Information Technology Reference
In-Depth Information
reference does not apply for these objects because they lack clear spatio-temporal
limits, thus preventing the use of direct reference in interactions. Furthermore,
everyday concepts like those illustrated above are notoriously ill-defined; making
shared meaning even more mysterious [10]. In our current work, we hold the view
that shared meaning is possible because meaning is conventional, i.e., there is a
limited set of meanings that apply to a given situation [11], [12], [13]. Constraining
the number of concepts that apply on a given occasion, makes agreement a tractable
problem. However, even if a group of people has developed conceptual conventions,
the likely case is that each person instantiates a somewhat different version of those
concepts (e.g., people may conceptualize “leadership” in slightly different ways).
Furthermore, even if a group of people has conventions about more or less
dichotomous concepts (e.g., “cowardice” and “courage”), a person could still be
wrong about which one is being deployed by someone else at a given moment (e.g., if
someone says “suicide”, she may be thinking of “cowardice” while I may be thinking
of “courage”). Consequently, an individual can never know for sure whether someone
else agrees or not with his conceptualization of a given event (even when being
explicit). Agreement is a probabilistic inference [14].
The ABM we report here focuses on two probabilities that represent the above
mentioned inference. First, the probability of
true agreement
(symbolized by
p(a1)
),
which stands for the probability that two agents (an
observer
and an
actor
) agree on
something given that they instantiate different versions of the same concept (i.e., the
“leadership” example above). Second, the probability of
illusory agreement
(symbolized by
p(a2)
), which stands for the probability that observer and actor agree,
given that they instantiate different concepts altogether (i.e., the “courage” or
“cowardice” example above).
2
Conceptual Description of the ABM
Our current ABM represents a social group which has a set of conventional
conceptual states that, for ease of exposition, we will call the
focal set
. These states
can represent different versions of the same concept (e.g., different versions of
“leadership”; or a set of closely related concepts, such as “miserly”, “stingy”,
“scrooge”). Our
p(a1)
probability reflects the degree of overlap between the different
versions in the focal set (greater overlap implies greater probability of true
agreement). Our
p(a2)
probability reflects the degree of contrast against alternative
conceptualizations (lower contrast implies greater probability of illusory agreement).
The system models the dynamical trajectories of concepts as they become
increasingly or decreasingly relevant for agents depending on their capacity to
generate agreement of any type.
In each simulation run, agents act as observers and actors. Observers seek evidence
that actors share their concept. Actors have a certain probability that they will or will
not act according to the focal set concept. If they act according to the focal set
concept, that specific interaction has a probability
p(a1)
of providing observers
evidence of a shared concept. If actors don't act according to the focal set concept
