limited rationality; they then select a particular mode of action, and evaluate a
provisional satisfactory solution before iterating the process if necessary [6]. The
comparison of potential scenarios results in mathematical formulations, developed
from an economic perspective. The choice is equivalent to the optimization of a
function (the utility). A set of axioms expresses postulates about the properties of
preferences (for a discussion of the various theories see [7]). If an objective
assessment of the probabilities of the consequences of an act is not possible, the
decision maker can use ‘subjective’ probabilities. The probability distribution is
therefore measured according to their knowledge of the possible states of the object
they are interested in and upon which they wish to act.
Nevertheless, the assessment of subjective probabilities can be arbitrary and, in
many practical situations, the behaviour of agents does not reflect preferences that
are consistent with these axioms, without being necessarily called an ‘error’ [1]. To
account for the assumed ignorance of the decision maker of certain states and their
aversion to uncertainty, Gilboa and Schmeidler [8] showed that decisions can
sometimes be seen as the maximization of a form of expected utility that takes into
account the worst case scenario, which is consistent with the ‘maxmin’ model. It is
also possible to broaden the spectrum of reactions in uncertain situations [9], in
order to model less paranoid attitudes [1].
From this perspective, the behaviour of the decision maker is interpreted
according to a concept of ambiguity relative to their knowledge of the world—he
could not be sure of the meaning of the few information he got regarding the state
of the facilities—, which is how the questions of the Japanese investigators should
be understood (see Sect. 2). Faced with ambiguity about the state of Reactor 1, it
could be argued that Yoshida violated expected utility theory by not deciding to
immediately cool the core; or, alternatively, he demonstrated an aversion to
uncertainty, by deciding to rely on information about the water level, while
simultaneously preparing cooling mechanisms.
5 The fact that he did not foresee
hydrogen leaks can be interpreted in two ways using the ‘maxmin’ model: either he
failed (cognitively) in his assessment based on all of the objective information
available; or his attitude or imperfect knowledge led him to limit his choices to a
subset of possible states. As for his decision to send staff to gather information on
the state of the reactors before considering operations such as venting, this can be
interpreted as an example of incomplete preference, where the status quo is
maintained until such time as a conclusively better, new alternative appears.
Can we conclude that Yoshida acted irrationally, or do the models provide an
incomplete description of such decisions? Gilboa [10] argues that we qualify
behaviour as ‘irrational’ if whoever violates its precepts regrets their actions. The
regret expressed by Yoshida about the confidence he placed in Reactor 1’s water
level indicators (see Sect. 2) would suggest that the decision was irrational.
5
The superintendent had to manage limited resources and set priorities. Therefore, in this case it is
not possible to apply a principle of dominance and conclude that the “ignore the water level”
option was better.
Decision-Making in Extreme Situations …
173
provisional satisfactory solution before iterating the process if necessary [6]. The
comparison of potential scenarios results in mathematical formulations, developed
from an economic perspective. The choice is equivalent to the optimization of a
function (the utility). A set of axioms expresses postulates about the properties of
preferences (for a discussion of the various theories see [7]). If an objective
assessment of the probabilities of the consequences of an act is not possible, the
decision maker can use ‘subjective’ probabilities. The probability distribution is
therefore measured according to their knowledge of the possible states of the object
they are interested in and upon which they wish to act.
Nevertheless, the assessment of subjective probabilities can be arbitrary and, in
many practical situations, the behaviour of agents does not reflect preferences that
are consistent with these axioms, without being necessarily called an ‘error’ [1]. To
account for the assumed ignorance of the decision maker of certain states and their
aversion to uncertainty, Gilboa and Schmeidler [8] showed that decisions can
sometimes be seen as the maximization of a form of expected utility that takes into
account the worst case scenario, which is consistent with the ‘maxmin’ model. It is
also possible to broaden the spectrum of reactions in uncertain situations [9], in
order to model less paranoid attitudes [1].
From this perspective, the behaviour of the decision maker is interpreted
according to a concept of ambiguity relative to their knowledge of the world—he
could not be sure of the meaning of the few information he got regarding the state
of the facilities—, which is how the questions of the Japanese investigators should
be understood (see Sect. 2). Faced with ambiguity about the state of Reactor 1, it
could be argued that Yoshida violated expected utility theory by not deciding to
immediately cool the core; or, alternatively, he demonstrated an aversion to
uncertainty, by deciding to rely on information about the water level, while
simultaneously preparing cooling mechanisms.
5 The fact that he did not foresee
hydrogen leaks can be interpreted in two ways using the ‘maxmin’ model: either he
failed (cognitively) in his assessment based on all of the objective information
available; or his attitude or imperfect knowledge led him to limit his choices to a
subset of possible states. As for his decision to send staff to gather information on
the state of the reactors before considering operations such as venting, this can be
interpreted as an example of incomplete preference, where the status quo is
maintained until such time as a conclusively better, new alternative appears.
Can we conclude that Yoshida acted irrationally, or do the models provide an
incomplete description of such decisions? Gilboa [10] argues that we qualify
behaviour as ‘irrational’ if whoever violates its precepts regrets their actions. The
regret expressed by Yoshida about the confidence he placed in Reactor 1’s water
level indicators (see Sect. 2) would suggest that the decision was irrational.
5
The superintendent had to manage limited resources and set priorities. Therefore, in this case it is
not possible to apply a principle of dominance and conclude that the “ignore the water level”
option was better.
Decision-Making in Extreme Situations …
173