prefer to randomize among equally valued prospects. Some significant theoretical
results in economics extend to a nonexpected utility world if agents’ preferences
satisfy betweenness (see Section 5.3).
Various models have been proposed that do not impose betweenness. Chew,
Epstein, and Segal (1991) propose
quadratic utility theory
, which relies on a
weakened form of betweenness called
mixture symmetry
: if
q
|
r
then (
q
,
p
;
r
,
(1
2
p
))
|
(
q
, (1
2
p
);
r
,
p
). In this model, indifference curves may switch from
concave to convex (or vice versa) as we move across the triangle. Joao Becker and
Rakesh Sarin (1987) propose a model with even weaker restrictions. Their
lottery-
dependent utility
assumes only ordering, continuity, and monotonicity. The basic
model is conventional theory for minimalists as, without further restriction, it has
virtually no empirical content. The authors discuss a particular “exponential
form,” which implies fanning out.
An important subset of the betweenness nonconforming theories has an addi-
tional feature absent from the models discussed so far. Upto this point we have
considered a variety of conventional theories, each of which generates the prop-
erty of fanning out. Although they achieve it in different ways, there is one struc-
tural similarity between these theories: each operates by assigning subjective
weights—or utilities—to consequences; the value assigned to any given prospect
is then determined by some function that combines these utilities with
objective
probabilities. Another variant of the conventional
strategy involves the use of
probability transformation functions that convert objective probabilities into sub-
jective
decision weights
. An important feature of these models is that, excepting
special cases, betweenness does not hold.
4.1.2. THEORIES WITH DECISION WEIGHTS
There is evidence for the view that individuals have subjective attitudes to probabil-
ities that are distinct from attitudes to consequences. For instance, according to
Nick Pidgeon et al. (1992), when people are asked to make judgments about the
likelihood of death occurring from different causes, they tend to underestimate the
number of deaths from relatively frequent causes, while overestimating deaths due
to relatively infrequent causes. Similarly, apparent biases in the subjective odds re-
vealed in studies of racetrack betting have been explained as bettors being either
oversensitive to the chances of winning on long shots (Ali 1977; Thaler and Ziemba
1988), or oversensitive to the chances of losing on favorites (Jullien and Salanié
1997). These effects might be revealing misperception of objective probabilities or
a tendency for individuals subjectively to weight objective probabilities. Either way,
in
principle, such effects could be captured in models incorporating
decision
weights
. A number of such theories can be understood as variants of the following
functional form where the
w
i
terms represent decision weights:
V
(
q
)
5
•
i
p
i
?
u
(
x
i
).
(5)
I will call this the
decision-weighted form
. Theories of this type were first dis-
cussed by Ward Edwards (1955, 1962). In its most basic form, consequences are
Do'stlaringiz bilan baham: