Information Technology Reference
In-Depth Information
The Recursive Nature of Trust as Mental Representation
One might claim that in a strict sense trust just is or implies the positive
expectation
, which in
its turn is based on positive evaluations; but that (i)
the evaluations are not trust or
(ii)
they
are not implied by trust
and
they are not strictly necessary
: One might trust
Y
without specific
and explicit beliefs on his skills or reliability, just on the base of experience and learning.
10
However, this is not fully true.
As for (ii),
Y
's trustworthiness or the trust
in
Y
always imply some (perhaps vague) attribution
of internal qualities (such that the task could be achieved/realized); some 'kripta' which makes
Y
apt and reliable (Section 2.2.5).
As for (i), the evaluations of
Y
, when used as the bases and reasons for a trust expectation
and decision, are
subsumed
and
rewritten
as aspects and forms of 'trust'. Given that on the
basis of those features (
qualities
)
X
trusts
Y
, then
X
trusts those features of
Y
. 'I trust (in)
his competence', 'I trust (in) his benevolence', 'I trust (in) his morality', 'I trust (in) his
punctuality', and so on.
Let us ground this on the general theory of goals and actions. Given her goal
g
X
(
X
's aim
or end)
X
searches for some possible action
(usually of
X
) able to achieve
g
X
(where
g
X
is
included in the outcomes
p
, the post-conditions of
α
useful for
g
X
,this
action in order to be successfully performed requires some condition
C
to be true. If
C
holds the
subject can perform
α
). Given an action
α
;if
C
does not hold it becomes the new goal
g
1
X
of
X
, subordinated and
instrumental to
g
X
: a sub-goal. This obviously is the essential abstract principle of planning.
Now, given that
X
has the goal of realizing
g
X
, and that she is dependent on
Y
and needs
an action
α
α
Y
of
Y
, she has the sub-goal that
Y
successfully performs
α
Y
. However, certain
conditions are needed for both (i)
Y
successfully performing
α
Y
; (ii)
X
can decide to count on
this and counts on this.
Y
's valuation is precisely the epistemic activity aimed at those conditions; and the same
hold for
X
's predictions about
Y
doing or not
α
Y
. As we have underlined, evaluations are
about goals (something is or isn't 'good
for
'), and predictions in trust are in fact not simple
'predictions' (beliefs about the future) but more rich 'expectations' (involving goals). Actually,
since
X
wishes to achieve
g
X
through
Y
's action, and has the goal that
Y
be able, in condition,
and predictable in performing
α
Y
, all these necessary conditions (also those 'internal' to
Y
)
are new (sub)goals for
X
:
X wishes
that
Y
is skilled enough, competent enough, not hostile or
dangerous, willing and reliable, and so on. Relative to those sub-goals she evaluates
Y
and has
(or not) trust in
Y
. She trusts
Yfor
being competent,
for
being persistent, etc.
So trust in
Y
as for action
α
Y
for goal
g
X
, (at least implicitly - but frequently explicitly)
entails sub-trust supporting the broad trust about the action; in a recursive way. Since any
new goal about
Y
might have its sub-conditions and needed sub-qualities on which
X
relies;
thus potential sub-trusts are generated. For example,
X
can trust
Y
for being really willing to
cooperate with her, because she knows and
trusts in Y
's friendship, or because she knows and
trusts in Y
's generosity and empathy, or because
Y
is morally obliged and she knows and
trusts
in Y
's morality.
This is not only the real use of the word and the commonsense on trust; it is a logical and
principled use, as we have just explained.
10
We thank Fabio Paglieri for this observation. This is also one of Andrew Jones' criticisms to our model (Jones,
2002); see Section 2.2.3 for our reply.
Search WWH ::
Custom Search