Trust, Control and Autonomy: A Dialectic Relationship - Trust Theory: A Socio-Cognitive and Computational Model - page 191

External
(world)
factors

Expected
results of
Control

Core Trust x y τ

Strict Trust notion

Broad Trust notion

Figure 7.3 Control complements strict trust

(that I expect for its result g ). But the problem is: are such an ability and willingness (the
'internal' factors) enough for realizing g ? What about conditions for successfully executing α

(i.e. the opportunities)? What about other concurrent causes (forces, actions, causal process
consequent to Y 's action)? If my trust is enough for delegating to Y , this means that I expect
(trust) that g will probably be realized.
We propose a broader notion of trust including all my expectations (about Y and the world;
including actions of other agents, and including possible control activity on Y ) such that g will
be eventually true thanks (also) to Y 's action; and a strict notion of trust as 'trust in' Y , relative
only to the internal factors (see Figure 7.3).
This strict notion is similar to that defined by Mayer (apart from the lack of the competence
ingredient), and it is in contrast, in conflict with the notion of control. If there is control then
there is no trust. But on the other hand they are also two complementary parts, as for the
broad/global trust: control supplements trust. 7
In this model, trust in Y and control of Y are antagonistic : where there is trust there is no
control, and vice versa; the larger the trust the less room for control, and vice versa; but they
are also supplementary : one remedies to the lack of the other; they are parts of one and the
same entity. What is this attitude that can either be built out of trust or out of control? It is
confidence, i.e. trust again, but in a broader sense, as we formalized it. 8
In our view we need these two levels and notions of trust. With this in mind, notice that
control is both antagonist to (one form of trust: the strict one) and consituent to (another form
of trust: the broader one). Obviously, this schema is very simplistic and just intuitive. We will
make this idea more precise. However, let us note immediately that this is not the only relation
between strict-trust and control. Control is not only aimed at supplementing and 'completing'
trust (when trust in Y would not be enough); it can also be aimed precisely at augmenting the
internal trust in Y , Y 's trustworthiness.

7 Control - especially in collaboration - cannot be completely eliminated and lost, and delegation and autonomy
cannot be complete. This, not only for reasons of confidence and trust, but for reasons of distribution of goals, of
knowledge, of competence, and for an effective collaboration. The trustor usually has at least to know whether and
when the goal has been realized or not (Castelfranchi and Falcone, 1994).
8 This holds for a fully delegated task. It is clear that for coordination between X and Y in a multi-agent plan, X
has to monitor Y (and vice versa) even if she trusts him a lot.