Trust does not need to involve belief in the good character, vices, or morals of the other party. Persons engaged in a criminal activity usually trust each other to some extent. Also trust does not need to include an action that you and the other party are mutually engaged in. Trust is a prediction of reliance on an action, based on what a party knows about the other party. Trust is a statement about what is otherwise unknown -- for example, because it is far away, cannot be verified, or is in the future.
From this perspective, trust is a mental state, which cannot be measured directly. Confidence in the results of trusting may be measured through behavior, or alternatively, one can measure self-reported trust (with all the caveat surrounding that method). Trust may be considered a moral choice, or at least a heuristic, allowing the human to deal with complexities that outgo rationalistic reasoning. In this case, machine-human trust is meaningless, because computers have no moral sense and rely on rational computations. Any trust in a device under this characterization is computer-mediated trust of the user of the machine in the designer and creator of the device; who has implemented the rational rules into the device. Francis Fukuyama and Tyler are academics who advocate this conception of trust – as moral and not directly observable.
The strength of Coleman's definition is that it allows for discussion of trust behavior. These discussions have been particularly useful in reasoning about human-computer trust, and trust behaviors.
A critical element in studies of trust behavior is power. One who is in a position of dependence cannot be said to trust another in a moral sense, but can be defined as trusting another in the most strict behavioral sense. Trusting another party when one is compelled to do so is sometimes called reliance, to indicate that the belief in benevolence and competence may be absent, while the behaviors are present. Others refer only to coercion.
Coleman's definition does not account for the distinction between trust(worthiness) as a moral attribute and trustworthiness as mere reliability. It is Annette Baier (Ethics, 1987) who characterizes contexts of trust as structures of interaction in which moral obligations act upon the trustees.
The substantive conflict in the social sciences is whether trust is entirely internal, and only confidence is observable, or whether trust behaviors (and self reported levels of trust) can meaningfully measure trust in the absence of coercion. Note however that many languages (e.g. Dutch or German) do not distinguish between the words trust and confidence, which is complicating this issue. The distinction between trust and confidence is an unsolved issue in current trust/confidence research.
In general, trust is essential as Social institutions (governments), economies, and communities require trust to function. Therefore trust and altruism are areas of study for economists, although these concepts go beyond strict rational economics.
Increasingly much research has been done on the notion of trust and its social implications:
Kini, A., & Choobineh, J. (1998). Trust in electronic commerce:
Kelton, Kari; Fleischmann, Kenneth R. & Wallace, William A. (2008). Trust in Digital Information. JOURNAL OF THE AMERICAN SOCIETY FOR INFORMATION SCIENCE AND TECHNOLOGY, 59(3):363–374.
Kini, A., & Choobineh, J. (1998, January). Trust in electronic commerce: Definition and theoretical considerations. Paper presented at the Thirty-FirstnHawaii International Conference on System Sciences, Kohala Coast, HI.