(ii) A trustworthy technology deserving and eliciting trust disposition, which is not at all just a matter of "security", like engineers currently believes.