Reviewed by
Corinna Elsenbroich
Department of Sociology, University of Surrey
What do the sentences, "the computer programme will do the calculation", "my mum will always be there for me", "the agent cooperated with last round so I will cooperate now", and "England will win the next football World Cup" have in common? They all express a form of trust into another entity, a programme, a mother, another agent and the England football team. These kinds of trust are, however, subtly different. The first is purely competence based, the second is emotional trust resulting from positive attachment and experience, the third is reciprocal trust as can be found in game theory and public goods games, the last could be classified as "blind trust" resulting from wishful thinking. Although all sentences describe trust, the kind of recipient, the relationship between agents, the cause of the trust or to sum it up, the meanings of the concept "trust" are different. Trust Theory: A Socio-Cognitive and Computational Model sets out to develop a general, universal model of trust, a model that brings all these different notions of trust under one roof. It sets the scene in Chapter 1 by discussing existing theories, concluding that they are all too domain specific, only ever tackling one aspect of trust. The authors conclude that the literature lacks a general theory of trust leading to many disparate or even inconsistent notions and definitions. In Chapter 2 the authors develop their own general theory and give a quantitative analysis of it in Chapter 3. In Chapters 4-7 they discuss their theory with respect to specific features, namely distrust, emotions, dynamics of trust, and with respect to control and autonomy. In these chapters the added value of a general model becomes clear as a wide range of phenomena can be captured. In Chapters 8, 9 and 10 the general model is put into context of economic theory (trust as irrational), social order (trust as the "glue" that keeps society together) and social capital (which here becomes relational capital between trustees and trustors). Chapters 11 and 12 are extensions of the general model, one using fuzziness and the other giving a general idea for the use of the socio-cognitive model of trust for implementation into Distributed AI, Multi-Agent Systems, Human Computer Interaction and other technological research fields. Finally Chapter 13 concludes on future issues of trust in neuroscience, politics and in particular the political situation of Italy, with surprisingly high trust in Berlusconi's government.
The book is not an easy read or introductory text, although parts could be used for teaching model construction and trust. That it is rather difficult to read is partly due to what it sets out to do, partly to the way it is done. Just like other classic works in philosophy, developing a general theory that spans all domains demands a level of abstraction (of the concept) as well as detail (in the definitions) that can prevent easy accessibility. The authors also choose to stay on a rather abstract level in their discussion, not relating their theory to empirical findings or real life settings, which might provide a more intuitive embedding for the abstract discussion and assist understanding.
The criticism of lacking empirical embedding is also brought forward by Herbert Gintis (see his review on Amazon at: http://www.amazon.com/Trust-Theory-Socio-Cognitive-Computational-Technology/dp/0470028750). He also points out the almost complete lack of discussion of behavioural game theory and its influence on trust research, other than as a reductive notion that does not serve the purpose. Although not necessarily convinced that behavioural game theory is the best assessment tool for a theory of trust, I agree with Gintis that more embedding in current research, in particular empirical research would have been helpful.
I am not convinced that a general, all encompassing theory of trust can be constructed and this book has not changed my mind about it. The authors do certainly manage to construct a very broad model that encompasses many salient features of trust but it is less clear how this broad model would actually be implemented and whether such a broad model would be needed for specific implementations.
That said, the book does make a very important contribution to trust research as it does serve to highlight the motley crew of concepts of trust that need to be somehow held together and points at the problematic use of trust both in everyday notions and scientific research. On this basis it will be seen whether a universal model of trust is desirable at all, whether it can be established or whether a web of definitions will be more fruitful in which different fields highlight different aspects.
I highly recommend this book to anyone who wants to delve deep into the concept of trust, in particular for use in computational applications and social simulations to strengthen awareness of complexity, complications and conundrums of trust.
Return to Contents
of this issue
© Copyright Journal of Artificial Societies and Social Simulation, 2011