Information Technology Reference
In-Depth Information
Given the resulting
amalgam
these representations of the future are charged of value, their
intention or content has a 'valence': it is positive, or negative.
26
More precisely, expectations
can be:
positive
(goal conformable): [(Bel
X
t
p
t
)
(Goal
X
t
p
t
)]
p
t
)
p
t
)]
∧
∨
[(Bel
X
t
¬
∧
(Goal
X
t
¬
negative
(goal opposite):): [(Bel
X
t
p
t
)
∧
(Goal
X
t
p
t
)]
p
t
)
p
t
)]
¬
∨
[(Bel
X
t
¬
∧
(Goal
X
t
neutral
:[(Bel
X
t
p
t
)
(Goal
X
t
p
t
)
(Goal
X
t
p
t
)]
[(Bel
X
t
p
t
)
(Goal
X
t
p
t
)
∧¬
∧¬
¬
∨
¬
∧¬
p
t
)]
ambivalent
:[(Bel
X
t
(Goal
X
t
∧¬
¬
p
t
)
(Goal
X
t
p
t
)
(Goal
X
t
p
t
)]
[(Bel
X
t
p
t
)
(Goal
X
t
p
t
)
∧
∧
¬
∨
¬
∧
∧
(Goal
X
t
p
t
)]
where
t
>
¬
t
.
2.3.3 The Quantitative Aspects of Mental Attitudes
Decomposing in terms of beliefs and goals is not enough. We need 'quantitative' parame-
ters. Frustration and pain have an
intensity
, can be more or less severe; the same holds for
surprise, disappointment, relief, hope, joy,
Since they are clearly related with what the
agent believes, expects, likes, pursues, can we account for those dimensions on the basis
of our (de)composition of those mental states, and of the basic epistemic and motivational
representations? We claim so.
Given the two basic ingredients of any Exp (defined as different from simple forecast or
prediction) Beliefs
...
+
Goals, we postulate that:
P1:
Beliefs
and
Goals
have specific quantitative dimensions; which are basically
independent from each other.
Beliefs have strength, a degree of subjective certainty
; the subject is more or less sure and
committed about their content.
Goals have a value, a subjective importance for the agent.
To simplify, we may have very important goals combined with uncertain predictions; pretty
sure forecasts for not very relevant objectives; etc. Thus, we should explicitly represent these
Either, the expectation entails a cognitive evaluation. In fact, since the realization of
p
coincides with a goal, it
is “good”; while if the belief is the opposite of the goal, it implies a belief that the outcome of the world will
be 'bad'.
Or the expectation produces an implicit, intuitive appraisal, simply by activating associated affective responses
or somatic markers; or both.
Or the expected result will produce a
reward
for the agent, and - although not strictly driving its behavior, it
is positive for it since it will satisfy a drive and reinforce the behavior.
26
We analyze here only the expectations in a strong sense, with an explicit goal; but we mentioned expectations in
those forms of reactive, rule-based behaviors, first in order to stress how the notion of expectation always involves
theideaofa
valence
and of the agent being concerned and monitoring the world; second, to give an idea of more
elementary and forerunner forms of this construct. It is in fact the case of proto-expectations or expectations in
'Anticipatory-Classifiers' based behaviors, strictly conceived as reactive (not really goal-driven) behaviors, but based
on anticipatory representation of the outcomes (Butz and Hoffman, 2002), (Castelfranchi, Tummolini and Pezzulo,
2005), (Butz, 2002), (Drescher, 1991), (Pezzulo et al., 2008).
Search WWH ::
Custom Search